* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 01/14] iscsi: Resolve iscsi_proto.h naming conflicts with drivers/target/iscsi Nicholas A. Bellinger
` (14 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi
Cc: James Bottomley, Christoph Hellwig, Mike Christie,
Hannes Reinecke, FUJITA Tomonori, Boaz Harrosh
On Tue, 2011-03-22 at 20:06 -0700, Nicholas A. Bellinger wrote:
> From: Nicholas Bellinger <nab@linux-iscsi.org>
>
> Hi James and Co,
>
> Here is an updated v2 mergeable + squashable version of iscsi_target_mod based
> upon Mike Christie's recent feedback which contains two additional minor changes
> to iscsi_proto.h and target core. Thanks for your review Mike!
>
> This patch series is made against the following scsi-misc-2.6.git/master HEAD:
>
> commit aa7131dcdb262cd08bc39a15ebd788a6bc0cda2f
> Author: James Bottomley <James.Bottomley@suse.de>
> Date: Fri Mar 18 11:24:23 2011 -0400
>
> [SCSI] ses: add subenclosure support
>
> Please merge patch #1 + #2 as a seperate iscsi commits after Mike's review
> and signoff. Patch #3 is a seperate target_core int_to_scsilun() conversion
> patch so we can drop target_core_fabric_ops->pack_lun() for the initial merge
> of iscsi-target.
>
> >From there squash #4 -> #14 using the commit message message of patch #4 for
> an initial commit for iscsi-target.
>
> I will follow with any further comments from Christoph, Mike & Co.
>
Here is the missing v1 -> v2 changelog btw:
*) Remove mixed case in iscsi_target_auth.c
*) Remove unnecessary ____cacheline_aligned usage
*) Cleanup code to use iscsit_release_cmd()
*) Convert to use scsilun_to_int and int_to_scsilun
*) Convert iscsi_session->session_[usage_count,waiting_on_uc] to int
*) Convert iscsi_conn->conn_[usage_count, waiting_on_uc] to int
*) Convert to use iscsi_proto.h iscsi_sna_gt
*) Remove unused iscsit_ntoa
*) Remove unused inet conversions
*) Remove unused struct net_device state checks
*) Convert to use kernel_[send,recv]msg() and struct kvec
*) Use single np->np_ip and conn->login_ip for AF_INET*
*) Use list_is_last() in iscsit_[free,stop]_session()
*) Fix up lio_target_call_delnpfromtpg output
Thanks again to Mike Christie for his review!
--nab
^ permalink raw reply [flat|nested] 40+ messages in thread* [PATCH-v2 01/14] iscsi: Resolve iscsi_proto.h naming conflicts with drivers/target/iscsi
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
2011-03-23 3:06 ` Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 02/14] iscsi: Add Serial Number Arithmetic LT and GT into iscsi_proto.h Nicholas A. Bellinger
` (13 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch renames the following iscsi_proto.h structures to avoid
namespace issues with drivers/target/iscsi/iscsi_target_core.h:
*) struct iscsi_cmd -> struct iscsi_scsi_req
*) struct iscsi_cmd_rsp -> struct iscsi_scsi_rsp
*) struct iscsi_login -> struct iscsi_login_req
This patch includes useful ISCSI_FLAG_LOGIN_[CURRENT,NEXT]_STAGE*,
and ISCSI_FLAG_SNACK_TYPE_* definitions used by iscsi_target_mod, and
fixes the incorrect definition of struct iscsi_snack to following
RFC-3720 Section 10.16. SNACK Request.
Also, this patch updates libiscsi, iSER, be2iscsi, and bn2xi to
use the updated structure definitions in a handful of locations.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/infiniband/ulp/iser/iser_initiator.c | 2 +-
drivers/scsi/be2iscsi/be_main.h | 4 +-
drivers/scsi/bnx2i/bnx2i_hwi.c | 8 +++---
drivers/scsi/bnx2i/bnx2i_iscsi.c | 2 +-
drivers/scsi/libiscsi.c | 6 ++--
include/scsi/iscsi_proto.h | 30 +++++++++++++++++++-------
6 files changed, 33 insertions(+), 19 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c
index 95a08a8..d73bab3 100644
--- a/drivers/infiniband/ulp/iser/iser_initiator.c
+++ b/drivers/infiniband/ulp/iser/iser_initiator.c
@@ -271,7 +271,7 @@ int iser_send_command(struct iscsi_conn *conn,
unsigned long edtl;
int err;
struct iser_data_buf *data_buf;
- struct iscsi_cmd *hdr = (struct iscsi_cmd *)task->hdr;
+ struct iscsi_scsi_req *hdr = (struct iscsi_scsi_req *)task->hdr;
struct scsi_cmnd *sc = task->sc;
struct iser_tx_desc *tx_desc = &iser_task->desc;
diff --git a/drivers/scsi/be2iscsi/be_main.h b/drivers/scsi/be2iscsi/be_main.h
index 90eb74f..205c502 100644
--- a/drivers/scsi/be2iscsi/be_main.h
+++ b/drivers/scsi/be2iscsi/be_main.h
@@ -398,7 +398,7 @@ struct amap_pdu_data_out {
};
struct be_cmd_bhs {
- struct iscsi_cmd iscsi_hdr;
+ struct iscsi_scsi_req iscsi_hdr;
unsigned char pad1[16];
struct pdu_data_out iscsi_data_pdu;
unsigned char pad2[BE_SENSE_INFO_SIZE -
@@ -429,7 +429,7 @@ struct be_nonio_bhs {
};
struct be_status_bhs {
- struct iscsi_cmd iscsi_hdr;
+ struct iscsi_scsi_req iscsi_hdr;
unsigned char pad1[16];
/**
* The plus 2 below is to hold the sense info length that gets
diff --git a/drivers/scsi/bnx2i/bnx2i_hwi.c b/drivers/scsi/bnx2i/bnx2i_hwi.c
index 1da34c0..665ebdc 100644
--- a/drivers/scsi/bnx2i/bnx2i_hwi.c
+++ b/drivers/scsi/bnx2i/bnx2i_hwi.c
@@ -326,11 +326,11 @@ int bnx2i_send_iscsi_login(struct bnx2i_conn *bnx2i_conn,
{
struct bnx2i_cmd *bnx2i_cmd;
struct bnx2i_login_request *login_wqe;
- struct iscsi_login *login_hdr;
+ struct iscsi_login_req *login_hdr;
u32 dword;
bnx2i_cmd = (struct bnx2i_cmd *)task->dd_data;
- login_hdr = (struct iscsi_login *)task->hdr;
+ login_hdr = (struct iscsi_login_req *)task->hdr;
login_wqe = (struct bnx2i_login_request *)
bnx2i_conn->ep->qp.sq_prod_qe;
@@ -1341,7 +1341,7 @@ static int bnx2i_process_scsi_cmd_resp(struct iscsi_session *session,
struct bnx2i_cmd_response *resp_cqe;
struct bnx2i_cmd *bnx2i_cmd;
struct iscsi_task *task;
- struct iscsi_cmd_rsp *hdr;
+ struct iscsi_scsi_rsp *hdr;
u32 datalen = 0;
resp_cqe = (struct bnx2i_cmd_response *)cqe;
@@ -1368,7 +1368,7 @@ static int bnx2i_process_scsi_cmd_resp(struct iscsi_session *session,
}
bnx2i_iscsi_unmap_sg_list(bnx2i_cmd);
- hdr = (struct iscsi_cmd_rsp *)task->hdr;
+ hdr = (struct iscsi_scsi_rsp *)task->hdr;
resp_cqe = (struct bnx2i_cmd_response *)cqe;
hdr->opcode = resp_cqe->op_code;
hdr->max_cmdsn = cpu_to_be32(resp_cqe->max_cmd_sn);
diff --git a/drivers/scsi/bnx2i/bnx2i_iscsi.c b/drivers/scsi/bnx2i/bnx2i_iscsi.c
index 1809f9c..9b50aaf 100644
--- a/drivers/scsi/bnx2i/bnx2i_iscsi.c
+++ b/drivers/scsi/bnx2i/bnx2i_iscsi.c
@@ -1206,7 +1206,7 @@ static int bnx2i_task_xmit(struct iscsi_task *task)
struct bnx2i_conn *bnx2i_conn = conn->dd_data;
struct scsi_cmnd *sc = task->sc;
struct bnx2i_cmd *cmd = task->dd_data;
- struct iscsi_cmd *hdr = (struct iscsi_cmd *) task->hdr;
+ struct iscsi_scsi_req *hdr = (struct iscsi_scsi_req *) task->hdr;
/*
* If there is no scsi_cmnd this must be a mgmt task
diff --git a/drivers/scsi/libiscsi.c b/drivers/scsi/libiscsi.c
index 0c550d5..9c3cb4e 100644
--- a/drivers/scsi/libiscsi.c
+++ b/drivers/scsi/libiscsi.c
@@ -360,7 +360,7 @@ static int iscsi_prep_scsi_cmd_pdu(struct iscsi_task *task)
struct iscsi_conn *conn = task->conn;
struct iscsi_session *session = conn->session;
struct scsi_cmnd *sc = task->sc;
- struct iscsi_cmd *hdr;
+ struct iscsi_scsi_req *hdr;
unsigned hdrlength, cmd_len;
itt_t itt;
int rc;
@@ -374,7 +374,7 @@ static int iscsi_prep_scsi_cmd_pdu(struct iscsi_task *task)
if (rc)
return rc;
}
- hdr = (struct iscsi_cmd *) task->hdr;
+ hdr = (struct iscsi_scsi_req *) task->hdr;
itt = hdr->itt;
memset(hdr, 0, sizeof(*hdr));
@@ -830,7 +830,7 @@ static void iscsi_scsi_cmd_rsp(struct iscsi_conn *conn, struct iscsi_hdr *hdr,
struct iscsi_task *task, char *data,
int datalen)
{
- struct iscsi_cmd_rsp *rhdr = (struct iscsi_cmd_rsp *)hdr;
+ struct iscsi_scsi_rsp *rhdr = (struct iscsi_scsi_rsp *)hdr;
struct iscsi_session *session = conn->session;
struct scsi_cmnd *sc = task->sc;
diff --git a/include/scsi/iscsi_proto.h b/include/scsi/iscsi_proto.h
index dd0a52c..a9c9058 100644
--- a/include/scsi/iscsi_proto.h
+++ b/include/scsi/iscsi_proto.h
@@ -116,7 +116,7 @@ struct iscsi_ahs_hdr {
#define ISCSI_CDB_SIZE 16
/* iSCSI PDU Header */
-struct iscsi_cmd {
+struct iscsi_scsi_req {
uint8_t opcode;
uint8_t flags;
__be16 rsvd2;
@@ -161,7 +161,7 @@ struct iscsi_ecdb_ahdr {
};
/* SCSI Response Header */
-struct iscsi_cmd_rsp {
+struct iscsi_scsi_rsp {
uint8_t opcode;
uint8_t flags;
uint8_t response;
@@ -406,7 +406,7 @@ struct iscsi_text_rsp {
};
/* Login Header */
-struct iscsi_login {
+struct iscsi_login_req {
uint8_t opcode;
uint8_t flags;
uint8_t max_version; /* Max. version supported */
@@ -427,7 +427,13 @@ struct iscsi_login {
#define ISCSI_FLAG_LOGIN_TRANSIT 0x80
#define ISCSI_FLAG_LOGIN_CONTINUE 0x40
#define ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK 0x0C /* 2 bits */
+#define ISCSI_FLAG_LOGIN_CURRENT_STAGE1 0x04
+#define ISCSI_FLAG_LOGIN_CURRENT_STAGE2 0x08
+#define ISCSI_FLAG_LOGIN_CURRENT_STAGE3 0x0C
#define ISCSI_FLAG_LOGIN_NEXT_STAGE_MASK 0x03 /* 2 bits */
+#define ISCSI_FLAG_LOGIN_NEXT_STAGE1 0x01
+#define ISCSI_FLAG_LOGIN_NEXT_STAGE2 0x02
+#define ISCSI_FLAG_LOGIN_NEXT_STAGE3 0x03
#define ISCSI_LOGIN_CURRENT_STAGE(flags) \
((flags & ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK) >> 2)
@@ -550,17 +556,25 @@ struct iscsi_logout_rsp {
struct iscsi_snack {
uint8_t opcode;
uint8_t flags;
- uint8_t rsvd2[14];
+ uint8_t rsvd2[2];
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t lun[8];
itt_t itt;
+ __be32 ttt;
+ uint8_t rsvd3[4];
+ __be32 exp_statsn;
+ uint8_t rsvd4[8];
__be32 begrun;
__be32 runlength;
- __be32 exp_statsn;
- __be32 rsvd3;
- __be32 exp_datasn;
- uint8_t rsvd6[8];
};
/* SNACK PDU flags */
+#define ISCSI_FLAG_SNACK_TYPE_DATA 0
+#define ISCSI_FLAG_SNACK_TYPE_R2T 0
+#define ISCSI_FLAG_SNACK_TYPE_STATUS 1
+#define ISCSI_FLAG_SNACK_TYPE_DATA_ACK 2
+#define ISCSI_FLAG_SNACK_TYPE_RDATA 3
#define ISCSI_FLAG_SNACK_TYPE_MASK 0x0F /* 4 bits */
/* Reject Message Header */
--
1.7.4.1
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 02/14] iscsi: Add Serial Number Arithmetic LT and GT into iscsi_proto.h
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 01/14] iscsi: Resolve iscsi_proto.h naming conflicts with drivers/target/iscsi Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 03/14] target: Convert REPORT_LUNs to use int_to_scsilun Nicholas A. Bellinger
` (12 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch moves the iscsi_sna_lt() and iscsi_sna_lte(), along with
iscsi_sna_gt() and iscsi_sna_gte() from iscsi_target_mod into
static inlines inside of include/scsi/iscsi_proto.h
Reported-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/scsi/libiscsi.c | 16 ----------------
include/scsi/iscsi_proto.h | 29 +++++++++++++++++++++++++++++
2 files changed, 29 insertions(+), 16 deletions(-)
diff --git a/drivers/scsi/libiscsi.c b/drivers/scsi/libiscsi.c
index 9c3cb4e..3461a03 100644
--- a/drivers/scsi/libiscsi.c
+++ b/drivers/scsi/libiscsi.c
@@ -84,22 +84,6 @@ MODULE_PARM_DESC(debug_libiscsi_eh,
__func__, ##arg); \
} while (0);
-/* Serial Number Arithmetic, 32 bits, less than, RFC1982 */
-#define SNA32_CHECK 2147483648UL
-
-static int iscsi_sna_lt(u32 n1, u32 n2)
-{
- return n1 != n2 && ((n1 < n2 && (n2 - n1 < SNA32_CHECK)) ||
- (n1 > n2 && (n2 - n1 < SNA32_CHECK)));
-}
-
-/* Serial Number Arithmetic, 32 bits, less than, RFC1982 */
-static int iscsi_sna_lte(u32 n1, u32 n2)
-{
- return n1 == n2 || ((n1 < n2 && (n2 - n1 < SNA32_CHECK)) ||
- (n1 > n2 && (n2 - n1 < SNA32_CHECK)));
-}
-
inline void iscsi_conn_queue_work(struct iscsi_conn *conn)
{
struct Scsi_Host *shost = conn->session->host;
diff --git a/include/scsi/iscsi_proto.h b/include/scsi/iscsi_proto.h
index a9c9058..626b0344 100644
--- a/include/scsi/iscsi_proto.h
+++ b/include/scsi/iscsi_proto.h
@@ -33,6 +33,35 @@
#define ISCSI_PAD_LEN 4
/*
+ * Serial Number Arithmetic, 32 bits, RFC1982
+ */
+#define SNA32_CHECK 2147483648UL
+
+static inline int iscsi_sna_lt(u32 n1, u32 n2)
+{
+ return n1 != n2 && ((n1 < n2 && (n2 - n1 < SNA32_CHECK)) ||
+ (n1 > n2 && (n2 - n1 < SNA32_CHECK)));
+}
+
+static inline int iscsi_sna_lte(u32 n1, u32 n2)
+{
+ return n1 == n2 || ((n1 < n2 && (n2 - n1 < SNA32_CHECK)) ||
+ (n1 > n2 && (n2 - n1 < SNA32_CHECK)));
+}
+
+static inline int iscsi_sna_gt(u32 n1, u32 n2)
+{
+ return n1 != n2 && (((n1 < n2) && ((n2 - n1) > SNA32_CHECK)) ||
+ ((n1 > n2) && ((n1 - n2) < SNA32_CHECK)));
+}
+
+static inline int iscsi_sna_gte(u32 n1, u32 n2)
+{
+ return n1 == n2 || (((n1 < n2) && ((n2 - n1) > SNA32_CHECK)) ||
+ ((n1 > n2) && ((n1 - n2) < SNA32_CHECK)));
+}
+
+/*
* useful common(control and data pathes) macro
*/
#define ntoh24(p) (((p)[0] << 16) | ((p)[1] << 8) | ((p)[2]))
--
1.7.4.1
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 03/14] target: Convert REPORT_LUNs to use int_to_scsilun
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (2 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 02/14] iscsi: Add Serial Number Arithmetic LT and GT into iscsi_proto.h Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 04/14] iscsi-target: Add iSCSI fabric support for target v4 Nicholas A. Bellinger
` (11 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch converts transport_core_report_lun_response() to use
drivers/scsi/scsi_scan.c:int_to_scsilun instead of using the
struct target_core_fabric_ops->pack_lun() fabric provided API vector.
It also removes the tfo->pack_lun check from target_fabric_tf_ops_check()
and will be removed from struct target_core_fabric_ops in a seperate
patch.
Reported-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/target_core_configfs.c | 4 ----
drivers/target/target_core_device.c | 26 +++++---------------------
2 files changed, 5 insertions(+), 25 deletions(-)
diff --git a/drivers/target/target_core_configfs.c b/drivers/target/target_core_configfs.c
index a5f44a6..ee6fad9 100644
--- a/drivers/target/target_core_configfs.c
+++ b/drivers/target/target_core_configfs.c
@@ -497,10 +497,6 @@ static int target_fabric_tf_ops_check(
printk(KERN_ERR "Missing tfo->is_state_remove()\n");
return -EINVAL;
}
- if (!(tfo->pack_lun)) {
- printk(KERN_ERR "Missing tfo->pack_lun()\n");
- return -EINVAL;
- }
/*
* We at least require tfo->fabric_make_wwn(), tfo->fabric_drop_wwn()
* tfo->fabric_make_tpg() and tfo->fabric_drop_tpg() in
diff --git a/drivers/target/target_core_device.c b/drivers/target/target_core_device.c
index 3fb8e32..7605967 100644
--- a/drivers/target/target_core_device.c
+++ b/drivers/target/target_core_device.c
@@ -38,6 +38,7 @@
#include <net/sock.h>
#include <net/tcp.h>
#include <scsi/scsi.h>
+#include <scsi/scsi_device.h>
#include <target/target_core_base.h>
#include <target/target_core_device.h>
@@ -658,8 +659,7 @@ int transport_core_report_lun_response(struct se_cmd *se_cmd)
struct se_session *se_sess = SE_SESS(se_cmd);
struct se_task *se_task;
unsigned char *buf = (unsigned char *)T_TASK(se_cmd)->t_task_buf;
- u32 cdb_offset = 0, lun_count = 0, offset = 8;
- u64 i, lun;
+ u32 cdb_offset = 0, lun_count = 0, offset = 8, i;
list_for_each_entry(se_task, &T_TASK(se_cmd)->t_task_list, t_list)
break;
@@ -675,15 +675,7 @@ int transport_core_report_lun_response(struct se_cmd *se_cmd)
* a $FABRIC_MOD. In that case, report LUN=0 only.
*/
if (!(se_sess)) {
- lun = 0;
- buf[offset++] = ((lun >> 56) & 0xff);
- buf[offset++] = ((lun >> 48) & 0xff);
- buf[offset++] = ((lun >> 40) & 0xff);
- buf[offset++] = ((lun >> 32) & 0xff);
- buf[offset++] = ((lun >> 24) & 0xff);
- buf[offset++] = ((lun >> 16) & 0xff);
- buf[offset++] = ((lun >> 8) & 0xff);
- buf[offset++] = (lun & 0xff);
+ int_to_scsilun(0, (struct scsi_lun *)&buf[offset]);
lun_count = 1;
goto done;
}
@@ -703,16 +695,8 @@ int transport_core_report_lun_response(struct se_cmd *se_cmd)
if ((cdb_offset + 8) >= se_cmd->data_length)
continue;
- lun = cpu_to_be64(CMD_TFO(se_cmd)->pack_lun(deve->mapped_lun));
- buf[offset++] = ((lun >> 56) & 0xff);
- buf[offset++] = ((lun >> 48) & 0xff);
- buf[offset++] = ((lun >> 40) & 0xff);
- buf[offset++] = ((lun >> 32) & 0xff);
- buf[offset++] = ((lun >> 24) & 0xff);
- buf[offset++] = ((lun >> 16) & 0xff);
- buf[offset++] = ((lun >> 8) & 0xff);
- buf[offset++] = (lun & 0xff);
- cdb_offset += 8;
+ int_to_scsilun(deve->mapped_lun, (struct scsi_lun *)&buf[offset]);
+ cdb_offset += 16;
}
spin_unlock_irq(&SE_NODE_ACL(se_sess)->device_list_lock);
--
1.7.4.1
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 04/14] iscsi-target: Add iSCSI fabric support for target v4
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (3 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 03/14] target: Convert REPORT_LUNs to use int_to_scsilun Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-04-07 7:34 ` Mike Christie
2011-03-23 3:06 ` [PATCH-v2 05/14] iscsi-target: Add TCM v4 compatiable ConfigFS control plane Nicholas A. Bellinger
` (10 subsequent siblings)
15 siblings, 1 reply; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
The RisingTide Systems iSCSI target module is a full featured in-kernel
software implementation of iSCSI target mode (RFC-3720) for the mainline
target v4 infrastructure code. More information can be found here:
http://linux-iscsi.org/wiki/ISCSI
This includes support for:
* RFC-3720 defined request / response state machines and support for
all defined iSCSI operation codes from Section 10.2.1.2 using libiscsi
include/scsi/iscsi_proto.h PDU definitions
* Target v4 compatible control plane using the generic layout in
target_core_fabric_configfs.c and fabric dependent attributes
within /sys/kernel/config/target/iscsi/ subdirectories.
* Target v4 compatible iSCSI statistics based on RFC-4544 (iSCSI MIBS)
* Support for IPv6 and IPv4 network portals in M:N mapping to TPGs
* iSCSI Error Recovery Hierarchy support
* Per iSCSI connection RX/TX thread pair scheduling affinity
* crc32c + crc32c_intel SSEv4 instruction offload support using libcrypto
* CHAP Authentication support using libcrypto
Significant feedback for mainline code cleanups by Christoph Hellwig and
Mike Christie.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target.c | 5017 ++++++++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target.h | 32 +
drivers/target/iscsi/iscsi_target_core.h | 881 ++++++
3 files changed, 5930 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target.c
create mode 100644 drivers/target/iscsi/iscsi_target.h
create mode 100644 drivers/target/iscsi/iscsi_target_core.h
diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
new file mode 100644
index 0000000..c81f5cc
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -0,0 +1,5017 @@
+/*******************************************************************************
+ * This file contains main functions related to the iSCSI Target Core Driver.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/kthread.h>
+#include <linux/crypto.h>
+#include <linux/completion.h>
+#include <asm/unaligned.h>
+#include <scsi/scsi_device.h>
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_tmr.h>
+#include <target/target_core_transport.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_parameters.h"
+#include "iscsi_target_seq_pdu_list.h"
+#include "iscsi_target_tq.h"
+#include "iscsi_target_configfs.h"
+#include "iscsi_target_datain_values.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_erl1.h"
+#include "iscsi_target_erl2.h"
+#include "iscsi_target_login.h"
+#include "iscsi_target_tmr.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target.h"
+#include "iscsi_target_device.h"
+#include "iscsi_target_stat.h"
+
+LIST_HEAD(g_tiqn_list);
+LIST_HEAD(g_np_list);
+DEFINE_SPINLOCK(tiqn_lock);
+DEFINE_SPINLOCK(np_lock);
+
+static struct idr tiqn_idr;
+struct idr sess_idr;
+struct mutex auth_id_lock;
+spinlock_t sess_idr_lock;
+
+struct iscsi_global *iscsi_global;
+
+struct kmem_cache *lio_cmd_cache;
+struct kmem_cache *lio_qr_cache;
+struct kmem_cache *lio_dr_cache;
+struct kmem_cache *lio_ooo_cache;
+struct kmem_cache *lio_r2t_cache;
+
+static void iscsit_rx_thread_wait_for_tcp(struct iscsi_conn *);
+
+static int iscsit_handle_immediate_data(struct iscsi_cmd *,
+ unsigned char *buf, u32);
+static inline int iscsit_send_data_in(struct iscsi_cmd *, struct iscsi_conn *,
+ struct se_unmap_sg *, int *);
+static inline int iscsit_send_logout_response(struct iscsi_cmd *, struct iscsi_conn *);
+static inline int iscsit_send_nopin_response(struct iscsi_cmd *, struct iscsi_conn *);
+static inline int iscsit_send_status(struct iscsi_cmd *, struct iscsi_conn *);
+static int iscsit_send_task_mgt_rsp(struct iscsi_cmd *, struct iscsi_conn *);
+static int iscsit_send_text_rsp(struct iscsi_cmd *, struct iscsi_conn *);
+static int iscsit_send_reject(struct iscsi_cmd *, struct iscsi_conn *);
+static int iscsit_logout_post_handler(struct iscsi_cmd *, struct iscsi_conn *);
+
+struct iscsi_tiqn *iscsit_get_tiqn_for_login(unsigned char *buf)
+{
+ struct iscsi_tiqn *tiqn = NULL;
+
+ spin_lock(&tiqn_lock);
+ list_for_each_entry(tiqn, &g_tiqn_list, tiqn_list) {
+ if (!strcmp(tiqn->tiqn, buf)) {
+
+ spin_lock(&tiqn->tiqn_state_lock);
+ if (tiqn->tiqn_state == TIQN_STATE_ACTIVE) {
+ tiqn->tiqn_access_count++;
+ spin_unlock(&tiqn->tiqn_state_lock);
+ spin_unlock(&tiqn_lock);
+ return tiqn;
+ }
+ spin_unlock(&tiqn->tiqn_state_lock);
+ }
+ }
+ spin_unlock(&tiqn_lock);
+
+ return NULL;
+}
+
+static int iscsit_set_tiqn_shutdown(struct iscsi_tiqn *tiqn)
+{
+ spin_lock(&tiqn->tiqn_state_lock);
+ if (tiqn->tiqn_state == TIQN_STATE_ACTIVE) {
+ tiqn->tiqn_state = TIQN_STATE_SHUTDOWN;
+ spin_unlock(&tiqn->tiqn_state_lock);
+ return 0;
+ }
+ spin_unlock(&tiqn->tiqn_state_lock);
+
+ return -1;
+}
+
+void iscsit_put_tiqn_for_login(struct iscsi_tiqn *tiqn)
+{
+ spin_lock(&tiqn->tiqn_state_lock);
+ tiqn->tiqn_access_count--;
+ spin_unlock(&tiqn->tiqn_state_lock);
+}
+
+/*
+ * Note that IQN formatting is expected to be done in userspace, and
+ * no explict IQN format checks are done here.
+ */
+struct iscsi_tiqn *iscsit_add_tiqn(unsigned char *buf, int *ret)
+{
+ struct iscsi_tiqn *tiqn = NULL;
+
+ if (strlen(buf) > ISCSI_IQN_LEN) {
+ printk(KERN_ERR "Target IQN exceeds %d bytes\n",
+ ISCSI_IQN_LEN);
+ *ret = -1;
+ return NULL;
+ }
+
+ spin_lock(&tiqn_lock);
+ list_for_each_entry(tiqn, &g_tiqn_list, tiqn_list) {
+ if (!strcmp(tiqn->tiqn, buf)) {
+ printk(KERN_ERR "Target IQN: %s already exists in Core\n",
+ tiqn->tiqn);
+ spin_unlock(&tiqn_lock);
+ *ret = -1;
+ return NULL;
+ }
+ }
+ spin_unlock(&tiqn_lock);
+
+ tiqn = kzalloc(sizeof(struct iscsi_tiqn), GFP_KERNEL);
+ if (!tiqn) {
+ printk(KERN_ERR "Unable to allocate struct iscsi_tiqn\n");
+ *ret = -1;
+ return NULL;
+ }
+
+ sprintf(tiqn->tiqn, "%s", buf);
+ INIT_LIST_HEAD(&tiqn->tiqn_list);
+ INIT_LIST_HEAD(&tiqn->tiqn_tpg_list);
+ spin_lock_init(&tiqn->tiqn_state_lock);
+ spin_lock_init(&tiqn->tiqn_tpg_lock);
+ spin_lock_init(&tiqn->sess_err_stats.lock);
+ spin_lock_init(&tiqn->login_stats.lock);
+ spin_lock_init(&tiqn->logout_stats.lock);
+
+ if (!idr_pre_get(&tiqn_idr, GFP_KERNEL)) {
+ printk(KERN_ERR "idr_pre_get() for tiqn_idr failed\n");
+ kfree(tiqn);
+ *ret = -1;
+ return NULL;
+ }
+ tiqn->tiqn_state = TIQN_STATE_ACTIVE;
+
+ spin_lock(&tiqn_lock);
+ idr_get_new(&tiqn_idr, NULL, &tiqn->tiqn_index);
+ list_add_tail(&tiqn->tiqn_list, &g_tiqn_list);
+ spin_unlock(&tiqn_lock);
+
+ printk(KERN_INFO "CORE[0] - Added iSCSI Target IQN: %s\n", tiqn->tiqn);
+
+ return tiqn;
+
+}
+
+void __iscsit_del_tiqn(struct iscsi_tiqn *tiqn)
+{
+ spin_lock(&tiqn_lock);
+ list_del(&tiqn->tiqn_list);
+ idr_remove(&tiqn_idr, tiqn->tiqn_index);
+ spin_unlock(&tiqn_lock);
+
+ printk(KERN_INFO "CORE[0] - Deleted iSCSI Target IQN: %s\n",
+ tiqn->tiqn);
+ kfree(tiqn);
+}
+
+static void iscsit_wait_for_tiqn(struct iscsi_tiqn *tiqn)
+{
+ /*
+ * Wait for accesses to said struct iscsi_tiqn to end.
+ */
+ spin_lock(&tiqn->tiqn_state_lock);
+ while (tiqn->tiqn_access_count != 0) {
+ spin_unlock(&tiqn->tiqn_state_lock);
+ msleep(10);
+ spin_lock(&tiqn->tiqn_state_lock);
+ }
+ spin_unlock(&tiqn->tiqn_state_lock);
+}
+
+void iscsit_del_tiqn(struct iscsi_tiqn *tiqn)
+{
+ /*
+ * iscsit_set_tiqn_shutdown sets tiqn->tiqn_state = TIQN_STATE_SHUTDOWN
+ * while holding tiqn->tiqn_state_lock. This means that all subsequent
+ * attempts to access this struct iscsi_tiqn will fail from both transport
+ * fabric and control code paths.
+ */
+ if (iscsit_set_tiqn_shutdown(tiqn) < 0) {
+ printk(KERN_ERR "iscsit_set_tiqn_shutdown() failed\n");
+ return;
+ }
+
+ iscsit_wait_for_tiqn(tiqn);
+ __iscsit_del_tiqn(tiqn);
+}
+
+int iscsit_access_np(struct iscsi_np *np, struct iscsi_portal_group *tpg)
+{
+ int ret;
+ /*
+ * Determine if the network portal is accepting storage traffic.
+ */
+ spin_lock_bh(&np->np_thread_lock);
+ if (np->np_thread_state != ISCSI_NP_THREAD_ACTIVE) {
+ spin_unlock_bh(&np->np_thread_lock);
+ return -1;
+ }
+ if (np->np_login_tpg) {
+ printk(KERN_ERR "np->np_login_tpg() is not NULL!\n");
+ spin_unlock_bh(&np->np_thread_lock);
+ return -1;
+ }
+ spin_unlock_bh(&np->np_thread_lock);
+ /*
+ * Determine if the portal group is accepting storage traffic.
+ */
+ spin_lock_bh(&tpg->tpg_state_lock);
+ if (tpg->tpg_state != TPG_STATE_ACTIVE) {
+ spin_unlock_bh(&tpg->tpg_state_lock);
+ return -1;
+ }
+ spin_unlock_bh(&tpg->tpg_state_lock);
+
+ /*
+ * Here we serialize access across the TIQN+TPG Tuple.
+ */
+ ret = mutex_lock_interruptible(&tpg->np_login_lock);
+ if ((ret != 0) || signal_pending(current))
+ return -1;
+
+ spin_lock_bh(&tpg->tpg_state_lock);
+ if (tpg->tpg_state != TPG_STATE_ACTIVE) {
+ spin_unlock_bh(&tpg->tpg_state_lock);
+ return -1;
+ }
+ spin_unlock_bh(&tpg->tpg_state_lock);
+
+ spin_lock_bh(&np->np_thread_lock);
+ np->np_login_tpg = tpg;
+ spin_unlock_bh(&np->np_thread_lock);
+
+ return 0;
+}
+
+int iscsit_deaccess_np(struct iscsi_np *np, struct iscsi_portal_group *tpg)
+{
+ struct iscsi_tiqn *tiqn = tpg->tpg_tiqn;
+
+ spin_lock_bh(&np->np_thread_lock);
+ np->np_login_tpg = NULL;
+ spin_unlock_bh(&np->np_thread_lock);
+
+ mutex_unlock(&tpg->np_login_lock);
+
+ if (tiqn)
+ iscsit_put_tiqn_for_login(tiqn);
+
+ return 0;
+}
+
+static struct iscsi_np *iscsit_get_np(
+ unsigned char *ipv6,
+ u32 ipv4,
+ u16 port,
+ int af_inet,
+ int network_transport)
+{
+ struct iscsi_np *np;
+ int ip_match = 0;
+
+ spin_lock_bh(&np_lock);
+ list_for_each_entry(np, &g_np_list, np_list) {
+ spin_lock(&np->np_thread_lock);
+ if (np->np_thread_state != ISCSI_NP_THREAD_ACTIVE) {
+ spin_unlock(&np->np_thread_lock);
+ continue;
+ }
+
+ if (af_inet == AF_INET6) {
+ if (!strcmp(&np->np_ip[0], &ipv6[0]))
+ ip_match = 1;
+ } else {
+ if (np->np_ipv4 == ipv4)
+ ip_match = 1;
+ }
+
+ if ((ip_match == 1) && (np->np_port == port) &&
+ (np->np_network_transport == network_transport)) {
+ /*
+ * Increment the np_exports reference count now to
+ * prevent iscsit_del_np() below from being called
+ * while iscsi_tpg_add_network_portal() is called.
+ */
+ np->np_exports++;
+ spin_unlock(&np->np_thread_lock);
+ spin_unlock_bh(&np_lock);
+ return np;
+ }
+ spin_unlock(&np->np_thread_lock);
+ }
+ spin_unlock_bh(&np_lock);
+
+ return NULL;
+}
+
+struct iscsi_np *iscsit_add_np(
+ struct iscsi_np_addr *np_addr,
+ int network_transport,
+ int af_inet)
+{
+ struct iscsi_np *np;
+ int ret;
+ /*
+ * Locate the existing struct iscsi_np if already active..
+ */
+ np = iscsit_get_np(np_addr->np_ipv6, np_addr->np_ipv4,
+ np_addr->np_port, af_inet, network_transport);
+ if (np)
+ return np;
+
+ np = kzalloc(sizeof(struct iscsi_np), GFP_KERNEL);
+ if (!np) {
+ printk(KERN_ERR "Unable to allocate memory for struct iscsi_np\n");
+ return ERR_PTR(-ENOMEM);
+ }
+
+ np->np_flags |= NPF_IP_NETWORK;
+ if (af_inet == AF_INET6) {
+ snprintf(np->np_ip, IPV6_ADDRESS_SPACE, "%s", np_addr->np_ipv6);
+ } else {
+ sprintf(np->np_ip, "%u.%u.%u.%u",
+ ((np_addr->np_ipv4 >> 24) & 0xff),
+ ((np_addr->np_ipv4 >> 16) & 0xff),
+ ((np_addr->np_ipv4 >> 8) & 0xff),
+ np_addr->np_ipv4 & 0xff);
+ np->np_ipv4 = np_addr->np_ipv4;
+ }
+
+ np->np_port = np_addr->np_port;
+ np->np_network_transport = network_transport;
+ spin_lock_init(&np->np_thread_lock);
+ init_completion(&np->np_restart_comp);
+ INIT_LIST_HEAD(&np->np_list);
+
+ ret = iscsi_target_setup_login_socket(np, af_inet);
+ if (ret != 0) {
+ kfree(np);
+ return ERR_PTR(ret);
+ }
+
+ np->np_thread = kthread_run(iscsi_target_login_thread, np, "iscsi_np");
+ if (IS_ERR(np->np_thread)) {
+ printk(KERN_ERR "Unable to create kthread: iscsi_np\n");
+ ret = PTR_ERR(np->np_thread);
+ kfree(np);
+ return ERR_PTR(ret);
+ }
+ /*
+ * Increment the np_exports reference count now to prevent
+ * iscsit_del_np() below from being run while a new call to
+ * iscsi_tpg_add_network_portal() for a matching iscsi_np is
+ * active. We don't need to hold np->np_thread_lock at this
+ * point because iscsi_np has not been added to g_np_list yet.
+ */
+ np->np_exports = 1;
+
+ spin_lock_bh(&np_lock);
+ list_add_tail(&np->np_list, &g_np_list);
+ spin_unlock_bh(&np_lock);
+
+ printk(KERN_INFO "CORE[0] - Added Network Portal: %s:%hu on %s on"
+ " network device: %s\n", np->np_ip, np->np_port,
+ (np->np_network_transport == ISCSI_TCP) ?
+ "TCP" : "SCTP", (strlen(np->np_net_dev)) ?
+ (char *)np->np_net_dev : "None");
+
+ return np;
+}
+
+int iscsit_reset_np_thread(
+ struct iscsi_np *np,
+ struct iscsi_tpg_np *tpg_np,
+ struct iscsi_portal_group *tpg)
+{
+ spin_lock_bh(&np->np_thread_lock);
+ if (tpg && tpg_np) {
+ /*
+ * The reset operation need only be performed when the
+ * passed struct iscsi_portal_group has a login in progress
+ * to one of the network portals.
+ */
+ if (tpg_np->tpg_np->np_login_tpg != tpg) {
+ spin_unlock_bh(&np->np_thread_lock);
+ return 0;
+ }
+ }
+ if (np->np_thread_state == ISCSI_NP_THREAD_INACTIVE) {
+ spin_unlock_bh(&np->np_thread_lock);
+ return 0;
+ }
+ np->np_thread_state = ISCSI_NP_THREAD_RESET;
+
+ if (np->np_thread) {
+ spin_unlock_bh(&np->np_thread_lock);
+ send_sig(SIGINT, np->np_thread, 1);
+ wait_for_completion(&np->np_restart_comp);
+ spin_lock_bh(&np->np_thread_lock);
+ }
+ spin_unlock_bh(&np->np_thread_lock);
+
+ return 0;
+}
+
+int iscsit_del_np_comm(struct iscsi_np *np)
+{
+ if (!np->np_socket)
+ return 0;
+
+ /*
+ * Some network transports allocate their own struct sock->file,
+ * see if we need to free any additional allocated resources.
+ */
+ if (np->np_flags & NPF_SCTP_STRUCT_FILE) {
+ kfree(np->np_socket->file);
+ np->np_socket->file = NULL;
+ }
+
+ sock_release(np->np_socket);
+ return 0;
+}
+
+int iscsit_del_np(struct iscsi_np *np)
+{
+ spin_lock_bh(&np->np_thread_lock);
+ if (!(--np->np_exports == 0)) {
+ spin_unlock_bh(&np->np_thread_lock);
+ return 0;
+ }
+ np->np_thread_state = ISCSI_NP_THREAD_SHUTDOWN;
+ spin_unlock_bh(&np->np_thread_lock);
+
+ if (np->np_thread) {
+ /*
+ * We need to send the signal to wakeup Linux/Net
+ * which may be sleeping in sock_accept()..
+ */
+ send_sig(SIGINT, np->np_thread, 1);
+ kthread_stop(np->np_thread);
+ }
+ iscsit_del_np_comm(np);
+
+ spin_lock_bh(&np_lock);
+ list_del(&np->np_list);
+ spin_unlock_bh(&np_lock);
+
+ printk(KERN_INFO "CORE[0] - Removed Network Portal: %s:%hu on %s on"
+ " network device: %s\n", np->np_ip, np->np_port,
+ (np->np_network_transport == ISCSI_TCP) ?
+ "TCP" : "SCTP", (strlen(np->np_net_dev)) ?
+ (char *)np->np_net_dev : "None");
+
+ kfree(np);
+ return 0;
+}
+
+static int __init iscsi_target_init_module(void)
+{
+ int ret = 0;
+
+ printk(KERN_INFO "RisingTide Systems iSCSI-Target "ISCSI_VERSION"\n");
+
+ iscsi_global = kzalloc(sizeof(struct iscsi_global), GFP_KERNEL);
+ if (!iscsi_global) {
+ printk(KERN_ERR "Unable to allocate memory for iscsi_global\n");
+ return -1;
+ }
+ mutex_init(&auth_id_lock);
+ spin_lock_init(&sess_idr_lock);
+ idr_init(&tiqn_idr);
+ idr_init(&sess_idr);
+
+ ret = iscsi_target_register_configfs();
+ if (ret < 0)
+ goto out;
+
+ ret = iscsi_thread_set_init();
+ if (ret < 0)
+ goto configfs_out;
+
+ if (iscsi_allocate_thread_sets(TARGET_THREAD_SET_COUNT) !=
+ TARGET_THREAD_SET_COUNT) {
+ printk(KERN_ERR "iscsi_allocate_thread_sets() returned"
+ " unexpected value!\n");
+ goto ts_out1;
+ }
+
+ lio_cmd_cache = kmem_cache_create("lio_cmd_cache",
+ sizeof(struct iscsi_cmd), __alignof__(struct iscsi_cmd),
+ 0, NULL);
+ if (!lio_cmd_cache) {
+ printk(KERN_ERR "Unable to kmem_cache_create() for"
+ " lio_cmd_cache\n");
+ goto ts_out2;
+ }
+
+ lio_qr_cache = kmem_cache_create("lio_qr_cache",
+ sizeof(struct iscsi_queue_req),
+ __alignof__(struct iscsi_queue_req), 0, NULL);
+ if (!lio_qr_cache) {
+ printk(KERN_ERR "nable to kmem_cache_create() for"
+ " lio_qr_cache\n");
+ goto cmd_out;
+ }
+
+ lio_dr_cache = kmem_cache_create("lio_dr_cache",
+ sizeof(struct iscsi_datain_req),
+ __alignof__(struct iscsi_datain_req), 0, NULL);
+ if (!lio_dr_cache) {
+ printk(KERN_ERR "Unable to kmem_cache_create() for"
+ " lio_dr_cache\n");
+ goto qr_out;
+ }
+
+ lio_ooo_cache = kmem_cache_create("lio_ooo_cache",
+ sizeof(struct iscsi_ooo_cmdsn),
+ __alignof__(struct iscsi_ooo_cmdsn), 0, NULL);
+ if (!lio_ooo_cache) {
+ printk(KERN_ERR "Unable to kmem_cache_create() for"
+ " lio_ooo_cache\n");
+ goto dr_out;
+ }
+
+ lio_r2t_cache = kmem_cache_create("lio_r2t_cache",
+ sizeof(struct iscsi_r2t), __alignof__(struct iscsi_r2t),
+ 0, NULL);
+ if (!lio_r2t_cache) {
+ printk(KERN_ERR "Unable to kmem_cache_create() for"
+ " lio_r2t_cache\n");
+ goto ooo_out;
+ }
+
+ if (iscsit_load_discovery_tpg() < 0)
+ goto r2t_out;
+
+ printk("Loading Complete.\n");
+
+ return ret;
+r2t_out:
+ kmem_cache_destroy(lio_r2t_cache);
+ooo_out:
+ kmem_cache_destroy(lio_ooo_cache);
+dr_out:
+ kmem_cache_destroy(lio_dr_cache);
+qr_out:
+ kmem_cache_destroy(lio_qr_cache);
+cmd_out:
+ kmem_cache_destroy(lio_cmd_cache);
+ts_out2:
+ iscsi_deallocate_thread_sets();
+ts_out1:
+ iscsi_thread_set_free();
+configfs_out:
+ iscsi_target_deregister_configfs();
+out:
+ kfree(iscsi_global);
+ iscsi_global = NULL;
+
+ return -1;
+}
+
+static void __exit iscsi_target_cleanup_module(void)
+{
+ iscsi_deallocate_thread_sets();
+ iscsi_thread_set_free();
+ iscsit_release_discovery_tpg();
+ kmem_cache_destroy(lio_cmd_cache);
+ kmem_cache_destroy(lio_qr_cache);
+ kmem_cache_destroy(lio_dr_cache);
+ kmem_cache_destroy(lio_ooo_cache);
+ kmem_cache_destroy(lio_r2t_cache);
+
+ iscsi_target_deregister_configfs();
+
+ kfree(iscsi_global);
+ printk(KERN_INFO "Unloading Complete.\n");
+}
+
+int iscsit_add_reject(
+ u8 reason,
+ int fail_conn,
+ unsigned char *buf,
+ struct iscsi_conn *conn)
+{
+ struct iscsi_cmd *cmd;
+ struct iscsi_reject *hdr;
+ int ret;
+
+ cmd = iscsit_allocate_cmd(conn);
+ if (!cmd)
+ return -1;
+
+ cmd->iscsi_opcode = ISCSI_OP_REJECT;
+ if (fail_conn)
+ cmd->cmd_flags |= ICF_REJECT_FAIL_CONN;
+
+ hdr = (struct iscsi_reject *) cmd->pdu;
+ hdr->reason = reason;
+
+ cmd->buf_ptr = kzalloc(ISCSI_HDR_LEN, GFP_ATOMIC);
+ if (!cmd->buf_ptr) {
+ printk(KERN_ERR "Unable to allocate memory for cmd->buf_ptr\n");
+ iscsit_release_cmd(cmd);
+ return -1;
+ }
+ memcpy(cmd->buf_ptr, buf, ISCSI_HDR_LEN);
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ cmd->i_state = ISTATE_SEND_REJECT;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+
+ ret = wait_for_completion_interruptible(&cmd->reject_comp);
+ if (ret != 0)
+ return -1;
+
+ return (!fail_conn) ? 0 : -1;
+}
+
+int iscsit_add_reject_from_cmd(
+ u8 reason,
+ int fail_conn,
+ int add_to_conn,
+ unsigned char *buf,
+ struct iscsi_cmd *cmd)
+{
+ struct iscsi_conn *conn;
+ struct iscsi_reject *hdr;
+ int ret;
+
+ if (!cmd->conn) {
+ printk(KERN_ERR "cmd->conn is NULL for ITT: 0x%08x\n",
+ cmd->init_task_tag);
+ return -1;
+ }
+ conn = cmd->conn;
+
+ cmd->iscsi_opcode = ISCSI_OP_REJECT;
+ if (fail_conn)
+ cmd->cmd_flags |= ICF_REJECT_FAIL_CONN;
+
+ hdr = (struct iscsi_reject *) cmd->pdu;
+ hdr->reason = reason;
+
+ cmd->buf_ptr = kzalloc(ISCSI_HDR_LEN, GFP_ATOMIC);
+ if (!cmd->buf_ptr) {
+ printk(KERN_ERR "Unable to allocate memory for cmd->buf_ptr\n");
+ iscsit_release_cmd(cmd);
+ return -1;
+ }
+ memcpy(cmd->buf_ptr, buf, ISCSI_HDR_LEN);
+
+ if (add_to_conn) {
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+ }
+
+ cmd->i_state = ISTATE_SEND_REJECT;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+
+ ret = wait_for_completion_interruptible(&cmd->reject_comp);
+ if (ret != 0)
+ return -1;
+
+ return (!fail_conn) ? 0 : -1;
+}
+
+static inline void iscsit_calculate_map_segment(
+ u32 *data_length,
+ struct se_offset_map *lm)
+{
+ u32 sg_offset = 0;
+ struct se_mem *se_mem = lm->map_se_mem;
+ /*
+ * Still working on pages in the current struct se_mem.
+ */
+ if (!lm->map_reset) {
+ lm->iovec_length = (lm->sg_length > PAGE_SIZE) ?
+ PAGE_SIZE : lm->sg_length;
+ if (*data_length < lm->iovec_length)
+ lm->iovec_length = *data_length;
+
+ lm->iovec_base = page_address(lm->sg_page) + sg_offset;
+ return;
+ }
+
+ /*
+ * First run of an iscsi_linux_map_t.
+ *
+ * OR:
+ *
+ * Mapped all of the pages in the current scatterlist, move
+ * on to the next one.
+ */
+ lm->map_reset = 0;
+ sg_offset = se_mem->se_off;
+ lm->sg_page = se_mem->se_page;
+ lm->sg_length = se_mem->se_len;
+ /*
+ * Get the base and length of the current page for use with the iovec.
+ */
+recalc:
+ lm->iovec_length = (lm->sg_length > (PAGE_SIZE - sg_offset)) ?
+ (PAGE_SIZE - sg_offset) : lm->sg_length;
+ /*
+ * See if there is any iSCSI offset we need to deal with.
+ */
+ if (!lm->current_offset) {
+ lm->iovec_base = page_address(lm->sg_page) + sg_offset;
+
+ if (*data_length < lm->iovec_length)
+ lm->iovec_length = *data_length;
+
+ return;
+ }
+
+ /*
+ * We know the iSCSI offset is in the next page of the current
+ * scatterlist. Increase the lm->sg_page pointer and try again.
+ */
+ if (lm->current_offset >= lm->iovec_length) {
+ lm->current_offset -= lm->iovec_length;
+ lm->sg_length -= lm->iovec_length;
+ lm->sg_page++;
+ sg_offset = 0;
+
+ goto recalc;
+ }
+
+ /*
+ * The iSCSI offset is in the current page, increment the iovec
+ * base and reduce iovec length.
+ */
+ lm->iovec_base = page_address(lm->sg_page);
+
+ lm->iovec_base += sg_offset;
+ lm->iovec_base += lm->current_offset;
+
+ if ((lm->iovec_length - lm->current_offset) < *data_length)
+ lm->iovec_length -= lm->current_offset;
+ else
+ lm->iovec_length = *data_length;
+
+ if ((lm->sg_length - lm->current_offset) < *data_length)
+ lm->sg_length -= lm->current_offset;
+ else
+ lm->sg_length = *data_length;
+
+ lm->current_offset = 0;
+}
+
+static int iscsit_get_offset(
+ struct se_offset_map *lmap,
+ struct se_unmap_sg *usg)
+{
+ u32 current_length = 0, current_iscsi_offset = lmap->iscsi_offset;
+ u32 total_offset = 0;
+ struct se_cmd *cmd = usg->se_cmd;
+ struct se_mem *se_mem;
+
+ list_for_each_entry(se_mem, T_TASK(cmd)->t_mem_list, se_list)
+ break;
+
+ if (!se_mem) {
+ printk(KERN_ERR "Unable to locate se_mem from"
+ " T_TASK(cmd)->t_mem_list\n");
+ return -1;
+ }
+
+ /*
+ * Locate the current offset from the passed iSCSI Offset.
+ */
+ while (lmap->iscsi_offset != current_length) {
+ /*
+ * The iSCSI Offset is within the current struct se_mem.
+ *
+ * Or:
+ *
+ * The iSCSI Offset is outside of the current struct se_mem.
+ * Recalculate the values and obtain the next struct se_mem pointer.
+ */
+ total_offset += se_mem->se_len;
+
+ if (total_offset > lmap->iscsi_offset) {
+ current_length += current_iscsi_offset;
+ lmap->orig_offset = lmap->current_offset =
+ usg->t_offset = current_iscsi_offset;
+ } else {
+ current_length += se_mem->se_len;
+ current_iscsi_offset -= se_mem->se_len;
+
+ list_for_each_entry_continue(se_mem,
+ T_TASK(cmd)->t_mem_list, se_list)
+ break;
+
+ if (!se_mem) {
+ printk(KERN_ERR "Unable to locate struct se_mem\n");
+ return -1;
+ }
+ }
+ }
+ lmap->map_orig_se_mem = se_mem;
+ usg->cur_se_mem = se_mem;
+
+ return 0;
+}
+
+static int iscsit_set_iovec_ptrs(
+ struct se_map_sg *map_sg,
+ struct se_unmap_sg *unmap_sg)
+{
+ u32 i = 0 /* For iovecs */, j = 0 /* For scatterlists */;
+ struct se_cmd *cmd = map_sg->se_cmd;
+ struct iscsi_cmd *i_cmd = container_of(cmd, struct iscsi_cmd, se_cmd);
+ struct se_offset_map *lmap = &unmap_sg->lmap;
+ struct kvec *iov = map_sg->iov;
+
+ /*
+ * Used for non scatterlist operations, assume a single iovec.
+ */
+ if (!T_TASK(cmd)->t_tasks_se_num) {
+ iov[0].iov_base = (unsigned char *) T_TASK(cmd)->t_task_buf +
+ map_sg->data_offset;
+ iov[0].iov_len = map_sg->data_length;
+ return 1;
+ }
+
+ /*
+ * Set lmap->map_reset = 1 so the first call to
+ * iscsit_calculate_map_segment() sets up the initial
+ * values for struct se_offset_map.
+ */
+ lmap->map_reset = 1;
+ /*
+ * Get a pointer to the first used scatterlist based on the passed
+ * offset. Also set the rest of the needed values in iscsi_linux_map_t.
+ */
+ lmap->iscsi_offset = map_sg->data_offset;
+ if (map_sg->sg_kmap_active) {
+ unmap_sg->se_cmd = map_sg->se_cmd;
+ iscsit_get_offset(lmap, unmap_sg);
+ unmap_sg->data_length = map_sg->data_length;
+ } else {
+ lmap->current_offset = lmap->orig_offset;
+ }
+ lmap->map_se_mem = lmap->map_orig_se_mem;
+
+ while (map_sg->data_length) {
+ /*
+ * Time to get the virtual address for use with iovec pointers.
+ * This function will return the expected iovec_base address
+ * and iovec_length.
+ */
+ iscsit_calculate_map_segment(&map_sg->data_length, lmap);
+
+ /*
+ * Set the iov.iov_base and iov.iov_len from the current values
+ * in iscsi_linux_map_t.
+ */
+ iov[i].iov_base = lmap->iovec_base;
+ iov[i].iov_len = lmap->iovec_length;
+
+ /*
+ * Subtract the final iovec length from the total length to be
+ * mapped, and the length of the current scatterlist. Also
+ * perform the paranoid check to make sure we are not going to
+ * overflow the iovecs allocated for this command in the next
+ * pass.
+ */
+ map_sg->data_length -= iov[i].iov_len;
+ lmap->sg_length -= iov[i].iov_len;
+
+ if ((++i + 1) > i_cmd->orig_iov_data_count) {
+ printk(KERN_ERR "Current iovec count %u is greater than"
+ " struct se_cmd->orig_data_iov_count %u, cannot"
+ " continue.\n", i+1, i_cmd->orig_iov_data_count);
+ return -1;
+ }
+
+ /*
+ * All done mapping this scatterlist's pages, move on to
+ * the next scatterlist by setting lmap.map_reset = 1;
+ */
+ if (!lmap->sg_length || !map_sg->data_length) {
+ list_for_each_entry(lmap->map_se_mem,
+ &lmap->map_se_mem->se_list, se_list)
+ break;
+
+ if (!lmap->map_se_mem) {
+ printk(KERN_ERR "Unable to locate next"
+ " lmap->map_struct se_mem entry\n");
+ return -1;
+ }
+ j++;
+
+ lmap->sg_page = NULL;
+ lmap->map_reset = 1;
+ } else
+ lmap->sg_page++;
+ }
+
+ unmap_sg->sg_count = j;
+
+ return i;
+}
+
+static void iscsit_map_SG_segments(struct se_unmap_sg *unmap_sg)
+{
+ u32 i = 0;
+ struct se_cmd *cmd = unmap_sg->se_cmd;
+ struct se_mem *se_mem = unmap_sg->cur_se_mem;
+
+ if (!T_TASK(cmd)->t_tasks_se_num)
+ return;
+
+ list_for_each_entry_continue(se_mem, T_TASK(cmd)->t_mem_list, se_list) {
+ kmap(se_mem->se_page);
+
+ if (++i == unmap_sg->sg_count)
+ break;
+ }
+}
+
+static void iscsit_unmap_SG_segments(struct se_unmap_sg *unmap_sg)
+{
+ u32 i = 0;
+ struct se_cmd *cmd = unmap_sg->se_cmd;
+ struct se_mem *se_mem = unmap_sg->cur_se_mem;
+
+ if (!T_TASK(cmd)->t_tasks_se_num)
+ return;
+
+ list_for_each_entry_continue(se_mem, T_TASK(cmd)->t_mem_list, se_list) {
+ kunmap(se_mem->se_page);
+
+ if (++i == unmap_sg->sg_count)
+ break;
+ }
+}
+
+static inline void iscsit_ack_from_expstatsn(struct iscsi_conn *conn, u32 exp_statsn)
+{
+ struct iscsi_cmd *cmd;
+
+ conn->exp_statsn = exp_statsn;
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry(cmd, &conn->conn_cmd_list, i_list) {
+ spin_lock(&cmd->istate_lock);
+ if ((cmd->i_state == ISTATE_SENT_STATUS) &&
+ (cmd->stat_sn < exp_statsn)) {
+ cmd->i_state = ISTATE_REMOVE;
+ spin_unlock(&cmd->istate_lock);
+ iscsit_add_cmd_to_immediate_queue(cmd, conn,
+ cmd->i_state);
+ continue;
+ }
+ spin_unlock(&cmd->istate_lock);
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+}
+
+static inline int iscsit_handle_scsi_cmd(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ int data_direction, cmdsn_ret = 0, immed_ret, ret, transport_ret;
+ int dump_immediate_data = 0, send_check_condition = 0, payload_length;
+ struct iscsi_cmd *cmd = NULL;
+ struct iscsi_scsi_req *hdr;
+
+ spin_lock_bh(&conn->sess->session_stats_lock);
+ conn->sess->cmd_pdus++;
+ if (conn->sess->se_sess->se_node_acl) {
+ spin_lock(&conn->sess->se_sess->se_node_acl->stats_lock);
+ conn->sess->se_sess->se_node_acl->num_cmds++;
+ spin_unlock(&conn->sess->se_sess->se_node_acl->stats_lock);
+ }
+ spin_unlock_bh(&conn->sess->session_stats_lock);
+
+ hdr = (struct iscsi_scsi_req *) buf;
+ payload_length = ntoh24(hdr->dlength);
+ hdr->itt = be32_to_cpu(hdr->itt);
+ hdr->data_length = be32_to_cpu(hdr->data_length);
+ hdr->cmdsn = be32_to_cpu(hdr->cmdsn);
+ hdr->exp_statsn = be32_to_cpu(hdr->exp_statsn);
+
+ /* FIXME; Add checks for AdditionalHeaderSegment */
+
+ if (!(hdr->flags & ISCSI_FLAG_CMD_WRITE) &&
+ !(hdr->flags & ISCSI_FLAG_CMD_FINAL)) {
+ printk(KERN_ERR "ISCSI_FLAG_CMD_WRITE & ISCSI_FLAG_CMD_FINAL"
+ " not set. Bad iSCSI Initiator.\n");
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_INVALID, 1,
+ buf, conn);
+ }
+
+ if (((hdr->flags & ISCSI_FLAG_CMD_READ) ||
+ (hdr->flags & ISCSI_FLAG_CMD_WRITE)) && !hdr->data_length) {
+ /*
+ * Vmware ESX v3.0 uses a modified Cisco Initiator (v3.4.2)
+ * that adds support for RESERVE/RELEASE. There is a bug
+ * add with this new functionality that sets R/W bits when
+ * neither CDB carries any READ or WRITE datapayloads.
+ */
+ if ((hdr->cdb[0] == 0x16) || (hdr->cdb[0] == 0x17)) {
+ hdr->flags &= ~ISCSI_FLAG_CMD_READ;
+ hdr->flags &= ~ISCSI_FLAG_CMD_WRITE;
+ goto done;
+ }
+
+ printk(KERN_ERR "ISCSI_FLAG_CMD_READ or ISCSI_FLAG_CMD_WRITE"
+ " set when Expected Data Transfer Length is 0 for"
+ " CDB: 0x%02x. Bad iSCSI Initiator.\n", hdr->cdb[0]);
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_INVALID, 1,
+ buf, conn);
+ }
+done:
+
+ if (!(hdr->flags & ISCSI_FLAG_CMD_READ) &&
+ !(hdr->flags & ISCSI_FLAG_CMD_WRITE) && (hdr->data_length != 0)) {
+ printk(KERN_ERR "ISCSI_FLAG_CMD_READ and/or ISCSI_FLAG_CMD_WRITE"
+ " MUST be set if Expected Data Transfer Length is not 0."
+ " Bad iSCSI Initiator\n");
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_INVALID, 1,
+ buf, conn);
+ }
+
+ if ((hdr->flags & ISCSI_FLAG_CMD_READ) &&
+ (hdr->flags & ISCSI_FLAG_CMD_WRITE)) {
+ printk(KERN_ERR "Bidirectional operations not supported!\n");
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_INVALID, 1,
+ buf, conn);
+ }
+
+ if (hdr->opcode & ISCSI_OP_IMMEDIATE) {
+ printk(KERN_ERR "Illegally set Immediate Bit in iSCSI Initiator"
+ " Scsi Command PDU.\n");
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_INVALID, 1,
+ buf, conn);
+ }
+
+ if (payload_length && !conn->sess->sess_ops->ImmediateData) {
+ printk(KERN_ERR "ImmediateData=No but DataSegmentLength=%u,"
+ " protocol error.\n", payload_length);
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+#if 0
+ if (!(hdr->flags & ISCSI_FLAG_CMD_FINAL) &&
+ (hdr->flags & ISCSI_FLAG_CMD_WRITE) && conn->sess->sess_ops->InitialR2T) {
+ printk(KERN_ERR "ISCSI_FLAG_CMD_FINAL is not Set and"
+ " ISCSI_FLAG_CMD_WRITE Bit and InitialR2T=Yes,"
+ " protocol error\n");
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+#endif
+ if ((hdr->data_length == payload_length) &&
+ (!(hdr->flags & ISCSI_FLAG_CMD_FINAL))) {
+ printk(KERN_ERR "Expected Data Transfer Length and Length of"
+ " Immediate Data are the same, but ISCSI_FLAG_CMD_FINAL"
+ " bit is not set protocol error\n");
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ if (payload_length > hdr->data_length) {
+ printk(KERN_ERR "DataSegmentLength: %u is greater than"
+ " EDTL: %u, protocol error.\n", payload_length,
+ hdr->data_length);
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ if (payload_length > conn->conn_ops->MaxRecvDataSegmentLength) {
+ printk(KERN_ERR "DataSegmentLength: %u is greater than"
+ " MaxRecvDataSegmentLength: %u, protocol error.\n",
+ payload_length, conn->conn_ops->MaxRecvDataSegmentLength);
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ if (payload_length > conn->sess->sess_ops->FirstBurstLength) {
+ printk(KERN_ERR "DataSegmentLength: %u is greater than"
+ " FirstBurstLength: %u, protocol error.\n",
+ payload_length, conn->sess->sess_ops->FirstBurstLength);
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_INVALID, 1,
+ buf, conn);
+ }
+
+ data_direction = (hdr->flags & ISCSI_FLAG_CMD_WRITE) ? DMA_TO_DEVICE :
+ (hdr->flags & ISCSI_FLAG_CMD_READ) ? DMA_FROM_DEVICE :
+ DMA_NONE;
+
+ cmd = iscsit_allocate_se_cmd(conn, hdr->data_length, data_direction,
+ (hdr->flags & ISCSI_FLAG_CMD_ATTR_MASK));
+ if (!cmd)
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_NO_RESOURCES, 1,
+ buf, conn);
+
+ TRACE(TRACE_ISCSI, "Got SCSI Command, ITT: 0x%08x, CmdSN: 0x%08x,"
+ " ExpXferLen: %u, Length: %u, CID: %hu\n", hdr->itt,
+ hdr->cmdsn, hdr->data_length, payload_length, conn->cid);
+
+ cmd->iscsi_opcode = ISCSI_OP_SCSI_CMD;
+ cmd->i_state = ISTATE_NEW_CMD;
+ cmd->immediate_cmd = ((hdr->opcode & ISCSI_OP_IMMEDIATE) ? 1 : 0);
+ cmd->immediate_data = (payload_length) ? 1 : 0;
+ cmd->unsolicited_data = ((!(hdr->flags & ISCSI_FLAG_CMD_FINAL) &&
+ (hdr->flags & ISCSI_FLAG_CMD_WRITE)) ? 1 : 0);
+ if (cmd->unsolicited_data)
+ cmd->cmd_flags |= ICF_NON_IMMEDIATE_UNSOLICITED_DATA;
+
+ conn->sess->init_task_tag = cmd->init_task_tag = hdr->itt;
+ if (hdr->flags & ISCSI_FLAG_CMD_READ) {
+ spin_lock_bh(&conn->sess->ttt_lock);
+ cmd->targ_xfer_tag = conn->sess->targ_xfer_tag++;
+ if (cmd->targ_xfer_tag == 0xFFFFFFFF)
+ cmd->targ_xfer_tag = conn->sess->targ_xfer_tag++;
+ spin_unlock_bh(&conn->sess->ttt_lock);
+ } else if (hdr->flags & ISCSI_FLAG_CMD_WRITE)
+ cmd->targ_xfer_tag = 0xFFFFFFFF;
+ cmd->cmd_sn = hdr->cmdsn;
+ cmd->exp_stat_sn = hdr->exp_statsn;
+ cmd->first_burst_len = payload_length;
+
+ if (cmd->data_direction == DMA_FROM_DEVICE) {
+ struct iscsi_datain_req *dr;
+
+ dr = iscsit_allocate_datain_req();
+ if (!dr)
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_BOOKMARK_NO_RESOURCES,
+ 1, 1, buf, cmd);
+
+ iscsit_attach_datain_req(cmd, dr);
+ }
+
+ /*
+ * The CDB is going to an se_device_t.
+ */
+ ret = iscsit_get_lun_for_cmd(cmd, hdr->cdb,
+ get_unaligned_le64(&hdr->lun[0]));
+ if (ret < 0) {
+ if (SE_CMD(cmd)->scsi_sense_reason == TCM_NON_EXISTENT_LUN) {
+ TRACE(TRACE_VANITY, "Responding to non-acl'ed,"
+ " non-existent or non-exported iSCSI LUN:"
+ " 0x%016Lx\n", get_unaligned_le64(&hdr->lun[0]));
+ }
+ if (ret == PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES)
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_BOOKMARK_NO_RESOURCES,
+ 1, 1, buf, cmd);
+
+ send_check_condition = 1;
+ goto attach_cmd;
+ }
+ /*
+ * The Initiator Node has access to the LUN (the addressing method
+ * is handled inside of iscsit_get_lun_for_cmd()). Now it's time to
+ * allocate 1->N transport tasks (depending on sector count and
+ * maximum request size the physical HBA(s) can handle.
+ */
+ transport_ret = transport_generic_allocate_tasks(SE_CMD(cmd), hdr->cdb);
+ if (!transport_ret)
+ goto build_list;
+
+ if (transport_ret == -1) {
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_BOOKMARK_NO_RESOURCES,
+ 1, 1, buf, cmd);
+ } else if (transport_ret == -2) {
+ /*
+ * Unsupported SAM Opcode. CHECK_CONDITION will be sent
+ * in iscsit_execute_cmd() during the CmdSN OOO Execution
+ * Mechinism.
+ */
+ send_check_condition = 1;
+ goto attach_cmd;
+ }
+
+build_list:
+ if (iscsit_decide_list_to_build(cmd, payload_length) < 0)
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_BOOKMARK_NO_RESOURCES,
+ 1, 1, buf, cmd);
+attach_cmd:
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+ /*
+ * Check if we need to delay processing because of ALUA
+ * Active/NonOptimized primary access state..
+ */
+ core_alua_check_nonop_delay(SE_CMD(cmd));
+ /*
+ * Check the CmdSN against ExpCmdSN/MaxCmdSN here if
+ * the Immediate Bit is not set, and no Immediate
+ * Data is attached.
+ *
+ * A PDU/CmdSN carrying Immediate Data can only
+ * be processed after the DataCRC has passed.
+ * If the DataCRC fails, the CmdSN MUST NOT
+ * be acknowledged. (See below)
+ */
+ if (!cmd->immediate_data) {
+ cmdsn_ret = iscsit_check_received_cmdsn(conn,
+ cmd, hdr->cmdsn);
+ if ((cmdsn_ret == CMDSN_NORMAL_OPERATION) ||
+ (cmdsn_ret == CMDSN_HIGHER_THAN_EXP))
+ do {} while (0);
+ else if (cmdsn_ret == CMDSN_LOWER_THAN_EXP) {
+ cmd->i_state = ISTATE_REMOVE;
+ iscsit_add_cmd_to_immediate_queue(cmd,
+ conn, cmd->i_state);
+ return 0;
+ } else { /* (cmdsn_ret == CMDSN_ERROR_CANNOT_RECOVER) */
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ }
+ }
+ iscsit_ack_from_expstatsn(conn, hdr->exp_statsn);
+
+ /*
+ * If no Immediate Data is attached, it's OK to return now.
+ */
+ if (!cmd->immediate_data) {
+ if (send_check_condition)
+ return 0;
+
+ if (cmd->unsolicited_data) {
+ iscsit_set_dataout_sequence_values(cmd);
+
+ spin_lock_bh(&cmd->dataout_timeout_lock);
+ iscsit_start_dataout_timer(cmd, cmd->conn);
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ }
+
+ return 0;
+ }
+
+ /*
+ * Early CHECK_CONDITIONs never make it to the transport processing
+ * thread. They are processed in CmdSN order by
+ * iscsit_check_received_cmdsn() below.
+ */
+ if (send_check_condition) {
+ immed_ret = IMMEDIDATE_DATA_NORMAL_OPERATION;
+ dump_immediate_data = 1;
+ goto after_immediate_data;
+ }
+
+ /*
+ * Immediate Data is present, send to the transport and block until
+ * the underlying transport plugin has allocated the buffer to
+ * receive the Immediate Write Data into.
+ */
+ transport_generic_handle_cdb(SE_CMD(cmd));
+
+ wait_for_completion(&cmd->unsolicited_data_comp);
+
+ if (SE_CMD(cmd)->se_cmd_flags & SCF_SE_CMD_FAILED) {
+ immed_ret = IMMEDIDATE_DATA_NORMAL_OPERATION;
+ dump_immediate_data = 1;
+ goto after_immediate_data;
+ }
+
+ immed_ret = iscsit_handle_immediate_data(cmd, buf, payload_length);
+after_immediate_data:
+ if (immed_ret == IMMEDIDATE_DATA_NORMAL_OPERATION) {
+ /*
+ * A PDU/CmdSN carrying Immediate Data passed
+ * DataCRC, check against ExpCmdSN/MaxCmdSN if
+ * Immediate Bit is not set.
+ */
+ cmdsn_ret = iscsit_check_received_cmdsn(conn,
+ cmd, hdr->cmdsn);
+ /*
+ * Special case for Unsupported SAM WRITE Opcodes
+ * and ImmediateData=Yes.
+ */
+ if (dump_immediate_data) {
+ if (iscsit_dump_data_payload(conn, payload_length, 1) < 0)
+ return -1;
+ } else if (cmd->unsolicited_data) {
+ iscsit_set_dataout_sequence_values(cmd);
+
+ spin_lock_bh(&cmd->dataout_timeout_lock);
+ iscsit_start_dataout_timer(cmd, cmd->conn);
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ }
+
+ if (cmdsn_ret == CMDSN_NORMAL_OPERATION)
+ return 0;
+ else if (cmdsn_ret == CMDSN_HIGHER_THAN_EXP)
+ return 0;
+ else if (cmdsn_ret == CMDSN_LOWER_THAN_EXP) {
+ cmd->i_state = ISTATE_REMOVE;
+ iscsit_add_cmd_to_immediate_queue(cmd,
+ conn, cmd->i_state);
+ return 0;
+ } else { /* (cmdsn_ret == CMDSN_ERROR_CANNOT_RECOVER) */
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ }
+ } else if (immed_ret == IMMEDIDATE_DATA_ERL1_CRC_FAILURE) {
+ /*
+ * Immediate Data failed DataCRC and ERL>=1,
+ * silently drop this PDU and let the initiator
+ * plug the CmdSN gap.
+ *
+ * FIXME: Send Unsolicited NOPIN with reserved
+ * TTT here to help the initiator figure out
+ * the missing CmdSN, although they should be
+ * intelligent enough to determine the missing
+ * CmdSN and issue a retry to plug the sequence.
+ */
+ cmd->i_state = ISTATE_REMOVE;
+ iscsit_add_cmd_to_immediate_queue(cmd, conn, cmd->i_state);
+ } else /* immed_ret == IMMEDIDATE_DATA_CANNOT_RECOVER */
+ return -1;
+
+ return 0;
+}
+
+static inline int iscsit_handle_data_out(struct iscsi_conn *conn, unsigned char *buf)
+{
+ int iov_ret, ooo_cmdsn = 0, ret;
+ u8 data_crc_failed = 0, *pad_bytes[4];
+ u32 checksum, iov_count = 0, padding = 0, rx_got = 0;
+ u32 rx_size = 0, payload_length;
+ struct iscsi_cmd *cmd = NULL;
+ struct se_cmd *se_cmd;
+ struct se_map_sg map_sg;
+ struct se_unmap_sg unmap_sg;
+ struct iscsi_data *hdr;
+ struct kvec *iov;
+ unsigned long flags;
+
+ hdr = (struct iscsi_data *) buf;
+ payload_length = ntoh24(hdr->dlength);
+ hdr->itt = be32_to_cpu(hdr->itt);
+ hdr->ttt = be32_to_cpu(hdr->ttt);
+ hdr->exp_statsn = be32_to_cpu(hdr->exp_statsn);
+ hdr->datasn = be32_to_cpu(hdr->datasn);
+ hdr->offset = be32_to_cpu(hdr->offset);
+
+ if (!payload_length) {
+ printk(KERN_ERR "DataOUT payload is ZERO, protocol error.\n");
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ /* iSCSI write */
+ spin_lock_bh(&conn->sess->session_stats_lock);
+ conn->sess->rx_data_octets += payload_length;
+ if (conn->sess->se_sess->se_node_acl) {
+ spin_lock(&conn->sess->se_sess->se_node_acl->stats_lock);
+ conn->sess->se_sess->se_node_acl->write_bytes += payload_length;
+ spin_unlock(&conn->sess->se_sess->se_node_acl->stats_lock);
+ }
+ spin_unlock_bh(&conn->sess->session_stats_lock);
+
+ if (payload_length > conn->conn_ops->MaxRecvDataSegmentLength) {
+ printk(KERN_ERR "DataSegmentLength: %u is greater than"
+ " MaxRecvDataSegmentLength: %u\n", payload_length,
+ conn->conn_ops->MaxRecvDataSegmentLength);
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ cmd = iscsit_find_cmd_from_itt_or_dump(conn, hdr->itt,
+ payload_length);
+ if (!cmd)
+ return 0;
+
+ TRACE(TRACE_ISCSI, "Got DataOut ITT: 0x%08x, TTT: 0x%08x,"
+ " DataSN: 0x%08x, Offset: %u, Length: %u, CID: %hu\n",
+ hdr->itt, hdr->ttt, hdr->datasn, hdr->offset,
+ payload_length, conn->cid);
+
+ if (cmd->cmd_flags & ICF_GOT_LAST_DATAOUT) {
+ printk(KERN_ERR "Command ITT: 0x%08x received DataOUT after"
+ " last DataOUT received, dumping payload\n",
+ cmd->init_task_tag);
+ return iscsit_dump_data_payload(conn, payload_length, 1);
+ }
+
+ if (cmd->data_direction != DMA_TO_DEVICE) {
+ printk(KERN_ERR "Command ITT: 0x%08x received DataOUT for a"
+ " NON-WRITE command.\n", cmd->init_task_tag);
+ return iscsit_add_reject_from_cmd(ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ }
+ se_cmd = SE_CMD(cmd);
+ iscsit_mod_dataout_timer(cmd);
+
+ if ((hdr->offset + payload_length) > cmd->data_length) {
+ printk(KERN_ERR "DataOut Offset: %u, Length %u greater than"
+ " iSCSI Command EDTL %u, protocol error.\n",
+ hdr->offset, payload_length, cmd->data_length);
+ return iscsit_add_reject_from_cmd(ISCSI_REASON_BOOKMARK_INVALID,
+ 1, 0, buf, cmd);
+ }
+
+ if (cmd->unsolicited_data) {
+ int dump_unsolicited_data = 0, wait_for_transport = 0;
+
+ if (conn->sess->sess_ops->InitialR2T) {
+ printk(KERN_ERR "Received unexpected unsolicited data"
+ " while InitialR2T=Yes, protocol error.\n");
+ transport_send_check_condition_and_sense(SE_CMD(cmd),
+ TCM_UNEXPECTED_UNSOLICITED_DATA, 0);
+ return -1;
+ }
+ /*
+ * Special case for dealing with Unsolicited DataOUT
+ * and Unsupported SAM WRITE Opcodes and SE resource allocation
+ * failures;
+ */
+ spin_lock_irqsave(&T_TASK(se_cmd)->t_state_lock, flags);
+ /*
+ * Handle cases where we do or do not want to sleep on
+ * unsolicited_data_comp
+ *
+ * First, if TRANSPORT_WRITE_PENDING state has not been reached,
+ * we need assume we need to wait and sleep..
+ */
+ wait_for_transport =
+ (se_cmd->t_state != TRANSPORT_WRITE_PENDING);
+ /*
+ * For the ImmediateData=Yes cases, there will already be
+ * generic target memory allocated with the original
+ * ISCSI_OP_SCSI_CMD PDU, so do not sleep for that case.
+ *
+ * The last is a check for a delayed TASK_ABORTED status that
+ * means the data payload will be dropped because
+ * SCF_SE_CMD_FAILED has been set to indicate that an exception
+ * condition for this struct sse_cmd has occured in generic target
+ * code that requires us to drop payload.
+ */
+ wait_for_transport =
+ (se_cmd->t_state != TRANSPORT_WRITE_PENDING);
+ if ((cmd->immediate_data != 0) ||
+ (atomic_read(&T_TASK(se_cmd)->t_transport_aborted) != 0))
+ wait_for_transport = 0;
+ spin_unlock_irqrestore(&T_TASK(se_cmd)->t_state_lock, flags);
+
+ if (wait_for_transport)
+ wait_for_completion(&cmd->unsolicited_data_comp);
+
+ spin_lock_irqsave(&T_TASK(se_cmd)->t_state_lock, flags);
+ if (!(se_cmd->se_cmd_flags & SCF_SUPPORTED_SAM_OPCODE) ||
+ (se_cmd->se_cmd_flags & SCF_SE_CMD_FAILED))
+ dump_unsolicited_data = 1;
+ spin_unlock_irqrestore(&T_TASK(se_cmd)->t_state_lock, flags);
+
+ if (dump_unsolicited_data) {
+ /*
+ * Check if a delayed TASK_ABORTED status needs to
+ * be sent now if the ISCSI_FLAG_CMD_FINAL has been
+ * received with the unsolicitied data out.
+ */
+ if (hdr->flags & ISCSI_FLAG_CMD_FINAL)
+ iscsit_stop_dataout_timer(cmd);
+
+ transport_check_aborted_status(se_cmd,
+ (hdr->flags & ISCSI_FLAG_CMD_FINAL));
+ return iscsit_dump_data_payload(conn, payload_length, 1);
+ }
+ } else {
+ /*
+ * For the normal solicited data path:
+ *
+ * Check for a delayed TASK_ABORTED status and dump any
+ * incoming data out payload if one exists. Also, when the
+ * ISCSI_FLAG_CMD_FINAL is set to denote the end of the current
+ * data out sequence, we decrement outstanding_r2ts. Once
+ * outstanding_r2ts reaches zero, go ahead and send the delayed
+ * TASK_ABORTED status.
+ */
+ if (atomic_read(&T_TASK(se_cmd)->t_transport_aborted) != 0) {
+ if (hdr->flags & ISCSI_FLAG_CMD_FINAL)
+ if (--cmd->outstanding_r2ts < 1) {
+ iscsit_stop_dataout_timer(cmd);
+ transport_check_aborted_status(
+ se_cmd, 1);
+ }
+
+ return iscsit_dump_data_payload(conn, payload_length, 1);
+ }
+ }
+ /*
+ * Preform DataSN, DataSequenceInOrder, DataPDUInOrder, and
+ * within-command recovery checks before receiving the payload.
+ */
+ ret = iscsit_check_pre_dataout(cmd, buf);
+ if (ret == DATAOUT_WITHIN_COMMAND_RECOVERY)
+ return 0;
+ else if (ret == DATAOUT_CANNOT_RECOVER)
+ return -1;
+
+ rx_size += payload_length;
+ iov = &cmd->iov_data[0];
+
+ memset(&map_sg, 0, sizeof(struct se_map_sg));
+ memset(&unmap_sg, 0, sizeof(struct se_unmap_sg));
+ map_sg.fabric_cmd = (void *)cmd;
+ map_sg.se_cmd = SE_CMD(cmd);
+ map_sg.iov = iov;
+ map_sg.sg_kmap_active = 1;
+ map_sg.data_length = payload_length;
+ map_sg.data_offset = hdr->offset;
+ unmap_sg.fabric_cmd = (void *)cmd;
+ unmap_sg.se_cmd = SE_CMD(cmd);
+
+ iov_ret = iscsit_set_iovec_ptrs(&map_sg, &unmap_sg);
+ if (iov_ret < 0)
+ return -1;
+
+ iov_count += iov_ret;
+
+ padding = ((-payload_length) & 3);
+ if (padding != 0) {
+ iov[iov_count].iov_base = &pad_bytes;
+ iov[iov_count++].iov_len = padding;
+ rx_size += padding;
+ TRACE(TRACE_ISCSI, "Receiving %u padding bytes.\n", padding);
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ iov[iov_count].iov_base = &checksum;
+ iov[iov_count++].iov_len = CRC_LEN;
+ rx_size += CRC_LEN;
+ }
+
+ iscsit_map_SG_segments(&unmap_sg);
+
+ rx_got = rx_data(conn, &cmd->iov_data[0], iov_count, rx_size);
+
+ iscsit_unmap_SG_segments(&unmap_sg);
+
+ if (rx_got != rx_size)
+ return -1;
+
+ if (conn->conn_ops->DataDigest) {
+ u32 counter = payload_length, data_crc = 0;
+ struct kvec *iov_ptr = &cmd->iov_data[0];
+ struct scatterlist sg;
+ /*
+ * Thanks to the IP stack shitting on passed iovecs, we have to
+ * call set_iovec_data_ptrs() again in order to have a iMD/PSCSI
+ * agnostic way of doing datadigests computations.
+ */
+ memset(&map_sg, 0, sizeof(struct se_map_sg));
+ map_sg.fabric_cmd = (void *)cmd;
+ map_sg.se_cmd = SE_CMD(cmd);
+ map_sg.iov = iov_ptr;
+ map_sg.data_length = payload_length;
+ map_sg.data_offset = hdr->offset;
+
+ if (iscsit_set_iovec_ptrs(&map_sg, &unmap_sg) < 0)
+ return -1;
+
+ crypto_hash_init(&conn->conn_rx_hash);
+
+ while (counter > 0) {
+ sg_init_one(&sg, iov_ptr->iov_base,
+ iov_ptr->iov_len);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ iov_ptr->iov_len);
+
+ TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %zu"
+ " bytes, CRC 0x%08x\n", iov_ptr->iov_len,
+ data_crc);
+ counter -= iov_ptr->iov_len;
+ iov_ptr++;
+ }
+
+ if (padding) {
+ sg_init_one(&sg, (u8 *)&pad_bytes, padding);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ padding);
+ TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %d"
+ " bytes of padding, CRC 0x%08x\n",
+ padding, data_crc);
+ }
+ crypto_hash_final(&conn->conn_rx_hash, (u8 *)&data_crc);
+
+ if (checksum != data_crc) {
+ printk(KERN_ERR "ITT: 0x%08x, Offset: %u, Length: %u,"
+ " DataSN: 0x%08x, CRC32C DataDigest 0x%08x"
+ " does not match computed 0x%08x\n",
+ hdr->itt, hdr->offset, payload_length,
+ hdr->datasn, checksum, data_crc);
+ data_crc_failed = 1;
+ } else {
+ TRACE(TRACE_DIGEST, "Got CRC32C DataDigest 0x%08x for"
+ " %u bytes of Data Out\n", checksum,
+ payload_length);
+ }
+ }
+ /*
+ * Increment post receive data and CRC values or perform
+ * within-command recovery.
+ */
+ ret = iscsit_check_post_dataout(cmd, buf, data_crc_failed);
+ if ((ret == DATAOUT_NORMAL) || (ret == DATAOUT_WITHIN_COMMAND_RECOVERY))
+ return 0;
+ else if (ret == DATAOUT_SEND_R2T) {
+ iscsit_set_dataout_sequence_values(cmd);
+ iscsit_build_r2ts_for_cmd(cmd, conn, 0);
+ } else if (ret == DATAOUT_SEND_TO_TRANSPORT) {
+ /*
+ * Handle extra special case for out of order
+ * Unsolicited Data Out.
+ */
+ spin_lock_bh(&cmd->istate_lock);
+ ooo_cmdsn = (cmd->cmd_flags & ICF_OOO_CMDSN);
+ cmd->cmd_flags |= ICF_GOT_LAST_DATAOUT;
+ cmd->i_state = ISTATE_RECEIVED_LAST_DATAOUT;
+ spin_unlock_bh(&cmd->istate_lock);
+
+ iscsit_stop_dataout_timer(cmd);
+ return (!ooo_cmdsn) ? transport_generic_handle_data(
+ SE_CMD(cmd)) : 0;
+ } else /* DATAOUT_CANNOT_RECOVER */
+ return -1;
+
+ return 0;
+}
+
+static inline int iscsit_handle_nop_out(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ unsigned char *ping_data = NULL;
+ int cmdsn_ret, niov = 0, ret = 0, rx_got, rx_size;
+ u32 checksum, data_crc, padding = 0, payload_length;
+ u64 lun;
+ struct iscsi_cmd *cmd = NULL;
+ struct kvec *iov = NULL;
+ struct iscsi_nopout *hdr;
+ struct scatterlist sg;
+
+ hdr = (struct iscsi_nopout *) buf;
+ payload_length = ntoh24(hdr->dlength);
+ lun = get_unaligned_le64(&hdr->lun[0]);
+ hdr->itt = be32_to_cpu(hdr->itt);
+ hdr->ttt = be32_to_cpu(hdr->ttt);
+ hdr->cmdsn = be32_to_cpu(hdr->cmdsn);
+ hdr->exp_statsn = be32_to_cpu(hdr->exp_statsn);
+
+ if ((hdr->itt == 0xFFFFFFFF) && !(hdr->opcode & ISCSI_OP_IMMEDIATE)) {
+ printk(KERN_ERR "NOPOUT ITT is reserved, but Immediate Bit is"
+ " not set, protocol error.\n");
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ if (payload_length > conn->conn_ops->MaxRecvDataSegmentLength) {
+ printk(KERN_ERR "NOPOUT Ping Data DataSegmentLength: %u is"
+ " greater than MaxRecvDataSegmentLength: %u, protocol"
+ " error.\n", payload_length,
+ conn->conn_ops->MaxRecvDataSegmentLength);
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ TRACE(TRACE_ISCSI, "Got NOPOUT Ping %s ITT: 0x%08x, TTT: 0x%09x,"
+ " CmdSN: 0x%08x, ExpStatSN: 0x%08x, Length: %u\n",
+ (hdr->itt == 0xFFFFFFFF) ? "Response" : "Request",
+ hdr->itt, hdr->ttt, hdr->cmdsn, hdr->exp_statsn,
+ payload_length);
+ /*
+ * This is not a response to a Unsolicited NopIN, which means
+ * it can either be a NOPOUT ping request (with a valid ITT),
+ * or a NOPOUT not requesting a NOPIN (with a reserved ITT).
+ * Either way, make sure we allocate an struct iscsi_cmd, as both
+ * can contain ping data.
+ */
+ if (hdr->ttt == 0xFFFFFFFF) {
+ cmd = iscsit_allocate_cmd(conn);
+ if (!cmd)
+ return iscsit_add_reject(
+ ISCSI_REASON_BOOKMARK_NO_RESOURCES,
+ 1, buf, conn);
+
+ cmd->iscsi_opcode = ISCSI_OP_NOOP_OUT;
+ cmd->i_state = ISTATE_SEND_NOPIN;
+ cmd->immediate_cmd = ((hdr->opcode & ISCSI_OP_IMMEDIATE) ?
+ 1 : 0);
+ conn->sess->init_task_tag = cmd->init_task_tag = hdr->itt;
+ cmd->targ_xfer_tag = 0xFFFFFFFF;
+ cmd->cmd_sn = hdr->cmdsn;
+ cmd->exp_stat_sn = hdr->exp_statsn;
+ cmd->data_direction = DMA_NONE;
+ }
+
+ if (payload_length && (hdr->ttt == 0xFFFFFFFF)) {
+ rx_size = payload_length;
+ ping_data = kzalloc(payload_length + 1, GFP_KERNEL);
+ if (!ping_data) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " NOPOUT ping data.\n");
+ ret = -1;
+ goto out;
+ }
+
+ iov = &cmd->iov_misc[0];
+ iov[niov].iov_base = ping_data;
+ iov[niov++].iov_len = payload_length;
+
+ padding = ((-payload_length) & 3);
+ if (padding != 0) {
+ TRACE(TRACE_ISCSI, "Receiving %u additional bytes"
+ " for padding.\n", padding);
+ iov[niov].iov_base = &cmd->pad_bytes;
+ iov[niov++].iov_len = padding;
+ rx_size += padding;
+ }
+ if (conn->conn_ops->DataDigest) {
+ iov[niov].iov_base = &checksum;
+ iov[niov++].iov_len = CRC_LEN;
+ rx_size += CRC_LEN;
+ }
+
+ rx_got = rx_data(conn, &cmd->iov_misc[0], niov, rx_size);
+ if (rx_got != rx_size) {
+ ret = -1;
+ goto out;
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ crypto_hash_init(&conn->conn_rx_hash);
+
+ sg_init_one(&sg, (u8 *)ping_data, payload_length);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ payload_length);
+
+ if (padding) {
+ sg_init_one(&sg, (u8 *)&cmd->pad_bytes,
+ padding);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ padding);
+ }
+ crypto_hash_final(&conn->conn_rx_hash, (u8 *)&data_crc);
+
+ if (checksum != data_crc) {
+ printk(KERN_ERR "Ping data CRC32C DataDigest"
+ " 0x%08x does not match computed 0x%08x\n",
+ checksum, data_crc);
+ if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
+ printk(KERN_ERR "Unable to recover from"
+ " NOPOUT Ping DataCRC failure while in"
+ " ERL=0.\n");
+ ret = -1;
+ goto out;
+ } else {
+ /*
+ * Silently drop this PDU and let the
+ * initiator plug the CmdSN gap.
+ */
+ TRACE(TRACE_ERL1, "Dropping NOPOUT"
+ " Command CmdSN: 0x%08x due to"
+ " DataCRC error.\n", hdr->cmdsn);
+ ret = 0;
+ goto out;
+ }
+ } else {
+ TRACE(TRACE_DIGEST, "Got CRC32C DataDigest"
+ " 0x%08x for %u bytes of ping data.\n",
+ checksum, payload_length);
+ }
+ }
+
+ ping_data[payload_length] = '\0';
+ /*
+ * Attach ping data to struct iscsi_cmd->buf_ptr.
+ */
+ cmd->buf_ptr = (void *)ping_data;
+ cmd->buf_ptr_size = payload_length;
+
+ TRACE(TRACE_ISCSI, "Got %u bytes of NOPOUT ping"
+ " data.\n", payload_length);
+ TRACE(TRACE_ISCSI, "Ping Data: \"%s\"\n", ping_data);
+ }
+
+ if (hdr->itt != 0xFFFFFFFF) {
+ if (!cmd) {
+ printk(KERN_ERR "Checking CmdSN for NOPOUT,"
+ " but cmd is NULL!\n");
+ return -1;
+ }
+ /*
+ * Initiator is expecting a NopIN ping reply,
+ */
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ iscsit_ack_from_expstatsn(conn, hdr->exp_statsn);
+
+ if (hdr->opcode & ISCSI_OP_IMMEDIATE) {
+ iscsit_add_cmd_to_response_queue(cmd, conn,
+ cmd->i_state);
+ return 0;
+ }
+
+ cmdsn_ret = iscsit_check_received_cmdsn(conn, cmd, hdr->cmdsn);
+ if ((cmdsn_ret == CMDSN_NORMAL_OPERATION) ||
+ (cmdsn_ret == CMDSN_HIGHER_THAN_EXP)) {
+ return 0;
+ } else if (cmdsn_ret == CMDSN_LOWER_THAN_EXP) {
+ cmd->i_state = ISTATE_REMOVE;
+ iscsit_add_cmd_to_immediate_queue(cmd, conn,
+ cmd->i_state);
+ ret = 0;
+ goto ping_out;
+ } else { /* (cmdsn_ret == CMDSN_ERROR_CANNOT_RECOVER) */
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ ret = -1;
+ goto ping_out;
+ }
+
+ return 0;
+ }
+
+ if (hdr->ttt != 0xFFFFFFFF) {
+ /*
+ * This was a response to a unsolicited NOPIN ping.
+ */
+ cmd = iscsit_find_cmd_from_ttt(conn, hdr->ttt);
+ if (!cmd)
+ return -1;
+
+ iscsit_stop_nopin_response_timer(conn);
+
+ cmd->i_state = ISTATE_REMOVE;
+ iscsit_add_cmd_to_immediate_queue(cmd, conn, cmd->i_state);
+ iscsit_start_nopin_timer(conn);
+ } else {
+ /*
+ * Initiator is not expecting a NOPIN is response.
+ * Just ignore for now.
+ *
+ * iSCSI v19-91 10.18
+ * "A NOP-OUT may also be used to confirm a changed
+ * ExpStatSN if another PDU will not be available
+ * for a long time."
+ */
+ ret = 0;
+ goto out;
+ }
+
+ return 0;
+out:
+ if (cmd)
+ iscsit_release_cmd(cmd);
+ping_out:
+ kfree(ping_data);
+ return ret;
+}
+
+static inline int iscsit_handle_task_mgt_cmd(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ struct iscsi_cmd *cmd;
+ struct se_tmr_req *se_tmr;
+ struct iscsi_tmr_req *tmr_req;
+ struct iscsi_tm *hdr;
+ u32 payload_length;
+ int cmdsn_ret, out_of_order_cmdsn = 0, ret;
+ u8 function;
+
+ hdr = (struct iscsi_tm *) buf;
+ payload_length = ntoh24(hdr->dlength);
+ hdr->itt = be32_to_cpu(hdr->itt);
+ hdr->rtt = be32_to_cpu(hdr->rtt);
+ hdr->cmdsn = be32_to_cpu(hdr->cmdsn);
+ hdr->exp_statsn = be32_to_cpu(hdr->exp_statsn);
+ hdr->refcmdsn = be32_to_cpu(hdr->refcmdsn);
+ hdr->exp_datasn = be32_to_cpu(hdr->exp_datasn);
+ hdr->flags &= ~ISCSI_FLAG_CMD_FINAL;
+ function = hdr->flags;
+
+ TRACE(TRACE_ISCSI, "Got Task Management Request ITT: 0x%08x, CmdSN:"
+ " 0x%08x, Function: 0x%02x, RefTaskTag: 0x%08x, RefCmdSN:"
+ " 0x%08x, CID: %hu\n", hdr->itt, hdr->cmdsn, function,
+ hdr->rtt, hdr->refcmdsn, conn->cid);
+
+ if ((function != ISCSI_TM_FUNC_ABORT_TASK) &&
+ ((function != ISCSI_TM_FUNC_TASK_REASSIGN) &&
+ (hdr->rtt != ISCSI_RESERVED_TAG))) {
+ printk(KERN_ERR "RefTaskTag should be set to 0xFFFFFFFF.\n");
+ hdr->rtt = ISCSI_RESERVED_TAG;
+ }
+
+ if ((function == ISCSI_TM_FUNC_TASK_REASSIGN) &&
+ !(hdr->opcode & ISCSI_OP_IMMEDIATE)) {
+ printk(KERN_ERR "Task Management Request TASK_REASSIGN not"
+ " issued as immediate command, bad iSCSI Initiator"
+ "implementation\n");
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+ if ((function != ISCSI_TM_FUNC_ABORT_TASK) &&
+ (hdr->refcmdsn != ISCSI_RESERVED_TAG))
+ hdr->refcmdsn = ISCSI_RESERVED_TAG;
+
+ cmd = iscsit_allocate_se_cmd_for_tmr(conn, function);
+ if (!cmd)
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_NO_RESOURCES,
+ 1, buf, conn);
+
+ cmd->iscsi_opcode = ISCSI_OP_SCSI_TMFUNC;
+ cmd->i_state = ISTATE_SEND_TASKMGTRSP;
+ cmd->immediate_cmd = ((hdr->opcode & ISCSI_OP_IMMEDIATE) ? 1 : 0);
+ cmd->init_task_tag = hdr->itt;
+ cmd->targ_xfer_tag = 0xFFFFFFFF;
+ cmd->cmd_sn = hdr->cmdsn;
+ cmd->exp_stat_sn = hdr->exp_statsn;
+ se_tmr = SE_CMD(cmd)->se_tmr_req;
+ tmr_req = cmd->tmr_req;
+ /*
+ * Locate the struct se_lun for all TMRs not related to ERL=2 TASK_REASSIGN
+ */
+ if (function != ISCSI_TM_FUNC_TASK_REASSIGN) {
+ ret = iscsit_get_lun_for_tmr(cmd,
+ get_unaligned_le64(&hdr->lun[0]));
+ if (ret < 0) {
+ SE_CMD(cmd)->se_cmd_flags |= SCF_SCSI_CDB_EXCEPTION;
+ se_tmr->response = ISCSI_TMF_RSP_NO_LUN;
+ goto attach;
+ }
+ }
+
+ switch (function) {
+ case ISCSI_TM_FUNC_ABORT_TASK:
+ se_tmr->response = iscsit_tmr_abort_task(cmd, buf);
+ if (se_tmr->response != ISCSI_TMF_RSP_COMPLETE) {
+ SE_CMD(cmd)->se_cmd_flags |= SCF_SCSI_CDB_EXCEPTION;
+ goto attach;
+ }
+ break;
+ case ISCSI_TM_FUNC_ABORT_TASK_SET:
+ case ISCSI_TM_FUNC_CLEAR_ACA:
+ case ISCSI_TM_FUNC_CLEAR_TASK_SET:
+ case ISCSI_TM_FUNC_LOGICAL_UNIT_RESET:
+ break;
+ case ISCSI_TM_FUNC_TARGET_WARM_RESET:
+ if (iscsit_tmr_task_warm_reset(conn, tmr_req, buf) < 0) {
+ SE_CMD(cmd)->se_cmd_flags |= SCF_SCSI_CDB_EXCEPTION;
+ se_tmr->response = ISCSI_TMF_RSP_AUTH_FAILED;
+ goto attach;
+ }
+ break;
+ case ISCSI_TM_FUNC_TARGET_COLD_RESET:
+ if (iscsit_tmr_task_cold_reset(conn, tmr_req, buf) < 0) {
+ SE_CMD(cmd)->se_cmd_flags |= SCF_SCSI_CDB_EXCEPTION;
+ se_tmr->response = ISCSI_TMF_RSP_AUTH_FAILED;
+ goto attach;
+ }
+ break;
+ case ISCSI_TM_FUNC_TASK_REASSIGN:
+ se_tmr->response = iscsit_tmr_task_reassign(cmd, buf);
+ /*
+ * Perform sanity checks on the ExpDataSN only if the
+ * TASK_REASSIGN was successful.
+ */
+ if (se_tmr->response != ISCSI_TMF_RSP_COMPLETE)
+ break;
+
+ if (iscsit_check_task_reassign_expdatasn(tmr_req, conn) < 0)
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_BOOKMARK_INVALID, 1, 1,
+ buf, cmd);
+ break;
+ default:
+ printk(KERN_ERR "Unknown TMR function: 0x%02x, protocol"
+ " error.\n", function);
+ SE_CMD(cmd)->se_cmd_flags |= SCF_SCSI_CDB_EXCEPTION;
+ se_tmr->response = ISCSI_TMF_RSP_NOT_SUPPORTED;
+ goto attach;
+ }
+
+ if ((function != ISCSI_TM_FUNC_TASK_REASSIGN) &&
+ (se_tmr->response == ISCSI_TMF_RSP_COMPLETE))
+ se_tmr->call_transport = 1;
+attach:
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (!(hdr->opcode & ISCSI_OP_IMMEDIATE)) {
+ cmdsn_ret = iscsit_check_received_cmdsn(conn,
+ cmd, hdr->cmdsn);
+ if (cmdsn_ret == CMDSN_NORMAL_OPERATION)
+ do {} while (0);
+ else if (cmdsn_ret == CMDSN_HIGHER_THAN_EXP)
+ out_of_order_cmdsn = 1;
+ else if (cmdsn_ret == CMDSN_LOWER_THAN_EXP) {
+ cmd->i_state = ISTATE_REMOVE;
+ iscsit_add_cmd_to_immediate_queue(cmd, conn,
+ cmd->i_state);
+ return 0;
+ } else { /* (cmdsn_ret == CMDSN_ERROR_CANNOT_RECOVER) */
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ }
+ }
+ iscsit_ack_from_expstatsn(conn, hdr->exp_statsn);
+
+ if (out_of_order_cmdsn)
+ return 0;
+ /*
+ * Found the referenced task, send to transport for processing.
+ */
+ if (se_tmr->call_transport)
+ return transport_generic_handle_tmr(SE_CMD(cmd));
+
+ /*
+ * Could not find the referenced LUN, task, or Task Management
+ * command not authorized or supported. Change state and
+ * let the tx_thread send the response.
+ *
+ * For connection recovery, this is also the default action for
+ * TMR TASK_REASSIGN.
+ */
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ return 0;
+}
+
+/* #warning FIXME: Support Text Command parameters besides SendTargets */
+static inline int iscsit_handle_text_cmd(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ char *text_ptr, *text_in;
+ int cmdsn_ret, niov = 0, rx_got, rx_size;
+ u32 checksum = 0, data_crc = 0, payload_length;
+ u32 padding = 0, pad_bytes = 0, text_length = 0;
+ struct iscsi_cmd *cmd;
+ struct kvec iov[3];
+ struct iscsi_text *hdr;
+ struct scatterlist sg;
+
+ hdr = (struct iscsi_text *) buf;
+ payload_length = ntoh24(hdr->dlength);
+ hdr->itt = be32_to_cpu(hdr->itt);
+ hdr->ttt = be32_to_cpu(hdr->ttt);
+ hdr->cmdsn = be32_to_cpu(hdr->cmdsn);
+ hdr->exp_statsn = be32_to_cpu(hdr->exp_statsn);
+
+ if (payload_length > conn->conn_ops->MaxRecvDataSegmentLength) {
+ printk(KERN_ERR "Unable to accept text parameter length: %u"
+ "greater than MaxRecvDataSegmentLength %u.\n",
+ payload_length, conn->conn_ops->MaxRecvDataSegmentLength);
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ TRACE(TRACE_ISCSI, "Got Text Request: ITT: 0x%08x, CmdSN: 0x%08x,"
+ " ExpStatSN: 0x%08x, Length: %u\n", hdr->itt, hdr->cmdsn,
+ hdr->exp_statsn, payload_length);
+
+ rx_size = text_length = payload_length;
+ if (text_length) {
+ text_in = kzalloc(text_length, GFP_KERNEL);
+ if (!text_in) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " incoming text parameters\n");
+ return -1;
+ }
+
+ memset(iov, 0, 3 * sizeof(struct kvec));
+ iov[niov].iov_base = text_in;
+ iov[niov++].iov_len = text_length;
+
+ padding = ((-payload_length) & 3);
+ if (padding != 0) {
+ iov[niov].iov_base = &pad_bytes;
+ iov[niov++].iov_len = padding;
+ rx_size += padding;
+ TRACE(TRACE_ISCSI, "Receiving %u additional bytes"
+ " for padding.\n", padding);
+ }
+ if (conn->conn_ops->DataDigest) {
+ iov[niov].iov_base = &checksum;
+ iov[niov++].iov_len = CRC_LEN;
+ rx_size += CRC_LEN;
+ }
+
+ rx_got = rx_data(conn, &iov[0], niov, rx_size);
+ if (rx_got != rx_size) {
+ kfree(text_in);
+ return -1;
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ crypto_hash_init(&conn->conn_rx_hash);
+
+ sg_init_one(&sg, (u8 *)text_in, text_length);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ text_length);
+
+ if (padding) {
+ sg_init_one(&sg, (u8 *)&pad_bytes, padding);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ padding);
+ }
+ crypto_hash_final(&conn->conn_rx_hash, (u8 *)&data_crc);
+
+ if (checksum != data_crc) {
+ printk(KERN_ERR "Text data CRC32C DataDigest"
+ " 0x%08x does not match computed"
+ " 0x%08x\n", checksum, data_crc);
+ if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
+ printk(KERN_ERR "Unable to recover from"
+ " Text Data digest failure while in"
+ " ERL=0.\n");
+ kfree(text_in);
+ return -1;
+ } else {
+ /*
+ * Silently drop this PDU and let the
+ * initiator plug the CmdSN gap.
+ */
+ TRACE(TRACE_ERL1, "Dropping Text"
+ " Command CmdSN: 0x%08x due to"
+ " DataCRC error.\n", hdr->cmdsn);
+ kfree(text_in);
+ return 0;
+ }
+ } else {
+ TRACE(TRACE_DIGEST, "Got CRC32C DataDigest"
+ " 0x%08x for %u bytes of text data.\n",
+ checksum, text_length);
+ }
+ }
+ text_in[text_length - 1] = '\0';
+ TRACE(TRACE_ISCSI, "Successfully read %d bytes of text"
+ " data.\n", text_length);
+
+ if (strncmp("SendTargets", text_in, 11) != 0) {
+ printk(KERN_ERR "Received Text Data that is not"
+ " SendTargets, cannot continue.\n");
+ kfree(text_in);
+ return -1;
+ }
+ text_ptr = strchr(text_in, '=');
+ if (!text_ptr) {
+ printk(KERN_ERR "No \"=\" separator found in Text Data,"
+ " cannot continue.\n");
+ kfree(text_in);
+ return -1;
+ }
+ if (strncmp("=All", text_ptr, 4) != 0) {
+ printk(KERN_ERR "Unable to locate All value for"
+ " SendTargets key, cannot continue.\n");
+ kfree(text_in);
+ return -1;
+ }
+/*#warning Support SendTargets=(iSCSI Target Name/Nothing) values. */
+ kfree(text_in);
+ }
+
+ cmd = iscsit_allocate_cmd(conn);
+ if (!cmd)
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_NO_RESOURCES,
+ 1, buf, conn);
+
+ cmd->iscsi_opcode = ISCSI_OP_TEXT;
+ cmd->i_state = ISTATE_SEND_TEXTRSP;
+ cmd->immediate_cmd = ((hdr->opcode & ISCSI_OP_IMMEDIATE) ? 1 : 0);
+ conn->sess->init_task_tag = cmd->init_task_tag = hdr->itt;
+ cmd->targ_xfer_tag = 0xFFFFFFFF;
+ cmd->cmd_sn = hdr->cmdsn;
+ cmd->exp_stat_sn = hdr->exp_statsn;
+ cmd->data_direction = DMA_NONE;
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ iscsit_ack_from_expstatsn(conn, hdr->exp_statsn);
+
+ if (!(hdr->opcode & ISCSI_OP_IMMEDIATE)) {
+ cmdsn_ret = iscsit_check_received_cmdsn(conn, cmd, hdr->cmdsn);
+ if ((cmdsn_ret == CMDSN_NORMAL_OPERATION) ||
+ (cmdsn_ret == CMDSN_HIGHER_THAN_EXP))
+ return 0;
+ else if (cmdsn_ret == CMDSN_LOWER_THAN_EXP) {
+ iscsit_add_cmd_to_immediate_queue(cmd, conn,
+ ISTATE_REMOVE);
+ return 0;
+ } else { /* (cmdsn_ret == CMDSN_ERROR_CANNOT_RECOVER) */
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ }
+
+ return 0;
+ }
+
+ return iscsit_execute_cmd(cmd, 0);
+}
+
+int iscsit_logout_closesession(struct iscsi_cmd *cmd, struct iscsi_conn *conn)
+{
+ struct iscsi_conn *conn_p;
+ struct iscsi_session *sess = conn->sess;
+
+ TRACE(TRACE_ISCSI, "Received logout request CLOSESESSION on CID: %hu"
+ " for SID: %u.\n", conn->cid, conn->sess->sid);
+
+ atomic_set(&sess->session_logout, 1);
+ atomic_set(&conn->conn_logout_remove, 1);
+ conn->conn_logout_reason = ISCSI_LOGOUT_REASON_CLOSE_SESSION;
+
+ iscsit_inc_conn_usage_count(conn);
+ iscsit_inc_session_usage_count(sess);
+
+ spin_lock_bh(&sess->conn_lock);
+ list_for_each_entry(conn_p, &sess->sess_conn_list, conn_list) {
+ if (conn_p->conn_state != TARG_CONN_STATE_LOGGED_IN)
+ continue;
+
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_IN_LOGOUT.\n");
+ conn_p->conn_state = TARG_CONN_STATE_IN_LOGOUT;
+ }
+ spin_unlock_bh(&sess->conn_lock);
+
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+
+ return 0;
+}
+
+int iscsit_logout_closeconnection(struct iscsi_cmd *cmd, struct iscsi_conn *conn)
+{
+ struct iscsi_conn *l_conn;
+ struct iscsi_session *sess = conn->sess;
+
+ TRACE(TRACE_ISCSI, "Received logout request CLOSECONNECTION for CID:"
+ " %hu on CID: %hu.\n", cmd->logout_cid, conn->cid);
+
+ /*
+ * A Logout Request with a CLOSECONNECTION reason code for a CID
+ * can arrive on a connection with a differing CID.
+ */
+ if (conn->cid == cmd->logout_cid) {
+ spin_lock_bh(&conn->state_lock);
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_IN_LOGOUT.\n");
+ conn->conn_state = TARG_CONN_STATE_IN_LOGOUT;
+
+ atomic_set(&conn->conn_logout_remove, 1);
+ conn->conn_logout_reason = ISCSI_LOGOUT_REASON_CLOSE_CONNECTION;
+ iscsit_inc_conn_usage_count(conn);
+
+ spin_unlock_bh(&conn->state_lock);
+ } else {
+ /*
+ * Handle all different cid CLOSECONNECTION requests in
+ * iscsit_logout_post_handler_diffcid() as to give enough
+ * time for any non immediate command's CmdSN to be
+ * acknowledged on the connection in question.
+ *
+ * Here we simply make sure the CID is still around.
+ */
+ l_conn = iscsit_get_conn_from_cid(sess,
+ cmd->logout_cid);
+ if (!l_conn) {
+ cmd->logout_response = ISCSI_LOGOUT_CID_NOT_FOUND;
+ iscsit_add_cmd_to_response_queue(cmd, conn,
+ cmd->i_state);
+ return 0;
+ }
+
+ iscsit_dec_conn_usage_count(l_conn);
+ }
+
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+
+ return 0;
+}
+
+int iscsit_logout_removeconnforrecovery(struct iscsi_cmd *cmd, struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+
+ TRACE(TRACE_ERL2, "Received explicit REMOVECONNFORRECOVERY logout for"
+ " CID: %hu on CID: %hu.\n", cmd->logout_cid, conn->cid);
+
+ if (sess->sess_ops->ErrorRecoveryLevel != 2) {
+ printk(KERN_ERR "Received Logout Request REMOVECONNFORRECOVERY"
+ " while ERL!=2.\n");
+ cmd->logout_response = ISCSI_LOGOUT_RECOVERY_UNSUPPORTED;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ return 0;
+ }
+
+ if (conn->cid == cmd->logout_cid) {
+ printk(KERN_ERR "Received Logout Request REMOVECONNFORRECOVERY"
+ " with CID: %hu on CID: %hu, implementation error.\n",
+ cmd->logout_cid, conn->cid);
+ cmd->logout_response = ISCSI_LOGOUT_CLEANUP_FAILED;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ return 0;
+ }
+
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+
+ return 0;
+}
+
+static inline int iscsit_handle_logout_cmd(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ int cmdsn_ret, logout_remove = 0;
+ u8 reason_code = 0;
+ struct iscsi_cmd *cmd;
+ struct iscsi_logout *hdr;
+ struct iscsi_tiqn *tiqn = iscsit_snmp_get_tiqn(conn);
+
+ hdr = (struct iscsi_logout *) buf;
+ reason_code = (hdr->flags & 0x7f);
+ hdr->itt = be32_to_cpu(hdr->itt);
+ hdr->cid = be16_to_cpu(hdr->cid);
+ hdr->cmdsn = be32_to_cpu(hdr->cmdsn);
+ hdr->exp_statsn = be32_to_cpu(hdr->exp_statsn);
+
+ if (tiqn) {
+ spin_lock(&tiqn->logout_stats.lock);
+ if (reason_code == ISCSI_LOGOUT_REASON_CLOSE_SESSION)
+ tiqn->logout_stats.normal_logouts++;
+ else
+ tiqn->logout_stats.abnormal_logouts++;
+ spin_unlock(&tiqn->logout_stats.lock);
+ }
+
+ TRACE(TRACE_ISCSI, "Got Logout Request ITT: 0x%08x CmdSN: 0x%08x"
+ " ExpStatSN: 0x%08x Reason: 0x%02x CID: %hu on CID: %hu\n",
+ hdr->itt, hdr->cmdsn, hdr->exp_statsn, reason_code,
+ hdr->cid, conn->cid);
+
+ if (conn->conn_state != TARG_CONN_STATE_LOGGED_IN) {
+ printk(KERN_ERR "Received logout request on connection that"
+ " is not in logged in state, ignoring request.\n");
+ return 0;
+ }
+
+ cmd = iscsit_allocate_cmd(conn);
+ if (!cmd)
+ return iscsit_add_reject(ISCSI_REASON_BOOKMARK_NO_RESOURCES, 1,
+ buf, conn);
+
+ cmd->iscsi_opcode = ISCSI_OP_LOGOUT;
+ cmd->i_state = ISTATE_SEND_LOGOUTRSP;
+ cmd->immediate_cmd = ((hdr->opcode & ISCSI_OP_IMMEDIATE) ? 1 : 0);
+ conn->sess->init_task_tag = cmd->init_task_tag = hdr->itt;
+ cmd->targ_xfer_tag = 0xFFFFFFFF;
+ cmd->cmd_sn = hdr->cmdsn;
+ cmd->exp_stat_sn = hdr->exp_statsn;
+ cmd->logout_cid = hdr->cid;
+ cmd->logout_reason = reason_code;
+ cmd->data_direction = DMA_NONE;
+
+ /*
+ * We need to sleep in these cases (by returning 1) until the Logout
+ * Response gets sent in the tx thread.
+ */
+ if ((reason_code == ISCSI_LOGOUT_REASON_CLOSE_SESSION) ||
+ ((reason_code == ISCSI_LOGOUT_REASON_CLOSE_CONNECTION) &&
+ (hdr->cid == conn->cid)))
+ logout_remove = 1;
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (reason_code != ISCSI_LOGOUT_REASON_RECOVERY)
+ iscsit_ack_from_expstatsn(conn, hdr->exp_statsn);
+
+ /*
+ * Non-Immediate Logout Commands are executed in CmdSN order..
+ */
+ if (!(hdr->opcode & ISCSI_OP_IMMEDIATE)) {
+ cmdsn_ret = iscsit_check_received_cmdsn(conn, cmd, hdr->cmdsn);
+ if ((cmdsn_ret == CMDSN_NORMAL_OPERATION) ||
+ (cmdsn_ret == CMDSN_HIGHER_THAN_EXP))
+ return logout_remove;
+ else if (cmdsn_ret == CMDSN_LOWER_THAN_EXP) {
+ cmd->i_state = ISTATE_REMOVE;
+ iscsit_add_cmd_to_immediate_queue(cmd, conn,
+ cmd->i_state);
+ return 0;
+ } else { /* (cmdsn_ret == CMDSN_ERROR_CANNOT_RECOVER) */
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ }
+ }
+ /*
+ * Immediate Logout Commands are executed, well, Immediately.
+ */
+ if (iscsit_execute_cmd(cmd, 0) < 0)
+ return -1;
+
+ return logout_remove;
+}
+
+static inline int iscsit_handle_snack(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ u32 debug_type, unpacked_lun;
+ u64 lun;
+ struct iscsi_snack *hdr;
+
+ hdr = (struct iscsi_snack *) buf;
+ hdr->flags &= ~ISCSI_FLAG_CMD_FINAL;
+ lun = get_unaligned_le64(&hdr->lun[0]);
+ unpacked_lun = scsilun_to_int((struct scsi_lun *)&lun);
+ hdr->itt = be32_to_cpu(hdr->itt);
+ hdr->ttt = be32_to_cpu(hdr->ttt);
+ hdr->exp_statsn = be32_to_cpu(hdr->exp_statsn);
+ hdr->begrun = be32_to_cpu(hdr->begrun);
+ hdr->runlength = be32_to_cpu(hdr->runlength);
+
+ debug_type = (hdr->flags & 0x02) ? TRACE_ISCSI : TRACE_ERL1;
+ TRACE(debug_type, "Got ISCSI_INIT_SNACK, ITT: 0x%08x, ExpStatSN:"
+ " 0x%08x, Type: 0x%02x, BegRun: 0x%08x, RunLength: 0x%08x,"
+ " CID: %hu\n", hdr->itt, hdr->exp_statsn, hdr->flags,
+ hdr->begrun, hdr->runlength, conn->cid);
+
+ if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
+ printk(KERN_ERR "Initiator sent SNACK request while in"
+ " ErrorRecoveryLevel=0.\n");
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+ /*
+ * SNACK_DATA and SNACK_R2T are both 0, so check which function to
+ * call from inside iscsi_send_recovery_datain_or_r2t().
+ */
+ switch (hdr->flags & ISCSI_FLAG_SNACK_TYPE_MASK) {
+ case 0:
+ return iscsit_handle_recovery_datain_or_r2t(conn, buf,
+ hdr->itt, hdr->ttt, hdr->begrun, hdr->runlength);
+ return 0;
+ case ISCSI_FLAG_SNACK_TYPE_STATUS:
+ return iscsit_handle_status_snack(conn, hdr->itt, hdr->ttt,
+ hdr->begrun, hdr->runlength);
+ case ISCSI_FLAG_SNACK_TYPE_DATA_ACK:
+ return iscsit_handle_data_ack(conn, hdr->ttt, hdr->begrun,
+ hdr->runlength);
+ case ISCSI_FLAG_SNACK_TYPE_RDATA:
+ /* FIXME: Support R-Data SNACK */
+ printk(KERN_ERR "R-Data SNACK Not Supported.\n");
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ default:
+ printk(KERN_ERR "Unknown SNACK type 0x%02x, protocol"
+ " error.\n", hdr->flags & 0x0f);
+ return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buf, conn);
+ }
+
+ return 0;
+}
+
+static int iscsit_handle_immediate_data(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf,
+ u32 length)
+{
+ int iov_ret, rx_got = 0, rx_size = 0;
+ u32 checksum, iov_count = 0, padding = 0, pad_bytes = 0;
+ struct iscsi_conn *conn = cmd->conn;
+ struct se_map_sg map_sg;
+ struct se_unmap_sg unmap_sg;
+ struct kvec *iov;
+
+ memset(&map_sg, 0, sizeof(struct se_map_sg));
+ memset(&unmap_sg, 0, sizeof(struct se_unmap_sg));
+ map_sg.fabric_cmd = (void *)cmd;
+ map_sg.se_cmd = SE_CMD(cmd);
+ map_sg.sg_kmap_active = 1;
+ map_sg.iov = &cmd->iov_data[0];
+ map_sg.data_length = length;
+ map_sg.data_offset = cmd->write_data_done;
+ unmap_sg.fabric_cmd = (void *)cmd;
+ unmap_sg.se_cmd = SE_CMD(cmd);
+
+ iov_ret = iscsit_set_iovec_ptrs(&map_sg, &unmap_sg);
+ if (iov_ret < 0)
+ return IMMEDIDATE_DATA_CANNOT_RECOVER;
+
+ rx_size = length;
+ iov_count = iov_ret;
+ iov = &cmd->iov_data[0];
+
+ padding = ((-length) & 3);
+ if (padding != 0) {
+ iov[iov_count].iov_base = &pad_bytes;
+ iov[iov_count++].iov_len = padding;
+ rx_size += padding;
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ iov[iov_count].iov_base = &checksum;
+ iov[iov_count++].iov_len = CRC_LEN;
+ rx_size += CRC_LEN;
+ }
+
+ iscsit_map_SG_segments(&unmap_sg);
+
+ rx_got = rx_data(conn, &cmd->iov_data[0], iov_count, rx_size);
+
+ iscsit_unmap_SG_segments(&unmap_sg);
+
+ if (rx_got != rx_size) {
+ iscsit_rx_thread_wait_for_tcp(conn);
+ return IMMEDIDATE_DATA_CANNOT_RECOVER;
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ u32 counter = length, data_crc;
+ struct kvec *iov_ptr = &cmd->iov_data[0];
+ struct scatterlist sg;
+ /*
+ * Thanks to the IP stack shitting on passed iovecs, we have to
+ * call set_iovec_data_ptrs again in order to have a iMD/PSCSI
+ * agnostic way of doing datadigests computations.
+ */
+ memset(&map_sg, 0, sizeof(struct se_map_sg));
+ map_sg.fabric_cmd = (void *)cmd;
+ map_sg.se_cmd = SE_CMD(cmd);
+ map_sg.iov = iov_ptr;
+ map_sg.data_length = length;
+ map_sg.data_offset = cmd->write_data_done;
+
+ if (iscsit_set_iovec_ptrs(&map_sg, &unmap_sg) < 0)
+ return IMMEDIDATE_DATA_CANNOT_RECOVER;
+
+ crypto_hash_init(&conn->conn_rx_hash);
+
+ while (counter > 0) {
+ sg_init_one(&sg, iov_ptr->iov_base,
+ iov_ptr->iov_len);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ iov_ptr->iov_len);
+
+ TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %zu"
+ " bytes, CRC 0x%08x\n", iov_ptr->iov_len, data_crc);
+ counter -= iov_ptr->iov_len;
+ iov_ptr++;
+ }
+
+ if (padding) {
+ sg_init_one(&sg, (u8 *)&pad_bytes, padding);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ padding);
+ TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %d"
+ " bytes of padding, CRC 0x%08x\n", padding, data_crc);
+ }
+ crypto_hash_final(&conn->conn_rx_hash, (u8 *)&data_crc);
+
+ if (checksum != data_crc) {
+ printk(KERN_ERR "ImmediateData CRC32C DataDigest 0x%08x"
+ " does not match computed 0x%08x\n", checksum,
+ data_crc);
+
+ if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
+ printk(KERN_ERR "Unable to recover from"
+ " Immediate Data digest failure while"
+ " in ERL=0.\n");
+ iscsit_add_reject_from_cmd(
+ ISCSI_REASON_DATA_DIGEST_ERROR,
+ 1, 0, buf, cmd);
+ return IMMEDIDATE_DATA_CANNOT_RECOVER;
+ } else {
+ iscsit_add_reject_from_cmd(
+ ISCSI_REASON_DATA_DIGEST_ERROR,
+ 0, 0, buf, cmd);
+ return IMMEDIDATE_DATA_ERL1_CRC_FAILURE;
+ }
+ } else {
+ TRACE(TRACE_DIGEST, "Got CRC32C DataDigest 0x%08x for"
+ " %u bytes of Immediate Data\n", checksum,
+ length);
+ }
+ }
+
+ cmd->write_data_done += length;
+
+ if (cmd->write_data_done == cmd->data_length) {
+ spin_lock_bh(&cmd->istate_lock);
+ cmd->cmd_flags |= ICF_GOT_LAST_DATAOUT;
+ cmd->i_state = ISTATE_RECEIVED_LAST_DATAOUT;
+ spin_unlock_bh(&cmd->istate_lock);
+ }
+
+ return IMMEDIDATE_DATA_NORMAL_OPERATION;
+}
+
+int iscsit_send_async_msg(
+ struct iscsi_conn *conn,
+ u16 cid,
+ u8 async_event,
+ u8 async_vcode)
+{
+ u8 iscsi_hdr[ISCSI_HDR_LEN+CRC_LEN];
+ u32 tx_send = ISCSI_HDR_LEN, tx_sent = 0;
+ struct iscsi_async *hdr;
+ struct kvec iov;
+ struct scatterlist sg;
+
+ memset(&iov, 0, sizeof(struct kvec));
+ memset(&iscsi_hdr, 0, ISCSI_HDR_LEN+CRC_LEN);
+
+ hdr = (struct iscsi_async *)&iscsi_hdr;
+ hdr->opcode = ISCSI_OP_ASYNC_EVENT;
+ hdr->flags |= ISCSI_FLAG_CMD_FINAL;
+ hton24(hdr->dlength, 0);
+ put_unaligned_le64(0, &hdr->lun[0]);
+ put_unaligned_be64(0xffffffffffffffff, &hdr->rsvd4[0]);
+ hdr->statsn = cpu_to_be32(conn->stat_sn++);
+ spin_lock(&conn->sess->cmdsn_lock);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+ spin_unlock(&conn->sess->cmdsn_lock);
+ hdr->async_event = async_event;
+ hdr->async_vcode = async_vcode;
+
+ switch (async_event) {
+ case ISCSI_ASYNC_MSG_SCSI_EVENT:
+ printk(KERN_ERR "ISCSI_ASYNC_MSG_SCSI_EVENT: not supported yet.\n");
+ return -1;
+ case ISCSI_ASYNC_MSG_REQUEST_LOGOUT:
+ TRACE(TRACE_STATE, "Moving to"
+ " TARG_CONN_STATE_LOGOUT_REQUESTED.\n");
+ conn->conn_state = TARG_CONN_STATE_LOGOUT_REQUESTED;
+ hdr->param1 = 0;
+ hdr->param2 = 0;
+ hdr->param3 = cpu_to_be16(SECONDS_FOR_ASYNC_LOGOUT);
+ break;
+ case ISCSI_ASYNC_MSG_DROPPING_CONNECTION:
+ hdr->param1 = cpu_to_be16(cid);
+ hdr->param2 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Wait);
+ hdr->param3 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Retain);
+ break;
+ case ISCSI_ASYNC_MSG_DROPPING_ALL_CONNECTIONS:
+ hdr->param1 = 0;
+ hdr->param2 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Wait);
+ hdr->param3 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Retain);
+ break;
+ case ISCSI_ASYNC_MSG_PARAM_NEGOTIATION:
+ hdr->param1 = 0;
+ hdr->param2 = 0;
+ hdr->param3 = cpu_to_be16(SECONDS_FOR_ASYNC_TEXT);
+ break;
+ case ISCSI_ASYNC_MSG_VENDOR_SPECIFIC:
+ printk(KERN_ERR "ISCSI_ASYNC_MSG_VENDOR_SPECIFIC not"
+ " supported yet.\n");
+ return -1;
+ default:
+ printk(KERN_ERR "Unknown AsycnEvent 0x%02x, protocol"
+ " error.\n", async_event);
+ return -1;
+ }
+
+ iov.iov_base = &iscsi_hdr;
+ iov.iov_len = ISCSI_HDR_LEN;
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&iscsi_hdr[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)&iscsi_hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ iov.iov_len += CRC_LEN;
+ tx_send += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest for Async"
+ " Msg PDU 0x%08x\n", *header_digest);
+ }
+
+ TRACE(TRACE_ISCSI, "Built Async Message StatSN: 0x%08x, AsyncEvent:"
+ " 0x%02x, P1: 0x%04x, P2: 0x%04x, P3: 0x%04x\n",
+ ntohl(hdr->statsn), hdr->async_event, ntohs(hdr->param1),
+ ntohs(hdr->param2), ntohs(hdr->param3));
+
+ tx_sent = tx_data(conn, &iov, 1, tx_send);
+ if (tx_sent != tx_send) {
+ printk(KERN_ERR "tx_data returned %d expecting %d\n",
+ tx_sent, tx_send);
+ return -1;
+ }
+
+ if (async_event == ISCSI_ASYNC_MSG_REQUEST_LOGOUT) {
+ wait_for_completion_timeout(&conn->sess->async_msg_comp,
+ SECONDS_FOR_ASYNC_LOGOUT * HZ);
+
+ if (conn->conn_state == TARG_CONN_STATE_LOGOUT_REQUESTED) {
+ printk(KERN_ERR "Asynchronous message timer expired"
+ " without receiving a logout request, dropping"
+ " iSCSI session.\n");
+ iscsit_send_async_msg(conn, 0,
+ ISCSI_ASYNC_MSG_DROPPING_ALL_CONNECTIONS, 0);
+ iscsit_free_session(conn->sess);
+ }
+ }
+ return 0;
+}
+
+/*
+ * Called with sess->conn_lock held.
+ */
+/* #warning iscsi_build_conn_drop_async_message() only sends out on connections
+ with active network interface */
+static void iscsit_build_conn_drop_async_message(struct iscsi_conn *conn)
+{
+ struct iscsi_cmd *cmd;
+ struct iscsi_conn *conn_p;
+
+ /*
+ * Only send a Asynchronous Message on connections whos network
+ * interface is still functional.
+ */
+ list_for_each_entry(conn_p, &conn->sess->sess_conn_list, conn_list) {
+ if (conn_p->conn_state == TARG_CONN_STATE_LOGGED_IN) {
+ iscsit_inc_conn_usage_count(conn_p);
+ break;
+ }
+ }
+
+ if (!conn_p)
+ return;
+
+ cmd = iscsit_allocate_cmd(conn_p);
+ if (!cmd) {
+ iscsit_dec_conn_usage_count(conn_p);
+ return;
+ }
+
+ cmd->logout_cid = conn->cid;
+ cmd->iscsi_opcode = ISCSI_OP_ASYNC_EVENT;
+ cmd->i_state = ISTATE_SEND_ASYNCMSG;
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn_p->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ iscsit_add_cmd_to_response_queue(cmd, conn_p, cmd->i_state);
+ iscsit_dec_conn_usage_count(conn_p);
+}
+
+static int iscsit_send_conn_drop_async_message(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ struct iscsi_async *hdr;
+ struct scatterlist sg;
+
+ cmd->tx_size = ISCSI_HDR_LEN;
+ cmd->iscsi_opcode = ISCSI_OP_ASYNC_EVENT;
+
+ hdr = (struct iscsi_async *) cmd->pdu;
+ hdr->opcode = ISCSI_OP_ASYNC_EVENT;
+ hdr->flags = ISCSI_FLAG_CMD_FINAL;
+ cmd->init_task_tag = 0xFFFFFFFF;
+ cmd->targ_xfer_tag = 0xFFFFFFFF;
+ put_unaligned_be64(0xffffffffffffffff, &hdr->rsvd4[0]);
+ cmd->stat_sn = conn->stat_sn++;
+ hdr->statsn = cpu_to_be32(cmd->stat_sn);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+ hdr->async_event = ISCSI_ASYNC_MSG_DROPPING_CONNECTION;
+ hdr->param1 = cpu_to_be16(cmd->logout_cid);
+ hdr->param2 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Wait);
+ hdr->param3 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Retain);
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ cmd->tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32C HeaderDigest to"
+ " Async Message 0x%08x\n", *header_digest);
+ }
+
+ cmd->iov_misc[0].iov_base = cmd->pdu;
+ cmd->iov_misc[0].iov_len = cmd->tx_size;
+ cmd->iov_misc_count = 1;
+
+ TRACE(TRACE_ERL2, "Sending Connection Dropped Async Message StatSN:"
+ " 0x%08x, for CID: %hu on CID: %hu\n", cmd->stat_sn,
+ cmd->logout_cid, conn->cid);
+ return 0;
+}
+
+static inline int iscsit_send_data_in(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn,
+ struct se_unmap_sg *unmap_sg,
+ int *eodr)
+{
+ int iov_ret = 0, set_statsn = 0;
+ u8 *pad_bytes;
+ u32 iov_count = 0, tx_size = 0;
+ struct iscsi_datain datain;
+ struct iscsi_datain_req *dr;
+ struct se_map_sg map_sg;
+ struct iscsi_data_rsp *hdr;
+ struct kvec *iov;
+ struct scatterlist sg;
+
+ memset(&datain, 0, sizeof(struct iscsi_datain));
+ dr = iscsit_get_datain_values(cmd, &datain);
+ if (!dr) {
+ printk(KERN_ERR "iscsit_get_datain_values failed for ITT: 0x%08x\n",
+ cmd->init_task_tag);
+ return -1;
+ }
+
+ /*
+ * Be paranoid and double check the logic for now.
+ */
+ if ((datain.offset + datain.length) > cmd->data_length) {
+ printk(KERN_ERR "Command ITT: 0x%08x, datain.offset: %u and"
+ " datain.length: %u exceeds cmd->data_length: %u\n",
+ cmd->init_task_tag, datain.offset, datain.length,
+ cmd->data_length);
+ return -1;
+ }
+
+ spin_lock_bh(&conn->sess->session_stats_lock);
+ conn->sess->tx_data_octets += datain.length;
+ if (conn->sess->se_sess->se_node_acl) {
+ spin_lock(&conn->sess->se_sess->se_node_acl->stats_lock);
+ conn->sess->se_sess->se_node_acl->read_bytes += datain.length;
+ spin_unlock(&conn->sess->se_sess->se_node_acl->stats_lock);
+ }
+ spin_unlock_bh(&conn->sess->session_stats_lock);
+ /*
+ * Special case for successfully execution w/ both DATAIN
+ * and Sense Data.
+ */
+ if ((datain.flags & ISCSI_FLAG_DATA_STATUS) &&
+ (SE_CMD(cmd)->se_cmd_flags & SCF_TRANSPORT_TASK_SENSE))
+ datain.flags &= ~ISCSI_FLAG_DATA_STATUS;
+ else {
+ if ((dr->dr_complete == DATAIN_COMPLETE_NORMAL) ||
+ (dr->dr_complete == DATAIN_COMPLETE_CONNECTION_RECOVERY)) {
+ iscsit_increment_maxcmdsn(cmd, conn->sess);
+ cmd->stat_sn = conn->stat_sn++;
+ set_statsn = 1;
+ } else if (dr->dr_complete ==
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY)
+ set_statsn = 1;
+ }
+
+ hdr = (struct iscsi_data_rsp *) cmd->pdu;
+ memset(hdr, 0, ISCSI_HDR_LEN);
+ hdr->opcode = ISCSI_OP_SCSI_DATA_IN;
+ hdr->flags = datain.flags;
+ if (hdr->flags & ISCSI_FLAG_DATA_STATUS) {
+ if (SE_CMD(cmd)->se_cmd_flags & SCF_OVERFLOW_BIT) {
+ hdr->flags |= ISCSI_FLAG_DATA_OVERFLOW;
+ hdr->residual_count = cpu_to_be32(cmd->residual_count);
+ } else if (SE_CMD(cmd)->se_cmd_flags & SCF_UNDERFLOW_BIT) {
+ hdr->flags |= ISCSI_FLAG_DATA_UNDERFLOW;
+ hdr->residual_count = cpu_to_be32(cmd->residual_count);
+ }
+ }
+ hton24(hdr->dlength, datain.length);
+ if (hdr->flags & ISCSI_FLAG_DATA_ACK)
+ int_to_scsilun(SE_CMD(cmd)->orig_fe_lun,
+ (struct scsi_lun *)&hdr->lun[0]);
+ else
+ put_unaligned_le64(0xFFFFFFFFFFFFFFFFULL, &hdr->lun[0]);
+
+ hdr->itt = cpu_to_be32(cmd->init_task_tag);
+ hdr->ttt = (hdr->flags & ISCSI_FLAG_DATA_ACK) ?
+ cpu_to_be32(cmd->targ_xfer_tag) :
+ 0xFFFFFFFF;
+ hdr->statsn = (set_statsn) ? cpu_to_be32(cmd->stat_sn) :
+ 0xFFFFFFFF;
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+ hdr->datasn = cpu_to_be32(datain.data_sn);
+ hdr->offset = cpu_to_be32(datain.offset);
+
+ iov = &cmd->iov_data[0];
+ iov[iov_count].iov_base = cmd->pdu;
+ iov[iov_count++].iov_len = ISCSI_HDR_LEN;
+ tx_size += ISCSI_HDR_LEN;
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ iov[0].iov_len += CRC_LEN;
+ tx_size += CRC_LEN;
+
+ TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest"
+ " for DataIN PDU 0x%08x\n", *header_digest);
+ }
+
+ memset(&map_sg, 0, sizeof(struct se_map_sg));
+ map_sg.fabric_cmd = (void *)cmd;
+ map_sg.se_cmd = SE_CMD(cmd);
+ map_sg.sg_kmap_active = 1;
+ map_sg.iov = &cmd->iov_data[1];
+ map_sg.data_length = datain.length;
+ map_sg.data_offset = datain.offset;
+
+ iov_ret = iscsit_set_iovec_ptrs(&map_sg, unmap_sg);
+ if (iov_ret < 0)
+ return -1;
+
+ iov_count += iov_ret;
+ tx_size += datain.length;
+
+ unmap_sg->padding = ((-datain.length) & 3);
+ if (unmap_sg->padding != 0) {
+ pad_bytes = kzalloc(unmap_sg->padding * sizeof(u8),
+ GFP_KERNEL);
+ if (!pad_bytes) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " pad_bytes.\n");
+ return -1;
+ }
+ cmd->buf_ptr = pad_bytes;
+ iov[iov_count].iov_base = pad_bytes;
+ iov[iov_count++].iov_len = unmap_sg->padding;
+ tx_size += unmap_sg->padding;
+
+ TRACE(TRACE_ISCSI, "Attaching %u padding bytes\n",
+ unmap_sg->padding);
+ }
+ if (conn->conn_ops->DataDigest) {
+ u32 counter = (datain.length + unmap_sg->padding);
+ struct kvec *iov_ptr = &cmd->iov_data[1];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ while (counter > 0) {
+ sg_init_one(&sg, iov_ptr->iov_base,
+ iov_ptr->iov_len);
+ crypto_hash_update(&conn->conn_tx_hash, &sg,
+ iov_ptr->iov_len);
+
+ TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %zu"
+ " bytes, crc 0x%08x\n", iov_ptr->iov_len,
+ cmd->data_crc);
+ counter -= iov_ptr->iov_len;
+ iov_ptr++;
+ }
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)&cmd->data_crc);
+
+ iov[iov_count].iov_base = &cmd->data_crc;
+ iov[iov_count++].iov_len = CRC_LEN;
+ tx_size += CRC_LEN;
+
+ TRACE(TRACE_DIGEST, "Attached CRC32C DataDigest %d bytes, crc"
+ " 0x%08x\n", datain.length+unmap_sg->padding,
+ cmd->data_crc);
+ }
+
+ cmd->iov_data_count = iov_count;
+ cmd->tx_size = tx_size;
+
+ TRACE(TRACE_ISCSI, "Built DataIN ITT: 0x%08x, StatSN: 0x%08x,"
+ " DataSN: 0x%08x, Offset: %u, Length: %u, CID: %hu\n",
+ cmd->init_task_tag, ntohl(hdr->statsn), ntohl(hdr->datasn),
+ ntohl(hdr->offset), datain.length, conn->cid);
+
+ if (dr->dr_complete) {
+ *eodr = (SE_CMD(cmd)->se_cmd_flags & SCF_TRANSPORT_TASK_SENSE) ?
+ 2 : 1;
+ iscsit_free_datain_req(cmd, dr);
+ }
+
+ return 0;
+}
+
+static inline int iscsit_send_logout_response(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ int niov = 0, tx_size;
+ struct iscsi_conn *logout_conn = NULL;
+ struct iscsi_conn_recovery *cr = NULL;
+ struct iscsi_session *sess = conn->sess;
+ struct kvec *iov;
+ struct iscsi_logout_rsp *hdr;
+ struct scatterlist sg;
+ /*
+ * The actual shutting down of Sessions and/or Connections
+ * for CLOSESESSION and CLOSECONNECTION Logout Requests
+ * is done in scsi_logout_post_handler().
+ */
+ switch (cmd->logout_reason) {
+ case ISCSI_LOGOUT_REASON_CLOSE_SESSION:
+ TRACE(TRACE_ISCSI, "iSCSI session logout successful, setting"
+ " logout response to ISCSI_LOGOUT_SUCCESS.\n");
+ cmd->logout_response = ISCSI_LOGOUT_SUCCESS;
+ break;
+ case ISCSI_LOGOUT_REASON_CLOSE_CONNECTION:
+ if (cmd->logout_response == ISCSI_LOGOUT_CID_NOT_FOUND)
+ break;
+ /*
+ * For CLOSECONNECTION logout requests carrying
+ * a matching logout CID -> local CID, the reference
+ * for the local CID will have been incremented in
+ * iscsi_logout_closeconnection().
+ *
+ * For CLOSECONNECTION logout requests carrying
+ * a different CID than the connection it arrived
+ * on, the connection responding to cmd->logout_cid
+ * is stopped in iscsit_logout_post_handler_diffcid().
+ */
+
+ TRACE(TRACE_ISCSI, "iSCSI CID: %hu logout on CID: %hu"
+ " successful.\n", cmd->logout_cid, conn->cid);
+ cmd->logout_response = ISCSI_LOGOUT_SUCCESS;
+ break;
+ case ISCSI_LOGOUT_REASON_RECOVERY:
+ if ((cmd->logout_response == ISCSI_LOGOUT_RECOVERY_UNSUPPORTED) ||
+ (cmd->logout_response == ISCSI_LOGOUT_CLEANUP_FAILED))
+ break;
+ /*
+ * If the connection is still active from our point of view
+ * force connection recovery to occur.
+ */
+ logout_conn = iscsit_get_conn_from_cid_rcfr(sess,
+ cmd->logout_cid);
+ if ((logout_conn)) {
+ iscsit_connection_reinstatement_rcfr(logout_conn);
+ iscsit_dec_conn_usage_count(logout_conn);
+ }
+
+ cr = iscsit_get_inactive_connection_recovery_entry(
+ conn->sess, cmd->logout_cid);
+ if (!cr) {
+ printk(KERN_ERR "Unable to locate CID: %hu for"
+ " REMOVECONNFORRECOVERY Logout Request.\n",
+ cmd->logout_cid);
+ cmd->logout_response = ISCSI_LOGOUT_CID_NOT_FOUND;
+ break;
+ }
+
+ iscsit_discard_cr_cmds_by_expstatsn(cr, cmd->exp_stat_sn);
+
+ TRACE(TRACE_ERL2, "iSCSI REMOVECONNFORRECOVERY logout"
+ " for recovery for CID: %hu on CID: %hu successful.\n",
+ cmd->logout_cid, conn->cid);
+ cmd->logout_response = ISCSI_LOGOUT_SUCCESS;
+ break;
+ default:
+ printk(KERN_ERR "Unknown cmd->logout_reason: 0x%02x\n",
+ cmd->logout_reason);
+ return -1;
+ }
+
+ tx_size = ISCSI_HDR_LEN;
+ hdr = (struct iscsi_logout_rsp *)cmd->pdu;
+ memset(hdr, 0, ISCSI_HDR_LEN);
+ hdr->opcode = ISCSI_OP_LOGOUT_RSP;
+ hdr->flags |= ISCSI_FLAG_CMD_FINAL;
+ hdr->response = cmd->logout_response;
+ hdr->itt = cpu_to_be32(cmd->init_task_tag);
+ cmd->stat_sn = conn->stat_sn++;
+ hdr->statsn = cpu_to_be32(cmd->stat_sn);
+
+ iscsit_increment_maxcmdsn(cmd, conn->sess);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+
+ iov = &cmd->iov_misc[0];
+ iov[niov].iov_base = cmd->pdu;
+ iov[niov++].iov_len = ISCSI_HDR_LEN;
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ iov[0].iov_len += CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32C HeaderDigest to"
+ " Logout Response 0x%08x\n", *header_digest);
+ }
+ cmd->iov_misc_count = niov;
+ cmd->tx_size = tx_size;
+
+ TRACE(TRACE_ISCSI, "Sending Logout Response ITT: 0x%08x StatSN:"
+ " 0x%08x Response: 0x%02x CID: %hu on CID: %hu\n",
+ cmd->init_task_tag, cmd->stat_sn, hdr->response,
+ cmd->logout_cid, conn->cid);
+
+ return 0;
+}
+
+/*
+ * Unsolicited NOPIN, either requesting a response or not.
+ */
+static inline int iscsit_send_unsolicited_nopin(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn,
+ int want_response)
+{
+ int tx_size = ISCSI_HDR_LEN;
+ struct iscsi_nopin *hdr;
+ struct scatterlist sg;
+
+ hdr = (struct iscsi_nopin *) cmd->pdu;
+ memset(hdr, 0, ISCSI_HDR_LEN);
+ hdr->opcode = ISCSI_OP_NOOP_IN;
+ hdr->flags |= ISCSI_FLAG_CMD_FINAL;
+ hdr->itt = cpu_to_be32(cmd->init_task_tag);
+ hdr->ttt = cpu_to_be32(cmd->targ_xfer_tag);
+ cmd->stat_sn = conn->stat_sn;
+ hdr->statsn = cpu_to_be32(cmd->stat_sn);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32C HeaderDigest to"
+ " NopIN 0x%08x\n", *header_digest);
+ }
+
+ cmd->iov_misc[0].iov_base = cmd->pdu;
+ cmd->iov_misc[0].iov_len = tx_size;
+ cmd->iov_misc_count = 1;
+ cmd->tx_size = tx_size;
+
+ TRACE(TRACE_ISCSI, "Sending Unsolicited NOPIN TTT: 0x%08x StatSN:"
+ " 0x%08x CID: %hu\n", hdr->ttt, cmd->stat_sn, conn->cid);
+
+ return 0;
+}
+
+static inline int iscsit_send_nopin_response(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ int niov = 0, tx_size;
+ u32 padding = 0;
+ struct kvec *iov;
+ struct iscsi_nopin *hdr;
+ struct scatterlist sg;
+
+ tx_size = ISCSI_HDR_LEN;
+ hdr = (struct iscsi_nopin *) cmd->pdu;
+ memset(hdr, 0, ISCSI_HDR_LEN);
+ hdr->opcode = ISCSI_OP_NOOP_IN;
+ hdr->flags |= ISCSI_FLAG_CMD_FINAL;
+ hton24(hdr->dlength, cmd->buf_ptr_size);
+ put_unaligned_le64(0xFFFFFFFFFFFFFFFFULL, &hdr->lun[0]);
+ hdr->itt = cpu_to_be32(cmd->init_task_tag);
+ hdr->ttt = cpu_to_be32(cmd->targ_xfer_tag);
+ cmd->stat_sn = conn->stat_sn++;
+ hdr->statsn = cpu_to_be32(cmd->stat_sn);
+
+ iscsit_increment_maxcmdsn(cmd, conn->sess);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+
+ iov = &cmd->iov_misc[0];
+ iov[niov].iov_base = cmd->pdu;
+ iov[niov++].iov_len = ISCSI_HDR_LEN;
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ iov[0].iov_len += CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32C HeaderDigest"
+ " to NopIn 0x%08x\n", *header_digest);
+ }
+
+ /*
+ * NOPOUT Ping Data is attached to struct iscsi_cmd->buf_ptr.
+ * NOPOUT DataSegmentLength is at struct iscsi_cmd->buf_ptr_size.
+ */
+ if (cmd->buf_ptr_size) {
+ iov[niov].iov_base = cmd->buf_ptr;
+ iov[niov++].iov_len = cmd->buf_ptr_size;
+ tx_size += cmd->buf_ptr_size;
+
+ TRACE(TRACE_ISCSI, "Echoing back %u bytes of ping"
+ " data.\n", cmd->buf_ptr_size);
+
+ padding = ((-cmd->buf_ptr_size) & 3);
+ if (padding != 0) {
+ iov[niov].iov_base = &cmd->pad_bytes;
+ iov[niov++].iov_len = padding;
+ tx_size += padding;
+ TRACE(TRACE_ISCSI, "Attaching %u additional"
+ " padding bytes.\n", padding);
+ }
+ if (conn->conn_ops->DataDigest) {
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)cmd->buf_ptr,
+ cmd->buf_ptr_size);
+ crypto_hash_update(&conn->conn_tx_hash, &sg,
+ cmd->buf_ptr_size);
+
+ if (padding) {
+ sg_init_one(&sg, (u8 *)&cmd->pad_bytes, padding);
+ crypto_hash_update(&conn->conn_tx_hash, &sg,
+ padding);
+ }
+
+ crypto_hash_final(&conn->conn_tx_hash,
+ (u8 *)&cmd->data_crc);
+
+ iov[niov].iov_base = &cmd->data_crc;
+ iov[niov++].iov_len = CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attached DataDigest for %u"
+ " bytes of ping data, CRC 0x%08x\n",
+ cmd->buf_ptr_size, cmd->data_crc);
+ }
+ }
+
+ cmd->iov_misc_count = niov;
+ cmd->tx_size = tx_size;
+
+ TRACE(TRACE_ISCSI, "Sending NOPIN Response ITT: 0x%08x, TTT:"
+ " 0x%08x, StatSN: 0x%08x, Length %u\n", cmd->init_task_tag,
+ cmd->targ_xfer_tag, cmd->stat_sn, cmd->buf_ptr_size);
+
+ return 0;
+}
+
+int iscsit_send_r2t(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ int tx_size = 0;
+ u32 trace_type;
+ struct iscsi_r2t *r2t;
+ struct iscsi_r2t_rsp *hdr;
+ struct scatterlist sg;
+
+ r2t = iscsit_get_r2t_from_list(cmd);
+ if (!r2t)
+ return -1;
+
+ hdr = (struct iscsi_r2t_rsp *) cmd->pdu;
+ memset(hdr, 0, ISCSI_HDR_LEN);
+ hdr->opcode = ISCSI_OP_R2T;
+ hdr->flags |= ISCSI_FLAG_CMD_FINAL;
+ int_to_scsilun(SE_CMD(cmd)->orig_fe_lun,
+ (struct scsi_lun *)&hdr->lun[0]);
+ hdr->itt = cpu_to_be32(cmd->init_task_tag);
+ spin_lock_bh(&conn->sess->ttt_lock);
+ r2t->targ_xfer_tag = conn->sess->targ_xfer_tag++;
+ if (r2t->targ_xfer_tag == 0xFFFFFFFF)
+ r2t->targ_xfer_tag = conn->sess->targ_xfer_tag++;
+ spin_unlock_bh(&conn->sess->ttt_lock);
+ hdr->ttt = cpu_to_be32(r2t->targ_xfer_tag);
+ hdr->statsn = cpu_to_be32(conn->stat_sn);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+ hdr->r2tsn = cpu_to_be32(r2t->r2t_sn);
+ hdr->data_offset = cpu_to_be32(r2t->offset);
+ hdr->data_length = cpu_to_be32(r2t->xfer_len);
+
+ cmd->iov_misc[0].iov_base = cmd->pdu;
+ cmd->iov_misc[0].iov_len = ISCSI_HDR_LEN;
+ tx_size += ISCSI_HDR_LEN;
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ cmd->iov_misc[0].iov_len += CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest for R2T"
+ " PDU 0x%08x\n", *header_digest);
+ }
+
+ trace_type = (!r2t->recovery_r2t) ? TRACE_ISCSI : TRACE_ERL1;
+ TRACE(trace_type, "Built %sR2T, ITT: 0x%08x, TTT: 0x%08x, StatSN:"
+ " 0x%08x, R2TSN: 0x%08x, Offset: %u, DDTL: %u, CID: %hu\n",
+ (!r2t->recovery_r2t) ? "" : "Recovery ", cmd->init_task_tag,
+ r2t->targ_xfer_tag, ntohl(hdr->statsn), r2t->r2t_sn,
+ r2t->offset, r2t->xfer_len, conn->cid);
+
+ cmd->iov_misc_count = 1;
+ cmd->tx_size = tx_size;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ r2t->sent_r2t = 1;
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ return 0;
+}
+
+/*
+ * type 0: Normal Operation.
+ * type 1: Called from Storage Transport.
+ * type 2: Called from iscsi_task_reassign_complete_write() for
+ * connection recovery.
+ */
+int iscsit_build_r2ts_for_cmd(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn,
+ int type)
+{
+ int first_r2t = 1;
+ u32 offset = 0, xfer_len = 0;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ if (cmd->cmd_flags & ICF_SENT_LAST_R2T) {
+ spin_unlock_bh(&cmd->r2t_lock);
+ return 0;
+ }
+
+ if (conn->sess->sess_ops->DataSequenceInOrder && (type != 2))
+ if (cmd->r2t_offset < cmd->write_data_done)
+ cmd->r2t_offset = cmd->write_data_done;
+
+ while (cmd->outstanding_r2ts < conn->sess->sess_ops->MaxOutstandingR2T) {
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ offset = cmd->r2t_offset;
+
+ if (first_r2t && (type == 2)) {
+ xfer_len = ((offset +
+ (conn->sess->sess_ops->MaxBurstLength -
+ cmd->next_burst_len) >
+ cmd->data_length) ?
+ (cmd->data_length - offset) :
+ (conn->sess->sess_ops->MaxBurstLength -
+ cmd->next_burst_len));
+ } else {
+ xfer_len = ((offset +
+ conn->sess->sess_ops->MaxBurstLength) >
+ cmd->data_length) ?
+ (cmd->data_length - offset) :
+ conn->sess->sess_ops->MaxBurstLength;
+ }
+ cmd->r2t_offset += xfer_len;
+
+ if (cmd->r2t_offset == cmd->data_length)
+ cmd->cmd_flags |= ICF_SENT_LAST_R2T;
+ } else {
+ struct iscsi_seq *seq;
+
+ seq = iscsit_get_seq_holder_for_r2t(cmd);
+ if (!seq) {
+ spin_unlock_bh(&cmd->r2t_lock);
+ return -1;
+ }
+
+ offset = seq->offset;
+ xfer_len = seq->xfer_len;
+
+ if (cmd->seq_send_order == cmd->seq_count)
+ cmd->cmd_flags |= ICF_SENT_LAST_R2T;
+ }
+ cmd->outstanding_r2ts++;
+ first_r2t = 0;
+
+ if (iscsit_add_r2t_to_list(cmd, offset, xfer_len, 0, 0) < 0) {
+ spin_unlock_bh(&cmd->r2t_lock);
+ return -1;
+ }
+
+ if (cmd->cmd_flags & ICF_SENT_LAST_R2T)
+ break;
+ }
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ return 0;
+}
+
+static inline int iscsit_send_status(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ u8 iov_count = 0, recovery;
+ u32 padding = 0, trace_type, tx_size = 0;
+ struct iscsi_scsi_rsp *hdr;
+ struct kvec *iov;
+ struct scatterlist sg;
+
+ recovery = (cmd->i_state != ISTATE_SEND_STATUS);
+ if (!recovery)
+ cmd->stat_sn = conn->stat_sn++;
+
+ spin_lock_bh(&conn->sess->session_stats_lock);
+ conn->sess->rsp_pdus++;
+ spin_unlock_bh(&conn->sess->session_stats_lock);
+
+ hdr = (struct iscsi_scsi_rsp *) cmd->pdu;
+ memset(hdr, 0, ISCSI_HDR_LEN);
+ hdr->opcode = ISCSI_OP_SCSI_CMD_RSP;
+ hdr->flags |= ISCSI_FLAG_CMD_FINAL;
+ if (SE_CMD(cmd)->se_cmd_flags & SCF_OVERFLOW_BIT) {
+ hdr->flags |= ISCSI_FLAG_CMD_OVERFLOW;
+ hdr->residual_count = cpu_to_be32(cmd->residual_count);
+ } else if (SE_CMD(cmd)->se_cmd_flags & SCF_UNDERFLOW_BIT) {
+ hdr->flags |= ISCSI_FLAG_CMD_UNDERFLOW;
+ hdr->residual_count = cpu_to_be32(cmd->residual_count);
+ }
+ hdr->response = cmd->iscsi_response;
+ hdr->cmd_status = SE_CMD(cmd)->scsi_status;
+ hdr->itt = cpu_to_be32(cmd->init_task_tag);
+ hdr->statsn = cpu_to_be32(cmd->stat_sn);
+
+ iscsit_increment_maxcmdsn(cmd, conn->sess);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+
+ iov = &cmd->iov_misc[0];
+ iov[iov_count].iov_base = cmd->pdu;
+ iov[iov_count++].iov_len = ISCSI_HDR_LEN;
+ tx_size += ISCSI_HDR_LEN;
+
+ /*
+ * Attach SENSE DATA payload to iSCSI Response PDU
+ */
+ if (SE_CMD(cmd)->sense_buffer &&
+ ((SE_CMD(cmd)->se_cmd_flags & SCF_TRANSPORT_TASK_SENSE) ||
+ (SE_CMD(cmd)->se_cmd_flags & SCF_EMULATED_TASK_SENSE))) {
+ padding = -(SE_CMD(cmd)->scsi_sense_length) & 3;
+ hton24(hdr->dlength, SE_CMD(cmd)->scsi_sense_length);
+ iov[iov_count].iov_base = SE_CMD(cmd)->sense_buffer;
+ iov[iov_count++].iov_len =
+ (SE_CMD(cmd)->scsi_sense_length + padding);
+ tx_size += SE_CMD(cmd)->scsi_sense_length;
+
+ if (padding) {
+ memset(SE_CMD(cmd)->sense_buffer +
+ SE_CMD(cmd)->scsi_sense_length, 0, padding);
+ tx_size += padding;
+ TRACE(TRACE_ISCSI, "Adding %u bytes of padding to"
+ " SENSE.\n", padding);
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)SE_CMD(cmd)->sense_buffer,
+ (SE_CMD(cmd)->scsi_sense_length + padding));
+ crypto_hash_update(&conn->conn_tx_hash, &sg,
+ (SE_CMD(cmd)->scsi_sense_length + padding));
+
+ crypto_hash_final(&conn->conn_tx_hash,
+ (u8 *)&cmd->data_crc);
+
+ iov[iov_count].iov_base = &cmd->data_crc;
+ iov[iov_count++].iov_len = CRC_LEN;
+ tx_size += CRC_LEN;
+
+ TRACE(TRACE_DIGEST, "Attaching CRC32 DataDigest for"
+ " SENSE, %u bytes CRC 0x%08x\n",
+ (SE_CMD(cmd)->scsi_sense_length + padding),
+ cmd->data_crc);
+ }
+
+ TRACE(TRACE_ISCSI, "Attaching SENSE DATA: %u bytes to iSCSI"
+ " Response PDU\n",
+ SE_CMD(cmd)->scsi_sense_length);
+ }
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ iov[0].iov_len += CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest for Response"
+ " PDU 0x%08x\n", *header_digest);
+ }
+
+ cmd->iov_misc_count = iov_count;
+ cmd->tx_size = tx_size;
+
+ trace_type = (!recovery) ? TRACE_ISCSI : TRACE_ERL1;
+ TRACE(trace_type, "Built %sSCSI Response, ITT: 0x%08x, StatSN: 0x%08x,"
+ " Response: 0x%02x, SAM Status: 0x%02x, CID: %hu\n",
+ (!recovery) ? "" : "Recovery ", cmd->init_task_tag,
+ cmd->stat_sn, 0x00, cmd->se_cmd.scsi_status, conn->cid);
+
+ return 0;
+}
+
+static inline u8 iscsit_convert_tcm_tmr_rsp(struct se_tmr_req *se_tmr)
+{
+ switch (se_tmr->response) {
+ case TMR_FUNCTION_COMPLETE:
+ return ISCSI_TMF_RSP_COMPLETE;
+ case TMR_TASK_DOES_NOT_EXIST:
+ return ISCSI_TMF_RSP_NO_TASK;
+ case TMR_LUN_DOES_NOT_EXIST:
+ return ISCSI_TMF_RSP_NO_LUN;
+ case TMR_TASK_MGMT_FUNCTION_NOT_SUPPORTED:
+ return ISCSI_TMF_RSP_NOT_SUPPORTED;
+ case TMR_FUNCTION_AUTHORIZATION_FAILED:
+ return ISCSI_TMF_RSP_AUTH_FAILED;
+ case TMR_FUNCTION_REJECTED:
+ default:
+ return ISCSI_TMF_RSP_REJECTED;
+ }
+}
+
+static int iscsit_send_task_mgt_rsp(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ struct se_tmr_req *se_tmr = SE_CMD(cmd)->se_tmr_req;
+ struct iscsi_tm_rsp *hdr;
+ struct scatterlist sg;
+ u32 tx_size = 0;
+
+ hdr = (struct iscsi_tm_rsp *) cmd->pdu;
+ memset(hdr, 0, ISCSI_HDR_LEN);
+ hdr->opcode = ISCSI_OP_SCSI_TMFUNC_RSP;
+ hdr->response = iscsit_convert_tcm_tmr_rsp(se_tmr);
+ hdr->itt = cpu_to_be32(cmd->init_task_tag);
+ cmd->stat_sn = conn->stat_sn++;
+ hdr->statsn = cpu_to_be32(cmd->stat_sn);
+
+ iscsit_increment_maxcmdsn(cmd, conn->sess);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+
+ cmd->iov_misc[0].iov_base = cmd->pdu;
+ cmd->iov_misc[0].iov_len = ISCSI_HDR_LEN;
+ tx_size += ISCSI_HDR_LEN;
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ cmd->iov_misc[0].iov_len += CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest for Task"
+ " Mgmt Response PDU 0x%08x\n", *header_digest);
+ }
+
+ cmd->iov_misc_count = 1;
+ cmd->tx_size = tx_size;
+
+ TRACE(TRACE_ERL2, "Built Task Management Response ITT: 0x%08x,"
+ " StatSN: 0x%08x, Response: 0x%02x, CID: %hu\n",
+ cmd->init_task_tag, cmd->stat_sn, hdr->response, conn->cid);
+
+ return 0;
+}
+
+/*
+ *
+ * FIXME: Add support for F_BIT and C_BIT when the length is longer than
+ * MaxRecvDataSegmentLength.
+ */
+static int iscsit_send_text_rsp(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ u8 iov_count = 0;
+ u32 padding = 0, text_length = 0, tx_size = 0;
+ struct iscsi_text_rsp *hdr;
+ struct kvec *iov;
+ struct scatterlist sg;
+
+ text_length = iscsit_build_sendtargets_response(cmd);
+
+ padding = ((-text_length) & 3);
+ if (padding != 0) {
+ memset(cmd->buf_ptr + text_length, 0, padding);
+ TRACE(TRACE_ISCSI, "Attaching %u additional bytes for"
+ " padding.\n", padding);
+ }
+
+ hdr = (struct iscsi_text_rsp *) cmd->pdu;
+ memset(hdr, 0, ISCSI_HDR_LEN);
+ hdr->opcode = ISCSI_OP_TEXT_RSP;
+ hdr->flags |= ISCSI_FLAG_CMD_FINAL;
+ hton24(hdr->dlength, text_length);
+ hdr->itt = cpu_to_be32(cmd->init_task_tag);
+ hdr->ttt = cpu_to_be32(cmd->targ_xfer_tag);
+ cmd->stat_sn = conn->stat_sn++;
+ hdr->statsn = cpu_to_be32(cmd->stat_sn);
+
+ iscsit_increment_maxcmdsn(cmd, conn->sess);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+
+ iov = &cmd->iov_misc[0];
+
+ iov[iov_count].iov_base = cmd->pdu;
+ iov[iov_count++].iov_len = ISCSI_HDR_LEN;
+ iov[iov_count].iov_base = cmd->buf_ptr;
+ iov[iov_count++].iov_len = text_length + padding;
+
+ tx_size += (ISCSI_HDR_LEN + text_length + padding);
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ iov[0].iov_len += CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest for"
+ " Text Response PDU 0x%08x\n", *header_digest);
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)cmd->buf_ptr, (text_length + padding));
+ crypto_hash_update(&conn->conn_tx_hash, &sg,
+ (text_length + padding));
+
+ crypto_hash_final(&conn->conn_tx_hash,
+ (u8 *)&cmd->data_crc);
+
+ iov[iov_count].iov_base = &cmd->data_crc;
+ iov[iov_count++].iov_len = CRC_LEN;
+ tx_size += CRC_LEN;
+
+ TRACE(TRACE_DIGEST, "Attaching DataDigest for %u bytes of text"
+ " data, CRC 0x%08x\n", (text_length + padding),
+ cmd->data_crc);
+ }
+
+ cmd->iov_misc_count = iov_count;
+ cmd->tx_size = tx_size;
+
+ TRACE(TRACE_ISCSI, "Built Text Response: ITT: 0x%08x, StatSN: 0x%08x,"
+ " Length: %u, CID: %hu\n", cmd->init_task_tag, cmd->stat_sn,
+ text_length, conn->cid);
+ return 0;
+}
+
+static int iscsit_send_reject(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ u32 iov_count = 0, tx_size = 0;
+ struct iscsi_reject *hdr;
+ struct kvec *iov;
+ struct scatterlist sg;
+
+ hdr = (struct iscsi_reject *) cmd->pdu;
+ hdr->opcode = ISCSI_OP_REJECT;
+ hdr->flags |= ISCSI_FLAG_CMD_FINAL;
+ hton24(hdr->dlength, ISCSI_HDR_LEN);
+ cmd->stat_sn = conn->stat_sn++;
+ hdr->statsn = cpu_to_be32(cmd->stat_sn);
+ hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+
+ iov = &cmd->iov_misc[0];
+
+ iov[iov_count].iov_base = cmd->pdu;
+ iov[iov_count++].iov_len = ISCSI_HDR_LEN;
+ iov[iov_count].iov_base = cmd->buf_ptr;
+ iov[iov_count++].iov_len = ISCSI_HDR_LEN;
+
+ tx_size = (ISCSI_HDR_LEN + ISCSI_HDR_LEN);
+
+ if (conn->conn_ops->HeaderDigest) {
+ u32 *header_digest = (u32 *)&cmd->pdu[ISCSI_HDR_LEN];
+
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)hdr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg, ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
+
+ iov[0].iov_len += CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest for"
+ " REJECT PDU 0x%08x\n", *header_digest);
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ crypto_hash_init(&conn->conn_tx_hash);
+
+ sg_init_one(&sg, (u8 *)cmd->buf_ptr, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_tx_hash, &sg,
+ ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_tx_hash,
+ (u8 *)&cmd->data_crc);
+
+ iov[iov_count].iov_base = &cmd->data_crc;
+ iov[iov_count++].iov_len = CRC_LEN;
+ tx_size += CRC_LEN;
+ TRACE(TRACE_DIGEST, "Attaching CRC32 DataDigest for REJECT"
+ " PDU 0x%08x\n", cmd->data_crc);
+ }
+
+ cmd->iov_misc_count = iov_count;
+ cmd->tx_size = tx_size;
+
+ TRACE(TRACE_ISCSI, "Built Reject PDU StatSN: 0x%08x, Reason: 0x%02x,"
+ " CID: %hu\n", ntohl(hdr->statsn), hdr->reason, conn->cid);
+
+ return 0;
+}
+
+static void iscsit_tx_thread_wait_for_tcp(struct iscsi_conn *conn)
+{
+ if ((conn->sock->sk->sk_shutdown & SEND_SHUTDOWN) ||
+ (conn->sock->sk->sk_shutdown & RCV_SHUTDOWN)) {
+ wait_for_completion_interruptible_timeout(
+ &conn->tx_half_close_comp,
+ ISCSI_TX_THREAD_TCP_TIMEOUT * HZ);
+ }
+}
+
+#ifdef CONFIG_SMP
+
+void iscsit_thread_get_cpumask(struct iscsi_conn *conn)
+{
+ struct iscsi_thread_set *ts = conn->thread_set;
+ int ord, cpu;
+ /*
+ * thread_id is assigned from iscsi_global->ts_bitmap from
+ * within iscsi_thread_set.c:iscsi_allocate_thread_sets()
+ *
+ * Here we use thread_id to determine which CPU that this
+ * iSCSI connection's iscsi_thread_set will be scheduled to
+ * execute upon.
+ */
+ ord = ts->thread_id % cpumask_weight(cpu_online_mask);
+#if 0
+ printk(KERN_INFO ">>>>>>>>>>>>>>>>>>>> Generated ord: %d from"
+ " thread_id: %d\n", ord, ts->thread_id);
+#endif
+ for_each_online_cpu(cpu) {
+ if (ord-- == 0) {
+ cpumask_set_cpu(cpu, conn->conn_cpumask);
+ return;
+ }
+ }
+ /*
+ * This should never be reached..
+ */
+ dump_stack();
+ cpumask_setall(conn->conn_cpumask);
+}
+
+static inline void iscsit_thread_check_cpumask(
+ struct iscsi_conn *conn,
+ struct task_struct *p,
+ int mode)
+{
+ char buf[128];
+ /*
+ * mode == 1 signals iscsi_target_tx_thread() usage.
+ * mode == 0 signals iscsi_target_rx_thread() usage.
+ */
+ if (mode == 1) {
+ if (!conn->conn_tx_reset_cpumask)
+ return;
+ conn->conn_tx_reset_cpumask = 0;
+ } else {
+ if (!conn->conn_rx_reset_cpumask)
+ return;
+ conn->conn_rx_reset_cpumask = 0;
+ }
+ /*
+ * Update the CPU mask for this single kthread so that
+ * both TX and RX kthreads are scheduled to run on the
+ * same CPU.
+ */
+ memset(buf, 0, 128);
+ cpumask_scnprintf(buf, 128, conn->conn_cpumask);
+#if 0
+ printk(KERN_INFO ">>>>>>>>>>>>>> Calling set_cpus_allowed_ptr():"
+ " %s for %s\n", buf, p->comm);
+#endif
+ set_cpus_allowed_ptr(p, conn->conn_cpumask);
+}
+
+#else
+#define iscsit_thread_get_cpumask(X) ({})
+#define iscsit_thread_check_cpumask(X, Y, Z) ({})
+#endif /* CONFIG_SMP */
+
+int iscsi_target_tx_thread(void *arg)
+{
+ u8 state;
+ int eodr = 0, map_sg = 0, ret = 0, sent_status = 0, use_misc = 0;
+ struct iscsi_cmd *cmd = NULL;
+ struct iscsi_conn *conn;
+ struct iscsi_queue_req *qr = NULL;
+ struct se_cmd *se_cmd;
+ struct iscsi_thread_set *ts = (struct iscsi_thread_set *)arg;
+ struct se_unmap_sg unmap_sg;
+ /*
+ * Bump up the task_struct priority for RX/TX thread set pairs,
+ * and allow ourselves to be interrupted by SIGINT so that a
+ * connection recovery / failure event can be triggered externally.
+ */
+ set_user_nice(current, -20);
+ allow_signal(SIGINT);
+
+restart:
+ conn = iscsi_tx_thread_pre_handler(ts);
+ if (!conn)
+ goto out;
+
+ eodr = map_sg = ret = sent_status = use_misc = 0;
+
+ while (!kthread_should_stop()) {
+ /*
+ * Ensure that both TX and RX per connection kthreads
+ * are scheduled to run on the same CPU.
+ */
+ iscsit_thread_check_cpumask(conn, current, 1);
+
+ schedule_timeout_interruptible(MAX_SCHEDULE_TIMEOUT);
+
+ if ((ts->status == ISCSI_THREAD_SET_RESET) ||
+ signal_pending(current))
+ goto transport_err;
+
+get_immediate:
+ qr = iscsit_get_cmd_from_immediate_queue(conn);
+ if (qr) {
+ atomic_set(&conn->check_immediate_queue, 0);
+ cmd = qr->cmd;
+ state = qr->state;
+ kmem_cache_free(lio_qr_cache, qr);
+
+ spin_lock_bh(&cmd->istate_lock);
+ switch (state) {
+ case ISTATE_SEND_R2T:
+ spin_unlock_bh(&cmd->istate_lock);
+ ret = iscsit_send_r2t(cmd, conn);
+ break;
+ case ISTATE_REMOVE:
+ spin_unlock_bh(&cmd->istate_lock);
+
+ if (cmd->data_direction == DMA_TO_DEVICE)
+ iscsit_stop_dataout_timer(cmd);
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_del(&cmd->i_list);
+ spin_unlock_bh(&conn->cmd_lock);
+ /*
+ * Determine if a struct se_cmd is assoicated with
+ * this struct iscsi_cmd.
+ */
+ if (!(SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD) &&
+ !(cmd->tmr_req))
+ iscsit_release_cmd(cmd);
+ else
+ transport_generic_free_cmd(SE_CMD(cmd),
+ 1, 1, 0);
+ goto get_immediate;
+ case ISTATE_SEND_NOPIN_WANT_RESPONSE:
+ spin_unlock_bh(&cmd->istate_lock);
+ iscsit_mod_nopin_response_timer(conn);
+ ret = iscsit_send_unsolicited_nopin(cmd,
+ conn, 1);
+ break;
+ case ISTATE_SEND_NOPIN_NO_RESPONSE:
+ spin_unlock_bh(&cmd->istate_lock);
+ ret = iscsit_send_unsolicited_nopin(cmd,
+ conn, 0);
+ break;
+ default:
+ printk(KERN_ERR "Unknown Opcode: 0x%02x ITT:"
+ " 0x%08x, i_state: %d on CID: %hu\n",
+ cmd->iscsi_opcode, cmd->init_task_tag, state,
+ conn->cid);
+ spin_unlock_bh(&cmd->istate_lock);
+ goto transport_err;
+ }
+ if (ret < 0) {
+ conn->tx_immediate_queue = 0;
+ goto transport_err;
+ }
+
+ if (iscsit_send_tx_data(cmd, conn, 1) < 0) {
+ conn->tx_immediate_queue = 0;
+ iscsit_tx_thread_wait_for_tcp(conn);
+ goto transport_err;
+ }
+
+ spin_lock_bh(&cmd->istate_lock);
+ switch (state) {
+ case ISTATE_SEND_R2T:
+ spin_unlock_bh(&cmd->istate_lock);
+ spin_lock_bh(&cmd->dataout_timeout_lock);
+ iscsit_start_dataout_timer(cmd, conn);
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ break;
+ case ISTATE_SEND_NOPIN_WANT_RESPONSE:
+ cmd->i_state = ISTATE_SENT_NOPIN_WANT_RESPONSE;
+ spin_unlock_bh(&cmd->istate_lock);
+ break;
+ case ISTATE_SEND_NOPIN_NO_RESPONSE:
+ cmd->i_state = ISTATE_SENT_STATUS;
+ spin_unlock_bh(&cmd->istate_lock);
+ break;
+ default:
+ printk(KERN_ERR "Unknown Opcode: 0x%02x ITT:"
+ " 0x%08x, i_state: %d on CID: %hu\n",
+ cmd->iscsi_opcode, cmd->init_task_tag,
+ state, conn->cid);
+ spin_unlock_bh(&cmd->istate_lock);
+ goto transport_err;
+ }
+ goto get_immediate;
+ } else
+ conn->tx_immediate_queue = 0;
+
+get_response:
+ qr = iscsit_get_cmd_from_response_queue(conn);
+ if (qr) {
+ cmd = qr->cmd;
+ state = qr->state;
+ kmem_cache_free(lio_qr_cache, qr);
+
+ spin_lock_bh(&cmd->istate_lock);
+check_rsp_state:
+ switch (state) {
+ case ISTATE_SEND_DATAIN:
+ spin_unlock_bh(&cmd->istate_lock);
+ memset(&unmap_sg, 0,
+ sizeof(struct se_unmap_sg));
+ unmap_sg.fabric_cmd = (void *)cmd;
+ unmap_sg.se_cmd = SE_CMD(cmd);
+ map_sg = 1;
+ ret = iscsit_send_data_in(cmd, conn,
+ &unmap_sg, &eodr);
+ break;
+ case ISTATE_SEND_STATUS:
+ case ISTATE_SEND_STATUS_RECOVERY:
+ spin_unlock_bh(&cmd->istate_lock);
+ use_misc = 1;
+ ret = iscsit_send_status(cmd, conn);
+ break;
+ case ISTATE_SEND_LOGOUTRSP:
+ spin_unlock_bh(&cmd->istate_lock);
+ use_misc = 1;
+ ret = iscsit_send_logout_response(cmd, conn);
+ break;
+ case ISTATE_SEND_ASYNCMSG:
+ spin_unlock_bh(&cmd->istate_lock);
+ use_misc = 1;
+ ret = iscsit_send_conn_drop_async_message(
+ cmd, conn);
+ break;
+ case ISTATE_SEND_NOPIN:
+ spin_unlock_bh(&cmd->istate_lock);
+ use_misc = 1;
+ ret = iscsit_send_nopin_response(cmd, conn);
+ break;
+ case ISTATE_SEND_REJECT:
+ spin_unlock_bh(&cmd->istate_lock);
+ use_misc = 1;
+ ret = iscsit_send_reject(cmd, conn);
+ break;
+ case ISTATE_SEND_TASKMGTRSP:
+ spin_unlock_bh(&cmd->istate_lock);
+ use_misc = 1;
+ ret = iscsit_send_task_mgt_rsp(cmd, conn);
+ if (ret != 0)
+ break;
+ ret = iscsit_tmr_post_handler(cmd, conn);
+ if (ret != 0)
+ iscsit_fall_back_to_erl0(conn->sess);
+ break;
+ case ISTATE_SEND_TEXTRSP:
+ spin_unlock_bh(&cmd->istate_lock);
+ use_misc = 1;
+ ret = iscsit_send_text_rsp(cmd, conn);
+ break;
+ default:
+ printk(KERN_ERR "Unknown Opcode: 0x%02x ITT:"
+ " 0x%08x, i_state: %d on CID: %hu\n",
+ cmd->iscsi_opcode, cmd->init_task_tag,
+ state, conn->cid);
+ spin_unlock_bh(&cmd->istate_lock);
+ goto transport_err;
+ }
+ if (ret < 0) {
+ conn->tx_response_queue = 0;
+ goto transport_err;
+ }
+
+ se_cmd = &cmd->se_cmd;
+
+ if (map_sg && !conn->conn_ops->IFMarker &&
+ T_TASK(se_cmd)->t_tasks_se_num) {
+ iscsit_map_SG_segments(&unmap_sg);
+ if (iscsit_fe_sendpage_sg(&unmap_sg, conn) < 0) {
+ conn->tx_response_queue = 0;
+ iscsit_tx_thread_wait_for_tcp(conn);
+ iscsit_unmap_SG_segments(&unmap_sg);
+ goto transport_err;
+ }
+ iscsit_unmap_SG_segments(&unmap_sg);
+ map_sg = 0;
+ } else {
+ if (map_sg)
+ iscsit_map_SG_segments(&unmap_sg);
+ if (iscsit_send_tx_data(cmd, conn, use_misc) < 0) {
+ conn->tx_response_queue = 0;
+ iscsit_tx_thread_wait_for_tcp(conn);
+ if (map_sg)
+ iscsit_unmap_SG_segments(&unmap_sg);
+ goto transport_err;
+ }
+ if (map_sg) {
+ iscsit_unmap_SG_segments(&unmap_sg);
+ map_sg = 0;
+ }
+ }
+
+ spin_lock_bh(&cmd->istate_lock);
+ switch (state) {
+ case ISTATE_SEND_DATAIN:
+ if (!eodr)
+ goto check_rsp_state;
+
+ if (eodr == 1) {
+ cmd->i_state = ISTATE_SENT_LAST_DATAIN;
+ sent_status = 1;
+ eodr = use_misc = 0;
+ } else if (eodr == 2) {
+ cmd->i_state = state =
+ ISTATE_SEND_STATUS;
+ sent_status = 0;
+ eodr = use_misc = 0;
+ goto check_rsp_state;
+ }
+ break;
+ case ISTATE_SEND_STATUS:
+ use_misc = 0;
+ sent_status = 1;
+ break;
+ case ISTATE_SEND_ASYNCMSG:
+ case ISTATE_SEND_NOPIN:
+ case ISTATE_SEND_STATUS_RECOVERY:
+ case ISTATE_SEND_TEXTRSP:
+ use_misc = 0;
+ sent_status = 1;
+ break;
+ case ISTATE_SEND_REJECT:
+ use_misc = 0;
+ if (cmd->cmd_flags & ICF_REJECT_FAIL_CONN) {
+ cmd->cmd_flags &= ~ICF_REJECT_FAIL_CONN;
+ spin_unlock_bh(&cmd->istate_lock);
+ complete(&cmd->reject_comp);
+ goto transport_err;
+ }
+ complete(&cmd->reject_comp);
+ break;
+ case ISTATE_SEND_TASKMGTRSP:
+ use_misc = 0;
+ sent_status = 1;
+ break;
+ case ISTATE_SEND_LOGOUTRSP:
+ spin_unlock_bh(&cmd->istate_lock);
+ if (!(iscsit_logout_post_handler(cmd, conn)))
+ goto restart;
+ spin_lock_bh(&cmd->istate_lock);
+ use_misc = 0;
+ sent_status = 1;
+ break;
+ default:
+ printk(KERN_ERR "Unknown Opcode: 0x%02x ITT:"
+ " 0x%08x, i_state: %d on CID: %hu\n",
+ cmd->iscsi_opcode, cmd->init_task_tag,
+ cmd->i_state, conn->cid);
+ spin_unlock_bh(&cmd->istate_lock);
+ goto transport_err;
+ }
+
+ if (sent_status) {
+ cmd->i_state = ISTATE_SENT_STATUS;
+ sent_status = 0;
+ }
+ spin_unlock_bh(&cmd->istate_lock);
+
+ if (atomic_read(&conn->check_immediate_queue))
+ goto get_immediate;
+
+ goto get_response;
+ } else
+ conn->tx_response_queue = 0;
+ }
+
+transport_err:
+ iscsit_take_action_for_connection_exit(conn);
+ goto restart;
+out:
+ return 0;
+}
+
+static void iscsit_rx_thread_wait_for_tcp(struct iscsi_conn *conn)
+{
+ if ((conn->sock->sk->sk_shutdown & SEND_SHUTDOWN) ||
+ (conn->sock->sk->sk_shutdown & RCV_SHUTDOWN)) {
+ wait_for_completion_interruptible_timeout(
+ &conn->rx_half_close_comp,
+ ISCSI_RX_THREAD_TCP_TIMEOUT * HZ);
+ }
+}
+
+int iscsi_target_rx_thread(void *arg)
+{
+ int ret;
+ u8 buffer[ISCSI_HDR_LEN], opcode;
+ u32 checksum = 0, digest = 0;
+ struct iscsi_conn *conn = NULL;
+ struct iscsi_thread_set *ts = (struct iscsi_thread_set *)arg;
+ struct kvec iov;
+ struct scatterlist sg;
+ /*
+ * Bump up the task_struct priority for RX/TX thread set pairs,
+ * and allow ourselves to be interrupted by SIGINT so that a
+ * connection recovery / failure event can be triggered externally.
+ */
+ set_user_nice(current, -20);
+ allow_signal(SIGINT);
+
+restart:
+ conn = iscsi_rx_thread_pre_handler(ts);
+ if (!conn)
+ goto out;
+
+ while (!kthread_should_stop()) {
+ /*
+ * Ensure that both TX and RX per connection kthreads
+ * are scheduled to run on the same CPU.
+ */
+ iscsit_thread_check_cpumask(conn, current, 0);
+
+ memset(buffer, 0, ISCSI_HDR_LEN);
+ memset(&iov, 0, sizeof(struct kvec));
+
+ iov.iov_base = buffer;
+ iov.iov_len = ISCSI_HDR_LEN;
+
+ ret = rx_data(conn, &iov, 1, ISCSI_HDR_LEN);
+ if (ret != ISCSI_HDR_LEN) {
+ iscsit_rx_thread_wait_for_tcp(conn);
+ goto transport_err;
+ }
+
+ /*
+ * Set conn->bad_hdr for use with REJECT PDUs.
+ */
+ memcpy(&conn->bad_hdr, &buffer, ISCSI_HDR_LEN);
+
+ if (conn->conn_ops->HeaderDigest) {
+ iov.iov_base = &digest;
+ iov.iov_len = CRC_LEN;
+
+ ret = rx_data(conn, &iov, 1, CRC_LEN);
+ if (ret != CRC_LEN) {
+ iscsit_rx_thread_wait_for_tcp(conn);
+ goto transport_err;
+ }
+ crypto_hash_init(&conn->conn_rx_hash);
+
+ sg_init_one(&sg, (u8 *)buffer, ISCSI_HDR_LEN);
+ crypto_hash_update(&conn->conn_rx_hash, &sg,
+ ISCSI_HDR_LEN);
+
+ crypto_hash_final(&conn->conn_rx_hash, (u8 *)&checksum);
+
+ if (digest != checksum) {
+ printk(KERN_ERR "HeaderDigest CRC32C failed,"
+ " received 0x%08x, computed 0x%08x\n",
+ digest, checksum);
+ /*
+ * Set the PDU to 0xff so it will intentionally
+ * hit default in the switch below.
+ */
+ memset(buffer, 0xff, ISCSI_HDR_LEN);
+ spin_lock_bh(&conn->sess->session_stats_lock);
+ conn->sess->conn_digest_errors++;
+ spin_unlock_bh(&conn->sess->session_stats_lock);
+ } else {
+ TRACE(TRACE_DIGEST, "Got HeaderDigest CRC32C"
+ " 0x%08x\n", checksum);
+ }
+ }
+
+ if (conn->conn_state == TARG_CONN_STATE_IN_LOGOUT)
+ goto transport_err;
+
+ opcode = buffer[0] & ISCSI_OPCODE_MASK;
+
+ if (conn->sess->sess_ops->SessionType &&
+ ((!(opcode & ISCSI_OP_TEXT)) ||
+ (!(opcode & ISCSI_OP_LOGOUT)))) {
+ printk(KERN_ERR "Received illegal iSCSI Opcode: 0x%02x"
+ " while in Discovery Session, rejecting.\n", opcode);
+ iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
+ buffer, conn);
+ goto transport_err;
+ }
+
+ switch (opcode) {
+ case ISCSI_OP_SCSI_CMD:
+ if (iscsit_handle_scsi_cmd(conn, buffer) < 0)
+ goto transport_err;
+ break;
+ case ISCSI_OP_SCSI_DATA_OUT:
+ if (iscsit_handle_data_out(conn, buffer) < 0)
+ goto transport_err;
+ break;
+ case ISCSI_OP_NOOP_OUT:
+ if (iscsit_handle_nop_out(conn, buffer) < 0)
+ goto transport_err;
+ break;
+ case ISCSI_OP_SCSI_TMFUNC:
+ if (iscsit_handle_task_mgt_cmd(conn, buffer) < 0)
+ goto transport_err;
+ break;
+ case ISCSI_OP_TEXT:
+ if (iscsit_handle_text_cmd(conn, buffer) < 0)
+ goto transport_err;
+ break;
+ case ISCSI_OP_LOGOUT:
+ ret = iscsit_handle_logout_cmd(conn, buffer);
+ if (ret > 0) {
+ wait_for_completion(&conn->conn_logout_comp);
+ goto transport_err;
+ } else if (ret < 0)
+ goto transport_err;
+ break;
+ case ISCSI_OP_SNACK:
+ if (iscsit_handle_snack(conn, buffer) < 0)
+ goto transport_err;
+ break;
+ default:
+ printk(KERN_ERR "Got unknown iSCSI OpCode: 0x%02x\n",
+ opcode);
+ if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
+ printk(KERN_ERR "Cannot recover from unknown"
+ " opcode while ERL=0, closing iSCSI connection"
+ ".\n");
+ goto transport_err;
+ }
+ if (!conn->conn_ops->OFMarker) {
+ printk(KERN_ERR "Unable to recover from unknown"
+ " opcode while OFMarker=No, closing iSCSI"
+ " connection.\n");
+ goto transport_err;
+ }
+ if (iscsit_recover_from_unknown_opcode(conn) < 0) {
+ printk(KERN_ERR "Unable to recover from unknown"
+ " opcode, closing iSCSI connection.\n");
+ goto transport_err;
+ }
+ break;
+ }
+ }
+
+transport_err:
+ if (!signal_pending(current))
+ atomic_set(&conn->transport_failed, 1);
+ iscsit_take_action_for_connection_exit(conn);
+ goto restart;
+out:
+ return 0;
+}
+
+static void iscsit_release_commands_from_conn(struct iscsi_conn *conn)
+{
+ struct iscsi_cmd *cmd = NULL, *cmd_tmp = NULL;
+ struct iscsi_session *sess = conn->sess;
+ struct se_cmd *se_cmd;
+ /*
+ * We expect this function to only ever be called from either RX or TX
+ * thread context via iscsit_close_connection() once the other context
+ * has been reset -> returned sleeping pre-handler state.
+ */
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry_safe(cmd, cmd_tmp, &conn->conn_cmd_list, i_list) {
+ if (!(SE_CMD(cmd)) ||
+ !(SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD)) {
+
+ list_del(&cmd->i_list);
+ spin_unlock_bh(&conn->cmd_lock);
+ iscsit_increment_maxcmdsn(cmd, sess);
+ se_cmd = SE_CMD(cmd);
+ /*
+ * Special cases for active iSCSI TMR, and
+ * transport_get_lun_for_cmd() failing from
+ * iscsit_get_lun_for_cmd() in iscsit_handle_scsi_cmd().
+ */
+ if (cmd->tmr_req && se_cmd->transport_wait_for_tasks)
+ se_cmd->transport_wait_for_tasks(se_cmd, 1, 1);
+ else if (SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD)
+ transport_release_cmd_to_pool(se_cmd);
+ else
+ iscsit_release_cmd(cmd);
+
+ spin_lock_bh(&conn->cmd_lock);
+ continue;
+ }
+ list_del(&cmd->i_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ iscsit_increment_maxcmdsn(cmd, sess);
+ se_cmd = SE_CMD(cmd);
+
+ if (se_cmd->transport_wait_for_tasks)
+ se_cmd->transport_wait_for_tasks(se_cmd, 1, 1);
+
+ spin_lock_bh(&conn->cmd_lock);
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+}
+
+static void iscsit_stop_timers_for_cmds(
+ struct iscsi_conn *conn)
+{
+ struct iscsi_cmd *cmd;
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry(cmd, &conn->conn_cmd_list, i_list) {
+ if (cmd->data_direction == DMA_TO_DEVICE)
+ iscsit_stop_dataout_timer(cmd);
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+}
+
+int iscsit_close_connection(
+ struct iscsi_conn *conn)
+{
+ int conn_logout = (conn->conn_state == TARG_CONN_STATE_IN_LOGOUT);
+ struct iscsi_session *sess = conn->sess;
+
+ TRACE(TRACE_ISCSI, "Closing iSCSI connection CID %hu on SID:"
+ " %u\n", conn->cid, sess->sid);
+ /*
+ * Always up conn_logout_comp just in case the RX Thread is sleeping
+ * and the logout response never got sent because the connection
+ * failed.
+ */
+ complete(&conn->conn_logout_comp);
+
+ iscsi_release_thread_set(conn);
+
+ iscsit_stop_timers_for_cmds(conn);
+ iscsit_stop_nopin_response_timer(conn);
+ iscsit_stop_nopin_timer(conn);
+ iscsit_free_queue_reqs_for_conn(conn);
+
+ /*
+ * During Connection recovery drop unacknowledged out of order
+ * commands for this connection, and prepare the other commands
+ * for realligence.
+ *
+ * During normal operation clear the out of order commands (but
+ * do not free the struct iscsi_ooo_cmdsn's) and release all
+ * struct iscsi_cmds.
+ */
+ if (atomic_read(&conn->connection_recovery)) {
+ iscsit_discard_unacknowledged_ooo_cmdsns_for_conn(conn);
+ iscsit_prepare_cmds_for_realligance(conn);
+ } else {
+ iscsit_clear_ooo_cmdsns_for_conn(conn);
+ iscsit_release_commands_from_conn(conn);
+ }
+
+ /*
+ * Handle decrementing session or connection usage count if
+ * a logout response was not able to be sent because the
+ * connection failed. Fall back to Session Recovery here.
+ */
+ if (atomic_read(&conn->conn_logout_remove)) {
+ if (conn->conn_logout_reason == ISCSI_LOGOUT_REASON_CLOSE_SESSION) {
+ iscsit_dec_conn_usage_count(conn);
+ iscsit_dec_session_usage_count(sess);
+ }
+ if (conn->conn_logout_reason == ISCSI_LOGOUT_REASON_CLOSE_CONNECTION)
+ iscsit_dec_conn_usage_count(conn);
+
+ atomic_set(&conn->conn_logout_remove, 0);
+ atomic_set(&sess->session_reinstatement, 0);
+ atomic_set(&sess->session_fall_back_to_erl0, 1);
+ }
+
+ spin_lock_bh(&sess->conn_lock);
+ list_del(&conn->conn_list);
+
+ /*
+ * Attempt to let the Initiator know this connection failed by
+ * sending an Connection Dropped Async Message on another
+ * active connection.
+ */
+ if (atomic_read(&conn->connection_recovery))
+ iscsit_build_conn_drop_async_message(conn);
+
+ spin_unlock_bh(&sess->conn_lock);
+
+ /*
+ * If connection reinstatement is being performed on this connection,
+ * up the connection reinstatement semaphore that is being blocked on
+ * in iscsit_cause_connection_reinstatement().
+ */
+ spin_lock_bh(&conn->state_lock);
+ if (atomic_read(&conn->sleep_on_conn_wait_comp)) {
+ spin_unlock_bh(&conn->state_lock);
+ complete(&conn->conn_wait_comp);
+ wait_for_completion(&conn->conn_post_wait_comp);
+ spin_lock_bh(&conn->state_lock);
+ }
+
+ /*
+ * If connection reinstatement is being performed on this connection
+ * by receiving a REMOVECONNFORRECOVERY logout request, up the
+ * connection wait rcfr semaphore that is being blocked on
+ * an iscsit_connection_reinstatement_rcfr().
+ */
+ if (atomic_read(&conn->connection_wait_rcfr)) {
+ spin_unlock_bh(&conn->state_lock);
+ complete(&conn->conn_wait_rcfr_comp);
+ wait_for_completion(&conn->conn_post_wait_comp);
+ spin_lock_bh(&conn->state_lock);
+ }
+ atomic_set(&conn->connection_reinstatement, 1);
+ spin_unlock_bh(&conn->state_lock);
+
+ /*
+ * If any other processes are accessing this connection pointer we
+ * must wait until they have completed.
+ */
+ iscsit_check_conn_usage_count(conn);
+
+ if (conn->conn_rx_hash.tfm)
+ crypto_free_hash(conn->conn_rx_hash.tfm);
+ if (conn->conn_tx_hash.tfm)
+ crypto_free_hash(conn->conn_tx_hash.tfm);
+
+ if (conn->conn_cpumask)
+ free_cpumask_var(conn->conn_cpumask);
+
+ kfree(conn->conn_ops);
+ conn->conn_ops = NULL;
+
+ if (conn->sock) {
+ if (conn->conn_flags & CONNFLAG_SCTP_STRUCT_FILE) {
+ kfree(conn->sock->file);
+ conn->sock->file = NULL;
+ }
+ sock_release(conn->sock);
+ }
+
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_FREE.\n");
+ conn->conn_state = TARG_CONN_STATE_FREE;
+ kfree(conn);
+
+ spin_lock_bh(&sess->conn_lock);
+ atomic_dec(&sess->nconn);
+ printk(KERN_INFO "Decremented iSCSI connection count to %hu from node:"
+ " %s\n", atomic_read(&sess->nconn),
+ sess->sess_ops->InitiatorName);
+ /*
+ * Make sure that if one connection fails in an non ERL=2 iSCSI
+ * Session that they all fail.
+ */
+ if ((sess->sess_ops->ErrorRecoveryLevel != 2) && !conn_logout &&
+ !atomic_read(&sess->session_logout))
+ atomic_set(&sess->session_fall_back_to_erl0, 1);
+
+ /*
+ * If this was not the last connection in the session, and we are
+ * performing session reinstatement or falling back to ERL=0, call
+ * iscsit_stop_session() without sleeping to shutdown the other
+ * active connections.
+ */
+ if (atomic_read(&sess->nconn)) {
+ if (!atomic_read(&sess->session_reinstatement) &&
+ !atomic_read(&sess->session_fall_back_to_erl0)) {
+ spin_unlock_bh(&sess->conn_lock);
+ return 0;
+ }
+ if (!atomic_read(&sess->session_stop_active)) {
+ atomic_set(&sess->session_stop_active, 1);
+ spin_unlock_bh(&sess->conn_lock);
+ iscsit_stop_session(sess, 0, 0);
+ return 0;
+ }
+ spin_unlock_bh(&sess->conn_lock);
+ return 0;
+ }
+
+ /*
+ * If this was the last connection in the session and one of the
+ * following is occurring:
+ *
+ * Session Reinstatement is not being performed, and are falling back
+ * to ERL=0 call iscsit_close_session().
+ *
+ * Session Logout was requested. iscsit_close_session() will be called
+ * elsewhere.
+ *
+ * Session Continuation is not being performed, start the Time2Retain
+ * handler and check if sleep_on_sess_wait_sem is active.
+ */
+ if (!atomic_read(&sess->session_reinstatement) &&
+ atomic_read(&sess->session_fall_back_to_erl0)) {
+ spin_unlock_bh(&sess->conn_lock);
+ iscsit_close_session(sess);
+
+ return 0;
+ } else if (atomic_read(&sess->session_logout)) {
+ TRACE(TRACE_STATE, "Moving to TARG_SESS_STATE_FREE.\n");
+ sess->session_state = TARG_SESS_STATE_FREE;
+ spin_unlock_bh(&sess->conn_lock);
+
+ if (atomic_read(&sess->sleep_on_sess_wait_comp))
+ complete(&sess->session_wait_comp);
+
+ return 0;
+ } else {
+ TRACE(TRACE_STATE, "Moving to TARG_SESS_STATE_FAILED.\n");
+ sess->session_state = TARG_SESS_STATE_FAILED;
+
+ if (!atomic_read(&sess->session_continuation)) {
+ spin_unlock_bh(&sess->conn_lock);
+ iscsit_start_time2retain_handler(sess);
+ } else
+ spin_unlock_bh(&sess->conn_lock);
+
+ if (atomic_read(&sess->sleep_on_sess_wait_comp))
+ complete(&sess->session_wait_comp);
+
+ return 0;
+ }
+ spin_unlock_bh(&sess->conn_lock);
+
+ return 0;
+}
+
+int iscsit_close_session(struct iscsi_session *sess)
+{
+ struct iscsi_portal_group *tpg = ISCSI_TPG_S(sess);
+ struct se_portal_group *se_tpg = &tpg->tpg_se_tpg;
+
+ if (atomic_read(&sess->nconn)) {
+ printk(KERN_ERR "%d connection(s) still exist for iSCSI session"
+ " to %s\n", atomic_read(&sess->nconn),
+ sess->sess_ops->InitiatorName);
+ BUG();
+ }
+
+ spin_lock_bh(&se_tpg->session_lock);
+ atomic_set(&sess->session_logout, 1);
+ atomic_set(&sess->session_reinstatement, 1);
+ iscsit_stop_time2retain_timer(sess);
+ spin_unlock_bh(&se_tpg->session_lock);
+
+ /*
+ * transport_deregister_session_configfs() will clear the
+ * struct se_node_acl->nacl_sess pointer now as a iscsi_np process context
+ * can be setting it again with __transport_register_session() in
+ * iscsi_post_login_handler() again after the iscsit_stop_session()
+ * completes in iscsi_np context.
+ */
+ transport_deregister_session_configfs(sess->se_sess);
+
+ /*
+ * If any other processes are accessing this session pointer we must
+ * wait until they have completed. If we are in an interrupt (the
+ * time2retain handler) and contain and active session usage count we
+ * restart the timer and exit.
+ */
+ if (!in_interrupt()) {
+ if (iscsit_check_session_usage_count(sess) == 1)
+ iscsit_stop_session(sess, 1, 1);
+ } else {
+ if (iscsit_check_session_usage_count(sess) == 2) {
+ atomic_set(&sess->session_logout, 0);
+ iscsit_start_time2retain_handler(sess);
+ return 0;
+ }
+ }
+
+ transport_deregister_session(sess->se_sess);
+
+ if (sess->sess_ops->ErrorRecoveryLevel == 2)
+ iscsit_free_connection_recovery_entires(sess);
+
+ iscsit_free_all_ooo_cmdsns(sess);
+
+ spin_lock_bh(&se_tpg->session_lock);
+ TRACE(TRACE_STATE, "Moving to TARG_SESS_STATE_FREE.\n");
+ sess->session_state = TARG_SESS_STATE_FREE;
+ printk(KERN_INFO "Released iSCSI session from node: %s\n",
+ sess->sess_ops->InitiatorName);
+ tpg->nsessions--;
+ if (tpg->tpg_tiqn)
+ tpg->tpg_tiqn->tiqn_nsessions--;
+
+ printk(KERN_INFO "Decremented number of active iSCSI Sessions on"
+ " iSCSI TPG: %hu to %u\n", tpg->tpgt, tpg->nsessions);
+
+ spin_lock(&sess_idr_lock);
+ idr_remove(&sess_idr, sess->session_index);
+ spin_unlock(&sess_idr_lock);
+
+ kfree(sess->sess_ops);
+ sess->sess_ops = NULL;
+ spin_unlock_bh(&se_tpg->session_lock);
+
+ kfree(sess);
+ return 0;
+}
+
+static void iscsit_logout_post_handler_closesession(
+ struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+
+ iscsi_set_thread_clear(conn, ISCSI_CLEAR_TX_THREAD);
+ iscsi_set_thread_set_signal(conn, ISCSI_SIGNAL_TX_THREAD);
+
+ atomic_set(&conn->conn_logout_remove, 0);
+ complete(&conn->conn_logout_comp);
+
+ iscsit_dec_conn_usage_count(conn);
+ iscsit_stop_session(sess, 1, 1);
+ iscsit_dec_session_usage_count(sess);
+ iscsit_close_session(sess);
+}
+
+static void iscsit_logout_post_handler_samecid(
+ struct iscsi_conn *conn)
+{
+ iscsi_set_thread_clear(conn, ISCSI_CLEAR_TX_THREAD);
+ iscsi_set_thread_set_signal(conn, ISCSI_SIGNAL_TX_THREAD);
+
+ atomic_set(&conn->conn_logout_remove, 0);
+ complete(&conn->conn_logout_comp);
+
+ iscsit_cause_connection_reinstatement(conn, 1);
+ iscsit_dec_conn_usage_count(conn);
+}
+
+static void iscsit_logout_post_handler_diffcid(
+ struct iscsi_conn *conn,
+ u16 cid)
+{
+ struct iscsi_conn *l_conn;
+ struct iscsi_session *sess = conn->sess;
+
+ if (!sess)
+ return;
+
+ spin_lock_bh(&sess->conn_lock);
+ list_for_each_entry(l_conn, &sess->sess_conn_list, conn_list) {
+ if (l_conn->cid == cid) {
+ iscsit_inc_conn_usage_count(l_conn);
+ break;
+ }
+ }
+ spin_unlock_bh(&sess->conn_lock);
+
+ if (!l_conn)
+ return;
+
+ if (l_conn->sock)
+ l_conn->sock->ops->shutdown(l_conn->sock, RCV_SHUTDOWN);
+
+ spin_lock_bh(&l_conn->state_lock);
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_IN_LOGOUT.\n");
+ l_conn->conn_state = TARG_CONN_STATE_IN_LOGOUT;
+ spin_unlock_bh(&l_conn->state_lock);
+
+ iscsit_cause_connection_reinstatement(l_conn, 1);
+ iscsit_dec_conn_usage_count(l_conn);
+}
+
+/*
+ * Return of 0 causes the TX thread to restart.
+ */
+static int iscsit_logout_post_handler(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ int ret = 0;
+
+ switch (cmd->logout_reason) {
+ case ISCSI_LOGOUT_REASON_CLOSE_SESSION:
+ switch (cmd->logout_response) {
+ case ISCSI_LOGOUT_SUCCESS:
+ case ISCSI_LOGOUT_CLEANUP_FAILED:
+ default:
+ iscsit_logout_post_handler_closesession(conn);
+ break;
+ }
+ ret = 0;
+ break;
+ case ISCSI_LOGOUT_REASON_CLOSE_CONNECTION:
+ if (conn->cid == cmd->logout_cid) {
+ switch (cmd->logout_response) {
+ case ISCSI_LOGOUT_SUCCESS:
+ case ISCSI_LOGOUT_CLEANUP_FAILED:
+ default:
+ iscsit_logout_post_handler_samecid(conn);
+ break;
+ }
+ ret = 0;
+ } else {
+ switch (cmd->logout_response) {
+ case ISCSI_LOGOUT_SUCCESS:
+ iscsit_logout_post_handler_diffcid(conn,
+ cmd->logout_cid);
+ break;
+ case ISCSI_LOGOUT_CID_NOT_FOUND:
+ case ISCSI_LOGOUT_CLEANUP_FAILED:
+ default:
+ break;
+ }
+ ret = 1;
+ }
+ break;
+ case ISCSI_LOGOUT_REASON_RECOVERY:
+ switch (cmd->logout_response) {
+ case ISCSI_LOGOUT_SUCCESS:
+ case ISCSI_LOGOUT_CID_NOT_FOUND:
+ case ISCSI_LOGOUT_RECOVERY_UNSUPPORTED:
+ case ISCSI_LOGOUT_CLEANUP_FAILED:
+ default:
+ break;
+ }
+ ret = 1;
+ break;
+ default:
+ break;
+
+ }
+ return ret;
+}
+
+void iscsit_fail_session(struct iscsi_session *sess)
+{
+ struct iscsi_conn *conn;
+
+ spin_lock_bh(&sess->conn_lock);
+ list_for_each_entry(conn, &sess->sess_conn_list, conn_list) {
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_CLEANUP_WAIT.\n");
+ conn->conn_state = TARG_CONN_STATE_CLEANUP_WAIT;
+ }
+ spin_unlock_bh(&sess->conn_lock);
+
+ TRACE(TRACE_STATE, "Moving to TARG_SESS_STATE_FAILED.\n");
+ sess->session_state = TARG_SESS_STATE_FAILED;
+}
+
+int iscsit_free_session(struct iscsi_session *sess)
+{
+ u16 conn_count = atomic_read(&sess->nconn);
+ struct iscsi_conn *conn, *conn_tmp = NULL;
+ int is_last;
+
+ spin_lock_bh(&sess->conn_lock);
+ atomic_set(&sess->sleep_on_sess_wait_comp, 1);
+
+ list_for_each_entry_safe(conn, conn_tmp, &sess->sess_conn_list,
+ conn_list) {
+ if (conn_count == 0)
+ break;
+
+ if (list_is_last(&conn->conn_list, &sess->sess_conn_list)) {
+ is_last = 1;
+ } else {
+ iscsit_inc_conn_usage_count(conn_tmp);
+ is_last = 0;
+ }
+ iscsit_inc_conn_usage_count(conn);
+
+ spin_unlock_bh(&sess->conn_lock);
+ iscsit_cause_connection_reinstatement(conn, 1);
+ spin_lock_bh(&sess->conn_lock);
+
+ iscsit_dec_conn_usage_count(conn);
+ if (is_last == 0)
+ iscsit_dec_conn_usage_count(conn_tmp);
+
+ conn_count--;
+ }
+
+ if (atomic_read(&sess->nconn)) {
+ spin_unlock_bh(&sess->conn_lock);
+ wait_for_completion(&sess->session_wait_comp);
+ } else
+ spin_unlock_bh(&sess->conn_lock);
+
+ iscsit_close_session(sess);
+ return 0;
+}
+
+void iscsit_stop_session(
+ struct iscsi_session *sess,
+ int session_sleep,
+ int connection_sleep)
+{
+ u16 conn_count = atomic_read(&sess->nconn);
+ struct iscsi_conn *conn, *conn_tmp = NULL;
+ int is_last;
+
+ spin_lock_bh(&sess->conn_lock);
+ if (session_sleep)
+ atomic_set(&sess->sleep_on_sess_wait_comp, 1);
+
+ if (connection_sleep) {
+ list_for_each_entry_safe(conn, conn_tmp, &sess->sess_conn_list,
+ conn_list) {
+ if (conn_count == 0)
+ break;
+
+ if (list_is_last(&conn->conn_list, &sess->sess_conn_list)) {
+ is_last = 1;
+ } else {
+ iscsit_inc_conn_usage_count(conn_tmp);
+ is_last = 0;
+ }
+ iscsit_inc_conn_usage_count(conn);
+
+ spin_unlock_bh(&sess->conn_lock);
+ iscsit_cause_connection_reinstatement(conn, 1);
+ spin_lock_bh(&sess->conn_lock);
+
+ iscsit_dec_conn_usage_count(conn);
+ if (is_last == 0)
+ iscsit_dec_conn_usage_count(conn_tmp);
+ conn_count--;
+ }
+ } else {
+ list_for_each_entry(conn, &sess->sess_conn_list, conn_list)
+ iscsit_cause_connection_reinstatement(conn, 0);
+ }
+
+ if (session_sleep && atomic_read(&sess->nconn)) {
+ spin_unlock_bh(&sess->conn_lock);
+ wait_for_completion(&sess->session_wait_comp);
+ } else
+ spin_unlock_bh(&sess->conn_lock);
+}
+
+int iscsit_release_sessions_for_tpg(struct iscsi_portal_group *tpg, int force)
+{
+ struct iscsi_session *sess;
+ struct se_portal_group *se_tpg = &tpg->tpg_se_tpg;
+ struct se_session *se_sess, *se_sess_tmp;
+ int session_count = 0;
+
+ spin_lock_bh(&se_tpg->session_lock);
+ if (tpg->nsessions && !force) {
+ spin_unlock_bh(&se_tpg->session_lock);
+ return -1;
+ }
+
+ list_for_each_entry_safe(se_sess, se_sess_tmp, &se_tpg->tpg_sess_list,
+ sess_list) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+
+ spin_lock(&sess->conn_lock);
+ if (atomic_read(&sess->session_fall_back_to_erl0) ||
+ atomic_read(&sess->session_logout) ||
+ (sess->time2retain_timer_flags & ISCSI_TF_EXPIRED)) {
+ spin_unlock(&sess->conn_lock);
+ continue;
+ }
+ atomic_set(&sess->session_reinstatement, 1);
+ spin_unlock(&sess->conn_lock);
+ spin_unlock_bh(&se_tpg->session_lock);
+
+ iscsit_free_session(sess);
+ spin_lock_bh(&se_tpg->session_lock);
+
+ session_count++;
+ }
+ spin_unlock_bh(&se_tpg->session_lock);
+
+ TRACE(TRACE_ISCSI, "Released %d iSCSI Session(s) from Target Portal"
+ " Group: %hu\n", session_count, tpg->tpgt);
+ return 0;
+}
+
+MODULE_DESCRIPTION("RisingTide Systems iSCSI-Target Driver 4.x.x Release");
+MODULE_AUTHOR("nab@Linux-iSCSI.org");
+MODULE_LICENSE("GPL");
+
+module_init(iscsi_target_init_module);
+module_exit(iscsi_target_cleanup_module);
diff --git a/drivers/target/iscsi/iscsi_target.h b/drivers/target/iscsi/iscsi_target.h
new file mode 100644
index 0000000..c85de9f
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target.h
@@ -0,0 +1,32 @@
+#ifndef ISCSI_TARGET_H
+#define ISCSI_TARGET_H
+
+extern struct iscsi_tiqn *iscsit_get_tiqn_for_login(unsigned char *);
+extern struct iscsi_tiqn *iscsit_get_tiqn(unsigned char *, int);
+extern void iscsit_put_tiqn_for_login(struct iscsi_tiqn *);
+extern struct iscsi_tiqn *iscsit_add_tiqn(unsigned char *, int *);
+extern void iscsit_del_tiqn(struct iscsi_tiqn *);
+extern int iscsit_access_np(struct iscsi_np *, struct iscsi_portal_group *);
+extern int iscsit_deaccess_np(struct iscsi_np *, struct iscsi_portal_group *);
+extern struct iscsi_np *iscsit_add_np(struct iscsi_np_addr *, int, int);
+extern int iscsit_reset_np_thread(struct iscsi_np *, struct iscsi_tpg_np *,
+ struct iscsi_portal_group *);
+extern int iscsit_del_np(struct iscsi_np *);
+extern int iscsit_add_reject_from_cmd(u8, int, int, unsigned char *, struct iscsi_cmd *);
+extern int iscsit_logout_closesession(struct iscsi_cmd *, struct iscsi_conn *);
+extern int iscsit_logout_closeconnection(struct iscsi_cmd *, struct iscsi_conn *);
+extern int iscsit_logout_removeconnforrecovery(struct iscsi_cmd *, struct iscsi_conn *);
+extern int iscsit_send_async_msg(struct iscsi_conn *, u16, u8, u8);
+extern int iscsit_send_r2t(struct iscsi_cmd *, struct iscsi_conn *);
+extern int iscsit_build_r2ts_for_cmd(struct iscsi_cmd *, struct iscsi_conn *, int);
+extern void iscsit_thread_get_cpumask(struct iscsi_conn *);
+extern int iscsi_target_tx_thread(void *);
+extern int iscsi_target_rx_thread(void *);
+extern int iscsit_close_connection(struct iscsi_conn *);
+extern int iscsit_close_session(struct iscsi_session *);
+extern void iscsit_fail_session(struct iscsi_session *);
+extern int iscsit_free_session(struct iscsi_session *);
+extern void iscsit_stop_session(struct iscsi_session *, int, int);
+extern int iscsit_release_sessions_for_tpg(struct iscsi_portal_group *, int);
+
+#endif /*** ISCSI_TARGET_H ***/
diff --git a/drivers/target/iscsi/iscsi_target_core.h b/drivers/target/iscsi/iscsi_target_core.h
new file mode 100644
index 0000000..20d094e
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_core.h
@@ -0,0 +1,881 @@
+#ifndef ISCSI_TARGET_CORE_H
+#define ISCSI_TARGET_CORE_H
+
+#include <linux/in.h>
+#include <linux/configfs.h>
+#include <net/sock.h>
+#include <net/tcp.h>
+#include <scsi/scsi_cmnd.h>
+#include <target/target_core_base.h>
+
+#define ISCSI_VERSION "v4.1.0-rc1"
+#define ISCSI_MAX_DATASN_MISSING_COUNT 16
+#define ISCSI_TX_THREAD_TCP_TIMEOUT 2
+#define ISCSI_RX_THREAD_TCP_TIMEOUT 2
+#define SECONDS_FOR_ASYNC_LOGOUT 10
+#define SECONDS_FOR_ASYNC_TEXT 10
+#define IPV4_ADDRESS_SPACE 4
+#define IPV4_BUF_SIZE 18
+#define ISCSI_HDR_LEN 48
+#define CRC_LEN 4
+#define MAX_KEY_NAME_LENGTH 63
+#define MAX_KEY_VALUE_LENGTH 255
+#define WHITE_SPACE " \t\v\f\n\r"
+
+/* struct iscsi_node_attrib sanity values */
+#define NA_DATAOUT_TIMEOUT 3
+#define NA_DATAOUT_TIMEOUT_MAX 60
+#define NA_DATAOUT_TIMEOUT_MIX 2
+#define NA_DATAOUT_TIMEOUT_RETRIES 5
+#define NA_DATAOUT_TIMEOUT_RETRIES_MAX 15
+#define NA_DATAOUT_TIMEOUT_RETRIES_MIN 1
+#define NA_NOPIN_TIMEOUT 5
+#define NA_NOPIN_TIMEOUT_MAX 60
+#define NA_NOPIN_TIMEOUT_MIN 3
+#define NA_NOPIN_RESPONSE_TIMEOUT 5
+#define NA_NOPIN_RESPONSE_TIMEOUT_MAX 60
+#define NA_NOPIN_RESPONSE_TIMEOUT_MIN 3
+#define NA_RANDOM_DATAIN_PDU_OFFSETS 0
+#define NA_RANDOM_DATAIN_SEQ_OFFSETS 0
+#define NA_RANDOM_R2T_OFFSETS 0
+#define NA_DEFAULT_ERL 0
+#define NA_DEFAULT_ERL_MAX 2
+#define NA_DEFAULT_ERL_MIN 0
+
+/* struct iscsi_tpg_attrib sanity values */
+#define TA_AUTHENTICATION 1
+#define TA_LOGIN_TIMEOUT 15
+#define TA_LOGIN_TIMEOUT_MAX 30
+#define TA_LOGIN_TIMEOUT_MIN 5
+#define TA_NETIF_TIMEOUT 2
+#define TA_NETIF_TIMEOUT_MAX 15
+#define TA_NETIF_TIMEOUT_MIN 2
+#define TA_GENERATE_NODE_ACLS 0
+#define TA_DEFAULT_CMDSN_DEPTH 16
+#define TA_DEFAULT_CMDSN_DEPTH_MAX 512
+#define TA_DEFAULT_CMDSN_DEPTH_MIN 1
+#define TA_CACHE_DYNAMIC_ACLS 0
+/* Enabled by default in demo mode (generic_node_acls=1) */
+#define TA_DEMO_MODE_WRITE_PROTECT 1
+/* Disabled by default in production mode w/ explict ACLs */
+#define TA_PROD_MODE_WRITE_PROTECT 0
+#define TA_CACHE_CORE_NPS 0
+
+enum tpg_np_network_transport_table {
+ ISCSI_TCP = 0,
+ ISCSI_SCTP_TCP = 1,
+ ISCSI_SCTP_UDP = 2,
+ ISCSI_IWARP_TCP = 3,
+ ISCSI_IWARP_SCTP = 4,
+ ISCSI_INFINIBAND = 5,
+};
+
+/* RFC-3720 7.1.4 Standard Connection State Diagram for a Target */
+enum target_conn_state_table {
+ TARG_CONN_STATE_FREE = 0x1,
+ TARG_CONN_STATE_XPT_UP = 0x3,
+ TARG_CONN_STATE_IN_LOGIN = 0x4,
+ TARG_CONN_STATE_LOGGED_IN = 0x5,
+ TARG_CONN_STATE_IN_LOGOUT = 0x6,
+ TARG_CONN_STATE_LOGOUT_REQUESTED = 0x7,
+ TARG_CONN_STATE_CLEANUP_WAIT = 0x8,
+};
+
+/* RFC-3720 7.3.2 Session State Diagram for a Target */
+enum target_sess_state_table {
+ TARG_SESS_STATE_FREE = 0x1,
+ TARG_SESS_STATE_ACTIVE = 0x2,
+ TARG_SESS_STATE_LOGGED_IN = 0x3,
+ TARG_SESS_STATE_FAILED = 0x4,
+ TARG_SESS_STATE_IN_CONTINUE = 0x5,
+};
+
+/* struct iscsi_data_count->type */
+enum data_count_type {
+ ISCSI_RX_DATA = 1,
+ ISCSI_TX_DATA = 2,
+};
+
+/* struct iscsi_datain_req->dr_complete */
+enum datain_req_comp_table {
+ DATAIN_COMPLETE_NORMAL = 1,
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY = 2,
+ DATAIN_COMPLETE_CONNECTION_RECOVERY = 3,
+};
+
+/* struct iscsi_datain_req->recovery */
+enum datain_req_rec_table {
+ DATAIN_WITHIN_COMMAND_RECOVERY = 1,
+ DATAIN_CONNECTION_RECOVERY = 2,
+};
+
+/* struct iscsi_portal_group->state */
+enum tpg_state_table {
+ TPG_STATE_FREE = 0,
+ TPG_STATE_ACTIVE = 1,
+ TPG_STATE_INACTIVE = 2,
+ TPG_STATE_COLD_RESET = 3,
+};
+
+/* struct iscsi_tiqn->tiqn_state */
+enum tiqn_state_table {
+ TIQN_STATE_ACTIVE = 1,
+ TIQN_STATE_SHUTDOWN = 2,
+};
+
+/* struct iscsi_cmd->cmd_flags */
+enum cmd_flags_table {
+ ICF_GOT_LAST_DATAOUT = 0x00000001,
+ ICF_GOT_DATACK_SNACK = 0x00000002,
+ ICF_NON_IMMEDIATE_UNSOLICITED_DATA = 0x00000004,
+ ICF_SENT_LAST_R2T = 0x00000008,
+ ICF_WITHIN_COMMAND_RECOVERY = 0x00000010,
+ ICF_CONTIG_MEMORY = 0x00000020,
+ ICF_ATTACHED_TO_RQUEUE = 0x00000040,
+ ICF_OOO_CMDSN = 0x00000080,
+ ICF_REJECT_FAIL_CONN = 0x00000100,
+};
+
+/* struct iscsi_cmd->i_state */
+enum cmd_i_state_table {
+ ISTATE_NO_STATE = 0,
+ ISTATE_NEW_CMD = 1,
+ ISTATE_DEFERRED_CMD = 2,
+ ISTATE_UNSOLICITED_DATA = 3,
+ ISTATE_RECEIVE_DATAOUT = 4,
+ ISTATE_RECEIVE_DATAOUT_RECOVERY = 5,
+ ISTATE_RECEIVED_LAST_DATAOUT = 6,
+ ISTATE_WITHIN_DATAOUT_RECOVERY = 7,
+ ISTATE_IN_CONNECTION_RECOVERY = 8,
+ ISTATE_RECEIVED_TASKMGT = 9,
+ ISTATE_SEND_ASYNCMSG = 10,
+ ISTATE_SENT_ASYNCMSG = 11,
+ ISTATE_SEND_DATAIN = 12,
+ ISTATE_SEND_LAST_DATAIN = 13,
+ ISTATE_SENT_LAST_DATAIN = 14,
+ ISTATE_SEND_LOGOUTRSP = 15,
+ ISTATE_SENT_LOGOUTRSP = 16,
+ ISTATE_SEND_NOPIN = 17,
+ ISTATE_SENT_NOPIN = 18,
+ ISTATE_SEND_REJECT = 19,
+ ISTATE_SENT_REJECT = 20,
+ ISTATE_SEND_R2T = 21,
+ ISTATE_SENT_R2T = 22,
+ ISTATE_SEND_R2T_RECOVERY = 23,
+ ISTATE_SENT_R2T_RECOVERY = 24,
+ ISTATE_SEND_LAST_R2T = 25,
+ ISTATE_SENT_LAST_R2T = 26,
+ ISTATE_SEND_LAST_R2T_RECOVERY = 27,
+ ISTATE_SENT_LAST_R2T_RECOVERY = 28,
+ ISTATE_SEND_STATUS = 29,
+ ISTATE_SEND_STATUS_BROKEN_PC = 30,
+ ISTATE_SENT_STATUS = 31,
+ ISTATE_SEND_STATUS_RECOVERY = 32,
+ ISTATE_SENT_STATUS_RECOVERY = 33,
+ ISTATE_SEND_TASKMGTRSP = 34,
+ ISTATE_SENT_TASKMGTRSP = 35,
+ ISTATE_SEND_TEXTRSP = 36,
+ ISTATE_SENT_TEXTRSP = 37,
+ ISTATE_SEND_NOPIN_WANT_RESPONSE = 38,
+ ISTATE_SENT_NOPIN_WANT_RESPONSE = 39,
+ ISTATE_SEND_NOPIN_NO_RESPONSE = 40,
+ ISTATE_REMOVE = 41,
+ ISTATE_FREE = 42,
+};
+
+/* Used for iscsi_recover_cmdsn() return values */
+enum recover_cmdsn_ret_table {
+ CMDSN_ERROR_CANNOT_RECOVER = -1,
+ CMDSN_NORMAL_OPERATION = 0,
+ CMDSN_LOWER_THAN_EXP = 1,
+ CMDSN_HIGHER_THAN_EXP = 2,
+};
+
+/* Used for iscsi_handle_immediate_data() return values */
+enum immedate_data_ret_table {
+ IMMEDIDATE_DATA_CANNOT_RECOVER = -1,
+ IMMEDIDATE_DATA_NORMAL_OPERATION = 0,
+ IMMEDIDATE_DATA_ERL1_CRC_FAILURE = 1,
+};
+
+/* Used for iscsi_decide_dataout_action() return values */
+enum dataout_action_ret_table {
+ DATAOUT_CANNOT_RECOVER = -1,
+ DATAOUT_NORMAL = 0,
+ DATAOUT_SEND_R2T = 1,
+ DATAOUT_SEND_TO_TRANSPORT = 2,
+ DATAOUT_WITHIN_COMMAND_RECOVERY = 3,
+};
+
+/* Used for struct iscsi_node_auth->naf_flags */
+enum naf_flags_table {
+ NAF_USERID_SET = 0x01,
+ NAF_PASSWORD_SET = 0x02,
+ NAF_USERID_IN_SET = 0x04,
+ NAF_PASSWORD_IN_SET = 0x08,
+};
+
+/* Used by various struct timer_list to manage iSCSI specific state */
+enum iscsi_timer_flags_table {
+ ISCSI_TF_RUNNING = 0x01,
+ ISCSI_TF_STOP = 0x02,
+ ISCSI_TF_EXPIRED = 0x04,
+};
+
+/* Used for struct iscsi_np->np_flags */
+enum np_flags_table {
+ NPF_IP_NETWORK = 0x00,
+ NPF_SCTP_STRUCT_FILE = 0x01 /* Bugfix */
+};
+
+/* Used for struct iscsi_np->np_thread_state */
+enum np_thread_state_table {
+ ISCSI_NP_THREAD_ACTIVE = 1,
+ ISCSI_NP_THREAD_INACTIVE = 2,
+ ISCSI_NP_THREAD_RESET = 3,
+ ISCSI_NP_THREAD_SHUTDOWN = 4,
+ ISCSI_NP_THREAD_EXIT = 5,
+};
+
+struct iscsi_conn_ops {
+ u8 HeaderDigest; /* [0,1] == [None,CRC32C] */
+ u8 DataDigest; /* [0,1] == [None,CRC32C] */
+ u32 MaxRecvDataSegmentLength; /* [512..2**24-1] */
+ u8 OFMarker; /* [0,1] == [No,Yes] */
+ u8 IFMarker; /* [0,1] == [No,Yes] */
+ u32 OFMarkInt; /* [1..65535] */
+ u32 IFMarkInt; /* [1..65535] */
+};
+
+struct iscsi_sess_ops {
+ char InitiatorName[224];
+ char InitiatorAlias[256];
+ char TargetName[224];
+ char TargetAlias[256];
+ char TargetAddress[256];
+ u16 TargetPortalGroupTag; /* [0..65535] */
+ u16 MaxConnections; /* [1..65535] */
+ u8 InitialR2T; /* [0,1] == [No,Yes] */
+ u8 ImmediateData; /* [0,1] == [No,Yes] */
+ u32 MaxBurstLength; /* [512..2**24-1] */
+ u32 FirstBurstLength; /* [512..2**24-1] */
+ u16 DefaultTime2Wait; /* [0..3600] */
+ u16 DefaultTime2Retain; /* [0..3600] */
+ u16 MaxOutstandingR2T; /* [1..65535] */
+ u8 DataPDUInOrder; /* [0,1] == [No,Yes] */
+ u8 DataSequenceInOrder; /* [0,1] == [No,Yes] */
+ u8 ErrorRecoveryLevel; /* [0..2] */
+ u8 SessionType; /* [0,1] == [Normal,Discovery]*/
+};
+
+struct iscsi_queue_req {
+ int state;
+ struct se_obj_lun_type_s *queue_se_obj_api;
+ struct iscsi_cmd *cmd;
+ struct list_head qr_list;
+};
+
+struct iscsi_data_count {
+ int data_length;
+ int sync_and_steering;
+ enum data_count_type type;
+ u32 iov_count;
+ u32 ss_iov_count;
+ u32 ss_marker_count;
+ struct kvec *iov;
+};
+
+struct iscsi_param_list {
+ struct list_head param_list;
+ struct list_head extra_response_list;
+};
+
+struct iscsi_datain_req {
+ enum datain_req_comp_table dr_complete;
+ int generate_recovery_values;
+ enum datain_req_rec_table recovery;
+ u32 begrun;
+ u32 runlength;
+ u32 data_length;
+ u32 data_offset;
+ u32 data_offset_end;
+ u32 data_sn;
+ u32 next_burst_len;
+ u32 read_data_done;
+ u32 seq_send_order;
+ struct list_head dr_list;
+} ____cacheline_aligned;
+
+struct iscsi_ooo_cmdsn {
+ u16 cid;
+ u32 batch_count;
+ u32 cmdsn;
+ u32 exp_cmdsn;
+ struct iscsi_cmd *cmd;
+ struct list_head ooo_list;
+} ____cacheline_aligned;
+
+struct iscsi_datain {
+ u8 flags;
+ u32 data_sn;
+ u32 length;
+ u32 offset;
+} ____cacheline_aligned;
+
+struct iscsi_r2t {
+ int seq_complete;
+ int recovery_r2t;
+ int sent_r2t;
+ u32 r2t_sn;
+ u32 offset;
+ u32 targ_xfer_tag;
+ u32 xfer_len;
+ struct list_head r2t_list;
+} ____cacheline_aligned;
+
+struct iscsi_cmd {
+ enum iscsi_timer_flags_table dataout_timer_flags;
+ /* DataOUT timeout retries */
+ u8 dataout_timeout_retries;
+ /* Within command recovery count */
+ u8 error_recovery_count;
+ /* iSCSI dependent state for out or order CmdSNs */
+ enum cmd_i_state_table deferred_i_state;
+ /* iSCSI dependent state */
+ enum cmd_i_state_table i_state;
+ /* Command is an immediate command (ISCSI_OP_IMMEDIATE set) */
+ u8 immediate_cmd;
+ /* Immediate data present */
+ u8 immediate_data;
+ /* iSCSI Opcode */
+ u8 iscsi_opcode;
+ /* iSCSI Response Code */
+ u8 iscsi_response;
+ /* Logout reason when iscsi_opcode == ISCSI_INIT_LOGOUT_CMND */
+ u8 logout_reason;
+ /* Logout response code when iscsi_opcode == ISCSI_INIT_LOGOUT_CMND */
+ u8 logout_response;
+ /* MaxCmdSN has been incremented */
+ u8 maxcmdsn_inc;
+ /* Immediate Unsolicited Dataout */
+ u8 unsolicited_data;
+ /* CID contained in logout PDU when opcode == ISCSI_INIT_LOGOUT_CMND */
+ u16 logout_cid;
+ /* Command flags */
+ enum cmd_flags_table cmd_flags;
+ /* Initiator Task Tag assigned from Initiator */
+ u32 init_task_tag;
+ /* Target Transfer Tag assigned from Target */
+ u32 targ_xfer_tag;
+ /* CmdSN assigned from Initiator */
+ u32 cmd_sn;
+ /* ExpStatSN assigned from Initiator */
+ u32 exp_stat_sn;
+ /* StatSN assigned to this ITT */
+ u32 stat_sn;
+ /* DataSN Counter */
+ u32 data_sn;
+ /* R2TSN Counter */
+ u32 r2t_sn;
+ /* Last DataSN acknowledged via DataAck SNACK */
+ u32 acked_data_sn;
+ /* Used for echoing NOPOUT ping data */
+ u32 buf_ptr_size;
+ /* Used to store DataDigest */
+ u32 data_crc;
+ /* Total size in bytes associated with command */
+ u32 data_length;
+ /* Counter for MaxOutstandingR2T */
+ u32 outstanding_r2ts;
+ /* Next R2T Offset when DataSequenceInOrder=Yes */
+ u32 r2t_offset;
+ /* Iovec current and orig count for iscsi_cmd->iov_data */
+ u32 iov_data_count;
+ u32 orig_iov_data_count;
+ /* Number of miscellaneous iovecs used for IP stack calls */
+ u32 iov_misc_count;
+ /* Bytes used for 32-bit word padding */
+ u32 pad_bytes;
+ /* Number of struct iscsi_pdu in struct iscsi_cmd->pdu_list */
+ u32 pdu_count;
+ /* Next struct iscsi_pdu to send in struct iscsi_cmd->pdu_list */
+ u32 pdu_send_order;
+ /* Current struct iscsi_pdu in struct iscsi_cmd->pdu_list */
+ u32 pdu_start;
+ u32 residual_count;
+ /* Next struct iscsi_seq to send in struct iscsi_cmd->seq_list */
+ u32 seq_send_order;
+ /* Number of struct iscsi_seq in struct iscsi_cmd->seq_list */
+ u32 seq_count;
+ /* Current struct iscsi_seq in struct iscsi_cmd->seq_list */
+ u32 seq_no;
+ /* Lowest offset in current DataOUT sequence */
+ u32 seq_start_offset;
+ /* Highest offset in current DataOUT sequence */
+ u32 seq_end_offset;
+ /* Total size in bytes received so far of READ data */
+ u32 read_data_done;
+ /* Total size in bytes received so far of WRITE data */
+ u32 write_data_done;
+ /* Counter for FirstBurstLength key */
+ u32 first_burst_len;
+ /* Counter for MaxBurstLength key */
+ u32 next_burst_len;
+ /* Transfer size used for IP stack calls */
+ u32 tx_size;
+ /* Buffer used for various purposes */
+ void *buf_ptr;
+ /* See include/linux/dma-mapping.h */
+ enum dma_data_direction data_direction;
+ /* iSCSI PDU Header + CRC */
+ unsigned char pdu[ISCSI_HDR_LEN + CRC_LEN];
+ /* Number of times struct iscsi_cmd is present in immediate queue */
+ atomic_t immed_queue_count;
+ atomic_t response_queue_count;
+ atomic_t transport_sent;
+ spinlock_t datain_lock;
+ spinlock_t dataout_timeout_lock;
+ /* spinlock for protecting struct iscsi_cmd->i_state */
+ spinlock_t istate_lock;
+ /* spinlock for adding within command recovery entries */
+ spinlock_t error_lock;
+ /* spinlock for adding R2Ts */
+ spinlock_t r2t_lock;
+ /* DataIN List */
+ struct list_head datain_list;
+ /* R2T List */
+ struct list_head cmd_r2t_list;
+ struct completion reject_comp;
+ /* Semaphore used for allocating buffer */
+ struct completion unsolicited_data_comp;
+ /* Timer for DataOUT */
+ struct timer_list dataout_timer;
+ /* Iovecs for SCSI data payload RX/TX w/ kernel level sockets */
+ struct kvec *iov_data;
+ /* Iovecs for miscellaneous purposes */
+#define ISCSI_MISC_IOVECS 5
+ struct kvec iov_misc[ISCSI_MISC_IOVECS];
+ /* Array of struct iscsi_pdu used for DataPDUInOrder=No */
+ struct iscsi_pdu *pdu_list;
+ /* Current struct iscsi_pdu used for DataPDUInOrder=No */
+ struct iscsi_pdu *pdu_ptr;
+ /* Array of struct iscsi_seq used for DataSequenceInOrder=No */
+ struct iscsi_seq *seq_list;
+ /* Current struct iscsi_seq used for DataSequenceInOrder=No */
+ struct iscsi_seq *seq_ptr;
+ /* TMR Request when iscsi_opcode == ISCSI_OP_SCSI_TMFUNC */
+ struct iscsi_tmr_req *tmr_req;
+ /* Connection this command is alligient to */
+ struct iscsi_conn *conn;
+ /* Pointer to connection recovery entry */
+ struct iscsi_conn_recovery *cr;
+ /* Session the command is part of, used for connection recovery */
+ struct iscsi_session *sess;
+ /* Next command in the session pool */
+ struct iscsi_cmd *next;
+ /* list_head for connection list */
+ struct list_head i_list;
+ /* Next command in DAS transport list */
+ struct iscsi_cmd *t_next;
+ /* Previous command in DAS transport list */
+ struct iscsi_cmd *t_prev;
+ /* The TCM I/O descriptor that is accessed via container_of() */
+ struct se_cmd se_cmd;
+ /* Sense buffer that will be mapped into outgoing status */
+#define ISCSI_SENSE_BUFFER_LEN (TRANSPORT_SENSE_BUFFER + 2)
+ unsigned char sense_buffer[ISCSI_SENSE_BUFFER_LEN];
+} ____cacheline_aligned;
+
+#define SE_CMD(cmd) (&(cmd)->se_cmd)
+
+struct iscsi_tmr_req {
+ bool task_reassign:1;
+ u32 ref_cmd_sn;
+ u32 exp_data_sn;
+ struct iscsi_conn_recovery *conn_recovery;
+ struct se_tmr_req *se_tmr_req;
+};
+
+struct iscsi_conn {
+#define ISCSI_NETDEV_NAME_SIZE 12
+ char net_dev[ISCSI_NETDEV_NAME_SIZE];
+ /* Authentication Successful for this connection */
+ u8 auth_complete;
+ /* State connection is currently in */
+ u8 conn_state;
+ u8 conn_logout_reason;
+ u8 network_transport;
+ enum iscsi_timer_flags_table nopin_timer_flags;
+ enum iscsi_timer_flags_table nopin_response_timer_flags;
+ u8 tx_immediate_queue;
+ u8 tx_response_queue;
+ /* Used to know what thread encountered a transport failure */
+ u8 which_thread;
+ /* connection id assigned by the Initiator */
+ u16 cid;
+ /* Remote TCP Port */
+ u16 login_port;
+ int net_size;
+ u32 auth_id;
+#define CONNFLAG_SCTP_STRUCT_FILE 0x01
+ u32 conn_flags;
+ /* Remote TCP IPv4 address */
+ u32 login_ipv4;
+ /* Used for iscsi_tx_login_rsp() */
+ u32 login_itt;
+ u32 exp_statsn;
+ /* Per connection status sequence number */
+ u32 stat_sn;
+ /* IFMarkInt's Current Value */
+ u32 if_marker;
+ /* OFMarkInt's Current Value */
+ u32 of_marker;
+ /* Used for calculating OFMarker offset to next PDU */
+ u32 of_marker_offset;
+ /* Complete Bad PDU for sending reject */
+ unsigned char bad_hdr[ISCSI_HDR_LEN];
+#define IPV6_ADDRESS_SPACE 48
+ unsigned char login_ip[IPV6_ADDRESS_SPACE];
+ u16 local_port;
+ u32 local_ip;
+ int conn_usage_count;
+ int conn_waiting_on_uc;
+ atomic_t check_immediate_queue;
+ atomic_t conn_logout_remove;
+ atomic_t connection_exit;
+ atomic_t connection_recovery;
+ atomic_t connection_reinstatement;
+ atomic_t connection_wait;
+ atomic_t connection_wait_rcfr;
+ atomic_t sleep_on_conn_wait_comp;
+ atomic_t transport_failed;
+ struct net_device *net_if;
+ struct completion conn_post_wait_comp;
+ struct completion conn_wait_comp;
+ struct completion conn_wait_rcfr_comp;
+ struct completion conn_waiting_on_uc_comp;
+ struct completion conn_logout_comp;
+ struct completion tx_half_close_comp;
+ struct completion rx_half_close_comp;
+ /* socket used by this connection */
+ struct socket *sock;
+ struct timer_list nopin_timer;
+ struct timer_list nopin_response_timer;
+ struct timer_list transport_timer;
+ /* Spinlock used for add/deleting cmd's from conn_cmd_list */
+ spinlock_t cmd_lock;
+ spinlock_t conn_usage_lock;
+ spinlock_t immed_queue_lock;
+ spinlock_t nopin_timer_lock;
+ spinlock_t response_queue_lock;
+ spinlock_t state_lock;
+ /* libcrypto RX and TX contexts for crc32c */
+ struct hash_desc conn_rx_hash;
+ struct hash_desc conn_tx_hash;
+ /* Used for scheduling TX and RX connection kthreads */
+ cpumask_var_t conn_cpumask;
+ int conn_rx_reset_cpumask:1;
+ int conn_tx_reset_cpumask:1;
+ /* list_head of struct iscsi_cmd for this connection */
+ struct list_head conn_cmd_list;
+ struct list_head immed_queue_list;
+ struct list_head response_queue_list;
+ struct iscsi_conn_ops *conn_ops;
+ struct iscsi_param_list *param_list;
+ /* Used for per connection auth state machine */
+ void *auth_protocol;
+ struct iscsi_login_thread_s *login_thread;
+ struct iscsi_portal_group *tpg;
+ /* Pointer to parent session */
+ struct iscsi_session *sess;
+ /* Pointer to thread_set in use for this conn's threads */
+ struct iscsi_thread_set *thread_set;
+ /* list_head for session connection list */
+ struct list_head conn_list;
+} ____cacheline_aligned;
+
+struct iscsi_conn_recovery {
+ u16 cid;
+ u32 cmd_count;
+ u32 maxrecvdatasegmentlength;
+ int ready_for_reallegiance;
+ struct list_head conn_recovery_cmd_list;
+ spinlock_t conn_recovery_cmd_lock;
+ struct completion time2wait_comp;
+ struct timer_list time2retain_timer;
+ struct iscsi_session *sess;
+ struct list_head cr_list;
+} ____cacheline_aligned;
+
+struct iscsi_session {
+ u8 cmdsn_outoforder;
+ u8 initiator_vendor;
+ u8 isid[6];
+ enum iscsi_timer_flags_table time2retain_timer_flags;
+ u8 version_active;
+ u16 cid_called;
+ u16 conn_recovery_count;
+ u16 tsih;
+ /* state session is currently in */
+ u32 session_state;
+ /* session wide counter: initiator assigned task tag */
+ u32 init_task_tag;
+ /* session wide counter: target assigned task tag */
+ u32 targ_xfer_tag;
+ u32 cmdsn_window;
+ /* session wide counter: expected command sequence number */
+ u32 exp_cmd_sn;
+ /* session wide counter: maximum allowed command sequence number */
+ u32 max_cmd_sn;
+ u32 ooo_cmdsn_count;
+ /* LIO specific session ID */
+ u32 sid;
+ char auth_type[8];
+ /* unique within the target */
+ int session_index;
+ /* Used for session reference counting */
+ int session_usage_count;
+ int session_waiting_on_uc;
+ u32 cmd_pdus;
+ u32 rsp_pdus;
+ u64 tx_data_octets;
+ u64 rx_data_octets;
+ u32 conn_digest_errors;
+ u32 conn_timeout_errors;
+ u64 creation_time;
+ spinlock_t session_stats_lock;
+ /* Number of active connections */
+ atomic_t nconn;
+ atomic_t session_continuation;
+ atomic_t session_fall_back_to_erl0;
+ atomic_t session_logout;
+ atomic_t session_reinstatement;
+ atomic_t session_stop_active;
+ atomic_t sleep_on_sess_wait_comp;
+ atomic_t transport_wait_cmds;
+ /* connection list */
+ struct list_head sess_conn_list;
+ struct list_head cr_active_list;
+ struct list_head cr_inactive_list;
+ spinlock_t cmdsn_lock;
+ spinlock_t conn_lock;
+ spinlock_t cr_a_lock;
+ spinlock_t cr_i_lock;
+ spinlock_t session_usage_lock;
+ spinlock_t ttt_lock;
+ struct list_head sess_ooo_cmdsn_list;
+ struct completion async_msg_comp;
+ struct completion reinstatement_comp;
+ struct completion session_wait_comp;
+ struct completion session_waiting_on_uc_comp;
+ struct timer_list time2retain_timer;
+ struct iscsi_sess_ops *sess_ops;
+ struct se_session *se_sess;
+ struct iscsi_portal_group *tpg;
+} ____cacheline_aligned;
+
+struct iscsi_login {
+ u8 auth_complete;
+ u8 checked_for_existing;
+ u8 current_stage;
+ u8 leading_connection;
+ u8 first_request;
+ u8 version_min;
+ u8 version_max;
+ char isid[6];
+ u32 cmd_sn;
+ u32 init_task_tag;
+ u32 initial_exp_statsn;
+ u32 rsp_length;
+ u16 cid;
+ u16 tsih;
+ char *req;
+ char *rsp;
+ char *req_buf;
+ char *rsp_buf;
+} ____cacheline_aligned;
+
+struct iscsi_node_attrib {
+ u32 dataout_timeout;
+ u32 dataout_timeout_retries;
+ u32 default_erl;
+ u32 nopin_timeout;
+ u32 nopin_response_timeout;
+ u32 random_datain_pdu_offsets;
+ u32 random_datain_seq_offsets;
+ u32 random_r2t_offsets;
+ u32 tmr_cold_reset;
+ u32 tmr_warm_reset;
+ struct iscsi_node_acl *nacl;
+};
+
+struct se_dev_entry_s;
+
+struct iscsi_node_auth {
+ enum naf_flags_table naf_flags;
+ int authenticate_target;
+ /* Used for iscsi_global->discovery_auth,
+ * set to zero (auth disabled) by default */
+ int enforce_discovery_auth;
+#define MAX_USER_LEN 256
+#define MAX_PASS_LEN 256
+ char userid[MAX_USER_LEN];
+ char password[MAX_PASS_LEN];
+ char userid_mutual[MAX_USER_LEN];
+ char password_mutual[MAX_PASS_LEN];
+};
+
+#include "iscsi_target_stat.h"
+
+struct iscsi_node_stat_grps {
+ struct config_group iscsi_sess_stats_group;
+ struct config_group iscsi_conn_stats_group;
+};
+
+struct iscsi_node_acl {
+ struct iscsi_node_attrib node_attrib;
+ struct iscsi_node_auth node_auth;
+ struct iscsi_node_stat_grps node_stat_grps;
+ struct se_node_acl se_node_acl;
+};
+
+#define NODE_STAT_GRPS(nacl) (&(nacl)->node_stat_grps)
+
+#define ISCSI_NODE_ATTRIB(t) (&(t)->node_attrib)
+#define ISCSI_NODE_AUTH(t) (&(t)->node_auth)
+
+struct iscsi_tpg_attrib {
+ u32 authentication;
+ u32 login_timeout;
+ u32 netif_timeout;
+ u32 generate_node_acls;
+ u32 cache_dynamic_acls;
+ u32 default_cmdsn_depth;
+ u32 demo_mode_write_protect;
+ u32 prod_mode_write_protect;
+ struct iscsi_portal_group *tpg;
+};
+
+struct iscsi_np {
+ unsigned char np_net_dev[ISCSI_NETDEV_NAME_SIZE];
+ int np_network_transport;
+ int np_ip_proto;
+ int np_sock_type;
+ enum np_thread_state_table np_thread_state;
+ enum iscsi_timer_flags_table np_login_timer_flags;
+ u32 np_exports;
+ enum np_flags_table np_flags;
+ u32 np_ipv4;
+ unsigned char np_ip[IPV6_ADDRESS_SPACE];
+ u16 np_port;
+ spinlock_t np_thread_lock;
+ struct completion np_restart_comp;
+ struct socket *np_socket;
+ struct __kernel_sockaddr_storage np_sockaddr;
+ struct task_struct *np_thread;
+ struct timer_list np_login_timer;
+ struct iscsi_portal_group *np_login_tpg;
+ struct list_head np_list;
+} ____cacheline_aligned;
+
+struct iscsi_tpg_np {
+ struct iscsi_np *tpg_np;
+ struct iscsi_portal_group *tpg;
+ struct iscsi_tpg_np *tpg_np_parent;
+ struct list_head tpg_np_list;
+ struct list_head tpg_np_child_list;
+ struct list_head tpg_np_parent_list;
+ struct se_tpg_np se_tpg_np;
+ spinlock_t tpg_np_parent_lock;
+};
+
+struct iscsi_np_addr {
+ u16 np_port;
+ u32 np_flags;
+ u32 np_ipv4;
+ unsigned char np_ipv6[IPV6_ADDRESS_SPACE];
+} ____cacheline_aligned;
+
+struct iscsi_portal_group {
+ unsigned char tpg_chap_id;
+ /* TPG State */
+ enum tpg_state_table tpg_state;
+ /* Target Portal Group Tag */
+ u16 tpgt;
+ /* Id assigned to target sessions */
+ u16 ntsih;
+ /* Number of active sessions */
+ u32 nsessions;
+ /* Number of Network Portals available for this TPG */
+ u32 num_tpg_nps;
+ /* Per TPG LIO specific session ID. */
+ u32 sid;
+ /* Spinlock for adding/removing Network Portals */
+ spinlock_t tpg_np_lock;
+ spinlock_t tpg_state_lock;
+ struct se_portal_group tpg_se_tpg;
+ struct mutex tpg_access_lock;
+ struct mutex np_login_lock;
+ struct iscsi_tpg_attrib tpg_attrib;
+ /* Pointer to default list of iSCSI parameters for TPG */
+ struct iscsi_param_list *param_list;
+ struct iscsi_tiqn *tpg_tiqn;
+ struct list_head tpg_gnp_list;
+ struct list_head tpg_list;
+} ____cacheline_aligned;
+
+#define ISCSI_TPG_C(c) ((struct iscsi_portal_group *)(c)->tpg)
+#define ISCSI_TPG_LUN(c, l) ((iscsi_tpg_list_t *)(c)->tpg->tpg_lun_list_t[l])
+#define ISCSI_TPG_S(s) ((struct iscsi_portal_group *)(s)->tpg)
+#define ISCSI_TPG_ATTRIB(t) (&(t)->tpg_attrib)
+#define SE_TPG(tpg) (&(tpg)->tpg_se_tpg)
+
+struct iscsi_wwn_stat_grps {
+ struct config_group iscsi_stat_group;
+ struct config_group iscsi_instance_group;
+ struct config_group iscsi_sess_err_group;
+ struct config_group iscsi_tgt_attr_group;
+ struct config_group iscsi_login_stats_group;
+ struct config_group iscsi_logout_stats_group;
+};
+
+struct iscsi_tiqn {
+#define ISCSI_IQN_LEN 224
+ unsigned char tiqn[ISCSI_IQN_LEN];
+ enum tiqn_state_table tiqn_state;
+ int tiqn_access_count;
+ u32 tiqn_active_tpgs;
+ u32 tiqn_ntpgs;
+ u32 tiqn_num_tpg_nps;
+ u32 tiqn_nsessions;
+ struct list_head tiqn_list;
+ struct list_head tiqn_tpg_list;
+ spinlock_t tiqn_state_lock;
+ spinlock_t tiqn_tpg_lock;
+ struct se_wwn tiqn_wwn;
+ struct iscsi_wwn_stat_grps tiqn_stat_grps;
+ int tiqn_index;
+ struct iscsi_sess_err_stats sess_err_stats;
+ struct iscsi_login_stats login_stats;
+ struct iscsi_logout_stats logout_stats;
+} ____cacheline_aligned;
+
+#define WWN_STAT_GRPS(tiqn) (&(tiqn)->tiqn_stat_grps)
+
+struct iscsi_global {
+ /* In core shutdown */
+ u32 in_shutdown;
+ u32 active_ts;
+ /* Unique identifier used for the authentication daemon */
+ u32 auth_id;
+ u32 inactive_ts;
+ /* Thread Set bitmap count */
+ int ts_bitmap_count;
+ /* Thread Set bitmap pointer */
+ unsigned long *ts_bitmap;
+ /* Used for iSCSI discovery session authentication */
+ struct iscsi_node_acl discovery_acl;
+ struct iscsi_portal_group *discovery_tpg;
+};
+
+#endif /* ISCSI_TARGET_CORE_H */
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* Re: [PATCH-v2 04/14] iscsi-target: Add iSCSI fabric support for target v4
2011-03-23 3:06 ` [PATCH-v2 04/14] iscsi-target: Add iSCSI fabric support for target v4 Nicholas A. Bellinger
@ 2011-04-07 7:34 ` Mike Christie
2011-05-09 16:22 ` Nicholas A. Bellinger
0 siblings, 1 reply; 40+ messages in thread
From: Mike Christie @ 2011-04-07 7:34 UTC (permalink / raw)
To: Nicholas A. Bellinger
Cc: linux-scsi, James Bottomley, Christoph Hellwig, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh
On 03/22/2011 10:06 PM, Nicholas A. Bellinger wrote:
> From: Nicholas Bellinger<nab@linux-iscsi.org>
>
> The RisingTide Systems iSCSI target module is a full featured in-kernel
> software implementation of iSCSI target mode (RFC-3720) for the mainline
> target v4 infrastructure code. More information can be found here:
>
> http://linux-iscsi.org/wiki/ISCSI
>
> This includes support for:
>
> * RFC-3720 defined request / response state machines and support for
> all defined iSCSI operation codes from Section 10.2.1.2 using libiscsi
> include/scsi/iscsi_proto.h PDU definitions
> * Target v4 compatible control plane using the generic layout in
> target_core_fabric_configfs.c and fabric dependent attributes
> within /sys/kernel/config/target/iscsi/ subdirectories.
> * Target v4 compatible iSCSI statistics based on RFC-4544 (iSCSI MIBS)
> * Support for IPv6 and IPv4 network portals in M:N mapping to TPGs
> * iSCSI Error Recovery Hierarchy support
> * Per iSCSI connection RX/TX thread pair scheduling affinity
> * crc32c + crc32c_intel SSEv4 instruction offload support using libcrypto
> * CHAP Authentication support using libcrypto
>
> Significant feedback for mainline code cleanups by Christoph Hellwig and
> Mike Christie.
>
> Signed-off-by: Nicholas A. Bellinger<nab@linux-iscsi.org>
> ---
> drivers/target/iscsi/iscsi_target.c | 5017 ++++++++++++++++++++++++++++++
> drivers/target/iscsi/iscsi_target.h | 32 +
> drivers/target/iscsi/iscsi_target_core.h | 881 ++++++
> 3 files changed, 5930 insertions(+), 0 deletions(-)
> create mode 100644 drivers/target/iscsi/iscsi_target.c
> create mode 100644 drivers/target/iscsi/iscsi_target.h
> create mode 100644 drivers/target/iscsi/iscsi_target_core.h
>
> diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
> new file mode 100644
> index 0000000..c81f5cc
> --- /dev/null
> +++ b/drivers/target/iscsi/iscsi_target.c
> @@ -0,0 +1,5017 @@
> +/*******************************************************************************
> + * This file contains main functions related to the iSCSI Target Core Driver.
> + *
> + * © Copyright 2007-2011 RisingTide Systems LLC.
> + *
> + * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
> + *
> + * Author: Nicholas A. Bellinger<nab@linux-iscsi.org>
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + ******************************************************************************/
> +
> +#include<linux/kthread.h>
> +#include<linux/crypto.h>
> +#include<linux/completion.h>
> +#include<asm/unaligned.h>
> +#include<scsi/scsi_device.h>
> +#include<scsi/iscsi_proto.h>
> +#include<target/target_core_base.h>
> +#include<target/target_core_tmr.h>
> +#include<target/target_core_transport.h>
> +
> +#include "iscsi_target_debug.h"
> +#include "iscsi_target_core.h"
> +#include "iscsi_target_parameters.h"
> +#include "iscsi_target_seq_pdu_list.h"
> +#include "iscsi_target_tq.h"
> +#include "iscsi_target_configfs.h"
> +#include "iscsi_target_datain_values.h"
> +#include "iscsi_target_erl0.h"
> +#include "iscsi_target_erl1.h"
> +#include "iscsi_target_erl2.h"
> +#include "iscsi_target_login.h"
> +#include "iscsi_target_tmr.h"
> +#include "iscsi_target_tpg.h"
> +#include "iscsi_target_util.h"
> +#include "iscsi_target.h"
> +#include "iscsi_target_device.h"
> +#include "iscsi_target_stat.h"
> +
> +LIST_HEAD(g_tiqn_list);
> +LIST_HEAD(g_np_list);
static
> +DEFINE_SPINLOCK(tiqn_lock);
> +DEFINE_SPINLOCK(np_lock);
static
> +
> +static struct idr tiqn_idr;
> +struct idr sess_idr;
> +struct mutex auth_id_lock;
> +spinlock_t sess_idr_lock;
> +
> +struct iscsi_global *iscsi_global;
I think that should be iscsit_global, because in other places you use
iscsit as a prefix.
Also you do not need to extern that in every header.
> +
> +struct kmem_cache *lio_cmd_cache;
> +struct kmem_cache *lio_qr_cache;
> +struct kmem_cache *lio_dr_cache;
> +struct kmem_cache *lio_ooo_cache;
> +struct kmem_cache *lio_r2t_cache;
> +
> +static void iscsit_rx_thread_wait_for_tcp(struct iscsi_conn *);
> +
> +static int iscsit_handle_immediate_data(struct iscsi_cmd *,
> + unsigned char *buf, u32);
> +static inline int iscsit_send_data_in(struct iscsi_cmd *, struct iscsi_conn *,
> + struct se_unmap_sg *, int *);
> +static inline int iscsit_send_logout_response(struct iscsi_cmd *, struct iscsi_conn *);
> +static inline int iscsit_send_nopin_response(struct iscsi_cmd *, struct iscsi_conn *);
> +static inline int iscsit_send_status(struct iscsi_cmd *, struct iscsi_conn *);
> +static int iscsit_send_task_mgt_rsp(struct iscsi_cmd *, struct iscsi_conn *);
> +static int iscsit_send_text_rsp(struct iscsi_cmd *, struct iscsi_conn *);
> +static int iscsit_send_reject(struct iscsi_cmd *, struct iscsi_conn *);
> +static int iscsit_logout_post_handler(struct iscsi_cmd *, struct iscsi_conn *);
> +
You do not need some of these to be forward declared, because they are
used after they are implemented.
Some of them could be trivially moved around so you do not have to
forward declare them.
Some are needed though,
> +/*
> + * Note that IQN formatting is expected to be done in userspace, and
> + * no explict IQN format checks are done here.
> + */
> +struct iscsi_tiqn *iscsit_add_tiqn(unsigned char *buf, int *ret)
> +{
> +
> + spin_lock(&tiqn_lock);
> + idr_get_new(&tiqn_idr, NULL,&tiqn->tiqn_index);
Need to check the return value.
> + list_add_tail(&tiqn->tiqn_list,&g_tiqn_list);
> + spin_unlock(&tiqn_lock);
> +
> + printk(KERN_INFO "CORE[0] - Added iSCSI Target IQN: %s\n", tiqn->tiqn);
> +
> + return tiqn;
> +
> +}
> +
> +void __iscsit_del_tiqn(struct iscsi_tiqn *tiqn)
Should be static or rolled up into iscsit_del_tiqn.
> +{
> + spin_lock(&tiqn_lock);
> + list_del(&tiqn->tiqn_list);
> + idr_remove(&tiqn_idr, tiqn->tiqn_index);
> + spin_unlock(&tiqn_lock);
> +
> + printk(KERN_INFO "CORE[0] - Deleted iSCSI Target IQN: %s\n",
> + tiqn->tiqn);
> + kfree(tiqn);
> +}
> +
> +static void iscsit_wait_for_tiqn(struct iscsi_tiqn *tiqn)
> +{
> + /*
> + * Wait for accesses to said struct iscsi_tiqn to end.
> + */
> + spin_lock(&tiqn->tiqn_state_lock);
> + while (tiqn->tiqn_access_count != 0) {
> + spin_unlock(&tiqn->tiqn_state_lock);
> + msleep(10);
> + spin_lock(&tiqn->tiqn_state_lock);
> + }
> + spin_unlock(&tiqn->tiqn_state_lock);
> +}
> +
> +void iscsit_del_tiqn(struct iscsi_tiqn *tiqn)
> +{
> + /*
> + * iscsit_set_tiqn_shutdown sets tiqn->tiqn_state = TIQN_STATE_SHUTDOWN
> + * while holding tiqn->tiqn_state_lock. This means that all subsequent
> + * attempts to access this struct iscsi_tiqn will fail from both transport
> + * fabric and control code paths.
> + */
> + if (iscsit_set_tiqn_shutdown(tiqn)< 0) {
> + printk(KERN_ERR "iscsit_set_tiqn_shutdown() failed\n");
> + return;
> + }
> +
> + iscsit_wait_for_tiqn(tiqn);
> + __iscsit_del_tiqn(tiqn);
> +}
> +
> +int iscsit_access_np(struct iscsi_np *np, struct iscsi_portal_group *tpg)
> +{
> + int ret;
> + /*
> + * Determine if the network portal is accepting storage traffic.
> + */
> + spin_lock_bh(&np->np_thread_lock);
> + if (np->np_thread_state != ISCSI_NP_THREAD_ACTIVE) {
> + spin_unlock_bh(&np->np_thread_lock);
> + return -1;
> + }
> + if (np->np_login_tpg) {
> + printk(KERN_ERR "np->np_login_tpg() is not NULL!\n");
> + spin_unlock_bh(&np->np_thread_lock);
> + return -1;
> + }
> + spin_unlock_bh(&np->np_thread_lock);
> + /*
> + * Determine if the portal group is accepting storage traffic.
> + */
> + spin_lock_bh(&tpg->tpg_state_lock);
> + if (tpg->tpg_state != TPG_STATE_ACTIVE) {
> + spin_unlock_bh(&tpg->tpg_state_lock);
> + return -1;
> + }
> + spin_unlock_bh(&tpg->tpg_state_lock);
> +
Do you mean to duplicate this check below? Does not seem like a very
useful optimization.
> + /*
> + * Here we serialize access across the TIQN+TPG Tuple.
> + */
> + ret = mutex_lock_interruptible(&tpg->np_login_lock);
> + if ((ret != 0) || signal_pending(current))
> + return -1;
> +
> + spin_lock_bh(&tpg->tpg_state_lock);
> + if (tpg->tpg_state != TPG_STATE_ACTIVE) {
> + spin_unlock_bh(&tpg->tpg_state_lock);
> + return -1;
> + }
> + spin_unlock_bh(&tpg->tpg_state_lock);
> +
> + spin_lock_bh(&np->np_thread_lock);
> + np->np_login_tpg = tpg;
> + spin_unlock_bh(&np->np_thread_lock);
> +
> + return 0;
> +}
> +}
> +
> +static struct iscsi_np *iscsit_get_np(
> + unsigned char *ipv6,
> + u32 ipv4,
> + u16 port,
> + int af_inet,
> + int network_transport)
> +{
> + struct iscsi_np *np;
> + int ip_match = 0;
> +
> + spin_lock_bh(&np_lock);
> + list_for_each_entry(np,&g_np_list, np_list) {
> + spin_lock(&np->np_thread_lock);
> + if (np->np_thread_state != ISCSI_NP_THREAD_ACTIVE) {
> + spin_unlock(&np->np_thread_lock);
> + continue;
> + }
> +
> + if (af_inet == AF_INET6) {
> + if (!strcmp(&np->np_ip[0],&ipv6[0]))
> + ip_match = 1;
> + } else {
> + if (np->np_ipv4 == ipv4)
> + ip_match = 1;
> + }
> +
> + if ((ip_match == 1)&& (np->np_port == port)&&
> + (np->np_network_transport == network_transport)) {
> + /*
> + * Increment the np_exports reference count now to
> + * prevent iscsit_del_np() below from being called
> + * while iscsi_tpg_add_network_portal() is called.
> + */
> + np->np_exports++;
> + spin_unlock(&np->np_thread_lock);
> + spin_unlock_bh(&np_lock);
> + return np;
> + }
> + spin_unlock(&np->np_thread_lock);
> + }
> + spin_unlock_bh(&np_lock);
> +
> + return NULL;
> +}
> +
> +struct iscsi_np *iscsit_add_np(
> + struct iscsi_np_addr *np_addr,
> + int network_transport,
> + int af_inet)
> +{
> + struct iscsi_np *np;
> + int ret;
> + /*
> + * Locate the existing struct iscsi_np if already active..
> + */
> + np = iscsit_get_np(np_addr->np_ipv6, np_addr->np_ipv4,
> + np_addr->np_port, af_inet, network_transport);
I thought you were going to sync up naming and types? np_addr->np_ipv6
is a text representation but np_addr->np_ipv4 is a binary one.
Why don't you put the address in a sockaddr when you get it from
configfs and then use that internally? Can keep np_ip for the text
representation buffer though to make it easier to print and use for
sendtargets.
> + if (np)
> + return np;
> +
> + np = kzalloc(sizeof(struct iscsi_np), GFP_KERNEL);
> + if (!np) {
> + printk(KERN_ERR "Unable to allocate memory for struct iscsi_np\n");
> + return ERR_PTR(-ENOMEM);
> + }
> +
> + np->np_flags |= NPF_IP_NETWORK;
> + if (af_inet == AF_INET6) {
> + snprintf(np->np_ip, IPV6_ADDRESS_SPACE, "%s", np_addr->np_ipv6);
> + } else {
> + sprintf(np->np_ip, "%u.%u.%u.%u",
%pI4
> + ((np_addr->np_ipv4>> 24)& 0xff),
> + ((np_addr->np_ipv4>> 16)& 0xff),
> + ((np_addr->np_ipv4>> 8)& 0xff),
> + np_addr->np_ipv4& 0xff);
> + np->np_ipv4 = np_addr->np_ipv4;
> + }
> +
> + np->np_port = np_addr->np_port;
> + np->np_network_transport = network_transport;
> + spin_lock_init(&np->np_thread_lock);
> + init_completion(&np->np_restart_comp);
> + INIT_LIST_HEAD(&np->np_list);
> +
> + ret = iscsi_target_setup_login_socket(np, af_inet);
> + if (ret != 0) {
> + kfree(np);
> + return ERR_PTR(ret);
> + }
> +
> + np->np_thread = kthread_run(iscsi_target_login_thread, np, "iscsi_np");
Did someone give some sort of comment about workqueues already?
> + if (IS_ERR(np->np_thread)) {
> + printk(KERN_ERR "Unable to create kthread: iscsi_np\n");
> + ret = PTR_ERR(np->np_thread);
> + kfree(np);
> + return ERR_PTR(ret);
> + }
> + /*
> + * Increment the np_exports reference count now to prevent
> + * iscsit_del_np() below from being run while a new call to
> + * iscsi_tpg_add_network_portal() for a matching iscsi_np is
> + * active. We don't need to hold np->np_thread_lock at this
> + * point because iscsi_np has not been added to g_np_list yet.
> + */
> + np->np_exports = 1;
> +
> + spin_lock_bh(&np_lock);
> + list_add_tail(&np->np_list,&g_np_list);
> + spin_unlock_bh(&np_lock);
> +
> + printk(KERN_INFO "CORE[0] - Added Network Portal: %s:%hu on %s on"
> + " network device: %s\n", np->np_ip, np->np_port,
> + (np->np_network_transport == ISCSI_TCP) ?
> + "TCP" : "SCTP", (strlen(np->np_net_dev)) ?
> + (char *)np->np_net_dev : "None");
> +
> + return np;
> +}
> +
> +int iscsit_del_np(struct iscsi_np *np)
> +{
> + spin_lock_bh(&np->np_thread_lock);
> + if (!(--np->np_exports == 0)) {
> + spin_unlock_bh(&np->np_thread_lock);
> + return 0;
> + }
> + np->np_thread_state = ISCSI_NP_THREAD_SHUTDOWN;
> + spin_unlock_bh(&np->np_thread_lock);
> +
> + if (np->np_thread) {
> + /*
> + * We need to send the signal to wakeup Linux/Net
> + * which may be sleeping in sock_accept()..
> + */
> + send_sig(SIGINT, np->np_thread, 1);
> + kthread_stop(np->np_thread);
> + }
> + iscsit_del_np_comm(np);
> +
> + spin_lock_bh(&np_lock);
> + list_del(&np->np_list);
> + spin_unlock_bh(&np_lock);
> +
> + printk(KERN_INFO "CORE[0] - Removed Network Portal: %s:%hu on %s on"
> + " network device: %s\n", np->np_ip, np->np_port,
> + (np->np_network_transport == ISCSI_TCP) ?
> + "TCP" : "SCTP", (strlen(np->np_net_dev)) ?
> + (char *)np->np_net_dev : "None");
Not now, but in the future you should add some nicer logging macros. It
seems you duplicate stuff, and sometimes there is logging without a
connection or session identifier so you do not know where it came from.
> +
> + kfree(np);
> + return 0;
> +}
> +
> +static int __init iscsi_target_init_module(void)
> +{
> + int ret = 0;
> +
> + printk(KERN_INFO "RisingTide Systems iSCSI-Target "ISCSI_VERSION"\n");
In general code like this we do not normally add company names. There is
not even a open-iscsi initiator string in the kernel (except for mailing
list, maintianer/auther and copyright stuff). Just "iSCSI Initiator over
TCP/IP". Or for fcoe we do not have open-fcoe or intel's fcoe, we just
have "FCoE Driver".
> +
> + iscsi_global = kzalloc(sizeof(struct iscsi_global), GFP_KERNEL);
> + if (!iscsi_global) {
> + printk(KERN_ERR "Unable to allocate memory for iscsi_global\n");
> + return -1;
> + }
> + mutex_init(&auth_id_lock);
> + spin_lock_init(&sess_idr_lock);
> + idr_init(&tiqn_idr);
> + idr_init(&sess_idr);
> +
> + ret = iscsi_target_register_configfs();
> + if (ret< 0)
> + goto out;
> +
> + ret = iscsi_thread_set_init();
> + if (ret< 0)
> + goto configfs_out;
> +
> + if (iscsi_allocate_thread_sets(TARGET_THREAD_SET_COUNT) !=
> + TARGET_THREAD_SET_COUNT) {
> + printk(KERN_ERR "iscsi_allocate_thread_sets() returned"
> + " unexpected value!\n");
> + goto ts_out1;
> + }
> +
> + lio_cmd_cache = kmem_cache_create("lio_cmd_cache",
> + sizeof(struct iscsi_cmd), __alignof__(struct iscsi_cmd),
> + 0, NULL);
> + if (!lio_cmd_cache) {
> + printk(KERN_ERR "Unable to kmem_cache_create() for"
> + " lio_cmd_cache\n");
> + goto ts_out2;
> + }
> +
> + lio_qr_cache = kmem_cache_create("lio_qr_cache",
> + sizeof(struct iscsi_queue_req),
> + __alignof__(struct iscsi_queue_req), 0, NULL);
> + if (!lio_qr_cache) {
> + printk(KERN_ERR "nable to kmem_cache_create() for"
> + " lio_qr_cache\n");
> + goto cmd_out;
> + }
> +
> + lio_dr_cache = kmem_cache_create("lio_dr_cache",
> + sizeof(struct iscsi_datain_req),
> + __alignof__(struct iscsi_datain_req), 0, NULL);
> + if (!lio_dr_cache) {
> + printk(KERN_ERR "Unable to kmem_cache_create() for"
> + " lio_dr_cache\n");
> + goto qr_out;
> + }
> +
> + lio_ooo_cache = kmem_cache_create("lio_ooo_cache",
> + sizeof(struct iscsi_ooo_cmdsn),
> + __alignof__(struct iscsi_ooo_cmdsn), 0, NULL);
> + if (!lio_ooo_cache) {
> + printk(KERN_ERR "Unable to kmem_cache_create() for"
> + " lio_ooo_cache\n");
> + goto dr_out;
> + }
> +
> + lio_r2t_cache = kmem_cache_create("lio_r2t_cache",
> + sizeof(struct iscsi_r2t), __alignof__(struct iscsi_r2t),
> + 0, NULL);
> + if (!lio_r2t_cache) {
> + printk(KERN_ERR "Unable to kmem_cache_create() for"
> + " lio_r2t_cache\n");
> + goto ooo_out;
> + }
> +
> + if (iscsit_load_discovery_tpg()< 0)
> + goto r2t_out;
> +
> + printk("Loading Complete.\n");
Is this a debug printk?
> +
> + return ret;
> +r2t_out:
> + kmem_cache_destroy(lio_r2t_cache);
> +ooo_out:
> + kmem_cache_destroy(lio_ooo_cache);
> +dr_out:
> + kmem_cache_destroy(lio_dr_cache);
> +qr_out:
> + kmem_cache_destroy(lio_qr_cache);
> +cmd_out:
> + kmem_cache_destroy(lio_cmd_cache);
> +ts_out2:
> + iscsi_deallocate_thread_sets();
> +ts_out1:
> + iscsi_thread_set_free();
> +configfs_out:
> + iscsi_target_deregister_configfs();
> +out:
> + kfree(iscsi_global);
> + iscsi_global = NULL;
No need to set to NULL. You are pretty screwed if you need it.
> +
> + return -1;
I think we normally return a errno.h type of value from module_init.
> +}
> +
> +static void __exit iscsi_target_cleanup_module(void)
> +{
> + iscsi_deallocate_thread_sets();
> + iscsi_thread_set_free();
> + iscsit_release_discovery_tpg();
> + kmem_cache_destroy(lio_cmd_cache);
> + kmem_cache_destroy(lio_qr_cache);
> + kmem_cache_destroy(lio_dr_cache);
> + kmem_cache_destroy(lio_ooo_cache);
> + kmem_cache_destroy(lio_r2t_cache);
> +
> + iscsi_target_deregister_configfs();
> +
> + kfree(iscsi_global);
> + printk(KERN_INFO "Unloading Complete.\n");
> +}
> +
> +int iscsit_add_reject(
> + u8 reason,
> + int fail_conn,
> + unsigned char *buf,
> + struct iscsi_conn *conn)
> +{
> + struct iscsi_cmd *cmd;
> + struct iscsi_reject *hdr;
> + int ret;
> +
> + cmd = iscsit_allocate_cmd(conn);
> + if (!cmd)
> + return -1;
> +
> + cmd->iscsi_opcode = ISCSI_OP_REJECT;
> + if (fail_conn)
> + cmd->cmd_flags |= ICF_REJECT_FAIL_CONN;
> +
> + hdr = (struct iscsi_reject *) cmd->pdu;
> + hdr->reason = reason;
> +
> + cmd->buf_ptr = kzalloc(ISCSI_HDR_LEN, GFP_ATOMIC);
Use GFP_KERNEL. If you can't sleep the wait below is wrong.
> + if (!cmd->buf_ptr) {
> + printk(KERN_ERR "Unable to allocate memory for cmd->buf_ptr\n");
> + iscsit_release_cmd(cmd);
> + return -1;
> + }
> + memcpy(cmd->buf_ptr, buf, ISCSI_HDR_LEN);
> +
> + spin_lock_bh(&conn->cmd_lock);
> + list_add_tail(&cmd->i_list,&conn->conn_cmd_list);
> + spin_unlock_bh(&conn->cmd_lock);
> +
> + cmd->i_state = ISTATE_SEND_REJECT;
> + iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
> +
> + ret = wait_for_completion_interruptible(&cmd->reject_comp);
> + if (ret != 0)
> + return -1;
> +
> + return (!fail_conn) ? 0 : -1;
> +}
> +
> +int iscsit_add_reject_from_cmd(
> + u8 reason,
> + int fail_conn,
> + int add_to_conn,
> + unsigned char *buf,
> + struct iscsi_cmd *cmd)
> +{
> + struct iscsi_conn *conn;
> + struct iscsi_reject *hdr;
> + int ret;
> +
> + if (!cmd->conn) {
> + printk(KERN_ERR "cmd->conn is NULL for ITT: 0x%08x\n",
> + cmd->init_task_tag);
> + return -1;
> + }
> + conn = cmd->conn;
> +
> + cmd->iscsi_opcode = ISCSI_OP_REJECT;
> + if (fail_conn)
> + cmd->cmd_flags |= ICF_REJECT_FAIL_CONN;
> +
> + hdr = (struct iscsi_reject *) cmd->pdu;
> + hdr->reason = reason;
> +
> + cmd->buf_ptr = kzalloc(ISCSI_HDR_LEN, GFP_ATOMIC);
Same as above. You do not want to use GFP_ATOMIC when you do not have
to. It can fail more easily than GFP_NOIO and GFP_KERNEL.
Could you check your other IO allocations. I think you can make
functions like iscsit_allocate_cmd take a gfp_t.
> + if (!cmd->buf_ptr) {
> + printk(KERN_ERR "Unable to allocate memory for cmd->buf_ptr\n");
> + iscsit_release_cmd(cmd);
> + return -1;
> + }
> + memcpy(cmd->buf_ptr, buf, ISCSI_HDR_LEN);
> +
> + if (add_to_conn) {
> + spin_lock_bh(&conn->cmd_lock);
> + list_add_tail(&cmd->i_list,&conn->conn_cmd_list);
> + spin_unlock_bh(&conn->cmd_lock);
> + }
> +
> + cmd->i_state = ISTATE_SEND_REJECT;
> + iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
> +
> + ret = wait_for_completion_interruptible(&cmd->reject_comp);
> + if (ret != 0)
> + return -1;
> +
> + return (!fail_conn) ? 0 : -1;
> +}
> +
> +static inline int iscsit_handle_scsi_cmd(
I thought the kernel coding style rules had something about not using
inline when functions are static and only used once. Compilers do the
right thing for you already? Could you check your other inline uses.
> + struct iscsi_conn *conn,
> + unsigned char *buf)
> +{
> +#if 0
> + if (!(hdr->flags& ISCSI_FLAG_CMD_FINAL)&&
> + (hdr->flags& ISCSI_FLAG_CMD_WRITE)&& conn->sess->sess_ops->InitialR2T) {
> + printk(KERN_ERR "ISCSI_FLAG_CMD_FINAL is not Set and"
> + " ISCSI_FLAG_CMD_WRITE Bit and InitialR2T=Yes,"
> + " protocol error\n");
> + return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
> + buf, conn);
> + }
> +#endif
??
> + if (ret == PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES)
Is PYX the target this is based on or something else?
> +
> +static inline int iscsit_handle_snack(
snacks om nom nom nom nom
> +
> +static int iscsit_handle_immediate_data(
> +
> + crypto_hash_init(&conn->conn_rx_hash);
> +
> + while (counter> 0) {
> + sg_init_one(&sg, iov_ptr->iov_base,
> + iov_ptr->iov_len);
> + crypto_hash_update(&conn->conn_rx_hash,&sg,
> + iov_ptr->iov_len);
> +
> + TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %zu"
> + " bytes, CRC 0x%08x\n", iov_ptr->iov_len, data_crc);
> + counter -= iov_ptr->iov_len;
> + iov_ptr++;
> + }
> +
> + if (padding) {
> + sg_init_one(&sg, (u8 *)&pad_bytes, padding);
> + crypto_hash_update(&conn->conn_rx_hash,&sg,
> + padding);
> + TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %d"
> + " bytes of padding, CRC 0x%08x\n", padding, data_crc);
> + }
> + crypto_hash_final(&conn->conn_rx_hash, (u8 *)&data_crc);
It seems you can could make this a function. I keep seeing the hash and
padding code over and over.
> +
> + if (checksum != data_crc) {
> + printk(KERN_ERR "ImmediateData CRC32C DataDigest 0x%08x"
> + " does not match computed 0x%08x\n", checksum,
> + data_crc);
> +
> + if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
> + printk(KERN_ERR "Unable to recover from"
> + " Immediate Data digest failure while"
> + " in ERL=0.\n");
> + iscsit_add_reject_from_cmd(
> + ISCSI_REASON_DATA_DIGEST_ERROR,
> + 1, 0, buf, cmd);
> + return IMMEDIDATE_DATA_CANNOT_RECOVER;
> + } else {
> + iscsit_add_reject_from_cmd(
> + ISCSI_REASON_DATA_DIGEST_ERROR,
> + 0, 0, buf, cmd);
> + return IMMEDIDATE_DATA_ERL1_CRC_FAILURE;
> + }
> + } else {
> + TRACE(TRACE_DIGEST, "Got CRC32C DataDigest 0x%08x for"
> + " %u bytes of Immediate Data\n", checksum,
> + length);
> + }
> + }
> +
> + cmd->write_data_done += length;
> +
> + if (cmd->write_data_done == cmd->data_length) {
> + spin_lock_bh(&cmd->istate_lock);
> + cmd->cmd_flags |= ICF_GOT_LAST_DATAOUT;
> + cmd->i_state = ISTATE_RECEIVED_LAST_DATAOUT;
> + spin_unlock_bh(&cmd->istate_lock);
> + }
> +
> + return IMMEDIDATE_DATA_NORMAL_OPERATION;
> +}
> +
> +int iscsit_send_async_msg(
> + struct iscsi_conn *conn,
> + u16 cid,
> + u8 async_event,
> + u8 async_vcode)
> +{
> + u8 iscsi_hdr[ISCSI_HDR_LEN+CRC_LEN];
> + u32 tx_send = ISCSI_HDR_LEN, tx_sent = 0;
> + struct iscsi_async *hdr;
> + struct kvec iov;
> + struct scatterlist sg;
> +
> + memset(&iov, 0, sizeof(struct kvec));
> + memset(&iscsi_hdr, 0, ISCSI_HDR_LEN+CRC_LEN);
> +
> + hdr = (struct iscsi_async *)&iscsi_hdr;
> + hdr->opcode = ISCSI_OP_ASYNC_EVENT;
> + hdr->flags |= ISCSI_FLAG_CMD_FINAL;
> + hton24(hdr->dlength, 0);
> + put_unaligned_le64(0,&hdr->lun[0]);
> + put_unaligned_be64(0xffffffffffffffff,&hdr->rsvd4[0]);
> + hdr->statsn = cpu_to_be32(conn->stat_sn++);
> + spin_lock(&conn->sess->cmdsn_lock);
> + hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
> + hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
> + spin_unlock(&conn->sess->cmdsn_lock);
> + hdr->async_event = async_event;
> + hdr->async_vcode = async_vcode;
> +
> + switch (async_event) {
> + case ISCSI_ASYNC_MSG_SCSI_EVENT:
> + printk(KERN_ERR "ISCSI_ASYNC_MSG_SCSI_EVENT: not supported yet.\n");
> + return -1;
> + case ISCSI_ASYNC_MSG_REQUEST_LOGOUT:
> + TRACE(TRACE_STATE, "Moving to"
> + " TARG_CONN_STATE_LOGOUT_REQUESTED.\n");
> + conn->conn_state = TARG_CONN_STATE_LOGOUT_REQUESTED;
> + hdr->param1 = 0;
> + hdr->param2 = 0;
> + hdr->param3 = cpu_to_be16(SECONDS_FOR_ASYNC_LOGOUT);
> + break;
> + case ISCSI_ASYNC_MSG_DROPPING_CONNECTION:
> + hdr->param1 = cpu_to_be16(cid);
> + hdr->param2 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Wait);
> + hdr->param3 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Retain);
> + break;
> + case ISCSI_ASYNC_MSG_DROPPING_ALL_CONNECTIONS:
> + hdr->param1 = 0;
> + hdr->param2 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Wait);
> + hdr->param3 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Retain);
> + break;
> + case ISCSI_ASYNC_MSG_PARAM_NEGOTIATION:
> + hdr->param1 = 0;
> + hdr->param2 = 0;
> + hdr->param3 = cpu_to_be16(SECONDS_FOR_ASYNC_TEXT);
> + break;
> + case ISCSI_ASYNC_MSG_VENDOR_SPECIFIC:
> + printk(KERN_ERR "ISCSI_ASYNC_MSG_VENDOR_SPECIFIC not"
> + " supported yet.\n");
> + return -1;
> + default:
> + printk(KERN_ERR "Unknown AsycnEvent 0x%02x, protocol"
> + " error.\n", async_event);
> + return -1;
> + }
> +
> + iov.iov_base =&iscsi_hdr;
> + iov.iov_len = ISCSI_HDR_LEN;
> +
> + if (conn->conn_ops->HeaderDigest) {
> + u32 *header_digest = (u32 *)&iscsi_hdr[ISCSI_HDR_LEN];
> +
> + crypto_hash_init(&conn->conn_tx_hash);
> +
> + sg_init_one(&sg, (u8 *)&iscsi_hdr, ISCSI_HDR_LEN);
> + crypto_hash_update(&conn->conn_tx_hash,&sg, ISCSI_HDR_LEN);
> +
> + crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
> +
I think you could make this a function too.
> + iov.iov_len += CRC_LEN;
> + tx_send += CRC_LEN;
> + TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest for Async"
> + " Msg PDU 0x%08x\n", *header_digest);
> + }
> +
> + TRACE(TRACE_ISCSI, "Built Async Message StatSN: 0x%08x, AsyncEvent:"
> + " 0x%02x, P1: 0x%04x, P2: 0x%04x, P3: 0x%04x\n",
> + ntohl(hdr->statsn), hdr->async_event, ntohs(hdr->param1),
> + ntohs(hdr->param2), ntohs(hdr->param3));
> +
> + tx_sent = tx_data(conn,&iov, 1, tx_send);
> + if (tx_sent != tx_send) {
> + printk(KERN_ERR "tx_data returned %d expecting %d\n",
> + tx_sent, tx_send);
> + return -1;
> + }
> +
> + if (async_event == ISCSI_ASYNC_MSG_REQUEST_LOGOUT) {
> + wait_for_completion_timeout(&conn->sess->async_msg_comp,
> + SECONDS_FOR_ASYNC_LOGOUT * HZ);
> +
> + if (conn->conn_state == TARG_CONN_STATE_LOGOUT_REQUESTED) {
> + printk(KERN_ERR "Asynchronous message timer expired"
> + " without receiving a logout request, dropping"
> + " iSCSI session.\n");
> + iscsit_send_async_msg(conn, 0,
> + ISCSI_ASYNC_MSG_DROPPING_ALL_CONNECTIONS, 0);
> + iscsit_free_session(conn->sess);
> + }
> + }
> + return 0;
> +}
> +
> +/*
> + * Called with sess->conn_lock held.
> + */
> +/* #warning iscsi_build_conn_drop_async_message() only sends out on connections
> + with active network interface */
> +static void iscsit_build_conn_drop_async_message(struct iscsi_conn *conn)
> +{
> + struct iscsi_cmd *cmd;
> + struct iscsi_conn *conn_p;
Why do you switch between conn_p and conn? Do you use conn_p when you
use it for lists? Kind of strange. If it makes it easier I guess ok. If
it was just due to code rework fix it up.
> +
> +#ifdef CONFIG_SMP
> +
> +void iscsit_thread_get_cpumask(struct iscsi_conn *conn)
Did Christoph or someone already say something about using workqueues or
per cpu threads with nonblocking IO and running from the softirq on the
recv path? I am assuming so, and skipping this threading code.
> +
> +int iscsi_target_tx_thread(void *arg)
> +{
> + * Bump up the task_struct priority for RX/TX thread set pairs,
> + * and allow ourselves to be interrupted by SIGINT so that a
> + * connection recovery / failure event can be triggered externally.
> + */
> + set_user_nice(current, -20);
What is the reason for using -20 on the target side? On the initiator
side I thought people did it to avoid some sort of starvation bug/issue
but I do not think we have that problem on the target do we?
Did you do it for performance reasons? Does it help in current kernels
and by how much?
> +
> +MODULE_DESCRIPTION("RisingTide Systems iSCSI-Target Driver 4.x.x Release");
I don't think we put names like RisingTide in module descriptions
either. Put it in MODULE_AUTHOR info.
Version info would go in MODULE_VERSION.
> diff --git a/drivers/target/iscsi/iscsi_target_core.h b/drivers/target/iscsi/iscsi_target_core.h
> new file mode 100644
> index 0000000..20d094e
> --- /dev/null
> +++ b/drivers/target/iscsi/iscsi_target_core.h
> @@ -0,0 +1,881 @@
> +#ifndef ISCSI_TARGET_CORE_H
> +#define ISCSI_TARGET_CORE_H
> +
> +#include<linux/in.h>
> +#include<linux/configfs.h>
> +#include<net/sock.h>
> +#include<net/tcp.h>
> +#include<scsi/scsi_cmnd.h>
> +#include<target/target_core_base.h>
> +
> +#define ISCSI_VERSION "v4.1.0-rc1"
Should probably be ISCSIT_VERSION.
> +#define ISCSI_MAX_DATASN_MISSING_COUNT 16
> +#define ISCSI_TX_THREAD_TCP_TIMEOUT 2
> +#define ISCSI_RX_THREAD_TCP_TIMEOUT 2
> +#define SECONDS_FOR_ASYNC_LOGOUT 10
> +#define SECONDS_FOR_ASYNC_TEXT 10
> +#define IPV4_ADDRESS_SPACE 4
> +#define IPV4_BUF_SIZE 18
IPV4 ones not used.
> +#define ISCSI_HDR_LEN 48
Should go in iscsi_proto.h?
> +#define CRC_LEN 4
> +#define MAX_KEY_NAME_LENGTH 63
> +#define MAX_KEY_VALUE_LENGTH 255
Where did these come from? Were they iscsi rfc values?
> +#define WHITE_SPACE " \t\v\f\n\r"
> +
> +/* struct iscsi_node_attrib sanity values */
> +#define NA_DATAOUT_TIMEOUT 3
> +#define NA_DATAOUT_TIMEOUT_MAX 60
> +#define NA_DATAOUT_TIMEOUT_MIX 2
> +#define NA_DATAOUT_TIMEOUT_RETRIES 5
> +#define NA_DATAOUT_TIMEOUT_RETRIES_MAX 15
> +#define NA_DATAOUT_TIMEOUT_RETRIES_MIN 1
> +#define NA_NOPIN_TIMEOUT 5
> +#define NA_NOPIN_TIMEOUT_MAX 60
> +#define NA_NOPIN_TIMEOUT_MIN 3
> +#define NA_NOPIN_RESPONSE_TIMEOUT 5
> +#define NA_NOPIN_RESPONSE_TIMEOUT_MAX 60
> +#define NA_NOPIN_RESPONSE_TIMEOUT_MIN 3
> +#define NA_RANDOM_DATAIN_PDU_OFFSETS 0
> +#define NA_RANDOM_DATAIN_SEQ_OFFSETS 0
> +#define NA_RANDOM_R2T_OFFSETS 0
> +#define NA_DEFAULT_ERL 0
> +#define NA_DEFAULT_ERL_MAX 2
> +#define NA_DEFAULT_ERL_MIN 0
> +
> +/* struct iscsi_tpg_attrib sanity values */
> +#define TA_AUTHENTICATION 1
> +#define TA_LOGIN_TIMEOUT 15
> +#define TA_LOGIN_TIMEOUT_MAX 30
> +#define TA_LOGIN_TIMEOUT_MIN 5
> +#define TA_NETIF_TIMEOUT 2
> +#define TA_NETIF_TIMEOUT_MAX 15
> +#define TA_NETIF_TIMEOUT_MIN 2
> +#define TA_GENERATE_NODE_ACLS 0
> +#define TA_DEFAULT_CMDSN_DEPTH 16
> +#define TA_DEFAULT_CMDSN_DEPTH_MAX 512
> +#define TA_DEFAULT_CMDSN_DEPTH_MIN 1
> +#define TA_CACHE_DYNAMIC_ACLS 0
> +/* Enabled by default in demo mode (generic_node_acls=1) */
> +#define TA_DEMO_MODE_WRITE_PROTECT 1
> +/* Disabled by default in production mode w/ explict ACLs */
> +#define TA_PROD_MODE_WRITE_PROTECT 0
> +#define TA_CACHE_CORE_NPS 0
> +
> +enum tpg_np_network_transport_table {
> + ISCSI_TCP = 0,
> + ISCSI_SCTP_TCP = 1,
> + ISCSI_SCTP_UDP = 2,
> + ISCSI_IWARP_TCP = 3,
> + ISCSI_IWARP_SCTP = 4,
> + ISCSI_INFINIBAND = 5,
> +};
> +
> +/* RFC-3720 7.1.4 Standard Connection State Diagram for a Target */
> +enum target_conn_state_table {
> + TARG_CONN_STATE_FREE = 0x1,
> + TARG_CONN_STATE_XPT_UP = 0x3,
> + TARG_CONN_STATE_IN_LOGIN = 0x4,
> + TARG_CONN_STATE_LOGGED_IN = 0x5,
> + TARG_CONN_STATE_IN_LOGOUT = 0x6,
> + TARG_CONN_STATE_LOGOUT_REQUESTED = 0x7,
> + TARG_CONN_STATE_CLEANUP_WAIT = 0x8,
> +};
> +
> +/* RFC-3720 7.3.2 Session State Diagram for a Target */
> +enum target_sess_state_table {
> + TARG_SESS_STATE_FREE = 0x1,
> + TARG_SESS_STATE_ACTIVE = 0x2,
> + TARG_SESS_STATE_LOGGED_IN = 0x3,
> + TARG_SESS_STATE_FAILED = 0x4,
> + TARG_SESS_STATE_IN_CONTINUE = 0x5,
> +};
Should the two above be in iscsi_proto.h?
> +
> +struct iscsi_conn {
> +#define ISCSI_NETDEV_NAME_SIZE 12
> + char net_dev[ISCSI_NETDEV_NAME_SIZE];
Were you going to remove this?
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 40+ messages in thread* Re: [PATCH-v2 04/14] iscsi-target: Add iSCSI fabric support for target v4
2011-04-07 7:34 ` Mike Christie
@ 2011-05-09 16:22 ` Nicholas A. Bellinger
0 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-05-09 16:22 UTC (permalink / raw)
To: Mike Christie
Cc: linux-scsi, James Bottomley, Christoph Hellwig, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh
On Thu, 2011-04-07 at 02:34 -0500, Mike Christie wrote:
> On 03/22/2011 10:06 PM, Nicholas A. Bellinger wrote:
> > From: Nicholas Bellinger<nab@linux-iscsi.org>
> >
> > The RisingTide Systems iSCSI target module is a full featured in-kernel
> > software implementation of iSCSI target mode (RFC-3720) for the mainline
> > target v4 infrastructure code. More information can be found here:
> >
<SNIP>
> > Significant feedback for mainline code cleanups by Christoph Hellwig and
> > Mike Christie.
> >
Hey Mike and Co,
Sorry for the delayed response on your last review. My comments are
inline below along with a diffstat at the very bottom for the changes
based on your feedback pushed into lio-core-2.6.git/lio-4.1.
These changes will be included in PATCH-v3 series for iscsi-target going
out this week.
> > Signed-off-by: Nicholas A. Bellinger<nab@linux-iscsi.org>
> > ---
> > drivers/target/iscsi/iscsi_target.c | 5017 ++++++++++++++++++++++++++++++
> > drivers/target/iscsi/iscsi_target.h | 32 +
> > drivers/target/iscsi/iscsi_target_core.h | 881 ++++++
> > 3 files changed, 5930 insertions(+), 0 deletions(-)
> > create mode 100644 drivers/target/iscsi/iscsi_target.c
> > create mode 100644 drivers/target/iscsi/iscsi_target.h
> > create mode 100644 drivers/target/iscsi/iscsi_target_core.h
> >
> > diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
> > new file mode 100644
> > index 0000000..c81f5cc
> > --- /dev/null
> > +++ b/drivers/target/iscsi/iscsi_target.c
> > @@ -0,0 +1,5017 @@
> > +/*******************************************************************************
> > + * This file contains main functions related to the iSCSI Target Core Driver.
> > + *
> > + * © Copyright 2007-2011 RisingTide Systems LLC.
> > + *
> > + * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
> > + *
> > + * Author: Nicholas A. Bellinger<nab@linux-iscsi.org>
> > + *
> > + * This program is free software; you can redistribute it and/or modify
> > + * it under the terms of the GNU General Public License as published by
> > + * the Free Software Foundation; either version 2 of the License, or
> > + * (at your option) any later version.
> > + *
> > + * This program is distributed in the hope that it will be useful,
> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> > + * GNU General Public License for more details.
> > + ******************************************************************************/
> > +
> > +#include<linux/kthread.h>
> > +#include<linux/crypto.h>
> > +#include<linux/completion.h>
> > +#include<asm/unaligned.h>
> > +#include<scsi/scsi_device.h>
> > +#include<scsi/iscsi_proto.h>
> > +#include<target/target_core_base.h>
> > +#include<target/target_core_tmr.h>
> > +#include<target/target_core_transport.h>
> > +
> > +#include "iscsi_target_debug.h"
> > +#include "iscsi_target_core.h"
> > +#include "iscsi_target_parameters.h"
> > +#include "iscsi_target_seq_pdu_list.h"
> > +#include "iscsi_target_tq.h"
> > +#include "iscsi_target_configfs.h"
> > +#include "iscsi_target_datain_values.h"
> > +#include "iscsi_target_erl0.h"
> > +#include "iscsi_target_erl1.h"
> > +#include "iscsi_target_erl2.h"
> > +#include "iscsi_target_login.h"
> > +#include "iscsi_target_tmr.h"
> > +#include "iscsi_target_tpg.h"
> > +#include "iscsi_target_util.h"
> > +#include "iscsi_target.h"
> > +#include "iscsi_target_device.h"
> > +#include "iscsi_target_stat.h"
> > +
> > +LIST_HEAD(g_tiqn_list);
> > +LIST_HEAD(g_np_list);
>
> static
>
Done
> > +DEFINE_SPINLOCK(tiqn_lock);
> > +DEFINE_SPINLOCK(np_lock);
>
> static
>
>
Done
> > +
> > +static struct idr tiqn_idr;
> > +struct idr sess_idr;
> > +struct mutex auth_id_lock;
> > +spinlock_t sess_idr_lock;
> > +
> > +struct iscsi_global *iscsi_global;
>
>
> I think that should be iscsit_global, because in other places you use
> iscsit as a prefix.
>
Fair enough. Renamed all usage of *iscsi_global to *iscsit_global.
> Also you do not need to extern that in every header.
>
>
<nod> Added a single set of prototypes of these in iscsi_target.h, and
dropped the extern usage in the other headers.
>
> > +
> > +struct kmem_cache *lio_cmd_cache;
> > +struct kmem_cache *lio_qr_cache;
> > +struct kmem_cache *lio_dr_cache;
> > +struct kmem_cache *lio_ooo_cache;
> > +struct kmem_cache *lio_r2t_cache;
> > +
> > +static void iscsit_rx_thread_wait_for_tcp(struct iscsi_conn *);
> > +
> > +static int iscsit_handle_immediate_data(struct iscsi_cmd *,
> > + unsigned char *buf, u32);
> > +static inline int iscsit_send_data_in(struct iscsi_cmd *, struct iscsi_conn *,
> > + struct se_unmap_sg *, int *);
> > +static inline int iscsit_send_logout_response(struct iscsi_cmd *, struct iscsi_conn *);
> > +static inline int iscsit_send_nopin_response(struct iscsi_cmd *, struct iscsi_conn *);
> > +static inline int iscsit_send_status(struct iscsi_cmd *, struct iscsi_conn *);
> > +static int iscsit_send_task_mgt_rsp(struct iscsi_cmd *, struct iscsi_conn *);
> > +static int iscsit_send_text_rsp(struct iscsi_cmd *, struct iscsi_conn *);
> > +static int iscsit_send_reject(struct iscsi_cmd *, struct iscsi_conn *);
> > +static int iscsit_logout_post_handler(struct iscsi_cmd *, struct iscsi_conn *);
> > +
>
>
> You do not need some of these to be forward declared, because they are
> used after they are implemented.
>
> Some of them could be trivially moved around so you do not have to
> forward declare them.
>
> Some are needed though,
>
>
<nod>, dropped the unnecessary + trivally moved forward declarations
here.
>
>
>
> > +/*
> > + * Note that IQN formatting is expected to be done in userspace, and
> > + * no explict IQN format checks are done here.
> > + */
> > +struct iscsi_tiqn *iscsit_add_tiqn(unsigned char *buf, int *ret)
> > +{
>
>
>
> > +
> > + spin_lock(&tiqn_lock);
> > + idr_get_new(&tiqn_idr, NULL,&tiqn->tiqn_index);
>
>
> Need to check the return value.
>
>
Fixed
>
> > + list_add_tail(&tiqn->tiqn_list,&g_tiqn_list);
> > + spin_unlock(&tiqn_lock);
> > +
> > + printk(KERN_INFO "CORE[0] - Added iSCSI Target IQN: %s\n", tiqn->tiqn);
> > +
> > + return tiqn;
> > +
> > +}
> > +
> > +void __iscsit_del_tiqn(struct iscsi_tiqn *tiqn)
>
> Should be static or rolled up into iscsit_del_tiqn.
>
>
Rolled into iscsit_del_tiqn.
>
> > +{
> > + spin_lock(&tiqn_lock);
> > + list_del(&tiqn->tiqn_list);
> > + idr_remove(&tiqn_idr, tiqn->tiqn_index);
> > + spin_unlock(&tiqn_lock);
> > +
> > + printk(KERN_INFO "CORE[0] - Deleted iSCSI Target IQN: %s\n",
> > + tiqn->tiqn);
> > + kfree(tiqn);
> > +}
> > +
> > +static void iscsit_wait_for_tiqn(struct iscsi_tiqn *tiqn)
> > +{
> > + /*
> > + * Wait for accesses to said struct iscsi_tiqn to end.
> > + */
> > + spin_lock(&tiqn->tiqn_state_lock);
> > + while (tiqn->tiqn_access_count != 0) {
> > + spin_unlock(&tiqn->tiqn_state_lock);
> > + msleep(10);
> > + spin_lock(&tiqn->tiqn_state_lock);
> > + }
> > + spin_unlock(&tiqn->tiqn_state_lock);
> > +}
> > +
> > +void iscsit_del_tiqn(struct iscsi_tiqn *tiqn)
> > +{
> > + /*
> > + * iscsit_set_tiqn_shutdown sets tiqn->tiqn_state = TIQN_STATE_SHUTDOWN
> > + * while holding tiqn->tiqn_state_lock. This means that all subsequent
> > + * attempts to access this struct iscsi_tiqn will fail from both transport
> > + * fabric and control code paths.
> > + */
> > + if (iscsit_set_tiqn_shutdown(tiqn)< 0) {
> > + printk(KERN_ERR "iscsit_set_tiqn_shutdown() failed\n");
> > + return;
> > + }
> > +
> > + iscsit_wait_for_tiqn(tiqn);
> > + __iscsit_del_tiqn(tiqn);
> > +}
> > +
> > +int iscsit_access_np(struct iscsi_np *np, struct iscsi_portal_group *tpg)
> > +{
> > + int ret;
> > + /*
> > + * Determine if the network portal is accepting storage traffic.
> > + */
> > + spin_lock_bh(&np->np_thread_lock);
> > + if (np->np_thread_state != ISCSI_NP_THREAD_ACTIVE) {
> > + spin_unlock_bh(&np->np_thread_lock);
> > + return -1;
> > + }
> > + if (np->np_login_tpg) {
> > + printk(KERN_ERR "np->np_login_tpg() is not NULL!\n");
> > + spin_unlock_bh(&np->np_thread_lock);
> > + return -1;
> > + }
> > + spin_unlock_bh(&np->np_thread_lock);
> > + /*
> > + * Determine if the portal group is accepting storage traffic.
> > + */
> > + spin_lock_bh(&tpg->tpg_state_lock);
> > + if (tpg->tpg_state != TPG_STATE_ACTIVE) {
> > + spin_unlock_bh(&tpg->tpg_state_lock);
> > + return -1;
> > + }
> > + spin_unlock_bh(&tpg->tpg_state_lock);
> > +
>
>
> Do you mean to duplicate this check below? Does not seem like a very
> useful optimization.
>
>
AFAICT the duplicate check is unnecessary. Dropped
>
> > + /*
> > + * Here we serialize access across the TIQN+TPG Tuple.
> > + */
> > + ret = mutex_lock_interruptible(&tpg->np_login_lock);
> > + if ((ret != 0) || signal_pending(current))
> > + return -1;
> > +
> > + spin_lock_bh(&tpg->tpg_state_lock);
> > + if (tpg->tpg_state != TPG_STATE_ACTIVE) {
> > + spin_unlock_bh(&tpg->tpg_state_lock);
> > + return -1;
> > + }
> > + spin_unlock_bh(&tpg->tpg_state_lock);
> > +
> > + spin_lock_bh(&np->np_thread_lock);
> > + np->np_login_tpg = tpg;
> > + spin_unlock_bh(&np->np_thread_lock);
> > +
> > + return 0;
> > +}
>
>
> > +}
> > +
> > +static struct iscsi_np *iscsit_get_np(
> > + unsigned char *ipv6,
> > + u32 ipv4,
> > + u16 port,
> > + int af_inet,
> > + int network_transport)
> > +{
> > + struct iscsi_np *np;
> > + int ip_match = 0;
> > +
> > + spin_lock_bh(&np_lock);
> > + list_for_each_entry(np,&g_np_list, np_list) {
> > + spin_lock(&np->np_thread_lock);
> > + if (np->np_thread_state != ISCSI_NP_THREAD_ACTIVE) {
> > + spin_unlock(&np->np_thread_lock);
> > + continue;
> > + }
> > +
> > + if (af_inet == AF_INET6) {
> > + if (!strcmp(&np->np_ip[0],&ipv6[0]))
> > + ip_match = 1;
> > + } else {
> > + if (np->np_ipv4 == ipv4)
> > + ip_match = 1;
> > + }
> > +
> > + if ((ip_match == 1)&& (np->np_port == port)&&
> > + (np->np_network_transport == network_transport)) {
> > + /*
> > + * Increment the np_exports reference count now to
> > + * prevent iscsit_del_np() below from being called
> > + * while iscsi_tpg_add_network_portal() is called.
> > + */
> > + np->np_exports++;
> > + spin_unlock(&np->np_thread_lock);
> > + spin_unlock_bh(&np_lock);
> > + return np;
> > + }
> > + spin_unlock(&np->np_thread_lock);
> > + }
> > + spin_unlock_bh(&np_lock);
> > +
> > + return NULL;
> > +}
> > +
> > +struct iscsi_np *iscsit_add_np(
> > + struct iscsi_np_addr *np_addr,
> > + int network_transport,
> > + int af_inet)
> > +{
> > + struct iscsi_np *np;
> > + int ret;
> > + /*
> > + * Locate the existing struct iscsi_np if already active..
> > + */
> > + np = iscsit_get_np(np_addr->np_ipv6, np_addr->np_ipv4,
> > + np_addr->np_port, af_inet, network_transport);
>
>
> I thought you were going to sync up naming and types? np_addr->np_ipv6
> is a text representation but np_addr->np_ipv4 is a binary one.
>
> Why don't you put the address in a sockaddr when you get it from
> configfs and then use that internally? Can keep np_ip for the text
> representation buffer though to make it easier to print and use for
> sendtargets.
>
>
Done.
Removed all struct iscsi_np_addr usage, and made iscsit_get_np(),
iscsit_add_np(), iscsi_target_setup_login_socket() accept a sockaddr
from iscsi_target_configfs.c:lio_target_call_addnptotpg().
>
>
> > + if (np)
> > + return np;
> > +
> > + np = kzalloc(sizeof(struct iscsi_np), GFP_KERNEL);
> > + if (!np) {
> > + printk(KERN_ERR "Unable to allocate memory for struct iscsi_np\n");
> > + return ERR_PTR(-ENOMEM);
> > + }
> > +
> > + np->np_flags |= NPF_IP_NETWORK;
> > + if (af_inet == AF_INET6) {
> > + snprintf(np->np_ip, IPV6_ADDRESS_SPACE, "%s", np_addr->np_ipv6);
> > + } else {
> > + sprintf(np->np_ip, "%u.%u.%u.%u",
>
>
>
> %pI4
>
Fixed
>
> > + ((np_addr->np_ipv4>> 24)& 0xff),
> > + ((np_addr->np_ipv4>> 16)& 0xff),
> > + ((np_addr->np_ipv4>> 8)& 0xff),
> > + np_addr->np_ipv4& 0xff);
> > + np->np_ipv4 = np_addr->np_ipv4;
> > + }
> > +
> > + np->np_port = np_addr->np_port;
> > + np->np_network_transport = network_transport;
> > + spin_lock_init(&np->np_thread_lock);
> > + init_completion(&np->np_restart_comp);
> > + INIT_LIST_HEAD(&np->np_list);
> > +
> > + ret = iscsi_target_setup_login_socket(np, af_inet);
> > + if (ret != 0) {
> > + kfree(np);
> > + return ERR_PTR(ret);
> > + }
> > +
> > + np->np_thread = kthread_run(iscsi_target_login_thread, np, "iscsi_np");
>
> Did someone give some sort of comment about workqueues already?
>
>
I don't think it has been mentioned wrt iscsi_target_login_thread()
context, but the main reason this is not happened yet is the dependancy
on signals.
I think converting this to a workqueue would be possible with some more
effort.
> > + if (IS_ERR(np->np_thread)) {
> > + printk(KERN_ERR "Unable to create kthread: iscsi_np\n");
> > + ret = PTR_ERR(np->np_thread);
> > + kfree(np);
> > + return ERR_PTR(ret);
> > + }
> > + /*
> > + * Increment the np_exports reference count now to prevent
> > + * iscsit_del_np() below from being run while a new call to
> > + * iscsi_tpg_add_network_portal() for a matching iscsi_np is
> > + * active. We don't need to hold np->np_thread_lock at this
> > + * point because iscsi_np has not been added to g_np_list yet.
> > + */
> > + np->np_exports = 1;
> > +
> > + spin_lock_bh(&np_lock);
> > + list_add_tail(&np->np_list,&g_np_list);
> > + spin_unlock_bh(&np_lock);
> > +
> > + printk(KERN_INFO "CORE[0] - Added Network Portal: %s:%hu on %s on"
> > + " network device: %s\n", np->np_ip, np->np_port,
> > + (np->np_network_transport == ISCSI_TCP) ?
> > + "TCP" : "SCTP", (strlen(np->np_net_dev)) ?
> > + (char *)np->np_net_dev : "None");
> > +
> > + return np;
> > +}
>
>
>
> > +
> > +int iscsit_del_np(struct iscsi_np *np)
> > +{
> > + spin_lock_bh(&np->np_thread_lock);
> > + if (!(--np->np_exports == 0)) {
> > + spin_unlock_bh(&np->np_thread_lock);
> > + return 0;
> > + }
> > + np->np_thread_state = ISCSI_NP_THREAD_SHUTDOWN;
> > + spin_unlock_bh(&np->np_thread_lock);
> > +
> > + if (np->np_thread) {
> > + /*
> > + * We need to send the signal to wakeup Linux/Net
> > + * which may be sleeping in sock_accept()..
> > + */
> > + send_sig(SIGINT, np->np_thread, 1);
> > + kthread_stop(np->np_thread);
> > + }
> > + iscsit_del_np_comm(np);
> > +
> > + spin_lock_bh(&np_lock);
> > + list_del(&np->np_list);
> > + spin_unlock_bh(&np_lock);
> > +
> > + printk(KERN_INFO "CORE[0] - Removed Network Portal: %s:%hu on %s on"
> > + " network device: %s\n", np->np_ip, np->np_port,
> > + (np->np_network_transport == ISCSI_TCP) ?
> > + "TCP" : "SCTP", (strlen(np->np_net_dev)) ?
> > + (char *)np->np_net_dev : "None");
>
>
> Not now, but in the future you should add some nicer logging macros. It
> seems you duplicate stuff, and sometimes there is logging without a
> connection or session identifier so you do not know where it came from.
>
>
<nod>, I would like to eventually convert this all over to dprink().
>
> > +
> > + kfree(np);
> > + return 0;
> > +}
> > +
> > +static int __init iscsi_target_init_module(void)
> > +{
> > + int ret = 0;
> > +
> > + printk(KERN_INFO "RisingTide Systems iSCSI-Target "ISCSI_VERSION"\n");
>
> In general code like this we do not normally add company names. There is
> not even a open-iscsi initiator string in the kernel (except for mailing
> list, maintianer/auther and copyright stuff). Just "iSCSI Initiator over
> TCP/IP". Or for fcoe we do not have open-fcoe or intel's fcoe, we just
> have "FCoE Driver".
>
>
<nod>, dropped
> > +
> > + iscsi_global = kzalloc(sizeof(struct iscsi_global), GFP_KERNEL);
> > + if (!iscsi_global) {
> > + printk(KERN_ERR "Unable to allocate memory for iscsi_global\n");
> > + return -1;
> > + }
> > + mutex_init(&auth_id_lock);
> > + spin_lock_init(&sess_idr_lock);
> > + idr_init(&tiqn_idr);
> > + idr_init(&sess_idr);
> > +
> > + ret = iscsi_target_register_configfs();
> > + if (ret< 0)
> > + goto out;
> > +
> > + ret = iscsi_thread_set_init();
> > + if (ret< 0)
> > + goto configfs_out;
> > +
> > + if (iscsi_allocate_thread_sets(TARGET_THREAD_SET_COUNT) !=
> > + TARGET_THREAD_SET_COUNT) {
> > + printk(KERN_ERR "iscsi_allocate_thread_sets() returned"
> > + " unexpected value!\n");
> > + goto ts_out1;
> > + }
> > +
> > + lio_cmd_cache = kmem_cache_create("lio_cmd_cache",
> > + sizeof(struct iscsi_cmd), __alignof__(struct iscsi_cmd),
> > + 0, NULL);
> > + if (!lio_cmd_cache) {
> > + printk(KERN_ERR "Unable to kmem_cache_create() for"
> > + " lio_cmd_cache\n");
> > + goto ts_out2;
> > + }
> > +
> > + lio_qr_cache = kmem_cache_create("lio_qr_cache",
> > + sizeof(struct iscsi_queue_req),
> > + __alignof__(struct iscsi_queue_req), 0, NULL);
> > + if (!lio_qr_cache) {
> > + printk(KERN_ERR "nable to kmem_cache_create() for"
> > + " lio_qr_cache\n");
> > + goto cmd_out;
> > + }
> > +
> > + lio_dr_cache = kmem_cache_create("lio_dr_cache",
> > + sizeof(struct iscsi_datain_req),
> > + __alignof__(struct iscsi_datain_req), 0, NULL);
> > + if (!lio_dr_cache) {
> > + printk(KERN_ERR "Unable to kmem_cache_create() for"
> > + " lio_dr_cache\n");
> > + goto qr_out;
> > + }
> > +
> > + lio_ooo_cache = kmem_cache_create("lio_ooo_cache",
> > + sizeof(struct iscsi_ooo_cmdsn),
> > + __alignof__(struct iscsi_ooo_cmdsn), 0, NULL);
> > + if (!lio_ooo_cache) {
> > + printk(KERN_ERR "Unable to kmem_cache_create() for"
> > + " lio_ooo_cache\n");
> > + goto dr_out;
> > + }
> > +
> > + lio_r2t_cache = kmem_cache_create("lio_r2t_cache",
> > + sizeof(struct iscsi_r2t), __alignof__(struct iscsi_r2t),
> > + 0, NULL);
> > + if (!lio_r2t_cache) {
> > + printk(KERN_ERR "Unable to kmem_cache_create() for"
> > + " lio_r2t_cache\n");
> > + goto ooo_out;
> > + }
> > +
> > + if (iscsit_load_discovery_tpg()< 0)
> > + goto r2t_out;
> > +
> > + printk("Loading Complete.\n");
>
> Is this a debug printk?
>
>
Removed this useless noise..
> > +
> > + return ret;
> > +r2t_out:
> > + kmem_cache_destroy(lio_r2t_cache);
> > +ooo_out:
> > + kmem_cache_destroy(lio_ooo_cache);
> > +dr_out:
> > + kmem_cache_destroy(lio_dr_cache);
> > +qr_out:
> > + kmem_cache_destroy(lio_qr_cache);
> > +cmd_out:
> > + kmem_cache_destroy(lio_cmd_cache);
> > +ts_out2:
> > + iscsi_deallocate_thread_sets();
> > +ts_out1:
> > + iscsi_thread_set_free();
> > +configfs_out:
> > + iscsi_target_deregister_configfs();
> > +out:
> > + kfree(iscsi_global);
> > + iscsi_global = NULL;
>
>
> No need to set to NULL. You are pretty screwed if you need it.
>
Dropped
> > +
> > + return -1;
>
>
> I think we normally return a errno.h type of value from module_init.
>
Changed to -ENOMEM
> > +}
> > +
> > +static void __exit iscsi_target_cleanup_module(void)
> > +{
> > + iscsi_deallocate_thread_sets();
> > + iscsi_thread_set_free();
> > + iscsit_release_discovery_tpg();
> > + kmem_cache_destroy(lio_cmd_cache);
> > + kmem_cache_destroy(lio_qr_cache);
> > + kmem_cache_destroy(lio_dr_cache);
> > + kmem_cache_destroy(lio_ooo_cache);
> > + kmem_cache_destroy(lio_r2t_cache);
> > +
> > + iscsi_target_deregister_configfs();
> > +
> > + kfree(iscsi_global);
> > + printk(KERN_INFO "Unloading Complete.\n");
> > +}
> > +
> > +int iscsit_add_reject(
> > + u8 reason,
> > + int fail_conn,
> > + unsigned char *buf,
> > + struct iscsi_conn *conn)
> > +{
> > + struct iscsi_cmd *cmd;
> > + struct iscsi_reject *hdr;
> > + int ret;
> > +
> > + cmd = iscsit_allocate_cmd(conn);
> > + if (!cmd)
> > + return -1;
> > +
> > + cmd->iscsi_opcode = ISCSI_OP_REJECT;
> > + if (fail_conn)
> > + cmd->cmd_flags |= ICF_REJECT_FAIL_CONN;
> > +
> > + hdr = (struct iscsi_reject *) cmd->pdu;
> > + hdr->reason = reason;
> > +
> > + cmd->buf_ptr = kzalloc(ISCSI_HDR_LEN, GFP_ATOMIC);
>
>
> Use GFP_KERNEL. If you can't sleep the wait below is wrong.
>
>
Converted to GFP_KERNEL
>
>
> > + if (!cmd->buf_ptr) {
> > + printk(KERN_ERR "Unable to allocate memory for cmd->buf_ptr\n");
> > + iscsit_release_cmd(cmd);
> > + return -1;
> > + }
> > + memcpy(cmd->buf_ptr, buf, ISCSI_HDR_LEN);
> > +
> > + spin_lock_bh(&conn->cmd_lock);
> > + list_add_tail(&cmd->i_list,&conn->conn_cmd_list);
> > + spin_unlock_bh(&conn->cmd_lock);
> > +
> > + cmd->i_state = ISTATE_SEND_REJECT;
> > + iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
> > +
> > + ret = wait_for_completion_interruptible(&cmd->reject_comp);
> > + if (ret != 0)
> > + return -1;
> > +
> > + return (!fail_conn) ? 0 : -1;
> > +}
> > +
> > +int iscsit_add_reject_from_cmd(
> > + u8 reason,
> > + int fail_conn,
> > + int add_to_conn,
> > + unsigned char *buf,
> > + struct iscsi_cmd *cmd)
> > +{
> > + struct iscsi_conn *conn;
> > + struct iscsi_reject *hdr;
> > + int ret;
> > +
> > + if (!cmd->conn) {
> > + printk(KERN_ERR "cmd->conn is NULL for ITT: 0x%08x\n",
> > + cmd->init_task_tag);
> > + return -1;
> > + }
> > + conn = cmd->conn;
> > +
> > + cmd->iscsi_opcode = ISCSI_OP_REJECT;
> > + if (fail_conn)
> > + cmd->cmd_flags |= ICF_REJECT_FAIL_CONN;
> > +
> > + hdr = (struct iscsi_reject *) cmd->pdu;
> > + hdr->reason = reason;
> > +
> > + cmd->buf_ptr = kzalloc(ISCSI_HDR_LEN, GFP_ATOMIC);
>
>
> Same as above. You do not want to use GFP_ATOMIC when you do not have
> to. It can fail more easily than GFP_NOIO and GFP_KERNEL.
>
>
Also converted to GFP_KERNEL
> Could you check your other IO allocations. I think you can make
> functions like iscsit_allocate_cmd take a gfp_t.
>
>
Done.
Note that iscsit_add_nopin() is the only caller of iscsit_allocate_cmd()
w/ GFP_ATOMIC called from softirq context for
conn->nopin_response_timer.
>
> > + if (!cmd->buf_ptr) {
> > + printk(KERN_ERR "Unable to allocate memory for cmd->buf_ptr\n");
> > + iscsit_release_cmd(cmd);
> > + return -1;
> > + }
> > + memcpy(cmd->buf_ptr, buf, ISCSI_HDR_LEN);
> > +
> > + if (add_to_conn) {
> > + spin_lock_bh(&conn->cmd_lock);
> > + list_add_tail(&cmd->i_list,&conn->conn_cmd_list);
> > + spin_unlock_bh(&conn->cmd_lock);
> > + }
> > +
> > + cmd->i_state = ISTATE_SEND_REJECT;
> > + iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
> > +
> > + ret = wait_for_completion_interruptible(&cmd->reject_comp);
> > + if (ret != 0)
> > + return -1;
> > +
> > + return (!fail_conn) ? 0 : -1;
> > +}
>
>
>
>
>
> > +
> > +static inline int iscsit_handle_scsi_cmd(
>
>
>
> I thought the kernel coding style rules had something about not using
> inline when functions are static and only used once. Compilers do the
> right thing for you already? Could you check your other inline uses.
>
>
>
<nod>, removed the unnecessary inlines here in iscsi_target.c and
aduited the other source files..
>
>
> > + struct iscsi_conn *conn,
> > + unsigned char *buf)
> > +{
>
>
>
> > +#if 0
> > + if (!(hdr->flags& ISCSI_FLAG_CMD_FINAL)&&
> > + (hdr->flags& ISCSI_FLAG_CMD_WRITE)&& conn->sess->sess_ops->InitialR2T) {
> > + printk(KERN_ERR "ISCSI_FLAG_CMD_FINAL is not Set and"
> > + " ISCSI_FLAG_CMD_WRITE Bit and InitialR2T=Yes,"
> > + " protocol error\n");
> > + return iscsit_add_reject(ISCSI_REASON_PROTOCOL_ERROR, 1,
> > + buf, conn);
> > + }
> > +#endif
>
>
> ??
>
>
This was an old overzealous check. Dropped
> > + if (ret == PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES)
>
>
> Is PYX the target this is based on or something else?
>
>
>
This is actually from include/target/target_core_transport.h and Andy
has already added a bugzilla entry about this bit..
I am going to go ahead and change this in LIO v4.1 upstream code, and
will post a patch to remove this prefix seperately for the mainline
target and modules.
>
>
>
>
>
> > +
> > +static inline int iscsit_handle_snack(
>
> snacks om nom nom nom nom
>
>
>
> > +
> > +static int iscsit_handle_immediate_data(
>
>
>
> > +
> > + crypto_hash_init(&conn->conn_rx_hash);
> > +
> > + while (counter> 0) {
> > + sg_init_one(&sg, iov_ptr->iov_base,
> > + iov_ptr->iov_len);
> > + crypto_hash_update(&conn->conn_rx_hash,&sg,
> > + iov_ptr->iov_len);
> > +
> > + TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %zu"
> > + " bytes, CRC 0x%08x\n", iov_ptr->iov_len, data_crc);
> > + counter -= iov_ptr->iov_len;
> > + iov_ptr++;
> > + }
> > +
> > + if (padding) {
> > + sg_init_one(&sg, (u8 *)&pad_bytes, padding);
> > + crypto_hash_update(&conn->conn_rx_hash,&sg,
> > + padding);
> > + TRACE(TRACE_DIGEST, "Computed CRC32C DataDigest %d"
> > + " bytes of padding, CRC 0x%08x\n", padding, data_crc);
> > + }
> > + crypto_hash_final(&conn->conn_rx_hash, (u8 *)&data_crc);
>
>
> It seems you can could make this a function. I keep seeing the hash and
> padding code over and over.
>
>
Done.
Added iscsit_do_crypto_hash_iovec() handler, and converted
iscsit_handle_immediate_data() to use this code.
Also converted all other conn->conn_rx_hash uses to the the appropriate
new iscsit_do_crypto_hash_*() calls.
> > +
> > + if (checksum != data_crc) {
> > + printk(KERN_ERR "ImmediateData CRC32C DataDigest 0x%08x"
> > + " does not match computed 0x%08x\n", checksum,
> > + data_crc);
> > +
> > + if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
> > + printk(KERN_ERR "Unable to recover from"
> > + " Immediate Data digest failure while"
> > + " in ERL=0.\n");
> > + iscsit_add_reject_from_cmd(
> > + ISCSI_REASON_DATA_DIGEST_ERROR,
> > + 1, 0, buf, cmd);
> > + return IMMEDIDATE_DATA_CANNOT_RECOVER;
> > + } else {
> > + iscsit_add_reject_from_cmd(
> > + ISCSI_REASON_DATA_DIGEST_ERROR,
> > + 0, 0, buf, cmd);
> > + return IMMEDIDATE_DATA_ERL1_CRC_FAILURE;
> > + }
> > + } else {
> > + TRACE(TRACE_DIGEST, "Got CRC32C DataDigest 0x%08x for"
> > + " %u bytes of Immediate Data\n", checksum,
> > + length);
> > + }
> > + }
> > +
> > + cmd->write_data_done += length;
> > +
> > + if (cmd->write_data_done == cmd->data_length) {
> > + spin_lock_bh(&cmd->istate_lock);
> > + cmd->cmd_flags |= ICF_GOT_LAST_DATAOUT;
> > + cmd->i_state = ISTATE_RECEIVED_LAST_DATAOUT;
> > + spin_unlock_bh(&cmd->istate_lock);
> > + }
> > +
> > + return IMMEDIDATE_DATA_NORMAL_OPERATION;
> > +}
> > +
> > +int iscsit_send_async_msg(
> > + struct iscsi_conn *conn,
> > + u16 cid,
> > + u8 async_event,
> > + u8 async_vcode)
> > +{
> > + u8 iscsi_hdr[ISCSI_HDR_LEN+CRC_LEN];
> > + u32 tx_send = ISCSI_HDR_LEN, tx_sent = 0;
> > + struct iscsi_async *hdr;
> > + struct kvec iov;
> > + struct scatterlist sg;
> > +
> > + memset(&iov, 0, sizeof(struct kvec));
> > + memset(&iscsi_hdr, 0, ISCSI_HDR_LEN+CRC_LEN);
> > +
> > + hdr = (struct iscsi_async *)&iscsi_hdr;
> > + hdr->opcode = ISCSI_OP_ASYNC_EVENT;
> > + hdr->flags |= ISCSI_FLAG_CMD_FINAL;
> > + hton24(hdr->dlength, 0);
> > + put_unaligned_le64(0,&hdr->lun[0]);
> > + put_unaligned_be64(0xffffffffffffffff,&hdr->rsvd4[0]);
> > + hdr->statsn = cpu_to_be32(conn->stat_sn++);
> > + spin_lock(&conn->sess->cmdsn_lock);
> > + hdr->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
> > + hdr->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
> > + spin_unlock(&conn->sess->cmdsn_lock);
> > + hdr->async_event = async_event;
> > + hdr->async_vcode = async_vcode;
> > +
> > + switch (async_event) {
> > + case ISCSI_ASYNC_MSG_SCSI_EVENT:
> > + printk(KERN_ERR "ISCSI_ASYNC_MSG_SCSI_EVENT: not supported yet.\n");
> > + return -1;
> > + case ISCSI_ASYNC_MSG_REQUEST_LOGOUT:
> > + TRACE(TRACE_STATE, "Moving to"
> > + " TARG_CONN_STATE_LOGOUT_REQUESTED.\n");
> > + conn->conn_state = TARG_CONN_STATE_LOGOUT_REQUESTED;
> > + hdr->param1 = 0;
> > + hdr->param2 = 0;
> > + hdr->param3 = cpu_to_be16(SECONDS_FOR_ASYNC_LOGOUT);
> > + break;
> > + case ISCSI_ASYNC_MSG_DROPPING_CONNECTION:
> > + hdr->param1 = cpu_to_be16(cid);
> > + hdr->param2 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Wait);
> > + hdr->param3 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Retain);
> > + break;
> > + case ISCSI_ASYNC_MSG_DROPPING_ALL_CONNECTIONS:
> > + hdr->param1 = 0;
> > + hdr->param2 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Wait);
> > + hdr->param3 = cpu_to_be16(conn->sess->sess_ops->DefaultTime2Retain);
> > + break;
> > + case ISCSI_ASYNC_MSG_PARAM_NEGOTIATION:
> > + hdr->param1 = 0;
> > + hdr->param2 = 0;
> > + hdr->param3 = cpu_to_be16(SECONDS_FOR_ASYNC_TEXT);
> > + break;
> > + case ISCSI_ASYNC_MSG_VENDOR_SPECIFIC:
> > + printk(KERN_ERR "ISCSI_ASYNC_MSG_VENDOR_SPECIFIC not"
> > + " supported yet.\n");
> > + return -1;
> > + default:
> > + printk(KERN_ERR "Unknown AsycnEvent 0x%02x, protocol"
> > + " error.\n", async_event);
> > + return -1;
> > + }
> > +
> > + iov.iov_base =&iscsi_hdr;
> > + iov.iov_len = ISCSI_HDR_LEN;
> > +
> > + if (conn->conn_ops->HeaderDigest) {
> > + u32 *header_digest = (u32 *)&iscsi_hdr[ISCSI_HDR_LEN];
> > +
> > + crypto_hash_init(&conn->conn_tx_hash);
> > +
> > + sg_init_one(&sg, (u8 *)&iscsi_hdr, ISCSI_HDR_LEN);
> > + crypto_hash_update(&conn->conn_tx_hash,&sg, ISCSI_HDR_LEN);
> > +
> > + crypto_hash_final(&conn->conn_tx_hash, (u8 *)header_digest);
> > +
>
> I think you could make this a function too.
>
Done.
Added iscsit_do_crypto_hash_buf() and converted iscsit_send_async_msg()
to use this code as well.
Also converted all other conn->conn_tx_hash uses to the the appropriate
new iscsit_do_crypto_hash_*() calls.
>
> > + iov.iov_len += CRC_LEN;
> > + tx_send += CRC_LEN;
> > + TRACE(TRACE_DIGEST, "Attaching CRC32 HeaderDigest for Async"
> > + " Msg PDU 0x%08x\n", *header_digest);
> > + }
> > +
> > + TRACE(TRACE_ISCSI, "Built Async Message StatSN: 0x%08x, AsyncEvent:"
> > + " 0x%02x, P1: 0x%04x, P2: 0x%04x, P3: 0x%04x\n",
> > + ntohl(hdr->statsn), hdr->async_event, ntohs(hdr->param1),
> > + ntohs(hdr->param2), ntohs(hdr->param3));
> > +
> > + tx_sent = tx_data(conn,&iov, 1, tx_send);
> > + if (tx_sent != tx_send) {
> > + printk(KERN_ERR "tx_data returned %d expecting %d\n",
> > + tx_sent, tx_send);
> > + return -1;
> > + }
> > +
> > + if (async_event == ISCSI_ASYNC_MSG_REQUEST_LOGOUT) {
> > + wait_for_completion_timeout(&conn->sess->async_msg_comp,
> > + SECONDS_FOR_ASYNC_LOGOUT * HZ);
> > +
> > + if (conn->conn_state == TARG_CONN_STATE_LOGOUT_REQUESTED) {
> > + printk(KERN_ERR "Asynchronous message timer expired"
> > + " without receiving a logout request, dropping"
> > + " iSCSI session.\n");
> > + iscsit_send_async_msg(conn, 0,
> > + ISCSI_ASYNC_MSG_DROPPING_ALL_CONNECTIONS, 0);
> > + iscsit_free_session(conn->sess);
> > + }
> > + }
> > + return 0;
> > +}
> > +
> > +/*
> > + * Called with sess->conn_lock held.
> > + */
> > +/* #warning iscsi_build_conn_drop_async_message() only sends out on connections
> > + with active network interface */
> > +static void iscsit_build_conn_drop_async_message(struct iscsi_conn *conn)
> > +{
> > + struct iscsi_cmd *cmd;
> > + struct iscsi_conn *conn_p;
>
>
> Why do you switch between conn_p and conn? Do you use conn_p when you
> use it for lists? Kind of strange. If it makes it easier I guess ok. If
> it was just due to code rework fix it up.
>
Yes. Fixed a bug here for conn->cmd_lock access which should be
conn_p->cmd_lock.
> > +
> > +#ifdef CONFIG_SMP
> > +
> > +void iscsit_thread_get_cpumask(struct iscsi_conn *conn)
>
>
>
> Did Christoph or someone already say something about using workqueues or
> per cpu threads with nonblocking IO and running from the softirq on the
> recv path? I am assuming so, and skipping this threading code.
>
>
>
Christoph had mentioned getting rid of iscsit_thread_get_cpumask() at
one point, but converting to workqueues for iscsi_target_tx_thread() and
iscsi_target_rx_thread() context is still quite difficult with the
current SIGINT usage for handling iSCSI connection failure cases.
I am still open to considering this as a post-merge improvement, or
something else more exotic like non blocking IO in rx side softirq
context..
>
>
> > +
> > +int iscsi_target_tx_thread(void *arg)
> > +{
>
>
> > + * Bump up the task_struct priority for RX/TX thread set pairs,
> > + * and allow ourselves to be interrupted by SIGINT so that a
> > + * connection recovery / failure event can be triggered externally.
> > + */
> > + set_user_nice(current, -20);
>
>
> What is the reason for using -20 on the target side? On the initiator
> side I thought people did it to avoid some sort of starvation bug/issue
> but I do not think we have that problem on the target do we?
>
> Did you do it for performance reasons? Does it help in current kernels
> and by how much?
>
I was following the original initiator code here as well. I am OK with
dropping this for now..
>
> > +
> > +MODULE_DESCRIPTION("RisingTide Systems iSCSI-Target Driver 4.x.x Release");
>
> I don't think we put names like RisingTide in module descriptions
> either. Put it in MODULE_AUTHOR info.
>
> Version info would go in MODULE_VERSION.
>
>
Fixed
>
>
>
> > diff --git a/drivers/target/iscsi/iscsi_target_core.h b/drivers/target/iscsi/iscsi_target_core.h
> > new file mode 100644
> > index 0000000..20d094e
> > --- /dev/null
> > +++ b/drivers/target/iscsi/iscsi_target_core.h
> > @@ -0,0 +1,881 @@
> > +#ifndef ISCSI_TARGET_CORE_H
> > +#define ISCSI_TARGET_CORE_H
> > +
> > +#include<linux/in.h>
> > +#include<linux/configfs.h>
> > +#include<net/sock.h>
> > +#include<net/tcp.h>
> > +#include<scsi/scsi_cmnd.h>
> > +#include<target/target_core_base.h>
> > +
> > +#define ISCSI_VERSION "v4.1.0-rc1"
>
> Should probably be ISCSIT_VERSION.
>
Done
> > +#define ISCSI_MAX_DATASN_MISSING_COUNT 16
> > +#define ISCSI_TX_THREAD_TCP_TIMEOUT 2
> > +#define ISCSI_RX_THREAD_TCP_TIMEOUT 2
> > +#define SECONDS_FOR_ASYNC_LOGOUT 10
> > +#define SECONDS_FOR_ASYNC_TEXT 10
> > +#define IPV4_ADDRESS_SPACE 4
> > +#define IPV4_BUF_SIZE 18
>
>
> IPV4 ones not used.
>
Dropped
> > +#define ISCSI_HDR_LEN 48
>
>
> Should go in iscsi_proto.h?
>
>
I think so. Moved into iscsi_proto.h
> > +#define CRC_LEN 4
> > +#define MAX_KEY_NAME_LENGTH 63
> > +#define MAX_KEY_VALUE_LENGTH 255
>
>
> Where did these come from? Were they iscsi rfc values?
>
Correct. Moving CRC_LEN to iscsi_proto.h:ISCSI_CRC_LEN..
Also, the value for MAX_KEY_NAME_LENGTH=63 is mentioned in RFC-3720
Section 5.1. Text Format:
standard-label: A string of one or more characters that consist of
letters, digits, dot, minus, plus, commercial at, or underscore.
A standard-label MUST begin with a capital letter and must not
exceed 63 characters.
key-name: A standard-label.
and the value for MAX_KEY_VALUE_LENGTH=255 is also mentioned in the
same section:
If not otherwise specified, the maximum length of a simple-value (not
its encoded representation) is 255 bytes, not including the delimiter
(comma or zero byte).
Btw, Both of these already exist via KEY_MAXLEN and VALUE_MAXLEN in
iscsi_proto.h. Converted iscsi-target to use instead..
>
> > +#define WHITE_SPACE " \t\v\f\n\r"
> > +
> > +/* struct iscsi_node_attrib sanity values */
> > +#define NA_DATAOUT_TIMEOUT 3
> > +#define NA_DATAOUT_TIMEOUT_MAX 60
> > +#define NA_DATAOUT_TIMEOUT_MIX 2
> > +#define NA_DATAOUT_TIMEOUT_RETRIES 5
> > +#define NA_DATAOUT_TIMEOUT_RETRIES_MAX 15
> > +#define NA_DATAOUT_TIMEOUT_RETRIES_MIN 1
> > +#define NA_NOPIN_TIMEOUT 5
> > +#define NA_NOPIN_TIMEOUT_MAX 60
> > +#define NA_NOPIN_TIMEOUT_MIN 3
> > +#define NA_NOPIN_RESPONSE_TIMEOUT 5
> > +#define NA_NOPIN_RESPONSE_TIMEOUT_MAX 60
> > +#define NA_NOPIN_RESPONSE_TIMEOUT_MIN 3
> > +#define NA_RANDOM_DATAIN_PDU_OFFSETS 0
> > +#define NA_RANDOM_DATAIN_SEQ_OFFSETS 0
> > +#define NA_RANDOM_R2T_OFFSETS 0
> > +#define NA_DEFAULT_ERL 0
> > +#define NA_DEFAULT_ERL_MAX 2
> > +#define NA_DEFAULT_ERL_MIN 0
> > +
> > +/* struct iscsi_tpg_attrib sanity values */
> > +#define TA_AUTHENTICATION 1
> > +#define TA_LOGIN_TIMEOUT 15
> > +#define TA_LOGIN_TIMEOUT_MAX 30
> > +#define TA_LOGIN_TIMEOUT_MIN 5
> > +#define TA_NETIF_TIMEOUT 2
> > +#define TA_NETIF_TIMEOUT_MAX 15
> > +#define TA_NETIF_TIMEOUT_MIN 2
> > +#define TA_GENERATE_NODE_ACLS 0
> > +#define TA_DEFAULT_CMDSN_DEPTH 16
> > +#define TA_DEFAULT_CMDSN_DEPTH_MAX 512
> > +#define TA_DEFAULT_CMDSN_DEPTH_MIN 1
> > +#define TA_CACHE_DYNAMIC_ACLS 0
> > +/* Enabled by default in demo mode (generic_node_acls=1) */
> > +#define TA_DEMO_MODE_WRITE_PROTECT 1
> > +/* Disabled by default in production mode w/ explict ACLs */
> > +#define TA_PROD_MODE_WRITE_PROTECT 0
> > +#define TA_CACHE_CORE_NPS 0
> > +
> > +enum tpg_np_network_transport_table {
> > + ISCSI_TCP = 0,
> > + ISCSI_SCTP_TCP = 1,
> > + ISCSI_SCTP_UDP = 2,
> > + ISCSI_IWARP_TCP = 3,
> > + ISCSI_IWARP_SCTP = 4,
> > + ISCSI_INFINIBAND = 5,
> > +};
> > +
> > +/* RFC-3720 7.1.4 Standard Connection State Diagram for a Target */
> > +enum target_conn_state_table {
> > + TARG_CONN_STATE_FREE = 0x1,
> > + TARG_CONN_STATE_XPT_UP = 0x3,
> > + TARG_CONN_STATE_IN_LOGIN = 0x4,
> > + TARG_CONN_STATE_LOGGED_IN = 0x5,
> > + TARG_CONN_STATE_IN_LOGOUT = 0x6,
> > + TARG_CONN_STATE_LOGOUT_REQUESTED = 0x7,
> > + TARG_CONN_STATE_CLEANUP_WAIT = 0x8,
> > +};
> > +
> > +/* RFC-3720 7.3.2 Session State Diagram for a Target */
> > +enum target_sess_state_table {
> > + TARG_SESS_STATE_FREE = 0x1,
> > + TARG_SESS_STATE_ACTIVE = 0x2,
> > + TARG_SESS_STATE_LOGGED_IN = 0x3,
> > + TARG_SESS_STATE_FAILED = 0x4,
> > + TARG_SESS_STATE_IN_CONTINUE = 0x5,
> > +};
>
>
> Should the two above be in iscsi_proto.h?
>
>
Since these are specific to target mode, I think it makes sense to leave
them in iscsi_target_core.h for now..
>
>
>
> > +
> > +struct iscsi_conn {
> > +#define ISCSI_NETDEV_NAME_SIZE 12
> > + char net_dev[ISCSI_NETDEV_NAME_SIZE];
>
>
> Were you going to remove this?
Dropped, along with iscsi_conn->net_if and iscsi_np->np_net_dev.
So with the above changes the diffstat pushed into lio-4.1 looks like:
drivers/target/iscsi/iscsi_target.c | 776 ++++++++++-----------
drivers/target/iscsi/iscsi_target.h | 14 +-
drivers/target/iscsi/iscsi_target_configfs.c | 97 ++--
drivers/target/iscsi/iscsi_target_core.h | 31 +-
drivers/target/iscsi/iscsi_target_datain_values.c | 9 +-
drivers/target/iscsi/iscsi_target_datain_values.h | 4 -
drivers/target/iscsi/iscsi_target_erl0.c | 18 +-
drivers/target/iscsi/iscsi_target_erl0.h | 2 -
drivers/target/iscsi/iscsi_target_erl1.c | 12 +-
drivers/target/iscsi/iscsi_target_erl1.h | 2 -
drivers/target/iscsi/iscsi_target_login.c | 53 +-
drivers/target/iscsi/iscsi_target_login.h | 5 +-
drivers/target/iscsi/iscsi_target_nego.c | 4 +-
drivers/target/iscsi/iscsi_target_nego.h | 2 -
drivers/target/iscsi/iscsi_target_parameters.c | 16 +-
drivers/target/iscsi/iscsi_target_parameters.h | 2 -
drivers/target/iscsi/iscsi_target_seq_pdu_list.c | 14 +-
drivers/target/iscsi/iscsi_target_seq_pdu_list.h | 2 -
drivers/target/iscsi/iscsi_target_stat.c | 2 +-
drivers/target/iscsi/iscsi_target_tpg.c | 34 +-
drivers/target/iscsi/iscsi_target_tpg.h | 6 +-
drivers/target/iscsi/iscsi_target_tq.c | 47 +-
drivers/target/iscsi/iscsi_target_tq.h | 1 -
drivers/target/iscsi/iscsi_target_util.c | 108 +---
drivers/target/iscsi/iscsi_target_util.h | 10 +-
include/scsi/iscsi_proto.h | 6 +
26 files changed, 549 insertions(+), 728 deletions(-)
Thanks for your comments Mike!
--nab
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 40+ messages in thread
* [PATCH-v2 05/14] iscsi-target: Add TCM v4 compatiable ConfigFS control plane
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (4 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 04/14] iscsi-target: Add iSCSI fabric support for target v4 Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 06/14] iscsi-target: Add configfs fabric dependent statistics Nicholas A. Bellinger
` (9 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch adds support for /sys/kernel/config/target/iscsi using
TCM v4.0 compatiable calls following target_core_fabric_configfs.c
This includes a number of iSCSI fabric dependent attributes upon
target_core_fabric_configfs.c provided struct config_item_types from
include/target/target_core_configfs.hstruct target_fabric_configfs_template
It also includes iscsi_target_nodeattrib.[c,h] for handling the
lio_target_nacl_attrib_attrs[] store/show for iSCSI fabric dependent
attributes.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_configfs.c | 1914 ++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_configfs.h | 7 +
drivers/target/iscsi/iscsi_target_nodeattrib.c | 264 ++++
drivers/target/iscsi/iscsi_target_nodeattrib.h | 14 +
4 files changed, 2199 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_configfs.c
create mode 100644 drivers/target/iscsi/iscsi_target_configfs.h
create mode 100644 drivers/target/iscsi/iscsi_target_nodeattrib.c
create mode 100644 drivers/target/iscsi/iscsi_target_nodeattrib.h
diff --git a/drivers/target/iscsi/iscsi_target_configfs.c b/drivers/target/iscsi/iscsi_target_configfs.c
new file mode 100644
index 0000000..7402391
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_configfs.c
@@ -0,0 +1,1914 @@
+/*******************************************************************************
+ * This file contains the configfs implementation for iSCSI Target mode
+ * from the LIO-Target Project.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ****************************************************************************/
+
+#include <linux/configfs.h>
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+#include <target/target_core_fabric_ops.h>
+#include <target/target_core_fabric_configfs.h>
+#include <target/target_core_fabric_lib.h>
+#include <target/target_core_device.h>
+#include <target/target_core_tpg.h>
+#include <target/target_core_configfs.h>
+#include <target/configfs_macros.h>
+
+#include "iscsi_target_core.h"
+#include "iscsi_target_parameters.h"
+#include "iscsi_target_device.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_nodeattrib.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target.h"
+#include "iscsi_target_stat.h"
+#include "iscsi_target_configfs.h"
+
+struct target_fabric_configfs *lio_target_fabric_configfs;
+
+struct lio_target_configfs_attribute {
+ struct configfs_attribute attr;
+ ssize_t (*show)(void *, char *);
+ ssize_t (*store)(void *, const char *, size_t);
+};
+
+struct iscsi_portal_group *lio_get_tpg_from_tpg_item(
+ struct config_item *item,
+ struct iscsi_tiqn **tiqn_out)
+{
+ struct se_portal_group *se_tpg = container_of(to_config_group(item),
+ struct se_portal_group, tpg_group);
+ struct iscsi_portal_group *tpg =
+ (struct iscsi_portal_group *)se_tpg->se_tpg_fabric_ptr;
+ int ret;
+
+ if (!tpg) {
+ printk(KERN_ERR "Unable to locate struct iscsi_portal_group "
+ "pointer\n");
+ return NULL;
+ }
+ ret = iscsit_get_tpg(tpg);
+ if (ret < 0)
+ return NULL;
+
+ *tiqn_out = tpg->tpg_tiqn;
+ return tpg;
+}
+
+/* Start items for lio_target_portal_cit */
+
+static ssize_t lio_target_np_show_sctp(
+ struct se_tpg_np *se_tpg_np,
+ char *page)
+{
+ struct iscsi_tpg_np *tpg_np = container_of(se_tpg_np,
+ struct iscsi_tpg_np, se_tpg_np);
+ struct iscsi_tpg_np *tpg_np_sctp;
+ ssize_t rb;
+
+ tpg_np_sctp = iscsit_tpg_locate_child_np(tpg_np, ISCSI_SCTP_TCP);
+ if (tpg_np_sctp)
+ rb = sprintf(page, "1\n");
+ else
+ rb = sprintf(page, "0\n");
+
+ return rb;
+}
+
+static ssize_t lio_target_np_store_sctp(
+ struct se_tpg_np *se_tpg_np,
+ const char *page,
+ size_t count)
+{
+ struct iscsi_np *np;
+ struct iscsi_portal_group *tpg;
+ struct iscsi_tpg_np *tpg_np = container_of(se_tpg_np,
+ struct iscsi_tpg_np, se_tpg_np);
+ struct iscsi_tpg_np *tpg_np_sctp = NULL;
+ struct iscsi_np_addr np_addr;
+ char *endptr;
+ u32 op;
+ int ret;
+
+ op = simple_strtoul(page, &endptr, 0);
+ if ((op != 1) && (op != 0)) {
+ printk(KERN_ERR "Illegal value for tpg_enable: %u\n", op);
+ return -EINVAL;
+ }
+ np = tpg_np->tpg_np;
+ if (!np) {
+ printk(KERN_ERR "Unable to locate struct iscsi_np from"
+ " struct iscsi_tpg_np\n");
+ return -EINVAL;
+ }
+
+ tpg = tpg_np->tpg;
+ if (iscsit_get_tpg(tpg) < 0)
+ return -EINVAL;
+
+ if (op) {
+ memset(&np_addr, 0, sizeof(struct iscsi_np_addr));
+ if (np->np_sockaddr.ss_family == AF_INET6)
+ snprintf(np_addr.np_ipv6, IPV6_ADDRESS_SPACE,
+ "%s", np->np_ip);
+ else
+ np_addr.np_ipv4 = np->np_ipv4;
+ np_addr.np_flags = np->np_flags;
+ np_addr.np_port = np->np_port;
+
+ tpg_np_sctp = iscsit_tpg_add_network_portal(tpg, &np_addr,
+ tpg_np, ISCSI_SCTP_TCP,
+ np->np_sockaddr.ss_family);
+ if (!tpg_np_sctp || IS_ERR(tpg_np_sctp))
+ goto out;
+ } else {
+ tpg_np_sctp = iscsit_tpg_locate_child_np(tpg_np, ISCSI_SCTP_TCP);
+ if (!tpg_np_sctp)
+ goto out;
+
+ ret = iscsit_tpg_del_network_portal(tpg, tpg_np_sctp);
+ if (ret < 0)
+ goto out;
+ }
+
+ iscsit_put_tpg(tpg);
+ return count;
+out:
+ iscsit_put_tpg(tpg);
+ return -EINVAL;
+}
+
+TF_NP_BASE_ATTR(lio_target, sctp, S_IRUGO | S_IWUSR);
+
+static struct configfs_attribute *lio_target_portal_attrs[] = {
+ &lio_target_np_sctp.attr,
+ NULL,
+};
+
+/* Stop items for lio_target_portal_cit */
+
+/* Start items for lio_target_np_cit */
+
+#define MAX_PORTAL_LEN 256
+
+struct se_tpg_np *lio_target_call_addnptotpg(
+ struct se_portal_group *se_tpg,
+ struct config_group *group,
+ const char *name)
+{
+ struct iscsi_portal_group *tpg;
+ struct iscsi_tpg_np *tpg_np;
+ char *str, *str2, *end_ptr, *ip_str, *port_str;
+ struct iscsi_np_addr np_addr;
+ u32 ipv4 = 0;
+ int ret, af_inet;
+ char buf[MAX_PORTAL_LEN + 1];
+
+ if (strlen(name) > MAX_PORTAL_LEN) {
+ printk(KERN_ERR "strlen(name): %d exceeds MAX_PORTAL_LEN: %d\n",
+ (int)strlen(name), MAX_PORTAL_LEN);
+ return ERR_PTR(-EOVERFLOW);
+ }
+ memset(buf, 0, MAX_PORTAL_LEN + 1);
+ snprintf(buf, MAX_PORTAL_LEN, "%s", name);
+
+ memset(&np_addr, 0, sizeof(struct iscsi_np_addr));
+
+ str = strstr(buf, "[");
+ if (str) {
+ str2 = strstr(str, "]");
+ if (!str2) {
+ printk(KERN_ERR "Unable to locate trailing \"]\""
+ " in IPv6 iSCSI network portal address\n");
+ return ERR_PTR(-EINVAL);
+ }
+ str++; /* Skip over leading "[" */
+ *str2 = '\0'; /* Terminate the IPv6 address */
+ str2++; /* Skip over the "]" */
+ port_str = strstr(str2, ":");
+ if (!port_str) {
+ printk(KERN_ERR "Unable to locate \":port\""
+ " in IPv6 iSCSI network portal address\n");
+ return ERR_PTR(-EINVAL);
+ }
+ *port_str = '\0'; /* Terminate string for IP */
+ port_str++; /* Skip over ":" */
+ np_addr.np_port = simple_strtoul(port_str, &end_ptr, 0);
+
+ snprintf(np_addr.np_ipv6, IPV6_ADDRESS_SPACE, "%s", str);
+ af_inet = AF_INET6;
+ } else {
+ ip_str = &buf[0];
+ port_str = strstr(ip_str, ":");
+ if (!port_str) {
+ printk(KERN_ERR "Unable to locate \":port\""
+ " in IPv4 iSCSI network portal address\n");
+ return ERR_PTR(-EINVAL);
+ }
+ *port_str = '\0'; /* Terminate string for IP */
+ port_str++; /* Skip over ":" */
+ np_addr.np_port = simple_strtoul(port_str, &end_ptr, 0);
+
+ ipv4 = in_aton(ip_str);
+ np_addr.np_ipv4 = htonl(ipv4);
+ af_inet = AF_INET;
+ }
+ tpg = container_of(se_tpg, struct iscsi_portal_group, tpg_se_tpg);
+ ret = iscsit_get_tpg(tpg);
+ if (ret < 0)
+ return ERR_PTR(-EINVAL);
+
+ printk(KERN_INFO "LIO_Target_ConfigFS: REGISTER -> %s TPGT: %hu"
+ " PORTAL: %s\n",
+ config_item_name(&se_tpg->se_tpg_wwn->wwn_group.cg_item),
+ tpg->tpgt, name);
+ /*
+ * Assume ISCSI_TCP by default. Other network portals for other
+ * iSCSI fabrics:
+ *
+ * Traditional iSCSI over SCTP (initial support)
+ * iSER/TCP (TODO, hardware available)
+ * iSER/SCTP (TODO, software emulation with osc-iwarp)
+ * iSER/IB (TODO, hardware available)
+ *
+ * can be enabled with atributes under
+ * sys/kernel/config/iscsi/$IQN/$TPG/np/$IP:$PORT/
+ *
+ */
+ tpg_np = iscsit_tpg_add_network_portal(tpg, &np_addr, NULL,
+ ISCSI_TCP, af_inet);
+ if (IS_ERR(tpg_np)) {
+ iscsit_put_tpg(tpg);
+ return ERR_PTR(PTR_ERR(tpg_np));
+ }
+ printk(KERN_INFO "LIO_Target_ConfigFS: addnptotpg done!\n");
+
+ iscsit_put_tpg(tpg);
+ return &tpg_np->se_tpg_np;
+}
+
+static void lio_target_call_delnpfromtpg(
+ struct se_tpg_np *se_tpg_np)
+{
+ struct iscsi_portal_group *tpg;
+ struct iscsi_tpg_np *tpg_np;
+ struct se_portal_group *se_tpg;
+ int ret;
+
+ tpg_np = container_of(se_tpg_np, struct iscsi_tpg_np, se_tpg_np);
+ tpg = tpg_np->tpg;
+ ret = iscsit_get_tpg(tpg);
+ if (ret < 0)
+ return;
+
+ se_tpg = &tpg->tpg_se_tpg;
+ printk(KERN_INFO "LIO_Target_ConfigFS: DEREGISTER -> %s TPGT: %hu"
+ " PORTAL: %s:%hu\n", config_item_name(&se_tpg->se_tpg_wwn->wwn_group.cg_item),
+ tpg->tpgt, tpg_np->tpg_np->np_ip, tpg_np->tpg_np->np_port);
+
+ ret = iscsit_tpg_del_network_portal(tpg, tpg_np);
+ if (ret < 0)
+ goto out;
+
+ printk(KERN_INFO "LIO_Target_ConfigFS: delnpfromtpg done!\n");
+out:
+ iscsit_put_tpg(tpg);
+}
+
+/* End items for lio_target_np_cit */
+
+/* Start items for lio_target_nacl_attrib_cit */
+
+#define DEF_NACL_ATTRIB(name) \
+static ssize_t iscsi_nacl_attrib_show_##name( \
+ struct se_node_acl *se_nacl, \
+ char *page) \
+{ \
+ struct iscsi_node_acl *nacl = container_of(se_nacl, struct iscsi_node_acl, \
+ se_node_acl); \
+ \
+ return sprintf(page, "%u\n", ISCSI_NODE_ATTRIB(nacl)->name); \
+} \
+ \
+static ssize_t iscsi_nacl_attrib_store_##name( \
+ struct se_node_acl *se_nacl, \
+ const char *page, \
+ size_t count) \
+{ \
+ struct iscsi_node_acl *nacl = container_of(se_nacl, struct iscsi_node_acl, \
+ se_node_acl); \
+ char *endptr; \
+ u32 val; \
+ int ret; \
+ \
+ val = simple_strtoul(page, &endptr, 0); \
+ ret = iscsit_na_##name(nacl, val); \
+ if (ret < 0) \
+ return ret; \
+ \
+ return count; \
+}
+
+#define NACL_ATTR(_name, _mode) TF_NACL_ATTRIB_ATTR(iscsi, _name, _mode);
+/*
+ * Define iscsi_node_attrib_s_dataout_timeout
+ */
+DEF_NACL_ATTRIB(dataout_timeout);
+NACL_ATTR(dataout_timeout, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_node_attrib_s_dataout_timeout_retries
+ */
+DEF_NACL_ATTRIB(dataout_timeout_retries);
+NACL_ATTR(dataout_timeout_retries, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_node_attrib_s_default_erl
+ */
+DEF_NACL_ATTRIB(default_erl);
+NACL_ATTR(default_erl, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_node_attrib_s_nopin_timeout
+ */
+DEF_NACL_ATTRIB(nopin_timeout);
+NACL_ATTR(nopin_timeout, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_node_attrib_s_nopin_response_timeout
+ */
+DEF_NACL_ATTRIB(nopin_response_timeout);
+NACL_ATTR(nopin_response_timeout, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_node_attrib_s_random_datain_pdu_offsets
+ */
+DEF_NACL_ATTRIB(random_datain_pdu_offsets);
+NACL_ATTR(random_datain_pdu_offsets, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_node_attrib_s_random_datain_seq_offsets
+ */
+DEF_NACL_ATTRIB(random_datain_seq_offsets);
+NACL_ATTR(random_datain_seq_offsets, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_node_attrib_s_random_r2t_offsets
+ */
+DEF_NACL_ATTRIB(random_r2t_offsets);
+NACL_ATTR(random_r2t_offsets, S_IRUGO | S_IWUSR);
+
+static struct configfs_attribute *lio_target_nacl_attrib_attrs[] = {
+ &iscsi_nacl_attrib_dataout_timeout.attr,
+ &iscsi_nacl_attrib_dataout_timeout_retries.attr,
+ &iscsi_nacl_attrib_default_erl.attr,
+ &iscsi_nacl_attrib_nopin_timeout.attr,
+ &iscsi_nacl_attrib_nopin_response_timeout.attr,
+ &iscsi_nacl_attrib_random_datain_pdu_offsets.attr,
+ &iscsi_nacl_attrib_random_datain_seq_offsets.attr,
+ &iscsi_nacl_attrib_random_r2t_offsets.attr,
+ NULL,
+};
+
+/* End items for lio_target_nacl_attrib_cit */
+
+/* Start items for lio_target_nacl_auth_cit */
+
+#define __DEF_NACL_AUTH_STR(prefix, name, flags) \
+static ssize_t __iscsi_##prefix##_show_##name( \
+ struct iscsi_node_acl *nacl, \
+ char *page) \
+{ \
+ struct iscsi_node_auth *auth = &nacl->node_auth; \
+ \
+ if (!capable(CAP_SYS_ADMIN)) \
+ return -EPERM; \
+ return snprintf(page, PAGE_SIZE, "%s\n", auth->name); \
+} \
+ \
+static ssize_t __iscsi_##prefix##_store_##name( \
+ struct iscsi_node_acl *nacl, \
+ const char *page, \
+ size_t count) \
+{ \
+ struct iscsi_node_auth *auth = &nacl->node_auth; \
+ \
+ if (!capable(CAP_SYS_ADMIN)) \
+ return -EPERM; \
+ \
+ snprintf(auth->name, PAGE_SIZE, "%s", page); \
+ if (!strncmp("NULL", auth->name, 4)) \
+ auth->naf_flags &= ~flags; \
+ else \
+ auth->naf_flags |= flags; \
+ \
+ if ((auth->naf_flags & NAF_USERID_IN_SET) && \
+ (auth->naf_flags & NAF_PASSWORD_IN_SET)) \
+ auth->authenticate_target = 1; \
+ else \
+ auth->authenticate_target = 0; \
+ \
+ return count; \
+}
+
+#define __DEF_NACL_AUTH_INT(prefix, name) \
+static ssize_t __iscsi_##prefix##_show_##name( \
+ struct iscsi_node_acl *nacl, \
+ char *page) \
+{ \
+ struct iscsi_node_auth *auth = &nacl->node_auth; \
+ \
+ if (!capable(CAP_SYS_ADMIN)) \
+ return -EPERM; \
+ \
+ return snprintf(page, PAGE_SIZE, "%d\n", auth->name); \
+}
+
+#define DEF_NACL_AUTH_STR(name, flags) \
+ __DEF_NACL_AUTH_STR(nacl_auth, name, flags) \
+static ssize_t iscsi_nacl_auth_show_##name( \
+ struct se_node_acl *nacl, \
+ char *page) \
+{ \
+ return __iscsi_nacl_auth_show_##name(container_of(nacl, \
+ struct iscsi_node_acl, se_node_acl), page); \
+} \
+static ssize_t iscsi_nacl_auth_store_##name( \
+ struct se_node_acl *nacl, \
+ const char *page, \
+ size_t count) \
+{ \
+ return __iscsi_nacl_auth_store_##name(container_of(nacl, \
+ struct iscsi_node_acl, se_node_acl), page, count); \
+}
+
+#define DEF_NACL_AUTH_INT(name) \
+ __DEF_NACL_AUTH_INT(nacl_auth, name) \
+static ssize_t iscsi_nacl_auth_show_##name( \
+ struct se_node_acl *nacl, \
+ char *page) \
+{ \
+ return __iscsi_nacl_auth_show_##name(container_of(nacl, \
+ struct iscsi_node_acl, se_node_acl), page); \
+}
+
+#define AUTH_ATTR(_name, _mode) TF_NACL_AUTH_ATTR(iscsi, _name, _mode);
+#define AUTH_ATTR_RO(_name) TF_NACL_AUTH_ATTR_RO(iscsi, _name);
+
+/*
+ * One-way authentication userid
+ */
+DEF_NACL_AUTH_STR(userid, NAF_USERID_SET);
+AUTH_ATTR(userid, S_IRUGO | S_IWUSR);
+/*
+ * One-way authentication password
+ */
+DEF_NACL_AUTH_STR(password, NAF_PASSWORD_SET);
+AUTH_ATTR(password, S_IRUGO | S_IWUSR);
+/*
+ * Enforce mutual authentication
+ */
+DEF_NACL_AUTH_INT(authenticate_target);
+AUTH_ATTR_RO(authenticate_target);
+/*
+ * Mutual authentication userid
+ */
+DEF_NACL_AUTH_STR(userid_mutual, NAF_USERID_IN_SET);
+AUTH_ATTR(userid_mutual, S_IRUGO | S_IWUSR);
+/*
+ * Mutual authentication password
+ */
+DEF_NACL_AUTH_STR(password_mutual, NAF_PASSWORD_IN_SET);
+AUTH_ATTR(password_mutual, S_IRUGO | S_IWUSR);
+
+static struct configfs_attribute *lio_target_nacl_auth_attrs[] = {
+ &iscsi_nacl_auth_userid.attr,
+ &iscsi_nacl_auth_password.attr,
+ &iscsi_nacl_auth_authenticate_target.attr,
+ &iscsi_nacl_auth_userid_mutual.attr,
+ &iscsi_nacl_auth_password_mutual.attr,
+ NULL,
+};
+
+/* End items for lio_target_nacl_auth_cit */
+
+/* Start items for lio_target_nacl_param_cit */
+
+#define DEF_NACL_PARAM(name) \
+static ssize_t iscsi_nacl_param_show_##name( \
+ struct se_node_acl *se_nacl, \
+ char *page) \
+{ \
+ struct iscsi_session *sess; \
+ struct se_session *se_sess; \
+ ssize_t rb; \
+ \
+ spin_lock_bh(&se_nacl->nacl_sess_lock); \
+ se_sess = se_nacl->nacl_sess; \
+ if (!se_sess) { \
+ rb = snprintf(page, PAGE_SIZE, \
+ "No Active iSCSI Session\n"); \
+ } else { \
+ sess = se_sess->fabric_sess_ptr; \
+ rb = snprintf(page, PAGE_SIZE, "%u\n", \
+ (u32)sess->sess_ops->name); \
+ } \
+ spin_unlock_bh(&se_nacl->nacl_sess_lock); \
+ \
+ return rb; \
+}
+
+#define NACL_PARAM_ATTR(_name) TF_NACL_PARAM_ATTR_RO(iscsi, _name);
+
+DEF_NACL_PARAM(MaxConnections);
+NACL_PARAM_ATTR(MaxConnections);
+
+DEF_NACL_PARAM(InitialR2T);
+NACL_PARAM_ATTR(InitialR2T);
+
+DEF_NACL_PARAM(ImmediateData);
+NACL_PARAM_ATTR(ImmediateData);
+
+DEF_NACL_PARAM(MaxBurstLength);
+NACL_PARAM_ATTR(MaxBurstLength);
+
+DEF_NACL_PARAM(FirstBurstLength);
+NACL_PARAM_ATTR(FirstBurstLength);
+
+DEF_NACL_PARAM(DefaultTime2Wait);
+NACL_PARAM_ATTR(DefaultTime2Wait);
+
+DEF_NACL_PARAM(DefaultTime2Retain);
+NACL_PARAM_ATTR(DefaultTime2Retain);
+
+DEF_NACL_PARAM(MaxOutstandingR2T);
+NACL_PARAM_ATTR(MaxOutstandingR2T);
+
+DEF_NACL_PARAM(DataPDUInOrder);
+NACL_PARAM_ATTR(DataPDUInOrder);
+
+DEF_NACL_PARAM(DataSequenceInOrder);
+NACL_PARAM_ATTR(DataSequenceInOrder);
+
+DEF_NACL_PARAM(ErrorRecoveryLevel);
+NACL_PARAM_ATTR(ErrorRecoveryLevel);
+
+static struct configfs_attribute *lio_target_nacl_param_attrs[] = {
+ &iscsi_nacl_param_MaxConnections.attr,
+ &iscsi_nacl_param_InitialR2T.attr,
+ &iscsi_nacl_param_ImmediateData.attr,
+ &iscsi_nacl_param_MaxBurstLength.attr,
+ &iscsi_nacl_param_FirstBurstLength.attr,
+ &iscsi_nacl_param_DefaultTime2Wait.attr,
+ &iscsi_nacl_param_DefaultTime2Retain.attr,
+ &iscsi_nacl_param_MaxOutstandingR2T.attr,
+ &iscsi_nacl_param_DataPDUInOrder.attr,
+ &iscsi_nacl_param_DataSequenceInOrder.attr,
+ &iscsi_nacl_param_ErrorRecoveryLevel.attr,
+ NULL,
+};
+
+/* End items for lio_target_nacl_param_cit */
+
+/* Start items for lio_target_acl_cit */
+
+static ssize_t lio_target_nacl_show_info(
+ struct se_node_acl *se_nacl,
+ char *page)
+{
+ struct iscsi_session *sess;
+ struct iscsi_conn *conn;
+ struct se_session *se_sess;
+ ssize_t rb = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (!se_sess) {
+ rb += sprintf(page+rb, "No active iSCSI Session for Initiator"
+ " Endpoint: %s\n", se_nacl->initiatorname);
+ } else {
+ sess = se_sess->fabric_sess_ptr;
+
+ if (sess->sess_ops->InitiatorName)
+ rb += sprintf(page+rb, "InitiatorName: %s\n",
+ sess->sess_ops->InitiatorName);
+ if (sess->sess_ops->InitiatorAlias)
+ rb += sprintf(page+rb, "InitiatorAlias: %s\n",
+ sess->sess_ops->InitiatorAlias);
+
+ rb += sprintf(page+rb, "LIO Session ID: %u "
+ "ISID: 0x%02x %02x %02x %02x %02x %02x "
+ "TSIH: %hu ", sess->sid,
+ sess->isid[0], sess->isid[1], sess->isid[2],
+ sess->isid[3], sess->isid[4], sess->isid[5],
+ sess->tsih);
+ rb += sprintf(page+rb, "SessionType: %s\n",
+ (sess->sess_ops->SessionType) ?
+ "Discovery" : "Normal");
+ rb += sprintf(page+rb, "Session State: ");
+ switch (sess->session_state) {
+ case TARG_SESS_STATE_FREE:
+ rb += sprintf(page+rb, "TARG_SESS_FREE\n");
+ break;
+ case TARG_SESS_STATE_ACTIVE:
+ rb += sprintf(page+rb, "TARG_SESS_STATE_ACTIVE\n");
+ break;
+ case TARG_SESS_STATE_LOGGED_IN:
+ rb += sprintf(page+rb, "TARG_SESS_STATE_LOGGED_IN\n");
+ break;
+ case TARG_SESS_STATE_FAILED:
+ rb += sprintf(page+rb, "TARG_SESS_STATE_FAILED\n");
+ break;
+ case TARG_SESS_STATE_IN_CONTINUE:
+ rb += sprintf(page+rb, "TARG_SESS_STATE_IN_CONTINUE\n");
+ break;
+ default:
+ rb += sprintf(page+rb, "ERROR: Unknown Session"
+ " State!\n");
+ break;
+ }
+
+ rb += sprintf(page+rb, "---------------------[iSCSI Session"
+ " Values]-----------------------\n");
+ rb += sprintf(page+rb, " CmdSN/WR : CmdSN/WC : ExpCmdSN"
+ " : MaxCmdSN : ITT : TTT\n");
+ rb += sprintf(page+rb, " 0x%08x 0x%08x 0x%08x 0x%08x"
+ " 0x%08x 0x%08x\n",
+ sess->cmdsn_window,
+ (sess->max_cmd_sn - sess->exp_cmd_sn) + 1,
+ sess->exp_cmd_sn, sess->max_cmd_sn,
+ sess->init_task_tag, sess->targ_xfer_tag);
+ rb += sprintf(page+rb, "----------------------[iSCSI"
+ " Connections]-------------------------\n");
+
+ spin_lock(&sess->conn_lock);
+ list_for_each_entry(conn, &sess->sess_conn_list, conn_list) {
+ rb += sprintf(page+rb, "CID: %hu Connection"
+ " State: ", conn->cid);
+ switch (conn->conn_state) {
+ case TARG_CONN_STATE_FREE:
+ rb += sprintf(page+rb,
+ "TARG_CONN_STATE_FREE\n");
+ break;
+ case TARG_CONN_STATE_XPT_UP:
+ rb += sprintf(page+rb,
+ "TARG_CONN_STATE_XPT_UP\n");
+ break;
+ case TARG_CONN_STATE_IN_LOGIN:
+ rb += sprintf(page+rb,
+ "TARG_CONN_STATE_IN_LOGIN\n");
+ break;
+ case TARG_CONN_STATE_LOGGED_IN:
+ rb += sprintf(page+rb,
+ "TARG_CONN_STATE_LOGGED_IN\n");
+ break;
+ case TARG_CONN_STATE_IN_LOGOUT:
+ rb += sprintf(page+rb,
+ "TARG_CONN_STATE_IN_LOGOUT\n");
+ break;
+ case TARG_CONN_STATE_LOGOUT_REQUESTED:
+ rb += sprintf(page+rb,
+ "TARG_CONN_STATE_LOGOUT_REQUESTED\n");
+ break;
+ case TARG_CONN_STATE_CLEANUP_WAIT:
+ rb += sprintf(page+rb,
+ "TARG_CONN_STATE_CLEANUP_WAIT\n");
+ break;
+ default:
+ rb += sprintf(page+rb,
+ "ERROR: Unknown Connection State!\n");
+ break;
+ }
+
+ rb += sprintf(page+rb, " Address %s %s", conn->login_ip,
+ (conn->network_transport == ISCSI_TCP) ?
+ "TCP" : "SCTP");
+ rb += sprintf(page+rb, " StatSN: 0x%08x\n",
+ conn->stat_sn);
+ }
+ spin_unlock(&sess->conn_lock);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return rb;
+}
+
+TF_NACL_BASE_ATTR_RO(lio_target, info);
+
+static ssize_t lio_target_nacl_show_cmdsn_depth(
+ struct se_node_acl *se_nacl,
+ char *page)
+{
+ return sprintf(page, "%u\n", se_nacl->queue_depth);
+}
+
+static ssize_t lio_target_nacl_store_cmdsn_depth(
+ struct se_node_acl *se_nacl,
+ const char *page,
+ size_t count)
+{
+ struct se_portal_group *se_tpg = se_nacl->se_tpg;
+ struct iscsi_portal_group *tpg = container_of(se_tpg,
+ struct iscsi_portal_group, tpg_se_tpg);
+ struct config_item *acl_ci, *tpg_ci, *wwn_ci;
+ char *endptr;
+ u32 cmdsn_depth = 0;
+ int ret;
+
+ cmdsn_depth = simple_strtoul(page, &endptr, 0);
+ if (cmdsn_depth > TA_DEFAULT_CMDSN_DEPTH_MAX) {
+ printk(KERN_ERR "Passed cmdsn_depth: %u exceeds"
+ " TA_DEFAULT_CMDSN_DEPTH_MAX: %u\n", cmdsn_depth,
+ TA_DEFAULT_CMDSN_DEPTH_MAX);
+ return -EINVAL;
+ }
+ acl_ci = &se_nacl->acl_group.cg_item;
+ if (!acl_ci) {
+ printk(KERN_ERR "Unable to locatel acl_ci\n");
+ return -EINVAL;
+ }
+ tpg_ci = &acl_ci->ci_parent->ci_group->cg_item;
+ if (!tpg_ci) {
+ printk(KERN_ERR "Unable to locate tpg_ci\n");
+ return -EINVAL;
+ }
+ wwn_ci = &tpg_ci->ci_group->cg_item;
+ if (!wwn_ci) {
+ printk(KERN_ERR "Unable to locate config_item wwn_ci\n");
+ return -EINVAL;
+ }
+
+ if (iscsit_get_tpg(tpg) < 0)
+ return -EINVAL;
+ /*
+ * iscsit_tpg_set_initiator_node_queue_depth() assumes force=1
+ */
+ ret = iscsit_tpg_set_initiator_node_queue_depth(tpg,
+ config_item_name(acl_ci), cmdsn_depth, 1);
+
+ printk(KERN_INFO "LIO_Target_ConfigFS: %s/%s Set CmdSN Window: %u for"
+ "InitiatorName: %s\n", config_item_name(wwn_ci),
+ config_item_name(tpg_ci), cmdsn_depth,
+ config_item_name(acl_ci));
+
+ iscsit_put_tpg(tpg);
+ return (!ret) ? count : (ssize_t)ret;
+}
+
+TF_NACL_BASE_ATTR(lio_target, cmdsn_depth, S_IRUGO | S_IWUSR);
+
+static struct configfs_attribute *lio_target_initiator_attrs[] = {
+ &lio_target_nacl_info.attr,
+ &lio_target_nacl_cmdsn_depth.attr,
+ NULL,
+};
+
+static struct se_node_acl *lio_tpg_alloc_fabric_acl(
+ struct se_portal_group *se_tpg)
+{
+ struct iscsi_node_acl *acl;
+
+ acl = kzalloc(sizeof(struct iscsi_node_acl), GFP_KERNEL);
+ if (!acl) {
+ printk(KERN_ERR "Unable to allocate memory for struct iscsi_node_acl\n");
+ return NULL;
+ }
+
+ return &acl->se_node_acl;
+}
+
+static struct se_node_acl *lio_target_make_nodeacl(
+ struct se_portal_group *se_tpg,
+ struct config_group *group,
+ const char *name)
+{
+ struct config_group *stats_cg;
+ struct iscsi_node_acl *acl;
+ struct se_node_acl *se_nacl_new, *se_nacl;
+ struct iscsi_portal_group *tpg = container_of(se_tpg,
+ struct iscsi_portal_group, tpg_se_tpg);
+ u32 cmdsn_depth;
+
+ se_nacl_new = lio_tpg_alloc_fabric_acl(se_tpg);
+ if (!se_nacl_new)
+ return ERR_PTR(-ENOMEM);
+
+ acl = container_of(se_nacl_new, struct iscsi_node_acl,
+ se_node_acl);
+
+ cmdsn_depth = ISCSI_TPG_ATTRIB(tpg)->default_cmdsn_depth;
+ /*
+ * se_nacl_new may be released by core_tpg_add_initiator_node_acl()
+ * when converting a NdoeACL from demo mode -> explict
+ */
+ se_nacl = core_tpg_add_initiator_node_acl(se_tpg, se_nacl_new,
+ name, cmdsn_depth);
+ if (IS_ERR(se_nacl))
+ return se_nacl;
+
+ stats_cg = &acl->se_node_acl.acl_fabric_stat_group;
+
+ stats_cg->default_groups = kzalloc(sizeof(struct config_group) * 2,
+ GFP_KERNEL);
+ if (!stats_cg->default_groups) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " stats_cg->default_groups\n");
+ core_tpg_del_initiator_node_acl(se_tpg, se_nacl, 1);
+ kfree(acl);
+ return ERR_PTR(-ENOMEM);
+ }
+
+ stats_cg->default_groups[0] = &NODE_STAT_GRPS(acl)->iscsi_sess_stats_group;
+ stats_cg->default_groups[1] = NULL;
+ config_group_init_type_name(&NODE_STAT_GRPS(acl)->iscsi_sess_stats_group,
+ "iscsi_sess_stats", &iscsi_stat_sess_cit);
+
+ return se_nacl;
+}
+
+static void lio_target_drop_nodeacl(
+ struct se_node_acl *se_nacl)
+{
+ struct se_portal_group *se_tpg = se_nacl->se_tpg;
+ struct iscsi_node_acl *acl = container_of(se_nacl,
+ struct iscsi_node_acl, se_node_acl);
+ struct config_item *df_item;
+ struct config_group *stats_cg;
+ int i;
+
+ stats_cg = &acl->se_node_acl.acl_fabric_stat_group;
+ for (i = 0; stats_cg->default_groups[i]; i++) {
+ df_item = &stats_cg->default_groups[i]->cg_item;
+ stats_cg->default_groups[i] = NULL;
+ config_item_put(df_item);
+ }
+ kfree(stats_cg->default_groups);
+
+ core_tpg_del_initiator_node_acl(se_tpg, se_nacl, 1);
+ kfree(acl);
+}
+
+/* End items for lio_target_acl_cit */
+
+/* Start items for lio_target_tpg_attrib_cit */
+
+#define DEF_TPG_ATTRIB(name) \
+ \
+static ssize_t iscsi_tpg_attrib_show_##name( \
+ struct se_portal_group *se_tpg, \
+ char *page) \
+{ \
+ struct iscsi_portal_group *tpg = container_of(se_tpg, \
+ struct iscsi_portal_group, tpg_se_tpg); \
+ ssize_t rb; \
+ \
+ if (iscsit_get_tpg(tpg) < 0) \
+ return -EINVAL; \
+ \
+ rb = sprintf(page, "%u\n", ISCSI_TPG_ATTRIB(tpg)->name); \
+ iscsit_put_tpg(tpg); \
+ return rb; \
+} \
+ \
+static ssize_t iscsi_tpg_attrib_store_##name( \
+ struct se_portal_group *se_tpg, \
+ const char *page, \
+ size_t count) \
+{ \
+ struct iscsi_portal_group *tpg = container_of(se_tpg, \
+ struct iscsi_portal_group, tpg_se_tpg); \
+ char *endptr; \
+ u32 val; \
+ int ret; \
+ \
+ if (iscsit_get_tpg(tpg) < 0) \
+ return -EINVAL; \
+ \
+ val = simple_strtoul(page, &endptr, 0); \
+ ret = iscsit_ta_##name(tpg, val); \
+ if (ret < 0) \
+ goto out; \
+ \
+ iscsit_put_tpg(tpg); \
+ return count; \
+out: \
+ iscsit_put_tpg(tpg); \
+ return ret; \
+}
+
+#define TPG_ATTR(_name, _mode) TF_TPG_ATTRIB_ATTR(iscsi, _name, _mode);
+
+/*
+ * Define iscsi_tpg_attrib_s_authentication
+ */
+DEF_TPG_ATTRIB(authentication);
+TPG_ATTR(authentication, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_tpg_attrib_s_login_timeout
+ */
+DEF_TPG_ATTRIB(login_timeout);
+TPG_ATTR(login_timeout, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_tpg_attrib_s_netif_timeout
+ */
+DEF_TPG_ATTRIB(netif_timeout);
+TPG_ATTR(netif_timeout, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_tpg_attrib_s_generate_node_acls
+ */
+DEF_TPG_ATTRIB(generate_node_acls);
+TPG_ATTR(generate_node_acls, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_tpg_attrib_s_default_cmdsn_depth
+ */
+DEF_TPG_ATTRIB(default_cmdsn_depth);
+TPG_ATTR(default_cmdsn_depth, S_IRUGO | S_IWUSR);
+/*
+ Define iscsi_tpg_attrib_s_cache_dynamic_acls
+ */
+DEF_TPG_ATTRIB(cache_dynamic_acls);
+TPG_ATTR(cache_dynamic_acls, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_tpg_attrib_s_demo_mode_write_protect
+ */
+DEF_TPG_ATTRIB(demo_mode_write_protect);
+TPG_ATTR(demo_mode_write_protect, S_IRUGO | S_IWUSR);
+/*
+ * Define iscsi_tpg_attrib_s_prod_mode_write_protect
+ */
+DEF_TPG_ATTRIB(prod_mode_write_protect);
+TPG_ATTR(prod_mode_write_protect, S_IRUGO | S_IWUSR);
+
+static struct configfs_attribute *lio_target_tpg_attrib_attrs[] = {
+ &iscsi_tpg_attrib_authentication.attr,
+ &iscsi_tpg_attrib_login_timeout.attr,
+ &iscsi_tpg_attrib_netif_timeout.attr,
+ &iscsi_tpg_attrib_generate_node_acls.attr,
+ &iscsi_tpg_attrib_default_cmdsn_depth.attr,
+ &iscsi_tpg_attrib_cache_dynamic_acls.attr,
+ &iscsi_tpg_attrib_demo_mode_write_protect.attr,
+ &iscsi_tpg_attrib_prod_mode_write_protect.attr,
+ NULL,
+};
+
+/* End items for lio_target_tpg_attrib_cit */
+
+/* Start items for lio_target_tpg_param_cit */
+
+#define DEF_TPG_PARAM(name) \
+static ssize_t iscsi_tpg_param_show_##name( \
+ struct se_portal_group *se_tpg, \
+ char *page) \
+{ \
+ struct iscsi_portal_group *tpg = container_of(se_tpg, \
+ struct iscsi_portal_group, tpg_se_tpg); \
+ struct iscsi_param *param; \
+ ssize_t rb; \
+ \
+ if (iscsit_get_tpg(tpg) < 0) \
+ return -EINVAL; \
+ \
+ param = iscsi_find_param_from_key(__stringify(name), \
+ tpg->param_list); \
+ if (!param) { \
+ iscsit_put_tpg(tpg); \
+ return -EINVAL; \
+ } \
+ rb = snprintf(page, PAGE_SIZE, "%s\n", param->value); \
+ \
+ iscsit_put_tpg(tpg); \
+ return rb; \
+} \
+static ssize_t iscsi_tpg_param_store_##name( \
+ struct se_portal_group *se_tpg, \
+ const char *page, \
+ size_t count) \
+{ \
+ struct iscsi_portal_group *tpg = container_of(se_tpg, \
+ struct iscsi_portal_group, tpg_se_tpg); \
+ char *buf; \
+ int ret; \
+ \
+ buf = kzalloc(PAGE_SIZE, GFP_KERNEL); \
+ if (!buf) \
+ return -ENOMEM; \
+ snprintf(buf, PAGE_SIZE, "%s=%s", __stringify(name), page); \
+ buf[strlen(buf)-1] = '\0'; /* Kill newline */ \
+ \
+ if (iscsit_get_tpg(tpg) < 0) { \
+ kfree(buf); \
+ return -EINVAL; \
+ } \
+ \
+ ret = iscsi_change_param_value(buf, tpg->param_list, 1); \
+ if (ret < 0) \
+ goto out; \
+ \
+ kfree(buf); \
+ iscsit_put_tpg(tpg); \
+ return count; \
+out: \
+ kfree(buf); \
+ iscsit_put_tpg(tpg); \
+ return -EINVAL; \
+}
+
+#define TPG_PARAM_ATTR(_name, _mode) TF_TPG_PARAM_ATTR(iscsi, _name, _mode);
+
+DEF_TPG_PARAM(AuthMethod);
+TPG_PARAM_ATTR(AuthMethod, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(HeaderDigest);
+TPG_PARAM_ATTR(HeaderDigest, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(DataDigest);
+TPG_PARAM_ATTR(DataDigest, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(MaxConnections);
+TPG_PARAM_ATTR(MaxConnections, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(TargetAlias);
+TPG_PARAM_ATTR(TargetAlias, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(InitialR2T);
+TPG_PARAM_ATTR(InitialR2T, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(ImmediateData);
+TPG_PARAM_ATTR(ImmediateData, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(MaxRecvDataSegmentLength);
+TPG_PARAM_ATTR(MaxRecvDataSegmentLength, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(MaxBurstLength);
+TPG_PARAM_ATTR(MaxBurstLength, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(FirstBurstLength);
+TPG_PARAM_ATTR(FirstBurstLength, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(DefaultTime2Wait);
+TPG_PARAM_ATTR(DefaultTime2Wait, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(DefaultTime2Retain);
+TPG_PARAM_ATTR(DefaultTime2Retain, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(MaxOutstandingR2T);
+TPG_PARAM_ATTR(MaxOutstandingR2T, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(DataPDUInOrder);
+TPG_PARAM_ATTR(DataPDUInOrder, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(DataSequenceInOrder);
+TPG_PARAM_ATTR(DataSequenceInOrder, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(ErrorRecoveryLevel);
+TPG_PARAM_ATTR(ErrorRecoveryLevel, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(IFMarker);
+TPG_PARAM_ATTR(IFMarker, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(OFMarker);
+TPG_PARAM_ATTR(OFMarker, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(IFMarkInt);
+TPG_PARAM_ATTR(IFMarkInt, S_IRUGO | S_IWUSR);
+
+DEF_TPG_PARAM(OFMarkInt);
+TPG_PARAM_ATTR(OFMarkInt, S_IRUGO | S_IWUSR);
+
+static struct configfs_attribute *lio_target_tpg_param_attrs[] = {
+ &iscsi_tpg_param_AuthMethod.attr,
+ &iscsi_tpg_param_HeaderDigest.attr,
+ &iscsi_tpg_param_DataDigest.attr,
+ &iscsi_tpg_param_MaxConnections.attr,
+ &iscsi_tpg_param_TargetAlias.attr,
+ &iscsi_tpg_param_InitialR2T.attr,
+ &iscsi_tpg_param_ImmediateData.attr,
+ &iscsi_tpg_param_MaxRecvDataSegmentLength.attr,
+ &iscsi_tpg_param_MaxBurstLength.attr,
+ &iscsi_tpg_param_FirstBurstLength.attr,
+ &iscsi_tpg_param_DefaultTime2Wait.attr,
+ &iscsi_tpg_param_DefaultTime2Retain.attr,
+ &iscsi_tpg_param_MaxOutstandingR2T.attr,
+ &iscsi_tpg_param_DataPDUInOrder.attr,
+ &iscsi_tpg_param_DataSequenceInOrder.attr,
+ &iscsi_tpg_param_ErrorRecoveryLevel.attr,
+ &iscsi_tpg_param_IFMarker.attr,
+ &iscsi_tpg_param_OFMarker.attr,
+ &iscsi_tpg_param_IFMarkInt.attr,
+ &iscsi_tpg_param_OFMarkInt.attr,
+ NULL,
+};
+
+/* End items for lio_target_tpg_param_cit */
+
+/* Start items for lio_target_tpg_cit */
+
+static ssize_t lio_target_tpg_show_enable(
+ struct se_portal_group *se_tpg,
+ char *page)
+{
+ struct iscsi_portal_group *tpg = container_of(se_tpg,
+ struct iscsi_portal_group, tpg_se_tpg);
+ ssize_t len;
+
+ spin_lock(&tpg->tpg_state_lock);
+ len = sprintf(page, "%d\n",
+ (tpg->tpg_state == TPG_STATE_ACTIVE) ? 1 : 0);
+ spin_unlock(&tpg->tpg_state_lock);
+
+ return len;
+}
+
+static ssize_t lio_target_tpg_store_enable(
+ struct se_portal_group *se_tpg,
+ const char *page,
+ size_t count)
+{
+ struct iscsi_portal_group *tpg = container_of(se_tpg,
+ struct iscsi_portal_group, tpg_se_tpg);
+ char *endptr;
+ u32 op;
+ int ret = 0;
+
+ op = simple_strtoul(page, &endptr, 0);
+ if ((op != 1) && (op != 0)) {
+ printk(KERN_ERR "Illegal value for tpg_enable: %u\n", op);
+ return -EINVAL;
+ }
+
+ ret = iscsit_get_tpg(tpg);
+ if (ret < 0)
+ return -EINVAL;
+
+ if (op) {
+ ret = iscsit_tpg_enable_portal_group(tpg);
+ if (ret < 0)
+ goto out;
+ } else {
+ /*
+ * iscsit_tpg_disable_portal_group() assumes force=1
+ */
+ ret = iscsit_tpg_disable_portal_group(tpg, 1);
+ if (ret < 0)
+ goto out;
+ }
+
+ iscsit_put_tpg(tpg);
+ return count;
+out:
+ iscsit_put_tpg(tpg);
+ return -EINVAL;
+}
+
+TF_TPG_BASE_ATTR(lio_target, enable, S_IRUGO | S_IWUSR);
+
+static struct configfs_attribute *lio_target_tpg_attrs[] = {
+ &lio_target_tpg_enable.attr,
+ NULL,
+};
+
+/* End items for lio_target_tpg_cit */
+
+/* Start items for lio_target_tiqn_cit */
+
+struct se_portal_group *lio_target_tiqn_addtpg(
+ struct se_wwn *wwn,
+ struct config_group *group,
+ const char *name)
+{
+ struct iscsi_portal_group *tpg;
+ struct iscsi_tiqn *tiqn;
+ char *tpgt_str, *end_ptr;
+ int ret = 0;
+ unsigned short int tpgt;
+
+ tiqn = container_of(wwn, struct iscsi_tiqn, tiqn_wwn);
+ /*
+ * Only tpgt_# directory groups can be created below
+ * target/iscsi/iqn.superturodiskarry/
+ */
+ tpgt_str = strstr(name, "tpgt_");
+ if (!tpgt_str) {
+ printk(KERN_ERR "Unable to locate \"tpgt_#\" directory"
+ " group\n");
+ return NULL;
+ }
+ tpgt_str += 5; /* Skip ahead of "tpgt_" */
+ tpgt = (unsigned short int) simple_strtoul(tpgt_str, &end_ptr, 0);
+
+ tpg = iscsit_alloc_portal_group(tiqn, tpgt);
+ if (!tpg)
+ return NULL;
+
+ ret = core_tpg_register(
+ &lio_target_fabric_configfs->tf_ops,
+ wwn, &tpg->tpg_se_tpg, (void *)tpg,
+ TRANSPORT_TPG_TYPE_NORMAL);
+ if (ret < 0)
+ return NULL;
+
+ ret = iscsit_tpg_add_portal_group(tiqn, tpg);
+ if (ret != 0)
+ goto out;
+
+ printk(KERN_INFO "LIO_Target_ConfigFS: REGISTER -> %s\n", tiqn->tiqn);
+ printk(KERN_INFO "LIO_Target_ConfigFS: REGISTER -> Allocated TPG: %s\n",
+ name);
+ return &tpg->tpg_se_tpg;
+out:
+ core_tpg_deregister(&tpg->tpg_se_tpg);
+ kfree(tpg);
+ return NULL;
+}
+
+void lio_target_tiqn_deltpg(struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg;
+ struct iscsi_tiqn *tiqn;
+
+ tpg = container_of(se_tpg, struct iscsi_portal_group, tpg_se_tpg);
+ tiqn = tpg->tpg_tiqn;
+ /*
+ * iscsit_tpg_del_portal_group() assumes force=1
+ */
+ printk(KERN_INFO "LIO_Target_ConfigFS: DEREGISTER -> Releasing TPG\n");
+ iscsit_tpg_del_portal_group(tiqn, tpg, 1);
+}
+
+/* End items for lio_target_tiqn_cit */
+
+/* Start LIO-Target TIQN struct contig_item lio_target_cit */
+
+static ssize_t lio_target_wwn_show_attr_lio_version(
+ struct target_fabric_configfs *tf,
+ char *page)
+{
+ return sprintf(page, "RisingTide Systems Linux-iSCSI Target "ISCSI_VERSION"\n");
+}
+
+TF_WWN_ATTR_RO(lio_target, lio_version);
+
+static struct configfs_attribute *lio_target_wwn_attrs[] = {
+ &lio_target_wwn_lio_version.attr,
+ NULL,
+};
+
+struct se_wwn *lio_target_call_coreaddtiqn(
+ struct target_fabric_configfs *tf,
+ struct config_group *group,
+ const char *name)
+{
+ struct config_group *stats_cg;
+ struct iscsi_tiqn *tiqn;
+ int ret = 0;
+
+ tiqn = iscsit_add_tiqn((unsigned char *)name, &ret);
+ if (!tiqn)
+ return NULL;
+ /*
+ * Setup struct iscsi_wwn_stat_grps for se_wwn->fabric_stat_group.
+ */
+ stats_cg = &tiqn->tiqn_wwn.fabric_stat_group;
+
+ stats_cg->default_groups = kzalloc(sizeof(struct config_group) * 6,
+ GFP_KERNEL);
+ if (!stats_cg->default_groups) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " stats_cg->default_groups\n");
+ iscsit_del_tiqn(tiqn);
+ return ERR_PTR(-ENOMEM);
+ }
+
+ stats_cg->default_groups[0] = &WWN_STAT_GRPS(tiqn)->iscsi_instance_group;
+ stats_cg->default_groups[1] = &WWN_STAT_GRPS(tiqn)->iscsi_sess_err_group;
+ stats_cg->default_groups[2] = &WWN_STAT_GRPS(tiqn)->iscsi_tgt_attr_group;
+ stats_cg->default_groups[3] = &WWN_STAT_GRPS(tiqn)->iscsi_login_stats_group;
+ stats_cg->default_groups[4] = &WWN_STAT_GRPS(tiqn)->iscsi_logout_stats_group;
+ stats_cg->default_groups[5] = NULL;
+ config_group_init_type_name(&WWN_STAT_GRPS(tiqn)->iscsi_instance_group,
+ "iscsi_instance", &iscsi_stat_instance_cit);
+ config_group_init_type_name(&WWN_STAT_GRPS(tiqn)->iscsi_sess_err_group,
+ "iscsi_sess_err", &iscsi_stat_sess_err_cit);
+ config_group_init_type_name(&WWN_STAT_GRPS(tiqn)->iscsi_tgt_attr_group,
+ "iscsi_tgt_attr", &iscsi_stat_tgt_attr_cit);
+ config_group_init_type_name(&WWN_STAT_GRPS(tiqn)->iscsi_login_stats_group,
+ "iscsi_login_stats", &iscsi_stat_login_cit);
+ config_group_init_type_name(&WWN_STAT_GRPS(tiqn)->iscsi_logout_stats_group,
+ "iscsi_logout_stats", &iscsi_stat_logout_cit);
+
+ printk(KERN_INFO "LIO_Target_ConfigFS: REGISTER -> %s\n", tiqn->tiqn);
+ printk(KERN_INFO "LIO_Target_ConfigFS: REGISTER -> Allocated Node:"
+ " %s\n", name);
+ return &tiqn->tiqn_wwn;
+}
+
+void lio_target_call_coredeltiqn(
+ struct se_wwn *wwn)
+{
+ struct iscsi_tiqn *tiqn = container_of(wwn, struct iscsi_tiqn, tiqn_wwn);
+ struct config_item *df_item;
+ struct config_group *stats_cg;
+ int i;
+
+ stats_cg = &tiqn->tiqn_wwn.fabric_stat_group;
+ for (i = 0; stats_cg->default_groups[i]; i++) {
+ df_item = &stats_cg->default_groups[i]->cg_item;
+ stats_cg->default_groups[i] = NULL;
+ config_item_put(df_item);
+ }
+ kfree(stats_cg->default_groups);
+
+ printk(KERN_INFO "LIO_Target_ConfigFS: DEREGISTER -> %s\n",
+ tiqn->tiqn);
+ iscsit_del_tiqn(tiqn);
+}
+
+/* End LIO-Target TIQN struct contig_lio_target_cit */
+
+/* Start lio_target_discovery_auth_cit */
+
+#define DEF_DISC_AUTH_STR(name, flags) \
+ __DEF_NACL_AUTH_STR(disc, name, flags) \
+static ssize_t iscsi_disc_show_##name( \
+ struct target_fabric_configfs *tf, \
+ char *page) \
+{ \
+ return __iscsi_disc_show_##name(&iscsi_global->discovery_acl, \
+ page); \
+} \
+static ssize_t iscsi_disc_store_##name( \
+ struct target_fabric_configfs *tf, \
+ const char *page, \
+ size_t count) \
+{ \
+ return __iscsi_disc_store_##name(&iscsi_global->discovery_acl, \
+ page, count); \
+}
+
+#define DEF_DISC_AUTH_INT(name) \
+ __DEF_NACL_AUTH_INT(disc, name) \
+static ssize_t iscsi_disc_show_##name( \
+ struct target_fabric_configfs *tf, \
+ char *page) \
+{ \
+ return __iscsi_disc_show_##name(&iscsi_global->discovery_acl, \
+ page); \
+}
+
+#define DISC_AUTH_ATTR(_name, _mode) TF_DISC_ATTR(iscsi, _name, _mode)
+#define DISC_AUTH_ATTR_RO(_name) TF_DISC_ATTR_RO(iscsi, _name)
+
+/*
+ * One-way authentication userid
+ */
+DEF_DISC_AUTH_STR(userid, NAF_USERID_SET);
+DISC_AUTH_ATTR(userid, S_IRUGO | S_IWUSR);
+/*
+ * One-way authentication password
+ */
+DEF_DISC_AUTH_STR(password, NAF_PASSWORD_SET);
+DISC_AUTH_ATTR(password, S_IRUGO | S_IWUSR);
+/*
+ * Enforce mutual authentication
+ */
+DEF_DISC_AUTH_INT(authenticate_target);
+DISC_AUTH_ATTR_RO(authenticate_target);
+/*
+ * Mutual authentication userid
+ */
+DEF_DISC_AUTH_STR(userid_mutual, NAF_USERID_IN_SET);
+DISC_AUTH_ATTR(userid_mutual, S_IRUGO | S_IWUSR);
+/*
+ * Mutual authentication password
+ */
+DEF_DISC_AUTH_STR(password_mutual, NAF_PASSWORD_IN_SET);
+DISC_AUTH_ATTR(password_mutual, S_IRUGO | S_IWUSR);
+
+/*
+ * enforce_discovery_auth
+ */
+static ssize_t iscsi_disc_show_enforce_discovery_auth(
+ struct target_fabric_configfs *tf,
+ char *page)
+{
+ struct iscsi_node_auth *discovery_auth = &iscsi_global->discovery_acl.node_auth;
+
+ return sprintf(page, "%d\n", discovery_auth->enforce_discovery_auth);
+}
+
+static ssize_t iscsi_disc_store_enforce_discovery_auth(
+ struct target_fabric_configfs *tf,
+ const char *page,
+ size_t count)
+{
+ struct iscsi_param *param;
+ struct iscsi_portal_group *discovery_tpg = iscsi_global->discovery_tpg;
+ char *endptr;
+ u32 op;
+
+ op = simple_strtoul(page, &endptr, 0);
+ if ((op != 1) && (op != 0)) {
+ printk(KERN_ERR "Illegal value for enforce_discovery_auth:"
+ " %u\n", op);
+ return -EINVAL;
+ }
+
+ if (!discovery_tpg) {
+ printk(KERN_ERR "iscsi_global->discovery_tpg is NULL\n");
+ return -EINVAL;
+ }
+
+ param = iscsi_find_param_from_key(AUTHMETHOD,
+ discovery_tpg->param_list);
+ if (!param)
+ return -EINVAL;
+
+ if (op) {
+ /*
+ * Reset the AuthMethod key to CHAP.
+ */
+ if (iscsi_update_param_value(param, CHAP) < 0)
+ return -EINVAL;
+
+ discovery_tpg->tpg_attrib.authentication = 1;
+ iscsi_global->discovery_acl.node_auth.enforce_discovery_auth = 1;
+ printk(KERN_INFO "LIO-CORE[0] Successfully enabled"
+ " authentication enforcement for iSCSI"
+ " Discovery TPG\n");
+ } else {
+ /*
+ * Reset the AuthMethod key to CHAP,None
+ */
+ if (iscsi_update_param_value(param, "CHAP,None") < 0)
+ return -EINVAL;
+
+ discovery_tpg->tpg_attrib.authentication = 0;
+ iscsi_global->discovery_acl.node_auth.enforce_discovery_auth = 0;
+ printk(KERN_INFO "LIO-CORE[0] Successfully disabled"
+ " authentication enforcement for iSCSI"
+ " Discovery TPG\n");
+ }
+
+ return count;
+}
+
+DISC_AUTH_ATTR(enforce_discovery_auth, S_IRUGO | S_IWUSR);
+
+static struct configfs_attribute *lio_target_discovery_auth_attrs[] = {
+ &iscsi_disc_userid.attr,
+ &iscsi_disc_password.attr,
+ &iscsi_disc_authenticate_target.attr,
+ &iscsi_disc_userid_mutual.attr,
+ &iscsi_disc_password_mutual.attr,
+ &iscsi_disc_enforce_discovery_auth.attr,
+ NULL,
+};
+
+/* End lio_target_discovery_auth_cit */
+
+/* Start functions for target_core_fabric_ops */
+
+static char *iscsi_get_fabric_name(void)
+{
+ return "iSCSI";
+}
+
+static u32 iscsi_get_task_tag(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ return cmd->init_task_tag;
+}
+
+static int iscsi_get_cmd_state(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ return cmd->i_state;
+}
+
+static void iscsi_new_cmd_failure(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ if (cmd->immediate_data || cmd->unsolicited_data)
+ complete(&cmd->unsolicited_data_comp);
+}
+
+static int iscsi_is_state_remove(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ return (cmd->i_state == ISTATE_REMOVE);
+}
+
+static int lio_sess_logged_in(struct se_session *se_sess)
+{
+ struct iscsi_session *sess = se_sess->fabric_sess_ptr;
+ int ret;
+ /*
+ * Called with spin_lock_bh(&se_global->se_tpg_lock); and
+ * spin_lock(&se_tpg->session_lock); held.
+ */
+ spin_lock(&sess->conn_lock);
+ ret = (sess->session_state != TARG_SESS_STATE_LOGGED_IN);
+ spin_unlock(&sess->conn_lock);
+
+ return ret;
+}
+
+static u32 lio_sess_get_index(struct se_session *se_sess)
+{
+ struct iscsi_session *sess = se_sess->fabric_sess_ptr;
+
+ return sess->session_index;
+}
+
+static u32 lio_sess_get_initiator_sid(
+ struct se_session *se_sess,
+ unsigned char *buf,
+ u32 size)
+{
+ struct iscsi_session *sess = se_sess->fabric_sess_ptr;
+ /*
+ * iSCSI Initiator Session Identifier from RFC-3720.
+ */
+ return snprintf(buf, size, "%02x%02x%02x%02x%02x%02x",
+ sess->isid[0], sess->isid[1], sess->isid[2],
+ sess->isid[3], sess->isid[4], sess->isid[5]);
+}
+
+static int lio_queue_data_in(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ cmd->i_state = ISTATE_SEND_DATAIN;
+ iscsit_add_cmd_to_response_queue(cmd, cmd->conn, cmd->i_state);
+ return 0;
+}
+
+static int lio_write_pending(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ if (cmd->immediate_data || cmd->unsolicited_data)
+ complete(&cmd->unsolicited_data_comp);
+ else {
+ if (iscsit_build_r2ts_for_cmd(cmd, cmd->conn, 1) < 0)
+ return PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES;
+ }
+
+ return 0;
+}
+
+static int lio_write_pending_status(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+ int ret;
+
+ spin_lock_bh(&cmd->istate_lock);
+ ret = !(cmd->cmd_flags & ICF_GOT_LAST_DATAOUT);
+ spin_unlock_bh(&cmd->istate_lock);
+
+ return ret;
+}
+
+static int lio_queue_status(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ cmd->i_state = ISTATE_SEND_STATUS;
+ iscsit_add_cmd_to_response_queue(cmd, cmd->conn, cmd->i_state);
+ return 0;
+}
+
+static u16 lio_set_fabric_sense_len(struct se_cmd *se_cmd, u32 sense_length)
+{
+ unsigned char *buffer = se_cmd->sense_buffer;
+ /*
+ * From RFC-3720 10.4.7. Data Segment - Sense and Response Data Segment
+ * 16-bit SenseLength.
+ */
+ buffer[0] = ((sense_length >> 8) & 0xff);
+ buffer[1] = (sense_length & 0xff);
+ /*
+ * Return two byte offset into allocated sense_buffer.
+ */
+ return 2;
+}
+
+static u16 lio_get_fabric_sense_len(void)
+{
+ /*
+ * Return two byte offset into allocated sense_buffer.
+ */
+ return 2;
+}
+
+static int lio_queue_tm_rsp(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ cmd->i_state = ISTATE_SEND_TASKMGTRSP;
+ iscsit_add_cmd_to_response_queue(cmd, cmd->conn, cmd->i_state);
+ return 0;
+}
+
+static char *lio_tpg_get_endpoint_wwn(struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg = se_tpg->se_tpg_fabric_ptr;
+
+ return &tpg->tpg_tiqn->tiqn[0];
+}
+
+static u16 lio_tpg_get_tag(struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg = se_tpg->se_tpg_fabric_ptr;
+
+ return tpg->tpgt;
+}
+
+static u32 lio_tpg_get_default_depth(struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg = se_tpg->se_tpg_fabric_ptr;
+
+ return ISCSI_TPG_ATTRIB(tpg)->default_cmdsn_depth;
+}
+
+static int lio_tpg_check_demo_mode(struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg = se_tpg->se_tpg_fabric_ptr;
+
+ return ISCSI_TPG_ATTRIB(tpg)->generate_node_acls;
+}
+
+static int lio_tpg_check_demo_mode_cache(struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg = se_tpg->se_tpg_fabric_ptr;
+
+ return ISCSI_TPG_ATTRIB(tpg)->cache_dynamic_acls;
+}
+
+static int lio_tpg_check_demo_mode_write_protect(
+ struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg = se_tpg->se_tpg_fabric_ptr;
+
+ return ISCSI_TPG_ATTRIB(tpg)->demo_mode_write_protect;
+}
+
+static int lio_tpg_check_prod_mode_write_protect(
+ struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg = se_tpg->se_tpg_fabric_ptr;
+
+ return ISCSI_TPG_ATTRIB(tpg)->prod_mode_write_protect;
+}
+
+static void lio_tpg_release_fabric_acl(
+ struct se_portal_group *se_tpg,
+ struct se_node_acl *se_acl)
+{
+ struct iscsi_node_acl *acl = container_of(se_acl,
+ struct iscsi_node_acl, se_node_acl);
+ kfree(acl);
+}
+
+/*
+ * Called with spin_lock_bh(struct se_portal_group->session_lock) held..
+ *
+ * Also, this function calls iscsit_inc_session_usage_count() on the
+ * struct iscsi_session in question.
+ */
+static int lio_tpg_shutdown_session(struct se_session *se_sess)
+{
+ struct iscsi_session *sess = se_sess->fabric_sess_ptr;
+
+ spin_lock(&sess->conn_lock);
+ if (atomic_read(&sess->session_fall_back_to_erl0) ||
+ atomic_read(&sess->session_logout) ||
+ (sess->time2retain_timer_flags & ISCSI_TF_EXPIRED)) {
+ spin_unlock(&sess->conn_lock);
+ return 0;
+ }
+ atomic_set(&sess->session_reinstatement, 1);
+ spin_unlock(&sess->conn_lock);
+
+ iscsit_inc_session_usage_count(sess);
+ iscsit_stop_time2retain_timer(sess);
+
+ return 1;
+}
+
+/*
+ * Calls iscsit_dec_session_usage_count() as inverse of
+ * lio_tpg_shutdown_session()
+ */
+static void lio_tpg_close_session(struct se_session *se_sess)
+{
+ struct iscsi_session *sess = se_sess->fabric_sess_ptr;
+ /*
+ * If the iSCSI Session for the iSCSI Initiator Node exists,
+ * forcefully shutdown the iSCSI NEXUS.
+ */
+ iscsit_stop_session(sess, 1, 1);
+ iscsit_dec_session_usage_count(sess);
+ iscsit_close_session(sess);
+}
+
+static void lio_tpg_stop_session(
+ struct se_session *se_sess,
+ int sess_sleep,
+ int conn_sleep)
+{
+ struct iscsi_session *sess = se_sess->fabric_sess_ptr;
+
+ iscsit_stop_session(sess, sess_sleep, conn_sleep);
+}
+
+static void lio_tpg_fall_back_to_erl0(struct se_session *se_sess)
+{
+ struct iscsi_session *sess = se_sess->fabric_sess_ptr;
+
+ iscsit_fall_back_to_erl0(sess);
+}
+
+static u32 lio_tpg_get_inst_index(struct se_portal_group *se_tpg)
+{
+ struct iscsi_portal_group *tpg = se_tpg->se_tpg_fabric_ptr;
+
+ return tpg->tpg_tiqn->tiqn_index;
+}
+
+static void lio_set_default_node_attributes(struct se_node_acl *se_acl)
+{
+ struct iscsi_node_acl *acl = container_of(se_acl, struct iscsi_node_acl,
+ se_node_acl);
+
+ ISCSI_NODE_ATTRIB(acl)->nacl = acl;
+ iscsit_set_default_node_attribues(acl);
+}
+
+static int iscsi_allocate_iovecs_for_cmd(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+ u32 iov_count = (T_TASK(se_cmd)->t_tasks_se_num == 0) ? 1 :
+ T_TASK(se_cmd)->t_tasks_se_num;
+
+ iov_count += TRANSPORT_IOV_DATA_BUFFER;
+
+ cmd->iov_data = kzalloc(iov_count * sizeof(struct kvec), GFP_KERNEL);
+ if (!cmd->iov_data)
+ return -ENOMEM;
+
+ cmd->orig_iov_data_count = iov_count;
+ return 0;
+}
+
+static void lio_release_cmd_direct(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ iscsit_release_cmd(cmd);
+}
+
+static void lio_release_cmd(struct se_cmd *se_cmd)
+{
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ iscsit_release_cmd(cmd);
+}
+
+/* End functions for target_core_fabric_ops */
+
+int iscsi_target_register_configfs(void)
+{
+ struct target_fabric_configfs *fabric;
+ int ret;
+
+ lio_target_fabric_configfs = NULL;
+ fabric = target_fabric_configfs_init(THIS_MODULE, "iscsi");
+ if (!fabric) {
+ printk(KERN_ERR "target_fabric_configfs_init() for"
+ " LIO-Target failed!\n");
+ return -1;
+ }
+ /*
+ * Setup the fabric API of function pointers used by target_core_mod..
+ */
+ fabric->tf_ops.get_fabric_name = &iscsi_get_fabric_name;
+ fabric->tf_ops.get_fabric_proto_ident = &iscsi_get_fabric_proto_ident;
+ fabric->tf_ops.tpg_get_wwn = &lio_tpg_get_endpoint_wwn;
+ fabric->tf_ops.tpg_get_tag = &lio_tpg_get_tag;
+ fabric->tf_ops.tpg_get_default_depth = &lio_tpg_get_default_depth;
+ fabric->tf_ops.tpg_get_pr_transport_id = &iscsi_get_pr_transport_id;
+ fabric->tf_ops.tpg_get_pr_transport_id_len =
+ &iscsi_get_pr_transport_id_len;
+ fabric->tf_ops.tpg_parse_pr_out_transport_id =
+ &iscsi_parse_pr_out_transport_id;
+ fabric->tf_ops.tpg_check_demo_mode = &lio_tpg_check_demo_mode;
+ fabric->tf_ops.tpg_check_demo_mode_cache =
+ &lio_tpg_check_demo_mode_cache;
+ fabric->tf_ops.tpg_check_demo_mode_write_protect =
+ &lio_tpg_check_demo_mode_write_protect;
+ fabric->tf_ops.tpg_check_prod_mode_write_protect =
+ &lio_tpg_check_prod_mode_write_protect;
+ fabric->tf_ops.tpg_alloc_fabric_acl = &lio_tpg_alloc_fabric_acl;
+ fabric->tf_ops.tpg_release_fabric_acl = &lio_tpg_release_fabric_acl;
+ fabric->tf_ops.tpg_get_inst_index = &lio_tpg_get_inst_index;
+ /*
+ * Use our local iscsi_allocate_iovecs_for_cmd() for the extra
+ * callback in transport_generic_new_cmd() to allocate
+ * iscsi_cmd->iov_data[] for Linux/Net kernel sockets operations.
+ */
+ fabric->tf_ops.alloc_cmd_iovecs = &iscsi_allocate_iovecs_for_cmd;
+ fabric->tf_ops.release_cmd_to_pool = &lio_release_cmd;
+ fabric->tf_ops.release_cmd_direct = &lio_release_cmd_direct;
+ fabric->tf_ops.shutdown_session = &lio_tpg_shutdown_session;
+ fabric->tf_ops.close_session = &lio_tpg_close_session;
+ fabric->tf_ops.stop_session = &lio_tpg_stop_session;
+ fabric->tf_ops.fall_back_to_erl0 = &lio_tpg_fall_back_to_erl0;
+ fabric->tf_ops.sess_logged_in = &lio_sess_logged_in;
+ fabric->tf_ops.sess_get_index = &lio_sess_get_index;
+ fabric->tf_ops.sess_get_initiator_sid = &lio_sess_get_initiator_sid;
+ fabric->tf_ops.write_pending = &lio_write_pending;
+ fabric->tf_ops.write_pending_status = &lio_write_pending_status;
+ fabric->tf_ops.set_default_node_attributes =
+ &lio_set_default_node_attributes;
+ fabric->tf_ops.get_task_tag = &iscsi_get_task_tag;
+ fabric->tf_ops.get_cmd_state = &iscsi_get_cmd_state;
+ fabric->tf_ops.new_cmd_failure = &iscsi_new_cmd_failure;
+ fabric->tf_ops.queue_data_in = &lio_queue_data_in;
+ fabric->tf_ops.queue_status = &lio_queue_status;
+ fabric->tf_ops.queue_tm_rsp = &lio_queue_tm_rsp;
+ fabric->tf_ops.set_fabric_sense_len = &lio_set_fabric_sense_len;
+ fabric->tf_ops.get_fabric_sense_len = &lio_get_fabric_sense_len;
+ fabric->tf_ops.is_state_remove = &iscsi_is_state_remove;
+ /*
+ * Setup function pointers for generic logic in target_core_fabric_configfs.c
+ */
+ fabric->tf_ops.fabric_make_wwn = &lio_target_call_coreaddtiqn;
+ fabric->tf_ops.fabric_drop_wwn = &lio_target_call_coredeltiqn;
+ fabric->tf_ops.fabric_make_tpg = &lio_target_tiqn_addtpg;
+ fabric->tf_ops.fabric_drop_tpg = &lio_target_tiqn_deltpg;
+ fabric->tf_ops.fabric_post_link = NULL;
+ fabric->tf_ops.fabric_pre_unlink = NULL;
+ fabric->tf_ops.fabric_make_np = &lio_target_call_addnptotpg;
+ fabric->tf_ops.fabric_drop_np = &lio_target_call_delnpfromtpg;
+ fabric->tf_ops.fabric_make_nodeacl = &lio_target_make_nodeacl;
+ fabric->tf_ops.fabric_drop_nodeacl = &lio_target_drop_nodeacl;
+ /*
+ * Setup default attribute lists for various fabric->tf_cit_tmpl
+ * sturct config_item_type's
+ */
+ TF_CIT_TMPL(fabric)->tfc_discovery_cit.ct_attrs = lio_target_discovery_auth_attrs;
+ TF_CIT_TMPL(fabric)->tfc_wwn_cit.ct_attrs = lio_target_wwn_attrs;
+ TF_CIT_TMPL(fabric)->tfc_tpg_base_cit.ct_attrs = lio_target_tpg_attrs;
+ TF_CIT_TMPL(fabric)->tfc_tpg_attrib_cit.ct_attrs = lio_target_tpg_attrib_attrs;
+ TF_CIT_TMPL(fabric)->tfc_tpg_param_cit.ct_attrs = lio_target_tpg_param_attrs;
+ TF_CIT_TMPL(fabric)->tfc_tpg_np_base_cit.ct_attrs = lio_target_portal_attrs;
+ TF_CIT_TMPL(fabric)->tfc_tpg_nacl_base_cit.ct_attrs = lio_target_initiator_attrs;
+ TF_CIT_TMPL(fabric)->tfc_tpg_nacl_attrib_cit.ct_attrs = lio_target_nacl_attrib_attrs;
+ TF_CIT_TMPL(fabric)->tfc_tpg_nacl_auth_cit.ct_attrs = lio_target_nacl_auth_attrs;
+ TF_CIT_TMPL(fabric)->tfc_tpg_nacl_param_cit.ct_attrs = lio_target_nacl_param_attrs;
+
+ ret = target_fabric_configfs_register(fabric);
+ if (ret < 0) {
+ printk(KERN_ERR "target_fabric_configfs_register() for"
+ " LIO-Target failed!\n");
+ target_fabric_configfs_free(fabric);
+ return -1;
+ }
+
+ lio_target_fabric_configfs = fabric;
+ printk(KERN_INFO "LIO_TARGET[0] - Set fabric ->"
+ " lio_target_fabric_configfs\n");
+ return 0;
+}
+
+
+void iscsi_target_deregister_configfs(void)
+{
+ if (!lio_target_fabric_configfs)
+ return;
+ /*
+ * Shutdown discovery sessions and disable discovery TPG
+ */
+ if (iscsi_global->discovery_tpg)
+ iscsit_tpg_disable_portal_group(iscsi_global->discovery_tpg, 1);
+
+ target_fabric_configfs_deregister(lio_target_fabric_configfs);
+ lio_target_fabric_configfs = NULL;
+ printk(KERN_INFO "LIO_TARGET[0] - Cleared"
+ " lio_target_fabric_configfs\n");
+}
diff --git a/drivers/target/iscsi/iscsi_target_configfs.h b/drivers/target/iscsi/iscsi_target_configfs.h
new file mode 100644
index 0000000..8cd5a63
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_configfs.h
@@ -0,0 +1,7 @@
+#ifndef ISCSI_TARGET_CONFIGFS_H
+#define ISCSI_TARGET_CONFIGFS_H
+
+extern int iscsi_target_register_configfs(void);
+extern void iscsi_target_deregister_configfs(void);
+
+#endif /* ISCSI_TARGET_CONFIGFS_H */
diff --git a/drivers/target/iscsi/iscsi_target_nodeattrib.c b/drivers/target/iscsi/iscsi_target_nodeattrib.c
new file mode 100644
index 0000000..2cb38c4
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_nodeattrib.c
@@ -0,0 +1,264 @@
+/*******************************************************************************
+ * This file contains the main functions related to Initiator Node Attributes.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_device.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target_nodeattrib.h"
+
+static inline char *iscsit_na_get_initiatorname(
+ struct iscsi_node_acl *nacl)
+{
+ struct se_node_acl *se_nacl = &nacl->se_node_acl;
+
+ return &se_nacl->initiatorname[0];
+}
+
+void iscsit_set_default_node_attribues(
+ struct iscsi_node_acl *acl)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+
+ a->dataout_timeout = NA_DATAOUT_TIMEOUT;
+ a->dataout_timeout_retries = NA_DATAOUT_TIMEOUT_RETRIES;
+ a->nopin_timeout = NA_NOPIN_TIMEOUT;
+ a->nopin_response_timeout = NA_NOPIN_RESPONSE_TIMEOUT;
+ a->random_datain_pdu_offsets = NA_RANDOM_DATAIN_PDU_OFFSETS;
+ a->random_datain_seq_offsets = NA_RANDOM_DATAIN_SEQ_OFFSETS;
+ a->random_r2t_offsets = NA_RANDOM_R2T_OFFSETS;
+ a->default_erl = NA_DEFAULT_ERL;
+}
+
+extern int iscsit_na_dataout_timeout(
+ struct iscsi_node_acl *acl,
+ u32 dataout_timeout)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+
+ if (dataout_timeout > NA_DATAOUT_TIMEOUT_MAX) {
+ printk(KERN_ERR "Requested DataOut Timeout %u larger than"
+ " maximum %u\n", dataout_timeout,
+ NA_DATAOUT_TIMEOUT_MAX);
+ return -EINVAL;
+ } else if (dataout_timeout < NA_DATAOUT_TIMEOUT_MIX) {
+ printk(KERN_ERR "Requested DataOut Timeout %u smaller than"
+ " minimum %u\n", dataout_timeout,
+ NA_DATAOUT_TIMEOUT_MIX);
+ return -EINVAL;
+ }
+
+ a->dataout_timeout = dataout_timeout;
+ TRACE(TRACE_NODEATTRIB, "Set DataOut Timeout to %u for Initiator Node"
+ " %s\n", a->dataout_timeout, iscsi_na_get_initiatorname(acl));
+
+ return 0;
+}
+
+extern int iscsit_na_dataout_timeout_retries(
+ struct iscsi_node_acl *acl,
+ u32 dataout_timeout_retries)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+
+ if (dataout_timeout_retries > NA_DATAOUT_TIMEOUT_RETRIES_MAX) {
+ printk(KERN_ERR "Requested DataOut Timeout Retries %u larger"
+ " than maximum %u", dataout_timeout_retries,
+ NA_DATAOUT_TIMEOUT_RETRIES_MAX);
+ return -EINVAL;
+ } else if (dataout_timeout_retries < NA_DATAOUT_TIMEOUT_RETRIES_MIN) {
+ printk(KERN_ERR "Requested DataOut Timeout Retries %u smaller"
+ " than minimum %u", dataout_timeout_retries,
+ NA_DATAOUT_TIMEOUT_RETRIES_MIN);
+ return -EINVAL;
+ }
+
+ a->dataout_timeout_retries = dataout_timeout_retries;
+ TRACE(TRACE_NODEATTRIB, "Set DataOut Timeout Retries to %u for"
+ " Initiator Node %s\n", a->dataout_timeout_retries,
+ iscsi_na_get_initiatorname(acl));
+
+ return 0;
+}
+
+extern int iscsit_na_nopin_timeout(
+ struct iscsi_node_acl *acl,
+ u32 nopin_timeout)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+ struct iscsi_session *sess;
+ struct iscsi_conn *conn;
+ struct se_node_acl *se_nacl = &a->nacl->se_node_acl;
+ struct se_session *se_sess;
+ u32 orig_nopin_timeout = a->nopin_timeout;
+
+ if (nopin_timeout > NA_NOPIN_TIMEOUT_MAX) {
+ printk(KERN_ERR "Requested NopIn Timeout %u larger than maximum"
+ " %u\n", nopin_timeout, NA_NOPIN_TIMEOUT_MAX);
+ return -EINVAL;
+ } else if ((nopin_timeout < NA_NOPIN_TIMEOUT_MIN) &&
+ (nopin_timeout != 0)) {
+ printk(KERN_ERR "Requested NopIn Timeout %u smaller than"
+ " minimum %u and not 0\n", nopin_timeout,
+ NA_NOPIN_TIMEOUT_MIN);
+ return -EINVAL;
+ }
+
+ a->nopin_timeout = nopin_timeout;
+ TRACE(TRACE_NODEATTRIB, "Set NopIn Timeout to %u for Initiator"
+ " Node %s\n", a->nopin_timeout,
+ iscsi_na_get_initiatorname(acl));
+ /*
+ * Reenable disabled nopin_timeout timer for all iSCSI connections.
+ */
+ if (!orig_nopin_timeout) {
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+
+ spin_lock(&sess->conn_lock);
+ list_for_each_entry(conn, &sess->sess_conn_list,
+ conn_list) {
+ if (conn->conn_state !=
+ TARG_CONN_STATE_LOGGED_IN)
+ continue;
+
+ spin_lock(&conn->nopin_timer_lock);
+ __iscsit_start_nopin_timer(conn);
+ spin_unlock(&conn->nopin_timer_lock);
+ }
+ spin_unlock(&sess->conn_lock);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+ }
+
+ return 0;
+}
+
+extern int iscsit_na_nopin_response_timeout(
+ struct iscsi_node_acl *acl,
+ u32 nopin_response_timeout)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+
+ if (nopin_response_timeout > NA_NOPIN_RESPONSE_TIMEOUT_MAX) {
+ printk(KERN_ERR "Requested NopIn Response Timeout %u larger"
+ " than maximum %u\n", nopin_response_timeout,
+ NA_NOPIN_RESPONSE_TIMEOUT_MAX);
+ return -EINVAL;
+ } else if (nopin_response_timeout < NA_NOPIN_RESPONSE_TIMEOUT_MIN) {
+ printk(KERN_ERR "Requested NopIn Response Timeout %u smaller"
+ " than minimum %u\n", nopin_response_timeout,
+ NA_NOPIN_RESPONSE_TIMEOUT_MIN);
+ return -EINVAL;
+ }
+
+ a->nopin_response_timeout = nopin_response_timeout;
+ TRACE(TRACE_NODEATTRIB, "Set NopIn Response Timeout to %u for"
+ " Initiator Node %s\n", a->nopin_timeout,
+ iscsi_na_get_initiatorname(acl));
+
+ return 0;
+}
+
+extern int iscsit_na_random_datain_pdu_offsets(
+ struct iscsi_node_acl *acl,
+ u32 random_datain_pdu_offsets)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+
+ if (random_datain_pdu_offsets != 0 && random_datain_pdu_offsets != 1) {
+ printk(KERN_ERR "Requested Random DataIN PDU Offsets: %u not"
+ " 0 or 1\n", random_datain_pdu_offsets);
+ return -EINVAL;
+ }
+
+ a->random_datain_pdu_offsets = random_datain_pdu_offsets;
+ TRACE(TRACE_NODEATTRIB, "Set Random DataIN PDU Offsets to %u for"
+ " Initiator Node %s\n", a->random_datain_pdu_offsets,
+ iscsi_na_get_initiatorname(acl));
+
+ return 0;
+}
+
+extern int iscsit_na_random_datain_seq_offsets(
+ struct iscsi_node_acl *acl,
+ u32 random_datain_seq_offsets)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+
+ if (random_datain_seq_offsets != 0 && random_datain_seq_offsets != 1) {
+ printk(KERN_ERR "Requested Random DataIN Sequence Offsets: %u"
+ " not 0 or 1\n", random_datain_seq_offsets);
+ return -EINVAL;
+ }
+
+ a->random_datain_seq_offsets = random_datain_seq_offsets;
+ TRACE(TRACE_NODEATTRIB, "Set Random DataIN Sequence Offsets to %u for"
+ " Initiator Node %s\n", a->random_datain_seq_offsets,
+ iscsi_na_get_initiatorname(acl));
+
+ return 0;
+}
+
+extern int iscsit_na_random_r2t_offsets(
+ struct iscsi_node_acl *acl,
+ u32 random_r2t_offsets)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+
+ if (random_r2t_offsets != 0 && random_r2t_offsets != 1) {
+ printk(KERN_ERR "Requested Random R2T Offsets: %u not"
+ " 0 or 1\n", random_r2t_offsets);
+ return -EINVAL;
+ }
+
+ a->random_r2t_offsets = random_r2t_offsets;
+ TRACE(TRACE_NODEATTRIB, "Set Random R2T Offsets to %u for"
+ " Initiator Node %s\n", a->random_r2t_offsets,
+ iscsi_na_get_initiatorname(acl));
+
+ return 0;
+}
+
+extern int iscsit_na_default_erl(
+ struct iscsi_node_acl *acl,
+ u32 default_erl)
+{
+ struct iscsi_node_attrib *a = &acl->node_attrib;
+
+ if (default_erl != 0 && default_erl != 1 && default_erl != 2) {
+ printk(KERN_ERR "Requested default ERL: %u not 0, 1, or 2\n",
+ default_erl);
+ return -EINVAL;
+ }
+
+ a->default_erl = default_erl;
+ TRACE(TRACE_NODEATTRIB, "Set use ERL0 flag to %u for Initiator"
+ " Node %s\n", a->default_erl,
+ iscsi_na_get_initiatorname(acl));
+
+ return 0;
+}
diff --git a/drivers/target/iscsi/iscsi_target_nodeattrib.h b/drivers/target/iscsi/iscsi_target_nodeattrib.h
new file mode 100644
index 0000000..c970b326
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_nodeattrib.h
@@ -0,0 +1,14 @@
+#ifndef ISCSI_TARGET_NODEATTRIB_H
+#define ISCSI_TARGET_NODEATTRIB_H
+
+extern void iscsit_set_default_node_attribues(struct iscsi_node_acl *);
+extern int iscsit_na_dataout_timeout(struct iscsi_node_acl *, u32);
+extern int iscsit_na_dataout_timeout_retries(struct iscsi_node_acl *, u32);
+extern int iscsit_na_nopin_timeout(struct iscsi_node_acl *, u32);
+extern int iscsit_na_nopin_response_timeout(struct iscsi_node_acl *, u32);
+extern int iscsit_na_random_datain_pdu_offsets(struct iscsi_node_acl *, u32);
+extern int iscsit_na_random_datain_seq_offsets(struct iscsi_node_acl *, u32);
+extern int iscsit_na_random_r2t_offsets(struct iscsi_node_acl *, u32);
+extern int iscsit_na_default_erl(struct iscsi_node_acl *, u32);
+
+#endif /* ISCSI_TARGET_NODEATTRIB_H */
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 06/14] iscsi-target: Add configfs fabric dependent statistics
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (5 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 05/14] iscsi-target: Add TCM v4 compatiable ConfigFS control plane Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 07/14] iscsi-target: Add TPG and Device logic Nicholas A. Bellinger
` (8 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch adds support for iSCSI fabric dependent configfs statistics
using TCM v4 default statistics groups.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_stat.c | 950 ++++++++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_stat.h | 64 ++
2 files changed, 1014 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_stat.c
create mode 100644 drivers/target/iscsi/iscsi_target_stat.h
diff --git a/drivers/target/iscsi/iscsi_target_stat.c b/drivers/target/iscsi/iscsi_target_stat.c
new file mode 100644
index 0000000..745a9e2
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_stat.c
@@ -0,0 +1,950 @@
+/*******************************************************************************
+ * Modern ConfigFS group context specific iSCSI statistics based on original
+ * iscsi_target_mib.c code
+ *
+ * Copyright (c) 2011 Rising Tide Systems
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/configfs.h>
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+#include <target/configfs_macros.h>
+
+#include "iscsi_target_core.h"
+#include "iscsi_target_parameters.h"
+#include "iscsi_target_device.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target_stat.h"
+
+#ifndef INITIAL_JIFFIES
+#define INITIAL_JIFFIES ((unsigned long)(unsigned int) (-300*HZ))
+#endif
+
+/* Instance Attributes Table */
+#define ISCSI_INST_NUM_NODES 1
+#define ISCSI_INST_DESCR "Storage Engine Target"
+#define ISCSI_INST_LAST_FAILURE_TYPE 0
+#define ISCSI_DISCONTINUITY_TIME 0
+
+#define ISCSI_NODE_INDEX 1
+
+#define ISPRINT(a) ((a >= ' ') && (a <= '~'))
+
+/****************************************************************************
+ * iSCSI MIB Tables
+ ****************************************************************************/
+/*
+ * Instance Attributes Table
+ */
+CONFIGFS_EATTR_STRUCT(iscsi_stat_instance, iscsi_wwn_stat_grps);
+#define ISCSI_STAT_INSTANCE_ATTR(_name, _mode) \
+static struct iscsi_stat_instance_attribute \
+ iscsi_stat_instance_##_name = \
+ __CONFIGFS_EATTR(_name, _mode, \
+ iscsi_stat_instance_show_attr_##_name, \
+ iscsi_stat_instance_store_attr_##_name);
+
+#define ISCSI_STAT_INSTANCE_ATTR_RO(_name) \
+static struct iscsi_stat_instance_attribute \
+ iscsi_stat_instance_##_name = \
+ __CONFIGFS_EATTR_RO(_name, \
+ iscsi_stat_instance_show_attr_##_name);
+
+static ssize_t iscsi_stat_instance_show_attr_inst(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", tiqn->tiqn_index);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(inst);
+
+static ssize_t iscsi_stat_instance_show_attr_min_ver(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%u\n", ISCSI_DRAFT20_VERSION);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(min_ver);
+
+static ssize_t iscsi_stat_instance_show_attr_max_ver(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%u\n", ISCSI_DRAFT20_VERSION);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(max_ver);
+
+static ssize_t iscsi_stat_instance_show_attr_portals(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", tiqn->tiqn_num_tpg_nps);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(portals);
+
+static ssize_t iscsi_stat_instance_show_attr_nodes(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%u\n", ISCSI_INST_NUM_NODES);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(nodes);
+
+static ssize_t iscsi_stat_instance_show_attr_sessions(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", tiqn->tiqn_nsessions);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(sessions);
+
+static ssize_t iscsi_stat_instance_show_attr_fail_sess(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_sess_err_stats *sess_err = &tiqn->sess_err_stats;
+ u32 sess_err_count;
+
+ spin_lock_bh(&sess_err->lock);
+ sess_err_count = (sess_err->digest_errors +
+ sess_err->cxn_timeout_errors +
+ sess_err->pdu_format_errors);
+ spin_unlock_bh(&sess_err->lock);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", sess_err_count);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(fail_sess);
+
+static ssize_t iscsi_stat_instance_show_attr_fail_type(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_sess_err_stats *sess_err = &tiqn->sess_err_stats;
+
+ return snprintf(page, PAGE_SIZE, "%u\n",
+ sess_err->last_sess_failure_type);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(fail_type);
+
+static ssize_t iscsi_stat_instance_show_attr_fail_rem_name(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_sess_err_stats *sess_err = &tiqn->sess_err_stats;
+
+ return snprintf(page, PAGE_SIZE, "%s\n",
+ sess_err->last_sess_fail_rem_name[0] ?
+ sess_err->last_sess_fail_rem_name : NONE);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(fail_rem_name);
+
+static ssize_t iscsi_stat_instance_show_attr_disc_time(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%u\n", ISCSI_DISCONTINUITY_TIME);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(disc_time);
+
+static ssize_t iscsi_stat_instance_show_attr_description(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%s\n", ISCSI_INST_DESCR);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(description);
+
+static ssize_t iscsi_stat_instance_show_attr_vendor(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "RisingTide Systems iSCSI-Target\n");
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(vendor);
+
+static ssize_t iscsi_stat_instance_show_attr_version(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%s\n", ISCSI_VERSION);
+}
+ISCSI_STAT_INSTANCE_ATTR_RO(version);
+
+CONFIGFS_EATTR_OPS(iscsi_stat_instance, iscsi_wwn_stat_grps,
+ iscsi_instance_group);
+
+static struct configfs_attribute *iscsi_stat_instance_attrs[] = {
+ &iscsi_stat_instance_inst.attr,
+ &iscsi_stat_instance_min_ver.attr,
+ &iscsi_stat_instance_max_ver.attr,
+ &iscsi_stat_instance_portals.attr,
+ &iscsi_stat_instance_nodes.attr,
+ &iscsi_stat_instance_sessions.attr,
+ &iscsi_stat_instance_fail_sess.attr,
+ &iscsi_stat_instance_fail_type.attr,
+ &iscsi_stat_instance_fail_rem_name.attr,
+ &iscsi_stat_instance_disc_time.attr,
+ &iscsi_stat_instance_description.attr,
+ &iscsi_stat_instance_vendor.attr,
+ &iscsi_stat_instance_version.attr,
+ NULL,
+};
+
+static struct configfs_item_operations iscsi_stat_instance_item_ops = {
+ .show_attribute = iscsi_stat_instance_attr_show,
+ .store_attribute = iscsi_stat_instance_attr_store,
+};
+
+struct config_item_type iscsi_stat_instance_cit = {
+ .ct_item_ops = &iscsi_stat_instance_item_ops,
+ .ct_attrs = iscsi_stat_instance_attrs,
+ .ct_owner = THIS_MODULE,
+};
+
+/*
+ * Instance Session Failure Stats Table
+ */
+CONFIGFS_EATTR_STRUCT(iscsi_stat_sess_err, iscsi_wwn_stat_grps);
+#define ISCSI_STAT_SESS_ERR_ATTR(_name, _mode) \
+static struct iscsi_stat_sess_err_attribute \
+ iscsi_stat_sess_err_##_name = \
+ __CONFIGFS_EATTR(_name, _mode, \
+ iscsi_stat_sess_err_show_attr_##_name, \
+ iscsi_stat_sess_err_store_attr_##_name);
+
+#define ISCSI_STAT_SESS_ERR_ATTR_RO(_name) \
+static struct iscsi_stat_sess_err_attribute \
+ iscsi_stat_sess_err_##_name = \
+ __CONFIGFS_EATTR_RO(_name, \
+ iscsi_stat_sess_err_show_attr_##_name);
+
+static ssize_t iscsi_stat_sess_err_show_attr_inst(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", tiqn->tiqn_index);
+}
+ISCSI_STAT_SESS_ERR_ATTR_RO(inst);
+
+static ssize_t iscsi_stat_sess_err_show_attr_digest_errors(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_sess_err_stats *sess_err = &tiqn->sess_err_stats;
+
+ return snprintf(page, PAGE_SIZE, "%u\n", sess_err->digest_errors);
+}
+ISCSI_STAT_SESS_ERR_ATTR_RO(digest_errors);
+
+static ssize_t iscsi_stat_sess_err_show_attr_cxn_errors(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_sess_err_stats *sess_err = &tiqn->sess_err_stats;
+
+ return snprintf(page, PAGE_SIZE, "%u\n", sess_err->cxn_timeout_errors);
+}
+ISCSI_STAT_SESS_ERR_ATTR_RO(cxn_errors);
+
+static ssize_t iscsi_stat_sess_err_show_attr_format_errors(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_sess_err_stats *sess_err = &tiqn->sess_err_stats;
+
+ return snprintf(page, PAGE_SIZE, "%u\n", sess_err->pdu_format_errors);
+}
+ISCSI_STAT_SESS_ERR_ATTR_RO(format_errors);
+
+CONFIGFS_EATTR_OPS(iscsi_stat_sess_err, iscsi_wwn_stat_grps,
+ iscsi_sess_err_group);
+
+static struct configfs_attribute *iscsi_stat_sess_err_attrs[] = {
+ &iscsi_stat_sess_err_inst.attr,
+ &iscsi_stat_sess_err_digest_errors.attr,
+ &iscsi_stat_sess_err_cxn_errors.attr,
+ &iscsi_stat_sess_err_format_errors.attr,
+ NULL,
+};
+
+static struct configfs_item_operations iscsi_stat_sess_err_item_ops = {
+ .show_attribute = iscsi_stat_sess_err_attr_show,
+ .store_attribute = iscsi_stat_sess_err_attr_store,
+};
+
+struct config_item_type iscsi_stat_sess_err_cit = {
+ .ct_item_ops = &iscsi_stat_sess_err_item_ops,
+ .ct_attrs = iscsi_stat_sess_err_attrs,
+ .ct_owner = THIS_MODULE,
+};
+
+/*
+ * Target Attributes Table
+ */
+CONFIGFS_EATTR_STRUCT(iscsi_stat_tgt_attr, iscsi_wwn_stat_grps);
+#define ISCSI_STAT_TGT_ATTR(_name, _mode) \
+static struct iscsi_stat_tgt_attr_attribute \
+ iscsi_stat_tgt_attr_##_name = \
+ __CONFIGFS_EATTR(_name, _mode, \
+ iscsi_stat_tgt-attr_show_attr_##_name, \
+ iscsi_stat_tgt_attr_store_attr_##_name);
+
+#define ISCSI_STAT_TGT_ATTR_RO(_name) \
+static struct iscsi_stat_tgt_attr_attribute \
+ iscsi_stat_tgt_attr_##_name = \
+ __CONFIGFS_EATTR_RO(_name, \
+ iscsi_stat_tgt_attr_show_attr_##_name);
+
+static ssize_t iscsi_stat_tgt_attr_show_attr_inst(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", tiqn->tiqn_index);
+}
+ISCSI_STAT_TGT_ATTR_RO(inst);
+
+static ssize_t iscsi_stat_tgt_attr_show_attr_indx(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%u\n", ISCSI_NODE_INDEX);
+}
+ISCSI_STAT_TGT_ATTR_RO(indx);
+
+static ssize_t iscsi_stat_tgt_attr_show_attr_login_fails(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ u32 fail_count;
+
+ spin_lock(&lstat->lock);
+ fail_count = (lstat->redirects + lstat->authorize_fails +
+ lstat->authenticate_fails + lstat->negotiate_fails +
+ lstat->other_fails);
+ spin_unlock(&lstat->lock);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", fail_count);
+}
+ISCSI_STAT_TGT_ATTR_RO(login_fails);
+
+static ssize_t iscsi_stat_tgt_attr_show_attr_last_fail_time(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ u32 last_fail_time;
+
+ spin_lock(&lstat->lock);
+ last_fail_time = lstat->last_fail_time ?
+ (u32)(((u32)lstat->last_fail_time -
+ INITIAL_JIFFIES) * 100 / HZ) : 0;
+ spin_unlock(&lstat->lock);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", last_fail_time);
+}
+ISCSI_STAT_TGT_ATTR_RO(last_fail_time);
+
+static ssize_t iscsi_stat_tgt_attr_show_attr_last_fail_type(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ u32 last_fail_type;
+
+ spin_lock(&lstat->lock);
+ last_fail_type = lstat->last_fail_type;
+ spin_unlock(&lstat->lock);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", last_fail_type);
+}
+ISCSI_STAT_TGT_ATTR_RO(last_fail_type);
+
+static ssize_t iscsi_stat_tgt_attr_show_attr_fail_intr_name(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ unsigned char buf[224];
+
+ spin_lock(&lstat->lock);
+ snprintf(buf, 224, "%s", lstat->last_intr_fail_name[0] ?
+ lstat->last_intr_fail_name : NONE);
+ spin_unlock(&lstat->lock);
+
+ return snprintf(page, PAGE_SIZE, "%s\n", buf);
+}
+ISCSI_STAT_TGT_ATTR_RO(fail_intr_name);
+
+static ssize_t iscsi_stat_tgt_attr_show_attr_fail_intr_addr_type(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ unsigned char buf[8];
+
+ spin_lock(&lstat->lock);
+ snprintf(buf, 8, "%s", (lstat->last_intr_fail_ip_addr != NULL) ?
+ "ipv6" : "ipv4");
+ spin_unlock(&lstat->lock);
+
+ return snprintf(page, PAGE_SIZE, "%s\n", buf);
+}
+ISCSI_STAT_TGT_ATTR_RO(fail_intr_addr_type);
+
+static ssize_t iscsi_stat_tgt_attr_show_attr_fail_intr_addr(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ unsigned char buf[32];
+
+ spin_lock(&lstat->lock);
+ if (lstat->last_intr_fail_ip_family == AF_INET6)
+ snprintf(buf, 32, "[%s]", lstat->last_intr_fail_ip_addr);
+ else
+ snprintf(buf, 32, "%s", lstat->last_intr_fail_ip_addr);
+ spin_unlock(&lstat->lock);
+
+ return snprintf(page, PAGE_SIZE, "%s\n", buf);
+}
+ISCSI_STAT_TGT_ATTR_RO(fail_intr_addr);
+
+CONFIGFS_EATTR_OPS(iscsi_stat_tgt_attr, iscsi_wwn_stat_grps,
+ iscsi_tgt_attr_group);
+
+static struct configfs_attribute *iscsi_stat_tgt_attr_attrs[] = {
+ &iscsi_stat_tgt_attr_inst.attr,
+ &iscsi_stat_tgt_attr_indx.attr,
+ &iscsi_stat_tgt_attr_login_fails.attr,
+ &iscsi_stat_tgt_attr_last_fail_time.attr,
+ &iscsi_stat_tgt_attr_last_fail_type.attr,
+ &iscsi_stat_tgt_attr_fail_intr_name.attr,
+ &iscsi_stat_tgt_attr_fail_intr_addr_type.attr,
+ &iscsi_stat_tgt_attr_fail_intr_addr.attr,
+ NULL,
+};
+
+static struct configfs_item_operations iscsi_stat_tgt_attr_item_ops = {
+ .show_attribute = iscsi_stat_tgt_attr_attr_show,
+ .store_attribute = iscsi_stat_tgt_attr_attr_store,
+};
+
+struct config_item_type iscsi_stat_tgt_attr_cit = {
+ .ct_item_ops = &iscsi_stat_tgt_attr_item_ops,
+ .ct_attrs = iscsi_stat_tgt_attr_attrs,
+ .ct_owner = THIS_MODULE,
+};
+
+/*
+ * Target Login Stats Table
+ */
+CONFIGFS_EATTR_STRUCT(iscsi_stat_login, iscsi_wwn_stat_grps);
+#define ISCSI_STAT_LOGIN(_name, _mode) \
+static struct iscsi_stat_login_attribute \
+ iscsi_stat_login_##_name = \
+ __CONFIGFS_EATTR(_name, _mode, \
+ iscsi_stat_login_show_attr_##_name, \
+ iscsi_stat_login_store_attr_##_name);
+
+#define ISCSI_STAT_LOGIN_RO(_name) \
+static struct iscsi_stat_login_attribute \
+ iscsi_stat_login_##_name = \
+ __CONFIGFS_EATTR_RO(_name, \
+ iscsi_stat_login_show_attr_##_name);
+
+static ssize_t iscsi_stat_login_show_attr_inst(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", tiqn->tiqn_index);
+}
+ISCSI_STAT_LOGIN_RO(inst);
+
+static ssize_t iscsi_stat_login_show_attr_indx(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%u\n", ISCSI_NODE_INDEX);
+}
+ISCSI_STAT_LOGIN_RO(indx);
+
+static ssize_t iscsi_stat_login_show_attr_accepts(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ ssize_t ret;
+
+ spin_lock(&lstat->lock);
+ ret = snprintf(page, PAGE_SIZE, "%u\n", lstat->accepts);
+ spin_unlock(&lstat->lock);
+
+ return ret;
+}
+ISCSI_STAT_LOGIN_RO(accepts);
+
+static ssize_t iscsi_stat_login_show_attr_other_fails(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ ssize_t ret;
+
+ spin_lock(&lstat->lock);
+ ret = snprintf(page, PAGE_SIZE, "%u\n", lstat->other_fails);
+ spin_unlock(&lstat->lock);
+
+ return ret;
+}
+ISCSI_STAT_LOGIN_RO(other_fails);
+
+static ssize_t iscsi_stat_login_show_attr_redirects(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ ssize_t ret;
+
+ spin_lock(&lstat->lock);
+ ret = snprintf(page, PAGE_SIZE, "%u\n", lstat->redirects);
+ spin_unlock(&lstat->lock);
+
+ return ret;
+}
+ISCSI_STAT_LOGIN_RO(redirects);
+
+static ssize_t iscsi_stat_login_show_attr_authorize_fails(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ ssize_t ret;
+
+ spin_lock(&lstat->lock);
+ ret = snprintf(page, PAGE_SIZE, "%u\n", lstat->authorize_fails);
+ spin_unlock(&lstat->lock);
+
+ return ret;
+}
+ISCSI_STAT_LOGIN_RO(authorize_fails);
+
+static ssize_t iscsi_stat_login_show_attr_authenticate_fails(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ ssize_t ret;
+
+ spin_lock(&lstat->lock);
+ ret = snprintf(page, PAGE_SIZE, "%u\n", lstat->authenticate_fails);
+ spin_unlock(&lstat->lock);
+
+ return ret;
+}
+ISCSI_STAT_LOGIN_RO(authenticate_fails);
+
+static ssize_t iscsi_stat_login_show_attr_negotiate_fails(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_login_stats *lstat = &tiqn->login_stats;
+ ssize_t ret;
+
+ spin_lock(&lstat->lock);
+ ret = snprintf(page, PAGE_SIZE, "%u\n", lstat->negotiate_fails);
+ spin_unlock(&lstat->lock);
+
+ return ret;
+}
+ISCSI_STAT_LOGIN_RO(negotiate_fails);
+
+CONFIGFS_EATTR_OPS(iscsi_stat_login, iscsi_wwn_stat_grps,
+ iscsi_login_stats_group);
+
+static struct configfs_attribute *iscsi_stat_login_stats_attrs[] = {
+ &iscsi_stat_login_inst.attr,
+ &iscsi_stat_login_indx.attr,
+ &iscsi_stat_login_accepts.attr,
+ &iscsi_stat_login_other_fails.attr,
+ &iscsi_stat_login_redirects.attr,
+ &iscsi_stat_login_authorize_fails.attr,
+ &iscsi_stat_login_authenticate_fails.attr,
+ &iscsi_stat_login_negotiate_fails.attr,
+ NULL,
+};
+
+static struct configfs_item_operations iscsi_stat_login_stats_item_ops = {
+ .show_attribute = iscsi_stat_login_attr_show,
+ .store_attribute = iscsi_stat_login_attr_store,
+};
+
+struct config_item_type iscsi_stat_login_cit = {
+ .ct_item_ops = &iscsi_stat_login_stats_item_ops,
+ .ct_attrs = iscsi_stat_login_stats_attrs,
+ .ct_owner = THIS_MODULE,
+};
+
+/*
+ * Target Logout Stats Table
+ */
+
+CONFIGFS_EATTR_STRUCT(iscsi_stat_logout, iscsi_wwn_stat_grps);
+#define ISCSI_STAT_LOGOUT(_name, _mode) \
+static struct iscsi_stat_logout_attribute \
+ iscsi_stat_logout_##_name = \
+ __CONFIGFS_EATTR(_name, _mode, \
+ iscsi_stat_logout_show_attr_##_name, \
+ iscsi_stat_logout_store_attr_##_name);
+
+#define ISCSI_STAT_LOGOUT_RO(_name) \
+static struct iscsi_stat_logout_attribute \
+ iscsi_stat_logout_##_name = \
+ __CONFIGFS_EATTR_RO(_name, \
+ iscsi_stat_logout_show_attr_##_name);
+
+static ssize_t iscsi_stat_logout_show_attr_inst(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", tiqn->tiqn_index);
+}
+ISCSI_STAT_LOGOUT_RO(inst);
+
+static ssize_t iscsi_stat_logout_show_attr_indx(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%u\n", ISCSI_NODE_INDEX);
+}
+ISCSI_STAT_LOGOUT_RO(indx);
+
+static ssize_t iscsi_stat_logout_show_attr_normal_logouts(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_logout_stats *lstats = &tiqn->logout_stats;
+
+ return snprintf(page, PAGE_SIZE, "%u\n", lstats->normal_logouts);
+}
+ISCSI_STAT_LOGOUT_RO(normal_logouts);
+
+static ssize_t iscsi_stat_logout_show_attr_abnormal_logouts(
+ struct iscsi_wwn_stat_grps *igrps, char *page)
+{
+ struct iscsi_tiqn *tiqn = container_of(igrps,
+ struct iscsi_tiqn, tiqn_stat_grps);
+ struct iscsi_logout_stats *lstats = &tiqn->logout_stats;
+
+ return snprintf(page, PAGE_SIZE, "%u\n", lstats->abnormal_logouts);
+}
+ISCSI_STAT_LOGOUT_RO(abnormal_logouts);
+
+CONFIGFS_EATTR_OPS(iscsi_stat_logout, iscsi_wwn_stat_grps,
+ iscsi_logout_stats_group);
+
+static struct configfs_attribute *iscsi_stat_logout_stats_attrs[] = {
+ &iscsi_stat_logout_inst.attr,
+ &iscsi_stat_logout_indx.attr,
+ &iscsi_stat_logout_normal_logouts.attr,
+ &iscsi_stat_logout_abnormal_logouts.attr,
+ NULL,
+};
+
+static struct configfs_item_operations iscsi_stat_logout_stats_item_ops = {
+ .show_attribute = iscsi_stat_logout_attr_show,
+ .store_attribute = iscsi_stat_logout_attr_store,
+};
+
+struct config_item_type iscsi_stat_logout_cit = {
+ .ct_item_ops = &iscsi_stat_logout_stats_item_ops,
+ .ct_attrs = iscsi_stat_logout_stats_attrs,
+ .ct_owner = THIS_MODULE,
+};
+
+/*
+ * Session Stats Table
+ */
+
+CONFIGFS_EATTR_STRUCT(iscsi_stat_sess, iscsi_node_stat_grps);
+#define ISCSI_STAT_SESS(_name, _mode) \
+static struct iscsi_stat_sess_attribute \
+ iscsi_stat_sess_##_name = \
+ __CONFIGFS_EATTR(_name, _mode, \
+ iscsi_stat_sess_show_attr_##_name, \
+ iscsi_stat_sess_store_attr_##_name);
+
+#define ISCSI_STAT_SESS_RO(_name) \
+static struct iscsi_stat_sess_attribute \
+ iscsi_stat_sess_##_name = \
+ __CONFIGFS_EATTR_RO(_name, \
+ iscsi_stat_sess_show_attr_##_name);
+
+static ssize_t iscsi_stat_sess_show_attr_inst(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_wwn *wwn = acl->se_node_acl.se_tpg->se_tpg_wwn;
+ struct iscsi_tiqn *tiqn = container_of(wwn,
+ struct iscsi_tiqn, tiqn_wwn);
+
+ return snprintf(page, PAGE_SIZE, "%u\n", tiqn->tiqn_index);
+}
+ISCSI_STAT_SESS_RO(inst);
+
+static ssize_t iscsi_stat_sess_show_attr_node(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_node_acl *se_nacl = &acl->se_node_acl;
+ struct iscsi_session *sess;
+ struct se_session *se_sess;
+ ssize_t ret = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (sess)
+ ret = snprintf(page, PAGE_SIZE, "%u\n",
+ sess->sess_ops->SessionType ? 0 : ISCSI_NODE_INDEX);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return ret;
+}
+ISCSI_STAT_SESS_RO(node);
+
+static ssize_t iscsi_stat_sess_show_attr_indx(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_node_acl *se_nacl = &acl->se_node_acl;
+ struct iscsi_session *sess;
+ struct se_session *se_sess;
+ ssize_t ret = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (sess)
+ ret = snprintf(page, PAGE_SIZE, "%u\n",
+ sess->session_index);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return ret;
+}
+ISCSI_STAT_SESS_RO(indx);
+
+static ssize_t iscsi_stat_sess_show_attr_cmd_pdus(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_node_acl *se_nacl = &acl->se_node_acl;
+ struct iscsi_session *sess;
+ struct se_session *se_sess;
+ ssize_t ret = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (sess)
+ ret = snprintf(page, PAGE_SIZE, "%u\n", sess->cmd_pdus);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return ret;
+}
+ISCSI_STAT_SESS_RO(cmd_pdus);
+
+static ssize_t iscsi_stat_sess_show_attr_rsp_pdus(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_node_acl *se_nacl = &acl->se_node_acl;
+ struct iscsi_session *sess;
+ struct se_session *se_sess;
+ ssize_t ret = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (sess)
+ ret = snprintf(page, PAGE_SIZE, "%u\n", sess->rsp_pdus);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return ret;
+}
+ISCSI_STAT_SESS_RO(rsp_pdus);
+
+static ssize_t iscsi_stat_sess_show_attr_txdata_octs(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_node_acl *se_nacl = &acl->se_node_acl;
+ struct iscsi_session *sess;
+ struct se_session *se_sess;
+ ssize_t ret = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (sess)
+ ret = snprintf(page, PAGE_SIZE, "%llu\n",
+ (unsigned long long)sess->tx_data_octets);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return ret;
+}
+ISCSI_STAT_SESS_RO(txdata_octs);
+
+static ssize_t iscsi_stat_sess_show_attr_rxdata_octs(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_node_acl *se_nacl = &acl->se_node_acl;
+ struct iscsi_session *sess;
+ struct se_session *se_sess;
+ ssize_t ret = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (sess)
+ ret = snprintf(page, PAGE_SIZE, "%llu\n",
+ (unsigned long long)sess->rx_data_octets);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return ret;
+}
+ISCSI_STAT_SESS_RO(rxdata_octs);
+
+static ssize_t iscsi_stat_sess_show_attr_conn_digest_errors(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_node_acl *se_nacl = &acl->se_node_acl;
+ struct iscsi_session *sess;
+ struct se_session *se_sess;
+ ssize_t ret = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (sess)
+ ret = snprintf(page, PAGE_SIZE, "%u\n",
+ sess->conn_digest_errors);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return ret;
+}
+ISCSI_STAT_SESS_RO(conn_digest_errors);
+
+static ssize_t iscsi_stat_sess_show_attr_conn_timeout_errors(
+ struct iscsi_node_stat_grps *igrps, char *page)
+{
+ struct iscsi_node_acl *acl = container_of(igrps,
+ struct iscsi_node_acl, node_stat_grps);
+ struct se_node_acl *se_nacl = &acl->se_node_acl;
+ struct iscsi_session *sess;
+ struct se_session *se_sess;
+ ssize_t ret = 0;
+
+ spin_lock_bh(&se_nacl->nacl_sess_lock);
+ se_sess = se_nacl->nacl_sess;
+ if (se_sess) {
+ sess = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (sess)
+ ret = snprintf(page, PAGE_SIZE, "%u\n",
+ sess->conn_timeout_errors);
+ }
+ spin_unlock_bh(&se_nacl->nacl_sess_lock);
+
+ return ret;
+}
+ISCSI_STAT_SESS_RO(conn_timeout_errors);
+
+CONFIGFS_EATTR_OPS(iscsi_stat_sess, iscsi_node_stat_grps,
+ iscsi_sess_stats_group);
+
+static struct configfs_attribute *iscsi_stat_sess_stats_attrs[] = {
+ &iscsi_stat_sess_inst.attr,
+ &iscsi_stat_sess_node.attr,
+ &iscsi_stat_sess_indx.attr,
+ &iscsi_stat_sess_cmd_pdus.attr,
+ &iscsi_stat_sess_rsp_pdus.attr,
+ &iscsi_stat_sess_txdata_octs.attr,
+ &iscsi_stat_sess_rxdata_octs.attr,
+ &iscsi_stat_sess_conn_digest_errors.attr,
+ &iscsi_stat_sess_conn_timeout_errors.attr,
+ NULL,
+};
+
+static struct configfs_item_operations iscsi_stat_sess_stats_item_ops = {
+ .show_attribute = iscsi_stat_sess_attr_show,
+ .store_attribute = iscsi_stat_sess_attr_store,
+};
+
+struct config_item_type iscsi_stat_sess_cit = {
+ .ct_item_ops = &iscsi_stat_sess_stats_item_ops,
+ .ct_attrs = iscsi_stat_sess_stats_attrs,
+ .ct_owner = THIS_MODULE,
+};
diff --git a/drivers/target/iscsi/iscsi_target_stat.h b/drivers/target/iscsi/iscsi_target_stat.h
new file mode 100644
index 0000000..3ff76b4
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_stat.h
@@ -0,0 +1,64 @@
+#ifndef ISCSI_TARGET_STAT_H
+#define ISCSI_TARGET_STAT_H
+
+/*
+ * For struct iscsi_tiqn->tiqn_wwn default groups
+ */
+extern struct config_item_type iscsi_stat_instance_cit;
+extern struct config_item_type iscsi_stat_sess_err_cit;
+extern struct config_item_type iscsi_stat_tgt_attr_cit;
+extern struct config_item_type iscsi_stat_login_cit;
+extern struct config_item_type iscsi_stat_logout_cit;
+
+/*
+ * For struct iscsi_session->se_sess default groups
+ */
+extern struct config_item_type iscsi_stat_sess_cit;
+
+/* iSCSI session error types */
+#define ISCSI_SESS_ERR_UNKNOWN 0
+#define ISCSI_SESS_ERR_DIGEST 1
+#define ISCSI_SESS_ERR_CXN_TIMEOUT 2
+#define ISCSI_SESS_ERR_PDU_FORMAT 3
+
+/* iSCSI session error stats */
+struct iscsi_sess_err_stats {
+ spinlock_t lock;
+ u32 digest_errors;
+ u32 cxn_timeout_errors;
+ u32 pdu_format_errors;
+ u32 last_sess_failure_type;
+ char last_sess_fail_rem_name[224];
+} ____cacheline_aligned;
+
+/* iSCSI login failure types (sub oids) */
+#define ISCSI_LOGIN_FAIL_OTHER 2
+#define ISCSI_LOGIN_FAIL_REDIRECT 3
+#define ISCSI_LOGIN_FAIL_AUTHORIZE 4
+#define ISCSI_LOGIN_FAIL_AUTHENTICATE 5
+#define ISCSI_LOGIN_FAIL_NEGOTIATE 6
+
+/* iSCSI login stats */
+struct iscsi_login_stats {
+ spinlock_t lock;
+ u32 accepts;
+ u32 other_fails;
+ u32 redirects;
+ u32 authorize_fails;
+ u32 authenticate_fails;
+ u32 negotiate_fails; /* used for notifications */
+ u64 last_fail_time; /* time stamp (jiffies) */
+ u32 last_fail_type;
+ int last_intr_fail_ip_family;
+ unsigned char last_intr_fail_ip_addr[IPV6_ADDRESS_SPACE];
+ char last_intr_fail_name[224];
+} ____cacheline_aligned;
+
+/* iSCSI logout stats */
+struct iscsi_logout_stats {
+ spinlock_t lock;
+ u32 normal_logouts;
+ u32 abnormal_logouts;
+} ____cacheline_aligned;
+
+#endif /*** ISCSI_TARGET_STAT_H ***/
--
1.7.4.1
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 07/14] iscsi-target: Add TPG and Device logic
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (6 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 06/14] iscsi-target: Add configfs fabric dependent statistics Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 08/14] iscsi-target: Add iSCSI Login Negotiation and Parameter logic Nicholas A. Bellinger
` (7 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch adds TPG and device logiced using for mapping iscsi-target
abstractions on top of TCM v4 struct se_portal_group and struct se_device
abstractions.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_device.c | 108 ++++
drivers/target/iscsi/iscsi_target_device.h | 9 +
drivers/target/iscsi/iscsi_target_tpg.c | 767 ++++++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_tpg.h | 43 ++
4 files changed, 927 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_device.c
create mode 100644 drivers/target/iscsi/iscsi_target_device.h
create mode 100644 drivers/target/iscsi/iscsi_target_tpg.c
create mode 100644 drivers/target/iscsi/iscsi_target_tpg.h
diff --git a/drivers/target/iscsi/iscsi_target_device.c b/drivers/target/iscsi/iscsi_target_device.c
new file mode 100644
index 0000000..b09ee93
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_device.c
@@ -0,0 +1,108 @@
+/*******************************************************************************
+ * This file contains the iSCSI Virtual Device and Disk Transport
+ * agnostic related functions.
+ *
+ © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <scsi/scsi_device.h>
+#include <target/target_core_base.h>
+#include <target/target_core_device.h>
+#include <target/target_core_transport.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_device.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+
+int iscsit_get_lun_for_tmr(
+ struct iscsi_cmd *cmd,
+ u64 lun)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_portal_group *tpg = ISCSI_TPG_C(conn);
+ u32 unpacked_lun;
+
+ unpacked_lun = scsilun_to_int((struct scsi_lun *)&lun);
+ if (unpacked_lun > (TRANSPORT_MAX_LUNS_PER_TPG-1)) {
+ printk(KERN_ERR "iSCSI LUN: %u exceeds TRANSPORT_MAX_LUNS_PER_TPG"
+ "-1: %u for Target Portal Group: %hu\n", unpacked_lun,
+ TRANSPORT_MAX_LUNS_PER_TPG-1, tpg->tpgt);
+ return -1;
+ }
+
+ return transport_get_lun_for_tmr(SE_CMD(cmd), unpacked_lun);
+}
+
+int iscsit_get_lun_for_cmd(
+ struct iscsi_cmd *cmd,
+ unsigned char *cdb,
+ u64 lun)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_portal_group *tpg = ISCSI_TPG_C(conn);
+ u32 unpacked_lun;
+
+ unpacked_lun = scsilun_to_int((struct scsi_lun *)&lun);
+ if (unpacked_lun > (TRANSPORT_MAX_LUNS_PER_TPG-1)) {
+ printk(KERN_ERR "iSCSI LUN: %u exceeds TRANSPORT_MAX_LUNS_PER_TPG"
+ "-1: %u for Target Portal Group: %hu\n", unpacked_lun,
+ TRANSPORT_MAX_LUNS_PER_TPG-1, tpg->tpgt);
+ return -1;
+ }
+
+ return transport_get_lun_for_cmd(SE_CMD(cmd), NULL, unpacked_lun);
+}
+
+void iscsit_determine_maxcmdsn(struct iscsi_session *sess)
+{
+ struct se_node_acl *se_nacl;
+
+ /*
+ * This is a discovery session, the single queue slot was already
+ * assigned in iscsi_login_zero_tsih(). Since only Logout and
+ * Text Opcodes are allowed during discovery we do not have to worry
+ * about the HBA's queue depth here.
+ */
+ if (sess->sess_ops->SessionType)
+ return;
+
+ se_nacl = sess->se_sess->se_node_acl;
+
+ /*
+ * This is a normal session, set the Session's CmdSN window to the
+ * struct se_node_acl->queue_depth. The value in struct se_node_acl->queue_depth
+ * has already been validated as a legal value in
+ * core_set_queue_depth_for_node().
+ */
+ sess->cmdsn_window = se_nacl->queue_depth;
+ sess->max_cmd_sn = (sess->max_cmd_sn + se_nacl->queue_depth) - 1;
+}
+
+void iscsit_increment_maxcmdsn(struct iscsi_cmd *cmd, struct iscsi_session *sess)
+{
+ if (cmd->immediate_cmd || cmd->maxcmdsn_inc)
+ return;
+
+ cmd->maxcmdsn_inc = 1;
+
+ spin_lock(&sess->cmdsn_lock);
+ sess->max_cmd_sn += 1;
+ TRACE(TRACE_ISCSI, "Updated MaxCmdSN to 0x%08x\n", sess->max_cmd_sn);
+ spin_unlock(&sess->cmdsn_lock);
+}
diff --git a/drivers/target/iscsi/iscsi_target_device.h b/drivers/target/iscsi/iscsi_target_device.h
new file mode 100644
index 0000000..bef1cad
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_device.h
@@ -0,0 +1,9 @@
+#ifndef ISCSI_TARGET_DEVICE_H
+#define ISCSI_TARGET_DEVICE_H
+
+extern int iscsit_get_lun_for_tmr(struct iscsi_cmd *, u64);
+extern int iscsit_get_lun_for_cmd(struct iscsi_cmd *, unsigned char *, u64);
+extern void iscsit_determine_maxcmdsn(struct iscsi_session *);
+extern void iscsit_increment_maxcmdsn(struct iscsi_cmd *, struct iscsi_session *);
+
+#endif /* ISCSI_TARGET_DEVICE_H */
diff --git a/drivers/target/iscsi/iscsi_target_tpg.c b/drivers/target/iscsi/iscsi_target_tpg.c
new file mode 100644
index 0000000..dcc1d99
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_tpg.c
@@ -0,0 +1,767 @@
+/*******************************************************************************
+ * This file contains iSCSI Target Portal Group related functions.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+#include <target/target_core_fabric_ops.h>
+#include <target/target_core_configfs.h>
+#include <target/target_core_tpg.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_login.h"
+#include "iscsi_target_nodeattrib.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target.h"
+#include "iscsi_target_parameters.h"
+
+struct iscsi_portal_group *iscsit_alloc_portal_group(struct iscsi_tiqn *tiqn, u16 tpgt)
+{
+ struct iscsi_portal_group *tpg;
+
+ tpg = kzalloc(sizeof(struct iscsi_portal_group), GFP_KERNEL);
+ if (!tpg) {
+ printk(KERN_ERR "Unable to allocate struct iscsi_portal_group\n");
+ return NULL;
+ }
+
+ tpg->tpgt = tpgt;
+ tpg->tpg_state = TPG_STATE_FREE;
+ tpg->tpg_tiqn = tiqn;
+ INIT_LIST_HEAD(&tpg->tpg_gnp_list);
+ INIT_LIST_HEAD(&tpg->tpg_list);
+ mutex_init(&tpg->tpg_access_lock);
+ mutex_init(&tpg->np_login_lock);
+ spin_lock_init(&tpg->tpg_state_lock);
+ spin_lock_init(&tpg->tpg_np_lock);
+
+ return tpg;
+}
+
+static void iscsit_set_default_tpg_attribs(struct iscsi_portal_group *);
+
+int iscsit_load_discovery_tpg(void)
+{
+ struct iscsi_param *param;
+ struct iscsi_portal_group *tpg;
+ int ret;
+
+ tpg = iscsit_alloc_portal_group(NULL, 1);
+ if (!tpg) {
+ printk(KERN_ERR "Unable to allocate struct iscsi_portal_group\n");
+ return -1;
+ }
+
+ ret = core_tpg_register(
+ &lio_target_fabric_configfs->tf_ops,
+ NULL, &tpg->tpg_se_tpg, (void *)tpg,
+ TRANSPORT_TPG_TYPE_DISCOVERY);
+ if (ret < 0) {
+ kfree(tpg);
+ return -1;
+ }
+
+ tpg->sid = 1; /* First Assigned LIO Session ID */
+ iscsit_set_default_tpg_attribs(tpg);
+
+ if (iscsi_create_default_params(&tpg->param_list) < 0)
+ goto out;
+ /*
+ * By default we disable authentication for discovery sessions,
+ * this can be changed with:
+ *
+ * /sys/kernel/config/target/iscsi/discovery_auth/enforce_discovery_auth
+ */
+ param = iscsi_find_param_from_key(AUTHMETHOD, tpg->param_list);
+ if (!param)
+ goto out;
+
+ if (iscsi_update_param_value(param, "CHAP,None") < 0)
+ goto out;
+
+ tpg->tpg_attrib.authentication = 0;
+
+ spin_lock(&tpg->tpg_state_lock);
+ tpg->tpg_state = TPG_STATE_ACTIVE;
+ spin_unlock(&tpg->tpg_state_lock);
+
+ iscsi_global->discovery_tpg = tpg;
+ printk(KERN_INFO "CORE[0] - Allocated Discovery TPG\n");
+
+ return 0;
+out:
+ if (tpg->sid == 1)
+ core_tpg_deregister(&tpg->tpg_se_tpg);
+ kfree(tpg);
+ return -1;
+}
+
+void iscsit_release_discovery_tpg(void)
+{
+ struct iscsi_portal_group *tpg = iscsi_global->discovery_tpg;
+
+ if (!tpg)
+ return;
+
+ core_tpg_deregister(&tpg->tpg_se_tpg);
+
+ kfree(tpg);
+ iscsi_global->discovery_tpg = NULL;
+}
+
+struct iscsi_portal_group *iscsit_get_tpg_from_np(
+ struct iscsi_tiqn *tiqn,
+ struct iscsi_np *np)
+{
+ struct iscsi_portal_group *tpg = NULL;
+ struct iscsi_tpg_np *tpg_np;
+
+ spin_lock(&tiqn->tiqn_tpg_lock);
+ list_for_each_entry(tpg, &tiqn->tiqn_tpg_list, tpg_list) {
+
+ spin_lock(&tpg->tpg_state_lock);
+ if (tpg->tpg_state == TPG_STATE_FREE) {
+ spin_unlock(&tpg->tpg_state_lock);
+ continue;
+ }
+ spin_unlock(&tpg->tpg_state_lock);
+
+ spin_lock(&tpg->tpg_np_lock);
+ list_for_each_entry(tpg_np, &tpg->tpg_gnp_list, tpg_np_list) {
+ if (tpg_np->tpg_np == np) {
+ spin_unlock(&tpg->tpg_np_lock);
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+ return tpg;
+ }
+ }
+ spin_unlock(&tpg->tpg_np_lock);
+ }
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+
+ return NULL;
+}
+
+int iscsit_get_tpg(
+ struct iscsi_portal_group *tpg)
+{
+ int ret;
+
+ ret = mutex_lock_interruptible(&tpg->tpg_access_lock);
+ return ((ret != 0) || signal_pending(current)) ? -1 : 0;
+}
+
+void iscsit_put_tpg(struct iscsi_portal_group *tpg)
+{
+ mutex_unlock(&tpg->tpg_access_lock);
+}
+
+static void iscsit_clear_tpg_np_login_thread(
+ struct iscsi_tpg_np *tpg_np,
+ struct iscsi_portal_group *tpg)
+{
+ if (!tpg_np->tpg_np) {
+ printk(KERN_ERR "struct iscsi_tpg_np->tpg_np is NULL!\n");
+ return;
+ }
+
+ iscsit_reset_np_thread(tpg_np->tpg_np, tpg_np, tpg);
+ return;
+}
+
+void iscsit_clear_tpg_np_login_threads(
+ struct iscsi_portal_group *tpg)
+{
+ struct iscsi_tpg_np *tpg_np;
+
+ spin_lock(&tpg->tpg_np_lock);
+ list_for_each_entry(tpg_np, &tpg->tpg_gnp_list, tpg_np_list) {
+ if (!tpg_np->tpg_np) {
+ printk(KERN_ERR "struct iscsi_tpg_np->tpg_np is NULL!\n");
+ continue;
+ }
+ spin_unlock(&tpg->tpg_np_lock);
+ iscsit_clear_tpg_np_login_thread(tpg_np, tpg);
+ spin_lock(&tpg->tpg_np_lock);
+ }
+ spin_unlock(&tpg->tpg_np_lock);
+}
+
+void iscsit_tpg_dump_params(struct iscsi_portal_group *tpg)
+{
+ iscsi_print_params(tpg->param_list);
+}
+
+static void iscsit_set_default_tpg_attribs(struct iscsi_portal_group *tpg)
+{
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ a->authentication = TA_AUTHENTICATION;
+ a->login_timeout = TA_LOGIN_TIMEOUT;
+ a->netif_timeout = TA_NETIF_TIMEOUT;
+ a->default_cmdsn_depth = TA_DEFAULT_CMDSN_DEPTH;
+ a->generate_node_acls = TA_GENERATE_NODE_ACLS;
+ a->cache_dynamic_acls = TA_CACHE_DYNAMIC_ACLS;
+ a->demo_mode_write_protect = TA_DEMO_MODE_WRITE_PROTECT;
+ a->prod_mode_write_protect = TA_PROD_MODE_WRITE_PROTECT;
+}
+
+int iscsit_tpg_add_portal_group(struct iscsi_tiqn *tiqn, struct iscsi_portal_group *tpg)
+{
+ if (tpg->tpg_state != TPG_STATE_FREE) {
+ printk(KERN_ERR "Unable to add iSCSI Target Portal Group: %d"
+ " while not in TPG_STATE_FREE state.\n", tpg->tpgt);
+ return -EEXIST;
+ }
+ iscsit_set_default_tpg_attribs(tpg);
+
+ if (iscsi_create_default_params(&tpg->param_list) < 0)
+ goto err_out;
+
+ ISCSI_TPG_ATTRIB(tpg)->tpg = tpg;
+
+ spin_lock(&tpg->tpg_state_lock);
+ tpg->tpg_state = TPG_STATE_INACTIVE;
+ spin_unlock(&tpg->tpg_state_lock);
+
+ spin_lock(&tiqn->tiqn_tpg_lock);
+ list_add_tail(&tpg->tpg_list, &tiqn->tiqn_tpg_list);
+ tiqn->tiqn_ntpgs++;
+ printk(KERN_INFO "CORE[%s]_TPG[%hu] - Added iSCSI Target Portal Group\n",
+ tiqn->tiqn, tpg->tpgt);
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+
+ return 0;
+err_out:
+ if (tpg->param_list) {
+ iscsi_release_param_list(tpg->param_list);
+ tpg->param_list = NULL;
+ }
+ kfree(tpg);
+ return -ENOMEM;
+}
+
+int iscsit_tpg_del_portal_group(
+ struct iscsi_tiqn *tiqn,
+ struct iscsi_portal_group *tpg,
+ int force)
+{
+ u8 old_state = tpg->tpg_state;
+
+ spin_lock(&tpg->tpg_state_lock);
+ tpg->tpg_state = TPG_STATE_INACTIVE;
+ spin_unlock(&tpg->tpg_state_lock);
+
+ if (iscsit_release_sessions_for_tpg(tpg, force) < 0) {
+ printk(KERN_ERR "Unable to delete iSCSI Target Portal Group:"
+ " %hu while active sessions exist, and force=0\n",
+ tpg->tpgt);
+ tpg->tpg_state = old_state;
+ return -EPERM;
+ }
+
+ core_tpg_clear_object_luns(&tpg->tpg_se_tpg);
+
+ if (tpg->param_list) {
+ iscsi_release_param_list(tpg->param_list);
+ tpg->param_list = NULL;
+ }
+
+ core_tpg_deregister(&tpg->tpg_se_tpg);
+
+ spin_lock(&tpg->tpg_state_lock);
+ tpg->tpg_state = TPG_STATE_FREE;
+ spin_unlock(&tpg->tpg_state_lock);
+
+ spin_lock(&tiqn->tiqn_tpg_lock);
+ tiqn->tiqn_ntpgs--;
+ list_del(&tpg->tpg_list);
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+
+ printk(KERN_INFO "CORE[%s]_TPG[%hu] - Deleted iSCSI Target Portal Group\n",
+ tiqn->tiqn, tpg->tpgt);
+
+ kfree(tpg);
+ return 0;
+}
+
+int iscsit_tpg_enable_portal_group(struct iscsi_portal_group *tpg)
+{
+ struct iscsi_param *param;
+ struct iscsi_tiqn *tiqn = tpg->tpg_tiqn;
+
+ spin_lock(&tpg->tpg_state_lock);
+ if (tpg->tpg_state == TPG_STATE_ACTIVE) {
+ printk(KERN_ERR "iSCSI target portal group: %hu is already"
+ " active, ignoring request.\n", tpg->tpgt);
+ spin_unlock(&tpg->tpg_state_lock);
+ return -EINVAL;
+ }
+ /*
+ * Make sure that AuthMethod does not contain None as an option
+ * unless explictly disabled. Set the default to CHAP if authentication
+ * is enforced (as per default), and remove the NONE option.
+ */
+ param = iscsi_find_param_from_key(AUTHMETHOD, tpg->param_list);
+ if (!param) {
+ spin_unlock(&tpg->tpg_state_lock);
+ return -ENOMEM;
+ }
+
+ if (ISCSI_TPG_ATTRIB(tpg)->authentication) {
+ if (!strcmp(param->value, NONE))
+ if (iscsi_update_param_value(param, CHAP) < 0) {
+ spin_unlock(&tpg->tpg_state_lock);
+ return -ENOMEM;
+ }
+ if (iscsit_ta_authentication(tpg, 1) < 0) {
+ spin_unlock(&tpg->tpg_state_lock);
+ return -ENOMEM;
+ }
+ }
+
+ tpg->tpg_state = TPG_STATE_ACTIVE;
+ spin_unlock(&tpg->tpg_state_lock);
+
+ spin_lock(&tiqn->tiqn_tpg_lock);
+ tiqn->tiqn_active_tpgs++;
+ printk(KERN_INFO "iSCSI_TPG[%hu] - Enabled iSCSI Target Portal Group\n",
+ tpg->tpgt);
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+
+ return 0;
+}
+
+int iscsit_tpg_disable_portal_group(struct iscsi_portal_group *tpg, int force)
+{
+ struct iscsi_tiqn *tiqn;
+ u8 old_state = tpg->tpg_state;
+
+ spin_lock(&tpg->tpg_state_lock);
+ if (tpg->tpg_state == TPG_STATE_INACTIVE) {
+ printk(KERN_ERR "iSCSI Target Portal Group: %hu is already"
+ " inactive, ignoring request.\n", tpg->tpgt);
+ spin_unlock(&tpg->tpg_state_lock);
+ return -EINVAL;
+ }
+ tpg->tpg_state = TPG_STATE_INACTIVE;
+ spin_unlock(&tpg->tpg_state_lock);
+
+ iscsit_clear_tpg_np_login_threads(tpg);
+
+ if (iscsit_release_sessions_for_tpg(tpg, force) < 0) {
+ spin_lock(&tpg->tpg_state_lock);
+ tpg->tpg_state = old_state;
+ spin_unlock(&tpg->tpg_state_lock);
+ printk(KERN_ERR "Unable to disable iSCSI Target Portal Group:"
+ " %hu while active sessions exist, and force=0\n",
+ tpg->tpgt);
+ return -EPERM;
+ }
+
+ tiqn = tpg->tpg_tiqn;
+ if (!tiqn || (tpg == iscsi_global->discovery_tpg))
+ return 0;
+
+ spin_lock(&tiqn->tiqn_tpg_lock);
+ tiqn->tiqn_active_tpgs--;
+ printk(KERN_INFO "iSCSI_TPG[%hu] - Disabled iSCSI Target Portal Group\n",
+ tpg->tpgt);
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+
+ return 0;
+}
+
+struct iscsi_node_attrib *iscsit_tpg_get_node_attrib(
+ struct iscsi_session *sess)
+{
+ struct se_session *se_sess = sess->se_sess;
+ struct se_node_acl *se_nacl = se_sess->se_node_acl;
+ struct iscsi_node_acl *acl = container_of(se_nacl, struct iscsi_node_acl,
+ se_node_acl);
+
+ return &acl->node_attrib;
+}
+
+struct iscsi_tpg_np *iscsit_tpg_locate_child_np(
+ struct iscsi_tpg_np *tpg_np,
+ int network_transport)
+{
+ struct iscsi_tpg_np *tpg_np_child, *tpg_np_child_tmp;
+
+ spin_lock(&tpg_np->tpg_np_parent_lock);
+ list_for_each_entry_safe(tpg_np_child, tpg_np_child_tmp,
+ &tpg_np->tpg_np_parent_list, tpg_np_child_list) {
+ if (tpg_np_child->tpg_np->np_network_transport ==
+ network_transport) {
+ spin_unlock(&tpg_np->tpg_np_parent_lock);
+ return tpg_np_child;
+ }
+ }
+ spin_unlock(&tpg_np->tpg_np_parent_lock);
+
+ return NULL;
+}
+
+struct iscsi_tpg_np *iscsit_tpg_add_network_portal(
+ struct iscsi_portal_group *tpg,
+ struct iscsi_np_addr *np_addr,
+ struct iscsi_tpg_np *tpg_np_parent,
+ int network_transport,
+ int af_inet)
+{
+ struct iscsi_np *np;
+ struct iscsi_tpg_np *tpg_np;
+
+ tpg_np = kzalloc(sizeof(struct iscsi_tpg_np), GFP_KERNEL);
+ if (!tpg_np) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_tpg_np.\n");
+ return ERR_PTR(-ENOMEM);
+ }
+
+ np = iscsit_add_np(np_addr, network_transport, af_inet);
+ if (IS_ERR(np)) {
+ kfree(tpg_np);
+ return ERR_CAST(np);
+ }
+
+ INIT_LIST_HEAD(&tpg_np->tpg_np_list);
+ INIT_LIST_HEAD(&tpg_np->tpg_np_child_list);
+ INIT_LIST_HEAD(&tpg_np->tpg_np_parent_list);
+ spin_lock_init(&tpg_np->tpg_np_parent_lock);
+ tpg_np->tpg_np = np;
+ tpg_np->tpg = tpg;
+
+ spin_lock(&tpg->tpg_np_lock);
+ list_add_tail(&tpg_np->tpg_np_list, &tpg->tpg_gnp_list);
+ tpg->num_tpg_nps++;
+ if (tpg->tpg_tiqn)
+ tpg->tpg_tiqn->tiqn_num_tpg_nps++;
+ spin_unlock(&tpg->tpg_np_lock);
+
+ if (tpg_np_parent) {
+ tpg_np->tpg_np_parent = tpg_np_parent;
+ spin_lock(&tpg_np_parent->tpg_np_parent_lock);
+ list_add_tail(&tpg_np->tpg_np_child_list,
+ &tpg_np_parent->tpg_np_parent_list);
+ spin_unlock(&tpg_np_parent->tpg_np_parent_lock);
+ }
+
+ printk(KERN_INFO "CORE[%s] - Added Network Portal: %s:%hu,%hu on %s on"
+ " network device: %s\n", tpg->tpg_tiqn->tiqn, np->np_ip,
+ np->np_port, tpg->tpgt,
+ (np->np_network_transport == ISCSI_TCP) ?
+ "TCP" : "SCTP", (strlen(np->np_net_dev)) ?
+ (char *)np->np_net_dev : "None");
+
+ return tpg_np;
+}
+
+static int iscsit_tpg_release_np(
+ struct iscsi_tpg_np *tpg_np,
+ struct iscsi_portal_group *tpg,
+ struct iscsi_np *np)
+{
+ iscsit_clear_tpg_np_login_thread(tpg_np, tpg);
+
+ printk(KERN_INFO "CORE[%s] - Removed Network Portal: %s:%hu,%hu on %s"
+ " on network device: %s\n", tpg->tpg_tiqn->tiqn, np->np_ip,
+ np->np_port, tpg->tpgt,
+ (np->np_network_transport == ISCSI_TCP) ?
+ "TCP" : "SCTP", (strlen(np->np_net_dev)) ?
+ (char *)np->np_net_dev : "None");
+
+ tpg_np->tpg_np = NULL;
+ tpg_np->tpg = NULL;
+ kfree(tpg_np);
+ /*
+ * iscsit_del_np() will shutdown struct iscsi_np when last TPG reference is released.
+ */
+ return iscsit_del_np(np);
+}
+
+int iscsit_tpg_del_network_portal(
+ struct iscsi_portal_group *tpg,
+ struct iscsi_tpg_np *tpg_np)
+{
+ struct iscsi_np *np;
+ struct iscsi_tpg_np *tpg_np_child, *tpg_np_child_tmp;
+ int ret = 0;
+
+ np = tpg_np->tpg_np;
+ if (!np) {
+ printk(KERN_ERR "Unable to locate struct iscsi_np from"
+ " struct iscsi_tpg_np\n");
+ return -EINVAL;
+ }
+
+ if (!tpg_np->tpg_np_parent) {
+ /*
+ * We are the parent tpg network portal. Release all of the
+ * child tpg_np's (eg: the non ISCSI_TCP ones) on our parent
+ * list first.
+ */
+ list_for_each_entry_safe(tpg_np_child, tpg_np_child_tmp,
+ &tpg_np->tpg_np_parent_list,
+ tpg_np_child_list) {
+ ret = iscsit_tpg_del_network_portal(tpg, tpg_np_child);
+ if (ret < 0)
+ printk(KERN_ERR "iscsit_tpg_del_network_portal()"
+ " failed: %d\n", ret);
+ }
+ } else {
+ /*
+ * We are not the parent ISCSI_TCP tpg network portal. Release
+ * our own network portals from the child list.
+ */
+ spin_lock(&tpg_np->tpg_np_parent->tpg_np_parent_lock);
+ list_del(&tpg_np->tpg_np_child_list);
+ spin_unlock(&tpg_np->tpg_np_parent->tpg_np_parent_lock);
+ }
+
+ spin_lock(&tpg->tpg_np_lock);
+ list_del(&tpg_np->tpg_np_list);
+ tpg->num_tpg_nps--;
+ if (tpg->tpg_tiqn)
+ tpg->tpg_tiqn->tiqn_num_tpg_nps--;
+ spin_unlock(&tpg->tpg_np_lock);
+
+ return iscsit_tpg_release_np(tpg_np, tpg, np);
+}
+
+int iscsit_tpg_set_initiator_node_queue_depth(
+ struct iscsi_portal_group *tpg,
+ unsigned char *initiatorname,
+ u32 queue_depth,
+ int force)
+{
+ return core_tpg_set_initiator_node_queue_depth(&tpg->tpg_se_tpg,
+ initiatorname, queue_depth, force);
+}
+
+int iscsit_ta_authentication(struct iscsi_portal_group *tpg, u32 authentication)
+{
+ unsigned char buf1[256], buf2[256], *none = NULL;
+ int len;
+ struct iscsi_param *param;
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ if ((authentication != 1) && (authentication != 0)) {
+ printk(KERN_ERR "Illegal value for authentication parameter:"
+ " %u, ignoring request.\n", authentication);
+ return -1;
+ }
+
+ memset(buf1, 0, sizeof(buf1));
+ memset(buf2, 0, sizeof(buf2));
+
+ param = iscsi_find_param_from_key(AUTHMETHOD, tpg->param_list);
+ if (!param)
+ return -EINVAL;
+
+ if (authentication) {
+ snprintf(buf1, sizeof(buf1), "%s", param->value);
+ none = strstr(buf1, NONE);
+ if (!none)
+ goto out;
+ if (!strncmp(none + 4, ",", 1)) {
+ if (!strcmp(buf1, none))
+ sprintf(buf2, "%s", none+5);
+ else {
+ none--;
+ *none = '\0';
+ len = sprintf(buf2, "%s", buf1);
+ none += 5;
+ sprintf(buf2 + len, "%s", none);
+ }
+ } else {
+ none--;
+ *none = '\0';
+ sprintf(buf2, "%s", buf1);
+ }
+ if (iscsi_update_param_value(param, buf2) < 0)
+ return -EINVAL;
+ } else {
+ snprintf(buf1, sizeof(buf1), "%s", param->value);
+ none = strstr(buf1, NONE);
+ if ((none))
+ goto out;
+ strncat(buf1, ",", strlen(","));
+ strncat(buf1, NONE, strlen(NONE));
+ if (iscsi_update_param_value(param, buf1) < 0)
+ return -EINVAL;
+ }
+
+out:
+ a->authentication = authentication;
+ printk(KERN_INFO "%s iSCSI Authentication Methods for TPG: %hu.\n",
+ a->authentication ? "Enforcing" : "Disabling", tpg->tpgt);
+
+ return 0;
+}
+
+int iscsit_ta_login_timeout(
+ struct iscsi_portal_group *tpg,
+ u32 login_timeout)
+{
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ if (login_timeout > TA_LOGIN_TIMEOUT_MAX) {
+ printk(KERN_ERR "Requested Login Timeout %u larger than maximum"
+ " %u\n", login_timeout, TA_LOGIN_TIMEOUT_MAX);
+ return -EINVAL;
+ } else if (login_timeout < TA_LOGIN_TIMEOUT_MIN) {
+ printk(KERN_ERR "Requested Logout Timeout %u smaller than"
+ " minimum %u\n", login_timeout, TA_LOGIN_TIMEOUT_MIN);
+ return -EINVAL;
+ }
+
+ a->login_timeout = login_timeout;
+ printk(KERN_INFO "Set Logout Timeout to %u for Target Portal Group"
+ " %hu\n", a->login_timeout, tpg->tpgt);
+
+ return 0;
+}
+
+int iscsit_ta_netif_timeout(
+ struct iscsi_portal_group *tpg,
+ u32 netif_timeout)
+{
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ if (netif_timeout > TA_NETIF_TIMEOUT_MAX) {
+ printk(KERN_ERR "Requested Network Interface Timeout %u larger"
+ " than maximum %u\n", netif_timeout,
+ TA_NETIF_TIMEOUT_MAX);
+ return -EINVAL;
+ } else if (netif_timeout < TA_NETIF_TIMEOUT_MIN) {
+ printk(KERN_ERR "Requested Network Interface Timeout %u smaller"
+ " than minimum %u\n", netif_timeout,
+ TA_NETIF_TIMEOUT_MIN);
+ return -EINVAL;
+ }
+
+ a->netif_timeout = netif_timeout;
+ printk(KERN_INFO "Set Network Interface Timeout to %u for"
+ " Target Portal Group %hu\n", a->netif_timeout, tpg->tpgt);
+
+ return 0;
+}
+
+int iscsit_ta_generate_node_acls(
+ struct iscsi_portal_group *tpg,
+ u32 flag)
+{
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ if ((flag != 0) && (flag != 1)) {
+ printk(KERN_ERR "Illegal value %d\n", flag);
+ return -EINVAL;
+ }
+
+ a->generate_node_acls = flag;
+ printk(KERN_INFO "iSCSI_TPG[%hu] - Generate Initiator Portal Group ACLs: %s\n",
+ tpg->tpgt, (a->generate_node_acls) ? "Enabled" : "Disabled");
+
+ return 0;
+}
+
+int iscsit_ta_default_cmdsn_depth(
+ struct iscsi_portal_group *tpg,
+ u32 tcq_depth)
+{
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ if (tcq_depth > TA_DEFAULT_CMDSN_DEPTH_MAX) {
+ printk(KERN_ERR "Requested Default Queue Depth: %u larger"
+ " than maximum %u\n", tcq_depth,
+ TA_DEFAULT_CMDSN_DEPTH_MAX);
+ return -EINVAL;
+ } else if (tcq_depth < TA_DEFAULT_CMDSN_DEPTH_MIN) {
+ printk(KERN_ERR "Requested Default Queue Depth: %u smaller"
+ " than minimum %u\n", tcq_depth,
+ TA_DEFAULT_CMDSN_DEPTH_MIN);
+ return -EINVAL;
+ }
+
+ a->default_cmdsn_depth = tcq_depth;
+ printk(KERN_INFO "iSCSI_TPG[%hu] - Set Default CmdSN TCQ Depth to %u\n",
+ tpg->tpgt, a->default_cmdsn_depth);
+
+ return 0;
+}
+
+int iscsit_ta_cache_dynamic_acls(
+ struct iscsi_portal_group *tpg,
+ u32 flag)
+{
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ if ((flag != 0) && (flag != 1)) {
+ printk(KERN_ERR "Illegal value %d\n", flag);
+ return -EINVAL;
+ }
+
+ a->cache_dynamic_acls = flag;
+ printk(KERN_INFO "iSCSI_TPG[%hu] - Cache Dynamic Initiator Portal Group"
+ " ACLs %s\n", tpg->tpgt, (a->cache_dynamic_acls) ?
+ "Enabled" : "Disabled");
+
+ return 0;
+}
+
+int iscsit_ta_demo_mode_write_protect(
+ struct iscsi_portal_group *tpg,
+ u32 flag)
+{
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ if ((flag != 0) && (flag != 1)) {
+ printk(KERN_ERR "Illegal value %d\n", flag);
+ return -EINVAL;
+ }
+
+ a->demo_mode_write_protect = flag;
+ printk(KERN_INFO "iSCSI_TPG[%hu] - Demo Mode Write Protect bit: %s\n",
+ tpg->tpgt, (a->demo_mode_write_protect) ? "ON" : "OFF");
+
+ return 0;
+}
+
+int iscsit_ta_prod_mode_write_protect(
+ struct iscsi_portal_group *tpg,
+ u32 flag)
+{
+ struct iscsi_tpg_attrib *a = &tpg->tpg_attrib;
+
+ if ((flag != 0) && (flag != 1)) {
+ printk(KERN_ERR "Illegal value %d\n", flag);
+ return -EINVAL;
+ }
+
+ a->prod_mode_write_protect = flag;
+ printk(KERN_INFO "iSCSI_TPG[%hu] - Production Mode Write Protect bit:"
+ " %s\n", tpg->tpgt, (a->prod_mode_write_protect) ?
+ "ON" : "OFF");
+
+ return 0;
+}
diff --git a/drivers/target/iscsi/iscsi_target_tpg.h b/drivers/target/iscsi/iscsi_target_tpg.h
new file mode 100644
index 0000000..31de29a
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_tpg.h
@@ -0,0 +1,43 @@
+#ifndef ISCSI_TARGET_TPG_H
+#define ISCSI_TARGET_TPG_H
+
+extern struct iscsi_portal_group *iscsit_alloc_portal_group(struct iscsi_tiqn *, u16);
+extern int iscsit_load_discovery_tpg(void);
+extern void iscsit_release_discovery_tpg(void);
+extern struct iscsi_portal_group *iscsit_get_tpg_from_np(struct iscsi_tiqn *,
+ struct iscsi_np *);
+extern int iscsit_get_tpg(struct iscsi_portal_group *);
+extern void iscsit_put_tpg(struct iscsi_portal_group *);
+extern void iscsit_clear_tpg_np_login_threads(struct iscsi_portal_group *);
+extern void iscsit_tpg_dump_params(struct iscsi_portal_group *);
+extern int iscsit_tpg_add_portal_group(struct iscsi_tiqn *, struct iscsi_portal_group *);
+extern int iscsit_tpg_del_portal_group(struct iscsi_tiqn *, struct iscsi_portal_group *,
+ int);
+extern int iscsit_tpg_enable_portal_group(struct iscsi_portal_group *);
+extern int iscsit_tpg_disable_portal_group(struct iscsi_portal_group *, int);
+extern struct iscsi_node_acl *iscsit_tpg_add_initiator_node_acl(
+ struct iscsi_portal_group *, const char *, u32);
+extern void iscsit_tpg_del_initiator_node_acl(struct iscsi_portal_group *,
+ struct se_node_acl *);
+extern struct iscsi_node_attrib *iscsit_tpg_get_node_attrib(struct iscsi_session *);
+extern void iscsit_tpg_del_external_nps(struct iscsi_tpg_np *);
+extern struct iscsi_tpg_np *iscsit_tpg_locate_child_np(struct iscsi_tpg_np *, int);
+extern struct iscsi_tpg_np *iscsit_tpg_add_network_portal(struct iscsi_portal_group *,
+ struct iscsi_np_addr *, struct iscsi_tpg_np *, int, int);
+extern int iscsit_tpg_del_network_portal(struct iscsi_portal_group *,
+ struct iscsi_tpg_np *);
+extern int iscsit_tpg_set_initiator_node_queue_depth(struct iscsi_portal_group *,
+ unsigned char *, u32, int);
+extern int iscsit_ta_authentication(struct iscsi_portal_group *, u32);
+extern int iscsit_ta_login_timeout(struct iscsi_portal_group *, u32);
+extern int iscsit_ta_netif_timeout(struct iscsi_portal_group *, u32);
+extern int iscsit_ta_generate_node_acls(struct iscsi_portal_group *, u32);
+extern int iscsit_ta_default_cmdsn_depth(struct iscsi_portal_group *, u32);
+extern int iscsit_ta_cache_dynamic_acls(struct iscsi_portal_group *, u32);
+extern int iscsit_ta_demo_mode_write_protect(struct iscsi_portal_group *, u32);
+extern int iscsit_ta_prod_mode_write_protect(struct iscsi_portal_group *, u32);
+
+extern struct iscsi_global *iscsi_global;
+extern struct target_fabric_configfs *lio_target_fabric_configfs;
+
+#endif /* ISCSI_TARGET_TPG_H */
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 08/14] iscsi-target: Add iSCSI Login Negotiation and Parameter logic
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (7 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 07/14] iscsi-target: Add TPG and Device logic Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 09/14] iscsi-target: Add CHAP Authentication support using libcrypto Nicholas A. Bellinger
` (6 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch adds the princple RFC-3720 compatiable iSCSI Login
phase negotiation for iscsi_target_mod. This also includes the
target RX/TX thread queue logic which is called directly from iSCSI
login associated code.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_login.c | 1245 +++++++++++++++
drivers/target/iscsi/iscsi_target_login.h | 13 +
drivers/target/iscsi/iscsi_target_nego.c | 1068 +++++++++++++
drivers/target/iscsi/iscsi_target_nego.h | 19 +
drivers/target/iscsi/iscsi_target_parameters.c | 1910 ++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_parameters.h | 271 ++++
drivers/target/iscsi/iscsi_target_tq.c | 552 +++++++
drivers/target/iscsi/iscsi_target_tq.h | 89 ++
8 files changed, 5167 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_login.c
create mode 100644 drivers/target/iscsi/iscsi_target_login.h
create mode 100644 drivers/target/iscsi/iscsi_target_nego.c
create mode 100644 drivers/target/iscsi/iscsi_target_nego.h
create mode 100644 drivers/target/iscsi/iscsi_target_parameters.c
create mode 100644 drivers/target/iscsi/iscsi_target_parameters.h
create mode 100644 drivers/target/iscsi/iscsi_target_tq.c
create mode 100644 drivers/target/iscsi/iscsi_target_tq.h
diff --git a/drivers/target/iscsi/iscsi_target_login.c b/drivers/target/iscsi/iscsi_target_login.c
new file mode 100644
index 0000000..7cb258b
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_login.c
@@ -0,0 +1,1245 @@
+/*******************************************************************************
+ * This file contains the login functions used by the iSCSI Target driver.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/kthread.h>
+#include <linux/crypto.h>
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_tq.h"
+#include "iscsi_target_device.h"
+#include "iscsi_target_nego.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_erl2.h"
+#include "iscsi_target_login.h"
+#include "iscsi_target_stat.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target.h"
+#include "iscsi_target_parameters.h"
+
+extern struct idr sess_idr;
+extern struct mutex auth_id_lock;
+extern spinlock_t sess_idr_lock;
+
+static int iscsi_login_init_conn(struct iscsi_conn *conn)
+{
+ INIT_LIST_HEAD(&conn->conn_list);
+ INIT_LIST_HEAD(&conn->conn_cmd_list);
+ INIT_LIST_HEAD(&conn->immed_queue_list);
+ INIT_LIST_HEAD(&conn->response_queue_list);
+ init_completion(&conn->conn_post_wait_comp);
+ init_completion(&conn->conn_wait_comp);
+ init_completion(&conn->conn_wait_rcfr_comp);
+ init_completion(&conn->conn_waiting_on_uc_comp);
+ init_completion(&conn->conn_logout_comp);
+ init_completion(&conn->rx_half_close_comp);
+ init_completion(&conn->tx_half_close_comp);
+ spin_lock_init(&conn->cmd_lock);
+ spin_lock_init(&conn->conn_usage_lock);
+ spin_lock_init(&conn->immed_queue_lock);
+ spin_lock_init(&conn->nopin_timer_lock);
+ spin_lock_init(&conn->response_queue_lock);
+ spin_lock_init(&conn->state_lock);
+
+ if (!(zalloc_cpumask_var(&conn->conn_cpumask, GFP_KERNEL))) {
+ printk(KERN_ERR "Unable to allocate conn->conn_cpumask\n");
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+/*
+ * Used by iscsi_target_nego.c:iscsi_target_locate_portal() to setup
+ * per struct iscsi_conn libcrypto contexts for crc32c and crc32-intel
+ */
+int iscsi_login_setup_crypto(struct iscsi_conn *conn)
+{
+ /*
+ * Setup slicing by CRC32C algorithm for RX and TX libcrypto contexts
+ * which will default to crc32c_intel.ko for cpu_has_xmm4_2, or fallback
+ * to software 1x8 byte slicing from crc32c.ko
+ */
+ conn->conn_rx_hash.flags = 0;
+ conn->conn_rx_hash.tfm = crypto_alloc_hash("crc32c", 0,
+ CRYPTO_ALG_ASYNC);
+ if (IS_ERR(conn->conn_rx_hash.tfm)) {
+ printk(KERN_ERR "crypto_alloc_hash() failed for conn_rx_tfm\n");
+ return -ENOMEM;
+ }
+
+ conn->conn_tx_hash.flags = 0;
+ conn->conn_tx_hash.tfm = crypto_alloc_hash("crc32c", 0,
+ CRYPTO_ALG_ASYNC);
+ if (IS_ERR(conn->conn_tx_hash.tfm)) {
+ printk(KERN_ERR "crypto_alloc_hash() failed for conn_tx_tfm\n");
+ crypto_free_hash(conn->conn_rx_hash.tfm);
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+static int iscsi_login_check_initiator_version(
+ struct iscsi_conn *conn,
+ u8 version_max,
+ u8 version_min)
+{
+ if ((version_max != 0x00) || (version_min != 0x00)) {
+ printk(KERN_ERR "Unsupported iSCSI IETF Pre-RFC Revision,"
+ " version Min/Max 0x%02x/0x%02x, rejecting login.\n",
+ version_min, version_max);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_NO_VERSION);
+ return -1;
+ }
+
+ return 0;
+}
+
+int iscsi_check_for_session_reinstatement(struct iscsi_conn *conn)
+{
+ int sessiontype;
+ struct iscsi_param *initiatorname_param = NULL, *sessiontype_param = NULL;
+ struct iscsi_portal_group *tpg = conn->tpg;
+ struct iscsi_session *sess = NULL, *sess_p = NULL;
+ struct se_portal_group *se_tpg = &tpg->tpg_se_tpg;
+ struct se_session *se_sess, *se_sess_tmp;
+
+ initiatorname_param = iscsi_find_param_from_key(
+ INITIATORNAME, conn->param_list);
+ if (!initiatorname_param)
+ return -1;
+
+ sessiontype_param = iscsi_find_param_from_key(
+ SESSIONTYPE, conn->param_list);
+ if (!sessiontype_param)
+ return -1;
+
+ sessiontype = (strncmp(sessiontype_param->value, NORMAL, 6)) ? 1 : 0;
+
+ spin_lock_bh(&se_tpg->session_lock);
+ list_for_each_entry_safe(se_sess, se_sess_tmp, &se_tpg->tpg_sess_list,
+ sess_list) {
+
+ sess_p = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ spin_lock(&sess_p->conn_lock);
+ if (atomic_read(&sess_p->session_fall_back_to_erl0) ||
+ atomic_read(&sess_p->session_logout) ||
+ (sess_p->time2retain_timer_flags & ISCSI_TF_EXPIRED)) {
+ spin_unlock(&sess_p->conn_lock);
+ continue;
+ }
+ if (!memcmp((void *)sess_p->isid, (void *)conn->sess->isid, 6) &&
+ (!strcmp((void *)sess_p->sess_ops->InitiatorName,
+ (void *)initiatorname_param->value) &&
+ (sess_p->sess_ops->SessionType == sessiontype))) {
+ atomic_set(&sess_p->session_reinstatement, 1);
+ spin_unlock(&sess_p->conn_lock);
+ iscsit_inc_session_usage_count(sess_p);
+ iscsit_stop_time2retain_timer(sess_p);
+ sess = sess_p;
+ break;
+ }
+ spin_unlock(&sess_p->conn_lock);
+ }
+ spin_unlock_bh(&se_tpg->session_lock);
+ /*
+ * If the Time2Retain handler has expired, the session is already gone.
+ */
+ if (!sess)
+ return 0;
+
+ TRACE(TRACE_ERL0, "%s iSCSI Session SID %u is still active for %s,"
+ " preforming session reinstatement.\n", (sessiontype) ?
+ "Discovery" : "Normal", sess->sid,
+ sess->sess_ops->InitiatorName);
+
+ spin_lock_bh(&sess->conn_lock);
+ if (sess->session_state == TARG_SESS_STATE_FAILED) {
+ spin_unlock_bh(&sess->conn_lock);
+ iscsit_dec_session_usage_count(sess);
+ return iscsit_close_session(sess);
+ }
+ spin_unlock_bh(&sess->conn_lock);
+
+ iscsit_stop_session(sess, 1, 1);
+ iscsit_dec_session_usage_count(sess);
+
+ return iscsit_close_session(sess);
+}
+
+static void iscsi_login_set_conn_values(
+ struct iscsi_session *sess,
+ struct iscsi_conn *conn,
+ u16 cid)
+{
+ conn->sess = sess;
+ conn->cid = cid;
+ /*
+ * Generate a random Status sequence number (statsn) for the new
+ * iSCSI connection.
+ */
+ get_random_bytes(&conn->stat_sn, sizeof(u32));
+
+ mutex_lock(&auth_id_lock);
+ conn->auth_id = iscsi_global->auth_id++;
+ mutex_unlock(&auth_id_lock);
+}
+
+/*
+ * This is the leading connection of a new session,
+ * or session reinstatement.
+ */
+static int iscsi_login_zero_tsih_s1(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ struct iscsi_session *sess = NULL;
+ struct iscsi_login_req *pdu = (struct iscsi_login_req *)buf;
+
+ sess = kzalloc(sizeof(struct iscsi_session), GFP_KERNEL);
+ if (!sess) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ printk(KERN_ERR "Could not allocate memory for session\n");
+ return -1;
+ }
+
+ iscsi_login_set_conn_values(sess, conn, pdu->cid);
+ sess->init_task_tag = pdu->itt;
+ memcpy((void *)&sess->isid, (void *)pdu->isid, 6);
+ sess->exp_cmd_sn = pdu->cmdsn;
+ INIT_LIST_HEAD(&sess->sess_conn_list);
+ INIT_LIST_HEAD(&sess->sess_ooo_cmdsn_list);
+ INIT_LIST_HEAD(&sess->cr_active_list);
+ INIT_LIST_HEAD(&sess->cr_inactive_list);
+ init_completion(&sess->async_msg_comp);
+ init_completion(&sess->reinstatement_comp);
+ init_completion(&sess->session_wait_comp);
+ spin_lock_init(&sess->cmdsn_lock);
+ spin_lock_init(&sess->conn_lock);
+ spin_lock_init(&sess->cr_a_lock);
+ spin_lock_init(&sess->cr_i_lock);
+ spin_lock_init(&sess->session_usage_lock);
+ spin_lock_init(&sess->ttt_lock);
+
+ if (!idr_pre_get(&sess_idr, GFP_KERNEL)) {
+ printk(KERN_ERR "idr_pre_get() for sess_idr failed\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+ spin_lock(&sess_idr_lock);
+ idr_get_new(&sess_idr, NULL, &sess->session_index);
+ spin_unlock(&sess_idr_lock);
+
+ sess->creation_time = get_jiffies_64();
+ spin_lock_init(&sess->session_stats_lock);
+ /*
+ * The FFP CmdSN window values will be allocated from the TPG's
+ * Initiator Node's ACL once the login has been successfully completed.
+ */
+ sess->max_cmd_sn = pdu->cmdsn;
+
+ sess->sess_ops = kzalloc(sizeof(struct iscsi_sess_ops), GFP_KERNEL);
+ if (!sess->sess_ops) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_sess_ops.\n");
+ return -1;
+ }
+
+ sess->se_sess = transport_init_session();
+ if (!sess->se_sess) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+
+ return 0;
+}
+
+static int iscsi_login_zero_tsih_s2(
+ struct iscsi_conn *conn)
+{
+ struct iscsi_node_attrib *na;
+ struct iscsi_session *sess = conn->sess;
+ unsigned char buf[32];
+
+ sess->tpg = conn->tpg;
+
+ /*
+ * Assign a new TPG Session Handle. Note this is protected with
+ * struct iscsi_portal_group->np_login_sem from iscsit_access_np().
+ */
+ sess->tsih = ++ISCSI_TPG_S(sess)->ntsih;
+ if (!sess->tsih)
+ sess->tsih = ++ISCSI_TPG_S(sess)->ntsih;
+
+ /*
+ * Create the default params from user defined values..
+ */
+ if (iscsi_copy_param_list(&conn->param_list,
+ ISCSI_TPG_C(conn)->param_list, 1) < 0) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+
+ iscsi_set_keys_to_negotiate(0, conn->param_list);
+
+ if (sess->sess_ops->SessionType)
+ return iscsi_set_keys_irrelevant_for_discovery(
+ conn->param_list);
+
+ na = iscsit_tpg_get_node_attrib(sess);
+
+ /*
+ * Need to send TargetPortalGroupTag back in first login response
+ * on any iSCSI connection where the Initiator provides TargetName.
+ * See 5.3.1. Login Phase Start
+ *
+ * In our case, we have already located the struct iscsi_tiqn at this point.
+ */
+ memset(buf, 0, 32);
+ sprintf(buf, "TargetPortalGroupTag=%hu", ISCSI_TPG_S(sess)->tpgt);
+ if (iscsi_change_param_value(buf, conn->param_list, 0) < 0) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+
+ /*
+ * Workaround for Initiators that have broken connection recovery logic.
+ *
+ * "We would really like to get rid of this." Linux-iSCSI.org team
+ */
+ memset(buf, 0, 32);
+ sprintf(buf, "ErrorRecoveryLevel=%d", na->default_erl);
+ if (iscsi_change_param_value(buf, conn->param_list, 0) < 0) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+
+ if (iscsi_login_disable_FIM_keys(conn->param_list, conn) < 0)
+ return -1;
+
+ return 0;
+}
+
+/*
+ * Remove PSTATE_NEGOTIATE for the four FIM related keys.
+ * The Initiator node will be able to enable FIM by proposing them itself.
+ */
+int iscsi_login_disable_FIM_keys(
+ struct iscsi_param_list *param_list,
+ struct iscsi_conn *conn)
+{
+ struct iscsi_param *param;
+
+ param = iscsi_find_param_from_key("OFMarker", param_list);
+ if (!param) {
+ printk(KERN_ERR "iscsi_find_param_from_key() for"
+ " OFMarker failed\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+ param->state &= ~PSTATE_NEGOTIATE;
+
+ param = iscsi_find_param_from_key("OFMarkInt", param_list);
+ if (!param) {
+ printk(KERN_ERR "iscsi_find_param_from_key() for"
+ " IFMarker failed\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+ param->state &= ~PSTATE_NEGOTIATE;
+
+ param = iscsi_find_param_from_key("IFMarker", param_list);
+ if (!param) {
+ printk(KERN_ERR "iscsi_find_param_from_key() for"
+ " IFMarker failed\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+ param->state &= ~PSTATE_NEGOTIATE;
+
+ param = iscsi_find_param_from_key("IFMarkInt", param_list);
+ if (!param) {
+ printk(KERN_ERR "iscsi_find_param_from_key() for"
+ " IFMarker failed\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+ param->state &= ~PSTATE_NEGOTIATE;
+
+ return 0;
+}
+
+static int iscsi_login_non_zero_tsih_s1(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ struct iscsi_login_req *pdu = (struct iscsi_login_req *)buf;
+
+ iscsi_login_set_conn_values(NULL, conn, pdu->cid);
+ return 0;
+}
+
+/*
+ * Add a new connection to an existing session.
+ */
+static int iscsi_login_non_zero_tsih_s2(
+ struct iscsi_conn *conn,
+ unsigned char *buf)
+{
+ struct iscsi_portal_group *tpg = conn->tpg;
+ struct iscsi_session *sess = NULL, *sess_p = NULL;
+ struct se_portal_group *se_tpg = &tpg->tpg_se_tpg;
+ struct se_session *se_sess, *se_sess_tmp;
+ struct iscsi_login_req *pdu = (struct iscsi_login_req *)buf;
+
+ spin_lock_bh(&se_tpg->session_lock);
+ list_for_each_entry_safe(se_sess, se_sess_tmp, &se_tpg->tpg_sess_list,
+ sess_list) {
+
+ sess_p = (struct iscsi_session *)se_sess->fabric_sess_ptr;
+ if (atomic_read(&sess_p->session_fall_back_to_erl0) ||
+ atomic_read(&sess_p->session_logout) ||
+ (sess_p->time2retain_timer_flags & ISCSI_TF_EXPIRED))
+ continue;
+ if (!(memcmp((const void *)sess_p->isid,
+ (const void *)pdu->isid, 6)) &&
+ (sess_p->tsih == pdu->tsih)) {
+ iscsit_inc_session_usage_count(sess_p);
+ iscsit_stop_time2retain_timer(sess_p);
+ sess = sess_p;
+ break;
+ }
+ }
+ spin_unlock_bh(&se_tpg->session_lock);
+
+ /*
+ * If the Time2Retain handler has expired, the session is already gone.
+ */
+ if (!sess) {
+ printk(KERN_ERR "Initiator attempting to add a connection to"
+ " a non-existent session, rejecting iSCSI Login.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_NO_SESSION);
+ return -1;
+ }
+
+ /*
+ * Stop the Time2Retain timer if this is a failed session, we restart
+ * the timer if the login is not successful.
+ */
+ spin_lock_bh(&sess->conn_lock);
+ if (sess->session_state == TARG_SESS_STATE_FAILED)
+ atomic_set(&sess->session_continuation, 1);
+ spin_unlock_bh(&sess->conn_lock);
+
+ iscsi_login_set_conn_values(sess, conn, pdu->cid);
+
+ if (iscsi_copy_param_list(&conn->param_list,
+ ISCSI_TPG_C(conn)->param_list, 0) < 0) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+
+ iscsi_set_keys_to_negotiate(0, conn->param_list);
+ /*
+ * Need to send TargetPortalGroupTag back in first login response
+ * on any iSCSI connection where the Initiator provides TargetName.
+ * See 5.3.1. Login Phase Start
+ *
+ * In our case, we have already located the struct iscsi_tiqn at this point.
+ */
+ memset(buf, 0, 32);
+ sprintf(buf, "TargetPortalGroupTag=%hu", ISCSI_TPG_S(sess)->tpgt);
+ if (iscsi_change_param_value(buf, conn->param_list, 0) < 0) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ return -1;
+ }
+
+ return iscsi_login_disable_FIM_keys(conn->param_list, conn);
+}
+
+int iscsi_login_post_auth_non_zero_tsih(
+ struct iscsi_conn *conn,
+ u16 cid,
+ u32 exp_statsn)
+{
+ struct iscsi_conn *conn_ptr = NULL;
+ struct iscsi_conn_recovery *cr = NULL;
+ struct iscsi_session *sess = conn->sess;
+
+ /*
+ * By following item 5 in the login table, if we have found
+ * an existing ISID and a valid/existing TSIH and an existing
+ * CID we do connection reinstatement. Currently we dont not
+ * support it so we send back an non-zero status class to the
+ * initiator and release the new connection.
+ */
+ conn_ptr = iscsit_get_conn_from_cid_rcfr(sess, cid);
+ if ((conn_ptr)) {
+ printk(KERN_ERR "Connection exists with CID %hu for %s,"
+ " performing connection reinstatement.\n",
+ conn_ptr->cid, sess->sess_ops->InitiatorName);
+
+ iscsit_connection_reinstatement_rcfr(conn_ptr);
+ iscsit_dec_conn_usage_count(conn_ptr);
+ }
+
+ /*
+ * Check for any connection recovery entires containing CID.
+ * We use the original ExpStatSN sent in the first login request
+ * to acknowledge commands for the failed connection.
+ *
+ * Also note that an explict logout may have already been sent,
+ * but the response may not be sent due to additional connection
+ * loss.
+ */
+ if (sess->sess_ops->ErrorRecoveryLevel == 2) {
+ cr = iscsit_get_inactive_connection_recovery_entry(
+ sess, cid);
+ if ((cr)) {
+ TRACE(TRACE_ERL2, "Performing implicit logout"
+ " for connection recovery on CID: %hu\n",
+ conn->cid);
+ iscsit_discard_cr_cmds_by_expstatsn(cr, exp_statsn);
+ }
+ }
+
+ /*
+ * Else we follow item 4 from the login table in that we have
+ * found an existing ISID and a valid/existing TSIH and a new
+ * CID we go ahead and continue to add a new connection to the
+ * session.
+ */
+ TRACE(TRACE_LOGIN, "Adding CID %hu to existing session for %s.\n",
+ cid, sess->sess_ops->InitiatorName);
+
+ if ((atomic_read(&sess->nconn) + 1) > sess->sess_ops->MaxConnections) {
+ printk(KERN_ERR "Adding additional connection to this session"
+ " would exceed MaxConnections %d, login failed.\n",
+ sess->sess_ops->MaxConnections);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_ISID_ERROR);
+ return -1;
+ }
+
+ return 0;
+}
+
+static void iscsi_post_login_start_timers(struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+
+ if (!sess->sess_ops->SessionType)
+ iscsit_start_nopin_timer(conn);
+}
+
+static int iscsi_post_login_handler(
+ struct iscsi_np *np,
+ struct iscsi_conn *conn,
+ u8 zero_tsih)
+{
+ int stop_timer = 0;
+ struct iscsi_session *sess = conn->sess;
+ struct se_session *se_sess = sess->se_sess;
+ struct iscsi_portal_group *tpg = ISCSI_TPG_S(sess);
+ struct se_portal_group *se_tpg = &tpg->tpg_se_tpg;
+ struct iscsi_thread_set *ts;
+
+ iscsit_inc_conn_usage_count(conn);
+
+ iscsit_collect_login_stats(conn, ISCSI_STATUS_CLS_SUCCESS,
+ ISCSI_LOGIN_STATUS_ACCEPT);
+
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_LOGGED_IN.\n");
+ conn->conn_state = TARG_CONN_STATE_LOGGED_IN;
+
+ iscsi_set_connection_parameters(conn->conn_ops, conn->param_list);
+ iscsit_set_sync_and_steering_values(conn);
+ /*
+ * SCSI Initiator -> SCSI Target Port Mapping
+ */
+ ts = iscsi_get_thread_set();
+ if (!zero_tsih) {
+ iscsi_set_session_parameters(sess->sess_ops,
+ conn->param_list, 0);
+ iscsi_release_param_list(conn->param_list);
+ conn->param_list = NULL;
+
+ spin_lock_bh(&sess->conn_lock);
+ atomic_set(&sess->session_continuation, 0);
+ if (sess->session_state == TARG_SESS_STATE_FAILED) {
+ TRACE(TRACE_STATE, "Moving to"
+ " TARG_SESS_STATE_LOGGED_IN.\n");
+ sess->session_state = TARG_SESS_STATE_LOGGED_IN;
+ stop_timer = 1;
+ }
+
+ printk(KERN_INFO "iSCSI Login successful on CID: %hu from %s to"
+ " %s:%hu,%hu\n", conn->cid, conn->login_ip, np->np_ip,
+ np->np_port, tpg->tpgt);
+
+ list_add_tail(&conn->conn_list, &sess->sess_conn_list);
+ atomic_inc(&sess->nconn);
+ printk(KERN_INFO "Incremented iSCSI Connection count to %hu"
+ " from node: %s\n", atomic_read(&sess->nconn),
+ sess->sess_ops->InitiatorName);
+ spin_unlock_bh(&sess->conn_lock);
+
+ iscsi_post_login_start_timers(conn);
+ iscsi_activate_thread_set(conn, ts);
+ /*
+ * Determine CPU mask to ensure connection's RX and TX kthreads
+ * are scheduled on the same CPU.
+ */
+ iscsit_thread_get_cpumask(conn);
+ conn->conn_rx_reset_cpumask = 1;
+ conn->conn_tx_reset_cpumask = 1;
+
+ iscsit_dec_conn_usage_count(conn);
+ if (stop_timer) {
+ spin_lock_bh(&se_tpg->session_lock);
+ iscsit_stop_time2retain_timer(sess);
+ spin_unlock_bh(&se_tpg->session_lock);
+ }
+ iscsit_dec_session_usage_count(sess);
+ return 0;
+ }
+
+ iscsi_set_session_parameters(sess->sess_ops, conn->param_list, 1);
+ iscsi_release_param_list(conn->param_list);
+ conn->param_list = NULL;
+
+ iscsit_determine_maxcmdsn(sess);
+
+ spin_lock_bh(&se_tpg->session_lock);
+ __transport_register_session(&sess->tpg->tpg_se_tpg,
+ se_sess->se_node_acl, se_sess, (void *)sess);
+ TRACE(TRACE_STATE, "Moving to TARG_SESS_STATE_LOGGED_IN.\n");
+ sess->session_state = TARG_SESS_STATE_LOGGED_IN;
+
+ printk(KERN_INFO "iSCSI Login successful on CID: %hu from %s to %s:%hu,%hu\n",
+ conn->cid, conn->login_ip, np->np_ip, np->np_port, tpg->tpgt);
+
+ spin_lock_bh(&sess->conn_lock);
+ list_add_tail(&conn->conn_list, &sess->sess_conn_list);
+ atomic_inc(&sess->nconn);
+ printk(KERN_INFO "Incremented iSCSI Connection count to %hu from node:"
+ " %s\n", atomic_read(&sess->nconn),
+ sess->sess_ops->InitiatorName);
+ spin_unlock_bh(&sess->conn_lock);
+
+ sess->sid = tpg->sid++;
+ if (!sess->sid)
+ sess->sid = tpg->sid++;
+ printk(KERN_INFO "Established iSCSI session from node: %s\n",
+ sess->sess_ops->InitiatorName);
+
+ tpg->nsessions++;
+ if (tpg->tpg_tiqn)
+ tpg->tpg_tiqn->tiqn_nsessions++;
+
+ printk(KERN_INFO "Incremented number of active iSCSI sessions to %u on"
+ " iSCSI Target Portal Group: %hu\n", tpg->nsessions, tpg->tpgt);
+ spin_unlock_bh(&se_tpg->session_lock);
+
+ iscsi_post_login_start_timers(conn);
+ iscsi_activate_thread_set(conn, ts);
+ /*
+ * Determine CPU mask to ensure connection's RX and TX kthreads
+ * are scheduled on the same CPU.
+ */
+ iscsit_thread_get_cpumask(conn);
+ conn->conn_rx_reset_cpumask = 1;
+ conn->conn_tx_reset_cpumask = 1;
+
+ iscsit_dec_conn_usage_count(conn);
+
+ return 0;
+}
+
+static void iscsi_handle_login_thread_timeout(unsigned long data)
+{
+ struct iscsi_np *np = (struct iscsi_np *) data;
+
+ spin_lock_bh(&np->np_thread_lock);
+ printk(KERN_ERR "iSCSI Login timeout on Network Portal %s:%hu\n",
+ np->np_ip, np->np_port);
+
+ if (np->np_login_timer_flags & ISCSI_TF_STOP) {
+ spin_unlock_bh(&np->np_thread_lock);
+ return;
+ }
+
+ if (np->np_thread)
+ send_sig(SIGINT, np->np_thread, 1);
+
+ np->np_login_timer_flags &= ~ISCSI_TF_RUNNING;
+ spin_unlock_bh(&np->np_thread_lock);
+}
+
+static void iscsi_start_login_thread_timer(struct iscsi_np *np)
+{
+ /*
+ * This used the TA_LOGIN_TIMEOUT constant because at this
+ * point we do not have access to ISCSI_TPG_ATTRIB(tpg)->login_timeout
+ */
+ spin_lock_bh(&np->np_thread_lock);
+ init_timer(&np->np_login_timer);
+ np->np_login_timer.expires = (get_jiffies_64() + TA_LOGIN_TIMEOUT * HZ);
+ np->np_login_timer.data = (unsigned long)np;
+ np->np_login_timer.function = iscsi_handle_login_thread_timeout;
+ np->np_login_timer_flags &= ~ISCSI_TF_STOP;
+ np->np_login_timer_flags |= ISCSI_TF_RUNNING;
+ add_timer(&np->np_login_timer);
+
+ TRACE(TRACE_LOGIN, "Added timeout timer to iSCSI login request for"
+ " %u seconds.\n", TA_LOGIN_TIMEOUT);
+ spin_unlock_bh(&np->np_thread_lock);
+}
+
+static void iscsi_stop_login_thread_timer(struct iscsi_np *np)
+{
+ spin_lock_bh(&np->np_thread_lock);
+ if (!(np->np_login_timer_flags & ISCSI_TF_RUNNING)) {
+ spin_unlock_bh(&np->np_thread_lock);
+ return;
+ }
+ np->np_login_timer_flags |= ISCSI_TF_STOP;
+ spin_unlock_bh(&np->np_thread_lock);
+
+ del_timer_sync(&np->np_login_timer);
+
+ spin_lock_bh(&np->np_thread_lock);
+ np->np_login_timer_flags &= ~ISCSI_TF_RUNNING;
+ spin_unlock_bh(&np->np_thread_lock);
+}
+
+int iscsi_target_setup_login_socket(struct iscsi_np *np, int af_inet)
+{
+ const char *end;
+ struct socket *sock;
+ int backlog = 5, ret, opt = 0, len;
+ struct sockaddr_in *sock_in;
+ struct sockaddr_in6 *sock_in6;
+
+ switch (np->np_network_transport) {
+ case ISCSI_TCP:
+ np->np_ip_proto = IPPROTO_TCP;
+ np->np_sock_type = SOCK_STREAM;
+ break;
+ case ISCSI_SCTP_TCP:
+ np->np_ip_proto = IPPROTO_SCTP;
+ np->np_sock_type = SOCK_STREAM;
+ break;
+ case ISCSI_SCTP_UDP:
+ np->np_ip_proto = IPPROTO_SCTP;
+ np->np_sock_type = SOCK_SEQPACKET;
+ break;
+ case ISCSI_IWARP_TCP:
+ case ISCSI_IWARP_SCTP:
+ case ISCSI_INFINIBAND:
+ default:
+ printk(KERN_ERR "Unsupported network_transport: %d\n",
+ np->np_network_transport);
+ return -EINVAL;
+ }
+
+ ret = sock_create(af_inet, np->np_sock_type, np->np_ip_proto, &sock);
+ if (ret < 0) {
+ printk(KERN_ERR "sock_create() failed.\n");
+ return ret;
+ }
+ np->np_socket = sock;
+ /*
+ * The SCTP stack needs struct socket->file.
+ */
+ if ((np->np_network_transport == ISCSI_SCTP_TCP) ||
+ (np->np_network_transport == ISCSI_SCTP_UDP)) {
+ if (!sock->file) {
+ sock->file = kzalloc(sizeof(struct file), GFP_KERNEL);
+ if (!sock->file) {
+ printk(KERN_ERR "Unable to allocate struct"
+ " file for SCTP\n");
+ ret = -ENOMEM;
+ goto fail;
+ }
+ np->np_flags |= NPF_SCTP_STRUCT_FILE;
+ }
+ }
+
+ if (af_inet == AF_INET6) {
+ sock_in6 = (struct sockaddr_in6 *)&np->np_sockaddr;
+ sock_in6->sin6_family = AF_INET6;
+ sock_in6->sin6_port = htons(np->np_port);
+ len = sizeof(struct sockaddr_in6);
+
+ ret = in6_pton(&np->np_ip[0], IPV6_ADDRESS_SPACE,
+ (void *)&sock_in6->sin6_addr.in6_u, -1, &end);
+ if (ret <= 0) {
+ printk(KERN_ERR "in6_pton returned: %d\n", ret);
+ ret = -EINVAL;
+ goto fail;
+ }
+ } else {
+ sock_in = (struct sockaddr_in *)&np->np_sockaddr;
+ sock_in->sin_family = AF_INET;
+ sock_in->sin_port = htons(np->np_port);
+ sock_in->sin_addr.s_addr = htonl(np->np_ipv4);
+ len = sizeof(struct sockaddr);
+ }
+ /*
+ * Set SO_REUSEADDR, and disable Nagel Algorithm with TCP_NODELAY.
+ */
+ opt = 1;
+ if (np->np_network_transport == ISCSI_TCP) {
+ ret = kernel_setsockopt(sock, IPPROTO_TCP, TCP_NODELAY,
+ (char *)&opt, sizeof(opt));
+ if (ret < 0) {
+ printk(KERN_ERR "kernel_setsockopt() for TCP_NODELAY"
+ " failed: %d\n", ret);
+ goto fail;
+ }
+ }
+
+ ret = kernel_setsockopt(sock, SOL_SOCKET, SO_REUSEADDR,
+ (char *)&opt, sizeof(opt));
+ if (ret < 0) {
+ printk(KERN_ERR "kernel_setsockopt() for SO_REUSEADDR"
+ " failed\n");
+ goto fail;
+ }
+
+ ret = kernel_bind(sock, (struct sockaddr *)&np->np_sockaddr, len);
+ if (ret < 0) {
+ printk(KERN_ERR "kernel_bind() failed: %d\n", ret);
+ goto fail;
+ }
+
+ ret = kernel_listen(sock, backlog);
+ if (ret != 0) {
+ printk(KERN_ERR "kernel_listen() failed: %d\n", ret);
+ goto fail;
+ }
+
+ return 0;
+
+fail:
+ np->np_socket = NULL;
+ if (sock) {
+ if (np->np_flags & NPF_SCTP_STRUCT_FILE) {
+ kfree(sock->file);
+ sock->file = NULL;
+ }
+
+ sock_release(sock);
+ }
+ return ret;
+}
+
+static int __iscsi_target_login_thread(struct iscsi_np *np)
+{
+ u8 buffer[ISCSI_HDR_LEN], iscsi_opcode, zero_tsih = 0;
+ int err, ret = 0, ip_proto, sock_type, set_sctp_conn_flag;
+ struct iscsi_conn *conn = NULL;
+ struct iscsi_login *login;
+ struct iscsi_portal_group *tpg = NULL;
+ struct socket *new_sock, *sock;
+ struct kvec iov;
+ struct iscsi_login_req *pdu;
+ struct sockaddr_in sock_in;
+ struct sockaddr_in6 sock_in6;
+
+ flush_signals(current);
+ set_sctp_conn_flag = 0;
+ sock = np->np_socket;
+ ip_proto = np->np_ip_proto;
+ sock_type = np->np_sock_type;
+
+ spin_lock_bh(&np->np_thread_lock);
+ if (np->np_thread_state == ISCSI_NP_THREAD_RESET) {
+ np->np_thread_state = ISCSI_NP_THREAD_ACTIVE;
+ complete(&np->np_restart_comp);
+ } else {
+ np->np_thread_state = ISCSI_NP_THREAD_ACTIVE;
+ }
+ spin_unlock_bh(&np->np_thread_lock);
+
+ if (kernel_accept(sock, &new_sock, 0) < 0) {
+ spin_lock_bh(&np->np_thread_lock);
+ if (np->np_thread_state == ISCSI_NP_THREAD_RESET) {
+ spin_unlock_bh(&np->np_thread_lock);
+ complete(&np->np_restart_comp);
+ /* Get another socket */
+ return 1;
+ }
+ spin_unlock_bh(&np->np_thread_lock);
+ goto out;
+ }
+ /*
+ * The SCTP stack needs struct socket->file.
+ */
+ if ((np->np_network_transport == ISCSI_SCTP_TCP) ||
+ (np->np_network_transport == ISCSI_SCTP_UDP)) {
+ if (!new_sock->file) {
+ new_sock->file = kzalloc(
+ sizeof(struct file), GFP_KERNEL);
+ if (!new_sock->file) {
+ printk(KERN_ERR "Unable to allocate struct"
+ " file for SCTP\n");
+ sock_release(new_sock);
+ /* Get another socket */
+ return 1;
+ }
+ set_sctp_conn_flag = 1;
+ }
+ }
+
+ iscsi_start_login_thread_timer(np);
+
+ conn = kzalloc(sizeof(struct iscsi_conn), GFP_KERNEL);
+ if (!conn) {
+ printk(KERN_ERR "Could not allocate memory for"
+ " new connection\n");
+ if (set_sctp_conn_flag) {
+ kfree(new_sock->file);
+ new_sock->file = NULL;
+ }
+ sock_release(new_sock);
+ /* Get another socket */
+ return 1;
+ }
+
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_FREE.\n");
+ conn->conn_state = TARG_CONN_STATE_FREE;
+ conn->sock = new_sock;
+
+ if (set_sctp_conn_flag)
+ conn->conn_flags |= CONNFLAG_SCTP_STRUCT_FILE;
+
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_XPT_UP.\n");
+ conn->conn_state = TARG_CONN_STATE_XPT_UP;
+
+ /*
+ * Allocate conn->conn_ops early as a failure calling
+ * iscsit_tx_login_rsp() below will call tx_data().
+ */
+ conn->conn_ops = kzalloc(sizeof(struct iscsi_conn_ops), GFP_KERNEL);
+ if (!conn->conn_ops) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_conn_ops.\n");
+ goto new_sess_out;
+ }
+ /*
+ * Perform the remaining iSCSI connection initialization items..
+ */
+ if (iscsi_login_init_conn(conn) < 0)
+ goto new_sess_out;
+
+ memset(buffer, 0, ISCSI_HDR_LEN);
+ memset(&iov, 0, sizeof(struct kvec));
+ iov.iov_base = buffer;
+ iov.iov_len = ISCSI_HDR_LEN;
+
+ if (rx_data(conn, &iov, 1, ISCSI_HDR_LEN) <= 0) {
+ printk(KERN_ERR "rx_data() returned an error.\n");
+ goto new_sess_out;
+ }
+
+ iscsi_opcode = (buffer[0] & ISCSI_OPCODE_MASK);
+ if (!(iscsi_opcode & ISCSI_OP_LOGIN)) {
+ printk(KERN_ERR "First opcode is not login request,"
+ " failing login request.\n");
+ goto new_sess_out;
+ }
+
+ pdu = (struct iscsi_login_req *) buffer;
+ pdu->cid = be16_to_cpu(pdu->cid);
+ pdu->tsih = be16_to_cpu(pdu->tsih);
+ pdu->itt = be32_to_cpu(pdu->itt);
+ pdu->cmdsn = be32_to_cpu(pdu->cmdsn);
+ pdu->exp_statsn = be32_to_cpu(pdu->exp_statsn);
+ /*
+ * Used by iscsit_tx_login_rsp() for Login Resonses PDUs
+ * when Status-Class != 0.
+ */
+ conn->login_itt = pdu->itt;
+
+ spin_lock_bh(&np->np_thread_lock);
+ if (np->np_thread_state != ISCSI_NP_THREAD_ACTIVE) {
+ spin_unlock_bh(&np->np_thread_lock);
+ printk(KERN_ERR "iSCSI Network Portal on %s:%hu currently not"
+ " active.\n", np->np_ip, np->np_port);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_SVC_UNAVAILABLE);
+ goto new_sess_out;
+ }
+ spin_unlock_bh(&np->np_thread_lock);
+
+ if (np->np_sockaddr.ss_family == AF_INET6) {
+ memset(&sock_in6, 0, sizeof(struct sockaddr_in6));
+
+ if (conn->sock->ops->getname(conn->sock,
+ (struct sockaddr *)&sock_in6, &err, 1) < 0) {
+ printk(KERN_ERR "sock_ops->getname() failed.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_TARGET_ERROR);
+ goto new_sess_out;
+ }
+#if 0
+ if (!(iscsi_ntop6((const unsigned char *)
+ &sock_in6.sin6_addr.in6_u,
+ (char *)&conn->ipv6_login_ip[0],
+ IPV6_ADDRESS_SPACE))) {
+ printk(KERN_ERR "iscsi_ntop6() failed\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_TARGET_ERROR);
+ goto new_sess_out;
+ }
+#else
+ printk(KERN_INFO "Skipping iscsi_ntop6()\n");
+#endif
+ } else {
+ u32 ipv4;
+ memset(&sock_in, 0, sizeof(struct sockaddr_in));
+
+ if (conn->sock->ops->getname(conn->sock,
+ (struct sockaddr *)&sock_in, &err, 1) < 0) {
+ printk(KERN_ERR "sock_ops->getname() failed.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_TARGET_ERROR);
+ goto new_sess_out;
+ }
+ ipv4 = ntohl(sock_in.sin_addr.s_addr);
+ sprintf(conn->login_ip, "%u.%u.%u.%u", ((ipv4 >> 24) & 0xff),
+ ((ipv4 >> 16) & 0xff), ((ipv4 >> 8) & 0xff),
+ ipv4 & 0xff);
+ conn->login_ipv4 = ipv4;
+ conn->login_port = ntohs(sock_in.sin_port);
+ }
+
+ conn->network_transport = np->np_network_transport;
+ snprintf(conn->net_dev, ISCSI_NETDEV_NAME_SIZE, "%s", np->np_net_dev);
+
+ conn->local_ip = np->np_ipv4;
+ conn->local_port = np->np_port;
+
+ printk(KERN_INFO "Received iSCSI login request from %s on %s Network"
+ " Portal %s:%hu\n", conn->login_ip,
+ (conn->network_transport == ISCSI_TCP) ? "TCP" : "SCTP",
+ np->np_ip, np->np_port);
+
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_IN_LOGIN.\n");
+ conn->conn_state = TARG_CONN_STATE_IN_LOGIN;
+
+ if (iscsi_login_check_initiator_version(conn, pdu->max_version,
+ pdu->min_version) < 0)
+ goto new_sess_out;
+
+ zero_tsih = (pdu->tsih == 0x0000);
+ if ((zero_tsih)) {
+ /*
+ * This is the leading connection of a new session.
+ * We wait until after authentication to check for
+ * session reinstatement.
+ */
+ if (iscsi_login_zero_tsih_s1(conn, buffer) < 0)
+ goto new_sess_out;
+ } else {
+ /*
+ * Add a new connection to an existing session.
+ * We check for a non-existant session in
+ * iscsi_login_non_zero_tsih_s2() below based
+ * on ISID/TSIH, but wait until after authentication
+ * to check for connection reinstatement, etc.
+ */
+ if (iscsi_login_non_zero_tsih_s1(conn, buffer) < 0)
+ goto new_sess_out;
+ }
+
+ /*
+ * This will process the first login request, and call
+ * iscsi_target_locate_portal(), and return a valid struct iscsi_login.
+ */
+ login = iscsi_target_init_negotiation(np, conn, buffer);
+ if (!login) {
+ tpg = conn->tpg;
+ goto new_sess_out;
+ }
+
+ tpg = conn->tpg;
+ if (!tpg) {
+ printk(KERN_ERR "Unable to locate struct iscsi_conn->tpg\n");
+ goto new_sess_out;
+ }
+
+ if (zero_tsih) {
+ if (iscsi_login_zero_tsih_s2(conn) < 0) {
+ iscsi_target_nego_release(login, conn);
+ goto new_sess_out;
+ }
+ } else {
+ if (iscsi_login_non_zero_tsih_s2(conn, buffer) < 0) {
+ iscsi_target_nego_release(login, conn);
+ goto old_sess_out;
+ }
+ }
+
+ if (iscsi_target_start_negotiation(login, conn) < 0)
+ goto new_sess_out;
+
+ if (!conn->sess) {
+ printk(KERN_ERR "struct iscsi_conn session pointer is NULL!\n");
+ goto new_sess_out;
+ }
+
+ iscsi_stop_login_thread_timer(np);
+
+ if (signal_pending(current))
+ goto new_sess_out;
+
+ ret = iscsi_post_login_handler(np, conn, zero_tsih);
+
+ if (ret < 0)
+ goto new_sess_out;
+
+ iscsit_deaccess_np(np, tpg);
+ tpg = NULL;
+ /* Get another socket */
+ return 1;
+
+new_sess_out:
+ printk(KERN_ERR "iSCSI Login negotiation failed.\n");
+ iscsit_collect_login_stats(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ if (!zero_tsih || !conn->sess)
+ goto old_sess_out;
+ if (conn->sess->se_sess)
+ transport_free_session(conn->sess->se_sess);
+ if (conn->sess->session_index != 0) {
+ spin_lock_bh(&sess_idr_lock);
+ idr_remove(&sess_idr, conn->sess->session_index);
+ spin_unlock_bh(&sess_idr_lock);
+ }
+ if (conn->sess->sess_ops)
+ kfree(conn->sess->sess_ops);
+ if (conn->sess)
+ kfree(conn->sess);
+old_sess_out:
+ iscsi_stop_login_thread_timer(np);
+ /*
+ * If login negotiation fails check if the Time2Retain timer
+ * needs to be restarted.
+ */
+ if (!zero_tsih && conn->sess) {
+ spin_lock_bh(&conn->sess->conn_lock);
+ if (conn->sess->session_state == TARG_SESS_STATE_FAILED) {
+ struct se_portal_group *se_tpg =
+ &ISCSI_TPG_C(conn)->tpg_se_tpg;
+
+ atomic_set(&conn->sess->session_continuation, 0);
+ spin_unlock_bh(&conn->sess->conn_lock);
+ spin_lock_bh(&se_tpg->session_lock);
+ iscsit_start_time2retain_handler(conn->sess);
+ spin_unlock_bh(&se_tpg->session_lock);
+ } else
+ spin_unlock_bh(&conn->sess->conn_lock);
+ iscsit_dec_session_usage_count(conn->sess);
+ }
+
+ if (!IS_ERR(conn->conn_rx_hash.tfm))
+ crypto_free_hash(conn->conn_rx_hash.tfm);
+ if (!IS_ERR(conn->conn_tx_hash.tfm))
+ crypto_free_hash(conn->conn_tx_hash.tfm);
+
+ if (conn->conn_cpumask)
+ free_cpumask_var(conn->conn_cpumask);
+
+ kfree(conn->conn_ops);
+
+ if (conn->param_list) {
+ iscsi_release_param_list(conn->param_list);
+ conn->param_list = NULL;
+ }
+ if (conn->sock) {
+ if (conn->conn_flags & CONNFLAG_SCTP_STRUCT_FILE) {
+ kfree(conn->sock->file);
+ conn->sock->file = NULL;
+ }
+ sock_release(conn->sock);
+ }
+ kfree(conn);
+
+ if (tpg) {
+ iscsit_deaccess_np(np, tpg);
+ tpg = NULL;
+ }
+
+out:
+ /* Wait for another socket.. */
+ if (!signal_pending(current) || !kthread_should_stop())
+ return 1;
+
+ iscsi_stop_login_thread_timer(np);
+ spin_lock_bh(&np->np_thread_lock);
+ np->np_thread_state = ISCSI_NP_THREAD_EXIT;
+ spin_unlock_bh(&np->np_thread_lock);
+ return 0;
+}
+
+int iscsi_target_login_thread(void *arg)
+{
+ struct iscsi_np *np = (struct iscsi_np *)arg;
+ int ret;
+
+ set_user_nice(current, -20);
+ allow_signal(SIGINT);
+
+ while (!kthread_should_stop()) {
+ ret = __iscsi_target_login_thread(np);
+ /*
+ * We break and exit here unless another sock_accept() call
+ * is expected.
+ */
+ if (ret != 1)
+ break;
+ }
+
+ return 0;
+}
diff --git a/drivers/target/iscsi/iscsi_target_login.h b/drivers/target/iscsi/iscsi_target_login.h
new file mode 100644
index 0000000..04f7a41
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_login.h
@@ -0,0 +1,13 @@
+#ifndef ISCSI_TARGET_LOGIN_H
+#define ISCSI_TARGET_LOGIN_H
+
+extern int iscsi_login_setup_crypto(struct iscsi_conn *);
+extern int iscsi_check_for_session_reinstatement(struct iscsi_conn *);
+extern int iscsi_login_post_auth_non_zero_tsih(struct iscsi_conn *, u16, u32);
+extern int iscsi_target_setup_login_socket(struct iscsi_np *, int);
+extern int iscsi_target_login_thread(void *);
+extern int iscsi_login_disable_FIM_keys(struct iscsi_param_list *, struct iscsi_conn *);
+
+extern struct iscsi_global *iscsi_global;
+
+#endif /*** ISCSI_TARGET_LOGIN_H ***/
diff --git a/drivers/target/iscsi/iscsi_target_nego.c b/drivers/target/iscsi/iscsi_target_nego.c
new file mode 100644
index 0000000..39b37ce
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_nego.c
@@ -0,0 +1,1068 @@
+/*******************************************************************************
+ * This file contains main functions related to iSCSI Parameter negotiation.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/ctype.h>
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_tpg.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_parameters.h"
+#include "iscsi_target_login.h"
+#include "iscsi_target_nego.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target.h"
+#include "iscsi_target_auth.h"
+
+#define MAX_LOGIN_PDUS 7
+#define TEXT_LEN 4096
+
+void convert_null_to_semi(char *buf, int len)
+{
+ int i;
+
+ for (i = 0; i < len; i++)
+ if (buf[i] == '\0')
+ buf[i] = ';';
+}
+
+int strlen_semi(char *buf)
+{
+ int i = 0;
+
+ while (buf[i] != '\0') {
+ if (buf[i] == ';')
+ return i;
+ i++;
+ }
+
+ return -1;
+}
+
+int extract_param(
+ const char *in_buf,
+ const char *pattern,
+ unsigned int max_length,
+ char *out_buf,
+ unsigned char *type)
+{
+ char *ptr;
+ int len;
+
+ if (!in_buf || !pattern || !out_buf || !type)
+ return -1;
+
+ ptr = strstr(in_buf, pattern);
+ if (!ptr)
+ return -1;
+
+ ptr = strstr(ptr, "=");
+ if (!ptr)
+ return -1;
+
+ ptr += 1;
+ if (*ptr == '0' && (*(ptr+1) == 'x' || *(ptr+1) == 'X')) {
+ ptr += 2; /* skip 0x */
+ *type = HEX;
+ } else
+ *type = DECIMAL;
+
+ len = strlen_semi(ptr);
+ if (len < 0)
+ return -1;
+
+ if (len > max_length) {
+ printk(KERN_ERR "Length of input: %d exeeds max_length:"
+ " %d\n", len, max_length);
+ return -1;
+ }
+ memcpy(out_buf, ptr, len);
+ out_buf[len] = '\0';
+
+ return 0;
+}
+
+static u32 iscsi_handle_authentication(
+ struct iscsi_conn *conn,
+ char *in_buf,
+ char *out_buf,
+ int in_length,
+ int *out_length,
+ unsigned char *authtype)
+{
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_auth *auth;
+ struct iscsi_node_acl *iscsi_nacl;
+ struct se_node_acl *se_nacl;
+
+ if (!sess->sess_ops->SessionType) {
+ /*
+ * For SessionType=Normal
+ */
+ se_nacl = conn->sess->se_sess->se_node_acl;
+ if (!se_nacl) {
+ printk(KERN_ERR "Unable to locate struct se_node_acl for"
+ " CHAP auth\n");
+ return -1;
+ }
+ iscsi_nacl = container_of(se_nacl, struct iscsi_node_acl,
+ se_node_acl);
+ if (!iscsi_nacl) {
+ printk(KERN_ERR "Unable to locate struct iscsi_node_acl for"
+ " CHAP auth\n");
+ return -1;
+ }
+
+ auth = ISCSI_NODE_AUTH(iscsi_nacl);
+ } else {
+ /*
+ * For SessionType=Discovery
+ */
+ auth = &iscsi_global->discovery_acl.node_auth;
+ }
+
+ if (strstr("CHAP", authtype))
+ strcpy(conn->sess->auth_type, "CHAP");
+ else
+ strcpy(conn->sess->auth_type, NONE);
+
+ if (strstr("None", authtype))
+ return 1;
+#ifdef CANSRP
+ else if (strstr("SRP", authtype))
+ return srp_main_loop(conn, auth, in_buf, out_buf,
+ &in_length, out_length);
+#endif
+ else if (strstr("CHAP", authtype))
+ return chap_main_loop(conn, auth, in_buf, out_buf,
+ &in_length, out_length);
+ else if (strstr("SPKM1", authtype))
+ return 2;
+ else if (strstr("SPKM2", authtype))
+ return 2;
+ else if (strstr("KRB5", authtype))
+ return 2;
+ else
+ return 2;
+}
+
+static void iscsi_remove_failed_auth_entry(struct iscsi_conn *conn)
+{
+ kfree(conn->auth_protocol);
+}
+
+static int iscsi_target_check_login_request(
+ struct iscsi_conn *conn,
+ struct iscsi_login *login)
+{
+ int req_csg, req_nsg, rsp_csg, rsp_nsg;
+ u32 payload_length;
+ struct iscsi_login_req *login_req;
+ struct iscsi_login_rsp *login_rsp;
+
+ login_req = (struct iscsi_login_req *) login->req;
+ login_rsp = (struct iscsi_login_rsp *) login->rsp;
+ payload_length = ntoh24(login_req->dlength);
+
+ switch (login_req->opcode & ISCSI_OPCODE_MASK) {
+ case ISCSI_OP_LOGIN:
+ break;
+ default:
+ printk(KERN_ERR "Received unknown opcode 0x%02x.\n",
+ login_req->opcode & ISCSI_OPCODE_MASK);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ return -1;
+ }
+
+ if ((login_req->flags & ISCSI_FLAG_LOGIN_CONTINUE) &&
+ (login_req->flags & ISCSI_FLAG_LOGIN_TRANSIT)) {
+ printk(KERN_ERR "Login request has both ISCSI_FLAG_LOGIN_CONTINUE"
+ " and ISCSI_FLAG_LOGIN_TRANSIT set, protocol error.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ return -1;
+ }
+
+ req_csg = (login_req->flags & ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK) >> 2;
+ rsp_csg = (login_rsp->flags & ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK) >> 2;
+ req_nsg = (login_req->flags & ISCSI_FLAG_LOGIN_NEXT_STAGE_MASK);
+ rsp_nsg = (login_rsp->flags & ISCSI_FLAG_LOGIN_NEXT_STAGE_MASK);
+
+ if (req_csg != login->current_stage) {
+ printk(KERN_ERR "Initiator unexpectedly changed login stage"
+ " from %d to %d, login failed.\n", login->current_stage,
+ req_csg);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ return -1;
+ }
+
+ if ((req_nsg == 2) || (req_csg >= 2) ||
+ ((login_req->flags & ISCSI_FLAG_LOGIN_TRANSIT) &&
+ (req_nsg <= req_csg))) {
+ printk(KERN_ERR "Illegal login_req->flags Combination, CSG: %d,"
+ " NSG: %d, ISCSI_FLAG_LOGIN_TRANSIT: %d.\n", req_csg,
+ req_nsg, (login_req->flags & ISCSI_FLAG_LOGIN_TRANSIT));
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ return -1;
+ }
+
+ if ((login_req->max_version != login->version_max) ||
+ (login_req->min_version != login->version_min)) {
+ printk(KERN_ERR "Login request changed Version Max/Nin"
+ " unexpectedly to 0x%02x/0x%02x, protocol error\n",
+ login_req->max_version, login_req->min_version);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ return -1;
+ }
+
+ if (memcmp(login_req->isid, login->isid, 6) != 0) {
+ printk(KERN_ERR "Login request changed ISID unexpectedly,"
+ " protocol error.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ return -1;
+ }
+
+ if (login_req->itt != login->init_task_tag) {
+ printk(KERN_ERR "Login request changed ITT unexpectedly to"
+ " 0x%08x, protocol error.\n", login_req->itt);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ return -1;
+ }
+
+ if (payload_length > MAX_KEY_VALUE_PAIRS) {
+ printk(KERN_ERR "Login request payload exceeds default"
+ " MaxRecvDataSegmentLength: %u, protocol error.\n",
+ MAX_KEY_VALUE_PAIRS);
+ return -1;
+ }
+
+ return 0;
+}
+
+static int iscsi_target_check_first_request(
+ struct iscsi_conn *conn,
+ struct iscsi_login *login)
+{
+ struct iscsi_param *param = NULL;
+ struct se_node_acl *se_nacl;
+
+ login->first_request = 0;
+
+ list_for_each_entry(param, &conn->param_list->param_list, p_list) {
+ if (!strncmp(param->name, SESSIONTYPE, 11)) {
+ if (!IS_PSTATE_ACCEPTOR(param)) {
+ printk(KERN_ERR "SessionType key not received"
+ " in first login request.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_MISSING_FIELDS);
+ return -1;
+ }
+ if (!strncmp(param->value, DISCOVERY, 9))
+ return 0;
+ }
+
+ if (!strncmp(param->name, INITIATORNAME, 13)) {
+ if (!IS_PSTATE_ACCEPTOR(param)) {
+ if (!login->leading_connection)
+ continue;
+
+ printk(KERN_ERR "InitiatorName key not received"
+ " in first login request.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_MISSING_FIELDS);
+ return -1;
+ }
+
+ /*
+ * For non-leading connections, double check that the
+ * received InitiatorName matches the existing session's
+ * struct iscsi_node_acl.
+ */
+ if (!login->leading_connection) {
+ se_nacl = conn->sess->se_sess->se_node_acl;
+ if (!se_nacl) {
+ printk(KERN_ERR "Unable to locate"
+ " struct se_node_acl\n");
+ iscsit_tx_login_rsp(conn,
+ ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_TGT_NOT_FOUND);
+ return -1;
+ }
+
+ if (strcmp(param->value,
+ se_nacl->initiatorname)) {
+ printk(KERN_ERR "Incorrect"
+ " InitiatorName: %s for this"
+ " iSCSI Initiator Node.\n",
+ param->value);
+ iscsit_tx_login_rsp(conn,
+ ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_TGT_NOT_FOUND);
+ return -1;
+ }
+ }
+ }
+ }
+
+ return 0;
+}
+
+static int iscsi_target_do_tx_login_io(struct iscsi_conn *conn, struct iscsi_login *login)
+{
+ u32 padding = 0;
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_login_rsp *login_rsp;
+
+ login_rsp = (struct iscsi_login_rsp *) login->rsp;
+
+ login_rsp->opcode = ISCSI_OP_LOGIN_RSP;
+ hton24(login_rsp->dlength, login->rsp_length);
+ memcpy(login_rsp->isid, login->isid, 6);
+ login_rsp->tsih = cpu_to_be16(login->tsih);
+ login_rsp->itt = cpu_to_be32(login->init_task_tag);
+ login_rsp->statsn = cpu_to_be32(conn->stat_sn++);
+ login_rsp->exp_cmdsn = cpu_to_be32(conn->sess->exp_cmd_sn);
+ login_rsp->max_cmdsn = cpu_to_be32(conn->sess->max_cmd_sn);
+
+ TRACE(TRACE_LOGIN, "Sending Login Response, Flags: 0x%02x, ITT: 0x%08x,"
+ " ExpCmdSN; 0x%08x, MaxCmdSN: 0x%08x, StatSN: 0x%08x, Length:"
+ " %u\n", login_rsp->flags, ntohl(login_rsp->itt),
+ ntohl(login_rsp->exp_cmdsn), ntohl(login_rsp->max_cmdsn),
+ ntohl(login_rsp->statsn), login->rsp_length);
+
+ padding = ((-login->rsp_length) & 3);
+
+ if (iscsi_login_tx_data(
+ conn,
+ login->rsp,
+ login->rsp_buf,
+ login->rsp_length + padding) < 0)
+ return -1;
+
+ login->rsp_length = 0;
+ login_rsp->tsih = be16_to_cpu(login_rsp->tsih);
+ login_rsp->itt = be32_to_cpu(login_rsp->itt);
+ login_rsp->statsn = be32_to_cpu(login_rsp->statsn);
+ spin_lock(&sess->cmdsn_lock);
+ login_rsp->exp_cmdsn = be32_to_cpu(sess->exp_cmd_sn);
+ login_rsp->max_cmdsn = be32_to_cpu(sess->max_cmd_sn);
+ spin_unlock(&sess->cmdsn_lock);
+
+ return 0;
+}
+
+static int iscsi_target_do_rx_login_io(struct iscsi_conn *conn, struct iscsi_login *login)
+{
+ u32 padding = 0, payload_length;
+ struct iscsi_login_req *login_req;
+
+ if (iscsi_login_rx_data(conn, login->req, ISCSI_HDR_LEN) < 0)
+ return -1;
+
+ login_req = (struct iscsi_login_req *) login->req;
+ payload_length = ntoh24(login_req->dlength);
+ login_req->tsih = be16_to_cpu(login_req->tsih);
+ login_req->itt = be32_to_cpu(login_req->itt);
+ login_req->cid = be16_to_cpu(login_req->cid);
+ login_req->cmdsn = be32_to_cpu(login_req->cmdsn);
+ login_req->exp_statsn = be32_to_cpu(login_req->exp_statsn);
+
+ TRACE(TRACE_LOGIN, "Got Login Command, Flags 0x%02x, ITT: 0x%08x,"
+ " CmdSN: 0x%08x, ExpStatSN: 0x%08x, CID: %hu, Length: %u\n",
+ login_req->flags, login_req->itt, login_req->cmdsn,
+ login_req->exp_statsn, login_req->cid, payload_length);
+
+ if (iscsi_target_check_login_request(conn, login) < 0)
+ return -1;
+
+ padding = ((-payload_length) & 3);
+ memset(login->req_buf, 0, MAX_KEY_VALUE_PAIRS);
+
+ if (iscsi_login_rx_data(
+ conn,
+ login->req_buf,
+ payload_length + padding) < 0)
+ return -1;
+
+ return 0;
+}
+
+static int iscsi_target_do_login_io(struct iscsi_conn *conn, struct iscsi_login *login)
+{
+ if (iscsi_target_do_tx_login_io(conn, login) < 0)
+ return -1;
+
+ if (iscsi_target_do_rx_login_io(conn, login) < 0)
+ return -1;
+
+ return 0;
+}
+
+static int iscsi_target_get_initial_payload(
+ struct iscsi_conn *conn,
+ struct iscsi_login *login)
+{
+ u32 padding = 0, payload_length;
+ struct iscsi_login_req *login_req;
+
+ login_req = (struct iscsi_login_req *) login->req;
+ payload_length = ntoh24(login_req->dlength);
+
+ TRACE(TRACE_LOGIN, "Got Login Command, Flags 0x%02x, ITT: 0x%08x,"
+ " CmdSN: 0x%08x, ExpStatSN: 0x%08x, Length: %u\n",
+ login_req->flags, login_req->itt, login_req->cmdsn,
+ login_req->exp_statsn, payload_length);
+
+ if (iscsi_target_check_login_request(conn, login) < 0)
+ return -1;
+
+ padding = ((-payload_length) & 3);
+
+ if (iscsi_login_rx_data(
+ conn,
+ login->req_buf,
+ payload_length + padding) < 0)
+ return -1;
+
+ return 0;
+}
+
+/*
+ * NOTE: We check for existing sessions or connections AFTER the initiator
+ * has been successfully authenticated in order to protect against faked
+ * ISID/TSIH combinations.
+ */
+static int iscsi_target_check_for_existing_instances(
+ struct iscsi_conn *conn,
+ struct iscsi_login *login)
+{
+ if (login->checked_for_existing)
+ return 0;
+
+ login->checked_for_existing = 1;
+
+ if (!login->tsih)
+ return iscsi_check_for_session_reinstatement(conn);
+ else
+ return iscsi_login_post_auth_non_zero_tsih(conn, login->cid,
+ login->initial_exp_statsn);
+}
+
+static int iscsi_target_do_authentication(
+ struct iscsi_conn *conn,
+ struct iscsi_login *login)
+{
+ int authret;
+ u32 payload_length;
+ struct iscsi_param *param;
+ struct iscsi_login_req *login_req;
+ struct iscsi_login_rsp *login_rsp;
+
+ login_req = (struct iscsi_login_req *) login->req;
+ login_rsp = (struct iscsi_login_rsp *) login->rsp;
+ payload_length = ntoh24(login_req->dlength);
+
+ param = iscsi_find_param_from_key(AUTHMETHOD, conn->param_list);
+ if (!param)
+ return -1;
+
+ authret = iscsi_handle_authentication(
+ conn,
+ login->req_buf,
+ login->rsp_buf,
+ payload_length,
+ &login->rsp_length,
+ param->value);
+ switch (authret) {
+ case 0:
+ printk(KERN_INFO "Received OK response"
+ " from LIO Authentication, continuing.\n");
+ break;
+ case 1:
+ printk(KERN_INFO "iSCSI security negotiation"
+ " completed sucessfully.\n");
+ login->auth_complete = 1;
+ if ((login_req->flags & ISCSI_FLAG_LOGIN_NEXT_STAGE1) &&
+ (login_req->flags & ISCSI_FLAG_LOGIN_TRANSIT)) {
+ login_rsp->flags |= (ISCSI_FLAG_LOGIN_NEXT_STAGE1 |
+ ISCSI_FLAG_LOGIN_TRANSIT);
+ login->current_stage = 1;
+ }
+ return iscsi_target_check_for_existing_instances(
+ conn, login);
+ case 2:
+ printk(KERN_ERR "Security negotiation"
+ " failed.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_AUTH_FAILED);
+ return -1;
+ default:
+ printk(KERN_ERR "Received unknown error %d from LIO"
+ " Authentication\n", authret);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_TARGET_ERROR);
+ return -1;
+ }
+
+ return 0;
+}
+
+static int iscsi_target_handle_csg_zero(
+ struct iscsi_conn *conn,
+ struct iscsi_login *login)
+{
+ int ret;
+ u32 payload_length;
+ struct iscsi_param *param;
+ struct iscsi_login_req *login_req;
+ struct iscsi_login_rsp *login_rsp;
+
+ login_req = (struct iscsi_login_req *) login->req;
+ login_rsp = (struct iscsi_login_rsp *) login->rsp;
+ payload_length = ntoh24(login_req->dlength);
+
+ param = iscsi_find_param_from_key(AUTHMETHOD, conn->param_list);
+ if (!param)
+ return -1;
+
+ ret = iscsi_decode_text_input(
+ PHASE_SECURITY|PHASE_DECLARATIVE,
+ SENDER_INITIATOR|SENDER_RECEIVER,
+ login->req_buf,
+ payload_length,
+ conn->param_list);
+ if (ret < 0)
+ return -1;
+
+ if (ret > 0) {
+ if (login->auth_complete) {
+ printk(KERN_ERR "Initiator has already been"
+ " successfully authenticated, but is still"
+ " sending %s keys.\n", param->value);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_INIT_ERR);
+ return -1;
+ }
+
+ goto do_auth;
+ }
+
+ if (login->first_request)
+ if (iscsi_target_check_first_request(conn, login) < 0)
+ return -1;
+
+ ret = iscsi_encode_text_output(
+ PHASE_SECURITY|PHASE_DECLARATIVE,
+ SENDER_TARGET,
+ login->rsp_buf,
+ &login->rsp_length,
+ conn->param_list);
+ if (ret < 0)
+ return -1;
+
+ if (!iscsi_check_negotiated_keys(conn->param_list)) {
+ if (ISCSI_TPG_ATTRIB(ISCSI_TPG_C(conn))->authentication &&
+ !strncmp(param->value, NONE, 4)) {
+ printk(KERN_ERR "Initiator sent AuthMethod=None but"
+ " Target is enforcing iSCSI Authentication,"
+ " login failed.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_AUTH_FAILED);
+ return -1;
+ }
+
+ if (ISCSI_TPG_ATTRIB(ISCSI_TPG_C(conn))->authentication &&
+ !login->auth_complete)
+ return 0;
+
+ if (strncmp(param->value, NONE, 4) && !login->auth_complete)
+ return 0;
+
+ if ((login_req->flags & ISCSI_FLAG_LOGIN_NEXT_STAGE1) &&
+ (login_req->flags & ISCSI_FLAG_LOGIN_TRANSIT)) {
+ login_rsp->flags |= ISCSI_FLAG_LOGIN_NEXT_STAGE1 |
+ ISCSI_FLAG_LOGIN_TRANSIT;
+ login->current_stage = 1;
+ }
+ }
+
+ return 0;
+do_auth:
+ return iscsi_target_do_authentication(conn, login);
+}
+
+static int iscsi_target_handle_csg_one(struct iscsi_conn *conn, struct iscsi_login *login)
+{
+ int ret;
+ u32 payload_length;
+ struct iscsi_login_req *login_req;
+ struct iscsi_login_rsp *login_rsp;
+
+ login_req = (struct iscsi_login_req *) login->req;
+ login_rsp = (struct iscsi_login_rsp *) login->rsp;
+ payload_length = ntoh24(login_req->dlength);
+
+ ret = iscsi_decode_text_input(
+ PHASE_OPERATIONAL|PHASE_DECLARATIVE,
+ SENDER_INITIATOR|SENDER_RECEIVER,
+ login->req_buf,
+ payload_length,
+ conn->param_list);
+ if (ret < 0)
+ return -1;
+
+ if (login->first_request)
+ if (iscsi_target_check_first_request(conn, login) < 0)
+ return -1;
+
+ if (iscsi_target_check_for_existing_instances(conn, login) < 0)
+ return -1;
+
+ ret = iscsi_encode_text_output(
+ PHASE_OPERATIONAL|PHASE_DECLARATIVE,
+ SENDER_TARGET,
+ login->rsp_buf,
+ &login->rsp_length,
+ conn->param_list);
+ if (ret < 0)
+ return -1;
+
+ if (!login->auth_complete &&
+ ISCSI_TPG_ATTRIB(ISCSI_TPG_C(conn))->authentication) {
+ printk(KERN_ERR "Initiator is requesting CSG: 1, has not been"
+ " successfully authenticated, and the Target is"
+ " enforcing iSCSI Authentication, login failed.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_AUTH_FAILED);
+ return -1;
+ }
+
+ if (!iscsi_check_negotiated_keys(conn->param_list))
+ if ((login_req->flags & ISCSI_FLAG_LOGIN_NEXT_STAGE3) &&
+ (login_req->flags & ISCSI_FLAG_LOGIN_TRANSIT))
+ login_rsp->flags |= ISCSI_FLAG_LOGIN_NEXT_STAGE3 |
+ ISCSI_FLAG_LOGIN_TRANSIT;
+
+ return 0;
+}
+
+static int iscsi_target_do_login(struct iscsi_conn *conn, struct iscsi_login *login)
+{
+ int pdu_count = 0;
+ struct iscsi_login_req *login_req;
+ struct iscsi_login_rsp *login_rsp;
+
+ login_req = (struct iscsi_login_req *) login->req;
+ login_rsp = (struct iscsi_login_rsp *) login->rsp;
+
+ while (1) {
+ if (++pdu_count > MAX_LOGIN_PDUS) {
+ printk(KERN_ERR "MAX_LOGIN_PDUS count reached.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_TARGET_ERROR);
+ return -1;
+ }
+
+ switch ((login_req->flags & ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK) >> 2) {
+ case 0:
+ login_rsp->flags |= (0 & ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK);
+ if (iscsi_target_handle_csg_zero(conn, login) < 0)
+ return -1;
+ break;
+ case 1:
+ login_rsp->flags |= ISCSI_FLAG_LOGIN_CURRENT_STAGE1;
+ if (iscsi_target_handle_csg_one(conn, login) < 0)
+ return -1;
+ if (login_rsp->flags & ISCSI_FLAG_LOGIN_TRANSIT) {
+ login->tsih = conn->sess->tsih;
+ if (iscsi_target_do_tx_login_io(conn,
+ login) < 0)
+ return -1;
+ return 0;
+ }
+ break;
+ default:
+ printk(KERN_ERR "Illegal CSG: %d received from"
+ " Initiator, protocol error.\n",
+ (login_req->flags & ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK)
+ >> 2);
+ break;
+ }
+
+ if (iscsi_target_do_login_io(conn, login) < 0)
+ return -1;
+
+ if (login_rsp->flags & ISCSI_FLAG_LOGIN_TRANSIT) {
+ login_rsp->flags &= ~ISCSI_FLAG_LOGIN_TRANSIT;
+ login_rsp->flags &= ~ISCSI_FLAG_LOGIN_NEXT_STAGE_MASK;
+ }
+ }
+
+ return 0;
+}
+
+static void iscsi_initiatorname_tolower(
+ char *param_buf)
+{
+ char *c;
+ u32 iqn_size = strlen(param_buf), i;
+
+ for (i = 0; i < iqn_size; i++) {
+ c = (char *)¶m_buf[i];
+ if (!isupper(*c))
+ continue;
+
+ *c = tolower(*c);
+ }
+}
+
+/*
+ * Processes the first Login Request..
+ */
+static int iscsi_target_locate_portal(
+ struct iscsi_np *np,
+ struct iscsi_conn *conn,
+ struct iscsi_login *login)
+{
+ char *i_buf = NULL, *s_buf = NULL, *t_buf = NULL;
+ char *tmpbuf, *start = NULL, *end = NULL, *key, *value;
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_tiqn *tiqn;
+ struct iscsi_login_req *login_req;
+ struct iscsi_targ_login_rsp *login_rsp;
+ u32 payload_length;
+ int sessiontype = 0, ret = 0;
+
+ login_req = (struct iscsi_login_req *) login->req;
+ login_rsp = (struct iscsi_targ_login_rsp *) login->rsp;
+ payload_length = ntoh24(login_req->dlength);
+
+ login->first_request = 1;
+ login->leading_connection = (!login_req->tsih) ? 1 : 0;
+ login->current_stage =
+ (login_req->flags & ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK) >> 2;
+ login->version_min = login_req->min_version;
+ login->version_max = login_req->max_version;
+ memcpy(login->isid, login_req->isid, 6);
+ login->cmd_sn = login_req->cmdsn;
+ login->init_task_tag = login_req->itt;
+ login->initial_exp_statsn = login_req->exp_statsn;
+ login->cid = login_req->cid;
+ login->tsih = login_req->tsih;
+
+ if (iscsi_target_get_initial_payload(conn, login) < 0)
+ return -1;
+
+ tmpbuf = kzalloc(payload_length + 1, GFP_KERNEL);
+ if (!tmpbuf) {
+ printk(KERN_ERR "Unable to allocate memory for tmpbuf.\n");
+ return -1;
+ }
+
+ memcpy(tmpbuf, login->req_buf, payload_length);
+ tmpbuf[payload_length] = '\0';
+ start = tmpbuf;
+ end = (start + payload_length);
+
+ /*
+ * Locate the initial keys expected from the Initiator node in
+ * the first login request in order to progress with the login phase.
+ */
+ while (start < end) {
+ if (iscsi_extract_key_value(start, &key, &value) < 0) {
+ ret = -1;
+ goto out;
+ }
+
+ if (!strncmp(key, "InitiatorName", 13))
+ i_buf = value;
+ else if (!strncmp(key, "SessionType", 11))
+ s_buf = value;
+ else if (!strncmp(key, "TargetName", 10))
+ t_buf = value;
+
+ start += strlen(key) + strlen(value) + 2;
+ }
+
+ /*
+ * See 5.3. Login Phase.
+ */
+ if (!i_buf) {
+ printk(KERN_ERR "InitiatorName key not received"
+ " in first login request.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_MISSING_FIELDS);
+ ret = -1;
+ goto out;
+ }
+ /*
+ * Convert the incoming InitiatorName to lowercase following
+ * RFC-3720 3.2.6.1. section c) that says that iSCSI IQNs
+ * are NOT case sensitive.
+ */
+ iscsi_initiatorname_tolower(i_buf);
+
+ if (!s_buf) {
+ if (!login->leading_connection)
+ goto get_target;
+
+ printk(KERN_ERR "SessionType key not received"
+ " in first login request.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_MISSING_FIELDS);
+ ret = -1;
+ goto out;
+ }
+
+ /*
+ * Use default portal group for discovery sessions.
+ */
+ sessiontype = strncmp(s_buf, DISCOVERY, 9);
+ if (!sessiontype) {
+ conn->tpg = iscsi_global->discovery_tpg;
+ if (!login->leading_connection)
+ goto get_target;
+
+ sess->sess_ops->SessionType = 1;
+ /*
+ * Setup crc32c modules from libcrypto
+ */
+ if (iscsi_login_setup_crypto(conn) < 0) {
+ printk(KERN_ERR "iscsi_login_setup_crypto() failed\n");
+ ret = -1;
+ goto out;
+ }
+ /*
+ * Serialize access across the discovery struct iscsi_portal_group to
+ * process login attempt.
+ */
+ if (iscsit_access_np(np, conn->tpg) < 0) {
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_SVC_UNAVAILABLE);
+ ret = -1;
+ goto out;
+ }
+ ret = 0;
+ goto out;
+ }
+
+get_target:
+ if (!t_buf) {
+ printk(KERN_ERR "TargetName key not received"
+ " in first login request while"
+ " SessionType=Normal.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_MISSING_FIELDS);
+ ret = -1;
+ goto out;
+ }
+
+ /*
+ * Locate Target IQN from Storage Node.
+ */
+ tiqn = iscsit_get_tiqn_for_login(t_buf);
+ if (!tiqn) {
+ printk(KERN_ERR "Unable to locate Target IQN: %s in"
+ " Storage Node\n", t_buf);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_SVC_UNAVAILABLE);
+ ret = -1;
+ goto out;
+ }
+ printk(KERN_INFO "Located Storage Object: %s\n", tiqn->tiqn);
+
+ /*
+ * Locate Target Portal Group from Storage Node.
+ */
+ conn->tpg = iscsit_get_tpg_from_np(tiqn, np);
+ if (!conn->tpg) {
+ printk(KERN_ERR "Unable to locate Target Portal Group"
+ " on %s\n", tiqn->tiqn);
+ iscsit_put_tiqn_for_login(tiqn);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_SVC_UNAVAILABLE);
+ ret = -1;
+ goto out;
+ }
+ printk(KERN_INFO "Located Portal Group Object: %hu\n", conn->tpg->tpgt);
+ /*
+ * Setup crc32c modules from libcrypto
+ */
+ if (iscsi_login_setup_crypto(conn) < 0) {
+ printk(KERN_ERR "iscsi_login_setup_crypto() failed\n");
+ ret = -1;
+ goto out;
+ }
+ /*
+ * Serialize access across the struct iscsi_portal_group to
+ * process login attempt.
+ */
+ if (iscsit_access_np(np, conn->tpg) < 0) {
+ iscsit_put_tiqn_for_login(tiqn);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_SVC_UNAVAILABLE);
+ ret = -1;
+ conn->tpg = NULL;
+ goto out;
+ }
+
+ /*
+ * conn->sess->node_acl will be set when the referenced
+ * struct iscsi_session is located from received ISID+TSIH in
+ * iscsi_login_non_zero_tsih_s2().
+ */
+ if (!login->leading_connection) {
+ ret = 0;
+ goto out;
+ }
+
+ /*
+ * This value is required in iscsi_login_zero_tsih_s2()
+ */
+ sess->sess_ops->SessionType = 0;
+
+ /*
+ * Locate incoming Initiator IQN reference from Storage Node.
+ */
+ sess->se_sess->se_node_acl = core_tpg_check_initiator_node_acl(
+ &conn->tpg->tpg_se_tpg, i_buf);
+ if (!sess->se_sess->se_node_acl) {
+ printk(KERN_ERR "iSCSI Initiator Node: %s is not authorized to"
+ " access iSCSI target portal group: %hu.\n",
+ i_buf, conn->tpg->tpgt);
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_INITIATOR_ERR,
+ ISCSI_LOGIN_STATUS_TGT_FORBIDDEN);
+ ret = -1;
+ goto out;
+ }
+
+ ret = 0;
+out:
+ kfree(tmpbuf);
+ return ret;
+}
+
+struct iscsi_login *iscsi_target_init_negotiation(
+ struct iscsi_np *np,
+ struct iscsi_conn *conn,
+ char *login_pdu)
+{
+ struct iscsi_login *login;
+
+ login = kzalloc(sizeof(struct iscsi_login), GFP_KERNEL);
+ if (!login) {
+ printk(KERN_ERR "Unable to allocate memory for struct iscsi_login.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ goto out;
+ }
+
+ login->req = kzalloc(ISCSI_HDR_LEN, GFP_KERNEL);
+ if (!login->req) {
+ printk(KERN_ERR "Unable to allocate memory for Login Request.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ goto out;
+ }
+ memcpy(login->req, login_pdu, ISCSI_HDR_LEN);
+
+ login->req_buf = kzalloc(MAX_KEY_VALUE_PAIRS, GFP_KERNEL);
+ if (!login->req_buf) {
+ printk(KERN_ERR "Unable to allocate memory for response buffer.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ goto out;
+ }
+ /*
+ * SessionType: Discovery
+ *
+ * Locates Default Portal
+ *
+ * SessionType: Normal
+ *
+ * Locates Target Portal from NP -> Target IQN
+ */
+ if (iscsi_target_locate_portal(np, conn, login) < 0) {
+ printk(KERN_ERR "iSCSI Login negotiation failed.\n");
+ goto out;
+ }
+
+ return login;
+out:
+ kfree(login->req);
+ kfree(login->req_buf);
+ kfree(login);
+
+ return NULL;
+}
+
+int iscsi_target_start_negotiation(
+ struct iscsi_login *login,
+ struct iscsi_conn *conn)
+{
+ int ret = -1;
+
+ login->rsp = kzalloc(ISCSI_HDR_LEN, GFP_KERNEL);
+ if (!login->rsp) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " Login Response.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ ret = -1;
+ goto out;
+ }
+
+ login->rsp_buf = kzalloc(MAX_KEY_VALUE_PAIRS, GFP_KERNEL);
+ if (!login->rsp_buf) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " request buffer.\n");
+ iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR,
+ ISCSI_LOGIN_STATUS_NO_RESOURCES);
+ ret = -1;
+ goto out;
+ }
+
+ ret = iscsi_target_do_login(conn, login);
+out:
+ if (ret != 0)
+ iscsi_remove_failed_auth_entry(conn);
+
+ iscsi_target_nego_release(login, conn);
+ return ret;
+}
+
+void iscsi_target_nego_release(
+ struct iscsi_login *login,
+ struct iscsi_conn *conn)
+{
+ kfree(login->req);
+ kfree(login->rsp);
+ kfree(login->req_buf);
+ kfree(login->rsp_buf);
+ kfree(login);
+}
diff --git a/drivers/target/iscsi/iscsi_target_nego.h b/drivers/target/iscsi/iscsi_target_nego.h
new file mode 100644
index 0000000..fd74f49
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_nego.h
@@ -0,0 +1,19 @@
+#ifndef ISCSI_TARGET_NEGO_H
+#define ISCSI_TARGET_NEGO_H
+
+#define DECIMAL 0
+#define HEX 1
+
+extern void convert_null_to_semi(char *, int);
+extern int extract_param(const char *, const char *, unsigned int, char *,
+ unsigned char *);
+extern struct iscsi_login *iscsi_target_init_negotiation(
+ struct iscsi_np *, struct iscsi_conn *, char *);
+extern int iscsi_target_start_negotiation(
+ struct iscsi_login *, struct iscsi_conn *);
+extern void iscsi_target_nego_release(
+ struct iscsi_login *, struct iscsi_conn *);
+
+extern struct iscsi_global *iscsi_global;
+
+#endif /* ISCSI_TARGET_NEGO_H */
diff --git a/drivers/target/iscsi/iscsi_target_parameters.c b/drivers/target/iscsi/iscsi_target_parameters.c
new file mode 100644
index 0000000..0dd229c
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_parameters.c
@@ -0,0 +1,1910 @@
+/*******************************************************************************
+ * This file contains main functions related to iSCSI Parameter negotiation.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/slab.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target_parameters.h"
+
+int iscsi_login_rx_data(
+ struct iscsi_conn *conn,
+ char *buf,
+ int length)
+{
+ int rx_got;
+ struct kvec iov;
+
+ memset(&iov, 0, sizeof(struct kvec));
+ iov.iov_len = length;
+ iov.iov_base = buf;
+
+ /*
+ * Initial Marker-less Interval.
+ * Add the values regardless of IFMarker/OFMarker, considering
+ * it may not be negoitated yet.
+ */
+ conn->of_marker += length;
+
+ rx_got = rx_data(conn, &iov, 1, length);
+ if (rx_got != length) {
+ printk(KERN_ERR "rx_data returned %d, expecting %d.\n",
+ rx_got, length);
+ return -1;
+ }
+
+ return 0 ;
+}
+
+int iscsi_login_tx_data(
+ struct iscsi_conn *conn,
+ char *pdu_buf,
+ char *text_buf,
+ int text_length)
+{
+ int length, tx_sent;
+ struct kvec iov[2];
+
+ length = (ISCSI_HDR_LEN + text_length);
+
+ memset(&iov[0], 0, 2 * sizeof(struct kvec));
+ iov[0].iov_len = ISCSI_HDR_LEN;
+ iov[0].iov_base = pdu_buf;
+ iov[1].iov_len = text_length;
+ iov[1].iov_base = text_buf;
+
+ /*
+ * Initial Marker-less Interval.
+ * Add the values regardless of IFMarker/OFMarker, considering
+ * it may not be negoitated yet.
+ */
+ conn->if_marker += length;
+
+ tx_sent = tx_data(conn, &iov[0], 2, length);
+ if (tx_sent != length) {
+ printk(KERN_ERR "tx_data returned %d, expecting %d.\n",
+ tx_sent, length);
+ return -1;
+ }
+
+ return 0;
+}
+
+void iscsi_dump_conn_ops(struct iscsi_conn_ops *conn_ops)
+{
+ printk(KERN_INFO "HeaderDigest: %s\n", (conn_ops->HeaderDigest) ?
+ "CRC32C" : "None");
+ printk(KERN_INFO "DataDigest: %s\n", (conn_ops->DataDigest) ?
+ "CRC32C" : "None");
+ printk(KERN_INFO "MaxRecvDataSegmentLength: %u\n",
+ conn_ops->MaxRecvDataSegmentLength);
+ printk(KERN_INFO "OFMarker: %s\n", (conn_ops->OFMarker) ? "Yes" : "No");
+ printk(KERN_INFO "IFMarker: %s\n", (conn_ops->IFMarker) ? "Yes" : "No");
+ if (conn_ops->OFMarker)
+ printk(KERN_INFO "OFMarkInt: %u\n", conn_ops->OFMarkInt);
+ if (conn_ops->IFMarker)
+ printk(KERN_INFO "IFMarkInt: %u\n", conn_ops->IFMarkInt);
+}
+
+void iscsi_dump_sess_ops(struct iscsi_sess_ops *sess_ops)
+{
+ printk(KERN_INFO "InitiatorName: %s\n", sess_ops->InitiatorName);
+ printk(KERN_INFO "InitiatorAlias: %s\n", sess_ops->InitiatorAlias);
+ printk(KERN_INFO "TargetName: %s\n", sess_ops->TargetName);
+ printk(KERN_INFO "TargetAlias: %s\n", sess_ops->TargetAlias);
+ printk(KERN_INFO "TargetPortalGroupTag: %hu\n",
+ sess_ops->TargetPortalGroupTag);
+ printk(KERN_INFO "MaxConnections: %hu\n", sess_ops->MaxConnections);
+ printk(KERN_INFO "InitialR2T: %s\n",
+ (sess_ops->InitialR2T) ? "Yes" : "No");
+ printk(KERN_INFO "ImmediateData: %s\n", (sess_ops->ImmediateData) ?
+ "Yes" : "No");
+ printk(KERN_INFO "MaxBurstLength: %u\n", sess_ops->MaxBurstLength);
+ printk(KERN_INFO "FirstBurstLength: %u\n", sess_ops->FirstBurstLength);
+ printk(KERN_INFO "DefaultTime2Wait: %hu\n", sess_ops->DefaultTime2Wait);
+ printk(KERN_INFO "DefaultTime2Retain: %hu\n",
+ sess_ops->DefaultTime2Retain);
+ printk(KERN_INFO "MaxOutstandingR2T: %hu\n",
+ sess_ops->MaxOutstandingR2T);
+ printk(KERN_INFO "DataPDUInOrder: %s\n",
+ (sess_ops->DataPDUInOrder) ? "Yes" : "No");
+ printk(KERN_INFO "DataSequenceInOrder: %s\n",
+ (sess_ops->DataSequenceInOrder) ? "Yes" : "No");
+ printk(KERN_INFO "ErrorRecoveryLevel: %hu\n",
+ sess_ops->ErrorRecoveryLevel);
+ printk(KERN_INFO "SessionType: %s\n", (sess_ops->SessionType) ?
+ "Discovery" : "Normal");
+}
+
+void iscsi_print_params(struct iscsi_param_list *param_list)
+{
+ struct iscsi_param *param;
+
+ list_for_each_entry(param, ¶m_list->param_list, p_list)
+ printk(KERN_INFO "%s: %s\n", param->name, param->value);
+}
+
+static struct iscsi_param *iscsi_set_default_param(struct iscsi_param_list *param_list,
+ char *name, char *value, u8 phase, u8 scope, u8 sender,
+ u16 type_range, u8 use)
+{
+ struct iscsi_param *param = NULL;
+
+ param = kzalloc(sizeof(struct iscsi_param), GFP_KERNEL);
+ if (!(param)) {
+ printk(KERN_ERR "Unable to allocate memory for parameter.\n");
+ goto out;
+ }
+ INIT_LIST_HEAD(¶m->p_list);
+
+ param->name = kzalloc(strlen(name) + 1, GFP_KERNEL);
+ if (!(param->name)) {
+ printk(KERN_ERR "Unable to allocate memory for parameter name.\n");
+ goto out;
+ }
+
+ param->value = kzalloc(strlen(value) + 1, GFP_KERNEL);
+ if (!(param->value)) {
+ printk(KERN_ERR "Unable to allocate memory for parameter value.\n");
+ goto out;
+ }
+
+ memcpy(param->name, name, strlen(name));
+ param->name[strlen(name)] = '\0';
+ memcpy(param->value, value, strlen(value));
+ param->value[strlen(value)] = '\0';
+ param->phase = phase;
+ param->scope = scope;
+ param->sender = sender;
+ param->use = use;
+ param->type_range = type_range;
+
+ switch (param->type_range) {
+ case TYPERANGE_BOOL_AND:
+ param->type = TYPE_BOOL_AND;
+ break;
+ case TYPERANGE_BOOL_OR:
+ param->type = TYPE_BOOL_OR;
+ break;
+ case TYPERANGE_0_TO_2:
+ case TYPERANGE_0_TO_3600:
+ case TYPERANGE_0_TO_32767:
+ case TYPERANGE_0_TO_65535:
+ case TYPERANGE_1_TO_65535:
+ case TYPERANGE_2_TO_3600:
+ case TYPERANGE_512_TO_16777215:
+ param->type = TYPE_NUMBER;
+ break;
+ case TYPERANGE_AUTH:
+ case TYPERANGE_DIGEST:
+ param->type = TYPE_VALUE_LIST | TYPE_STRING;
+ break;
+ case TYPERANGE_MARKINT:
+ param->type = TYPE_NUMBER_RANGE;
+ param->type_range |= TYPERANGE_1_TO_65535;
+ break;
+ case TYPERANGE_ISCSINAME:
+ case TYPERANGE_SESSIONTYPE:
+ case TYPERANGE_TARGETADDRESS:
+ case TYPERANGE_UTF8:
+ param->type = TYPE_STRING;
+ break;
+ default:
+ printk(KERN_ERR "Unknown type_range 0x%02x\n",
+ param->type_range);
+ goto out;
+ }
+ list_add_tail(¶m->p_list, ¶m_list->param_list);
+
+ return param;
+out:
+ if (param) {
+ kfree(param->value);
+ kfree(param->name);
+ kfree(param);
+ }
+
+ return NULL;
+}
+
+/* #warning Add extension keys */
+int iscsi_create_default_params(struct iscsi_param_list **param_list_ptr)
+{
+ struct iscsi_param *param = NULL;
+ struct iscsi_param_list *pl;
+
+ pl = kzalloc(sizeof(struct iscsi_param_list), GFP_KERNEL);
+ if (!(pl)) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_param_list.\n");
+ return -1 ;
+ }
+ INIT_LIST_HEAD(&pl->param_list);
+ INIT_LIST_HEAD(&pl->extra_response_list);
+
+ /*
+ * The format for setting the initial parameter definitions are:
+ *
+ * Parameter name:
+ * Initial value:
+ * Allowable phase:
+ * Scope:
+ * Allowable senders:
+ * Typerange:
+ * Use:
+ */
+ param = iscsi_set_default_param(pl, AUTHMETHOD, INITIAL_AUTHMETHOD,
+ PHASE_SECURITY, SCOPE_CONNECTION_ONLY, SENDER_BOTH,
+ TYPERANGE_AUTH, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, HEADERDIGEST, INITIAL_HEADERDIGEST,
+ PHASE_OPERATIONAL, SCOPE_CONNECTION_ONLY, SENDER_BOTH,
+ TYPERANGE_DIGEST, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, DATADIGEST, INITIAL_DATADIGEST,
+ PHASE_OPERATIONAL, SCOPE_CONNECTION_ONLY, SENDER_BOTH,
+ TYPERANGE_DIGEST, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, MAXCONNECTIONS,
+ INITIAL_MAXCONNECTIONS, PHASE_OPERATIONAL,
+ SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_1_TO_65535, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, SENDTARGETS, INITIAL_SENDTARGETS,
+ PHASE_FFP0, SCOPE_SESSION_WIDE, SENDER_INITIATOR,
+ TYPERANGE_UTF8, 0);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, TARGETNAME, INITIAL_TARGETNAME,
+ PHASE_DECLARATIVE, SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_ISCSINAME, USE_ALL);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, INITIATORNAME,
+ INITIAL_INITIATORNAME, PHASE_DECLARATIVE,
+ SCOPE_SESSION_WIDE, SENDER_INITIATOR,
+ TYPERANGE_ISCSINAME, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, TARGETALIAS, INITIAL_TARGETALIAS,
+ PHASE_DECLARATIVE, SCOPE_SESSION_WIDE, SENDER_TARGET,
+ TYPERANGE_UTF8, USE_ALL);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, INITIATORALIAS,
+ INITIAL_INITIATORALIAS, PHASE_DECLARATIVE,
+ SCOPE_SESSION_WIDE, SENDER_INITIATOR, TYPERANGE_UTF8,
+ USE_ALL);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, TARGETADDRESS,
+ INITIAL_TARGETADDRESS, PHASE_DECLARATIVE,
+ SCOPE_SESSION_WIDE, SENDER_TARGET,
+ TYPERANGE_TARGETADDRESS, USE_ALL);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, TARGETPORTALGROUPTAG,
+ INITIAL_TARGETPORTALGROUPTAG,
+ PHASE_DECLARATIVE, SCOPE_SESSION_WIDE, SENDER_TARGET,
+ TYPERANGE_0_TO_65535, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, INITIALR2T, INITIAL_INITIALR2T,
+ PHASE_OPERATIONAL, SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_BOOL_OR, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, IMMEDIATEDATA,
+ INITIAL_IMMEDIATEDATA, PHASE_OPERATIONAL,
+ SCOPE_SESSION_WIDE, SENDER_BOTH, TYPERANGE_BOOL_AND,
+ USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, MAXRECVDATASEGMENTLENGTH,
+ INITIAL_MAXRECVDATASEGMENTLENGTH,
+ PHASE_OPERATIONAL, SCOPE_CONNECTION_ONLY, SENDER_BOTH,
+ TYPERANGE_512_TO_16777215, USE_ALL);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, MAXBURSTLENGTH,
+ INITIAL_MAXBURSTLENGTH, PHASE_OPERATIONAL,
+ SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_512_TO_16777215, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, FIRSTBURSTLENGTH,
+ INITIAL_FIRSTBURSTLENGTH,
+ PHASE_OPERATIONAL, SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_512_TO_16777215, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, DEFAULTTIME2WAIT,
+ INITIAL_DEFAULTTIME2WAIT,
+ PHASE_OPERATIONAL, SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_0_TO_3600, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, DEFAULTTIME2RETAIN,
+ INITIAL_DEFAULTTIME2RETAIN,
+ PHASE_OPERATIONAL, SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_0_TO_3600, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, MAXOUTSTANDINGR2T,
+ INITIAL_MAXOUTSTANDINGR2T,
+ PHASE_OPERATIONAL, SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_1_TO_65535, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, DATAPDUINORDER,
+ INITIAL_DATAPDUINORDER, PHASE_OPERATIONAL,
+ SCOPE_SESSION_WIDE, SENDER_BOTH, TYPERANGE_BOOL_OR,
+ USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, DATASEQUENCEINORDER,
+ INITIAL_DATASEQUENCEINORDER,
+ PHASE_OPERATIONAL, SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_BOOL_OR, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, ERRORRECOVERYLEVEL,
+ INITIAL_ERRORRECOVERYLEVEL,
+ PHASE_OPERATIONAL, SCOPE_SESSION_WIDE, SENDER_BOTH,
+ TYPERANGE_0_TO_2, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, SESSIONTYPE, INITIAL_SESSIONTYPE,
+ PHASE_DECLARATIVE, SCOPE_SESSION_WIDE, SENDER_INITIATOR,
+ TYPERANGE_SESSIONTYPE, USE_LEADING_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, IFMARKER, INITIAL_IFMARKER,
+ PHASE_OPERATIONAL, SCOPE_CONNECTION_ONLY, SENDER_BOTH,
+ TYPERANGE_BOOL_AND, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, OFMARKER, INITIAL_OFMARKER,
+ PHASE_OPERATIONAL, SCOPE_CONNECTION_ONLY, SENDER_BOTH,
+ TYPERANGE_BOOL_AND, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, IFMARKINT, INITIAL_IFMARKINT,
+ PHASE_OPERATIONAL, SCOPE_CONNECTION_ONLY, SENDER_BOTH,
+ TYPERANGE_MARKINT, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ param = iscsi_set_default_param(pl, OFMARKINT, INITIAL_OFMARKINT,
+ PHASE_OPERATIONAL, SCOPE_CONNECTION_ONLY, SENDER_BOTH,
+ TYPERANGE_MARKINT, USE_INITIAL_ONLY);
+ if (!(param))
+ goto out;
+
+ *param_list_ptr = pl;
+ return 0;
+out:
+ iscsi_release_param_list(pl);
+ return -1;
+}
+
+int iscsi_set_keys_to_negotiate(
+ int sessiontype,
+ struct iscsi_param_list *param_list)
+{
+ struct iscsi_param *param;
+
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ param->state = 0;
+ if (!strcmp(param->name, AUTHMETHOD)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, HEADERDIGEST)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, DATADIGEST)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, MAXCONNECTIONS)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, TARGETNAME)) {
+ continue;
+ } else if (!strcmp(param->name, INITIATORNAME)) {
+ continue;
+ } else if (!strcmp(param->name, TARGETALIAS)) {
+ if (param->value)
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, INITIATORALIAS)) {
+ continue;
+ } else if (!strcmp(param->name, TARGETPORTALGROUPTAG)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, INITIALR2T)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, IMMEDIATEDATA)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, MAXRECVDATASEGMENTLENGTH)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, MAXBURSTLENGTH)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, FIRSTBURSTLENGTH)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, DEFAULTTIME2WAIT)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, DEFAULTTIME2RETAIN)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, MAXOUTSTANDINGR2T)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, DATAPDUINORDER)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, DATASEQUENCEINORDER)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, ERRORRECOVERYLEVEL)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, SESSIONTYPE)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, IFMARKER)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, OFMARKER)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, IFMARKINT)) {
+ SET_PSTATE_NEGOTIATE(param);
+ } else if (!strcmp(param->name, OFMARKINT)) {
+ SET_PSTATE_NEGOTIATE(param);
+ }
+ }
+
+ return 0;
+}
+
+int iscsi_set_keys_irrelevant_for_discovery(
+ struct iscsi_param_list *param_list)
+{
+ struct iscsi_param *param;
+
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ if (!strcmp(param->name, MAXCONNECTIONS))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, INITIALR2T))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, IMMEDIATEDATA))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, MAXBURSTLENGTH))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, FIRSTBURSTLENGTH))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, MAXOUTSTANDINGR2T))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, DATAPDUINORDER))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, DATASEQUENCEINORDER))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, ERRORRECOVERYLEVEL))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, DEFAULTTIME2WAIT))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, DEFAULTTIME2RETAIN))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, IFMARKER))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, OFMARKER))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, IFMARKINT))
+ param->state &= ~PSTATE_NEGOTIATE;
+ else if (!strcmp(param->name, OFMARKINT))
+ param->state &= ~PSTATE_NEGOTIATE;
+ }
+
+ return 0;
+}
+
+int iscsi_copy_param_list(
+ struct iscsi_param_list **dst_param_list,
+ struct iscsi_param_list *src_param_list,
+ int leading)
+{
+ struct iscsi_param *new_param = NULL, *param = NULL;
+ struct iscsi_param_list *param_list = NULL;
+
+ param_list = kzalloc(sizeof(struct iscsi_param_list), GFP_KERNEL);
+ if (!(param_list)) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_param_list.\n");
+ goto err_out;
+ }
+ INIT_LIST_HEAD(¶m_list->param_list);
+ INIT_LIST_HEAD(¶m_list->extra_response_list);
+
+ list_for_each_entry(param, &src_param_list->param_list, p_list) {
+ if (!leading && (param->scope & SCOPE_SESSION_WIDE)) {
+ if ((strcmp(param->name, "TargetName") != 0) &&
+ (strcmp(param->name, "InitiatorName") != 0) &&
+ (strcmp(param->name, "TargetPortalGroupTag") != 0))
+ continue;
+ }
+
+ new_param = kzalloc(sizeof(struct iscsi_param), GFP_KERNEL);
+ if (!(new_param)) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_param.\n");
+ goto err_out;
+ }
+
+ new_param->set_param = param->set_param;
+ new_param->phase = param->phase;
+ new_param->scope = param->scope;
+ new_param->sender = param->sender;
+ new_param->type = param->type;
+ new_param->use = param->use;
+ new_param->type_range = param->type_range;
+
+ new_param->name = kzalloc(strlen(param->name) + 1, GFP_KERNEL);
+ if (!(new_param->name)) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " parameter name.\n");
+ goto err_out;
+ }
+
+ new_param->value = kzalloc(strlen(param->value) + 1,
+ GFP_KERNEL);
+ if (!(new_param->value)) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " parameter value.\n");
+ goto err_out;
+ }
+
+ memcpy(new_param->name, param->name, strlen(param->name));
+ new_param->name[strlen(param->name)] = '\0';
+ memcpy(new_param->value, param->value, strlen(param->value));
+ new_param->value[strlen(param->value)] = '\0';
+
+ list_add_tail(&new_param->p_list, ¶m_list->param_list);
+ }
+
+ if (!(list_empty(¶m_list->param_list)))
+ *dst_param_list = param_list;
+ else {
+ printk(KERN_ERR "No parameters allocated.\n");
+ goto err_out;
+ }
+
+ return 0;
+
+err_out:
+ iscsi_release_param_list(param_list);
+ return -1;
+}
+
+static void iscsi_release_extra_responses(struct iscsi_param_list *param_list)
+{
+ struct iscsi_extra_response *er, *er_tmp;
+
+ list_for_each_entry_safe(er, er_tmp, ¶m_list->extra_response_list,
+ er_list) {
+ list_del(&er->er_list);
+ kfree(er);
+ }
+}
+
+void iscsi_release_param_list(struct iscsi_param_list *param_list)
+{
+ struct iscsi_param *param, *param_tmp;
+
+ list_for_each_entry_safe(param, param_tmp, ¶m_list->param_list,
+ p_list) {
+ list_del(¶m->p_list);
+
+ kfree(param->name);
+ param->name = NULL;
+ kfree(param->value);
+ param->value = NULL;
+ kfree(param);
+ param = NULL;
+ }
+
+ iscsi_release_extra_responses(param_list);
+
+ kfree(param_list);
+}
+
+struct iscsi_param *iscsi_find_param_from_key(
+ char *key,
+ struct iscsi_param_list *param_list)
+{
+ struct iscsi_param *param;
+
+ if (!key || !param_list) {
+ printk(KERN_ERR "Key or parameter list pointer is NULL.\n");
+ return NULL;
+ }
+
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ if (!strcmp(key, param->name))
+ break;
+ }
+
+ if (!param) {
+ printk(KERN_ERR "Unable to locate key \"%s\".\n", key);
+ return NULL;
+ }
+
+ return param;
+}
+
+int iscsi_extract_key_value(char *textbuf, char **key, char **value)
+{
+ *value = strchr(textbuf, '=');
+ if (!(*value)) {
+ printk(KERN_ERR "Unable to locate \"=\" seperator for key,"
+ " ignoring request.\n");
+ return -1;
+ }
+
+ *key = textbuf;
+ **value = '\0';
+ *value = *value + 1;
+
+ return 0;
+}
+
+int iscsi_update_param_value(struct iscsi_param *param, char *value)
+{
+ kfree(param->value);
+
+ param->value = kzalloc(strlen(value) + 1, GFP_KERNEL);
+ if (!(param->value)) {
+ printk(KERN_ERR "Unable to allocate memory for value.\n");
+ return -1;
+ }
+
+ memcpy(param->value, value, strlen(value));
+ param->value[strlen(value)] = '\0';
+
+ TRACE(TRACE_PARAM, "iSCSI Parameter updated to %s=%s\n",
+ param->name, param->value);
+ return 0;
+}
+
+static int iscsi_add_notunderstood_response(
+ char *key,
+ char *value,
+ struct iscsi_param_list *param_list)
+{
+ struct iscsi_extra_response *extra_response;
+
+ if (strlen(value) > MAX_KEY_VALUE_LENGTH) {
+ printk(KERN_ERR "Value for notunderstood key \"%s\" exceeds %d,"
+ " protocol error.\n", key, MAX_KEY_VALUE_LENGTH);
+ return -1;
+ }
+
+ extra_response = kzalloc(sizeof(struct iscsi_extra_response), GFP_KERNEL);
+ if (!(extra_response)) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_extra_response.\n");
+ return -1;
+ }
+ INIT_LIST_HEAD(&extra_response->er_list);
+
+ strncpy(extra_response->key, key, strlen(key) + 1);
+ strncpy(extra_response->value, NOTUNDERSTOOD,
+ strlen(NOTUNDERSTOOD) + 1);
+
+ list_add_tail(&extra_response->er_list,
+ ¶m_list->extra_response_list);
+ return 0;
+}
+
+static int iscsi_check_for_auth_key(char *key)
+{
+ /*
+ * RFC 1994
+ */
+ if (!strcmp(key, "CHAP_A") || !strcmp(key, "CHAP_I") ||
+ !strcmp(key, "CHAP_C") || !strcmp(key, "CHAP_N") ||
+ !strcmp(key, "CHAP_R"))
+ return 1;
+
+ /*
+ * RFC 2945
+ */
+ if (!strcmp(key, "SRP_U") || !strcmp(key, "SRP_N") ||
+ !strcmp(key, "SRP_g") || !strcmp(key, "SRP_s") ||
+ !strcmp(key, "SRP_A") || !strcmp(key, "SRP_B") ||
+ !strcmp(key, "SRP_M") || !strcmp(key, "SRP_HM"))
+ return 1;
+
+ return 0;
+}
+
+static void iscsi_check_proposer_for_optional_reply(struct iscsi_param *param)
+{
+ if (IS_TYPE_BOOL_AND(param)) {
+ if (!strcmp(param->value, NO))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ } else if (IS_TYPE_BOOL_OR(param)) {
+ if (!strcmp(param->value, YES))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ /*
+ * Required for gPXE iSCSI boot client
+ */
+ if (!strcmp(param->name, IMMEDIATEDATA))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ } else if (IS_TYPE_NUMBER(param)) {
+ if (!strcmp(param->name, MAXRECVDATASEGMENTLENGTH))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ /*
+ * The GlobalSAN iSCSI Initiator for MacOSX does
+ * not respond to MaxBurstLength, FirstBurstLength,
+ * DefaultTime2Wait or DefaultTime2Retain parameter keys.
+ * So, we set them to 'reply optional' here, and assume the
+ * the defaults from iscsi_parameters.h if the initiator
+ * is not RFC compliant and the keys are not negotiated.
+ */
+ if (!strcmp(param->name, MAXBURSTLENGTH))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ if (!strcmp(param->name, FIRSTBURSTLENGTH))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ if (!strcmp(param->name, DEFAULTTIME2WAIT))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ if (!strcmp(param->name, DEFAULTTIME2RETAIN))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ /*
+ * Required for gPXE iSCSI boot client
+ */
+ if (!strcmp(param->name, MAXCONNECTIONS))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ } else if (IS_PHASE_DECLARATIVE(param))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+}
+
+static int iscsi_check_boolean_value(struct iscsi_param *param, char *value)
+{
+ if (strcmp(value, YES) && strcmp(value, NO)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be either"
+ " \"%s\" or \"%s\".\n", param->name, YES, NO);
+ return -1;
+ }
+
+ return 0;
+}
+
+static int iscsi_check_numerical_value(struct iscsi_param *param, char *value_ptr)
+{
+ char *tmpptr;
+ int value = 0;
+
+ value = simple_strtoul(value_ptr, &tmpptr, 0);
+
+/* #warning FIXME: Fix this */
+#if 0
+ if (strspn(endptr, WHITE_SPACE) != strlen(endptr)) {
+ printk(KERN_ERR "Illegal value \"%s\" for \"%s\".\n",
+ value, param->name);
+ return -1;
+ }
+#endif
+ if (IS_TYPERANGE_0_TO_2(param)) {
+ if ((value < 0) || (value > 2)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " between 0 and 2.\n", param->name);
+ return -1;
+ }
+ return 0;
+ }
+ if (IS_TYPERANGE_0_TO_3600(param)) {
+ if ((value < 0) || (value > 3600)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " between 0 and 3600.\n", param->name);
+ return -1;
+ }
+ return 0;
+ }
+ if (IS_TYPERANGE_0_TO_32767(param)) {
+ if ((value < 0) || (value > 32767)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " between 0 and 32767.\n", param->name);
+ return -1;
+ }
+ return 0;
+ }
+ if (IS_TYPERANGE_0_TO_65535(param)) {
+ if ((value < 0) || (value > 65535)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " between 0 and 65535.\n", param->name);
+ return -1;
+ }
+ return 0;
+ }
+ if (IS_TYPERANGE_1_TO_65535(param)) {
+ if ((value < 1) || (value > 65535)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " between 1 and 65535.\n", param->name);
+ return -1;
+ }
+ return 0;
+ }
+ if (IS_TYPERANGE_2_TO_3600(param)) {
+ if ((value < 2) || (value > 3600)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " between 2 and 3600.\n", param->name);
+ return -1;
+ }
+ return 0;
+ }
+ if (IS_TYPERANGE_512_TO_16777215(param)) {
+ if ((value < 512) || (value > 16777215)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " between 512 and 16777215.\n", param->name);
+ return -1;
+ }
+ return 0;
+ }
+
+ return 0;
+}
+
+static int iscsi_check_numerical_range_value(struct iscsi_param *param, char *value)
+{
+ char *left_val_ptr = NULL, *right_val_ptr = NULL;
+ char *tilde_ptr = NULL, *tmp_ptr = NULL;
+ u32 left_val, right_val, local_left_val, local_right_val;
+
+ if ((strcmp(param->name, IFMARKINT)) &&
+ (strcmp(param->name, OFMARKINT))) {
+ printk(KERN_ERR "Only parameters \"%s\" or \"%s\" may contain a"
+ " numerical range value.\n", IFMARKINT, OFMARKINT);
+ return -1;
+ }
+
+ if (IS_PSTATE_PROPOSER(param))
+ return 0;
+
+ tilde_ptr = strchr(value, '~');
+ if (!(tilde_ptr)) {
+ printk(KERN_ERR "Unable to locate numerical range indicator"
+ " \"~\" for \"%s\".\n", param->name);
+ return -1;
+ }
+ *tilde_ptr = '\0';
+
+ left_val_ptr = value;
+ right_val_ptr = value + strlen(left_val_ptr) + 1;
+
+ if (iscsi_check_numerical_value(param, left_val_ptr) < 0)
+ return -1;
+ if (iscsi_check_numerical_value(param, right_val_ptr) < 0)
+ return -1;
+
+ left_val = simple_strtoul(left_val_ptr, &tmp_ptr, 0);
+ right_val = simple_strtoul(right_val_ptr, &tmp_ptr, 0);
+ *tilde_ptr = '~';
+
+ if (right_val < left_val) {
+ printk(KERN_ERR "Numerical range for parameter \"%s\" contains"
+ " a right value which is less than the left.\n",
+ param->name);
+ return -1;
+ }
+
+ /*
+ * For now, enforce reasonable defaults for [I,O]FMarkInt.
+ */
+ tilde_ptr = strchr(param->value, '~');
+ if (!(tilde_ptr)) {
+ printk(KERN_ERR "Unable to locate numerical range indicator"
+ " \"~\" for \"%s\".\n", param->name);
+ return -1;
+ }
+ *tilde_ptr = '\0';
+
+ left_val_ptr = param->value;
+ right_val_ptr = param->value + strlen(left_val_ptr) + 1;
+
+ local_left_val = simple_strtoul(left_val_ptr, &tmp_ptr, 0);
+ local_right_val = simple_strtoul(right_val_ptr, &tmp_ptr, 0);
+ *tilde_ptr = '~';
+
+ if (param->set_param) {
+ if ((left_val < local_left_val) ||
+ (right_val < local_left_val)) {
+ printk(KERN_ERR "Passed value range \"%u~%u\" is below"
+ " minimum left value \"%u\" for key \"%s\","
+ " rejecting.\n", left_val, right_val,
+ local_left_val, param->name);
+ return -1;
+ }
+ } else {
+ if ((left_val < local_left_val) &&
+ (right_val < local_left_val)) {
+ printk(KERN_ERR "Received value range \"%u~%u\" is"
+ " below minimum left value \"%u\" for key"
+ " \"%s\", rejecting.\n", left_val, right_val,
+ local_left_val, param->name);
+ SET_PSTATE_REJECT(param);
+ if (iscsi_update_param_value(param, REJECT) < 0)
+ return -1;
+ }
+ }
+
+ return 0;
+}
+
+static int iscsi_check_string_or_list_value(struct iscsi_param *param, char *value)
+{
+ if (IS_PSTATE_PROPOSER(param))
+ return 0;
+
+ if (IS_TYPERANGE_AUTH_PARAM(param)) {
+ if (strcmp(value, KRB5) && strcmp(value, SPKM1) &&
+ strcmp(value, SPKM2) && strcmp(value, SRP) &&
+ strcmp(value, CHAP) && strcmp(value, NONE)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " \"%s\", \"%s\", \"%s\", \"%s\", \"%s\""
+ " or \"%s\".\n", param->name, KRB5,
+ SPKM1, SPKM2, SRP, CHAP, NONE);
+ return -1;
+ }
+ }
+ if (IS_TYPERANGE_DIGEST_PARAM(param)) {
+ if (strcmp(value, CRC32C) && strcmp(value, NONE)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " \"%s\" or \"%s\".\n", param->name,
+ CRC32C, NONE);
+ return -1;
+ }
+ }
+ if (IS_TYPERANGE_SESSIONTYPE(param)) {
+ if (strcmp(value, DISCOVERY) && strcmp(value, NORMAL)) {
+ printk(KERN_ERR "Illegal value for \"%s\", must be"
+ " \"%s\" or \"%s\".\n", param->name,
+ DISCOVERY, NORMAL);
+ return -1;
+ }
+ }
+
+ return 0;
+}
+
+/*
+ * This function is used to pick a value range number, currently just
+ * returns the lesser of both right values.
+ */
+static char *iscsi_get_value_from_number_range(
+ struct iscsi_param *param,
+ char *value)
+{
+ char *end_ptr, *tilde_ptr1 = NULL, *tilde_ptr2 = NULL;
+ u32 acceptor_right_value, proposer_right_value;
+
+ tilde_ptr1 = strchr(value, '~');
+ if (!(tilde_ptr1))
+ return NULL;
+ *tilde_ptr1++ = '\0';
+ proposer_right_value = simple_strtoul(tilde_ptr1, &end_ptr, 0);
+
+ tilde_ptr2 = strchr(param->value, '~');
+ if (!(tilde_ptr2))
+ return NULL;
+ *tilde_ptr2++ = '\0';
+ acceptor_right_value = simple_strtoul(tilde_ptr2, &end_ptr, 0);
+
+ return (acceptor_right_value >= proposer_right_value) ?
+ tilde_ptr1 : tilde_ptr2;
+}
+
+static char *iscsi_check_valuelist_for_support(
+ struct iscsi_param *param,
+ char *value)
+{
+ char *tmp1 = NULL, *tmp2 = NULL;
+ char *acceptor_values = NULL, *proposer_values = NULL;
+
+ acceptor_values = param->value;
+ proposer_values = value;
+
+ do {
+ if (!proposer_values)
+ return NULL;
+ tmp1 = strchr(proposer_values, ',');
+ if ((tmp1))
+ *tmp1 = '\0';
+ acceptor_values = param->value;
+ do {
+ if (!acceptor_values) {
+ if (tmp1)
+ *tmp1 = ',';
+ return NULL;
+ }
+ tmp2 = strchr(acceptor_values, ',');
+ if ((tmp2))
+ *tmp2 = '\0';
+ if (!acceptor_values || !proposer_values) {
+ if (tmp1)
+ *tmp1 = ',';
+ if (tmp2)
+ *tmp2 = ',';
+ return NULL;
+ }
+ if (!strcmp(acceptor_values, proposer_values)) {
+ if (tmp2)
+ *tmp2 = ',';
+ goto out;
+ }
+ if (tmp2)
+ *tmp2++ = ',';
+
+ acceptor_values = tmp2;
+ if (!acceptor_values)
+ break;
+ } while (acceptor_values);
+ if (tmp1)
+ *tmp1++ = ',';
+ proposer_values = tmp1;
+ } while (proposer_values);
+
+out:
+ return proposer_values;
+}
+
+static int iscsi_check_acceptor_state(struct iscsi_param *param, char *value)
+{
+ u8 acceptor_boolean_value = 0, proposer_boolean_value = 0;
+ char *negoitated_value = NULL;
+
+ if (IS_PSTATE_ACCEPTOR(param)) {
+ printk(KERN_ERR "Received key \"%s\" twice, protocol error.\n",
+ param->name);
+ return -1;
+ }
+
+ if (IS_PSTATE_REJECT(param))
+ return 0;
+
+ if (IS_TYPE_BOOL_AND(param)) {
+ if (!strcmp(value, YES))
+ proposer_boolean_value = 1;
+ if (!strcmp(param->value, YES))
+ acceptor_boolean_value = 1;
+ if (acceptor_boolean_value && proposer_boolean_value)
+ do {} while (0);
+ else {
+ if (iscsi_update_param_value(param, NO) < 0)
+ return -1;
+ if (!proposer_boolean_value)
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ }
+ } else if (IS_TYPE_BOOL_OR(param)) {
+ if (!strcmp(value, YES))
+ proposer_boolean_value = 1;
+ if (!strcmp(param->value, YES))
+ acceptor_boolean_value = 1;
+ if (acceptor_boolean_value || proposer_boolean_value) {
+ if (iscsi_update_param_value(param, YES) < 0)
+ return -1;
+ if (proposer_boolean_value)
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ }
+ } else if (IS_TYPE_NUMBER(param)) {
+ char *tmpptr, buf[10];
+ u32 acceptor_value = simple_strtoul(param->value, &tmpptr, 0);
+ u32 proposer_value = simple_strtoul(value, &tmpptr, 0);
+
+ memset(buf, 0, 10);
+
+ if (!strcmp(param->name, MAXCONNECTIONS) ||
+ !strcmp(param->name, MAXBURSTLENGTH) ||
+ !strcmp(param->name, FIRSTBURSTLENGTH) ||
+ !strcmp(param->name, MAXOUTSTANDINGR2T) ||
+ !strcmp(param->name, DEFAULTTIME2RETAIN) ||
+ !strcmp(param->name, ERRORRECOVERYLEVEL)) {
+ if (proposer_value > acceptor_value) {
+ sprintf(buf, "%u", acceptor_value);
+ if (iscsi_update_param_value(param,
+ &buf[0]) < 0)
+ return -1;
+ } else {
+ if (iscsi_update_param_value(param, value) < 0)
+ return -1;
+ }
+ } else if (!strcmp(param->name, DEFAULTTIME2WAIT)) {
+ if (acceptor_value > proposer_value) {
+ sprintf(buf, "%u", acceptor_value);
+ if (iscsi_update_param_value(param,
+ &buf[0]) < 0)
+ return -1;
+ } else {
+ if (iscsi_update_param_value(param, value) < 0)
+ return -1;
+ }
+ } else {
+ if (iscsi_update_param_value(param, value) < 0)
+ return -1;
+ }
+
+ if (!strcmp(param->name, MAXRECVDATASEGMENTLENGTH))
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ } else if (IS_TYPE_NUMBER_RANGE(param)) {
+ negoitated_value = iscsi_get_value_from_number_range(
+ param, value);
+ if (!(negoitated_value))
+ return -1;
+ if (iscsi_update_param_value(param, negoitated_value) < 0)
+ return -1;
+ } else if (IS_TYPE_VALUE_LIST(param)) {
+ negoitated_value = iscsi_check_valuelist_for_support(
+ param, value);
+ if (!(negoitated_value)) {
+ printk(KERN_ERR "Proposer's value list \"%s\" contains"
+ " no valid values from Acceptor's value list"
+ " \"%s\".\n", value, param->value);
+ return -1;
+ }
+ if (iscsi_update_param_value(param, negoitated_value) < 0)
+ return -1;
+ } else if (IS_PHASE_DECLARATIVE(param)) {
+ if (iscsi_update_param_value(param, value) < 0)
+ return -1;
+ SET_PSTATE_REPLY_OPTIONAL(param);
+ }
+
+ return 0;
+}
+
+static int iscsi_check_proposer_state(struct iscsi_param *param, char *value)
+{
+ if (IS_PSTATE_RESPONSE_GOT(param)) {
+ printk(KERN_ERR "Received key \"%s\" twice, protocol error.\n",
+ param->name);
+ return -1;
+ }
+
+ if (IS_TYPE_NUMBER_RANGE(param)) {
+ u32 left_val = 0, right_val = 0, recieved_value = 0;
+ char *left_val_ptr = NULL, *right_val_ptr = NULL;
+ char *tilde_ptr = NULL, *tmp_ptr = NULL;
+
+ if (!strcmp(value, IRRELEVANT) || !strcmp(value, REJECT)) {
+ if (iscsi_update_param_value(param, value) < 0)
+ return -1;
+ return 0;
+ }
+
+ tilde_ptr = strchr(value, '~');
+ if ((tilde_ptr)) {
+ printk(KERN_ERR "Illegal \"~\" in response for \"%s\".\n",
+ param->name);
+ return -1;
+ }
+ tilde_ptr = strchr(param->value, '~');
+ if (!(tilde_ptr)) {
+ printk(KERN_ERR "Unable to locate numerical range"
+ " indicator \"~\" for \"%s\".\n", param->name);
+ return -1;
+ }
+ *tilde_ptr = '\0';
+
+ left_val_ptr = param->value;
+ right_val_ptr = param->value + strlen(left_val_ptr) + 1;
+ left_val = simple_strtoul(left_val_ptr, &tmp_ptr, 0);
+ right_val = simple_strtoul(right_val_ptr, &tmp_ptr, 0);
+ recieved_value = simple_strtoul(value, &tmp_ptr, 0);
+
+ *tilde_ptr = '~';
+
+ if ((recieved_value < left_val) ||
+ (recieved_value > right_val)) {
+ printk(KERN_ERR "Illegal response \"%s=%u\", value must"
+ " be between %u and %u.\n", param->name,
+ recieved_value, left_val, right_val);
+ return -1;
+ }
+ } else if (IS_TYPE_VALUE_LIST(param)) {
+ char *comma_ptr = NULL, *tmp_ptr = NULL;
+
+ comma_ptr = strchr(value, ',');
+ if ((comma_ptr)) {
+ printk(KERN_ERR "Illegal \",\" in response for \"%s\".\n",
+ param->name);
+ return -1;
+ }
+
+ tmp_ptr = iscsi_check_valuelist_for_support(param, value);
+ if (!(tmp_ptr))
+ return -1;
+ }
+
+ if (iscsi_update_param_value(param, value) < 0)
+ return -1;
+
+ return 0;
+}
+
+static int iscsi_check_value(struct iscsi_param *param, char *value)
+{
+ char *comma_ptr = NULL;
+
+ if (!strcmp(value, REJECT)) {
+ if (!strcmp(param->name, IFMARKINT) ||
+ !strcmp(param->name, OFMARKINT)) {
+ /*
+ * Reject is not fatal for [I,O]FMarkInt, and causes
+ * [I,O]FMarker to be reset to No. (See iSCSI v20 A.3.2)
+ */
+ SET_PSTATE_REJECT(param);
+ return 0;
+ }
+ printk(KERN_ERR "Received %s=%s\n", param->name, value);
+ return -1;
+ }
+ if (!strcmp(value, IRRELEVANT)) {
+ TRACE(TRACE_LOGIN, "Received %s=%s\n", param->name, value);
+ SET_PSTATE_IRRELEVANT(param);
+ return 0;
+ }
+ if (!strcmp(value, NOTUNDERSTOOD)) {
+ if (!IS_PSTATE_PROPOSER(param)) {
+ printk(KERN_ERR "Received illegal offer %s=%s\n",
+ param->name, value);
+ return -1;
+ }
+
+/* #warning FIXME: Add check for X-ExtensionKey here */
+ printk(KERN_ERR "Standard iSCSI key \"%s\" cannot be answered"
+ " with \"%s\", protocol error.\n", param->name, value);
+ return -1;
+ }
+
+ do {
+ comma_ptr = NULL;
+ comma_ptr = strchr(value, ',');
+
+ if (comma_ptr && !IS_TYPE_VALUE_LIST(param)) {
+ printk(KERN_ERR "Detected value seperator \",\", but"
+ " key \"%s\" does not allow a value list,"
+ " protocol error.\n", param->name);
+ return -1;
+ }
+ if (comma_ptr)
+ *comma_ptr = '\0';
+
+ if (strlen(value) > MAX_KEY_VALUE_LENGTH) {
+ printk(KERN_ERR "Value for key \"%s\" exceeds %d,"
+ " protocol error.\n", param->name,
+ MAX_KEY_VALUE_LENGTH);
+ return -1;
+ }
+
+ if (IS_TYPE_BOOL_AND(param) || IS_TYPE_BOOL_OR(param)) {
+ if (iscsi_check_boolean_value(param, value) < 0)
+ return -1;
+ } else if (IS_TYPE_NUMBER(param)) {
+ if (iscsi_check_numerical_value(param, value) < 0)
+ return -1;
+ } else if (IS_TYPE_NUMBER_RANGE(param)) {
+ if (iscsi_check_numerical_range_value(param, value) < 0)
+ return -1;
+ } else if (IS_TYPE_STRING(param) || IS_TYPE_VALUE_LIST(param)) {
+ if (iscsi_check_string_or_list_value(param, value) < 0)
+ return -1;
+ } else {
+ printk(KERN_ERR "Huh? 0x%02x\n", param->type);
+ return -1;
+ }
+
+ if (comma_ptr)
+ *comma_ptr++ = ',';
+
+ value = comma_ptr;
+ } while (value);
+
+ return 0;
+}
+
+static struct iscsi_param *__iscsi_check_key(
+ char *key,
+ int sender,
+ struct iscsi_param_list *param_list)
+{
+ struct iscsi_param *param;
+
+ if (strlen(key) > MAX_KEY_NAME_LENGTH) {
+ printk(KERN_ERR "Length of key name \"%s\" exceeds %d.\n",
+ key, MAX_KEY_NAME_LENGTH);
+ return NULL;
+ }
+
+ param = iscsi_find_param_from_key(key, param_list);
+ if (!(param))
+ return NULL;
+
+ if ((sender & SENDER_INITIATOR) && !IS_SENDER_INITIATOR(param)) {
+ printk(KERN_ERR "Key \"%s\" may not be sent to %s,"
+ " protocol error.\n", param->name,
+ (sender & SENDER_RECEIVER) ? "target" : "initiator");
+ return NULL;
+ }
+
+ if ((sender & SENDER_TARGET) && !IS_SENDER_TARGET(param)) {
+ printk(KERN_ERR "Key \"%s\" may not be sent to %s,"
+ " protocol error.\n", param->name,
+ (sender & SENDER_RECEIVER) ? "initiator" : "target");
+ return NULL;
+ }
+
+ return param;
+}
+
+static struct iscsi_param *iscsi_check_key(
+ char *key,
+ int phase,
+ int sender,
+ struct iscsi_param_list *param_list)
+{
+ struct iscsi_param *param;
+ /*
+ * Key name length must not exceed 63 bytes. (See iSCSI v20 5.1)
+ */
+ if (strlen(key) > MAX_KEY_NAME_LENGTH) {
+ printk(KERN_ERR "Length of key name \"%s\" exceeds %d.\n",
+ key, MAX_KEY_NAME_LENGTH);
+ return NULL;
+ }
+
+ param = iscsi_find_param_from_key(key, param_list);
+ if (!(param))
+ return NULL;
+
+ if ((sender & SENDER_INITIATOR) && !IS_SENDER_INITIATOR(param)) {
+ printk(KERN_ERR "Key \"%s\" may not be sent to %s,"
+ " protocol error.\n", param->name,
+ (sender & SENDER_RECEIVER) ? "target" : "initiator");
+ return NULL;
+ }
+ if ((sender & SENDER_TARGET) && !IS_SENDER_TARGET(param)) {
+ printk(KERN_ERR "Key \"%s\" may not be sent to %s,"
+ " protocol error.\n", param->name,
+ (sender & SENDER_RECEIVER) ? "initiator" : "target");
+ return NULL;
+ }
+
+ if (IS_PSTATE_ACCEPTOR(param)) {
+ printk(KERN_ERR "Key \"%s\" received twice, protocol error.\n",
+ key);
+ return NULL;
+ }
+
+ if (!phase)
+ return param;
+
+ if (!(param->phase & phase)) {
+ printk(KERN_ERR "Key \"%s\" may not be negotiated during ",
+ param->name);
+ switch (phase) {
+ case PHASE_SECURITY:
+ printk(KERN_INFO "Security phase.\n");
+ break;
+ case PHASE_OPERATIONAL:
+ printk(KERN_INFO "Operational phase.\n");
+ default:
+ printk(KERN_INFO "Unknown phase.\n");
+ }
+ return NULL;
+ }
+
+ return param;
+}
+
+static int iscsi_enforce_integrity_rules(
+ u8 phase,
+ struct iscsi_param_list *param_list)
+{
+ char *tmpptr;
+ u8 DataSequenceInOrder = 0;
+ u8 ErrorRecoveryLevel = 0, SessionType = 0;
+ u8 IFMarker = 0, OFMarker = 0;
+ u8 IFMarkInt_Reject = 0, OFMarkInt_Reject = 0;
+ u32 FirstBurstLength = 0, MaxBurstLength = 0;
+ struct iscsi_param *param = NULL;
+
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ if (!(param->phase & phase))
+ continue;
+ if (!strcmp(param->name, SESSIONTYPE))
+ if (!strcmp(param->value, NORMAL))
+ SessionType = 1;
+ if (!strcmp(param->name, ERRORRECOVERYLEVEL))
+ ErrorRecoveryLevel = simple_strtoul(param->value,
+ &tmpptr, 0);
+ if (!strcmp(param->name, DATASEQUENCEINORDER))
+ if (!strcmp(param->value, YES))
+ DataSequenceInOrder = 1;
+ if (!strcmp(param->name, MAXBURSTLENGTH))
+ MaxBurstLength = simple_strtoul(param->value,
+ &tmpptr, 0);
+ if (!strcmp(param->name, IFMARKER))
+ if (!strcmp(param->value, YES))
+ IFMarker = 1;
+ if (!strcmp(param->name, OFMARKER))
+ if (!strcmp(param->value, YES))
+ OFMarker = 1;
+ if (!strcmp(param->name, IFMARKINT))
+ if (!strcmp(param->value, REJECT))
+ IFMarkInt_Reject = 1;
+ if (!strcmp(param->name, OFMARKINT))
+ if (!strcmp(param->value, REJECT))
+ OFMarkInt_Reject = 1;
+ }
+
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ if (!(param->phase & phase))
+ continue;
+ if (!SessionType && (!IS_PSTATE_ACCEPTOR(param) &&
+ (strcmp(param->name, IFMARKER) &&
+ strcmp(param->name, OFMARKER) &&
+ strcmp(param->name, IFMARKINT) &&
+ strcmp(param->name, OFMARKINT))))
+ continue;
+ if (!strcmp(param->name, MAXOUTSTANDINGR2T) &&
+ DataSequenceInOrder && (ErrorRecoveryLevel > 0)) {
+ if (strcmp(param->value, "1")) {
+ if (iscsi_update_param_value(param, "1") < 0)
+ return -1;
+ TRACE(TRACE_PARAM, "Reset \"%s\" to \"%s\".\n",
+ param->name, param->value);
+ }
+ }
+ if (!strcmp(param->name, MAXCONNECTIONS) && !SessionType) {
+ if (strcmp(param->value, "1")) {
+ if (iscsi_update_param_value(param, "1") < 0)
+ return -1;
+ TRACE(TRACE_PARAM, "Reset \"%s\" to \"%s\".\n",
+ param->name, param->value);
+ }
+ }
+ if (!strcmp(param->name, FIRSTBURSTLENGTH)) {
+ FirstBurstLength = simple_strtoul(param->value,
+ &tmpptr, 0);
+ if (FirstBurstLength > MaxBurstLength) {
+ char tmpbuf[10];
+ memset(tmpbuf, 0, 10);
+ sprintf(tmpbuf, "%u", MaxBurstLength);
+ if (iscsi_update_param_value(param, tmpbuf))
+ return -1;
+ TRACE(TRACE_PARAM, "Reset \"%s\" to \"%s\".\n",
+ param->name, param->value);
+ }
+ }
+ if (!strcmp(param->name, IFMARKER) && IFMarkInt_Reject) {
+ if (iscsi_update_param_value(param, NO) < 0)
+ return -1;
+ IFMarker = 0;
+ TRACE(TRACE_PARAM, "Reset \"%s\" to \"%s\".\n",
+ param->name, param->value);
+ }
+ if (!strcmp(param->name, OFMARKER) && OFMarkInt_Reject) {
+ if (iscsi_update_param_value(param, NO) < 0)
+ return -1;
+ OFMarker = 0;
+ TRACE(TRACE_PARAM, "Reset \"%s\" to \"%s\".\n",
+ param->name, param->value);
+ }
+ if (!strcmp(param->name, IFMARKINT) && !IFMarker) {
+ if (!strcmp(param->value, REJECT))
+ continue;
+ param->state &= ~PSTATE_NEGOTIATE;
+ if (iscsi_update_param_value(param, IRRELEVANT) < 0)
+ return -1;
+ TRACE(TRACE_PARAM, "Reset \"%s\" to \"%s\".\n",
+ param->name, param->value);
+ }
+ if (!strcmp(param->name, OFMARKINT) && !OFMarker) {
+ if (!strcmp(param->value, REJECT))
+ continue;
+ param->state &= ~PSTATE_NEGOTIATE;
+ if (iscsi_update_param_value(param, IRRELEVANT) < 0)
+ return -1;
+ TRACE(TRACE_PARAM, "Reset \"%s\" to \"%s\".\n",
+ param->name, param->value);
+ }
+ }
+
+ return 0;
+}
+
+int iscsi_decode_text_input(
+ u8 phase,
+ u8 sender,
+ char *textbuf,
+ u32 length,
+ struct iscsi_param_list *param_list)
+{
+ char *tmpbuf, *start = NULL, *end = NULL;
+
+ tmpbuf = kzalloc(length + 1, GFP_KERNEL);
+ if (!(tmpbuf)) {
+ printk(KERN_ERR "Unable to allocate memory for tmpbuf.\n");
+ return -1;
+ }
+
+ memcpy(tmpbuf, textbuf, length);
+ tmpbuf[length] = '\0';
+ start = tmpbuf;
+ end = (start + length);
+
+ while (start < end) {
+ char *key, *value;
+ struct iscsi_param *param;
+
+ if (iscsi_extract_key_value(start, &key, &value) < 0) {
+ kfree(tmpbuf);
+ return -1;
+ }
+
+ TRACE(TRACE_PARAM, "Got key: %s=%s\n", key, value);
+
+ if (phase & PHASE_SECURITY) {
+ if (iscsi_check_for_auth_key(key) > 0) {
+ char *tmpptr = key + strlen(key);
+ *tmpptr = '=';
+ kfree(tmpbuf);
+ return 1;
+ }
+ }
+
+ param = iscsi_check_key(key, phase, sender, param_list);
+ if (!(param)) {
+ if (iscsi_add_notunderstood_response(key,
+ value, param_list) < 0) {
+ kfree(tmpbuf);
+ return -1;
+ }
+ start += strlen(key) + strlen(value) + 2;
+ continue;
+ }
+ if (iscsi_check_value(param, value) < 0) {
+ kfree(tmpbuf);
+ return -1;
+ }
+
+ start += strlen(key) + strlen(value) + 2;
+
+ if (IS_PSTATE_PROPOSER(param)) {
+ if (iscsi_check_proposer_state(param, value) < 0) {
+ kfree(tmpbuf);
+ return -1;
+ }
+ SET_PSTATE_RESPONSE_GOT(param);
+ } else {
+ if (iscsi_check_acceptor_state(param, value) < 0) {
+ kfree(tmpbuf);
+ return -1;
+ }
+ SET_PSTATE_ACCEPTOR(param);
+ }
+ }
+
+ kfree(tmpbuf);
+ return 0;
+}
+
+int iscsi_encode_text_output(
+ u8 phase,
+ u8 sender,
+ char *textbuf,
+ u32 *length,
+ struct iscsi_param_list *param_list)
+{
+ char *output_buf = NULL;
+ struct iscsi_extra_response *er;
+ struct iscsi_param *param;
+
+ output_buf = textbuf + *length;
+
+ if (iscsi_enforce_integrity_rules(phase, param_list) < 0)
+ return -1;
+
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ if (!(param->sender & sender))
+ continue;
+ if (IS_PSTATE_ACCEPTOR(param) &&
+ !IS_PSTATE_RESPONSE_SENT(param) &&
+ !IS_PSTATE_REPLY_OPTIONAL(param) &&
+ (param->phase & phase)) {
+ *length += sprintf(output_buf, "%s=%s",
+ param->name, param->value);
+ *length += 1;
+ output_buf = textbuf + *length;
+ SET_PSTATE_RESPONSE_SENT(param);
+ TRACE(TRACE_PARAM, "Sending key: %s=%s\n",
+ param->name, param->value);
+ continue;
+ }
+ if (IS_PSTATE_NEGOTIATE(param) &&
+ !IS_PSTATE_ACCEPTOR(param) &&
+ !IS_PSTATE_PROPOSER(param) &&
+ (param->phase & phase)) {
+ *length += sprintf(output_buf, "%s=%s",
+ param->name, param->value);
+ *length += 1;
+ output_buf = textbuf + *length;
+ SET_PSTATE_PROPOSER(param);
+ iscsi_check_proposer_for_optional_reply(param);
+ TRACE(TRACE_PARAM, "Sending key: %s=%s\n",
+ param->name, param->value);
+ }
+ }
+
+ list_for_each_entry(er, ¶m_list->extra_response_list, er_list) {
+ *length += sprintf(output_buf, "%s=%s", er->key, er->value);
+ *length += 1;
+ output_buf = textbuf + *length;
+ TRACE(TRACE_PARAM, "Sending key: %s=%s\n", er->key, er->value);
+ }
+ iscsi_release_extra_responses(param_list);
+
+ return 0;
+}
+
+int iscsi_check_negotiated_keys(struct iscsi_param_list *param_list)
+{
+ int ret = 0;
+ struct iscsi_param *param;
+
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ if (IS_PSTATE_NEGOTIATE(param) &&
+ IS_PSTATE_PROPOSER(param) &&
+ !IS_PSTATE_RESPONSE_GOT(param) &&
+ !IS_PSTATE_REPLY_OPTIONAL(param) &&
+ !IS_PHASE_DECLARATIVE(param)) {
+ printk(KERN_ERR "No response for proposed key \"%s\".\n",
+ param->name);
+ ret = -1;
+ }
+ }
+
+ return ret;
+}
+
+int iscsi_change_param_value(
+ char *keyvalue,
+ struct iscsi_param_list *param_list,
+ int check_key)
+{
+ char *key = NULL, *value = NULL;
+ struct iscsi_param *param;
+ int sender = 0;
+
+ if (iscsi_extract_key_value(keyvalue, &key, &value) < 0)
+ return -1;
+
+ if (!check_key) {
+ param = __iscsi_check_key(keyvalue, sender, param_list);
+ if (!(param))
+ return -1;
+ } else {
+ param = iscsi_check_key(keyvalue, 0, sender, param_list);
+ if (!(param))
+ return -1;
+
+ param->set_param = 1;
+ if (iscsi_check_value(param, value) < 0) {
+ param->set_param = 0;
+ return -1;
+ }
+ param->set_param = 0;
+ }
+
+ if (iscsi_update_param_value(param, value) < 0)
+ return -1;
+
+ return 0;
+}
+
+void iscsi_set_connection_parameters(
+ struct iscsi_conn_ops *ops,
+ struct iscsi_param_list *param_list)
+{
+ char *tmpptr;
+ struct iscsi_param *param;
+
+ printk(KERN_INFO "---------------------------------------------------"
+ "---------------\n");
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ if (!IS_PSTATE_ACCEPTOR(param) && !IS_PSTATE_PROPOSER(param))
+ continue;
+ if (!strcmp(param->name, AUTHMETHOD)) {
+ printk(KERN_INFO "AuthMethod: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, HEADERDIGEST)) {
+ ops->HeaderDigest = !strcmp(param->value, CRC32C);
+ printk(KERN_INFO "HeaderDigest: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, DATADIGEST)) {
+ ops->DataDigest = !strcmp(param->value, CRC32C);
+ printk(KERN_INFO "DataDigest: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, MAXRECVDATASEGMENTLENGTH)) {
+ ops->MaxRecvDataSegmentLength =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "MaxRecvDataSegmentLength: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, OFMARKER)) {
+ ops->OFMarker = !strcmp(param->value, YES);
+ printk(KERN_INFO "OFMarker: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, IFMARKER)) {
+ ops->IFMarker = !strcmp(param->value, YES);
+ printk(KERN_INFO "IFMarker: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, OFMARKINT)) {
+ ops->OFMarkInt =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "OFMarkInt: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, IFMARKINT)) {
+ ops->IFMarkInt =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "IFMarkInt: %s\n",
+ param->value);
+ }
+ }
+ printk(KERN_INFO "----------------------------------------------------"
+ "--------------\n");
+}
+
+void iscsi_set_session_parameters(
+ struct iscsi_sess_ops *ops,
+ struct iscsi_param_list *param_list,
+ int leading)
+{
+ char *tmpptr;
+ struct iscsi_param *param;
+
+ printk(KERN_INFO "----------------------------------------------------"
+ "--------------\n");
+ list_for_each_entry(param, ¶m_list->param_list, p_list) {
+ if (!IS_PSTATE_ACCEPTOR(param) && !IS_PSTATE_PROPOSER(param))
+ continue;
+ if (!strcmp(param->name, INITIATORNAME)) {
+ if (!param->value)
+ continue;
+ if (leading)
+ snprintf(ops->InitiatorName,
+ sizeof(ops->InitiatorName),
+ "%s", param->value);
+ printk(KERN_INFO "InitiatorName: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, INITIATORALIAS)) {
+ if (!param->value)
+ continue;
+ snprintf(ops->InitiatorAlias,
+ sizeof(ops->InitiatorAlias),
+ "%s", param->value);
+ printk(KERN_INFO "InitiatorAlias: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, TARGETNAME)) {
+ if (!param->value)
+ continue;
+ if (leading)
+ snprintf(ops->TargetName,
+ sizeof(ops->TargetName),
+ "%s", param->value);
+ printk(KERN_INFO "TargetName: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, TARGETALIAS)) {
+ if (!param->value)
+ continue;
+ snprintf(ops->TargetAlias, sizeof(ops->TargetAlias),
+ "%s", param->value);
+ printk(KERN_INFO "TargetAlias: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, TARGETPORTALGROUPTAG)) {
+ ops->TargetPortalGroupTag =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "TargetPortalGroupTag: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, MAXCONNECTIONS)) {
+ ops->MaxConnections =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "MaxConnections: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, INITIALR2T)) {
+ ops->InitialR2T = !strcmp(param->value, YES);
+ printk(KERN_INFO "InitialR2T: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, IMMEDIATEDATA)) {
+ ops->ImmediateData = !strcmp(param->value, YES);
+ printk(KERN_INFO "ImmediateData: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, MAXBURSTLENGTH)) {
+ ops->MaxBurstLength =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "MaxBurstLength: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, FIRSTBURSTLENGTH)) {
+ ops->FirstBurstLength =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "FirstBurstLength: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, DEFAULTTIME2WAIT)) {
+ ops->DefaultTime2Wait =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "DefaultTime2Wait: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, DEFAULTTIME2RETAIN)) {
+ ops->DefaultTime2Retain =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "DefaultTime2Retain: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, MAXOUTSTANDINGR2T)) {
+ ops->MaxOutstandingR2T =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "MaxOutstandingR2T: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, DATAPDUINORDER)) {
+ ops->DataPDUInOrder = !strcmp(param->value, YES);
+ printk(KERN_INFO "DataPDUInOrder: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, DATASEQUENCEINORDER)) {
+ ops->DataSequenceInOrder = !strcmp(param->value, YES);
+ printk(KERN_INFO "DataSequenceInOrder: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, ERRORRECOVERYLEVEL)) {
+ ops->ErrorRecoveryLevel =
+ simple_strtoul(param->value, &tmpptr, 0);
+ printk(KERN_INFO "ErrorRecoveryLevel: %s\n",
+ param->value);
+ } else if (!strcmp(param->name, SESSIONTYPE)) {
+ ops->SessionType = !strcmp(param->value, DISCOVERY);
+ printk(KERN_INFO "SessionType: %s\n",
+ param->value);
+ }
+ }
+ printk(KERN_INFO "----------------------------------------------------"
+ "--------------\n");
+
+}
diff --git a/drivers/target/iscsi/iscsi_target_parameters.h b/drivers/target/iscsi/iscsi_target_parameters.h
new file mode 100644
index 0000000..e6bde0c
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_parameters.h
@@ -0,0 +1,271 @@
+#ifndef ISCSI_PARAMETERS_H
+#define ISCSI_PARAMETERS_H
+
+struct iscsi_extra_response {
+ char key[64];
+ char value[32];
+ struct list_head er_list;
+} ____cacheline_aligned;
+
+struct iscsi_param {
+ char *name;
+ char *value;
+ u8 set_param;
+ u8 phase;
+ u8 scope;
+ u8 sender;
+ u8 type;
+ u8 use;
+ u16 type_range;
+ u32 state;
+ struct list_head p_list;
+} ____cacheline_aligned;
+
+extern struct iscsi_global *iscsi_global;
+
+extern int iscsi_login_rx_data(struct iscsi_conn *, char *, int);
+extern int iscsi_login_tx_data(struct iscsi_conn *, char *, char *, int);
+extern void iscsi_dump_conn_ops(struct iscsi_conn_ops *);
+extern void iscsi_dump_sess_ops(struct iscsi_sess_ops *);
+extern void iscsi_print_params(struct iscsi_param_list *);
+extern int iscsi_create_default_params(struct iscsi_param_list **);
+extern int iscsi_set_keys_to_negotiate(int, struct iscsi_param_list *);
+extern int iscsi_set_keys_irrelevant_for_discovery(struct iscsi_param_list *);
+extern int iscsi_copy_param_list(struct iscsi_param_list **,
+ struct iscsi_param_list *, int);
+extern int iscsi_change_param_value(char *, struct iscsi_param_list *, int);
+extern void iscsi_release_param_list(struct iscsi_param_list *);
+extern struct iscsi_param *iscsi_find_param_from_key(char *, struct iscsi_param_list *);
+extern int iscsi_extract_key_value(char *, char **, char **);
+extern int iscsi_update_param_value(struct iscsi_param *, char *);
+extern int iscsi_decode_text_input(u8, u8, char *, u32, struct iscsi_param_list *);
+extern int iscsi_encode_text_output(u8, u8, char *, u32 *,
+ struct iscsi_param_list *);
+extern int iscsi_check_negotiated_keys(struct iscsi_param_list *);
+extern void iscsi_set_connection_parameters(struct iscsi_conn_ops *,
+ struct iscsi_param_list *);
+extern void iscsi_set_session_parameters(struct iscsi_sess_ops *,
+ struct iscsi_param_list *, int);
+
+#define YES "Yes"
+#define NO "No"
+#define ALL "All"
+#define IRRELEVANT "Irrelevant"
+#define NONE "None"
+#define NOTUNDERSTOOD "NotUnderstood"
+#define REJECT "Reject"
+
+/*
+ * The Parameter Names.
+ */
+#define AUTHMETHOD "AuthMethod"
+#define HEADERDIGEST "HeaderDigest"
+#define DATADIGEST "DataDigest"
+#define MAXCONNECTIONS "MaxConnections"
+#define SENDTARGETS "SendTargets"
+#define TARGETNAME "TargetName"
+#define INITIATORNAME "InitiatorName"
+#define TARGETALIAS "TargetAlias"
+#define INITIATORALIAS "InitiatorAlias"
+#define TARGETADDRESS "TargetAddress"
+#define TARGETPORTALGROUPTAG "TargetPortalGroupTag"
+#define INITIALR2T "InitialR2T"
+#define IMMEDIATEDATA "ImmediateData"
+#define MAXRECVDATASEGMENTLENGTH "MaxRecvDataSegmentLength"
+#define MAXBURSTLENGTH "MaxBurstLength"
+#define FIRSTBURSTLENGTH "FirstBurstLength"
+#define DEFAULTTIME2WAIT "DefaultTime2Wait"
+#define DEFAULTTIME2RETAIN "DefaultTime2Retain"
+#define MAXOUTSTANDINGR2T "MaxOutstandingR2T"
+#define DATAPDUINORDER "DataPDUInOrder"
+#define DATASEQUENCEINORDER "DataSequenceInOrder"
+#define ERRORRECOVERYLEVEL "ErrorRecoveryLevel"
+#define SESSIONTYPE "SessionType"
+#define IFMARKER "IFMarker"
+#define OFMARKER "OFMarker"
+#define IFMARKINT "IFMarkInt"
+#define OFMARKINT "OFMarkInt"
+#define X_EXTENSIONKEY "X-com.sbei.version"
+#define X_EXTENSIONKEY_CISCO_NEW "X-com.cisco.protocol"
+#define X_EXTENSIONKEY_CISCO_OLD "X-com.cisco.iscsi.draft"
+
+/*
+ * For AuthMethod.
+ */
+#define KRB5 "KRB5"
+#define SPKM1 "SPKM1"
+#define SPKM2 "SPKM2"
+#define SRP "SRP"
+#define CHAP "CHAP"
+
+/*
+ * Initial values for Parameter Negotiation.
+ */
+#define INITIAL_AUTHMETHOD CHAP
+#define INITIAL_HEADERDIGEST "CRC32C,None"
+#define INITIAL_DATADIGEST "CRC32C,None"
+#define INITIAL_MAXCONNECTIONS "1"
+#define INITIAL_SENDTARGETS ALL
+#define INITIAL_TARGETNAME "LIO.Target"
+#define INITIAL_INITIATORNAME "LIO.Initiator"
+#define INITIAL_TARGETALIAS "LIO Target"
+#define INITIAL_INITIATORALIAS "LIO Initiator"
+#define INITIAL_TARGETADDRESS "0.0.0.0:0000,0"
+#define INITIAL_TARGETPORTALGROUPTAG "1"
+#define INITIAL_INITIALR2T YES
+#define INITIAL_IMMEDIATEDATA YES
+#define INITIAL_MAXRECVDATASEGMENTLENGTH "8192"
+#define INITIAL_MAXBURSTLENGTH "262144"
+#define INITIAL_FIRSTBURSTLENGTH "65536"
+#define INITIAL_DEFAULTTIME2WAIT "2"
+#define INITIAL_DEFAULTTIME2RETAIN "20"
+#define INITIAL_MAXOUTSTANDINGR2T "1"
+#define INITIAL_DATAPDUINORDER YES
+#define INITIAL_DATASEQUENCEINORDER YES
+#define INITIAL_ERRORRECOVERYLEVEL "0"
+#define INITIAL_SESSIONTYPE NORMAL
+#define INITIAL_IFMARKER NO
+#define INITIAL_OFMARKER NO
+#define INITIAL_IFMARKINT "2048~65535"
+#define INITIAL_OFMARKINT "2048~65535"
+
+/*
+ * For [Header,Data]Digests.
+ */
+#define CRC32C "CRC32C"
+
+/*
+ * For SessionType.
+ */
+#define DISCOVERY "Discovery"
+#define NORMAL "Normal"
+
+/*
+ * struct iscsi_param->use
+ */
+#define USE_LEADING_ONLY 0x01
+#define USE_INITIAL_ONLY 0x02
+#define USE_ALL 0x04
+
+#define IS_USE_LEADING_ONLY(p) ((p)->use & USE_LEADING_ONLY)
+#define IS_USE_INITIAL_ONLY(p) ((p)->use & USE_INITIAL_ONLY)
+#define IS_USE_ALL(p) ((p)->use & USE_ALL)
+
+#define SET_USE_INITIAL_ONLY(p) ((p)->use |= USE_INITIAL_ONLY)
+
+/*
+ * struct iscsi_param->sender
+ */
+#define SENDER_INITIATOR 0x01
+#define SENDER_TARGET 0x02
+#define SENDER_BOTH 0x03
+/* Used in iscsi_check_key() */
+#define SENDER_RECEIVER 0x04
+
+#define IS_SENDER_INITIATOR(p) ((p)->sender & SENDER_INITIATOR)
+#define IS_SENDER_TARGET(p) ((p)->sender & SENDER_TARGET)
+#define IS_SENDER_BOTH(p) ((p)->sender & SENDER_BOTH)
+
+/*
+ * struct iscsi_param->scope
+ */
+#define SCOPE_CONNECTION_ONLY 0x01
+#define SCOPE_SESSION_WIDE 0x02
+
+#define IS_SCOPE_CONNECTION_ONLY(p) ((p)->scope & SCOPE_CONNECTION_ONLY)
+#define IS_SCOPE_SESSION_WIDE(p) ((p)->scope & SCOPE_SESSION_WIDE)
+
+/*
+ * struct iscsi_param->phase
+ */
+#define PHASE_SECURITY 0x01
+#define PHASE_OPERATIONAL 0x02
+#define PHASE_DECLARATIVE 0x04
+#define PHASE_FFP0 0x08
+
+#define IS_PHASE_SECURITY(p) ((p)->phase & PHASE_SECURITY)
+#define IS_PHASE_OPERATIONAL(p) ((p)->phase & PHASE_OPERATIONAL)
+#define IS_PHASE_DECLARATIVE(p) ((p)->phase & PHASE_DECLARATIVE)
+#define IS_PHASE_FFP0(p) ((p)->phase & PHASE_FFP0)
+
+/*
+ * struct iscsi_param->type
+ */
+#define TYPE_BOOL_AND 0x01
+#define TYPE_BOOL_OR 0x02
+#define TYPE_NUMBER 0x04
+#define TYPE_NUMBER_RANGE 0x08
+#define TYPE_STRING 0x10
+#define TYPE_VALUE_LIST 0x20
+
+#define IS_TYPE_BOOL_AND(p) ((p)->type & TYPE_BOOL_AND)
+#define IS_TYPE_BOOL_OR(p) ((p)->type & TYPE_BOOL_OR)
+#define IS_TYPE_NUMBER(p) ((p)->type & TYPE_NUMBER)
+#define IS_TYPE_NUMBER_RANGE(p) ((p)->type & TYPE_NUMBER_RANGE)
+#define IS_TYPE_STRING(p) ((p)->type & TYPE_STRING)
+#define IS_TYPE_VALUE_LIST(p) ((p)->type & TYPE_VALUE_LIST)
+
+/*
+ * struct iscsi_param->type_range
+ */
+#define TYPERANGE_BOOL_AND 0x0001
+#define TYPERANGE_BOOL_OR 0x0002
+#define TYPERANGE_0_TO_2 0x0004
+#define TYPERANGE_0_TO_3600 0x0008
+#define TYPERANGE_0_TO_32767 0x0010
+#define TYPERANGE_0_TO_65535 0x0020
+#define TYPERANGE_1_TO_65535 0x0040
+#define TYPERANGE_2_TO_3600 0x0080
+#define TYPERANGE_512_TO_16777215 0x0100
+#define TYPERANGE_AUTH 0x0200
+#define TYPERANGE_DIGEST 0x0400
+#define TYPERANGE_ISCSINAME 0x0800
+#define TYPERANGE_MARKINT 0x1000
+#define TYPERANGE_SESSIONTYPE 0x2000
+#define TYPERANGE_TARGETADDRESS 0x4000
+#define TYPERANGE_UTF8 0x8000
+
+#define IS_TYPERANGE_0_TO_2(p) ((p)->type_range & TYPERANGE_0_TO_2)
+#define IS_TYPERANGE_0_TO_3600(p) ((p)->type_range & TYPERANGE_0_TO_3600)
+#define IS_TYPERANGE_0_TO_32767(p) ((p)->type_range & TYPERANGE_0_TO_32767)
+#define IS_TYPERANGE_0_TO_65535(p) ((p)->type_range & TYPERANGE_0_TO_65535)
+#define IS_TYPERANGE_1_TO_65535(p) ((p)->type_range & TYPERANGE_1_TO_65535)
+#define IS_TYPERANGE_2_TO_3600(p) ((p)->type_range & TYPERANGE_2_TO_3600)
+#define IS_TYPERANGE_512_TO_16777215(p) ((p)->type_range & \
+ TYPERANGE_512_TO_16777215)
+#define IS_TYPERANGE_AUTH_PARAM(p) ((p)->type_range & TYPERANGE_AUTH)
+#define IS_TYPERANGE_DIGEST_PARAM(p) ((p)->type_range & TYPERANGE_DIGEST)
+#define IS_TYPERANGE_SESSIONTYPE(p) ((p)->type_range & \
+ TYPERANGE_SESSIONTYPE)
+
+/*
+ * struct iscsi_param->state
+ */
+#define PSTATE_ACCEPTOR 0x01
+#define PSTATE_NEGOTIATE 0x02
+#define PSTATE_PROPOSER 0x04
+#define PSTATE_IRRELEVANT 0x08
+#define PSTATE_REJECT 0x10
+#define PSTATE_REPLY_OPTIONAL 0x20
+#define PSTATE_RESPONSE_GOT 0x40
+#define PSTATE_RESPONSE_SENT 0x80
+
+#define IS_PSTATE_ACCEPTOR(p) ((p)->state & PSTATE_ACCEPTOR)
+#define IS_PSTATE_NEGOTIATE(p) ((p)->state & PSTATE_NEGOTIATE)
+#define IS_PSTATE_PROPOSER(p) ((p)->state & PSTATE_PROPOSER)
+#define IS_PSTATE_IRRELEVANT(p) ((p)->state & PSTATE_IRRELEVANT)
+#define IS_PSTATE_REJECT(p) ((p)->state & PSTATE_REJECT)
+#define IS_PSTATE_REPLY_OPTIONAL(p) ((p)->state & PSTATE_REPLY_OPTIONAL)
+#define IS_PSTATE_RESPONSE_GOT(p) ((p)->state & PSTATE_RESPONSE_GOT)
+#define IS_PSTATE_RESPONSE_SENT(p) ((p)->state & PSTATE_RESPONSE_SENT)
+
+#define SET_PSTATE_ACCEPTOR(p) ((p)->state |= PSTATE_ACCEPTOR)
+#define SET_PSTATE_NEGOTIATE(p) ((p)->state |= PSTATE_NEGOTIATE)
+#define SET_PSTATE_PROPOSER(p) ((p)->state |= PSTATE_PROPOSER)
+#define SET_PSTATE_IRRELEVANT(p) ((p)->state |= PSTATE_IRRELEVANT)
+#define SET_PSTATE_REJECT(p) ((p)->state |= PSTATE_REJECT)
+#define SET_PSTATE_REPLY_OPTIONAL(p) ((p)->state |= PSTATE_REPLY_OPTIONAL)
+#define SET_PSTATE_RESPONSE_GOT(p) ((p)->state |= PSTATE_RESPONSE_GOT)
+#define SET_PSTATE_RESPONSE_SENT(p) ((p)->state |= PSTATE_RESPONSE_SENT)
+
+#endif /* ISCSI_PARAMETERS_H */
diff --git a/drivers/target/iscsi/iscsi_target_tq.c b/drivers/target/iscsi/iscsi_target_tq.c
new file mode 100644
index 0000000..dc86601
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_tq.c
@@ -0,0 +1,552 @@
+/*******************************************************************************
+ * This file contains the iSCSI Login Thread and Thread Queue functions.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/kthread.h>
+#include <linux/list.h>
+#include <linux/bitmap.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_tq.h"
+
+LIST_HEAD(active_ts_list);
+LIST_HEAD(inactive_ts_list);
+DEFINE_SPINLOCK(active_ts_lock);
+DEFINE_SPINLOCK(inactive_ts_lock);
+DEFINE_SPINLOCK(ts_bitmap_lock);
+
+static void iscsi_add_ts_to_active_list(struct iscsi_thread_set *ts)
+{
+ spin_lock(&active_ts_lock);
+ list_add_tail(&ts->ts_list, &active_ts_list);
+ iscsi_global->active_ts++;
+ spin_unlock(&active_ts_lock);
+}
+
+extern void iscsi_add_ts_to_inactive_list(struct iscsi_thread_set *ts)
+{
+ spin_lock(&inactive_ts_lock);
+ list_add_tail(&ts->ts_list, &inactive_ts_list);
+ iscsi_global->inactive_ts++;
+ spin_unlock(&inactive_ts_lock);
+}
+
+static void iscsi_del_ts_from_active_list(struct iscsi_thread_set *ts)
+{
+ spin_lock(&active_ts_lock);
+ list_del(&ts->ts_list);
+ iscsi_global->active_ts--;
+ spin_unlock(&active_ts_lock);
+}
+
+static struct iscsi_thread_set *iscsi_get_ts_from_inactive_list(void)
+{
+ struct iscsi_thread_set *ts;
+
+ spin_lock(&inactive_ts_lock);
+ if (list_empty(&inactive_ts_list)) {
+ spin_unlock(&inactive_ts_lock);
+ return NULL;
+ }
+
+ list_for_each_entry(ts, &inactive_ts_list, ts_list)
+ break;
+
+ list_del(&ts->ts_list);
+ iscsi_global->inactive_ts--;
+ spin_unlock(&inactive_ts_lock);
+
+ return ts;
+}
+
+extern int iscsi_allocate_thread_sets(u32 thread_pair_count)
+{
+ int allocated_thread_pair_count = 0, i, thread_id;
+ struct iscsi_thread_set *ts = NULL;
+
+ for (i = 0; i < thread_pair_count; i++) {
+ ts = kzalloc(sizeof(struct iscsi_thread_set), GFP_KERNEL);
+ if (!(ts)) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " thread set.\n");
+ return allocated_thread_pair_count;
+ }
+ /*
+ * Locate the next available regision in the thread_set_bitmap
+ */
+ spin_lock(&ts_bitmap_lock);
+ thread_id = bitmap_find_free_region(iscsi_global->ts_bitmap,
+ iscsi_global->ts_bitmap_count, get_order(1));
+ spin_unlock(&ts_bitmap_lock);
+ if (thread_id < 0) {
+ printk(KERN_ERR "bitmap_find_free_region() failed for"
+ " thread_set_bitmap\n");
+ kfree(ts);
+ return allocated_thread_pair_count;
+ }
+
+ ts->thread_id = thread_id;
+ ts->status = ISCSI_THREAD_SET_FREE;
+ INIT_LIST_HEAD(&ts->ts_list);
+ spin_lock_init(&ts->ts_state_lock);
+ init_completion(&ts->rx_post_start_comp);
+ init_completion(&ts->tx_post_start_comp);
+ init_completion(&ts->rx_restart_comp);
+ init_completion(&ts->tx_restart_comp);
+ init_completion(&ts->rx_start_comp);
+ init_completion(&ts->tx_start_comp);
+
+ ts->create_threads = 1;
+ ts->tx_thread = kthread_run(iscsi_target_tx_thread, ts, "%s",
+ ISCSI_TX_THREAD_NAME);
+ if (IS_ERR(ts->tx_thread)) {
+ dump_stack();
+ printk(KERN_ERR "Unable to start iscsi_target_tx_thread\n");
+ break;
+ }
+
+ ts->rx_thread = kthread_run(iscsi_target_rx_thread, ts, "%s",
+ ISCSI_RX_THREAD_NAME);
+ if (IS_ERR(ts->rx_thread)) {
+ kthread_stop(ts->tx_thread);
+ printk(KERN_ERR "Unable to start iscsi_target_rx_thread\n");
+ break;
+ }
+ ts->create_threads = 0;
+
+ iscsi_add_ts_to_inactive_list(ts);
+ allocated_thread_pair_count++;
+ }
+
+ printk(KERN_INFO "Spawned %d thread set(s) (%d total threads).\n",
+ allocated_thread_pair_count, allocated_thread_pair_count * 2);
+ return allocated_thread_pair_count;
+}
+
+extern void iscsi_deallocate_thread_sets(void)
+{
+ u32 released_count = 0;
+ struct iscsi_thread_set *ts = NULL;
+
+ while ((ts = iscsi_get_ts_from_inactive_list())) {
+
+ spin_lock_bh(&ts->ts_state_lock);
+ ts->status = ISCSI_THREAD_SET_DIE;
+ spin_unlock_bh(&ts->ts_state_lock);
+
+ if (ts->rx_thread) {
+ send_sig(SIGINT, ts->rx_thread, 1);
+ kthread_stop(ts->rx_thread);
+ }
+ if (ts->tx_thread) {
+ send_sig(SIGINT, ts->tx_thread, 1);
+ kthread_stop(ts->tx_thread);
+ }
+ /*
+ * Release this thread_id in the thread_set_bitmap
+ */
+ spin_lock(&ts_bitmap_lock);
+ bitmap_release_region(iscsi_global->ts_bitmap,
+ ts->thread_id, get_order(1));
+ spin_unlock(&ts_bitmap_lock);
+
+ released_count++;
+ kfree(ts);
+ }
+
+ if (released_count)
+ printk(KERN_INFO "Stopped %d thread set(s) (%d total threads)."
+ "\n", released_count, released_count * 2);
+}
+
+static void iscsi_deallocate_extra_thread_sets(void)
+{
+ u32 orig_count, released_count = 0;
+ struct iscsi_thread_set *ts = NULL;
+
+ orig_count = TARGET_THREAD_SET_COUNT;
+
+ while ((iscsi_global->inactive_ts + 1) > orig_count) {
+ ts = iscsi_get_ts_from_inactive_list();
+ if (!(ts))
+ break;
+
+ spin_lock_bh(&ts->ts_state_lock);
+ ts->status = ISCSI_THREAD_SET_DIE;
+ spin_unlock_bh(&ts->ts_state_lock);
+
+ if (ts->rx_thread) {
+ send_sig(SIGINT, ts->rx_thread, 1);
+ kthread_stop(ts->rx_thread);
+ }
+ if (ts->tx_thread) {
+ send_sig(SIGINT, ts->tx_thread, 1);
+ kthread_stop(ts->tx_thread);
+ }
+ /*
+ * Release this thread_id in the thread_set_bitmap
+ */
+ spin_lock(&ts_bitmap_lock);
+ bitmap_release_region(iscsi_global->ts_bitmap,
+ ts->thread_id, get_order(1));
+ spin_unlock(&ts_bitmap_lock);
+
+ released_count++;
+ kfree(ts);
+ }
+
+ if (released_count) {
+ printk(KERN_INFO "Stopped %d thread set(s) (%d total threads)."
+ "\n", released_count, released_count * 2);
+ }
+}
+
+void iscsi_activate_thread_set(struct iscsi_conn *conn, struct iscsi_thread_set *ts)
+{
+ iscsi_add_ts_to_active_list(ts);
+
+ spin_lock_bh(&ts->ts_state_lock);
+ conn->thread_set = ts;
+ ts->conn = conn;
+ spin_unlock_bh(&ts->ts_state_lock);
+ /*
+ * Start up the RX thread and wait on rx_post_start_comp. The RX
+ * Thread will then do the same for the TX Thread in
+ * iscsi_rx_thread_pre_handler().
+ */
+ complete(&ts->rx_start_comp);
+ wait_for_completion(&ts->rx_post_start_comp);
+}
+
+struct iscsi_thread_set *iscsi_get_thread_set(void)
+{
+ int allocate_ts = 0;
+ struct completion comp;
+ struct iscsi_thread_set *ts = NULL;
+ /*
+ * If no inactive thread set is available on the first call to
+ * iscsi_get_ts_from_inactive_list(), sleep for a second and
+ * try again. If still none are available after two attempts,
+ * allocate a set ourselves.
+ */
+get_set:
+ ts = iscsi_get_ts_from_inactive_list();
+ if (!(ts)) {
+ if (allocate_ts == 2)
+ iscsi_allocate_thread_sets(1);
+
+ init_completion(&comp);
+ wait_for_completion_timeout(&comp, 1 * HZ);
+
+ allocate_ts++;
+ goto get_set;
+ }
+
+ ts->delay_inactive = 1;
+ ts->signal_sent = 0;
+ ts->thread_count = 2;
+ init_completion(&ts->rx_restart_comp);
+ init_completion(&ts->tx_restart_comp);
+
+ return ts;
+}
+
+void iscsi_set_thread_clear(struct iscsi_conn *conn, u8 thread_clear)
+{
+ struct iscsi_thread_set *ts = NULL;
+
+ if (!conn->thread_set) {
+ printk(KERN_ERR "struct iscsi_conn->thread_set is NULL\n");
+ return;
+ }
+ ts = conn->thread_set;
+
+ spin_lock_bh(&ts->ts_state_lock);
+ ts->thread_clear &= ~thread_clear;
+
+ if ((thread_clear & ISCSI_CLEAR_RX_THREAD) &&
+ (ts->blocked_threads & ISCSI_BLOCK_RX_THREAD))
+ complete(&ts->rx_restart_comp);
+ else if ((thread_clear & ISCSI_CLEAR_TX_THREAD) &&
+ (ts->blocked_threads & ISCSI_BLOCK_TX_THREAD))
+ complete(&ts->tx_restart_comp);
+ spin_unlock_bh(&ts->ts_state_lock);
+}
+
+void iscsi_set_thread_set_signal(struct iscsi_conn *conn, u8 signal_sent)
+{
+ struct iscsi_thread_set *ts = NULL;
+
+ if (!conn->thread_set) {
+ printk(KERN_ERR "struct iscsi_conn->thread_set is NULL\n");
+ return;
+ }
+ ts = conn->thread_set;
+
+ spin_lock_bh(&ts->ts_state_lock);
+ ts->signal_sent |= signal_sent;
+ spin_unlock_bh(&ts->ts_state_lock);
+}
+
+int iscsi_release_thread_set(struct iscsi_conn *conn)
+{
+ int thread_called = 0;
+ struct iscsi_thread_set *ts = NULL;
+
+ if (!conn || !conn->thread_set) {
+ printk(KERN_ERR "connection or thread set pointer is NULL\n");
+ BUG();
+ }
+ ts = conn->thread_set;
+
+ spin_lock_bh(&ts->ts_state_lock);
+ ts->status = ISCSI_THREAD_SET_RESET;
+
+ if (!(strncmp(current->comm, ISCSI_RX_THREAD_NAME,
+ strlen(ISCSI_RX_THREAD_NAME))))
+ thread_called = ISCSI_RX_THREAD;
+ else if (!(strncmp(current->comm, ISCSI_TX_THREAD_NAME,
+ strlen(ISCSI_TX_THREAD_NAME))))
+ thread_called = ISCSI_TX_THREAD;
+
+ if (ts->rx_thread && (thread_called == ISCSI_TX_THREAD) &&
+ (ts->thread_clear & ISCSI_CLEAR_RX_THREAD)) {
+
+ if (!(ts->signal_sent & ISCSI_SIGNAL_RX_THREAD)) {
+ send_sig(SIGINT, ts->rx_thread, 1);
+ ts->signal_sent |= ISCSI_SIGNAL_RX_THREAD;
+ }
+ ts->blocked_threads |= ISCSI_BLOCK_RX_THREAD;
+ spin_unlock_bh(&ts->ts_state_lock);
+ wait_for_completion(&ts->rx_restart_comp);
+ spin_lock_bh(&ts->ts_state_lock);
+ ts->blocked_threads &= ~ISCSI_BLOCK_RX_THREAD;
+ }
+ if (ts->tx_thread && (thread_called == ISCSI_RX_THREAD) &&
+ (ts->thread_clear & ISCSI_CLEAR_TX_THREAD)) {
+
+ if (!(ts->signal_sent & ISCSI_SIGNAL_TX_THREAD)) {
+ send_sig(SIGINT, ts->tx_thread, 1);
+ ts->signal_sent |= ISCSI_SIGNAL_TX_THREAD;
+ }
+ ts->blocked_threads |= ISCSI_BLOCK_TX_THREAD;
+ spin_unlock_bh(&ts->ts_state_lock);
+ wait_for_completion(&ts->tx_restart_comp);
+ spin_lock_bh(&ts->ts_state_lock);
+ ts->blocked_threads &= ~ISCSI_BLOCK_TX_THREAD;
+ }
+
+ conn->thread_set = NULL;
+ ts->conn = NULL;
+ ts->status = ISCSI_THREAD_SET_FREE;
+ spin_unlock_bh(&ts->ts_state_lock);
+
+ return 0;
+}
+
+int iscsi_thread_set_force_reinstatement(struct iscsi_conn *conn)
+{
+ struct iscsi_thread_set *ts;
+
+ if (!conn->thread_set)
+ return -1;
+ ts = conn->thread_set;
+
+ spin_lock_bh(&ts->ts_state_lock);
+ if (ts->status != ISCSI_THREAD_SET_ACTIVE) {
+ spin_unlock_bh(&ts->ts_state_lock);
+ return -1;
+ }
+
+ if (ts->tx_thread && (!(ts->signal_sent & ISCSI_SIGNAL_TX_THREAD))) {
+ send_sig(SIGINT, ts->tx_thread, 1);
+ ts->signal_sent |= ISCSI_SIGNAL_TX_THREAD;
+ }
+ if (ts->rx_thread && (!(ts->signal_sent & ISCSI_SIGNAL_RX_THREAD))) {
+ send_sig(SIGINT, ts->rx_thread, 1);
+ ts->signal_sent |= ISCSI_SIGNAL_RX_THREAD;
+ }
+ spin_unlock_bh(&ts->ts_state_lock);
+
+ return 0;
+}
+
+static void iscsi_check_to_add_additional_sets(void)
+{
+ int thread_sets_add;
+
+ spin_lock(&inactive_ts_lock);
+ thread_sets_add = iscsi_global->inactive_ts;
+ spin_unlock(&inactive_ts_lock);
+ if (thread_sets_add == 1)
+ iscsi_allocate_thread_sets(1);
+}
+
+static int iscsi_signal_thread_pre_handler(struct iscsi_thread_set *ts)
+{
+ spin_lock_bh(&ts->ts_state_lock);
+ if ((ts->status == ISCSI_THREAD_SET_DIE) || signal_pending(current)) {
+ spin_unlock_bh(&ts->ts_state_lock);
+ return -1;
+ }
+ spin_unlock_bh(&ts->ts_state_lock);
+
+ return 0;
+}
+
+struct iscsi_conn *iscsi_rx_thread_pre_handler(struct iscsi_thread_set *ts)
+{
+ int ret;
+
+ spin_lock_bh(&ts->ts_state_lock);
+ if (ts->create_threads) {
+ spin_unlock_bh(&ts->ts_state_lock);
+ goto sleep;
+ }
+
+ flush_signals(current);
+
+ if (ts->delay_inactive && (--ts->thread_count == 0)) {
+ spin_unlock_bh(&ts->ts_state_lock);
+ iscsi_del_ts_from_active_list(ts);
+
+ if (!iscsi_global->in_shutdown)
+ iscsi_deallocate_extra_thread_sets();
+
+ iscsi_add_ts_to_inactive_list(ts);
+ spin_lock_bh(&ts->ts_state_lock);
+ }
+
+ if ((ts->status == ISCSI_THREAD_SET_RESET) &&
+ (ts->thread_clear & ISCSI_CLEAR_RX_THREAD))
+ complete(&ts->rx_restart_comp);
+
+ ts->thread_clear &= ~ISCSI_CLEAR_RX_THREAD;
+ spin_unlock_bh(&ts->ts_state_lock);
+sleep:
+ ret = wait_for_completion_interruptible(&ts->rx_start_comp);
+ if (ret != 0)
+ return NULL;
+
+ if (iscsi_signal_thread_pre_handler(ts) < 0)
+ return NULL;
+
+ if (!ts->conn) {
+ printk(KERN_ERR "struct iscsi_thread_set->conn is NULL for"
+ " thread_id: %d, going back to sleep\n", ts->thread_id);
+ goto sleep;
+ }
+ iscsi_check_to_add_additional_sets();
+ /*
+ * The RX Thread starts up the TX Thread and sleeps.
+ */
+ ts->thread_clear |= ISCSI_CLEAR_RX_THREAD;
+ complete(&ts->tx_start_comp);
+ wait_for_completion(&ts->tx_post_start_comp);
+
+ return ts->conn;
+}
+
+struct iscsi_conn *iscsi_tx_thread_pre_handler(struct iscsi_thread_set *ts)
+{
+ int ret;
+
+ spin_lock_bh(&ts->ts_state_lock);
+ if (ts->create_threads) {
+ spin_unlock_bh(&ts->ts_state_lock);
+ goto sleep;
+ }
+
+ flush_signals(current);
+
+ if (ts->delay_inactive && (--ts->thread_count == 0)) {
+ spin_unlock_bh(&ts->ts_state_lock);
+ iscsi_del_ts_from_active_list(ts);
+
+ if (!iscsi_global->in_shutdown)
+ iscsi_deallocate_extra_thread_sets();
+
+ iscsi_add_ts_to_inactive_list(ts);
+ spin_lock_bh(&ts->ts_state_lock);
+ }
+ if ((ts->status == ISCSI_THREAD_SET_RESET) &&
+ (ts->thread_clear & ISCSI_CLEAR_TX_THREAD))
+ complete(&ts->tx_restart_comp);
+
+ ts->thread_clear &= ~ISCSI_CLEAR_TX_THREAD;
+ spin_unlock_bh(&ts->ts_state_lock);
+sleep:
+ ret = wait_for_completion_interruptible(&ts->tx_start_comp);
+ if (ret != 0)
+ return NULL;
+
+ if (iscsi_signal_thread_pre_handler(ts) < 0)
+ return NULL;
+
+ if (!ts->conn) {
+ printk(KERN_ERR "struct iscsi_thread_set->conn is NULL for "
+ " thread_id: %d, going back to sleep\n",
+ ts->thread_id);
+ goto sleep;
+ }
+
+ iscsi_check_to_add_additional_sets();
+ /*
+ * From the TX thread, up the tx_post_start_comp that the RX Thread is
+ * sleeping on in iscsi_rx_thread_pre_handler(), then up the
+ * rx_post_start_comp that iscsi_activate_thread_set() is sleeping on.
+ */
+ ts->thread_clear |= ISCSI_CLEAR_TX_THREAD;
+ complete(&ts->tx_post_start_comp);
+ complete(&ts->rx_post_start_comp);
+
+ spin_lock_bh(&ts->ts_state_lock);
+ ts->status = ISCSI_THREAD_SET_ACTIVE;
+ spin_unlock_bh(&ts->ts_state_lock);
+
+ return ts->conn;
+}
+
+int iscsi_thread_set_init(void)
+{
+ int size;
+
+ iscsi_global->ts_bitmap_count = ISCSI_TS_BITMAP_BITS;
+
+ size = BITS_TO_LONGS(iscsi_global->ts_bitmap_count) * sizeof(long);
+ iscsi_global->ts_bitmap = kzalloc(size, GFP_KERNEL);
+ if (!(iscsi_global->ts_bitmap)) {
+ printk(KERN_ERR "Unable to allocate iscsi_global->ts_bitmap\n");
+ return -ENOMEM;
+ }
+
+ spin_lock_init(&active_ts_lock);
+ spin_lock_init(&inactive_ts_lock);
+ spin_lock_init(&ts_bitmap_lock);
+ INIT_LIST_HEAD(&active_ts_list);
+ INIT_LIST_HEAD(&inactive_ts_list);
+
+ return 0;
+}
+
+void iscsi_thread_set_free(void)
+{
+ kfree(iscsi_global->ts_bitmap);
+}
diff --git a/drivers/target/iscsi/iscsi_target_tq.h b/drivers/target/iscsi/iscsi_target_tq.h
new file mode 100644
index 0000000..e1fdc2c
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_tq.h
@@ -0,0 +1,89 @@
+#ifndef ISCSI_THREAD_QUEUE_H
+#define ISCSI_THREAD_QUEUE_H
+
+/*
+ * Defines for thread sets.
+ */
+extern int iscsi_thread_set_force_reinstatement(struct iscsi_conn *);
+extern void iscsi_add_ts_to_inactive_list(struct iscsi_thread_set *);
+extern int iscsi_allocate_thread_sets(u32);
+extern void iscsi_deallocate_thread_sets(void);
+extern void iscsi_activate_thread_set(struct iscsi_conn *, struct iscsi_thread_set *);
+extern struct iscsi_thread_set *iscsi_get_thread_set(void);
+extern void iscsi_set_thread_clear(struct iscsi_conn *, u8);
+extern void iscsi_set_thread_set_signal(struct iscsi_conn *, u8);
+extern int iscsi_release_thread_set(struct iscsi_conn *);
+extern struct iscsi_conn *iscsi_rx_thread_pre_handler(struct iscsi_thread_set *);
+extern struct iscsi_conn *iscsi_tx_thread_pre_handler(struct iscsi_thread_set *);
+extern int iscsi_thread_set_init(void);
+extern void iscsi_thread_set_free(void);
+
+extern int iscsi_target_tx_thread(void *);
+extern int iscsi_target_rx_thread(void *);
+extern struct iscsi_global *iscsi_global;
+
+#define TARGET_THREAD_SET_COUNT 4
+
+#define ISCSI_RX_THREAD 1
+#define ISCSI_TX_THREAD 2
+#define ISCSI_RX_THREAD_NAME "iscsi_trx"
+#define ISCSI_TX_THREAD_NAME "iscsi_ttx"
+#define ISCSI_BLOCK_RX_THREAD 0x1
+#define ISCSI_BLOCK_TX_THREAD 0x2
+#define ISCSI_CLEAR_RX_THREAD 0x1
+#define ISCSI_CLEAR_TX_THREAD 0x2
+#define ISCSI_SIGNAL_RX_THREAD 0x1
+#define ISCSI_SIGNAL_TX_THREAD 0x2
+
+/* struct iscsi_thread_set->status */
+#define ISCSI_THREAD_SET_FREE 1
+#define ISCSI_THREAD_SET_ACTIVE 2
+#define ISCSI_THREAD_SET_DIE 3
+#define ISCSI_THREAD_SET_RESET 4
+#define ISCSI_THREAD_SET_DEALLOCATE_THREADS 5
+
+/* By default allow a maximum of 32K iSCSI connections */
+#define ISCSI_TS_BITMAP_BITS 32768
+
+struct iscsi_thread_set {
+ /* flags used for blocking and restarting sets */
+ int blocked_threads;
+ /* flag for creating threads */
+ int create_threads;
+ /* flag for delaying readding to inactive list */
+ int delay_inactive;
+ /* status for thread set */
+ int status;
+ /* which threads have had signals sent */
+ int signal_sent;
+ /* flag for which threads exited first */
+ int thread_clear;
+ /* Active threads in the thread set */
+ int thread_count;
+ /* Unique thread ID */
+ u32 thread_id;
+ /* pointer to connection if set is active */
+ struct iscsi_conn *conn;
+ /* used for controlling ts state accesses */
+ spinlock_t ts_state_lock;
+ /* Used for rx side post startup */
+ struct completion rx_post_start_comp;
+ /* Used for tx side post startup */
+ struct completion tx_post_start_comp;
+ /* used for restarting thread queue */
+ struct completion rx_restart_comp;
+ /* used for restarting thread queue */
+ struct completion tx_restart_comp;
+ /* used for normal unused blocking */
+ struct completion rx_start_comp;
+ /* used for normal unused blocking */
+ struct completion tx_start_comp;
+ /* OS descriptor for rx thread */
+ struct task_struct *rx_thread;
+ /* OS descriptor for tx thread */
+ struct task_struct *tx_thread;
+ /* struct iscsi_thread_set in list list head*/
+ struct list_head ts_list;
+};
+
+#endif /*** ISCSI_THREAD_QUEUE_H ***/
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 09/14] iscsi-target: Add CHAP Authentication support using libcrypto
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (8 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 08/14] iscsi-target: Add iSCSI Login Negotiation and Parameter logic Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 10/14] iscsi-target: Add Sequence/PDU list + DataIN response logic Nicholas A. Bellinger
` (5 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch adds support for libcrypto md5 based iSCSI CHAP authentication
support for iscsi_target_mod. This includes support for mutual and one-way
NodeACL authentication for SessionType=Normal and SessionType=Discovery
via /sys/kernel/config/target/iscsi.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_auth.c | 496 ++++++++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_auth.h | 32 ++
2 files changed, 528 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_auth.c
create mode 100644 drivers/target/iscsi/iscsi_target_auth.h
diff --git a/drivers/target/iscsi/iscsi_target_auth.c b/drivers/target/iscsi/iscsi_target_auth.c
new file mode 100644
index 0000000..fc5e067
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_auth.c
@@ -0,0 +1,496 @@
+/*******************************************************************************
+ * This file houses the main functions for the iSCSI CHAP support
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/string.h>
+#include <linux/crypto.h>
+#include <linux/err.h>
+#include <linux/scatterlist.h>
+
+#include "iscsi_target_core.h"
+#include "iscsi_target_nego.h"
+#include "iscsi_target_auth.h"
+
+#ifdef DEBUG_CHAP
+#define PRINT(x...) printk(KERN_INFO x)
+#else
+#define PRINT(x...)
+#endif
+
+static unsigned char chap_asciihex_to_binaryhex(unsigned char val[2])
+{
+ unsigned char result = 0;
+ /*
+ * MSB
+ */
+ if ((val[0] >= 'a') && (val[0] <= 'f'))
+ result = ((val[0] - 'a' + 10) & 0xf) << 4;
+ else
+ if ((val[0] >= 'A') && (val[0] <= 'F'))
+ result = ((val[0] - 'A' + 10) & 0xf) << 4;
+ else /* digit */
+ result = ((val[0] - '0') & 0xf) << 4;
+ /*
+ * LSB
+ */
+ if ((val[1] >= 'a') && (val[1] <= 'f'))
+ result |= ((val[1] - 'a' + 10) & 0xf);
+ else
+ if ((val[1] >= 'A') && (val[1] <= 'F'))
+ result |= ((val[1] - 'A' + 10) & 0xf);
+ else /* digit */
+ result |= ((val[1] - '0') & 0xf);
+
+ return result;
+}
+
+static int chap_string_to_hex(unsigned char *dst, unsigned char *src, int len)
+{
+ int i, j = 0;
+
+ for (i = 0; i < len; i += 2) {
+ dst[j++] = (unsigned char) chap_asciihex_to_binaryhex(&src[i]);
+ }
+
+ dst[j] = '\0';
+ return j;
+}
+
+static void chap_binaryhex_to_asciihex(char *dst, char *src, int src_len)
+{
+ int i;
+
+ for (i = 0; i < src_len; i++) {
+ sprintf(&dst[i*2], "%02x", (int) src[i] & 0xff);
+ }
+}
+
+static void chap_set_random(char *data, int length)
+{
+ long r;
+ unsigned n;
+
+ while (length > 0) {
+ get_random_bytes(&r, sizeof(long));
+ r = r ^ (r >> 8);
+ r = r ^ (r >> 4);
+ n = r & 0x7;
+
+ get_random_bytes(&r, sizeof(long));
+ r = r ^ (r >> 8);
+ r = r ^ (r >> 5);
+ n = (n << 3) | (r & 0x7);
+
+ get_random_bytes(&r, sizeof(long));
+ r = r ^ (r >> 8);
+ r = r ^ (r >> 5);
+ n = (n << 2) | (r & 0x3);
+
+ *data++ = n;
+ length--;
+ }
+}
+
+static void chap_gen_challenge(
+ struct iscsi_conn *conn,
+ int caller,
+ char *c_str,
+ unsigned int *c_len)
+{
+ unsigned char challenge_asciihex[CHAP_CHALLENGE_LENGTH * 2 + 1];
+ struct iscsi_chap *chap = (struct iscsi_chap *) conn->auth_protocol;
+
+ memset(challenge_asciihex, 0, CHAP_CHALLENGE_LENGTH * 2 + 1);
+
+ chap_set_random(chap->challenge, CHAP_CHALLENGE_LENGTH);
+ chap_binaryhex_to_asciihex(challenge_asciihex, chap->challenge,
+ CHAP_CHALLENGE_LENGTH);
+ /*
+ * Set CHAP_C, and copy the generated challenge into c_str.
+ */
+ *c_len += sprintf(c_str + *c_len, "CHAP_C=0x%s", challenge_asciihex);
+ *c_len += 1;
+
+ PRINT("[%s] Sending CHAP_C=0x%s\n\n", (caller) ? "server" : "client",
+ challenge_asciihex);
+}
+
+
+static struct iscsi_chap *chap_server_open(
+ struct iscsi_conn *conn,
+ struct iscsi_node_auth *auth,
+ const char *A_str,
+ char *aic_str,
+ unsigned int *aic_len)
+{
+ struct iscsi_chap *chap;
+
+ if (!(auth->naf_flags & NAF_USERID_SET) ||
+ !(auth->naf_flags & NAF_PASSWORD_SET)) {
+ printk(KERN_ERR "CHAP user or password not set for"
+ " Initiator ACL\n");
+ return NULL;
+ }
+
+ conn->auth_protocol = kzalloc(sizeof(struct iscsi_chap), GFP_KERNEL);
+ if (!conn->auth_protocol)
+ return NULL;
+
+ chap = (struct iscsi_chap *) conn->auth_protocol;
+ /*
+ * We only support MD5 MDA presently.
+ */
+ if (strncmp(A_str, "CHAP_A=5", 8)) {
+ printk(KERN_ERR "CHAP_A is not MD5.\n");
+ return NULL;
+ }
+ PRINT("[server] Got CHAP_A=5\n");
+ /*
+ * Send back CHAP_A set to MD5.
+ */
+ *aic_len = sprintf(aic_str, "CHAP_A=5");
+ *aic_len += 1;
+ chap->digest_type = CHAP_DIGEST_MD5;
+ PRINT("[server] Sending CHAP_A=%d\n", chap->digest_type);
+ /*
+ * Set Identifier.
+ */
+ chap->id = ISCSI_TPG_C(conn)->tpg_chap_id++;
+ *aic_len += sprintf(aic_str + *aic_len, "CHAP_I=%d", chap->id);
+ *aic_len += 1;
+ PRINT("[server] Sending CHAP_I=%d\n", chap->id);
+ /*
+ * Generate Challenge.
+ */
+ chap_gen_challenge(conn, 1, aic_str, aic_len);
+
+ return chap;
+}
+
+static void chap_close(struct iscsi_conn *conn)
+{
+ kfree(conn->auth_protocol);
+ conn->auth_protocol = NULL;
+}
+
+static int chap_server_compute_md5(
+ struct iscsi_conn *conn,
+ struct iscsi_node_auth *auth,
+ char *nr_in_ptr,
+ char *nr_out_ptr,
+ unsigned int *nr_out_len)
+{
+ char *endptr;
+ unsigned char id, digest[MD5_SIGNATURE_SIZE];
+ unsigned char type, response[MD5_SIGNATURE_SIZE * 2 + 2];
+ unsigned char identifier[10], *challenge = NULL;
+ unsigned char *challenge_binhex = NULL;
+ unsigned char client_digest[MD5_SIGNATURE_SIZE];
+ unsigned char server_digest[MD5_SIGNATURE_SIZE];
+ unsigned char chap_n[MAX_CHAP_N_SIZE], chap_r[MAX_RESPONSE_LENGTH];
+ struct iscsi_chap *chap = (struct iscsi_chap *) conn->auth_protocol;
+ struct crypto_hash *tfm;
+ struct hash_desc desc;
+ struct scatterlist sg;
+ int auth_ret = -1, ret, challenge_len;
+
+ memset(identifier, 0, 10);
+ memset(chap_n, 0, MAX_CHAP_N_SIZE);
+ memset(chap_r, 0, MAX_RESPONSE_LENGTH);
+ memset(digest, 0, MD5_SIGNATURE_SIZE);
+ memset(response, 0, MD5_SIGNATURE_SIZE * 2 + 2);
+ memset(client_digest, 0, MD5_SIGNATURE_SIZE);
+ memset(server_digest, 0, MD5_SIGNATURE_SIZE);
+
+ challenge = kzalloc(CHAP_CHALLENGE_STR_LEN, GFP_KERNEL);
+ if (!challenge) {
+ printk(KERN_ERR "Unable to allocate challenge buffer\n");
+ goto out;
+ }
+
+ challenge_binhex = kzalloc(CHAP_CHALLENGE_STR_LEN, GFP_KERNEL);
+ if (!challenge_binhex) {
+ printk(KERN_ERR "Unable to allocate challenge_binhex buffer\n");
+ goto out;
+ }
+ /*
+ * Extract CHAP_N.
+ */
+ if (extract_param(nr_in_ptr, "CHAP_N", MAX_CHAP_N_SIZE, chap_n,
+ &type) < 0) {
+ printk(KERN_ERR "Could not find CHAP_N.\n");
+ goto out;
+ }
+ if (type == HEX) {
+ printk(KERN_ERR "Could not find CHAP_N.\n");
+ goto out;
+ }
+
+ if (memcmp(chap_n, auth->userid, strlen(auth->userid)) != 0) {
+ printk(KERN_ERR "CHAP_N values do not match!\n");
+ goto out;
+ }
+ PRINT("[server] Got CHAP_N=%s\n", chap_n);
+ /*
+ * Extract CHAP_R.
+ */
+ if (extract_param(nr_in_ptr, "CHAP_R", MAX_RESPONSE_LENGTH, chap_r,
+ &type) < 0) {
+ printk(KERN_ERR "Could not find CHAP_R.\n");
+ goto out;
+ }
+ if (type != HEX) {
+ printk(KERN_ERR "Could not find CHAP_R.\n");
+ goto out;
+ }
+
+ PRINT("[server] Got CHAP_R=%s\n", chap_r);
+ chap_string_to_hex(client_digest, chap_r, strlen(chap_r));
+
+ tfm = crypto_alloc_hash("md5", 0, CRYPTO_ALG_ASYNC);
+ if (IS_ERR(tfm)) {
+ printk(KERN_ERR "Unable to allocate struct crypto_hash\n");
+ goto out;
+ }
+ desc.tfm = tfm;
+ desc.flags = 0;
+
+ ret = crypto_hash_init(&desc);
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_init() failed\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+
+ sg_init_one(&sg, (void *)&chap->id, 1);
+ ret = crypto_hash_update(&desc, &sg, 1);
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_update() failed for id\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+
+ sg_init_one(&sg, (void *)&auth->password, strlen(auth->password));
+ ret = crypto_hash_update(&desc, &sg, strlen(auth->password));
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_update() failed for password\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+
+ sg_init_one(&sg, (void *)chap->challenge, strlen(chap->challenge));
+ ret = crypto_hash_update(&desc, &sg, strlen(chap->challenge));
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_update() failed for challenge\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+
+ ret = crypto_hash_final(&desc, server_digest);
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_final() failed for server digest\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+ crypto_free_hash(tfm);
+
+ chap_binaryhex_to_asciihex(response, server_digest, MD5_SIGNATURE_SIZE);
+ PRINT("[server] MD5 Server Digest: %s\n", response);
+
+ if (memcmp(server_digest, client_digest, MD5_SIGNATURE_SIZE) != 0) {
+ PRINT("[server] MD5 Digests do not match!\n\n");
+ goto out;
+ } else
+ PRINT("[server] MD5 Digests match, CHAP connetication"
+ " successful.\n\n");
+ /*
+ * One way authentication has succeeded, return now if mutual
+ * authentication is not enabled.
+ */
+ if (!auth->authenticate_target) {
+ kfree(challenge);
+ kfree(challenge_binhex);
+ return 0;
+ }
+ /*
+ * Get CHAP_I.
+ */
+ if (extract_param(nr_in_ptr, "CHAP_I", 10, identifier, &type) < 0) {
+ printk(KERN_ERR "Could not find CHAP_I.\n");
+ goto out;
+ }
+
+ if (type == HEX)
+ id = (unsigned char)simple_strtoul((char *)&identifier[2],
+ &endptr, 0);
+ else
+ id = (unsigned char)simple_strtoul(identifier, &endptr, 0);
+ /*
+ * RFC 1994 says Identifier is no more than octet (8 bits).
+ */
+ PRINT("[server] Got CHAP_I=%d\n", id);
+ /*
+ * Get CHAP_C.
+ */
+ if (extract_param(nr_in_ptr, "CHAP_C", CHAP_CHALLENGE_STR_LEN,
+ challenge, &type) < 0) {
+ printk(KERN_ERR "Could not find CHAP_C.\n");
+ goto out;
+ }
+
+ if (type != HEX) {
+ printk(KERN_ERR "Could not find CHAP_C.\n");
+ goto out;
+ }
+ PRINT("[server] Got CHAP_C=%s\n", challenge);
+ challenge_len = chap_string_to_hex(challenge_binhex, challenge,
+ strlen(challenge));
+ if (!challenge_len) {
+ printk(KERN_ERR "Unable to convert incoming challenge\n");
+ goto out;
+ }
+ /*
+ * Generate CHAP_N and CHAP_R for mutual authentication.
+ */
+ tfm = crypto_alloc_hash("md5", 0, CRYPTO_ALG_ASYNC);
+ if (IS_ERR(tfm)) {
+ printk(KERN_ERR "Unable to allocate struct crypto_hash\n");
+ goto out;
+ }
+ desc.tfm = tfm;
+ desc.flags = 0;
+
+ ret = crypto_hash_init(&desc);
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_init() failed\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+
+ sg_init_one(&sg, (void *)&id, 1);
+ ret = crypto_hash_update(&desc, &sg, 1);
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_update() failed for id\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+
+ sg_init_one(&sg, (void *)auth->password_mutual,
+ strlen(auth->password_mutual));
+ ret = crypto_hash_update(&desc, &sg, strlen(auth->password_mutual));
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_update() failed for"
+ " password_mutual\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+ /*
+ * Convert received challenge to binary hex.
+ */
+ sg_init_one(&sg, (void *)challenge_binhex, challenge_len);
+ ret = crypto_hash_update(&desc, &sg, challenge_len);
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_update() failed for ma challenge\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+
+ ret = crypto_hash_final(&desc, digest);
+ if (ret < 0) {
+ printk(KERN_ERR "crypto_hash_final() failed for ma digest\n");
+ crypto_free_hash(tfm);
+ goto out;
+ }
+ crypto_free_hash(tfm);
+ /*
+ * Generate CHAP_N and CHAP_R.
+ */
+ *nr_out_len = sprintf(nr_out_ptr, "CHAP_N=%s", auth->userid_mutual);
+ *nr_out_len += 1;
+ PRINT("[server] Sending CHAP_N=%s\n", auth->userid_mutual);
+ /*
+ * Convert response from binary hex to ascii hext.
+ */
+ chap_binaryhex_to_asciihex(response, digest, MD5_SIGNATURE_SIZE);
+ *nr_out_len += sprintf(nr_out_ptr + *nr_out_len, "CHAP_R=0x%s",
+ response);
+ *nr_out_len += 1;
+ PRINT("[server] Sending CHAP_R=0x%s\n", response);
+ auth_ret = 0;
+out:
+ kfree(challenge);
+ kfree(challenge_binhex);
+ return auth_ret;
+}
+
+static int chap_got_response(
+ struct iscsi_conn *conn,
+ struct iscsi_node_auth *auth,
+ char *nr_in_ptr,
+ char *nr_out_ptr,
+ unsigned int *nr_out_len)
+{
+ struct iscsi_chap *chap = (struct iscsi_chap *) conn->auth_protocol;
+
+ switch (chap->digest_type) {
+ case CHAP_DIGEST_MD5:
+ if (chap_server_compute_md5(conn, auth, nr_in_ptr,
+ nr_out_ptr, nr_out_len) < 0)
+ return -1;
+ return 0;
+ default:
+ printk(KERN_ERR "Unknown CHAP digest type %d!\n",
+ chap->digest_type);
+ return -1;
+ }
+}
+
+u32 chap_main_loop(
+ struct iscsi_conn *conn,
+ struct iscsi_node_auth *auth,
+ char *in_text,
+ char *out_text,
+ int *in_len,
+ int *out_len)
+{
+ struct iscsi_chap *chap = (struct iscsi_chap *) conn->auth_protocol;
+
+ if (!chap) {
+ chap = chap_server_open(conn, auth, in_text, out_text, out_len);
+ if (!chap)
+ return 2;
+ chap->chap_state = CHAP_STAGE_SERVER_AIC;
+ return 0;
+ } else if (chap->chap_state == CHAP_STAGE_SERVER_AIC) {
+ convert_null_to_semi(in_text, *in_len);
+ if (chap_got_response(conn, auth, in_text, out_text,
+ out_len) < 0) {
+ chap_close(conn);
+ return 2;
+ }
+ if (auth->authenticate_target)
+ chap->chap_state = CHAP_STAGE_SERVER_NR;
+ else
+ *out_len = 0;
+ chap_close(conn);
+ return 1;
+ }
+
+ return 2;
+}
diff --git a/drivers/target/iscsi/iscsi_target_auth.h b/drivers/target/iscsi/iscsi_target_auth.h
new file mode 100644
index 0000000..17b042d
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_auth.h
@@ -0,0 +1,32 @@
+#ifndef _ISCSI_CHAP_H_
+#define _ISCSI_CHAP_H_
+
+#define CHAP_DIGEST_MD5 5
+#define CHAP_DIGEST_SHA 6
+
+#define CHAP_CHALLENGE_LENGTH 16
+#define CHAP_CHALLENGE_STR_LEN 4096
+#define MAX_RESPONSE_LENGTH 64 /* sufficient for MD5 */
+#define MAX_CHAP_N_SIZE 512
+
+#define MD5_SIGNATURE_SIZE 16 /* 16 bytes in a MD5 message digest */
+
+#define CHAP_STAGE_CLIENT_A 1
+#define CHAP_STAGE_SERVER_AIC 2
+#define CHAP_STAGE_CLIENT_NR 3
+#define CHAP_STAGE_CLIENT_NRIC 4
+#define CHAP_STAGE_SERVER_NR 5
+
+extern u32 chap_main_loop(struct iscsi_conn *, struct iscsi_node_auth *, char *, char *,
+ int *, int *);
+
+struct iscsi_chap {
+ unsigned char digest_type;
+ unsigned char id;
+ unsigned char challenge[CHAP_CHALLENGE_LENGTH];
+ unsigned int challenge_len;
+ unsigned int authenticate_target;
+ unsigned int chap_state;
+} ____cacheline_aligned;
+
+#endif /*** _ISCSI_CHAP_H_ ***/
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 10/14] iscsi-target: Add Sequence/PDU list + DataIN response logic
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (9 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 09/14] iscsi-target: Add CHAP Authentication support using libcrypto Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 11/14] iscsi-target: Add iSCSI Error Recovery Hierarchy support Nicholas A. Bellinger
` (4 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch adds Sequence/PDU list logic used by RFC-3720 for
DataSequenceInOrder=[Yes,No] and DataPDUInOrder=[Yes,No]. It also
includes support for these modes of support for generating iSCSI
DataIN response data from iscsi_target.c:iscsi_send_data_in().
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_datain_values.c | 531 ++++++++++++++++
drivers/target/iscsi/iscsi_target_datain_values.h | 15 +
drivers/target/iscsi/iscsi_target_seq_pdu_list.c | 667 +++++++++++++++++++++
drivers/target/iscsi/iscsi_target_seq_pdu_list.h | 88 +++
4 files changed, 1301 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_datain_values.c
create mode 100644 drivers/target/iscsi/iscsi_target_datain_values.h
create mode 100644 drivers/target/iscsi/iscsi_target_seq_pdu_list.c
create mode 100644 drivers/target/iscsi/iscsi_target_seq_pdu_list.h
diff --git a/drivers/target/iscsi/iscsi_target_datain_values.c b/drivers/target/iscsi/iscsi_target_datain_values.c
new file mode 100644
index 0000000..67ecfe3
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_datain_values.c
@@ -0,0 +1,531 @@
+/*******************************************************************************
+ * This file contains the iSCSI Target DataIN value generation functions.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <scsi/iscsi_proto.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_seq_pdu_list.h"
+#include "iscsi_target_erl1.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target_datain_values.h"
+
+struct iscsi_datain_req *iscsit_allocate_datain_req(void)
+{
+ struct iscsi_datain_req *dr;
+
+ dr = kmem_cache_zalloc(lio_dr_cache, GFP_ATOMIC);
+ if (!dr) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_datain_req\n");
+ return NULL;
+ }
+ INIT_LIST_HEAD(&dr->dr_list);
+
+ return dr;
+}
+
+void iscsit_attach_datain_req(struct iscsi_cmd *cmd, struct iscsi_datain_req *dr)
+{
+ spin_lock(&cmd->datain_lock);
+ list_add_tail(&dr->dr_list, &cmd->datain_list);
+ spin_unlock(&cmd->datain_lock);
+}
+
+void iscsit_free_datain_req(struct iscsi_cmd *cmd, struct iscsi_datain_req *dr)
+{
+ spin_lock(&cmd->datain_lock);
+ list_del(&dr->dr_list);
+ spin_unlock(&cmd->datain_lock);
+
+ kmem_cache_free(lio_dr_cache, dr);
+}
+
+void iscsit_free_all_datain_reqs(struct iscsi_cmd *cmd)
+{
+ struct iscsi_datain_req *dr, *dr_tmp;
+
+ spin_lock(&cmd->datain_lock);
+ list_for_each_entry_safe(dr, dr_tmp, &cmd->datain_list, dr_list) {
+ list_del(&dr->dr_list);
+ kmem_cache_free(lio_dr_cache, dr);
+ }
+ spin_unlock(&cmd->datain_lock);
+}
+
+struct iscsi_datain_req *iscsit_get_datain_req(struct iscsi_cmd *cmd)
+{
+ struct iscsi_datain_req *dr;
+
+ if (list_empty(&cmd->datain_list)) {
+ printk(KERN_ERR "cmd->datain_list is empty for ITT:"
+ " 0x%08x\n", cmd->init_task_tag);
+ return NULL;
+ }
+ list_for_each_entry(dr, &cmd->datain_list, dr_list)
+ break;
+
+ return dr;
+}
+
+/*
+ * For Normal and Recovery DataSequenceInOrder=Yes and DataPDUInOrder=Yes.
+ */
+static inline struct iscsi_datain_req *iscsit_set_datain_values_yes_and_yes(
+ struct iscsi_cmd *cmd,
+ struct iscsi_datain *datain)
+{
+ u32 next_burst_len, read_data_done, read_data_left;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_datain_req *dr;
+
+ dr = iscsit_get_datain_req(cmd);
+ if (!dr)
+ return NULL;
+
+ if (dr->recovery && dr->generate_recovery_values) {
+ if (iscsit_create_recovery_datain_values_datasequenceinorder_yes(
+ cmd, dr) < 0)
+ return NULL;
+
+ dr->generate_recovery_values = 0;
+ }
+
+ next_burst_len = (!dr->recovery) ?
+ cmd->next_burst_len : dr->next_burst_len;
+ read_data_done = (!dr->recovery) ?
+ cmd->read_data_done : dr->read_data_done;
+
+ read_data_left = (cmd->data_length - read_data_done);
+ if (!read_data_left) {
+ printk(KERN_ERR "ITT: 0x%08x read_data_left is zero!\n",
+ cmd->init_task_tag);
+ return NULL;
+ }
+
+ if ((read_data_left <= conn->conn_ops->MaxRecvDataSegmentLength) &&
+ (read_data_left <= (conn->sess->sess_ops->MaxBurstLength -
+ next_burst_len))) {
+ datain->length = read_data_left;
+
+ datain->flags |= (ISCSI_FLAG_CMD_FINAL | ISCSI_FLAG_DATA_STATUS);
+ if (conn->sess->sess_ops->ErrorRecoveryLevel > 0)
+ datain->flags |= ISCSI_FLAG_DATA_ACK;
+ } else {
+ if ((next_burst_len +
+ conn->conn_ops->MaxRecvDataSegmentLength) <
+ conn->sess->sess_ops->MaxBurstLength) {
+ datain->length =
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ next_burst_len += datain->length;
+ } else {
+ datain->length = (conn->sess->sess_ops->MaxBurstLength -
+ next_burst_len);
+ next_burst_len = 0;
+
+ datain->flags |= ISCSI_FLAG_CMD_FINAL;
+ if (conn->sess->sess_ops->ErrorRecoveryLevel > 0)
+ datain->flags |= ISCSI_FLAG_DATA_ACK;
+ }
+ }
+
+ datain->data_sn = (!dr->recovery) ? cmd->data_sn++ : dr->data_sn++;
+ datain->offset = read_data_done;
+
+ if (!dr->recovery) {
+ cmd->next_burst_len = next_burst_len;
+ cmd->read_data_done += datain->length;
+ } else {
+ dr->next_burst_len = next_burst_len;
+ dr->read_data_done += datain->length;
+ }
+
+ if (!dr->recovery) {
+ if (datain->flags & ISCSI_FLAG_DATA_STATUS)
+ dr->dr_complete = DATAIN_COMPLETE_NORMAL;
+
+ return dr;
+ }
+
+ if (!dr->runlength) {
+ if (datain->flags & ISCSI_FLAG_DATA_STATUS) {
+ dr->dr_complete =
+ (dr->recovery == DATAIN_WITHIN_COMMAND_RECOVERY) ?
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY :
+ DATAIN_COMPLETE_CONNECTION_RECOVERY;
+ }
+ } else {
+ if ((dr->begrun + dr->runlength) == dr->data_sn) {
+ dr->dr_complete =
+ (dr->recovery == DATAIN_WITHIN_COMMAND_RECOVERY) ?
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY :
+ DATAIN_COMPLETE_CONNECTION_RECOVERY;
+ }
+ }
+
+ return dr;
+}
+
+/*
+ * For Normal and Recovery DataSequenceInOrder=No and DataPDUInOrder=Yes.
+ */
+static inline struct iscsi_datain_req *iscsit_set_datain_values_no_and_yes(
+ struct iscsi_cmd *cmd,
+ struct iscsi_datain *datain)
+{
+ u32 offset, read_data_done, read_data_left, seq_send_order;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_datain_req *dr;
+ struct iscsi_seq *seq;
+
+ dr = iscsit_get_datain_req(cmd);
+ if (!dr)
+ return NULL;
+
+ if (dr->recovery && dr->generate_recovery_values) {
+ if (iscsit_create_recovery_datain_values_datasequenceinorder_no(
+ cmd, dr) < 0)
+ return NULL;
+
+ dr->generate_recovery_values = 0;
+ }
+
+ read_data_done = (!dr->recovery) ?
+ cmd->read_data_done : dr->read_data_done;
+ seq_send_order = (!dr->recovery) ?
+ cmd->seq_send_order : dr->seq_send_order;
+
+ read_data_left = (cmd->data_length - read_data_done);
+ if (!read_data_left) {
+ printk(KERN_ERR "ITT: 0x%08x read_data_left is zero!\n",
+ cmd->init_task_tag);
+ return NULL;
+ }
+
+ seq = iscsit_get_seq_holder_for_datain(cmd, seq_send_order);
+ if (!seq)
+ return NULL;
+
+ seq->sent = 1;
+
+ if (!dr->recovery && !seq->next_burst_len)
+ seq->first_datasn = cmd->data_sn;
+
+ offset = (seq->offset + seq->next_burst_len);
+
+ if ((offset + conn->conn_ops->MaxRecvDataSegmentLength) >=
+ cmd->data_length) {
+ datain->length = (cmd->data_length - offset);
+ datain->offset = offset;
+
+ datain->flags |= ISCSI_FLAG_CMD_FINAL;
+ if (conn->sess->sess_ops->ErrorRecoveryLevel > 0)
+ datain->flags |= ISCSI_FLAG_DATA_ACK;
+
+ seq->next_burst_len = 0;
+ seq_send_order++;
+ } else {
+ if ((seq->next_burst_len +
+ conn->conn_ops->MaxRecvDataSegmentLength) <
+ conn->sess->sess_ops->MaxBurstLength) {
+ datain->length =
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ datain->offset = (seq->offset + seq->next_burst_len);
+
+ seq->next_burst_len += datain->length;
+ } else {
+ datain->length = (conn->sess->sess_ops->MaxBurstLength -
+ seq->next_burst_len);
+ datain->offset = (seq->offset + seq->next_burst_len);
+
+ datain->flags |= ISCSI_FLAG_CMD_FINAL;
+ if (conn->sess->sess_ops->ErrorRecoveryLevel > 0)
+ datain->flags |= ISCSI_FLAG_DATA_ACK;
+
+ seq->next_burst_len = 0;
+ seq_send_order++;
+ }
+ }
+
+ if ((read_data_done + datain->length) == cmd->data_length)
+ datain->flags |= ISCSI_FLAG_DATA_STATUS;
+
+ datain->data_sn = (!dr->recovery) ? cmd->data_sn++ : dr->data_sn++;
+ if (!dr->recovery) {
+ cmd->seq_send_order = seq_send_order;
+ cmd->read_data_done += datain->length;
+ } else {
+ dr->seq_send_order = seq_send_order;
+ dr->read_data_done += datain->length;
+ }
+
+ if (!dr->recovery) {
+ if (datain->flags & ISCSI_FLAG_CMD_FINAL)
+ seq->last_datasn = datain->data_sn;
+ if (datain->flags & ISCSI_FLAG_DATA_STATUS)
+ dr->dr_complete = DATAIN_COMPLETE_NORMAL;
+
+ return dr;
+ }
+
+ if (!dr->runlength) {
+ if (datain->flags & ISCSI_FLAG_DATA_STATUS) {
+ dr->dr_complete =
+ (dr->recovery == DATAIN_WITHIN_COMMAND_RECOVERY) ?
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY :
+ DATAIN_COMPLETE_CONNECTION_RECOVERY;
+ }
+ } else {
+ if ((dr->begrun + dr->runlength) == dr->data_sn) {
+ dr->dr_complete =
+ (dr->recovery == DATAIN_WITHIN_COMMAND_RECOVERY) ?
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY :
+ DATAIN_COMPLETE_CONNECTION_RECOVERY;
+ }
+ }
+
+ return dr;
+}
+
+/*
+ * For Normal and Recovery DataSequenceInOrder=Yes and DataPDUInOrder=No.
+ */
+static inline struct iscsi_datain_req *iscsit_set_datain_values_yes_and_no(
+ struct iscsi_cmd *cmd,
+ struct iscsi_datain *datain)
+{
+ u32 next_burst_len, read_data_done, read_data_left;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_datain_req *dr;
+ struct iscsi_pdu *pdu;
+
+ dr = iscsit_get_datain_req(cmd);
+ if (!dr)
+ return NULL;
+
+ if (dr->recovery && dr->generate_recovery_values) {
+ if (iscsit_create_recovery_datain_values_datasequenceinorder_yes(
+ cmd, dr) < 0)
+ return NULL;
+
+ dr->generate_recovery_values = 0;
+ }
+
+ next_burst_len = (!dr->recovery) ?
+ cmd->next_burst_len : dr->next_burst_len;
+ read_data_done = (!dr->recovery) ?
+ cmd->read_data_done : dr->read_data_done;
+
+ read_data_left = (cmd->data_length - read_data_done);
+ if (!read_data_left) {
+ printk(KERN_ERR "ITT: 0x%08x read_data_left is zero!\n",
+ cmd->init_task_tag);
+ return dr;
+ }
+
+ pdu = iscsit_get_pdu_holder_for_seq(cmd, NULL);
+ if (!pdu)
+ return dr;
+
+ if ((read_data_done + pdu->length) == cmd->data_length) {
+ pdu->flags |= (ISCSI_FLAG_CMD_FINAL | ISCSI_FLAG_DATA_STATUS);
+ if (conn->sess->sess_ops->ErrorRecoveryLevel > 0)
+ pdu->flags |= ISCSI_FLAG_DATA_ACK;
+
+ next_burst_len = 0;
+ } else {
+ if ((next_burst_len + conn->conn_ops->MaxRecvDataSegmentLength) <
+ conn->sess->sess_ops->MaxBurstLength)
+ next_burst_len += pdu->length;
+ else {
+ pdu->flags |= ISCSI_FLAG_CMD_FINAL;
+ if (conn->sess->sess_ops->ErrorRecoveryLevel > 0)
+ pdu->flags |= ISCSI_FLAG_DATA_ACK;
+
+ next_burst_len = 0;
+ }
+ }
+
+ pdu->data_sn = (!dr->recovery) ? cmd->data_sn++ : dr->data_sn++;
+ if (!dr->recovery) {
+ cmd->next_burst_len = next_burst_len;
+ cmd->read_data_done += pdu->length;
+ } else {
+ dr->next_burst_len = next_burst_len;
+ dr->read_data_done += pdu->length;
+ }
+
+ datain->flags = pdu->flags;
+ datain->length = pdu->length;
+ datain->offset = pdu->offset;
+ datain->data_sn = pdu->data_sn;
+
+ if (!dr->recovery) {
+ if (datain->flags & ISCSI_FLAG_DATA_STATUS)
+ dr->dr_complete = DATAIN_COMPLETE_NORMAL;
+
+ return dr;
+ }
+
+ if (!dr->runlength) {
+ if (datain->flags & ISCSI_FLAG_DATA_STATUS) {
+ dr->dr_complete =
+ (dr->recovery == DATAIN_WITHIN_COMMAND_RECOVERY) ?
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY :
+ DATAIN_COMPLETE_CONNECTION_RECOVERY;
+ }
+ } else {
+ if ((dr->begrun + dr->runlength) == dr->data_sn) {
+ dr->dr_complete =
+ (dr->recovery == DATAIN_WITHIN_COMMAND_RECOVERY) ?
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY :
+ DATAIN_COMPLETE_CONNECTION_RECOVERY;
+ }
+ }
+
+ return dr;
+}
+
+/*
+ * For Normal and Recovery DataSequenceInOrder=No and DataPDUInOrder=No.
+ */
+static inline struct iscsi_datain_req *iscsit_set_datain_values_no_and_no(
+ struct iscsi_cmd *cmd,
+ struct iscsi_datain *datain)
+{
+ u32 read_data_done, read_data_left, seq_send_order;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_datain_req *dr;
+ struct iscsi_pdu *pdu;
+ struct iscsi_seq *seq = NULL;
+
+ dr = iscsit_get_datain_req(cmd);
+ if (!dr)
+ return NULL;
+
+ if (dr->recovery && dr->generate_recovery_values) {
+ if (iscsit_create_recovery_datain_values_datasequenceinorder_no(
+ cmd, dr) < 0)
+ return NULL;
+
+ dr->generate_recovery_values = 0;
+ }
+
+ read_data_done = (!dr->recovery) ?
+ cmd->read_data_done : dr->read_data_done;
+ seq_send_order = (!dr->recovery) ?
+ cmd->seq_send_order : dr->seq_send_order;
+
+ read_data_left = (cmd->data_length - read_data_done);
+ if (!read_data_left) {
+ printk(KERN_ERR "ITT: 0x%08x read_data_left is zero!\n",
+ cmd->init_task_tag);
+ return NULL;
+ }
+
+ seq = iscsit_get_seq_holder_for_datain(cmd, seq_send_order);
+ if (!seq)
+ return NULL;
+
+ seq->sent = 1;
+
+ if (!dr->recovery && !seq->next_burst_len)
+ seq->first_datasn = cmd->data_sn;
+
+ pdu = iscsit_get_pdu_holder_for_seq(cmd, seq);
+ if (!pdu)
+ return NULL;
+
+ if (seq->pdu_send_order == seq->pdu_count) {
+ pdu->flags |= ISCSI_FLAG_CMD_FINAL;
+ if (conn->sess->sess_ops->ErrorRecoveryLevel > 0)
+ pdu->flags |= ISCSI_FLAG_DATA_ACK;
+
+ seq->next_burst_len = 0;
+ seq_send_order++;
+ } else
+ seq->next_burst_len += pdu->length;
+
+ if ((read_data_done + pdu->length) == cmd->data_length)
+ pdu->flags |= ISCSI_FLAG_DATA_STATUS;
+
+ pdu->data_sn = (!dr->recovery) ? cmd->data_sn++ : dr->data_sn++;
+ if (!dr->recovery) {
+ cmd->seq_send_order = seq_send_order;
+ cmd->read_data_done += pdu->length;
+ } else {
+ dr->seq_send_order = seq_send_order;
+ dr->read_data_done += pdu->length;
+ }
+
+ datain->flags = pdu->flags;
+ datain->length = pdu->length;
+ datain->offset = pdu->offset;
+ datain->data_sn = pdu->data_sn;
+
+ if (!dr->recovery) {
+ if (datain->flags & ISCSI_FLAG_CMD_FINAL)
+ seq->last_datasn = datain->data_sn;
+ if (datain->flags & ISCSI_FLAG_DATA_STATUS)
+ dr->dr_complete = DATAIN_COMPLETE_NORMAL;
+
+ return dr;
+ }
+
+ if (!dr->runlength) {
+ if (datain->flags & ISCSI_FLAG_DATA_STATUS) {
+ dr->dr_complete =
+ (dr->recovery == DATAIN_WITHIN_COMMAND_RECOVERY) ?
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY :
+ DATAIN_COMPLETE_CONNECTION_RECOVERY;
+ }
+ } else {
+ if ((dr->begrun + dr->runlength) == dr->data_sn) {
+ dr->dr_complete =
+ (dr->recovery == DATAIN_WITHIN_COMMAND_RECOVERY) ?
+ DATAIN_COMPLETE_WITHIN_COMMAND_RECOVERY :
+ DATAIN_COMPLETE_CONNECTION_RECOVERY;
+ }
+ }
+
+ return dr;
+}
+
+struct iscsi_datain_req *iscsit_get_datain_values(
+ struct iscsi_cmd *cmd,
+ struct iscsi_datain *datain)
+{
+ struct iscsi_conn *conn = cmd->conn;
+
+ if (conn->sess->sess_ops->DataSequenceInOrder &&
+ conn->sess->sess_ops->DataPDUInOrder)
+ return iscsit_set_datain_values_yes_and_yes(cmd, datain);
+ else if (!conn->sess->sess_ops->DataSequenceInOrder &&
+ conn->sess->sess_ops->DataPDUInOrder)
+ return iscsit_set_datain_values_no_and_yes(cmd, datain);
+ else if (conn->sess->sess_ops->DataSequenceInOrder &&
+ !conn->sess->sess_ops->DataPDUInOrder)
+ return iscsit_set_datain_values_yes_and_no(cmd, datain);
+ else if (!conn->sess->sess_ops->DataSequenceInOrder &&
+ !conn->sess->sess_ops->DataPDUInOrder)
+ return iscsit_set_datain_values_no_and_no(cmd, datain);
+
+ return NULL;
+}
diff --git a/drivers/target/iscsi/iscsi_target_datain_values.h b/drivers/target/iscsi/iscsi_target_datain_values.h
new file mode 100644
index 0000000..eb3d2ab
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_datain_values.h
@@ -0,0 +1,15 @@
+#ifndef ISCSI_TARGET_DATAIN_VALUES_H
+#define ISCSI_TARGET_DATAIN_VALUES_H
+
+extern struct iscsi_datain_req *iscsit_allocate_datain_req(void);
+extern void iscsit_attach_datain_req(struct iscsi_cmd *, struct iscsi_datain_req *);
+extern void iscsit_free_datain_req(struct iscsi_cmd *, struct iscsi_datain_req *);
+extern void iscsit_free_all_datain_reqs(struct iscsi_cmd *);
+extern struct iscsi_datain_req *iscsit_get_datain_req(struct iscsi_cmd *);
+extern struct iscsi_datain_req *iscsit_get_datain_values(struct iscsi_cmd *,
+ struct iscsi_datain *);
+
+extern struct iscsi_global *iscsi_global;
+extern struct kmem_cache *lio_dr_cache;
+
+#endif /*** ISCSI_TARGET_DATAIN_VALUES_H ***/
diff --git a/drivers/target/iscsi/iscsi_target_seq_pdu_list.c b/drivers/target/iscsi/iscsi_target_seq_pdu_list.c
new file mode 100644
index 0000000..4b9b7c1
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_seq_pdu_list.c
@@ -0,0 +1,667 @@
+/*******************************************************************************
+ * This file contains main functions related to iSCSI DataSequenceInOrder=No
+ * and DataPDUInOrder=No.
+ *
+ © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/slab.h>
+#include <linux/random.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target_seq_pdu_list.h"
+
+#define OFFLOAD_BUF_SIZE 32768
+
+void iscsit_dump_seq_list(struct iscsi_cmd *cmd)
+{
+ int i;
+ struct iscsi_seq *seq;
+
+ printk(KERN_INFO "Dumping Sequence List for ITT: 0x%08x:\n",
+ cmd->init_task_tag);
+
+ for (i = 0; i < cmd->seq_count; i++) {
+ seq = &cmd->seq_list[i];
+ printk(KERN_INFO "i: %d, pdu_start: %d, pdu_count: %d,"
+ " offset: %d, xfer_len: %d, seq_send_order: %d,"
+ " seq_no: %d\n", i, seq->pdu_start, seq->pdu_count,
+ seq->offset, seq->xfer_len, seq->seq_send_order,
+ seq->seq_no);
+ }
+}
+
+void iscsit_dump_pdu_list(struct iscsi_cmd *cmd)
+{
+ int i;
+ struct iscsi_pdu *pdu;
+
+ printk(KERN_INFO "Dumping PDU List for ITT: 0x%08x:\n",
+ cmd->init_task_tag);
+
+ for (i = 0; i < cmd->pdu_count; i++) {
+ pdu = &cmd->pdu_list[i];
+ printk(KERN_INFO "i: %d, offset: %d, length: %d,"
+ " pdu_send_order: %d, seq_no: %d\n", i, pdu->offset,
+ pdu->length, pdu->pdu_send_order, pdu->seq_no);
+ }
+}
+
+static inline void iscsit_ordered_seq_lists(
+ struct iscsi_cmd *cmd,
+ u8 type)
+{
+ u32 i, seq_count = 0;
+
+ for (i = 0; i < cmd->seq_count; i++) {
+ if (cmd->seq_list[i].type != SEQTYPE_NORMAL)
+ continue;
+ cmd->seq_list[i].seq_send_order = seq_count++;
+ }
+}
+
+static inline void iscsit_ordered_pdu_lists(
+ struct iscsi_cmd *cmd,
+ u8 type)
+{
+ u32 i, pdu_send_order = 0, seq_no = 0;
+
+ for (i = 0; i < cmd->pdu_count; i++) {
+redo:
+ if (cmd->pdu_list[i].seq_no == seq_no) {
+ cmd->pdu_list[i].pdu_send_order = pdu_send_order++;
+ continue;
+ }
+ seq_no++;
+ pdu_send_order = 0;
+ goto redo;
+ }
+}
+
+/*
+ * Generate count random values into array.
+ * Use 0x80000000 to mark generates valued in array[].
+ */
+static inline void iscsit_create_random_array(u32 *array, u32 count)
+{
+ int i, j, k;
+
+ if (count == 1) {
+ array[0] = 0;
+ return;
+ }
+
+ for (i = 0; i < count; i++) {
+redo:
+ get_random_bytes(&j, sizeof(u32));
+ j = (1 + (int) (9999 + 1) - j) % count;
+ for (k = 0; k < i + 1; k++) {
+ j |= 0x80000000;
+ if ((array[k] & 0x80000000) && (array[k] == j))
+ goto redo;
+ }
+ array[i] = j;
+ }
+
+ for (i = 0; i < count; i++)
+ array[i] &= ~0x80000000;
+
+ return;
+}
+
+static inline int iscsit_randomize_pdu_lists(
+ struct iscsi_cmd *cmd,
+ u8 type)
+{
+ int i = 0;
+ u32 *array, pdu_count, seq_count = 0, seq_no = 0, seq_offset = 0;
+
+ for (pdu_count = 0; pdu_count < cmd->pdu_count; pdu_count++) {
+redo:
+ if (cmd->pdu_list[pdu_count].seq_no == seq_no) {
+ seq_count++;
+ continue;
+ }
+ array = kzalloc(seq_count * sizeof(u32), GFP_KERNEL);
+ if (!(array)) {
+ printk(KERN_ERR "Unable to allocate memory"
+ " for random array.\n");
+ return -1;
+ }
+ iscsit_create_random_array(array, seq_count);
+
+ for (i = 0; i < seq_count; i++)
+ cmd->pdu_list[seq_offset+i].pdu_send_order = array[i];
+
+ kfree(array);
+
+ seq_offset += seq_count;
+ seq_count = 0;
+ seq_no++;
+ goto redo;
+ }
+
+ if (seq_count) {
+ array = kzalloc(seq_count * sizeof(u32), GFP_KERNEL);
+ if (!(array)) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " random array.\n");
+ return -1;
+ }
+ iscsit_create_random_array(array, seq_count);
+
+ for (i = 0; i < seq_count; i++)
+ cmd->pdu_list[seq_offset+i].pdu_send_order = array[i];
+
+ kfree(array);
+ }
+
+ return 0;
+}
+
+static inline int iscsit_randomize_seq_lists(
+ struct iscsi_cmd *cmd,
+ u8 type)
+{
+ int i, j = 0;
+ u32 *array, seq_count = cmd->seq_count;
+
+ if ((type == PDULIST_IMMEDIATE) || (type == PDULIST_UNSOLICITED))
+ seq_count--;
+ else if (type == PDULIST_IMMEDIATE_AND_UNSOLICITED)
+ seq_count -= 2;
+
+ if (!seq_count)
+ return 0;
+
+ array = kzalloc(seq_count * sizeof(u32), GFP_KERNEL);
+ if (!(array)) {
+ printk(KERN_ERR "Unable to allocate memory for random array.\n");
+ return -1;
+ }
+ iscsit_create_random_array(array, seq_count);
+
+ for (i = 0; i < cmd->seq_count; i++) {
+ if (cmd->seq_list[i].type != SEQTYPE_NORMAL)
+ continue;
+ cmd->seq_list[i].seq_send_order = array[j++];
+ }
+
+ kfree(array);
+ return 0;
+}
+
+static inline void iscsit_determine_counts_for_list(
+ struct iscsi_cmd *cmd,
+ struct iscsi_build_list *bl,
+ u32 *seq_count,
+ u32 *pdu_count)
+{
+ int check_immediate = 0;
+ u32 burstlength = 0, offset = 0;
+ u32 unsolicited_data_length = 0;
+ struct iscsi_conn *conn = cmd->conn;
+
+ if ((bl->type == PDULIST_IMMEDIATE) ||
+ (bl->type == PDULIST_IMMEDIATE_AND_UNSOLICITED))
+ check_immediate = 1;
+
+ if ((bl->type == PDULIST_UNSOLICITED) ||
+ (bl->type == PDULIST_IMMEDIATE_AND_UNSOLICITED))
+ unsolicited_data_length = (cmd->data_length >
+ conn->sess->sess_ops->FirstBurstLength) ?
+ conn->sess->sess_ops->FirstBurstLength : cmd->data_length;
+
+ while (offset < cmd->data_length) {
+ *pdu_count += 1;
+
+ if (check_immediate) {
+ check_immediate = 0;
+ offset += bl->immediate_data_length;
+ *seq_count += 1;
+ if (unsolicited_data_length)
+ unsolicited_data_length -=
+ bl->immediate_data_length;
+ continue;
+ }
+ if (unsolicited_data_length > 0) {
+ if ((offset + conn->conn_ops->MaxRecvDataSegmentLength)
+ >= cmd->data_length) {
+ unsolicited_data_length -=
+ (cmd->data_length - offset);
+ offset += (cmd->data_length - offset);
+ continue;
+ }
+ if ((offset + conn->conn_ops->MaxRecvDataSegmentLength)
+ >= conn->sess->sess_ops->FirstBurstLength) {
+ unsolicited_data_length -=
+ (conn->sess->sess_ops->FirstBurstLength -
+ offset);
+ offset += (conn->sess->sess_ops->FirstBurstLength -
+ offset);
+ burstlength = 0;
+ *seq_count += 1;
+ continue;
+ }
+
+ offset += conn->conn_ops->MaxRecvDataSegmentLength;
+ unsolicited_data_length -=
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ continue;
+ }
+ if ((offset + conn->conn_ops->MaxRecvDataSegmentLength) >=
+ cmd->data_length) {
+ offset += (cmd->data_length - offset);
+ continue;
+ }
+ if ((burstlength + conn->conn_ops->MaxRecvDataSegmentLength) >=
+ conn->sess->sess_ops->MaxBurstLength) {
+ offset += (conn->sess->sess_ops->MaxBurstLength -
+ burstlength);
+ burstlength = 0;
+ *seq_count += 1;
+ continue;
+ }
+
+ burstlength += conn->conn_ops->MaxRecvDataSegmentLength;
+ offset += conn->conn_ops->MaxRecvDataSegmentLength;
+ }
+}
+
+
+/*
+ * Builds PDU and/or Sequence list, called while DataSequenceInOrder=No
+ * and DataPDUInOrder=No.
+ */
+static inline int iscsit_build_pdu_and_seq_list(
+ struct iscsi_cmd *cmd,
+ struct iscsi_build_list *bl)
+{
+ int check_immediate = 0, datapduinorder, datasequenceinorder;
+ u32 burstlength = 0, offset = 0, i = 0;
+ u32 pdu_count = 0, seq_no = 0, unsolicited_data_length = 0;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_pdu *pdu = cmd->pdu_list;
+ struct iscsi_seq *seq = cmd->seq_list;
+
+ datapduinorder = conn->sess->sess_ops->DataPDUInOrder;
+ datasequenceinorder = conn->sess->sess_ops->DataSequenceInOrder;
+
+ if ((bl->type == PDULIST_IMMEDIATE) ||
+ (bl->type == PDULIST_IMMEDIATE_AND_UNSOLICITED))
+ check_immediate = 1;
+
+ if ((bl->type == PDULIST_UNSOLICITED) ||
+ (bl->type == PDULIST_IMMEDIATE_AND_UNSOLICITED))
+ unsolicited_data_length = (cmd->data_length >
+ conn->sess->sess_ops->FirstBurstLength) ?
+ conn->sess->sess_ops->FirstBurstLength : cmd->data_length;
+
+ while (offset < cmd->data_length) {
+ pdu_count++;
+ if (!datapduinorder) {
+ pdu[i].offset = offset;
+ pdu[i].seq_no = seq_no;
+ }
+ if (!datasequenceinorder && (pdu_count == 1)) {
+ seq[seq_no].pdu_start = i;
+ seq[seq_no].seq_no = seq_no;
+ seq[seq_no].offset = offset;
+ seq[seq_no].orig_offset = offset;
+ }
+
+ if (check_immediate) {
+ check_immediate = 0;
+ if (!datapduinorder) {
+ pdu[i].type = PDUTYPE_IMMEDIATE;
+ pdu[i++].length = bl->immediate_data_length;
+ }
+ if (!datasequenceinorder) {
+ seq[seq_no].type = SEQTYPE_IMMEDIATE;
+ seq[seq_no].pdu_count = 1;
+ seq[seq_no].xfer_len =
+ bl->immediate_data_length;
+ }
+ offset += bl->immediate_data_length;
+ pdu_count = 0;
+ seq_no++;
+ if (unsolicited_data_length)
+ unsolicited_data_length -=
+ bl->immediate_data_length;
+ continue;
+ }
+ if (unsolicited_data_length > 0) {
+ if ((offset +
+ conn->conn_ops->MaxRecvDataSegmentLength) >=
+ cmd->data_length) {
+ if (!datapduinorder) {
+ pdu[i].type = PDUTYPE_UNSOLICITED;
+ pdu[i].length =
+ (cmd->data_length - offset);
+ }
+ if (!datasequenceinorder) {
+ seq[seq_no].type = SEQTYPE_UNSOLICITED;
+ seq[seq_no].pdu_count = pdu_count;
+ seq[seq_no].xfer_len = (burstlength +
+ (cmd->data_length - offset));
+ }
+ unsolicited_data_length -=
+ (cmd->data_length - offset);
+ offset += (cmd->data_length - offset);
+ continue;
+ }
+ if ((offset +
+ conn->conn_ops->MaxRecvDataSegmentLength) >=
+ conn->sess->sess_ops->FirstBurstLength) {
+ if (!datapduinorder) {
+ pdu[i].type = PDUTYPE_UNSOLICITED;
+ pdu[i++].length =
+ (conn->sess->sess_ops->FirstBurstLength -
+ offset);
+ }
+ if (!datasequenceinorder) {
+ seq[seq_no].type = SEQTYPE_UNSOLICITED;
+ seq[seq_no].pdu_count = pdu_count;
+ seq[seq_no].xfer_len = (burstlength +
+ (conn->sess->sess_ops->FirstBurstLength -
+ offset));
+ }
+ unsolicited_data_length -=
+ (conn->sess->sess_ops->FirstBurstLength -
+ offset);
+ offset += (conn->sess->sess_ops->FirstBurstLength -
+ offset);
+ burstlength = 0;
+ pdu_count = 0;
+ seq_no++;
+ continue;
+ }
+
+ if (!datapduinorder) {
+ pdu[i].type = PDUTYPE_UNSOLICITED;
+ pdu[i++].length =
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ }
+ burstlength += conn->conn_ops->MaxRecvDataSegmentLength;
+ offset += conn->conn_ops->MaxRecvDataSegmentLength;
+ unsolicited_data_length -=
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ continue;
+ }
+ if ((offset + conn->conn_ops->MaxRecvDataSegmentLength) >=
+ cmd->data_length) {
+ if (!datapduinorder) {
+ pdu[i].type = PDUTYPE_NORMAL;
+ pdu[i].length = (cmd->data_length - offset);
+ }
+ if (!datasequenceinorder) {
+ seq[seq_no].type = SEQTYPE_NORMAL;
+ seq[seq_no].pdu_count = pdu_count;
+ seq[seq_no].xfer_len = (burstlength +
+ (cmd->data_length - offset));
+ }
+ offset += (cmd->data_length - offset);
+ continue;
+ }
+ if ((burstlength + conn->conn_ops->MaxRecvDataSegmentLength) >=
+ conn->sess->sess_ops->MaxBurstLength) {
+ if (!datapduinorder) {
+ pdu[i].type = PDUTYPE_NORMAL;
+ pdu[i++].length =
+ (conn->sess->sess_ops->MaxBurstLength -
+ burstlength);
+ }
+ if (!datasequenceinorder) {
+ seq[seq_no].type = SEQTYPE_NORMAL;
+ seq[seq_no].pdu_count = pdu_count;
+ seq[seq_no].xfer_len = (burstlength +
+ (conn->sess->sess_ops->MaxBurstLength -
+ burstlength));
+ }
+ offset += (conn->sess->sess_ops->MaxBurstLength -
+ burstlength);
+ burstlength = 0;
+ pdu_count = 0;
+ seq_no++;
+ continue;
+ }
+
+ if (!datapduinorder) {
+ pdu[i].type = PDUTYPE_NORMAL;
+ pdu[i++].length =
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ }
+ burstlength += conn->conn_ops->MaxRecvDataSegmentLength;
+ offset += conn->conn_ops->MaxRecvDataSegmentLength;
+ }
+
+ if (!datasequenceinorder) {
+ if (bl->data_direction & ISCSI_PDU_WRITE) {
+ if (bl->randomize & RANDOM_R2T_OFFSETS) {
+ if (iscsit_randomize_seq_lists(cmd, bl->type)
+ < 0)
+ return -1;
+ } else
+ iscsit_ordered_seq_lists(cmd, bl->type);
+ } else if (bl->data_direction & ISCSI_PDU_READ) {
+ if (bl->randomize & RANDOM_DATAIN_SEQ_OFFSETS) {
+ if (iscsit_randomize_seq_lists(cmd, bl->type)
+ < 0)
+ return -1;
+ } else
+ iscsit_ordered_seq_lists(cmd, bl->type);
+ }
+#if 0
+ iscsit_dump_seq_list(cmd);
+#endif
+ }
+ if (!datapduinorder) {
+ if (bl->data_direction & ISCSI_PDU_WRITE) {
+ if (bl->randomize & RANDOM_DATAOUT_PDU_OFFSETS) {
+ if (iscsit_randomize_pdu_lists(cmd, bl->type)
+ < 0)
+ return -1;
+ } else
+ iscsit_ordered_pdu_lists(cmd, bl->type);
+ } else if (bl->data_direction & ISCSI_PDU_READ) {
+ if (bl->randomize & RANDOM_DATAIN_PDU_OFFSETS) {
+ if (iscsit_randomize_pdu_lists(cmd, bl->type)
+ < 0)
+ return -1;
+ } else
+ iscsit_ordered_pdu_lists(cmd, bl->type);
+ }
+#if 0
+ iscsit_dump_pdu_list(cmd);
+#endif
+ }
+
+ return 0;
+}
+
+/*
+ * Only called while DataSequenceInOrder=No or DataPDUInOrder=No.
+ */
+int iscsit_do_build_list(
+ struct iscsi_cmd *cmd,
+ struct iscsi_build_list *bl)
+{
+ u32 pdu_count = 0, seq_count = 1;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_pdu *pdu = NULL;
+ struct iscsi_seq *seq = NULL;
+
+ iscsit_determine_counts_for_list(cmd, bl, &seq_count, &pdu_count);
+
+ if (!conn->sess->sess_ops->DataSequenceInOrder) {
+ seq = kzalloc(seq_count * sizeof(struct iscsi_seq), GFP_ATOMIC);
+ if (!(seq)) {
+ printk(KERN_ERR "Unable to allocate struct iscsi_seq list\n");
+ return -1;
+ }
+ cmd->seq_list = seq;
+ cmd->seq_count = seq_count;
+ }
+
+ if (!conn->sess->sess_ops->DataPDUInOrder) {
+ pdu = kzalloc(pdu_count * sizeof(struct iscsi_pdu), GFP_ATOMIC);
+ if (!(pdu)) {
+ printk(KERN_ERR "Unable to allocate struct iscsi_pdu list.\n");
+ kfree(seq);
+ return -1;
+ }
+ cmd->pdu_list = pdu;
+ cmd->pdu_count = pdu_count;
+ }
+
+ return iscsit_build_pdu_and_seq_list(cmd, bl);
+}
+
+struct iscsi_pdu *iscsit_get_pdu_holder(
+ struct iscsi_cmd *cmd,
+ u32 offset,
+ u32 length)
+{
+ u32 i;
+ struct iscsi_pdu *pdu = NULL;
+
+ if (!cmd->pdu_list) {
+ printk(KERN_ERR "struct iscsi_cmd->pdu_list is NULL!\n");
+ return NULL;
+ }
+
+ pdu = &cmd->pdu_list[0];
+
+ for (i = 0; i < cmd->pdu_count; i++)
+ if ((pdu[i].offset == offset) && (pdu[i].length == length))
+ return &pdu[i];
+
+ printk(KERN_ERR "Unable to locate PDU holder for ITT: 0x%08x, Offset:"
+ " %u, Length: %u\n", cmd->init_task_tag, offset, length);
+ return NULL;
+}
+
+struct iscsi_pdu *iscsit_get_pdu_holder_for_seq(
+ struct iscsi_cmd *cmd,
+ struct iscsi_seq *seq)
+{
+ u32 i;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_pdu *pdu = NULL;
+
+ if (!cmd->pdu_list) {
+ printk(KERN_ERR "struct iscsi_cmd->pdu_list is NULL!\n");
+ return NULL;
+ }
+
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+redo:
+ pdu = &cmd->pdu_list[cmd->pdu_start];
+
+ for (i = 0; pdu[i].seq_no != cmd->seq_no; i++) {
+#if 0
+ printk(KERN_INFO "pdu[i].seq_no: %d, pdu[i].pdu"
+ "_send_order: %d, pdu[i].offset: %d,"
+ " pdu[i].length: %d\n", pdu[i].seq_no,
+ pdu[i].pdu_send_order, pdu[i].offset,
+ pdu[i].length);
+#endif
+ if (pdu[i].pdu_send_order == cmd->pdu_send_order) {
+ cmd->pdu_send_order++;
+ return &pdu[i];
+ }
+ }
+
+ cmd->pdu_start += cmd->pdu_send_order;
+ cmd->pdu_send_order = 0;
+ cmd->seq_no++;
+
+ if (cmd->pdu_start < cmd->pdu_count)
+ goto redo;
+
+ printk(KERN_ERR "Command ITT: 0x%08x unable to locate"
+ " struct iscsi_pdu for cmd->pdu_send_order: %u.\n",
+ cmd->init_task_tag, cmd->pdu_send_order);
+ return NULL;
+ } else {
+ if (!seq) {
+ printk(KERN_ERR "struct iscsi_seq is NULL!\n");
+ return NULL;
+ }
+#if 0
+ printk(KERN_INFO "seq->pdu_start: %d, seq->pdu_count: %d,"
+ " seq->seq_no: %d\n", seq->pdu_start, seq->pdu_count,
+ seq->seq_no);
+#endif
+ pdu = &cmd->pdu_list[seq->pdu_start];
+
+ if (seq->pdu_send_order == seq->pdu_count) {
+ printk(KERN_ERR "Command ITT: 0x%08x seq->pdu_send"
+ "_order: %u equals seq->pdu_count: %u\n",
+ cmd->init_task_tag, seq->pdu_send_order,
+ seq->pdu_count);
+ return NULL;
+ }
+
+ for (i = 0; i < seq->pdu_count; i++) {
+ if (pdu[i].pdu_send_order == seq->pdu_send_order) {
+ seq->pdu_send_order++;
+ return &pdu[i];
+ }
+ }
+
+ printk(KERN_ERR "Command ITT: 0x%08x unable to locate iscsi"
+ "_pdu_t for seq->pdu_send_order: %u.\n",
+ cmd->init_task_tag, seq->pdu_send_order);
+ return NULL;
+ }
+
+ return NULL;
+}
+
+struct iscsi_seq *iscsit_get_seq_holder(
+ struct iscsi_cmd *cmd,
+ u32 offset,
+ u32 length)
+{
+ u32 i;
+
+ if (!cmd->seq_list) {
+ printk(KERN_ERR "struct iscsi_cmd->seq_list is NULL!\n");
+ return NULL;
+ }
+
+ for (i = 0; i < cmd->seq_count; i++) {
+#if 0
+ printk(KERN_INFO "seq_list[i].orig_offset: %d, seq_list[i]."
+ "xfer_len: %d, seq_list[i].seq_no %u\n",
+ cmd->seq_list[i].orig_offset, cmd->seq_list[i].xfer_len,
+ cmd->seq_list[i].seq_no);
+#endif
+ if ((cmd->seq_list[i].orig_offset +
+ cmd->seq_list[i].xfer_len) >=
+ (offset + length))
+ return &cmd->seq_list[i];
+ }
+
+ printk(KERN_ERR "Unable to locate Sequence holder for ITT: 0x%08x,"
+ " Offset: %u, Length: %u\n", cmd->init_task_tag, offset,
+ length);
+ return NULL;
+}
diff --git a/drivers/target/iscsi/iscsi_target_seq_pdu_list.h b/drivers/target/iscsi/iscsi_target_seq_pdu_list.h
new file mode 100644
index 0000000..8e97f77
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_seq_pdu_list.h
@@ -0,0 +1,88 @@
+#ifndef ISCSI_SEQ_AND_PDU_LIST_H
+#define ISCSI_SEQ_AND_PDU_LIST_H
+
+/* struct iscsi_pdu->status */
+#define DATAOUT_PDU_SENT 1
+
+/* struct iscsi_seq->type */
+#define SEQTYPE_IMMEDIATE 1
+#define SEQTYPE_UNSOLICITED 2
+#define SEQTYPE_NORMAL 3
+
+/* struct iscsi_seq->status */
+#define DATAOUT_SEQUENCE_GOT_R2T 1
+#define DATAOUT_SEQUENCE_WITHIN_COMMAND_RECOVERY 2
+#define DATAOUT_SEQUENCE_COMPLETE 3
+
+/* iscsi_determine_counts_for_list() type */
+#define PDULIST_NORMAL 1
+#define PDULIST_IMMEDIATE 2
+#define PDULIST_UNSOLICITED 3
+#define PDULIST_IMMEDIATE_AND_UNSOLICITED 4
+
+/* struct iscsi_pdu->type */
+#define PDUTYPE_IMMEDIATE 1
+#define PDUTYPE_UNSOLICITED 2
+#define PDUTYPE_NORMAL 3
+
+/* struct iscsi_pdu->status */
+#define ISCSI_PDU_NOT_RECEIVED 0
+#define ISCSI_PDU_RECEIVED_OK 1
+#define ISCSI_PDU_CRC_FAILED 2
+#define ISCSI_PDU_TIMED_OUT 3
+
+/* struct iscsi_build_list->randomize */
+#define RANDOM_DATAIN_PDU_OFFSETS 0x01
+#define RANDOM_DATAIN_SEQ_OFFSETS 0x02
+#define RANDOM_DATAOUT_PDU_OFFSETS 0x04
+#define RANDOM_R2T_OFFSETS 0x08
+
+/* struct iscsi_build_list->data_direction */
+#define ISCSI_PDU_READ 0x01
+#define ISCSI_PDU_WRITE 0x02
+
+struct iscsi_build_list {
+ int data_direction;
+ int randomize;
+ int type;
+ int immediate_data_length;
+};
+
+struct iscsi_pdu {
+ int status;
+ int type;
+ u8 flags;
+ u32 data_sn;
+ u32 length;
+ u32 offset;
+ u32 pdu_send_order;
+ u32 seq_no;
+} ____cacheline_aligned;
+
+struct iscsi_seq {
+ int sent;
+ int status;
+ int type;
+ u32 data_sn;
+ u32 first_datasn;
+ u32 last_datasn;
+ u32 next_burst_len;
+ u32 pdu_start;
+ u32 pdu_count;
+ u32 offset;
+ u32 orig_offset;
+ u32 pdu_send_order;
+ u32 r2t_sn;
+ u32 seq_send_order;
+ u32 seq_no;
+ u32 xfer_len;
+} ____cacheline_aligned;
+
+extern struct iscsi_global *iscsi_global;
+
+extern int iscsit_do_build_list(struct iscsi_cmd *, struct iscsi_build_list *);
+extern struct iscsi_pdu *iscsit_get_pdu_holder(struct iscsi_cmd *, u32, u32);
+extern struct iscsi_pdu *iscsit_get_pdu_holder_for_seq(struct iscsi_cmd *, struct iscsi_seq *);
+extern struct iscsi_seq *iscsit_get_seq_holder(struct iscsi_cmd *, u32, u32);
+
+#endif /* ISCSI_SEQ_AND_PDU_LIST_H */
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 11/14] iscsi-target: Add iSCSI Error Recovery Hierarchy support
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (10 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 10/14] iscsi-target: Add Sequence/PDU list + DataIN response logic Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 12/14] iscsi-target: Add support for task management operations Nicholas A. Bellinger
` (3 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch adds support for RFC-3720 compatiable ErrorRecoveryLevel
support as defined in Section 6.1.5. Error Recovery Hierarchy.
This includes support for iSCSI session reinstatement, iSCSI within
command and within connection recovery, and explict/implict connection
recovery (CSM-E and CSM-I) from state machines in Section 7 of RFC-3720.
These functions are called from iscsi_target.c to handle processing
based on the negotiated session-wide ErrorRecoveryLevel parameter.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_erl0.c | 1005 +++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_erl0.h | 17 +
drivers/target/iscsi/iscsi_target_erl1.c | 1305 ++++++++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_erl1.h | 28 +
drivers/target/iscsi/iscsi_target_erl2.c | 483 +++++++++++
drivers/target/iscsi/iscsi_target_erl2.h | 18 +
6 files changed, 2856 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_erl0.c
create mode 100644 drivers/target/iscsi/iscsi_target_erl0.h
create mode 100644 drivers/target/iscsi/iscsi_target_erl1.c
create mode 100644 drivers/target/iscsi/iscsi_target_erl1.h
create mode 100644 drivers/target/iscsi/iscsi_target_erl2.c
create mode 100644 drivers/target/iscsi/iscsi_target_erl2.h
diff --git a/drivers/target/iscsi/iscsi_target_erl0.c b/drivers/target/iscsi/iscsi_target_erl0.c
new file mode 100644
index 0000000..214e0ea
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_erl0.c
@@ -0,0 +1,1005 @@
+/******************************************************************************
+ * This file contains error recovery level zero functions used by
+ * the iSCSI Target driver.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_seq_pdu_list.h"
+#include "iscsi_target_tq.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_erl1.h"
+#include "iscsi_target_erl2.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target.h"
+
+/*
+ * Used to set values in struct iscsi_cmd that iscsit_dataout_check_sequence()
+ * checks against to determine a PDU's Offset+Length is within the current
+ * DataOUT Sequence. Used for DataSequenceInOrder=Yes only.
+ */
+void iscsit_set_dataout_sequence_values(
+ struct iscsi_cmd *cmd)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ /*
+ * Still set seq_start_offset and seq_end_offset for Unsolicited
+ * DataOUT, even if DataSequenceInOrder=No.
+ */
+ if (cmd->unsolicited_data) {
+ cmd->seq_start_offset = cmd->write_data_done;
+ cmd->seq_end_offset = (cmd->write_data_done +
+ (cmd->data_length >
+ conn->sess->sess_ops->FirstBurstLength) ?
+ conn->sess->sess_ops->FirstBurstLength : cmd->data_length);
+ return;
+ }
+
+ if (!conn->sess->sess_ops->DataSequenceInOrder)
+ return;
+
+ if (!cmd->seq_start_offset && !cmd->seq_end_offset) {
+ cmd->seq_start_offset = cmd->write_data_done;
+ cmd->seq_end_offset = (cmd->data_length >
+ conn->sess->sess_ops->MaxBurstLength) ?
+ (cmd->write_data_done +
+ conn->sess->sess_ops->MaxBurstLength) : cmd->data_length;
+ } else {
+ cmd->seq_start_offset = cmd->seq_end_offset;
+ cmd->seq_end_offset = ((cmd->seq_end_offset +
+ conn->sess->sess_ops->MaxBurstLength) >=
+ cmd->data_length) ? cmd->data_length :
+ (cmd->seq_end_offset +
+ conn->sess->sess_ops->MaxBurstLength);
+ }
+}
+
+static inline int iscsit_dataout_within_command_recovery_check(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+ /*
+ * We do the within-command recovery checks here as it is
+ * the first function called in iscsi_check_pre_dataout().
+ * Basically, if we are in within-command recovery and
+ * the PDU does not contain the offset the sequence needs,
+ * dump the payload.
+ *
+ * This only applies to DataPDUInOrder=Yes, for
+ * DataPDUInOrder=No we only re-request the failed PDU
+ * and check that all PDUs in a sequence are received
+ * upon end of sequence.
+ */
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ if ((cmd->cmd_flags & ICF_WITHIN_COMMAND_RECOVERY) &&
+ (cmd->write_data_done != hdr->offset))
+ goto dump;
+
+ cmd->cmd_flags &= ~ICF_WITHIN_COMMAND_RECOVERY;
+ } else {
+ struct iscsi_seq *seq;
+
+ seq = iscsit_get_seq_holder(cmd, hdr->offset, payload_length);
+ if (!seq)
+ return DATAOUT_CANNOT_RECOVER;
+ /*
+ * Set the struct iscsi_seq pointer to reuse later.
+ */
+ cmd->seq_ptr = seq;
+
+ if (conn->sess->sess_ops->DataPDUInOrder) {
+ if ((seq->status ==
+ DATAOUT_SEQUENCE_WITHIN_COMMAND_RECOVERY) &&
+ ((seq->offset != hdr->offset) ||
+ (seq->data_sn != hdr->datasn)))
+ goto dump;
+ } else {
+ if ((seq->status ==
+ DATAOUT_SEQUENCE_WITHIN_COMMAND_RECOVERY) &&
+ (seq->data_sn != hdr->datasn))
+ goto dump;
+ }
+
+ if (seq->status == DATAOUT_SEQUENCE_COMPLETE)
+ goto dump;
+
+ if (seq->status != DATAOUT_SEQUENCE_COMPLETE)
+ seq->status = 0;
+ }
+
+ return DATAOUT_NORMAL;
+
+dump:
+ printk(KERN_ERR "Dumping DataOUT PDU Offset: %u Length: %d DataSN:"
+ " 0x%08x\n", hdr->offset, payload_length, hdr->datasn);
+ return iscsit_dump_data_payload(conn, payload_length, 1);
+}
+
+static inline int iscsit_dataout_check_unsolicited_sequence(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ u32 first_burst_len;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+
+ if ((hdr->offset < cmd->seq_start_offset) ||
+ ((hdr->offset + payload_length) > cmd->seq_end_offset)) {
+ printk(KERN_ERR "Command ITT: 0x%08x with Offset: %u,"
+ " Length: %u outside of Unsolicited Sequence %u:%u while"
+ " DataSequenceInOrder=Yes.\n", cmd->init_task_tag,
+ hdr->offset, payload_length, cmd->seq_start_offset,
+ cmd->seq_end_offset);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+
+ first_burst_len = (cmd->first_burst_len + payload_length);
+
+ if (first_burst_len > conn->sess->sess_ops->FirstBurstLength) {
+ printk(KERN_ERR "Total %u bytes exceeds FirstBurstLength: %u"
+ " for this Unsolicited DataOut Burst.\n",
+ first_burst_len, conn->sess->sess_ops->FirstBurstLength);
+ transport_send_check_condition_and_sense(SE_CMD(cmd),
+ TCM_INCORRECT_AMOUNT_OF_DATA, 0);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+
+ /*
+ * Perform various MaxBurstLength and ISCSI_FLAG_CMD_FINAL sanity
+ * checks for the current Unsolicited DataOUT Sequence.
+ */
+ if (hdr->flags & ISCSI_FLAG_CMD_FINAL) {
+ /*
+ * Ignore ISCSI_FLAG_CMD_FINAL checks while DataPDUInOrder=No, end of
+ * sequence checks are handled in
+ * iscsit_dataout_datapduinorder_no_fbit().
+ */
+ if (!conn->sess->sess_ops->DataPDUInOrder)
+ goto out;
+
+ if ((first_burst_len != cmd->data_length) &&
+ (first_burst_len != conn->sess->sess_ops->FirstBurstLength)) {
+ printk(KERN_ERR "Unsolicited non-immediate data"
+ " received %u does not equal FirstBurstLength: %u, and"
+ " does not equal ExpXferLen %u.\n", first_burst_len,
+ conn->sess->sess_ops->FirstBurstLength,
+ cmd->data_length);
+ transport_send_check_condition_and_sense(SE_CMD(cmd),
+ TCM_INCORRECT_AMOUNT_OF_DATA, 0);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+ } else {
+ if (first_burst_len == conn->sess->sess_ops->FirstBurstLength) {
+ printk(KERN_ERR "Command ITT: 0x%08x reached"
+ " FirstBurstLength: %u, but ISCSI_FLAG_CMD_FINAL is not set. protocol"
+ " error.\n", cmd->init_task_tag,
+ conn->sess->sess_ops->FirstBurstLength);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+ if (first_burst_len == cmd->data_length) {
+ printk(KERN_ERR "Command ITT: 0x%08x reached"
+ " ExpXferLen: %u, but ISCSI_FLAG_CMD_FINAL is not set. protocol"
+ " error.\n", cmd->init_task_tag, cmd->data_length);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+ }
+
+out:
+ return DATAOUT_NORMAL;
+}
+
+static inline int iscsit_dataout_check_sequence(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ u32 next_burst_len;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_seq *seq = NULL;
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+ /*
+ * For DataSequenceInOrder=Yes: Check that the offset and offset+length
+ * is within range as defined by iscsi_set_dataout_sequence_values().
+ *
+ * For DataSequenceInOrder=No: Check that an struct iscsi_seq exists for
+ * offset+length tuple.
+ */
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ /*
+ * Due to possibility of recovery DataOUT sent by the initiator
+ * fullfilling an Recovery R2T, it's best to just dump the
+ * payload here, instead of erroring out.
+ */
+ if ((hdr->offset < cmd->seq_start_offset) ||
+ ((hdr->offset + payload_length) > cmd->seq_end_offset)) {
+ printk(KERN_ERR "Command ITT: 0x%08x with Offset: %u,"
+ " Length: %u outside of Sequence %u:%u while"
+ " DataSequenceInOrder=Yes.\n", cmd->init_task_tag,
+ hdr->offset, payload_length, cmd->seq_start_offset,
+ cmd->seq_end_offset);
+
+ if (iscsit_dump_data_payload(conn, payload_length, 1) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+ return DATAOUT_WITHIN_COMMAND_RECOVERY;
+ }
+
+ next_burst_len = (cmd->next_burst_len + payload_length);
+ } else {
+ seq = iscsit_get_seq_holder(cmd, hdr->offset, payload_length);
+ if (!seq)
+ return DATAOUT_CANNOT_RECOVER;
+ /*
+ * Set the struct iscsi_seq pointer to reuse later.
+ */
+ cmd->seq_ptr = seq;
+
+ if (seq->status == DATAOUT_SEQUENCE_COMPLETE) {
+ if (iscsit_dump_data_payload(conn, payload_length, 1) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+ return DATAOUT_WITHIN_COMMAND_RECOVERY;
+ }
+
+ next_burst_len = (seq->next_burst_len + payload_length);
+ }
+
+ if (next_burst_len > conn->sess->sess_ops->MaxBurstLength) {
+ printk(KERN_ERR "Command ITT: 0x%08x, NextBurstLength: %u and"
+ " Length: %u exceeds MaxBurstLength: %u. protocol"
+ " error.\n", cmd->init_task_tag,
+ (next_burst_len - payload_length),
+ payload_length, conn->sess->sess_ops->MaxBurstLength);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+
+ /*
+ * Perform various MaxBurstLength and ISCSI_FLAG_CMD_FINAL sanity
+ * checks for the current DataOUT Sequence.
+ */
+ if (hdr->flags & ISCSI_FLAG_CMD_FINAL) {
+ /*
+ * Ignore ISCSI_FLAG_CMD_FINAL checks while DataPDUInOrder=No, end of
+ * sequence checks are handled in
+ * iscsit_dataout_datapduinorder_no_fbit().
+ */
+ if (!conn->sess->sess_ops->DataPDUInOrder)
+ goto out;
+
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ if ((next_burst_len <
+ conn->sess->sess_ops->MaxBurstLength) &&
+ ((cmd->write_data_done + payload_length) <
+ cmd->data_length)) {
+ printk(KERN_ERR "Command ITT: 0x%08x set ISCSI_FLAG_CMD_FINAL"
+ " before end of DataOUT sequence, protocol"
+ " error.\n", cmd->init_task_tag);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+ } else {
+ if (next_burst_len < seq->xfer_len) {
+ printk(KERN_ERR "Command ITT: 0x%08x set ISCSI_FLAG_CMD_FINAL"
+ " before end of DataOUT sequence, protocol"
+ " error.\n", cmd->init_task_tag);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+ }
+ } else {
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ if (next_burst_len ==
+ conn->sess->sess_ops->MaxBurstLength) {
+ printk(KERN_ERR "Command ITT: 0x%08x reached"
+ " MaxBurstLength: %u, but ISCSI_FLAG_CMD_FINAL is"
+ " not set, protocol error.", cmd->init_task_tag,
+ conn->sess->sess_ops->MaxBurstLength);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+ if ((cmd->write_data_done + payload_length) ==
+ cmd->data_length) {
+ printk(KERN_ERR "Command ITT: 0x%08x reached"
+ " last DataOUT PDU in sequence but ISCSI_FLAG_"
+ "CMD_FINAL is not set, protocol error.\n",
+ cmd->init_task_tag);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+ } else {
+ if (next_burst_len == seq->xfer_len) {
+ printk(KERN_ERR "Command ITT: 0x%08x reached"
+ " last DataOUT PDU in sequence but ISCSI_FLAG_"
+ "CMD_FINAL is not set, protocol error.\n",
+ cmd->init_task_tag);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+ }
+ }
+
+out:
+ return DATAOUT_NORMAL;
+}
+
+static inline int iscsit_dataout_check_datasn(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ int dump = 0, recovery = 0;
+ u32 data_sn = 0;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+ /*
+ * Considering the target has no method of re-requesting DataOUT
+ * by DataSN, if we receieve a greater DataSN than expected we
+ * assume the functions for DataPDUInOrder=[Yes,No] below will
+ * handle it.
+ *
+ * If the DataSN is less than expected, dump the payload.
+ */
+ if (conn->sess->sess_ops->DataSequenceInOrder)
+ data_sn = cmd->data_sn;
+ else {
+ struct iscsi_seq *seq = cmd->seq_ptr;
+ data_sn = seq->data_sn;
+ }
+
+ if (hdr->datasn > data_sn) {
+ printk(KERN_ERR "Command ITT: 0x%08x, received DataSN: 0x%08x"
+ " higher than expected 0x%08x.\n", cmd->init_task_tag,
+ hdr->datasn, data_sn);
+ recovery = 1;
+ goto recover;
+ } else if (hdr->datasn < data_sn) {
+ printk(KERN_ERR "Command ITT: 0x%08x, received DataSN: 0x%08x"
+ " lower than expected 0x%08x, discarding payload.\n",
+ cmd->init_task_tag, hdr->datasn, data_sn);
+ dump = 1;
+ goto dump;
+ }
+
+ return DATAOUT_NORMAL;
+
+recover:
+ if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
+ printk(KERN_ERR "Unable to perform within-command recovery"
+ " while ERL=0.\n");
+ return DATAOUT_CANNOT_RECOVER;
+ }
+dump:
+ if (iscsit_dump_data_payload(conn, payload_length, 1) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+
+ return (recovery || dump) ? DATAOUT_WITHIN_COMMAND_RECOVERY :
+ DATAOUT_NORMAL;
+}
+
+static inline int iscsit_dataout_pre_datapduinorder_yes(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ int dump = 0, recovery = 0;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+ /*
+ * For DataSequenceInOrder=Yes: If the offset is greater than the global
+ * DataPDUInOrder=Yes offset counter in struct iscsi_cmd a protcol error has
+ * occured and fail the connection.
+ *
+ * For DataSequenceInOrder=No: If the offset is greater than the per
+ * sequence DataPDUInOrder=Yes offset counter in struct iscsi_seq a protocol
+ * error has occured and fail the connection.
+ */
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ if (hdr->offset != cmd->write_data_done) {
+ printk(KERN_ERR "Command ITT: 0x%08x, received offset"
+ " %u different than expected %u.\n", cmd->init_task_tag,
+ hdr->offset, cmd->write_data_done);
+ recovery = 1;
+ goto recover;
+ }
+ } else {
+ struct iscsi_seq *seq = cmd->seq_ptr;
+
+ if (hdr->offset > seq->offset) {
+ printk(KERN_ERR "Command ITT: 0x%08x, received offset"
+ " %u greater than expected %u.\n", cmd->init_task_tag,
+ hdr->offset, seq->offset);
+ recovery = 1;
+ goto recover;
+ } else if (hdr->offset < seq->offset) {
+ printk(KERN_ERR "Command ITT: 0x%08x, received offset"
+ " %u less than expected %u, discarding payload.\n",
+ cmd->init_task_tag, hdr->offset, seq->offset);
+ dump = 1;
+ goto dump;
+ }
+ }
+
+ return DATAOUT_NORMAL;
+
+recover:
+ if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
+ printk(KERN_ERR "Unable to perform within-command recovery"
+ " while ERL=0.\n");
+ return DATAOUT_CANNOT_RECOVER;
+ }
+dump:
+ if (iscsit_dump_data_payload(conn, payload_length, 1) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+
+ return (recovery) ? iscsit_recover_dataout_sequence(cmd,
+ hdr->offset, payload_length) :
+ (dump) ? DATAOUT_WITHIN_COMMAND_RECOVERY : DATAOUT_NORMAL;
+}
+
+static inline int iscsit_dataout_pre_datapduinorder_no(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ struct iscsi_pdu *pdu;
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+ pdu = iscsit_get_pdu_holder(cmd, hdr->offset, payload_length);
+ if (!pdu)
+ return DATAOUT_CANNOT_RECOVER;
+
+ cmd->pdu_ptr = pdu;
+
+ switch (pdu->status) {
+ case ISCSI_PDU_NOT_RECEIVED:
+ case ISCSI_PDU_CRC_FAILED:
+ case ISCSI_PDU_TIMED_OUT:
+ break;
+ case ISCSI_PDU_RECEIVED_OK:
+ printk(KERN_ERR "Command ITT: 0x%08x received already gotten"
+ " Offset: %u, Length: %u\n", cmd->init_task_tag,
+ hdr->offset, payload_length);
+ return iscsit_dump_data_payload(cmd->conn, payload_length, 1);
+ default:
+ return DATAOUT_CANNOT_RECOVER;
+ }
+
+ return DATAOUT_NORMAL;
+}
+
+static int iscsit_dataout_update_r2t(struct iscsi_cmd *cmd, u32 offset, u32 length)
+{
+ struct iscsi_r2t *r2t;
+
+ if (cmd->unsolicited_data)
+ return 0;
+
+ r2t = iscsit_get_r2t_for_eos(cmd, offset, length);
+ if (!r2t)
+ return -1;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ r2t->seq_complete = 1;
+ cmd->outstanding_r2ts--;
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ return 0;
+}
+
+static int iscsit_dataout_update_datapduinorder_no(
+ struct iscsi_cmd *cmd,
+ u32 data_sn,
+ int f_bit)
+{
+ int ret = 0;
+ struct iscsi_pdu *pdu = cmd->pdu_ptr;
+
+ pdu->data_sn = data_sn;
+
+ switch (pdu->status) {
+ case ISCSI_PDU_NOT_RECEIVED:
+ pdu->status = ISCSI_PDU_RECEIVED_OK;
+ break;
+ case ISCSI_PDU_CRC_FAILED:
+ pdu->status = ISCSI_PDU_RECEIVED_OK;
+ break;
+ case ISCSI_PDU_TIMED_OUT:
+ pdu->status = ISCSI_PDU_RECEIVED_OK;
+ break;
+ default:
+ return DATAOUT_CANNOT_RECOVER;
+ }
+
+ if (f_bit) {
+ ret = iscsit_dataout_datapduinorder_no_fbit(cmd, pdu);
+ if (ret == DATAOUT_CANNOT_RECOVER)
+ return ret;
+ }
+
+ return DATAOUT_NORMAL;
+}
+
+static inline int iscsit_dataout_post_crc_passed(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ int ret, send_r2t = 0;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_seq *seq = NULL;
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+ if (cmd->unsolicited_data) {
+ if ((cmd->first_burst_len + payload_length) ==
+ conn->sess->sess_ops->FirstBurstLength) {
+ if (iscsit_dataout_update_r2t(cmd, hdr->offset,
+ payload_length) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+ send_r2t = 1;
+ }
+
+ if (!conn->sess->sess_ops->DataPDUInOrder) {
+ ret = iscsit_dataout_update_datapduinorder_no(cmd,
+ hdr->datasn, (hdr->flags & ISCSI_FLAG_CMD_FINAL));
+ if (ret == DATAOUT_CANNOT_RECOVER)
+ return ret;
+ }
+
+ cmd->first_burst_len += payload_length;
+
+ if (conn->sess->sess_ops->DataSequenceInOrder)
+ cmd->data_sn++;
+ else {
+ seq = cmd->seq_ptr;
+ seq->data_sn++;
+ seq->offset += payload_length;
+ }
+
+ if (send_r2t) {
+ if (seq)
+ seq->status = DATAOUT_SEQUENCE_COMPLETE;
+ cmd->first_burst_len = 0;
+ cmd->unsolicited_data = 0;
+ }
+ } else {
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ if ((cmd->next_burst_len + payload_length) ==
+ conn->sess->sess_ops->MaxBurstLength) {
+ if (iscsit_dataout_update_r2t(cmd, hdr->offset,
+ payload_length) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+ send_r2t = 1;
+ }
+
+ if (!conn->sess->sess_ops->DataPDUInOrder) {
+ ret = iscsit_dataout_update_datapduinorder_no(
+ cmd, hdr->datasn,
+ (hdr->flags & ISCSI_FLAG_CMD_FINAL));
+ if (ret == DATAOUT_CANNOT_RECOVER)
+ return ret;
+ }
+
+ cmd->next_burst_len += payload_length;
+ cmd->data_sn++;
+
+ if (send_r2t)
+ cmd->next_burst_len = 0;
+ } else {
+ seq = cmd->seq_ptr;
+
+ if ((seq->next_burst_len + payload_length) ==
+ seq->xfer_len) {
+ if (iscsit_dataout_update_r2t(cmd, hdr->offset,
+ payload_length) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+ send_r2t = 1;
+ }
+
+ if (!conn->sess->sess_ops->DataPDUInOrder) {
+ ret = iscsit_dataout_update_datapduinorder_no(
+ cmd, hdr->datasn,
+ (hdr->flags & ISCSI_FLAG_CMD_FINAL));
+ if (ret == DATAOUT_CANNOT_RECOVER)
+ return ret;
+ }
+
+ seq->data_sn++;
+ seq->offset += payload_length;
+ seq->next_burst_len += payload_length;
+
+ if (send_r2t) {
+ seq->next_burst_len = 0;
+ seq->status = DATAOUT_SEQUENCE_COMPLETE;
+ }
+ }
+ }
+
+ if (send_r2t && conn->sess->sess_ops->DataSequenceInOrder)
+ cmd->data_sn = 0;
+
+ cmd->write_data_done += payload_length;
+
+ return (cmd->write_data_done == cmd->data_length) ?
+ DATAOUT_SEND_TO_TRANSPORT : (send_r2t) ?
+ DATAOUT_SEND_R2T : DATAOUT_NORMAL;
+}
+
+static inline int iscsit_dataout_post_crc_failed(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_pdu *pdu;
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+ if (conn->sess->sess_ops->DataPDUInOrder)
+ goto recover;
+ /*
+ * The rest of this function is only called when DataPDUInOrder=No.
+ */
+ pdu = cmd->pdu_ptr;
+
+ switch (pdu->status) {
+ case ISCSI_PDU_NOT_RECEIVED:
+ pdu->status = ISCSI_PDU_CRC_FAILED;
+ break;
+ case ISCSI_PDU_CRC_FAILED:
+ break;
+ case ISCSI_PDU_TIMED_OUT:
+ pdu->status = ISCSI_PDU_CRC_FAILED;
+ break;
+ default:
+ return DATAOUT_CANNOT_RECOVER;
+ }
+
+recover:
+ return iscsit_recover_dataout_sequence(cmd, hdr->offset, payload_length);
+}
+
+/*
+ * Called from iscsit_handle_data_out() before DataOUT Payload is received
+ * and CRC computed.
+ */
+extern int iscsit_check_pre_dataout(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ int ret;
+ struct iscsi_conn *conn = cmd->conn;
+
+ ret = iscsit_dataout_within_command_recovery_check(cmd, buf);
+ if ((ret == DATAOUT_WITHIN_COMMAND_RECOVERY) ||
+ (ret == DATAOUT_CANNOT_RECOVER))
+ return ret;
+
+ ret = iscsit_dataout_check_datasn(cmd, buf);
+ if ((ret == DATAOUT_WITHIN_COMMAND_RECOVERY) ||
+ (ret == DATAOUT_CANNOT_RECOVER))
+ return ret;
+
+ if (cmd->unsolicited_data) {
+ ret = iscsit_dataout_check_unsolicited_sequence(cmd, buf);
+ if ((ret == DATAOUT_WITHIN_COMMAND_RECOVERY) ||
+ (ret == DATAOUT_CANNOT_RECOVER))
+ return ret;
+ } else {
+ ret = iscsit_dataout_check_sequence(cmd, buf);
+ if ((ret == DATAOUT_WITHIN_COMMAND_RECOVERY) ||
+ (ret == DATAOUT_CANNOT_RECOVER))
+ return ret;
+ }
+
+ return (conn->sess->sess_ops->DataPDUInOrder) ?
+ iscsit_dataout_pre_datapduinorder_yes(cmd, buf) :
+ iscsit_dataout_pre_datapduinorder_no(cmd, buf);
+}
+
+/*
+ * Called from iscsit_handle_data_out() after DataOUT Payload is received
+ * and CRC computed.
+ */
+int iscsit_check_post_dataout(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf,
+ u8 data_crc_failed)
+{
+ struct iscsi_conn *conn = cmd->conn;
+
+ cmd->dataout_timeout_retries = 0;
+
+ if (!data_crc_failed)
+ return iscsit_dataout_post_crc_passed(cmd, buf);
+ else {
+ if (!conn->sess->sess_ops->ErrorRecoveryLevel) {
+ printk(KERN_ERR "Unable to recover from DataOUT CRC"
+ " failure while ERL=0, closing session.\n");
+ iscsit_add_reject_from_cmd(ISCSI_REASON_DATA_DIGEST_ERROR,
+ 1, 0, buf, cmd);
+ return DATAOUT_CANNOT_RECOVER;
+ }
+
+ iscsit_add_reject_from_cmd(ISCSI_REASON_DATA_DIGEST_ERROR,
+ 0, 0, buf, cmd);
+ return iscsit_dataout_post_crc_failed(cmd, buf);
+ }
+}
+
+static void iscsit_handle_time2retain_timeout(unsigned long data)
+{
+ struct iscsi_session *sess = (struct iscsi_session *) data;
+ struct iscsi_portal_group *tpg = ISCSI_TPG_S(sess);
+ struct se_portal_group *se_tpg = &tpg->tpg_se_tpg;
+
+ spin_lock_bh(&se_tpg->session_lock);
+ if (sess->time2retain_timer_flags & ISCSI_TF_STOP) {
+ spin_unlock_bh(&se_tpg->session_lock);
+ return;
+ }
+ if (atomic_read(&sess->session_reinstatement)) {
+ printk(KERN_ERR "Exiting Time2Retain handler because"
+ " session_reinstatement=1\n");
+ spin_unlock_bh(&se_tpg->session_lock);
+ return;
+ }
+ sess->time2retain_timer_flags |= ISCSI_TF_EXPIRED;
+
+ printk(KERN_ERR "Time2Retain timer expired for SID: %u, cleaning up"
+ " iSCSI session.\n", sess->sid);
+ {
+ struct iscsi_tiqn *tiqn = tpg->tpg_tiqn;
+
+ if (tiqn) {
+ spin_lock(&tiqn->sess_err_stats.lock);
+ strcpy(tiqn->sess_err_stats.last_sess_fail_rem_name,
+ (void *)sess->sess_ops->InitiatorName);
+ tiqn->sess_err_stats.last_sess_failure_type =
+ ISCSI_SESS_ERR_CXN_TIMEOUT;
+ tiqn->sess_err_stats.cxn_timeout_errors++;
+ sess->conn_timeout_errors++;
+ spin_unlock(&tiqn->sess_err_stats.lock);
+ }
+ }
+
+ spin_unlock_bh(&se_tpg->session_lock);
+ iscsit_close_session(sess);
+}
+
+extern void iscsit_start_time2retain_handler(struct iscsi_session *sess)
+{
+ int tpg_active;
+ /*
+ * Only start Time2Retain timer when the assoicated TPG is still in
+ * an ACTIVE (eg: not disabled or shutdown) state.
+ */
+ spin_lock(&ISCSI_TPG_S(sess)->tpg_state_lock);
+ tpg_active = (ISCSI_TPG_S(sess)->tpg_state == TPG_STATE_ACTIVE);
+ spin_unlock(&ISCSI_TPG_S(sess)->tpg_state_lock);
+
+ if (!tpg_active)
+ return;
+
+ if (sess->time2retain_timer_flags & ISCSI_TF_RUNNING)
+ return;
+
+ TRACE(TRACE_TIMER, "Starting Time2Retain timer for %u seconds on"
+ " SID: %u\n", sess->sess_ops->DefaultTime2Retain, sess->sid);
+
+ init_timer(&sess->time2retain_timer);
+ sess->time2retain_timer.expires =
+ (get_jiffies_64() + sess->sess_ops->DefaultTime2Retain * HZ);
+ sess->time2retain_timer.data = (unsigned long)sess;
+ sess->time2retain_timer.function = iscsit_handle_time2retain_timeout;
+ sess->time2retain_timer_flags &= ~ISCSI_TF_STOP;
+ sess->time2retain_timer_flags |= ISCSI_TF_RUNNING;
+ add_timer(&sess->time2retain_timer);
+}
+
+/*
+ * Called with spin_lock_bh(&struct se_portal_group->session_lock) held
+ */
+extern int iscsit_stop_time2retain_timer(struct iscsi_session *sess)
+{
+ struct iscsi_portal_group *tpg = ISCSI_TPG_S(sess);
+ struct se_portal_group *se_tpg = &tpg->tpg_se_tpg;
+
+ if (sess->time2retain_timer_flags & ISCSI_TF_EXPIRED)
+ return -1;
+
+ if (!(sess->time2retain_timer_flags & ISCSI_TF_RUNNING))
+ return 0;
+
+ sess->time2retain_timer_flags |= ISCSI_TF_STOP;
+ spin_unlock_bh(&se_tpg->session_lock);
+
+ del_timer_sync(&sess->time2retain_timer);
+
+ spin_lock_bh(&se_tpg->session_lock);
+ sess->time2retain_timer_flags &= ~ISCSI_TF_RUNNING;
+ TRACE(TRACE_TIMER, "Stopped Time2Retain Timer for SID: %u\n",
+ sess->sid);
+ return 0;
+}
+
+void iscsit_connection_reinstatement_rcfr(struct iscsi_conn *conn)
+{
+ spin_lock_bh(&conn->state_lock);
+ if (atomic_read(&conn->connection_exit)) {
+ spin_unlock_bh(&conn->state_lock);
+ goto sleep;
+ }
+
+ if (atomic_read(&conn->transport_failed)) {
+ spin_unlock_bh(&conn->state_lock);
+ goto sleep;
+ }
+ spin_unlock_bh(&conn->state_lock);
+
+ iscsi_thread_set_force_reinstatement(conn);
+
+sleep:
+ wait_for_completion(&conn->conn_wait_rcfr_comp);
+ complete(&conn->conn_post_wait_comp);
+}
+
+void iscsit_cause_connection_reinstatement(struct iscsi_conn *conn, int sleep)
+{
+ spin_lock_bh(&conn->state_lock);
+ if (atomic_read(&conn->connection_exit)) {
+ spin_unlock_bh(&conn->state_lock);
+ return;
+ }
+
+ if (atomic_read(&conn->transport_failed)) {
+ spin_unlock_bh(&conn->state_lock);
+ return;
+ }
+
+ if (atomic_read(&conn->connection_reinstatement)) {
+ spin_unlock_bh(&conn->state_lock);
+ return;
+ }
+
+ if (iscsi_thread_set_force_reinstatement(conn) < 0) {
+ spin_unlock_bh(&conn->state_lock);
+ return;
+ }
+
+ atomic_set(&conn->connection_reinstatement, 1);
+ if (!sleep) {
+ spin_unlock_bh(&conn->state_lock);
+ return;
+ }
+
+ atomic_set(&conn->sleep_on_conn_wait_comp, 1);
+ spin_unlock_bh(&conn->state_lock);
+
+ wait_for_completion(&conn->conn_wait_comp);
+ complete(&conn->conn_post_wait_comp);
+}
+
+void iscsit_fall_back_to_erl0(struct iscsi_session *sess)
+{
+ TRACE(TRACE_ERL0, "Falling back to ErrorRecoveryLevel=0 for SID:"
+ " %u\n", sess->sid);
+
+ atomic_set(&sess->session_fall_back_to_erl0, 1);
+}
+
+static void iscsit_handle_connection_cleanup(struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+
+ if ((sess->sess_ops->ErrorRecoveryLevel == 2) &&
+ !atomic_read(&sess->session_reinstatement) &&
+ !atomic_read(&sess->session_fall_back_to_erl0))
+ iscsit_connection_recovery_transport_reset(conn);
+ else {
+ TRACE(TRACE_ERL0, "Performing cleanup for failed iSCSI"
+ " Connection ID: %hu from %s\n", conn->cid,
+ sess->sess_ops->InitiatorName);
+ iscsit_close_connection(conn);
+ }
+}
+
+extern void iscsit_take_action_for_connection_exit(struct iscsi_conn *conn)
+{
+ spin_lock_bh(&conn->state_lock);
+ if (atomic_read(&conn->connection_exit)) {
+ spin_unlock_bh(&conn->state_lock);
+ return;
+ }
+ atomic_set(&conn->connection_exit, 1);
+
+ if (conn->conn_state == TARG_CONN_STATE_IN_LOGOUT) {
+ spin_unlock_bh(&conn->state_lock);
+ iscsit_close_connection(conn);
+ return;
+ }
+
+ if (conn->conn_state == TARG_CONN_STATE_CLEANUP_WAIT) {
+ spin_unlock_bh(&conn->state_lock);
+ return;
+ }
+
+ TRACE(TRACE_STATE, "Moving to TARG_CONN_STATE_CLEANUP_WAIT.\n");
+ conn->conn_state = TARG_CONN_STATE_CLEANUP_WAIT;
+ spin_unlock_bh(&conn->state_lock);
+
+ iscsit_handle_connection_cleanup(conn);
+}
+
+/*
+ * This is the simple function that makes the magic of
+ * sync and steering happen in the follow paradoxical order:
+ *
+ * 0) Receive conn->of_marker (bytes left until next OFMarker)
+ * bytes into an offload buffer. When we pass the exact number
+ * of bytes in conn->of_marker, iscsit_dump_data_payload() and hence
+ * rx_data() will automatically receive the identical u32 marker
+ * values and store it in conn->of_marker_offset;
+ * 1) Now conn->of_marker_offset will contain the offset to the start
+ * of the next iSCSI PDU. Dump these remaining bytes into another
+ * offload buffer.
+ * 2) We are done!
+ * Next byte in the TCP stream will contain the next iSCSI PDU!
+ * Cool Huh?!
+ */
+int iscsit_recover_from_unknown_opcode(struct iscsi_conn *conn)
+{
+ /*
+ * Make sure the remaining bytes to next maker is a sane value.
+ */
+ if (conn->of_marker > (conn->conn_ops->OFMarkInt * 4)) {
+ printk(KERN_ERR "Remaining bytes to OFMarker: %u exceeds"
+ " OFMarkInt bytes: %u.\n", conn->of_marker,
+ conn->conn_ops->OFMarkInt * 4);
+ return -1;
+ }
+
+ TRACE(TRACE_ERL1, "Advancing %u bytes in TCP stream to get to the"
+ " next OFMarker.\n", conn->of_marker);
+
+ if (iscsit_dump_data_payload(conn, conn->of_marker, 0) < 0)
+ return -1;
+
+ /*
+ * Make sure the offset marker we retrived is a valid value.
+ */
+ if (conn->of_marker_offset > (ISCSI_HDR_LEN + (CRC_LEN * 2) +
+ conn->conn_ops->MaxRecvDataSegmentLength)) {
+ printk(KERN_ERR "OfMarker offset value: %u exceeds limit.\n",
+ conn->of_marker_offset);
+ return -1;
+ }
+
+ TRACE(TRACE_ERL1, "Discarding %u bytes of TCP stream to get to the"
+ " next iSCSI Opcode.\n", conn->of_marker_offset);
+
+ if (iscsit_dump_data_payload(conn, conn->of_marker_offset, 0) < 0)
+ return -1;
+
+ return 0;
+}
diff --git a/drivers/target/iscsi/iscsi_target_erl0.h b/drivers/target/iscsi/iscsi_target_erl0.h
new file mode 100644
index 0000000..c29c59f
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_erl0.h
@@ -0,0 +1,17 @@
+#ifndef ISCSI_TARGET_ERL0_H
+#define ISCSI_TARGET_ERL0_H
+
+extern void iscsit_set_dataout_sequence_values(struct iscsi_cmd *);
+extern int iscsit_check_pre_dataout(struct iscsi_cmd *, unsigned char *);
+extern int iscsit_check_post_dataout(struct iscsi_cmd *, unsigned char *, u8);
+extern void iscsit_start_time2retain_handler(struct iscsi_session *);
+extern int iscsit_stop_time2retain_timer(struct iscsi_session *);
+extern void iscsit_connection_reinstatement_rcfr(struct iscsi_conn *);
+extern void iscsit_cause_connection_reinstatement(struct iscsi_conn *, int);
+extern void iscsit_fall_back_to_erl0(struct iscsi_session *);
+extern void iscsit_take_action_for_connection_exit(struct iscsi_conn *);
+extern int iscsit_recover_from_unknown_opcode(struct iscsi_conn *);
+
+extern struct iscsi_global *iscsi_global;
+
+#endif /*** ISCSI_TARGET_ERL0_H ***/
diff --git a/drivers/target/iscsi/iscsi_target_erl1.c b/drivers/target/iscsi/iscsi_target_erl1.c
new file mode 100644
index 0000000..b616e23
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_erl1.c
@@ -0,0 +1,1305 @@
+/*******************************************************************************
+ * This file contains error recovery level one used by the iSCSI Target driver.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/list.h>
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_seq_pdu_list.h"
+#include "iscsi_target_datain_values.h"
+#include "iscsi_target_device.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_erl1.h"
+#include "iscsi_target_erl2.h"
+#include "iscsi_target.h"
+
+#define OFFLOAD_BUF_SIZE 32768
+
+/*
+ * Used to dump excess datain payload for certain error recovery
+ * situations. Receive in OFFLOAD_BUF_SIZE max of datain per rx_data().
+ *
+ * dump_padding_digest denotes if padding and data digests need
+ * to be dumped.
+ */
+int iscsit_dump_data_payload(
+ struct iscsi_conn *conn,
+ u32 buf_len,
+ int dump_padding_digest)
+{
+ char *buf, pad_bytes[4];
+ int ret = DATAOUT_WITHIN_COMMAND_RECOVERY, rx_got;
+ u32 length, padding, offset = 0, size;
+ struct kvec iov;
+
+ length = (buf_len > OFFLOAD_BUF_SIZE) ? OFFLOAD_BUF_SIZE : buf_len;
+
+ buf = kzalloc(length, GFP_ATOMIC);
+ if (!buf) {
+ printk(KERN_ERR "Unable to allocate %u bytes for offload"
+ " buffer.\n", length);
+ return -1;
+ }
+ memset(&iov, 0, sizeof(struct kvec));
+
+ while (offset < buf_len) {
+ size = ((offset + length) > buf_len) ?
+ (buf_len - offset) : length;
+
+ iov.iov_len = size;
+ iov.iov_base = buf;
+
+ rx_got = rx_data(conn, &iov, 1, size);
+ if (rx_got != size) {
+ ret = DATAOUT_CANNOT_RECOVER;
+ goto out;
+ }
+
+ offset += size;
+ }
+
+ if (!dump_padding_digest)
+ goto out;
+
+ padding = ((-buf_len) & 3);
+ if (padding != 0) {
+ iov.iov_len = padding;
+ iov.iov_base = pad_bytes;
+
+ rx_got = rx_data(conn, &iov, 1, padding);
+ if (rx_got != padding) {
+ ret = DATAOUT_CANNOT_RECOVER;
+ goto out;
+ }
+ }
+
+ if (conn->conn_ops->DataDigest) {
+ u32 data_crc;
+
+ iov.iov_len = CRC_LEN;
+ iov.iov_base = &data_crc;
+
+ rx_got = rx_data(conn, &iov, 1, CRC_LEN);
+ if (rx_got != CRC_LEN) {
+ ret = DATAOUT_CANNOT_RECOVER;
+ goto out;
+ }
+ }
+
+out:
+ kfree(buf);
+ return ret;
+}
+
+/*
+ * Used for retransmitting R2Ts from a R2T SNACK request.
+ */
+static int iscsit_send_recovery_r2t_for_snack(
+ struct iscsi_cmd *cmd,
+ struct iscsi_r2t *r2t)
+{
+ /*
+ * If the struct iscsi_r2t has not been sent yet, we can safely
+ * ignore retransmission
+ * of the R2TSN in question.
+ */
+ spin_lock_bh(&cmd->r2t_lock);
+ if (!r2t->sent_r2t) {
+ spin_unlock_bh(&cmd->r2t_lock);
+ return 0;
+ }
+ r2t->sent_r2t = 0;
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ iscsit_add_cmd_to_immediate_queue(cmd, cmd->conn, ISTATE_SEND_R2T);
+
+ return 0;
+}
+
+static int iscsit_handle_r2t_snack(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf,
+ u32 begrun,
+ u32 runlength)
+{
+ u32 last_r2tsn;
+ struct iscsi_r2t *r2t;
+
+ /*
+ * Make sure the initiator is not requesting retransmission
+ * of R2TSNs already acknowledged by a TMR TASK_REASSIGN.
+ */
+ if ((cmd->cmd_flags & ICF_GOT_DATACK_SNACK) &&
+ (begrun <= cmd->acked_data_sn)) {
+ printk(KERN_ERR "ITT: 0x%08x, R2T SNACK requesting"
+ " retransmission of R2TSN: 0x%08x to 0x%08x but already"
+ " acked to R2TSN: 0x%08x by TMR TASK_REASSIGN,"
+ " protocol error.\n", cmd->init_task_tag, begrun,
+ (begrun + runlength), cmd->acked_data_sn);
+
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ }
+
+ if (runlength) {
+ if ((begrun + runlength) > cmd->r2t_sn) {
+ printk(KERN_ERR "Command ITT: 0x%08x received R2T SNACK"
+ " with BegRun: 0x%08x, RunLength: 0x%08x, exceeds"
+ " current R2TSN: 0x%08x, protocol error.\n",
+ cmd->init_task_tag, begrun, runlength, cmd->r2t_sn);
+ return iscsit_add_reject_from_cmd(
+ ISCSI_REASON_BOOKMARK_INVALID, 1, 0, buf, cmd);
+ }
+ last_r2tsn = (begrun + runlength);
+ } else
+ last_r2tsn = cmd->r2t_sn;
+
+ while (begrun < last_r2tsn) {
+ r2t = iscsit_get_holder_for_r2tsn(cmd, begrun);
+ if (!r2t)
+ return -1;
+ if (iscsit_send_recovery_r2t_for_snack(cmd, r2t) < 0)
+ return -1;
+
+ begrun++;
+ }
+
+ return 0;
+}
+
+/*
+ * Generates Offsets and NextBurstLength based on Begrun and Runlength
+ * carried in a Data SNACK or ExpDataSN in TMR TASK_REASSIGN.
+ *
+ * For DataSequenceInOrder=Yes and DataPDUInOrder=[Yes,No] only.
+ *
+ * FIXME: How is this handled for a RData SNACK?
+ */
+int iscsit_create_recovery_datain_values_datasequenceinorder_yes(
+ struct iscsi_cmd *cmd,
+ struct iscsi_datain_req *dr)
+{
+ u32 data_sn = 0, data_sn_count = 0;
+ u32 pdu_start = 0, seq_no = 0;
+ u32 begrun = dr->begrun;
+ struct iscsi_conn *conn = cmd->conn;
+
+ while (begrun > data_sn++) {
+ data_sn_count++;
+ if ((dr->next_burst_len +
+ conn->conn_ops->MaxRecvDataSegmentLength) <
+ conn->sess->sess_ops->MaxBurstLength) {
+ dr->read_data_done +=
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ dr->next_burst_len +=
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ } else {
+ dr->read_data_done +=
+ (conn->sess->sess_ops->MaxBurstLength -
+ dr->next_burst_len);
+ dr->next_burst_len = 0;
+ pdu_start += data_sn_count;
+ data_sn_count = 0;
+ seq_no++;
+ }
+ }
+
+ if (!conn->sess->sess_ops->DataPDUInOrder) {
+ cmd->seq_no = seq_no;
+ cmd->pdu_start = pdu_start;
+ cmd->pdu_send_order = data_sn_count;
+ }
+
+ return 0;
+}
+
+/*
+ * Generates Offsets and NextBurstLength based on Begrun and Runlength
+ * carried in a Data SNACK or ExpDataSN in TMR TASK_REASSIGN.
+ *
+ * For DataSequenceInOrder=No and DataPDUInOrder=[Yes,No] only.
+ *
+ * FIXME: How is this handled for a RData SNACK?
+ */
+int iscsit_create_recovery_datain_values_datasequenceinorder_no(
+ struct iscsi_cmd *cmd,
+ struct iscsi_datain_req *dr)
+{
+ int found_seq = 0, i;
+ u32 data_sn, read_data_done = 0, seq_send_order = 0;
+ u32 begrun = dr->begrun;
+ u32 runlength = dr->runlength;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_seq *first_seq = NULL, *seq = NULL;
+
+ if (!cmd->seq_list) {
+ printk(KERN_ERR "struct iscsi_cmd->seq_list is NULL!\n");
+ return -1;
+ }
+
+ /*
+ * Calculate read_data_done for all sequences containing a
+ * first_datasn and last_datasn less than the BegRun.
+ *
+ * Locate the struct iscsi_seq the BegRun lies within and calculate
+ * NextBurstLenghth up to the DataSN based on MaxRecvDataSegmentLength.
+ *
+ * Also use struct iscsi_seq->seq_send_order to determine where to start.
+ */
+ for (i = 0; i < cmd->seq_count; i++) {
+ seq = &cmd->seq_list[i];
+
+ if (!seq->seq_send_order)
+ first_seq = seq;
+
+ /*
+ * No data has been transferred for this DataIN sequence, so the
+ * seq->first_datasn and seq->last_datasn have not been set.
+ */
+ if (!seq->sent) {
+#if 0
+ printk(KERN_ERR "Ignoring non-sent sequence 0x%08x ->"
+ " 0x%08x\n\n", seq->first_datasn,
+ seq->last_datasn);
+#endif
+ continue;
+ }
+
+ /*
+ * This DataIN sequence is precedes the received BegRun, add the
+ * total xfer_len of the sequence to read_data_done and reset
+ * seq->pdu_send_order.
+ */
+ if ((seq->first_datasn < begrun) &&
+ (seq->last_datasn < begrun)) {
+#if 0
+ printk(KERN_ERR "Pre BegRun sequence 0x%08x ->"
+ " 0x%08x\n", seq->first_datasn,
+ seq->last_datasn);
+#endif
+ read_data_done += cmd->seq_list[i].xfer_len;
+ seq->next_burst_len = seq->pdu_send_order = 0;
+ continue;
+ }
+
+ /*
+ * The BegRun lies within this DataIN sequence.
+ */
+ if ((seq->first_datasn <= begrun) &&
+ (seq->last_datasn >= begrun)) {
+#if 0
+ printk(KERN_ERR "Found sequence begrun: 0x%08x in"
+ " 0x%08x -> 0x%08x\n", begrun,
+ seq->first_datasn, seq->last_datasn);
+#endif
+ seq_send_order = seq->seq_send_order;
+ data_sn = seq->first_datasn;
+ seq->next_burst_len = seq->pdu_send_order = 0;
+ found_seq = 1;
+
+ /*
+ * For DataPDUInOrder=Yes, while the first DataSN of
+ * the sequence is less than the received BegRun, add
+ * the MaxRecvDataSegmentLength to read_data_done and
+ * to the sequence's next_burst_len;
+ *
+ * For DataPDUInOrder=No, while the first DataSN of the
+ * sequence is less than the received BegRun, find the
+ * struct iscsi_pdu of the DataSN in question and add the
+ * MaxRecvDataSegmentLength to read_data_done and to the
+ * sequence's next_burst_len;
+ */
+ if (conn->sess->sess_ops->DataPDUInOrder) {
+ while (data_sn < begrun) {
+ seq->pdu_send_order++;
+ read_data_done +=
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ seq->next_burst_len +=
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ data_sn++;
+ }
+ } else {
+ int j;
+ struct iscsi_pdu *pdu;
+
+ while (data_sn < begrun) {
+ seq->pdu_send_order++;
+
+ for (j = 0; j < seq->pdu_count; j++) {
+ pdu = &cmd->pdu_list[
+ seq->pdu_start + j];
+ if (pdu->data_sn == data_sn) {
+ read_data_done +=
+ pdu->length;
+ seq->next_burst_len +=
+ pdu->length;
+ }
+ }
+ data_sn++;
+ }
+ }
+ continue;
+ }
+
+ /*
+ * This DataIN sequence is larger than the received BegRun,
+ * reset seq->pdu_send_order and continue.
+ */
+ if ((seq->first_datasn > begrun) ||
+ (seq->last_datasn > begrun)) {
+#if 0
+ printk(KERN_ERR "Post BegRun sequence 0x%08x -> 0x%08x\n",
+ seq->first_datasn, seq->last_datasn);
+#endif
+ seq->next_burst_len = seq->pdu_send_order = 0;
+ continue;
+ }
+ }
+
+ if (!found_seq) {
+ if (!begrun) {
+ if (!first_seq) {
+ printk(KERN_ERR "ITT: 0x%08x, Begrun: 0x%08x"
+ " but first_seq is NULL\n",
+ cmd->init_task_tag, begrun);
+ return -1;
+ }
+ seq_send_order = first_seq->seq_send_order;
+ seq->next_burst_len = seq->pdu_send_order = 0;
+ goto done;
+ }
+
+ printk(KERN_ERR "Unable to locate struct iscsi_seq for ITT: 0x%08x,"
+ " BegRun: 0x%08x, RunLength: 0x%08x while"
+ " DataSequenceInOrder=No and DataPDUInOrder=%s.\n",
+ cmd->init_task_tag, begrun, runlength,
+ (conn->sess->sess_ops->DataPDUInOrder) ? "Yes" : "No");
+ return -1;
+ }
+
+done:
+ dr->read_data_done = read_data_done;
+ dr->seq_send_order = seq_send_order;
+
+ return 0;
+}
+
+static inline int iscsit_handle_recovery_datain(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf,
+ u32 begrun,
+ u32 runlength)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_datain_req *dr;
+ struct se_cmd *se_cmd = &cmd->se_cmd;
+
+ if (!atomic_read(&T_TASK(se_cmd)->t_transport_complete)) {
+ printk(KERN_ERR "Ignoring ITT: 0x%08x Data SNACK\n",
+ cmd->init_task_tag);
+ return 0;
+ }
+
+ /*
+ * Make sure the initiator is not requesting retransmission
+ * of DataSNs already acknowledged by a Data ACK SNACK.
+ */
+ if ((cmd->cmd_flags & ICF_GOT_DATACK_SNACK) &&
+ (begrun <= cmd->acked_data_sn)) {
+ printk(KERN_ERR "ITT: 0x%08x, Data SNACK requesting"
+ " retransmission of DataSN: 0x%08x to 0x%08x but"
+ " already acked to DataSN: 0x%08x by Data ACK SNACK,"
+ " protocol error.\n", cmd->init_task_tag, begrun,
+ (begrun + runlength), cmd->acked_data_sn);
+
+ return iscsit_add_reject_from_cmd(ISCSI_REASON_PROTOCOL_ERROR,
+ 1, 0, buf, cmd);
+ }
+
+ /*
+ * Make sure BegRun and RunLength in the Data SNACK are sane.
+ * Note: (cmd->data_sn - 1) will carry the maximum DataSN sent.
+ */
+ if ((begrun + runlength) > (cmd->data_sn - 1)) {
+ printk(KERN_ERR "Initiator requesting BegRun: 0x%08x, RunLength"
+ ": 0x%08x greater than maximum DataSN: 0x%08x.\n",
+ begrun, runlength, (cmd->data_sn - 1));
+ return iscsit_add_reject_from_cmd(ISCSI_REASON_BOOKMARK_INVALID,
+ 1, 0, buf, cmd);
+ }
+
+ dr = iscsit_allocate_datain_req();
+ if (!dr)
+ return iscsit_add_reject_from_cmd(ISCSI_REASON_BOOKMARK_NO_RESOURCES,
+ 1, 0, buf, cmd);
+
+ dr->data_sn = dr->begrun = begrun;
+ dr->runlength = runlength;
+ dr->generate_recovery_values = 1;
+ dr->recovery = DATAIN_WITHIN_COMMAND_RECOVERY;
+
+ iscsit_attach_datain_req(cmd, dr);
+
+ cmd->i_state = ISTATE_SEND_DATAIN;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+
+ return 0;
+}
+
+int iscsit_handle_recovery_datain_or_r2t(
+ struct iscsi_conn *conn,
+ unsigned char *buf,
+ u32 init_task_tag,
+ u32 targ_xfer_tag,
+ u32 begrun,
+ u32 runlength)
+{
+ struct iscsi_cmd *cmd;
+
+ cmd = iscsit_find_cmd_from_itt(conn, init_task_tag);
+ if (!cmd)
+ return 0;
+
+ /*
+ * FIXME: This will not work for bidi commands.
+ */
+ switch (cmd->data_direction) {
+ case DMA_TO_DEVICE:
+ return iscsit_handle_r2t_snack(cmd, buf, begrun, runlength);
+ case DMA_FROM_DEVICE:
+ return iscsit_handle_recovery_datain(cmd, buf, begrun,
+ runlength);
+ default:
+ printk(KERN_ERR "Unknown cmd->data_direction: 0x%02x\n",
+ cmd->data_direction);
+ return -1;
+ }
+
+ return 0;
+}
+
+/* #warning FIXME: Status SNACK needs to be dependent on OPCODE!!! */
+int iscsit_handle_status_snack(
+ struct iscsi_conn *conn,
+ u32 init_task_tag,
+ u32 targ_xfer_tag,
+ u32 begrun,
+ u32 runlength)
+{
+ u32 last_statsn;
+ struct iscsi_cmd *cmd = NULL;
+
+ if (conn->exp_statsn > begrun) {
+ printk(KERN_ERR "Got Status SNACK Begrun: 0x%08x, RunLength:"
+ " 0x%08x but already got ExpStatSN: 0x%08x on CID:"
+ " %hu.\n", begrun, runlength, conn->exp_statsn,
+ conn->cid);
+ return 0;
+ }
+
+ last_statsn = (!runlength) ? conn->stat_sn : (begrun + runlength);
+
+ while (begrun < last_statsn) {
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry(cmd, &conn->conn_cmd_list, i_list) {
+ if (cmd->stat_sn == begrun)
+ break;
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (!cmd) {
+ printk(KERN_ERR "Unable to find StatSN: 0x%08x for"
+ " a Status SNACK, assuming this was a"
+ " protactic SNACK for an untransmitted"
+ " StatSN, ignoring.\n", begrun);
+ begrun++;
+ continue;
+ }
+
+ spin_lock_bh(&cmd->istate_lock);
+ if (cmd->i_state == ISTATE_SEND_DATAIN) {
+ spin_unlock_bh(&cmd->istate_lock);
+ printk(KERN_ERR "Ignoring Status SNACK for BegRun:"
+ " 0x%08x, RunLength: 0x%08x, assuming this was"
+ " a protactic SNACK for an untransmitted"
+ " StatSN\n", begrun, runlength);
+ begrun++;
+ continue;
+ }
+ spin_unlock_bh(&cmd->istate_lock);
+
+ cmd->i_state = ISTATE_SEND_STATUS_RECOVERY;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ begrun++;
+ }
+
+ return 0;
+}
+
+int iscsit_handle_data_ack(
+ struct iscsi_conn *conn,
+ u32 targ_xfer_tag,
+ u32 begrun,
+ u32 runlength)
+{
+ struct iscsi_cmd *cmd = NULL;
+
+ cmd = iscsit_find_cmd_from_ttt(conn, targ_xfer_tag);
+ if (!cmd) {
+ printk(KERN_ERR "Data ACK SNACK for TTT: 0x%08x is"
+ " invalid.\n", targ_xfer_tag);
+ return -1;
+ }
+
+ if (begrun <= cmd->acked_data_sn) {
+ printk(KERN_ERR "ITT: 0x%08x Data ACK SNACK BegRUN: 0x%08x is"
+ " less than the already acked DataSN: 0x%08x.\n",
+ cmd->init_task_tag, begrun, cmd->acked_data_sn);
+ return -1;
+ }
+
+ /*
+ * For Data ACK SNACK, BegRun is the next expected DataSN.
+ * (see iSCSI v19: 10.16.6)
+ */
+ cmd->cmd_flags |= ICF_GOT_DATACK_SNACK;
+ cmd->acked_data_sn = (begrun - 1);
+
+ TRACE(TRACE_ISCSI, "Received Data ACK SNACK for ITT: 0x%08x,"
+ " updated acked DataSN to 0x%08x.\n",
+ cmd->init_task_tag, cmd->acked_data_sn);
+
+ return 0;
+}
+
+static int iscsit_send_recovery_r2t(
+ struct iscsi_cmd *cmd,
+ u32 offset,
+ u32 xfer_len)
+{
+ int ret;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ ret = iscsit_add_r2t_to_list(cmd, offset, xfer_len, 1, 0);
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ return ret;
+}
+
+int iscsit_dataout_datapduinorder_no_fbit(
+ struct iscsi_cmd *cmd,
+ struct iscsi_pdu *pdu)
+{
+ int i, send_recovery_r2t = 0, recovery = 0;
+ u32 length = 0, offset = 0, pdu_count = 0, xfer_len = 0;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_pdu *first_pdu = NULL;
+
+ /*
+ * Get an struct iscsi_pdu pointer to the first PDU, and total PDU count
+ * of the DataOUT sequence.
+ */
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ for (i = 0; i < cmd->pdu_count; i++) {
+ if (cmd->pdu_list[i].seq_no == pdu->seq_no) {
+ if (!first_pdu)
+ first_pdu = &cmd->pdu_list[i];
+ xfer_len += cmd->pdu_list[i].length;
+ pdu_count++;
+ } else if (pdu_count)
+ break;
+ }
+ } else {
+ struct iscsi_seq *seq = cmd->seq_ptr;
+
+ first_pdu = &cmd->pdu_list[seq->pdu_start];
+ pdu_count = seq->pdu_count;
+ }
+
+ if (!first_pdu || !pdu_count)
+ return DATAOUT_CANNOT_RECOVER;
+
+ /*
+ * Loop through the ending DataOUT Sequence checking each struct iscsi_pdu.
+ * The following ugly logic does batching of not received PDUs.
+ */
+ for (i = 0; i < pdu_count; i++) {
+ if (first_pdu[i].status == ISCSI_PDU_RECEIVED_OK) {
+ if (!send_recovery_r2t)
+ continue;
+
+ if (iscsit_send_recovery_r2t(cmd, offset, length) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+
+ send_recovery_r2t = length = offset = 0;
+ continue;
+ }
+ /*
+ * Set recovery = 1 for any missing, CRC failed, or timed
+ * out PDUs to let the DataOUT logic know that this sequence
+ * has not been completed yet.
+ *
+ * Also, only send a Recovery R2T for ISCSI_PDU_NOT_RECEIVED.
+ * We assume if the PDU either failed CRC or timed out
+ * that a Recovery R2T has already been sent.
+ */
+ recovery = 1;
+
+ if (first_pdu[i].status != ISCSI_PDU_NOT_RECEIVED)
+ continue;
+
+ if (!offset)
+ offset = first_pdu[i].offset;
+ length += first_pdu[i].length;
+
+ send_recovery_r2t = 1;
+ }
+
+ if (send_recovery_r2t)
+ if (iscsit_send_recovery_r2t(cmd, offset, length) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+
+ return (!recovery) ? DATAOUT_NORMAL : DATAOUT_WITHIN_COMMAND_RECOVERY;
+}
+
+static int iscsit_recalculate_dataout_values(
+ struct iscsi_cmd *cmd,
+ u32 pdu_offset,
+ u32 pdu_length,
+ u32 *r2t_offset,
+ u32 *r2t_length)
+{
+ int i;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_pdu *pdu = NULL;
+
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ cmd->data_sn = 0;
+
+ if (conn->sess->sess_ops->DataPDUInOrder) {
+ *r2t_offset = cmd->write_data_done;
+ *r2t_length = (cmd->seq_end_offset -
+ cmd->write_data_done);
+ return 0;
+ }
+
+ *r2t_offset = cmd->seq_start_offset;
+ *r2t_length = (cmd->seq_end_offset - cmd->seq_start_offset);
+
+ for (i = 0; i < cmd->pdu_count; i++) {
+ pdu = &cmd->pdu_list[i];
+
+ if (pdu->status != ISCSI_PDU_RECEIVED_OK)
+ continue;
+
+ if ((pdu->offset >= cmd->seq_start_offset) &&
+ ((pdu->offset + pdu->length) <=
+ cmd->seq_end_offset)) {
+ if (!cmd->unsolicited_data)
+ cmd->next_burst_len -= pdu->length;
+ else
+ cmd->first_burst_len -= pdu->length;
+
+ cmd->write_data_done -= pdu->length;
+ pdu->status = ISCSI_PDU_NOT_RECEIVED;
+ }
+ }
+ } else {
+ struct iscsi_seq *seq = NULL;
+
+ seq = iscsit_get_seq_holder(cmd, pdu_offset, pdu_length);
+ if (!seq)
+ return -1;
+
+ *r2t_offset = seq->orig_offset;
+ *r2t_length = seq->xfer_len;
+
+ cmd->write_data_done -= (seq->offset - seq->orig_offset);
+ if (cmd->immediate_data)
+ cmd->first_burst_len = cmd->write_data_done;
+
+ seq->data_sn = 0;
+ seq->offset = seq->orig_offset;
+ seq->next_burst_len = 0;
+ seq->status = DATAOUT_SEQUENCE_WITHIN_COMMAND_RECOVERY;
+
+ if (conn->sess->sess_ops->DataPDUInOrder)
+ return 0;
+
+ for (i = 0; i < seq->pdu_count; i++) {
+ pdu = &cmd->pdu_list[i+seq->pdu_start];
+
+ if (pdu->status != ISCSI_PDU_RECEIVED_OK)
+ continue;
+
+ pdu->status = ISCSI_PDU_NOT_RECEIVED;
+ }
+ }
+
+ return 0;
+}
+
+int iscsit_recover_dataout_sequence(
+ struct iscsi_cmd *cmd,
+ u32 pdu_offset,
+ u32 pdu_length)
+{
+ u32 r2t_length = 0, r2t_offset = 0;
+
+ spin_lock_bh(&cmd->istate_lock);
+ cmd->cmd_flags |= ICF_WITHIN_COMMAND_RECOVERY;
+ spin_unlock_bh(&cmd->istate_lock);
+
+ if (iscsit_recalculate_dataout_values(cmd, pdu_offset, pdu_length,
+ &r2t_offset, &r2t_length) < 0)
+ return DATAOUT_CANNOT_RECOVER;
+
+ iscsit_send_recovery_r2t(cmd, r2t_offset, r2t_length);
+
+ return DATAOUT_WITHIN_COMMAND_RECOVERY;
+}
+
+static inline struct iscsi_ooo_cmdsn *iscsit_allocate_ooo_cmdsn(void)
+{
+ struct iscsi_ooo_cmdsn *ooo_cmdsn = NULL;
+
+ ooo_cmdsn = kmem_cache_zalloc(lio_ooo_cache, GFP_ATOMIC);
+ if (!ooo_cmdsn) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_ooo_cmdsn.\n");
+ return NULL;
+ }
+ INIT_LIST_HEAD(&ooo_cmdsn->ooo_list);
+
+ return ooo_cmdsn;
+}
+
+/*
+ * Called with sess->cmdsn_lock held.
+ */
+static inline int iscsit_attach_ooo_cmdsn(
+ struct iscsi_session *sess,
+ struct iscsi_ooo_cmdsn *ooo_cmdsn)
+{
+ struct iscsi_ooo_cmdsn *ooo_tail, *ooo_tmp;
+ /*
+ * We attach the struct iscsi_ooo_cmdsn entry to the out of order
+ * list in increasing CmdSN order.
+ * This allows iscsi_execute_ooo_cmdsns() to detect any
+ * additional CmdSN holes while performing delayed execution.
+ */
+ if (list_empty(&sess->sess_ooo_cmdsn_list))
+ list_add_tail(&ooo_cmdsn->ooo_list,
+ &sess->sess_ooo_cmdsn_list);
+ else {
+ ooo_tail = list_entry(sess->sess_ooo_cmdsn_list.prev,
+ typeof(*ooo_tail), ooo_list);
+ /*
+ * CmdSN is greater than the tail of the list.
+ */
+ if (ooo_tail->cmdsn < ooo_cmdsn->cmdsn)
+ list_add_tail(&ooo_cmdsn->ooo_list,
+ &sess->sess_ooo_cmdsn_list);
+ else {
+ /*
+ * CmdSN is either lower than the head, or somewhere
+ * in the middle.
+ */
+ list_for_each_entry(ooo_tmp, &sess->sess_ooo_cmdsn_list,
+ ooo_list) {
+ while (ooo_tmp->cmdsn < ooo_cmdsn->cmdsn)
+ continue;
+
+ list_add(&ooo_cmdsn->ooo_list,
+ &ooo_tmp->ooo_list);
+ break;
+ }
+ }
+ }
+ sess->ooo_cmdsn_count++;
+
+ TRACE(TRACE_CMDSN, "Set out of order CmdSN count for SID:"
+ " %u to %hu.\n", sess->sid, sess->ooo_cmdsn_count);
+
+ return 0;
+}
+
+/*
+ * Removes an struct iscsi_ooo_cmdsn from a session's list,
+ * called with struct iscsi_session->cmdsn_lock held.
+ */
+void iscsit_remove_ooo_cmdsn(
+ struct iscsi_session *sess,
+ struct iscsi_ooo_cmdsn *ooo_cmdsn)
+{
+ list_del(&ooo_cmdsn->ooo_list);
+ kmem_cache_free(lio_ooo_cache, ooo_cmdsn);
+}
+
+void iscsit_clear_ooo_cmdsns_for_conn(struct iscsi_conn *conn)
+{
+ struct iscsi_ooo_cmdsn *ooo_cmdsn;
+ struct iscsi_session *sess = conn->sess;
+
+ spin_lock(&sess->cmdsn_lock);
+ list_for_each_entry(ooo_cmdsn, &sess->sess_ooo_cmdsn_list, ooo_list) {
+ if (ooo_cmdsn->cid != conn->cid)
+ continue;
+
+ ooo_cmdsn->cmd = NULL;
+ }
+ spin_unlock(&sess->cmdsn_lock);
+}
+
+/*
+ * Called with sess->cmdsn_lock held.
+ */
+int iscsit_execute_ooo_cmdsns(struct iscsi_session *sess)
+{
+ int ooo_count = 0;
+ struct iscsi_cmd *cmd = NULL;
+ struct iscsi_ooo_cmdsn *ooo_cmdsn, *ooo_cmdsn_tmp;
+
+ list_for_each_entry_safe(ooo_cmdsn, ooo_cmdsn_tmp,
+ &sess->sess_ooo_cmdsn_list, ooo_list) {
+ if (ooo_cmdsn->cmdsn != sess->exp_cmd_sn)
+ continue;
+
+ if (!ooo_cmdsn->cmd) {
+ sess->exp_cmd_sn++;
+ iscsit_remove_ooo_cmdsn(sess, ooo_cmdsn);
+ continue;
+ }
+
+ cmd = ooo_cmdsn->cmd;
+ cmd->i_state = cmd->deferred_i_state;
+ ooo_count++;
+ sess->exp_cmd_sn++;
+ TRACE(TRACE_CMDSN, "Executing out of order CmdSN: 0x%08x,"
+ " incremented ExpCmdSN to 0x%08x.\n",
+ cmd->cmd_sn, sess->exp_cmd_sn);
+
+ iscsit_remove_ooo_cmdsn(sess, ooo_cmdsn);
+
+ if (iscsit_execute_cmd(cmd, 1) < 0)
+ return -1;
+
+ continue;
+ }
+
+ return ooo_count;
+}
+
+/*
+ * Called either:
+ *
+ * 1. With sess->cmdsn_lock held from iscsi_execute_ooo_cmdsns()
+ * or iscsi_check_received_cmdsn().
+ * 2. With no locks held directly from iscsi_handle_XXX_pdu() functions
+ * for immediate commands.
+ */
+int iscsit_execute_cmd(struct iscsi_cmd *cmd, int ooo)
+{
+ struct se_cmd *se_cmd = &cmd->se_cmd;
+ int lr = 0;
+
+ spin_lock_bh(&cmd->istate_lock);
+ if (ooo)
+ cmd->cmd_flags &= ~ICF_OOO_CMDSN;
+
+ switch (cmd->iscsi_opcode) {
+ case ISCSI_OP_SCSI_CMD:
+ /*
+ * Go ahead and send the CHECK_CONDITION status for
+ * any SCSI CDB exceptions that may have occurred, also
+ * handle the SCF_SCSI_RESERVATION_CONFLICT case here as well.
+ */
+ if (se_cmd->se_cmd_flags & SCF_SCSI_CDB_EXCEPTION) {
+ if (se_cmd->se_cmd_flags &
+ SCF_SCSI_RESERVATION_CONFLICT) {
+ cmd->i_state = ISTATE_SEND_STATUS;
+ spin_unlock_bh(&cmd->istate_lock);
+ iscsit_add_cmd_to_response_queue(cmd, cmd->conn,
+ cmd->i_state);
+ return 0;
+ }
+ spin_unlock_bh(&cmd->istate_lock);
+ /*
+ * Determine if delayed TASK_ABORTED status for WRITEs
+ * should be sent now if no unsolicited data out
+ * payloads are expected, or if the delayed status
+ * should be sent after unsolicited data out with
+ * ISCSI_FLAG_CMD_FINAL set in iscsi_handle_data_out()
+ */
+ if (transport_check_aborted_status(se_cmd,
+ (cmd->unsolicited_data == 0)) != 0)
+ return 0;
+ /*
+ * Otherwise send CHECK_CONDITION and sense for
+ * exception
+ */
+ return transport_send_check_condition_and_sense(se_cmd,
+ se_cmd->scsi_sense_reason, 0);
+ }
+ /*
+ * Special case for delayed CmdSN with Immediate
+ * Data and/or Unsolicited Data Out attached.
+ */
+ if (cmd->immediate_data) {
+ if (cmd->cmd_flags & ICF_GOT_LAST_DATAOUT) {
+ spin_unlock_bh(&cmd->istate_lock);
+ return transport_generic_handle_data(
+ &cmd->se_cmd);
+ }
+ spin_unlock_bh(&cmd->istate_lock);
+
+ if (!(cmd->cmd_flags &
+ ICF_NON_IMMEDIATE_UNSOLICITED_DATA)) {
+ /*
+ * Send the delayed TASK_ABORTED status for
+ * WRITEs if no more unsolicitied data is
+ * expected.
+ */
+ if (transport_check_aborted_status(se_cmd, 1)
+ != 0)
+ return 0;
+
+ iscsit_set_dataout_sequence_values(cmd);
+ iscsit_build_r2ts_for_cmd(cmd, cmd->conn, 0);
+ }
+ return 0;
+ }
+ /*
+ * The default handler.
+ */
+ spin_unlock_bh(&cmd->istate_lock);
+
+ if ((cmd->data_direction == DMA_TO_DEVICE) &&
+ !(cmd->cmd_flags & ICF_NON_IMMEDIATE_UNSOLICITED_DATA)) {
+ /*
+ * Send the delayed TASK_ABORTED status for WRITEs if
+ * no more nsolicitied data is expected.
+ */
+ if (transport_check_aborted_status(se_cmd, 1) != 0)
+ return 0;
+
+ iscsit_set_dataout_sequence_values(cmd);
+ spin_lock_bh(&cmd->dataout_timeout_lock);
+ iscsit_start_dataout_timer(cmd, cmd->conn);
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ }
+ return transport_generic_handle_cdb(&cmd->se_cmd);
+
+ case ISCSI_OP_NOOP_OUT:
+ case ISCSI_OP_TEXT:
+ spin_unlock_bh(&cmd->istate_lock);
+ iscsit_add_cmd_to_response_queue(cmd, cmd->conn, cmd->i_state);
+ break;
+ case ISCSI_OP_SCSI_TMFUNC:
+ if (se_cmd->se_cmd_flags & SCF_SCSI_CDB_EXCEPTION) {
+ spin_unlock_bh(&cmd->istate_lock);
+ iscsit_add_cmd_to_response_queue(cmd, cmd->conn,
+ cmd->i_state);
+ return 0;
+ }
+ spin_unlock_bh(&cmd->istate_lock);
+
+ return transport_generic_handle_tmr(SE_CMD(cmd));
+ case ISCSI_OP_LOGOUT:
+ spin_unlock_bh(&cmd->istate_lock);
+ switch (cmd->logout_reason) {
+ case ISCSI_LOGOUT_REASON_CLOSE_SESSION:
+ lr = iscsit_logout_closesession(cmd, cmd->conn);
+ break;
+ case ISCSI_LOGOUT_REASON_CLOSE_CONNECTION:
+ lr = iscsit_logout_closeconnection(cmd, cmd->conn);
+ break;
+ case ISCSI_LOGOUT_REASON_RECOVERY:
+ lr = iscsit_logout_removeconnforrecovery(cmd, cmd->conn);
+ break;
+ default:
+ printk(KERN_ERR "Unknown iSCSI Logout Request Code:"
+ " 0x%02x\n", cmd->logout_reason);
+ return -1;
+ }
+
+ return lr;
+ default:
+ spin_unlock_bh(&cmd->istate_lock);
+ printk(KERN_ERR "Cannot perform out of order execution for"
+ " unknown iSCSI Opcode: 0x%02x\n", cmd->iscsi_opcode);
+ return -1;
+ }
+
+ return 0;
+}
+
+void iscsit_free_all_ooo_cmdsns(struct iscsi_session *sess)
+{
+ struct iscsi_ooo_cmdsn *ooo_cmdsn, *ooo_cmdsn_tmp;
+
+ spin_lock(&sess->cmdsn_lock);
+ list_for_each_entry_safe(ooo_cmdsn, ooo_cmdsn_tmp,
+ &sess->sess_ooo_cmdsn_list, ooo_list) {
+
+ list_del(&ooo_cmdsn->ooo_list);
+ kmem_cache_free(lio_ooo_cache, ooo_cmdsn);
+ }
+ spin_unlock(&sess->cmdsn_lock);
+}
+
+int iscsit_handle_ooo_cmdsn(
+ struct iscsi_session *sess,
+ struct iscsi_cmd *cmd,
+ u32 cmdsn)
+{
+ int batch = 0;
+ struct iscsi_ooo_cmdsn *ooo_cmdsn = NULL, *ooo_tail = NULL;
+
+ sess->cmdsn_outoforder = 1;
+
+ cmd->deferred_i_state = cmd->i_state;
+ cmd->i_state = ISTATE_DEFERRED_CMD;
+ cmd->cmd_flags |= ICF_OOO_CMDSN;
+
+ if (list_empty(&sess->sess_ooo_cmdsn_list))
+ batch = 1;
+ else {
+ ooo_tail = list_entry(sess->sess_ooo_cmdsn_list.prev,
+ typeof(*ooo_tail), ooo_list);
+ if (ooo_tail->cmdsn != (cmdsn - 1))
+ batch = 1;
+ }
+
+ ooo_cmdsn = iscsit_allocate_ooo_cmdsn();
+ if (!ooo_cmdsn)
+ return CMDSN_ERROR_CANNOT_RECOVER;
+
+ ooo_cmdsn->cmd = cmd;
+ ooo_cmdsn->batch_count = (batch) ?
+ (cmdsn - sess->exp_cmd_sn) : 1;
+ ooo_cmdsn->cid = cmd->conn->cid;
+ ooo_cmdsn->exp_cmdsn = sess->exp_cmd_sn;
+ ooo_cmdsn->cmdsn = cmdsn;
+
+ if (iscsit_attach_ooo_cmdsn(sess, ooo_cmdsn) < 0) {
+ kmem_cache_free(lio_ooo_cache, ooo_cmdsn);
+ return CMDSN_ERROR_CANNOT_RECOVER;
+ }
+
+ return CMDSN_HIGHER_THAN_EXP;
+}
+
+static int iscsit_set_dataout_timeout_values(
+ struct iscsi_cmd *cmd,
+ u32 *offset,
+ u32 *length)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_r2t *r2t;
+
+ if (cmd->unsolicited_data) {
+ *offset = 0;
+ *length = (conn->sess->sess_ops->FirstBurstLength >
+ cmd->data_length) ?
+ cmd->data_length :
+ conn->sess->sess_ops->FirstBurstLength;
+ return 0;
+ }
+
+ spin_lock_bh(&cmd->r2t_lock);
+ if (list_empty(&cmd->cmd_r2t_list)) {
+ printk(KERN_ERR "cmd->cmd_r2t_list is empty!\n");
+ spin_unlock_bh(&cmd->r2t_lock);
+ return -1;
+ }
+
+ list_for_each_entry(r2t, &cmd->cmd_r2t_list, r2t_list)
+ if (r2t->sent_r2t && !r2t->recovery_r2t && !r2t->seq_complete)
+ break;
+
+ if (!r2t) {
+ printk(KERN_ERR "Unable to locate any incomplete DataOUT"
+ " sequences for ITT: 0x%08x.\n", cmd->init_task_tag);
+ spin_unlock_bh(&cmd->r2t_lock);
+ return -1;
+ }
+
+ *offset = r2t->offset;
+ *length = r2t->xfer_len;
+
+ spin_unlock_bh(&cmd->r2t_lock);
+ return 0;
+}
+
+/*
+ * NOTE: Called from interrupt (timer) context.
+ */
+static void iscsit_handle_dataout_timeout(unsigned long data)
+{
+ u32 pdu_length = 0, pdu_offset = 0;
+ u32 r2t_length = 0, r2t_offset = 0;
+ struct iscsi_cmd *cmd = (struct iscsi_cmd *) data;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_session *sess = NULL;
+ struct iscsi_node_attrib *na;
+
+ iscsit_inc_conn_usage_count(conn);
+
+ spin_lock_bh(&cmd->dataout_timeout_lock);
+ if (cmd->dataout_timer_flags & ISCSI_TF_STOP) {
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ iscsit_dec_conn_usage_count(conn);
+ return;
+ }
+ cmd->dataout_timer_flags &= ~ISCSI_TF_RUNNING;
+ sess = conn->sess;
+ na = iscsit_tpg_get_node_attrib(sess);
+
+ if (!sess->sess_ops->ErrorRecoveryLevel) {
+ TRACE(TRACE_ERL0, "Unable to recover from DataOut timeout while"
+ " in ERL=0.\n");
+ goto failure;
+ }
+
+ if (++cmd->dataout_timeout_retries == na->dataout_timeout_retries) {
+ TRACE(TRACE_TIMER, "Command ITT: 0x%08x exceeded max retries"
+ " for DataOUT timeout %u, closing iSCSI connection.\n",
+ cmd->init_task_tag, na->dataout_timeout_retries);
+ goto failure;
+ }
+
+ cmd->cmd_flags |= ICF_WITHIN_COMMAND_RECOVERY;
+
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ if (conn->sess->sess_ops->DataPDUInOrder) {
+ pdu_offset = cmd->write_data_done;
+ if ((pdu_offset + (conn->sess->sess_ops->MaxBurstLength -
+ cmd->next_burst_len)) > cmd->data_length)
+ pdu_length = (cmd->data_length -
+ cmd->write_data_done);
+ else
+ pdu_length = (conn->sess->sess_ops->MaxBurstLength -
+ cmd->next_burst_len);
+ } else {
+ pdu_offset = cmd->seq_start_offset;
+ pdu_length = (cmd->seq_end_offset -
+ cmd->seq_start_offset);
+ }
+ } else {
+ if (iscsit_set_dataout_timeout_values(cmd, &pdu_offset,
+ &pdu_length) < 0)
+ goto failure;
+ }
+
+ if (iscsit_recalculate_dataout_values(cmd, pdu_offset, pdu_length,
+ &r2t_offset, &r2t_length) < 0)
+ goto failure;
+
+ TRACE(TRACE_TIMER, "Command ITT: 0x%08x timed out waiting for"
+ " completion of %sDataOUT Sequence Offset: %u, Length: %u\n",
+ cmd->init_task_tag, (cmd->unsolicited_data) ? "Unsolicited " :
+ "", r2t_offset, r2t_length);
+
+ if (iscsit_send_recovery_r2t(cmd, r2t_offset, r2t_length) < 0)
+ goto failure;
+
+ iscsit_start_dataout_timer(cmd, conn);
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ iscsit_dec_conn_usage_count(conn);
+
+ return;
+
+failure:
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ iscsit_cause_connection_reinstatement(conn, 0);
+ iscsit_dec_conn_usage_count(conn);
+
+ return;
+}
+
+void iscsit_mod_dataout_timer(struct iscsi_cmd *cmd)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na = na = iscsit_tpg_get_node_attrib(sess);
+
+ spin_lock_bh(&cmd->dataout_timeout_lock);
+ if (!(cmd->dataout_timer_flags & ISCSI_TF_RUNNING)) {
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ return;
+ }
+
+ mod_timer(&cmd->dataout_timer,
+ (get_jiffies_64() + na->dataout_timeout * HZ));
+ TRACE(TRACE_TIMER, "Updated DataOUT timer for ITT: 0x%08x",
+ cmd->init_task_tag);
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+}
+
+/*
+ * Called with cmd->dataout_timeout_lock held.
+ */
+void iscsit_start_dataout_timer(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na = na = iscsit_tpg_get_node_attrib(sess);
+
+ if (cmd->dataout_timer_flags & ISCSI_TF_RUNNING)
+ return;
+
+ TRACE(TRACE_TIMER, "Starting DataOUT timer for ITT: 0x%08x on"
+ " CID: %hu.\n", cmd->init_task_tag, conn->cid);
+
+ init_timer(&cmd->dataout_timer);
+ cmd->dataout_timer.expires = (get_jiffies_64() + na->dataout_timeout * HZ);
+ cmd->dataout_timer.data = (unsigned long)cmd;
+ cmd->dataout_timer.function = iscsit_handle_dataout_timeout;
+ cmd->dataout_timer_flags &= ~ISCSI_TF_STOP;
+ cmd->dataout_timer_flags |= ISCSI_TF_RUNNING;
+ add_timer(&cmd->dataout_timer);
+}
+
+void iscsit_stop_dataout_timer(struct iscsi_cmd *cmd)
+{
+ spin_lock_bh(&cmd->dataout_timeout_lock);
+ if (!(cmd->dataout_timer_flags & ISCSI_TF_RUNNING)) {
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+ return;
+ }
+ cmd->dataout_timer_flags |= ISCSI_TF_STOP;
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+
+ del_timer_sync(&cmd->dataout_timer);
+
+ spin_lock_bh(&cmd->dataout_timeout_lock);
+ cmd->dataout_timer_flags &= ~ISCSI_TF_RUNNING;
+ TRACE(TRACE_TIMER, "Stopped DataOUT Timer for ITT: 0x%08x\n",
+ cmd->init_task_tag);
+ spin_unlock_bh(&cmd->dataout_timeout_lock);
+}
diff --git a/drivers/target/iscsi/iscsi_target_erl1.h b/drivers/target/iscsi/iscsi_target_erl1.h
new file mode 100644
index 0000000..3118a18
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_erl1.h
@@ -0,0 +1,28 @@
+#ifndef ISCSI_TARGET_ERL1_H
+#define ISCSI_TARGET_ERL1_H
+
+extern int iscsit_dump_data_payload(struct iscsi_conn *, u32, int);
+extern int iscsit_create_recovery_datain_values_datasequenceinorder_yes(
+ struct iscsi_cmd *, struct iscsi_datain_req *);
+extern int iscsit_create_recovery_datain_values_datasequenceinorder_no(
+ struct iscsi_cmd *, struct iscsi_datain_req *);
+extern int iscsit_handle_recovery_datain_or_r2t(struct iscsi_conn *, unsigned char *,
+ u32, u32, u32, u32);
+extern int iscsit_handle_status_snack(struct iscsi_conn *, u32, u32,
+ u32, u32);
+extern int iscsit_handle_data_ack(struct iscsi_conn *, u32, u32, u32);
+extern int iscsit_dataout_datapduinorder_no_fbit(struct iscsi_cmd *, struct iscsi_pdu *);
+extern int iscsit_recover_dataout_sequence(struct iscsi_cmd *, u32, u32);
+extern void iscsit_clear_ooo_cmdsns_for_conn(struct iscsi_conn *);
+extern void iscsit_free_all_ooo_cmdsns(struct iscsi_session *);
+extern int iscsit_execute_ooo_cmdsns(struct iscsi_session *);
+extern int iscsit_execute_cmd(struct iscsi_cmd *, int);
+extern int iscsit_handle_ooo_cmdsn(struct iscsi_session *, struct iscsi_cmd *, u32);
+extern void iscsit_remove_ooo_cmdsn(struct iscsi_session *, struct iscsi_ooo_cmdsn *);
+extern void iscsit_mod_dataout_timer(struct iscsi_cmd *);
+extern void iscsit_start_dataout_timer(struct iscsi_cmd *, struct iscsi_conn *);
+extern void iscsit_stop_dataout_timer(struct iscsi_cmd *);
+
+extern struct kmem_cache *lio_ooo_cache;
+
+#endif /* ISCSI_TARGET_ERL1_H */
diff --git a/drivers/target/iscsi/iscsi_target_erl2.c b/drivers/target/iscsi/iscsi_target_erl2.c
new file mode 100644
index 0000000..a239a08
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_erl2.c
@@ -0,0 +1,483 @@
+/*******************************************************************************
+ * This file contains error recovery level two functions used by
+ * the iSCSI Target driver.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_datain_values.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_erl1.h"
+#include "iscsi_target_erl2.h"
+#include "iscsi_target.h"
+
+/*
+ * FIXME: Does RData SNACK apply here as well?
+ */
+void iscsit_create_conn_recovery_datain_values(
+ struct iscsi_cmd *cmd,
+ u32 exp_data_sn)
+{
+ u32 data_sn = 0;
+ struct iscsi_conn *conn = cmd->conn;
+
+ cmd->next_burst_len = 0;
+ cmd->read_data_done = 0;
+
+ while (exp_data_sn > data_sn) {
+ if ((cmd->next_burst_len +
+ conn->conn_ops->MaxRecvDataSegmentLength) <
+ conn->sess->sess_ops->MaxBurstLength) {
+ cmd->read_data_done +=
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ cmd->next_burst_len +=
+ conn->conn_ops->MaxRecvDataSegmentLength;
+ } else {
+ cmd->read_data_done +=
+ (conn->sess->sess_ops->MaxBurstLength -
+ cmd->next_burst_len);
+ cmd->next_burst_len = 0;
+ }
+ data_sn++;
+ }
+}
+
+void iscsit_create_conn_recovery_dataout_values(
+ struct iscsi_cmd *cmd)
+{
+ u32 write_data_done = 0;
+ struct iscsi_conn *conn = cmd->conn;
+
+ cmd->data_sn = 0;
+ cmd->next_burst_len = 0;
+
+ while (cmd->write_data_done > write_data_done) {
+ if ((write_data_done + conn->sess->sess_ops->MaxBurstLength) <=
+ cmd->write_data_done)
+ write_data_done += conn->sess->sess_ops->MaxBurstLength;
+ else
+ break;
+ }
+
+ cmd->write_data_done = write_data_done;
+}
+
+static int iscsit_attach_active_connection_recovery_entry(
+ struct iscsi_session *sess,
+ struct iscsi_conn_recovery *cr)
+{
+ spin_lock(&sess->cr_a_lock);
+ list_add_tail(&cr->cr_list, &sess->cr_active_list);
+ spin_unlock(&sess->cr_a_lock);
+
+ return 0;
+}
+
+static int iscsit_attach_inactive_connection_recovery_entry(
+ struct iscsi_session *sess,
+ struct iscsi_conn_recovery *cr)
+{
+ spin_lock(&sess->cr_i_lock);
+ list_add_tail(&cr->cr_list, &sess->cr_inactive_list);
+
+ sess->conn_recovery_count++;
+ TRACE(TRACE_ERL2, "Incremented connection recovery count to %u for"
+ " SID: %u\n", sess->conn_recovery_count, sess->sid);
+ spin_unlock(&sess->cr_i_lock);
+
+ return 0;
+}
+
+struct iscsi_conn_recovery *iscsit_get_inactive_connection_recovery_entry(
+ struct iscsi_session *sess,
+ u16 cid)
+{
+ struct iscsi_conn_recovery *cr;
+
+ spin_lock(&sess->cr_i_lock);
+ list_for_each_entry(cr, &sess->cr_inactive_list, cr_list) {
+ if (cr->cid == cid)
+ break;
+ }
+ spin_unlock(&sess->cr_i_lock);
+
+ return (cr) ? cr : NULL;
+}
+
+void iscsit_free_connection_recovery_entires(struct iscsi_session *sess)
+{
+ struct iscsi_cmd *cmd, *cmd_tmp;
+ struct iscsi_conn_recovery *cr, *cr_tmp;
+
+ spin_lock(&sess->cr_a_lock);
+ list_for_each_entry_safe(cr, cr_tmp, &sess->cr_active_list, cr_list) {
+ list_del(&cr->cr_list);
+ spin_unlock(&sess->cr_a_lock);
+
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ list_for_each_entry_safe(cmd, cmd_tmp,
+ &cr->conn_recovery_cmd_list, i_list) {
+
+ list_del(&cmd->i_list);
+ cmd->conn = NULL;
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+ if (!(SE_CMD(cmd)) ||
+ !(SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD) ||
+ !(SE_CMD(cmd)->transport_wait_for_tasks))
+ iscsit_release_cmd(cmd);
+ else
+ SE_CMD(cmd)->transport_wait_for_tasks(
+ SE_CMD(cmd), 1, 1);
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ }
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+ spin_lock(&sess->cr_a_lock);
+
+ kfree(cr);
+ }
+ spin_unlock(&sess->cr_a_lock);
+
+ spin_lock(&sess->cr_i_lock);
+ list_for_each_entry_safe(cr, cr_tmp, &sess->cr_inactive_list, cr_list) {
+ list_del(&cr->cr_list);
+ spin_unlock(&sess->cr_i_lock);
+
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ list_for_each_entry_safe(cmd, cmd_tmp,
+ &cr->conn_recovery_cmd_list, i_list) {
+
+ list_del(&cmd->i_list);
+ cmd->conn = NULL;
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+ if (!(SE_CMD(cmd)) ||
+ !(SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD) ||
+ !(SE_CMD(cmd)->transport_wait_for_tasks))
+ iscsit_release_cmd(cmd);
+ else
+ SE_CMD(cmd)->transport_wait_for_tasks(
+ SE_CMD(cmd), 1, 1);
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ }
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+ spin_lock(&sess->cr_i_lock);
+
+ kfree(cr);
+ }
+ spin_unlock(&sess->cr_i_lock);
+}
+
+int iscsit_remove_active_connection_recovery_entry(
+ struct iscsi_conn_recovery *cr,
+ struct iscsi_session *sess)
+{
+ spin_lock(&sess->cr_a_lock);
+ list_del(&cr->cr_list);
+
+ sess->conn_recovery_count--;
+ TRACE(TRACE_ERL2, "Decremented connection recovery count to %u for"
+ " SID: %u\n", sess->conn_recovery_count, sess->sid);
+ spin_unlock(&sess->cr_a_lock);
+
+ kfree(cr);
+
+ return 0;
+}
+
+int iscsit_remove_inactive_connection_recovery_entry(
+ struct iscsi_conn_recovery *cr,
+ struct iscsi_session *sess)
+{
+ spin_lock(&sess->cr_i_lock);
+ list_del(&cr->cr_list);
+ spin_unlock(&sess->cr_i_lock);
+
+ return 0;
+}
+
+/*
+ * Called with cr->conn_recovery_cmd_lock help.
+ */
+int iscsit_remove_cmd_from_connection_recovery(
+ struct iscsi_cmd *cmd,
+ struct iscsi_session *sess)
+{
+ struct iscsi_conn_recovery *cr;
+
+ if (!cmd->cr) {
+ printk(KERN_ERR "struct iscsi_conn_recovery pointer for ITT: 0x%08x"
+ " is NULL!\n", cmd->init_task_tag);
+ BUG();
+ }
+ cr = cmd->cr;
+
+ list_del(&cmd->i_list);
+ return --cr->cmd_count;
+}
+
+void iscsit_discard_cr_cmds_by_expstatsn(
+ struct iscsi_conn_recovery *cr,
+ u32 exp_statsn)
+{
+ u32 dropped_count = 0;
+ struct iscsi_cmd *cmd, *cmd_tmp;
+ struct iscsi_session *sess = cr->sess;
+
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ list_for_each_entry_safe(cmd, cmd_tmp,
+ &cr->conn_recovery_cmd_list, i_list) {
+
+ if (((cmd->deferred_i_state != ISTATE_SENT_STATUS) &&
+ (cmd->deferred_i_state != ISTATE_REMOVE)) ||
+ (cmd->stat_sn >= exp_statsn)) {
+ continue;
+ }
+
+ dropped_count++;
+ TRACE(TRACE_ERL2, "Dropping Acknowledged ITT: 0x%08x, StatSN:"
+ " 0x%08x, CID: %hu.\n", cmd->init_task_tag,
+ cmd->stat_sn, cr->cid);
+
+ iscsit_remove_cmd_from_connection_recovery(cmd, sess);
+
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+ if (!(SE_CMD(cmd)) ||
+ !(SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD) ||
+ !(SE_CMD(cmd)->transport_wait_for_tasks))
+ iscsit_release_cmd(cmd);
+ else
+ SE_CMD(cmd)->transport_wait_for_tasks(
+ SE_CMD(cmd), 1, 0);
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ }
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+
+ TRACE(TRACE_ERL2, "Dropped %u total acknowledged commands on"
+ " CID: %hu less than old ExpStatSN: 0x%08x\n",
+ dropped_count, cr->cid, exp_statsn);
+
+ if (!cr->cmd_count) {
+ TRACE(TRACE_ERL2, "No commands to be reassigned for failed"
+ " connection CID: %hu on SID: %u\n",
+ cr->cid, sess->sid);
+ iscsit_remove_inactive_connection_recovery_entry(cr, sess);
+ iscsit_attach_active_connection_recovery_entry(sess, cr);
+ printk(KERN_INFO "iSCSI connection recovery successful for CID:"
+ " %hu on SID: %u\n", cr->cid, sess->sid);
+ iscsit_remove_active_connection_recovery_entry(cr, sess);
+ } else {
+ iscsit_remove_inactive_connection_recovery_entry(cr, sess);
+ iscsit_attach_active_connection_recovery_entry(sess, cr);
+ }
+
+ return;
+}
+
+int iscsit_discard_unacknowledged_ooo_cmdsns_for_conn(struct iscsi_conn *conn)
+{
+ u32 dropped_count = 0;
+ struct iscsi_cmd *cmd, *cmd_tmp;
+ struct iscsi_ooo_cmdsn *ooo_cmdsn, *ooo_cmdsn_tmp;
+ struct iscsi_session *sess = conn->sess;
+
+ spin_lock(&sess->cmdsn_lock);
+ list_for_each_entry_safe(ooo_cmdsn, ooo_cmdsn_tmp,
+ &sess->sess_ooo_cmdsn_list, ooo_list) {
+
+ if (ooo_cmdsn->cid != conn->cid)
+ continue;
+
+ dropped_count++;
+ TRACE(TRACE_ERL2, "Dropping unacknowledged CmdSN:"
+ " 0x%08x during connection recovery on CID: %hu\n",
+ ooo_cmdsn->cmdsn, conn->cid);
+ iscsit_remove_ooo_cmdsn(sess, ooo_cmdsn);
+ }
+ conn->sess->ooo_cmdsn_count -= dropped_count;
+ spin_unlock(&sess->cmdsn_lock);
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry_safe(cmd, cmd_tmp, &conn->conn_cmd_list, i_list) {
+ if (!(cmd->cmd_flags & ICF_OOO_CMDSN))
+ continue;
+
+ list_del(&cmd->i_list);
+
+ spin_unlock_bh(&conn->cmd_lock);
+ if (!(SE_CMD(cmd)) ||
+ !(SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD) ||
+ !(SE_CMD(cmd)->transport_wait_for_tasks))
+ iscsit_release_cmd(cmd);
+ else
+ SE_CMD(cmd)->transport_wait_for_tasks(
+ SE_CMD(cmd), 1, 1);
+ spin_lock_bh(&conn->cmd_lock);
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+
+ TRACE(TRACE_ERL2, "Dropped %u total unacknowledged commands on CID:"
+ " %hu for ExpCmdSN: 0x%08x.\n", dropped_count, conn->cid,
+ sess->exp_cmd_sn);
+ return 0;
+}
+
+int iscsit_prepare_cmds_for_realligance(struct iscsi_conn *conn)
+{
+ u32 cmd_count = 0;
+ struct iscsi_cmd *cmd, *cmd_tmp;
+ struct iscsi_conn_recovery *cr;
+
+ /*
+ * Allocate an struct iscsi_conn_recovery for this connection.
+ * Each struct iscsi_cmd contains an struct iscsi_conn_recovery pointer
+ * (struct iscsi_cmd->cr) so we need to allocate this before preparing the
+ * connection's command list for connection recovery.
+ */
+ cr = kzalloc(sizeof(struct iscsi_conn_recovery), GFP_KERNEL);
+ if (!cr) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_conn_recovery.\n");
+ return -1;
+ }
+ INIT_LIST_HEAD(&cr->cr_list);
+ INIT_LIST_HEAD(&cr->conn_recovery_cmd_list);
+ spin_lock_init(&cr->conn_recovery_cmd_lock);
+ /*
+ * Only perform connection recovery on ISCSI_OP_SCSI_CMD or
+ * ISCSI_OP_NOOP_OUT opcodes. For all other opcodes call
+ * list_del(&cmd->i_list); to release the command to the
+ * session pool and remove it from the connection's list.
+ *
+ * Also stop the DataOUT timer, which will be restarted after
+ * sending the TMR response.
+ */
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry_safe(cmd, cmd_tmp, &conn->conn_cmd_list, i_list) {
+
+ if ((cmd->iscsi_opcode != ISCSI_OP_SCSI_CMD) &&
+ (cmd->iscsi_opcode != ISCSI_OP_NOOP_OUT)) {
+ TRACE(TRACE_ERL2, "Not performing realligence on"
+ " Opcode: 0x%02x, ITT: 0x%08x, CmdSN: 0x%08x,"
+ " CID: %hu\n", cmd->iscsi_opcode,
+ cmd->init_task_tag, cmd->cmd_sn, conn->cid);
+
+ list_del(&cmd->i_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (!(SE_CMD(cmd)) ||
+ !(SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD) ||
+ !(SE_CMD(cmd)->transport_wait_for_tasks))
+ iscsit_release_cmd(cmd);
+ else
+ SE_CMD(cmd)->transport_wait_for_tasks(
+ SE_CMD(cmd), 1, 0);
+ spin_lock_bh(&conn->cmd_lock);
+ continue;
+ }
+
+ /*
+ * Special case where commands greater than or equal to
+ * the session's ExpCmdSN are attached to the connection
+ * list but not to the out of order CmdSN list. The one
+ * obvious case is when a command with immediate data
+ * attached must only check the CmdSN against ExpCmdSN
+ * after the data is received. The special case below
+ * is when the connection fails before data is received,
+ * but also may apply to other PDUs, so it has been
+ * made generic here.
+ */
+ if (!(cmd->cmd_flags & ICF_OOO_CMDSN) && !cmd->immediate_cmd &&
+ (cmd->cmd_sn >= conn->sess->exp_cmd_sn)) {
+ list_del(&cmd->i_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (!(SE_CMD(cmd)) ||
+ !(SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD) ||
+ !(SE_CMD(cmd)->transport_wait_for_tasks))
+ iscsit_release_cmd(cmd);
+ else
+ SE_CMD(cmd)->transport_wait_for_tasks(
+ SE_CMD(cmd), 1, 1);
+ spin_lock_bh(&conn->cmd_lock);
+ continue;
+ }
+
+ cmd_count++;
+ TRACE(TRACE_ERL2, "Preparing Opcode: 0x%02x, ITT: 0x%08x,"
+ " CmdSN: 0x%08x, StatSN: 0x%08x, CID: %hu for"
+ " realligence.\n", cmd->iscsi_opcode,
+ cmd->init_task_tag, cmd->cmd_sn, cmd->stat_sn,
+ conn->cid);
+
+ cmd->deferred_i_state = cmd->i_state;
+ cmd->i_state = ISTATE_IN_CONNECTION_RECOVERY;
+
+ if (cmd->data_direction == DMA_TO_DEVICE)
+ iscsit_stop_dataout_timer(cmd);
+
+ cmd->sess = conn->sess;
+
+ list_del(&cmd->i_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ iscsit_free_all_datain_reqs(cmd);
+
+ if ((SE_CMD(cmd)) &&
+ (SE_CMD(cmd)->se_cmd_flags & SCF_SE_LUN_CMD) &&
+ SE_CMD(cmd)->transport_wait_for_tasks)
+ SE_CMD(cmd)->transport_wait_for_tasks(SE_CMD(cmd),
+ 0, 0);
+ /*
+ * Add the struct iscsi_cmd to the connection recovery cmd list
+ */
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ list_add_tail(&cmd->i_list, &cr->conn_recovery_cmd_list);
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+
+ spin_lock_bh(&conn->cmd_lock);
+ cmd->cr = cr;
+ cmd->conn = NULL;
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+ /*
+ * Fill in the various values in the preallocated struct iscsi_conn_recovery.
+ */
+ cr->cid = conn->cid;
+ cr->cmd_count = cmd_count;
+ cr->maxrecvdatasegmentlength = conn->conn_ops->MaxRecvDataSegmentLength;
+ cr->sess = conn->sess;
+
+ iscsit_attach_inactive_connection_recovery_entry(conn->sess, cr);
+
+ return 0;
+}
+
+int iscsit_connection_recovery_transport_reset(struct iscsi_conn *conn)
+{
+ atomic_set(&conn->connection_recovery, 1);
+
+ if (iscsit_close_connection(conn) < 0)
+ return -1;
+
+ return 0;
+}
diff --git a/drivers/target/iscsi/iscsi_target_erl2.h b/drivers/target/iscsi/iscsi_target_erl2.h
new file mode 100644
index 0000000..22f8d24
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_erl2.h
@@ -0,0 +1,18 @@
+#ifndef ISCSI_TARGET_ERL2_H
+#define ISCSI_TARGET_ERL2_H
+
+extern void iscsit_create_conn_recovery_datain_values(struct iscsi_cmd *, u32);
+extern void iscsit_create_conn_recovery_dataout_values(struct iscsi_cmd *);
+extern struct iscsi_conn_recovery *iscsit_get_inactive_connection_recovery_entry(
+ struct iscsi_session *, u16);
+extern void iscsit_free_connection_recovery_entires(struct iscsi_session *);
+extern int iscsit_remove_active_connection_recovery_entry(
+ struct iscsi_conn_recovery *, struct iscsi_session *);
+extern int iscsit_remove_cmd_from_connection_recovery(struct iscsi_cmd *,
+ struct iscsi_session *);
+extern void iscsit_discard_cr_cmds_by_expstatsn(struct iscsi_conn_recovery *, u32);
+extern int iscsit_discard_unacknowledged_ooo_cmdsns_for_conn(struct iscsi_conn *);
+extern int iscsit_prepare_cmds_for_realligance(struct iscsi_conn *);
+extern int iscsit_connection_recovery_transport_reset(struct iscsi_conn *);
+
+#endif /*** ISCSI_TARGET_ERL2_H ***/
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 12/14] iscsi-target: Add support for task management operations
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (11 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 11/14] iscsi-target: Add iSCSI Error Recovery Hierarchy support Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 13/14] iscsi-target: Add misc utility and debug logic Nicholas A. Bellinger
` (2 subsequent siblings)
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This patch adds support for iSCSI task management operations called
directly from iscsi_target.c TMR request/response PDU logic, and
interfaces struct se_lun -> struct se_device for assoication
of TMRs to TCM backend devices.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_tmr.c | 854 +++++++++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_tmr.h | 14 +
2 files changed, 868 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_tmr.c
create mode 100644 drivers/target/iscsi/iscsi_target_tmr.h
diff --git a/drivers/target/iscsi/iscsi_target_tmr.c b/drivers/target/iscsi/iscsi_target_tmr.c
new file mode 100644
index 0000000..43121c0
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_tmr.c
@@ -0,0 +1,854 @@
+/*******************************************************************************
+ * This file contains the iSCSI Target specific Task Management functions.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <asm/unaligned.h>
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_seq_pdu_list.h"
+#include "iscsi_target_datain_values.h"
+#include "iscsi_target_device.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_erl1.h"
+#include "iscsi_target_erl2.h"
+#include "iscsi_target_tmr.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target.h"
+
+u8 iscsit_tmr_abort_task(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ struct iscsi_cmd *ref_cmd;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_tmr_req *tmr_req = cmd->tmr_req;
+ struct se_tmr_req *se_tmr = SE_CMD(cmd)->se_tmr_req;
+ struct iscsi_tm *hdr = (struct iscsi_tm *) buf;
+
+ ref_cmd = iscsit_find_cmd_from_itt(conn, hdr->rtt);
+ if (!ref_cmd) {
+ printk(KERN_ERR "Unable to locate RefTaskTag: 0x%08x on CID:"
+ " %hu.\n", hdr->rtt, conn->cid);
+ return ((hdr->refcmdsn >= conn->sess->exp_cmd_sn) &&
+ (hdr->refcmdsn <= conn->sess->max_cmd_sn)) ?
+ ISCSI_TMF_RSP_COMPLETE : ISCSI_TMF_RSP_NO_TASK;
+ }
+ if (ref_cmd->cmd_sn != hdr->refcmdsn) {
+ printk(KERN_ERR "RefCmdSN 0x%08x does not equal"
+ " task's CmdSN 0x%08x. Rejecting ABORT_TASK.\n",
+ hdr->refcmdsn, ref_cmd->cmd_sn);
+ return ISCSI_TMF_RSP_REJECTED;
+ }
+
+ se_tmr->ref_task_tag = hdr->rtt;
+ se_tmr->ref_cmd = &ref_cmd->se_cmd;
+ tmr_req->ref_cmd_sn = hdr->refcmdsn;
+ tmr_req->exp_data_sn = hdr->exp_datasn;
+
+ return ISCSI_TMF_RSP_COMPLETE;
+}
+
+/*
+ * Called from iscsit_handle_task_mgt_cmd().
+ */
+int iscsit_tmr_task_warm_reset(
+ struct iscsi_conn *conn,
+ struct iscsi_tmr_req *tmr_req,
+ unsigned char *buf)
+{
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na = iscsit_tpg_get_node_attrib(sess);
+#if 0
+ struct iscsi_init_task_mgt_cmnd *hdr =
+ (struct iscsi_init_task_mgt_cmnd *) buf;
+#endif
+ if (!na->tmr_warm_reset) {
+ printk(KERN_ERR "TMR Opcode TARGET_WARM_RESET authorization"
+ " failed for Initiator Node: %s\n",
+ sess->se_sess->se_node_acl->initiatorname);
+ return -1;
+ }
+ /*
+ * Do the real work in transport_generic_do_tmr().
+ */
+ return 0;
+}
+
+int iscsit_tmr_task_cold_reset(
+ struct iscsi_conn *conn,
+ struct iscsi_tmr_req *tmr_req,
+ unsigned char *buf)
+{
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na = iscsit_tpg_get_node_attrib(sess);
+
+ if (!na->tmr_cold_reset) {
+ printk(KERN_ERR "TMR Opcode TARGET_COLD_RESET authorization"
+ " failed for Initiator Node: %s\n",
+ sess->se_sess->se_node_acl->initiatorname);
+ return -1;
+ }
+ /*
+ * Do the real work in transport_generic_do_tmr().
+ */
+ return 0;
+}
+
+u8 iscsit_tmr_task_reassign(
+ struct iscsi_cmd *cmd,
+ unsigned char *buf)
+{
+ struct iscsi_cmd *ref_cmd = NULL;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_conn_recovery *cr = NULL;
+ struct iscsi_tmr_req *tmr_req = cmd->tmr_req;
+ struct se_tmr_req *se_tmr = SE_CMD(cmd)->se_tmr_req;
+ struct iscsi_tm *hdr = (struct iscsi_tm *) buf;
+ int ret;
+
+ TRACE(TRACE_ERL2, "Got TASK_REASSIGN TMR ITT: 0x%08x,"
+ " RefTaskTag: 0x%08x, ExpDataSN: 0x%08x, CID: %hu\n",
+ hdr->itt, hdr->rtt, hdr->exp_datasn, conn->cid);
+
+ if (conn->sess->sess_ops->ErrorRecoveryLevel != 2) {
+ printk(KERN_ERR "TMR TASK_REASSIGN not supported in ERL<2,"
+ " ignoring request.\n");
+ return ISCSI_TMF_RSP_NOT_SUPPORTED;
+ }
+
+ ret = iscsit_find_cmd_for_recovery(conn->sess, &ref_cmd, &cr, hdr->rtt);
+ if (ret == -2) {
+ printk(KERN_ERR "Command ITT: 0x%08x is still alligent to CID:"
+ " %hu\n", ref_cmd->init_task_tag, cr->cid);
+ return ISCSI_TMF_RSP_TASK_ALLEGIANT;
+ } else if (ret == -1) {
+ printk(KERN_ERR "Unable to locate RefTaskTag: 0x%08x in"
+ " connection recovery command list.\n", hdr->rtt);
+ return ISCSI_TMF_RSP_NO_TASK;
+ }
+ /*
+ * Temporary check to prevent connection recovery for
+ * connections with a differing MaxRecvDataSegmentLength.
+ */
+ if (cr->maxrecvdatasegmentlength !=
+ conn->conn_ops->MaxRecvDataSegmentLength) {
+ printk(KERN_ERR "Unable to perform connection recovery for"
+ " differing MaxRecvDataSegmentLength, rejecting"
+ " TMR TASK_REASSIGN.\n");
+ return ISCSI_TMF_RSP_REJECTED;
+ }
+
+ se_tmr->ref_task_tag = hdr->rtt;
+ se_tmr->ref_cmd = &ref_cmd->se_cmd;
+ se_tmr->ref_task_lun = get_unaligned_le64(&hdr->lun[0]);
+ tmr_req->ref_cmd_sn = hdr->refcmdsn;
+ tmr_req->exp_data_sn = hdr->exp_datasn;
+ tmr_req->conn_recovery = cr;
+ tmr_req->task_reassign = 1;
+ /*
+ * Command can now be reassigned to a new connection.
+ * The task management response must be sent before the
+ * reassignment actually happens. See iscsi_tmr_post_handler().
+ */
+ return ISCSI_TMF_RSP_COMPLETE;
+}
+
+static void iscsit_task_reassign_remove_cmd(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn_recovery *cr,
+ struct iscsi_session *sess)
+{
+ int ret;
+
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ ret = iscsit_remove_cmd_from_connection_recovery(cmd, sess);
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+ if (!ret) {
+ printk(KERN_INFO "iSCSI connection recovery successful for CID:"
+ " %hu on SID: %u\n", cr->cid, sess->sid);
+ iscsit_remove_active_connection_recovery_entry(cr, sess);
+ }
+
+ return;
+}
+
+static int iscsit_task_reassign_complete_nop_out(
+ struct iscsi_tmr_req *tmr_req,
+ struct iscsi_conn *conn)
+{
+ struct se_tmr_req *se_tmr = tmr_req->se_tmr_req;
+ struct se_cmd *se_cmd = se_tmr->ref_cmd;
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+ struct iscsi_conn_recovery *cr;
+
+ if (!cmd->cr) {
+ printk(KERN_ERR "struct iscsi_conn_recovery pointer for ITT: 0x%08x"
+ " is NULL!\n", cmd->init_task_tag);
+ return -1;
+ }
+ cr = cmd->cr;
+
+ /*
+ * Reset the StatSN so a new one for this commands new connection
+ * will be assigned.
+ * Reset the ExpStatSN as well so we may receive Status SNACKs.
+ */
+ cmd->stat_sn = cmd->exp_stat_sn = 0;
+
+ iscsit_task_reassign_remove_cmd(cmd, cr, conn->sess);
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ cmd->i_state = ISTATE_SEND_NOPIN;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ return 0;
+}
+
+static int iscsit_task_reassign_complete_write(
+ struct iscsi_cmd *cmd,
+ struct iscsi_tmr_req *tmr_req)
+{
+ int no_build_r2ts = 0;
+ u32 length = 0, offset = 0;
+ struct iscsi_conn *conn = cmd->conn;
+ struct se_cmd *se_cmd = SE_CMD(cmd);
+ /*
+ * The Initiator must not send a R2T SNACK with a Begrun less than
+ * the TMR TASK_REASSIGN's ExpDataSN.
+ */
+ if (!tmr_req->exp_data_sn) {
+ cmd->cmd_flags &= ~ICF_GOT_DATACK_SNACK;
+ cmd->acked_data_sn = 0;
+ } else {
+ cmd->cmd_flags |= ICF_GOT_DATACK_SNACK;
+ cmd->acked_data_sn = (tmr_req->exp_data_sn - 1);
+ }
+
+ /*
+ * The TMR TASK_REASSIGN's ExpDataSN contains the next R2TSN the
+ * Initiator is expecting. The Target controls all WRITE operations
+ * so if we have received all DataOUT we can safety ignore Initiator.
+ */
+ if (cmd->cmd_flags & ICF_GOT_LAST_DATAOUT) {
+ if (!atomic_read(&cmd->transport_sent)) {
+ TRACE(TRACE_ERL2, "WRITE ITT: 0x%08x: t_state: %d"
+ " never sent to transport\n",
+ cmd->init_task_tag, cmd->se_cmd.t_state);
+ return transport_generic_handle_data(se_cmd);
+ }
+
+ cmd->i_state = ISTATE_SEND_STATUS;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ return 0;
+ }
+
+ /*
+ * Special case to deal with DataSequenceInOrder=No and Non-Immeidate
+ * Unsolicited DataOut.
+ */
+ if (cmd->unsolicited_data) {
+ cmd->unsolicited_data = 0;
+
+ offset = cmd->next_burst_len = cmd->write_data_done;
+
+ if ((conn->sess->sess_ops->FirstBurstLength - offset) >=
+ cmd->data_length) {
+ no_build_r2ts = 1;
+ length = (cmd->data_length - offset);
+ } else
+ length = (conn->sess->sess_ops->FirstBurstLength - offset);
+
+ spin_lock_bh(&cmd->r2t_lock);
+ if (iscsit_add_r2t_to_list(cmd, offset, length, 0, 0) < 0) {
+ spin_unlock_bh(&cmd->r2t_lock);
+ return -1;
+ }
+ cmd->outstanding_r2ts++;
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ if (no_build_r2ts)
+ return 0;
+ }
+ /*
+ * iscsit_build_r2ts_for_cmd() can handle the rest from here.
+ */
+ return iscsit_build_r2ts_for_cmd(cmd, conn, 2);
+}
+
+static int iscsit_task_reassign_complete_read(
+ struct iscsi_cmd *cmd,
+ struct iscsi_tmr_req *tmr_req)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_datain_req *dr;
+ struct se_cmd *se_cmd = SE_CMD(cmd);
+ /*
+ * The Initiator must not send a Data SNACK with a BegRun less than
+ * the TMR TASK_REASSIGN's ExpDataSN.
+ */
+ if (!tmr_req->exp_data_sn) {
+ cmd->cmd_flags &= ~ICF_GOT_DATACK_SNACK;
+ cmd->acked_data_sn = 0;
+ } else {
+ cmd->cmd_flags |= ICF_GOT_DATACK_SNACK;
+ cmd->acked_data_sn = (tmr_req->exp_data_sn - 1);
+ }
+
+ if (!atomic_read(&cmd->transport_sent)) {
+ printk(KERN_INFO "READ ITT: 0x%08x: t_state: %d never sent to"
+ " transport\n", cmd->init_task_tag,
+ SE_CMD(cmd)->t_state);
+ transport_generic_handle_cdb(se_cmd);
+ return 0;
+ }
+
+ if (!atomic_read(&T_TASK(se_cmd)->t_transport_complete)) {
+ printk(KERN_ERR "READ ITT: 0x%08x: t_state: %d, never returned"
+ " from transport\n", cmd->init_task_tag,
+ SE_CMD(cmd)->t_state);
+ return -1;
+ }
+
+ dr = iscsit_allocate_datain_req();
+ if (!dr)
+ return -1;
+ /*
+ * The TMR TASK_REASSIGN's ExpDataSN contains the next DataSN the
+ * Initiator is expecting.
+ */
+ dr->data_sn = dr->begrun = tmr_req->exp_data_sn;
+ dr->runlength = 0;
+ dr->generate_recovery_values = 1;
+ dr->recovery = DATAIN_CONNECTION_RECOVERY;
+
+ iscsit_attach_datain_req(cmd, dr);
+
+ cmd->i_state = ISTATE_SEND_DATAIN;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ return 0;
+}
+
+static int iscsit_task_reassign_complete_none(
+ struct iscsi_cmd *cmd,
+ struct iscsi_tmr_req *tmr_req)
+{
+ struct iscsi_conn *conn = cmd->conn;
+
+ cmd->i_state = ISTATE_SEND_STATUS;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ return 0;
+}
+
+static int iscsit_task_reassign_complete_scsi_cmnd(
+ struct iscsi_tmr_req *tmr_req,
+ struct iscsi_conn *conn)
+{
+ struct se_tmr_req *se_tmr = tmr_req->se_tmr_req;
+ struct se_cmd *se_cmd = se_tmr->ref_cmd;
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+ struct iscsi_conn_recovery *cr;
+
+ if (!cmd->cr) {
+ printk(KERN_ERR "struct iscsi_conn_recovery pointer for ITT: 0x%08x"
+ " is NULL!\n", cmd->init_task_tag);
+ return -1;
+ }
+ cr = cmd->cr;
+
+ /*
+ * Reset the StatSN so a new one for this commands new connection
+ * will be assigned.
+ * Reset the ExpStatSN as well so we may receive Status SNACKs.
+ */
+ cmd->stat_sn = cmd->exp_stat_sn = 0;
+
+ iscsit_task_reassign_remove_cmd(cmd, cr, conn->sess);
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (se_cmd->se_cmd_flags & SCF_SENT_CHECK_CONDITION) {
+ cmd->i_state = ISTATE_SEND_STATUS;
+ iscsit_add_cmd_to_response_queue(cmd, conn, cmd->i_state);
+ return 0;
+ }
+
+ switch (cmd->data_direction) {
+ case DMA_TO_DEVICE:
+ return iscsit_task_reassign_complete_write(cmd, tmr_req);
+ case DMA_FROM_DEVICE:
+ return iscsit_task_reassign_complete_read(cmd, tmr_req);
+ case DMA_NONE:
+ return iscsit_task_reassign_complete_none(cmd, tmr_req);
+ default:
+ printk(KERN_ERR "Unknown cmd->data_direction: 0x%02x\n",
+ cmd->data_direction);
+ return -1;
+ }
+
+ return 0;
+}
+
+static int iscsit_task_reassign_complete(
+ struct iscsi_tmr_req *tmr_req,
+ struct iscsi_conn *conn)
+{
+ struct se_tmr_req *se_tmr = tmr_req->se_tmr_req;
+ struct se_cmd *se_cmd;
+ struct iscsi_cmd *cmd;
+ int ret = 0;
+
+ if (!se_tmr->ref_cmd) {
+ printk(KERN_ERR "TMR Request is missing a RefCmd struct iscsi_cmd.\n");
+ return -1;
+ }
+ se_cmd = se_tmr->ref_cmd;
+ cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ cmd->conn = conn;
+
+ switch (cmd->iscsi_opcode) {
+ case ISCSI_OP_NOOP_OUT:
+ ret = iscsit_task_reassign_complete_nop_out(tmr_req, conn);
+ break;
+ case ISCSI_OP_SCSI_CMD:
+ ret = iscsit_task_reassign_complete_scsi_cmnd(tmr_req, conn);
+ break;
+ default:
+ printk(KERN_ERR "Illegal iSCSI Opcode 0x%02x during"
+ " command realligence\n", cmd->iscsi_opcode);
+ return -1;
+ }
+
+ if (ret != 0)
+ return ret;
+
+ TRACE(TRACE_ERL2, "Completed connection realligence for Opcode: 0x%02x,"
+ " ITT: 0x%08x to CID: %hu.\n", cmd->iscsi_opcode,
+ cmd->init_task_tag, conn->cid);
+
+ return 0;
+}
+
+/*
+ * Handles special after-the-fact actions related to TMRs.
+ * Right now the only one that its really needed for is
+ * connection recovery releated TASK_REASSIGN.
+ */
+extern int iscsit_tmr_post_handler(struct iscsi_cmd *cmd, struct iscsi_conn *conn)
+{
+ struct iscsi_tmr_req *tmr_req = cmd->tmr_req;
+ struct se_tmr_req *se_tmr = SE_CMD(cmd)->se_tmr_req;
+
+ if (tmr_req->task_reassign &&
+ (se_tmr->response == ISCSI_TMF_RSP_COMPLETE))
+ return iscsit_task_reassign_complete(tmr_req, conn);
+
+ return 0;
+}
+
+/*
+ * Nothing to do here, but leave it for good measure. :-)
+ */
+int iscsit_task_reassign_prepare_read(
+ struct iscsi_tmr_req *tmr_req,
+ struct iscsi_conn *conn)
+{
+ return 0;
+}
+
+static void iscsit_task_reassign_prepare_unsolicited_dataout(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ int i, j;
+ struct iscsi_pdu *pdu = NULL;
+ struct iscsi_seq *seq = NULL;
+
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ cmd->data_sn = 0;
+
+ if (cmd->immediate_data)
+ cmd->r2t_offset += (cmd->first_burst_len -
+ cmd->seq_start_offset);
+
+ if (conn->sess->sess_ops->DataPDUInOrder) {
+ cmd->write_data_done -= (cmd->immediate_data) ?
+ (cmd->first_burst_len -
+ cmd->seq_start_offset) :
+ cmd->first_burst_len;
+ cmd->first_burst_len = 0;
+ return;
+ }
+
+ for (i = 0; i < cmd->pdu_count; i++) {
+ pdu = &cmd->pdu_list[i];
+
+ if (pdu->status != ISCSI_PDU_RECEIVED_OK)
+ continue;
+
+ if ((pdu->offset >= cmd->seq_start_offset) &&
+ ((pdu->offset + pdu->length) <=
+ cmd->seq_end_offset)) {
+ cmd->first_burst_len -= pdu->length;
+ cmd->write_data_done -= pdu->length;
+ pdu->status = ISCSI_PDU_NOT_RECEIVED;
+ }
+ }
+ } else {
+ for (i = 0; i < cmd->seq_count; i++) {
+ seq = &cmd->seq_list[i];
+
+ if (seq->type != SEQTYPE_UNSOLICITED)
+ continue;
+
+ cmd->write_data_done -=
+ (seq->offset - seq->orig_offset);
+ cmd->first_burst_len = 0;
+ seq->data_sn = 0;
+ seq->offset = seq->orig_offset;
+ seq->next_burst_len = 0;
+ seq->status = DATAOUT_SEQUENCE_WITHIN_COMMAND_RECOVERY;
+
+ if (conn->sess->sess_ops->DataPDUInOrder)
+ continue;
+
+ for (j = 0; j < seq->pdu_count; j++) {
+ pdu = &cmd->pdu_list[j+seq->pdu_start];
+
+ if (pdu->status != ISCSI_PDU_RECEIVED_OK)
+ continue;
+
+ pdu->status = ISCSI_PDU_NOT_RECEIVED;
+ }
+ }
+ }
+
+ return;
+}
+
+int iscsit_task_reassign_prepare_write(
+ struct iscsi_tmr_req *tmr_req,
+ struct iscsi_conn *conn)
+{
+ struct se_tmr_req *se_tmr = tmr_req->se_tmr_req;
+ struct se_cmd *se_cmd = se_tmr->ref_cmd;
+ struct iscsi_cmd *cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+ struct iscsi_pdu *pdu = NULL;
+ struct iscsi_r2t *r2t = NULL, *r2t_tmp;
+ int first_incomplete_r2t = 1, i = 0;
+
+ /*
+ * The command was in the process of receiving Unsolicited DataOUT when
+ * the connection failed.
+ */
+ if (cmd->unsolicited_data)
+ iscsit_task_reassign_prepare_unsolicited_dataout(cmd, conn);
+
+ /*
+ * The Initiator is requesting R2Ts starting from zero, skip
+ * checking acknowledged R2Ts and start checking struct iscsi_r2ts
+ * greater than zero.
+ */
+ if (!tmr_req->exp_data_sn)
+ goto drop_unacknowledged_r2ts;
+
+ /*
+ * We now check that the PDUs in DataOUT sequences below
+ * the TMR TASK_REASSIGN ExpDataSN (R2TSN the Initiator is
+ * expecting next) have all the DataOUT they require to complete
+ * the DataOUT sequence. First scan from R2TSN 0 to TMR
+ * TASK_REASSIGN ExpDataSN-1.
+ *
+ * If we have not received all DataOUT in question, we must
+ * make sure to make the appropriate changes to values in
+ * struct iscsi_cmd (and elsewhere depending on session parameters)
+ * so iscsit_build_r2ts_for_cmd() in iscsit_task_reassign_complete_write()
+ * will resend a new R2T for the DataOUT sequences in question.
+ */
+ spin_lock_bh(&cmd->r2t_lock);
+ if (list_empty(&cmd->cmd_r2t_list)) {
+ spin_unlock_bh(&cmd->r2t_lock);
+ return -1;
+ }
+
+ list_for_each_entry(r2t, &cmd->cmd_r2t_list, r2t_list) {
+
+ if (r2t->r2t_sn >= tmr_req->exp_data_sn)
+ continue;
+ /*
+ * Safely ignore Recovery R2Ts and R2Ts that have completed
+ * DataOUT sequences.
+ */
+ if (r2t->seq_complete)
+ continue;
+
+ if (r2t->recovery_r2t)
+ continue;
+
+ /*
+ * DataSequenceInOrder=Yes:
+ *
+ * Taking into account the iSCSI implementation requirement of
+ * MaxOutstandingR2T=1 while ErrorRecoveryLevel>0 and
+ * DataSequenceInOrder=Yes, we must take into consideration
+ * the following:
+ *
+ * DataSequenceInOrder=No:
+ *
+ * Taking into account that the Initiator controls the (possibly
+ * random) PDU Order in (possibly random) Sequence Order of
+ * DataOUT the target requests with R2Ts, we must take into
+ * consideration the following:
+ *
+ * DataPDUInOrder=Yes for DataSequenceInOrder=[Yes,No]:
+ *
+ * While processing non-complete R2T DataOUT sequence requests
+ * the Target will re-request only the total sequence length
+ * minus current received offset. This is because we must
+ * assume the initiator will continue sending DataOUT from the
+ * last PDU before the connection failed.
+ *
+ * DataPDUInOrder=No for DataSequenceInOrder=[Yes,No]:
+ *
+ * While processing non-complete R2T DataOUT sequence requests
+ * the Target will re-request the entire DataOUT sequence if
+ * any single PDU is missing from the sequence. This is because
+ * we have no logical method to determine the next PDU offset,
+ * and we must assume the Initiator will be sending any random
+ * PDU offset in the current sequence after TASK_REASSIGN
+ * has completed.
+ */
+ if (conn->sess->sess_ops->DataSequenceInOrder) {
+ if (!first_incomplete_r2t) {
+ cmd->r2t_offset -= r2t->xfer_len;
+ goto next;
+ }
+
+ if (conn->sess->sess_ops->DataPDUInOrder) {
+ cmd->data_sn = 0;
+ cmd->r2t_offset -= (r2t->xfer_len -
+ cmd->next_burst_len);
+ first_incomplete_r2t = 0;
+ goto next;
+ }
+
+ cmd->data_sn = 0;
+ cmd->r2t_offset -= r2t->xfer_len;
+
+ for (i = 0; i < cmd->pdu_count; i++) {
+ pdu = &cmd->pdu_list[i];
+
+ if (pdu->status != ISCSI_PDU_RECEIVED_OK)
+ continue;
+
+ if ((pdu->offset >= r2t->offset) &&
+ (pdu->offset < (r2t->offset +
+ r2t->xfer_len))) {
+ cmd->next_burst_len -= pdu->length;
+ cmd->write_data_done -= pdu->length;
+ pdu->status = ISCSI_PDU_NOT_RECEIVED;
+ }
+ }
+
+ first_incomplete_r2t = 0;
+ } else {
+ struct iscsi_seq *seq;
+
+ seq = iscsit_get_seq_holder(cmd, r2t->offset,
+ r2t->xfer_len);
+ if (!seq) {
+ spin_unlock_bh(&cmd->r2t_lock);
+ return -1;
+ }
+
+ cmd->write_data_done -=
+ (seq->offset - seq->orig_offset);
+ seq->data_sn = 0;
+ seq->offset = seq->orig_offset;
+ seq->next_burst_len = 0;
+ seq->status = DATAOUT_SEQUENCE_WITHIN_COMMAND_RECOVERY;
+
+ cmd->seq_send_order--;
+
+ if (conn->sess->sess_ops->DataPDUInOrder)
+ goto next;
+
+ for (i = 0; i < seq->pdu_count; i++) {
+ pdu = &cmd->pdu_list[i+seq->pdu_start];
+
+ if (pdu->status != ISCSI_PDU_RECEIVED_OK)
+ continue;
+
+ pdu->status = ISCSI_PDU_NOT_RECEIVED;
+ }
+ }
+
+next:
+ cmd->outstanding_r2ts--;
+ }
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ /*
+ * We now drop all unacknowledged R2Ts, ie: ExpDataSN from TMR
+ * TASK_REASSIGN to the last R2T in the list.. We are also careful
+ * to check that the Initiator is not requesting R2Ts for DataOUT
+ * sequences it has already completed.
+ *
+ * Free each R2T in question and adjust values in struct iscsi_cmd
+ * accordingly so iscsit_build_r2ts_for_cmd() do the rest of
+ * the work after the TMR TASK_REASSIGN Response is sent.
+ */
+drop_unacknowledged_r2ts:
+
+ cmd->cmd_flags &= ~ICF_SENT_LAST_R2T;
+ cmd->r2t_sn = tmr_req->exp_data_sn;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ list_for_each_entry_safe(r2t, r2t_tmp, &cmd->cmd_r2t_list, r2t_list) {
+ /*
+ * Skip up to the R2T Sequence number provided by the
+ * iSCSI TASK_REASSIGN TMR
+ */
+ if (r2t->r2t_sn < tmr_req->exp_data_sn)
+ continue;
+
+ if (r2t->seq_complete) {
+ printk(KERN_ERR "Initiator is requesting R2Ts from"
+ " R2TSN: 0x%08x, but R2TSN: 0x%08x, Offset: %u,"
+ " Length: %u is already complete."
+ " BAD INITIATOR ERL=2 IMPLEMENTATION!\n",
+ tmr_req->exp_data_sn, r2t->r2t_sn,
+ r2t->offset, r2t->xfer_len);
+ spin_unlock_bh(&cmd->r2t_lock);
+ return -1;
+ }
+
+ if (r2t->recovery_r2t) {
+ iscsit_free_r2t(r2t, cmd);
+ continue;
+ }
+
+ /* DataSequenceInOrder=Yes:
+ *
+ * Taking into account the iSCSI implementation requirement of
+ * MaxOutstandingR2T=1 while ErrorRecoveryLevel>0 and
+ * DataSequenceInOrder=Yes, it's safe to subtract the R2Ts
+ * entire transfer length from the commands R2T offset marker.
+ *
+ * DataSequenceInOrder=No:
+ *
+ * We subtract the difference from struct iscsi_seq between the
+ * current offset and original offset from cmd->write_data_done
+ * for account for DataOUT PDUs already received. Then reset
+ * the current offset to the original and zero out the current
+ * burst length, to make sure we re-request the entire DataOUT
+ * sequence.
+ */
+ if (conn->sess->sess_ops->DataSequenceInOrder)
+ cmd->r2t_offset -= r2t->xfer_len;
+ else
+ cmd->seq_send_order--;
+
+ cmd->outstanding_r2ts--;
+ iscsit_free_r2t(r2t, cmd);
+ }
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ return 0;
+}
+
+/*
+ * Performs sanity checks TMR TASK_REASSIGN's ExpDataSN for
+ * a given struct iscsi_cmd.
+ */
+int iscsit_check_task_reassign_expdatasn(
+ struct iscsi_tmr_req *tmr_req,
+ struct iscsi_conn *conn)
+{
+ struct se_tmr_req *se_tmr = tmr_req->se_tmr_req;
+ struct se_cmd *se_cmd = se_tmr->ref_cmd;
+ struct iscsi_cmd *ref_cmd = container_of(se_cmd, struct iscsi_cmd, se_cmd);
+
+ if (ref_cmd->iscsi_opcode != ISCSI_OP_SCSI_CMD)
+ return 0;
+
+ if (se_cmd->se_cmd_flags & SCF_SENT_CHECK_CONDITION)
+ return 0;
+
+ if (ref_cmd->data_direction == DMA_NONE)
+ return 0;
+
+ /*
+ * For READs the TMR TASK_REASSIGNs ExpDataSN contains the next DataSN
+ * of DataIN the Initiator is expecting.
+ *
+ * Also check that the Initiator is not re-requesting DataIN that has
+ * already been acknowledged with a DataAck SNACK.
+ */
+ if (ref_cmd->data_direction == DMA_FROM_DEVICE) {
+ if (tmr_req->exp_data_sn > ref_cmd->data_sn) {
+ printk(KERN_ERR "Received ExpDataSN: 0x%08x for READ"
+ " in TMR TASK_REASSIGN greater than command's"
+ " DataSN: 0x%08x.\n", tmr_req->exp_data_sn,
+ ref_cmd->data_sn);
+ return -1;
+ }
+ if ((ref_cmd->cmd_flags & ICF_GOT_DATACK_SNACK) &&
+ (tmr_req->exp_data_sn <= ref_cmd->acked_data_sn)) {
+ printk(KERN_ERR "Received ExpDataSN: 0x%08x for READ"
+ " in TMR TASK_REASSIGN for previously"
+ " acknowledged DataIN: 0x%08x,"
+ " protocol error\n", tmr_req->exp_data_sn,
+ ref_cmd->acked_data_sn);
+ return -1;
+ }
+ return iscsit_task_reassign_prepare_read(tmr_req, conn);
+ }
+
+ /*
+ * For WRITEs the TMR TASK_REASSIGNs ExpDataSN contains the next R2TSN
+ * for R2Ts the Initiator is expecting.
+ *
+ * Do the magic in iscsit_task_reassign_prepare_write().
+ */
+ if (ref_cmd->data_direction == DMA_TO_DEVICE) {
+ if (tmr_req->exp_data_sn > ref_cmd->r2t_sn) {
+ printk(KERN_ERR "Received ExpDataSN: 0x%08x for WRITE"
+ " in TMR TASK_REASSIGN greater than command's"
+ " R2TSN: 0x%08x.\n", tmr_req->exp_data_sn,
+ ref_cmd->r2t_sn);
+ return -1;
+ }
+ return iscsit_task_reassign_prepare_write(tmr_req, conn);
+ }
+
+ printk(KERN_ERR "Unknown iSCSI data_direction: 0x%02x\n",
+ ref_cmd->data_direction);
+
+ return -1;
+}
diff --git a/drivers/target/iscsi/iscsi_target_tmr.h b/drivers/target/iscsi/iscsi_target_tmr.h
new file mode 100644
index 0000000..142e992
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_tmr.h
@@ -0,0 +1,14 @@
+#ifndef ISCSI_TARGET_TMR_H
+#define ISCSI_TARGET_TMR_H
+
+extern u8 iscsit_tmr_abort_task(struct iscsi_cmd *, unsigned char *);
+extern int iscsit_tmr_task_warm_reset(struct iscsi_conn *, struct iscsi_tmr_req *,
+ unsigned char *);
+extern int iscsit_tmr_task_cold_reset(struct iscsi_conn *, struct iscsi_tmr_req *,
+ unsigned char *);
+extern u8 iscsit_tmr_task_reassign(struct iscsi_cmd *, unsigned char *);
+extern int iscsit_tmr_post_handler(struct iscsi_cmd *, struct iscsi_conn *);
+extern int iscsit_check_task_reassign_expdatasn(struct iscsi_tmr_req *,
+ struct iscsi_conn *);
+
+#endif /* ISCSI_TARGET_TMR_H */
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 13/14] iscsi-target: Add misc utility and debug logic
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (12 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 12/14] iscsi-target: Add support for task management operations Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 3:06 ` [PATCH-v2 14/14] iscsi-target: Add Makefile/Kconfig and update TCM top level Nicholas A. Bellinger
2011-03-23 8:54 ` [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge FUJITA Tomonori
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
This file adds iscsi_target_util.[c,h] code containing a number
of miscellaneous utility functions for iscsi_target_mod.
It also contains iscsi_debug.h macros for CONFIG_ISCSI_TARGET_DEBUG
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/iscsi/iscsi_target_debug.h | 113 ++
drivers/target/iscsi/iscsi_target_util.c | 1976 +++++++++++++++++++++++++++++
drivers/target/iscsi/iscsi_target_util.h | 105 ++
3 files changed, 2194 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/iscsi_target_debug.h
create mode 100644 drivers/target/iscsi/iscsi_target_util.c
create mode 100644 drivers/target/iscsi/iscsi_target_util.h
diff --git a/drivers/target/iscsi/iscsi_target_debug.h b/drivers/target/iscsi/iscsi_target_debug.h
new file mode 100644
index 0000000..1356146
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_debug.h
@@ -0,0 +1,113 @@
+#ifndef ISCSI_DEBUG_H
+#define ISCSI_DEBUG_H
+
+/*
+ * Debugging Support
+ */
+
+#define TRACE_DEBUG 0x00000001 /* Verbose debugging */
+#define TRACE_SCSI 0x00000002 /* Stuff related to SCSI Mid-layer */
+#define TRACE_ISCSI 0x00000004 /* Stuff related to iSCSI */
+#define TRACE_NET 0x00000008 /* Stuff related to network code */
+#define TRACE_BUFF 0x00000010 /* For dumping raw data */
+#define TRACE_FILE 0x00000020 /* Used for __FILE__ */
+#define TRACE_LINE 0x00000040 /* Used for __LINE__ */
+#define TRACE_FUNCTION 0x00000080 /* Used for __FUNCTION__ */
+#define TRACE_SEM 0x00000100 /* Stuff related to semaphores */
+#define TRACE_ENTER_LEAVE 0x00000200 /* For entering/leaving functions */
+#define TRACE_DIGEST 0x00000400 /* For Header/Data Digests */
+#define TRACE_PARAM 0x00000800 /* For parameters in parameters.c */
+#define TRACE_LOGIN 0x00001000 /* For login related code */
+#define TRACE_STATE 0x00002000 /* For conn/sess/cleanup states */
+#define TRACE_ERL0 0x00004000 /* For ErrorRecoveryLevel=0 */
+#define TRACE_ERL1 0x00008000 /* For ErrorRecoveryLevel=1 */
+#define TRACE_ERL2 0x00010000 /* For ErrorRecoveryLevel=2 */
+#define TRACE_TIMER 0x00020000 /* For various ERL timers */
+#define TRACE_R2T 0x00040000 /* For R2T callers */
+#define TRACE_SPINDLE 0x00080000 /* For Spindle callers */
+#define TRACE_SSLR 0x00100000 /* For SyncNSteering RX */
+#define TRACE_SSLT 0x00200000 /* For SyncNSteering TX */
+#define TRACE_CHANNEL 0x00400000 /* For SCSI Channels */
+#define TRACE_CMDSN 0x00800000 /* For Out of Order CmdSN execution */
+#define TRACE_NODEATTRIB 0x01000000 /* For Initiator Nodes */
+
+#define TRACE_VANITY 0x80000000 /* For all Vanity Noise */
+#define TRACE_ALL 0xffffffff /* Turn on all flags */
+#define TRACE_ENDING 0x00000000 /* foo */
+
+#ifdef CONFIG_ISCSI_TARGET_DEBUG
+/*
+ * TRACE_VANITY, is always last!
+ */
+static unsigned int iscsi_trace =
+/* TRACE_DEBUG | */
+/* TRACE_SCSI | */
+/* TRACE_ISCSI | */
+/* TRACE_NET | */
+/* TRACE_BUFF | */
+/* TRACE_FILE | */
+/* TRACE_LINE | */
+/* TRACE_FUNCTION | */
+/* TRACE_SEM | */
+
+/* TRACE_DIGEST | */
+/* TRACE_PARAM | */
+/* TRACE_LOGIN | */
+/* TRACE_STATE | */
+ TRACE_ERL0 |
+ TRACE_ERL1 |
+ TRACE_ERL2 |
+/* TRACE_TIMER | */
+/* TRACE_R2T | */
+/* TRACE_SPINDLE | */
+/* TRACE_SSLR | */
+/* TRACE_SSLT | */
+/* TRACE_CHANNEL | */
+/* TRACE_CMDSN | */
+/* TRACE_NODEATTRIB | */
+ TRACE_VANITY |
+ TRACE_ENDING;
+
+#define TRACE(trace, args...) \
+{ \
+static char iscsi_trace_buff[256]; \
+ \
+if (iscsi_trace & trace) { \
+ sprintf(iscsi_trace_buff, args); \
+ if (iscsi_trace & TRACE_FUNCTION) { \
+ printk(KERN_INFO "%s:%d: %s", __func__, __LINE__, \
+ iscsi_trace_buff); \
+ } else if (iscsi_trace&TRACE_FILE) { \
+ printk(KERN_INFO "%s::%d: %s", __FILE__, __LINE__, \
+ iscsi_trace_buff); \
+ } else if (iscsi_trace & TRACE_LINE) { \
+ printk(KERN_INFO "%d: %s", __LINE__, iscsi_trace_buff); \
+ } else { \
+ printk(KERN_INFO "%s", iscsi_trace_buff); \
+ } \
+} \
+}
+
+#define PRINT_BUFF(buff, len) \
+if (iscsi_trace & TRACE_BUFF) { \
+ int zzz; \
+ \
+ printk(KERN_INFO "%d:\n", __LINE__); \
+ for (zzz = 0; zzz < len; zzz++) { \
+ if (zzz % 16 == 0) { \
+ if (zzz) \
+ printk(KERN_INFO "\n"); \
+ printk(KERN_INFO "%4i: ", zzz); \
+ } \
+ printk(KERN_INFO "%02x ", (unsigned char) (buff)[zzz]); \
+ } \
+ if ((len + 1) % 16) \
+ printk(KERN_INFO "\n"); \
+}
+
+#else /* !CONFIG_ISCSI_TARGET_DEBUG */
+#define TRACE(trace, args...)
+#define PRINT_BUFF(buff, len)
+#endif /* CONFIG_ISCSI_TARGET_DEBUG */
+
+#endif /*** ISCSI_DEBUG_H ***/
diff --git a/drivers/target/iscsi/iscsi_target_util.c b/drivers/target/iscsi/iscsi_target_util.c
new file mode 100644
index 0000000..05338a7
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_util.c
@@ -0,0 +1,1976 @@
+/*******************************************************************************
+ * This file contains the iSCSI Target specific utility functions.
+ *
+ * © Copyright 2007-2011 RisingTide Systems LLC.
+ *
+ * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
+ *
+ * Author: Nicholas A. Bellinger <nab@linux-iscsi.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ ******************************************************************************/
+
+#include <linux/list.h>
+#include <scsi/libsas.h> /* For TASK_ATTR_* */
+#include <scsi/iscsi_proto.h>
+#include <target/target_core_base.h>
+#include <target/target_core_transport.h>
+#include <target/target_core_tmr.h>
+#include <target/target_core_fabric_ops.h>
+#include <target/target_core_configfs.h>
+
+#include "iscsi_target_debug.h"
+#include "iscsi_target_core.h"
+#include "iscsi_target_parameters.h"
+#include "iscsi_target_seq_pdu_list.h"
+#include "iscsi_target_datain_values.h"
+#include "iscsi_target_erl0.h"
+#include "iscsi_target_erl1.h"
+#include "iscsi_target_erl2.h"
+#include "iscsi_target_tpg.h"
+#include "iscsi_target_tq.h"
+#include "iscsi_target_util.h"
+#include "iscsi_target.h"
+
+extern struct list_head g_tiqn_list;
+extern spinlock_t tiqn_lock;
+
+/*
+ * Called with cmd->r2t_lock held.
+ */
+int iscsit_add_r2t_to_list(
+ struct iscsi_cmd *cmd,
+ u32 offset,
+ u32 xfer_len,
+ int recovery,
+ u32 r2t_sn)
+{
+ struct iscsi_r2t *r2t;
+
+ r2t = kmem_cache_zalloc(lio_r2t_cache, GFP_ATOMIC);
+ if (!r2t) {
+ printk(KERN_ERR "Unable to allocate memory for struct iscsi_r2t.\n");
+ return -1;
+ }
+ INIT_LIST_HEAD(&r2t->r2t_list);
+
+ r2t->recovery_r2t = recovery;
+ r2t->r2t_sn = (!r2t_sn) ? cmd->r2t_sn++ : r2t_sn;
+ r2t->offset = offset;
+ r2t->xfer_len = xfer_len;
+ list_add_tail(&r2t->r2t_list, &cmd->cmd_r2t_list);
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ iscsit_add_cmd_to_immediate_queue(cmd, cmd->conn, ISTATE_SEND_R2T);
+
+ spin_lock_bh(&cmd->r2t_lock);
+ return 0;
+}
+
+struct iscsi_r2t *iscsit_get_r2t_for_eos(
+ struct iscsi_cmd *cmd,
+ u32 offset,
+ u32 length)
+{
+ struct iscsi_r2t *r2t;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ list_for_each_entry(r2t, &cmd->cmd_r2t_list, r2t_list) {
+ if ((r2t->offset <= offset) &&
+ (r2t->offset + r2t->xfer_len) >= (offset + length))
+ break;
+ }
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ if (!r2t) {
+ printk(KERN_ERR "Unable to locate R2T for Offset: %u, Length:"
+ " %u\n", offset, length);
+ return NULL;
+ }
+
+ return r2t;
+}
+
+struct iscsi_r2t *iscsit_get_r2t_from_list(struct iscsi_cmd *cmd)
+{
+ struct iscsi_r2t *r2t;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ list_for_each_entry(r2t, &cmd->cmd_r2t_list, r2t_list) {
+ if (!r2t->sent_r2t)
+ break;
+ }
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ if (!r2t) {
+ printk(KERN_ERR "Unable to locate next R2T to send for ITT:"
+ " 0x%08x.\n", cmd->init_task_tag);
+ return NULL;
+ }
+
+ return r2t;
+}
+
+/*
+ * Called with cmd->r2t_lock held.
+ */
+void iscsit_free_r2t(struct iscsi_r2t *r2t, struct iscsi_cmd *cmd)
+{
+ list_del(&r2t->r2t_list);
+ kmem_cache_free(lio_r2t_cache, r2t);
+}
+
+void iscsit_free_r2ts_from_list(struct iscsi_cmd *cmd)
+{
+ struct iscsi_r2t *r2t, *r2t_tmp;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ list_for_each_entry_safe(r2t, r2t_tmp, &cmd->cmd_r2t_list, r2t_list) {
+ list_del(&r2t->r2t_list);
+ kmem_cache_free(lio_r2t_cache, r2t);
+ }
+ spin_unlock_bh(&cmd->r2t_lock);
+}
+
+/*
+ * May be called from software interrupt (timer) context for allocating
+ * iSCSI NopINs.
+ */
+struct iscsi_cmd *iscsit_allocate_cmd(struct iscsi_conn *conn)
+{
+ struct iscsi_cmd *cmd;
+
+ cmd = kmem_cache_zalloc(lio_cmd_cache, GFP_ATOMIC);
+ if (!cmd) {
+ printk(KERN_ERR "Unable to allocate memory for struct iscsi_cmd.\n");
+ return NULL;
+ }
+
+ cmd->conn = conn;
+ INIT_LIST_HEAD(&cmd->i_list);
+ INIT_LIST_HEAD(&cmd->datain_list);
+ INIT_LIST_HEAD(&cmd->cmd_r2t_list);
+ init_completion(&cmd->reject_comp);
+ init_completion(&cmd->unsolicited_data_comp);
+ spin_lock_init(&cmd->datain_lock);
+ spin_lock_init(&cmd->dataout_timeout_lock);
+ spin_lock_init(&cmd->istate_lock);
+ spin_lock_init(&cmd->error_lock);
+ spin_lock_init(&cmd->r2t_lock);
+
+ return cmd;
+}
+
+/*
+ * Called from iscsi_handle_scsi_cmd()
+ */
+struct iscsi_cmd *iscsit_allocate_se_cmd(
+ struct iscsi_conn *conn,
+ u32 data_length,
+ int data_direction,
+ int iscsi_task_attr)
+{
+ struct iscsi_cmd *cmd;
+ struct se_cmd *se_cmd;
+ int sam_task_attr;
+
+ cmd = iscsit_allocate_cmd(conn);
+ if (!cmd)
+ return NULL;
+
+ cmd->data_direction = data_direction;
+ cmd->data_length = data_length;
+ /*
+ * Figure out the SAM Task Attribute for the incoming SCSI CDB
+ */
+ if ((iscsi_task_attr == ISCSI_ATTR_UNTAGGED) ||
+ (iscsi_task_attr == ISCSI_ATTR_SIMPLE))
+ sam_task_attr = TASK_ATTR_SIMPLE;
+ else if (iscsi_task_attr == ISCSI_ATTR_ORDERED)
+ sam_task_attr = TASK_ATTR_ORDERED;
+ else if (iscsi_task_attr == ISCSI_ATTR_HEAD_OF_QUEUE)
+ sam_task_attr = TASK_ATTR_HOQ;
+ else if (iscsi_task_attr == ISCSI_ATTR_ACA)
+ sam_task_attr = TASK_ATTR_ACA;
+ else {
+ printk(KERN_INFO "Unknown iSCSI Task Attribute: 0x%02x, using"
+ " TASK_ATTR_SIMPLE\n", iscsi_task_attr);
+ sam_task_attr = TASK_ATTR_SIMPLE;
+ }
+
+ se_cmd = &cmd->se_cmd;
+ /*
+ * Initialize struct se_cmd descriptor from target_core_mod infrastructure
+ */
+ transport_init_se_cmd(se_cmd, &lio_target_fabric_configfs->tf_ops,
+ conn->sess->se_sess, data_length, data_direction,
+ sam_task_attr, &cmd->sense_buffer[0]);
+ return cmd;
+}
+
+struct iscsi_cmd *iscsit_allocate_se_cmd_for_tmr(
+ struct iscsi_conn *conn,
+ u8 function)
+{
+ struct iscsi_cmd *cmd;
+ struct se_cmd *se_cmd;
+ u8 tcm_function;
+
+ cmd = iscsit_allocate_cmd(conn);
+ if (!cmd)
+ return NULL;
+
+ cmd->data_direction = DMA_NONE;
+
+ cmd->tmr_req = kzalloc(sizeof(struct iscsi_tmr_req), GFP_KERNEL);
+ if (!cmd->tmr_req) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " Task Management command!\n");
+ return NULL;
+ }
+ /*
+ * TASK_REASSIGN for ERL=2 / connection stays inside of
+ * LIO-Target $FABRIC_MOD
+ */
+ if (function == ISCSI_TM_FUNC_TASK_REASSIGN)
+ return cmd;
+
+ se_cmd = &cmd->se_cmd;
+ /*
+ * Initialize struct se_cmd descriptor from target_core_mod infrastructure
+ */
+ transport_init_se_cmd(se_cmd, &lio_target_fabric_configfs->tf_ops,
+ conn->sess->se_sess, 0, DMA_NONE,
+ TASK_ATTR_SIMPLE, &cmd->sense_buffer[0]);
+
+ switch (function) {
+ case ISCSI_TM_FUNC_ABORT_TASK:
+ tcm_function = TMR_ABORT_TASK;
+ break;
+ case ISCSI_TM_FUNC_ABORT_TASK_SET:
+ tcm_function = TMR_ABORT_TASK_SET;
+ break;
+ case ISCSI_TM_FUNC_CLEAR_ACA:
+ tcm_function = TMR_CLEAR_ACA;
+ break;
+ case ISCSI_TM_FUNC_CLEAR_TASK_SET:
+ tcm_function = TMR_CLEAR_TASK_SET;
+ break;
+ case ISCSI_TM_FUNC_LOGICAL_UNIT_RESET:
+ tcm_function = TMR_LUN_RESET;
+ break;
+ case ISCSI_TM_FUNC_TARGET_WARM_RESET:
+ tcm_function = TMR_TARGET_WARM_RESET;
+ break;
+ case ISCSI_TM_FUNC_TARGET_COLD_RESET:
+ tcm_function = TMR_TARGET_COLD_RESET;
+ break;
+ default:
+ printk(KERN_ERR "Unknown iSCSI TMR Function:"
+ " 0x%02x\n", function);
+ goto out;
+ }
+
+ se_cmd->se_tmr_req = core_tmr_alloc_req(se_cmd,
+ (void *)cmd->tmr_req, tcm_function);
+ if (!se_cmd->se_tmr_req)
+ goto out;
+
+ cmd->tmr_req->se_tmr_req = se_cmd->se_tmr_req;
+
+ return cmd;
+out:
+ iscsit_release_cmd(cmd);
+ if (se_cmd)
+ transport_free_se_cmd(se_cmd);
+ return NULL;
+}
+
+int iscsit_decide_list_to_build(
+ struct iscsi_cmd *cmd,
+ u32 immediate_data_length)
+{
+ struct iscsi_build_list bl;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na;
+
+ if (sess->sess_ops->DataSequenceInOrder &&
+ sess->sess_ops->DataPDUInOrder)
+ return 0;
+
+ if (cmd->data_direction == DMA_NONE)
+ return 0;
+
+ na = iscsit_tpg_get_node_attrib(sess);
+ memset(&bl, 0, sizeof(struct iscsi_build_list));
+
+ if (cmd->data_direction == DMA_FROM_DEVICE) {
+ bl.data_direction = ISCSI_PDU_READ;
+ bl.type = PDULIST_NORMAL;
+ if (na->random_datain_pdu_offsets)
+ bl.randomize |= RANDOM_DATAIN_PDU_OFFSETS;
+ if (na->random_datain_seq_offsets)
+ bl.randomize |= RANDOM_DATAIN_SEQ_OFFSETS;
+ } else {
+ bl.data_direction = ISCSI_PDU_WRITE;
+ bl.immediate_data_length = immediate_data_length;
+ if (na->random_r2t_offsets)
+ bl.randomize |= RANDOM_R2T_OFFSETS;
+
+ if (!cmd->immediate_data && !cmd->unsolicited_data)
+ bl.type = PDULIST_NORMAL;
+ else if (cmd->immediate_data && !cmd->unsolicited_data)
+ bl.type = PDULIST_IMMEDIATE;
+ else if (!cmd->immediate_data && cmd->unsolicited_data)
+ bl.type = PDULIST_UNSOLICITED;
+ else if (cmd->immediate_data && cmd->unsolicited_data)
+ bl.type = PDULIST_IMMEDIATE_AND_UNSOLICITED;
+ }
+
+ return iscsit_do_build_list(cmd, &bl);
+}
+
+struct iscsi_seq *iscsit_get_seq_holder_for_datain(
+ struct iscsi_cmd *cmd,
+ u32 seq_send_order)
+{
+ u32 i;
+
+ for (i = 0; i < cmd->seq_count; i++)
+ if (cmd->seq_list[i].seq_send_order == seq_send_order)
+ return &cmd->seq_list[i];
+
+ return NULL;
+}
+
+struct iscsi_seq *iscsit_get_seq_holder_for_r2t(struct iscsi_cmd *cmd)
+{
+ u32 i;
+
+ if (!cmd->seq_list) {
+ printk(KERN_ERR "struct iscsi_cmd->seq_list is NULL!\n");
+ return NULL;
+ }
+
+ for (i = 0; i < cmd->seq_count; i++) {
+ if (cmd->seq_list[i].type != SEQTYPE_NORMAL)
+ continue;
+ if (cmd->seq_list[i].seq_send_order == cmd->seq_send_order) {
+ cmd->seq_send_order++;
+ return &cmd->seq_list[i];
+ }
+ }
+
+ return NULL;
+}
+
+struct iscsi_r2t *iscsit_get_holder_for_r2tsn(
+ struct iscsi_cmd *cmd,
+ u32 r2t_sn)
+{
+ struct iscsi_r2t *r2t;
+
+ spin_lock_bh(&cmd->r2t_lock);
+ list_for_each_entry(r2t, &cmd->cmd_r2t_list, r2t_list) {
+ if (r2t->r2t_sn == r2t_sn) {
+ spin_unlock_bh(&cmd->r2t_lock);
+ return r2t;
+ }
+ }
+ spin_unlock_bh(&cmd->r2t_lock);
+
+ return NULL;
+}
+
+inline int iscsit_check_received_cmdsn(
+ struct iscsi_conn *conn,
+ struct iscsi_cmd *cmd,
+ u32 cmdsn)
+{
+ int ret;
+ /*
+ * This is the proper method of checking received CmdSN against
+ * ExpCmdSN and MaxCmdSN values, as well as accounting for out
+ * or order CmdSNs due to multiple connection sessions and/or
+ * CRC failures.
+ */
+ spin_lock(&conn->sess->cmdsn_lock);
+ if (iscsi_sna_gt(cmdsn, conn->sess->max_cmd_sn)) {
+ printk(KERN_ERR "Received CmdSN: 0x%08x is greater than"
+ " MaxCmdSN: 0x%08x, protocol error.\n", cmdsn,
+ conn->sess->max_cmd_sn);
+ spin_unlock(&conn->sess->cmdsn_lock);
+ return CMDSN_ERROR_CANNOT_RECOVER;
+ }
+
+ if (!conn->sess->cmdsn_outoforder) {
+ if (cmdsn == conn->sess->exp_cmd_sn) {
+ conn->sess->exp_cmd_sn++;
+ TRACE(TRACE_CMDSN, "Received CmdSN matches ExpCmdSN,"
+ " incremented ExpCmdSN to: 0x%08x\n",
+ conn->sess->exp_cmd_sn);
+ ret = iscsit_execute_cmd(cmd, 0);
+ spin_unlock(&conn->sess->cmdsn_lock);
+
+ return (!ret) ? CMDSN_NORMAL_OPERATION :
+ CMDSN_ERROR_CANNOT_RECOVER;
+ } else if (iscsi_sna_gt(cmdsn, conn->sess->exp_cmd_sn)) {
+ TRACE(TRACE_CMDSN, "Received CmdSN: 0x%08x is greater"
+ " than ExpCmdSN: 0x%08x, not acknowledging.\n",
+ cmdsn, conn->sess->exp_cmd_sn);
+ goto ooo_cmdsn;
+ } else {
+ printk(KERN_ERR "Received CmdSN: 0x%08x is less than"
+ " ExpCmdSN: 0x%08x, ignoring.\n", cmdsn,
+ conn->sess->exp_cmd_sn);
+ spin_unlock(&conn->sess->cmdsn_lock);
+ return CMDSN_LOWER_THAN_EXP;
+ }
+ } else {
+ int counter = 0;
+ u32 old_expcmdsn = 0;
+ if (cmdsn == conn->sess->exp_cmd_sn) {
+ old_expcmdsn = conn->sess->exp_cmd_sn++;
+ TRACE(TRACE_CMDSN, "Got missing CmdSN: 0x%08x matches"
+ " ExpCmdSN, incremented ExpCmdSN to 0x%08x.\n",
+ cmdsn, conn->sess->exp_cmd_sn);
+
+ if (iscsit_execute_cmd(cmd, 0) < 0) {
+ spin_unlock(&conn->sess->cmdsn_lock);
+ return CMDSN_ERROR_CANNOT_RECOVER;
+ }
+ } else if (iscsi_sna_gt(cmdsn, conn->sess->exp_cmd_sn)) {
+ TRACE(TRACE_CMDSN, "CmdSN: 0x%08x greater than"
+ " ExpCmdSN: 0x%08x, not acknowledging.\n",
+ cmdsn, conn->sess->exp_cmd_sn);
+ goto ooo_cmdsn;
+ } else {
+ printk(KERN_ERR "CmdSN: 0x%08x less than ExpCmdSN:"
+ " 0x%08x, ignoring.\n", cmdsn,
+ conn->sess->exp_cmd_sn);
+ spin_unlock(&conn->sess->cmdsn_lock);
+ return CMDSN_LOWER_THAN_EXP;
+ }
+
+ counter = iscsit_execute_ooo_cmdsns(conn->sess);
+ if (counter < 0) {
+ spin_unlock(&conn->sess->cmdsn_lock);
+ return CMDSN_ERROR_CANNOT_RECOVER;
+ }
+
+ if (counter == conn->sess->ooo_cmdsn_count) {
+ if (conn->sess->ooo_cmdsn_count == 1) {
+ TRACE(TRACE_CMDSN, "Received final missing"
+ " CmdSN: 0x%08x.\n", old_expcmdsn);
+ } else {
+ TRACE(TRACE_CMDSN, "Received final missing"
+ " CmdSNs: 0x%08x->0x%08x.\n",
+ old_expcmdsn, (conn->sess->exp_cmd_sn - 1));
+ }
+
+ conn->sess->ooo_cmdsn_count = 0;
+ conn->sess->cmdsn_outoforder = 0;
+ } else {
+ conn->sess->ooo_cmdsn_count -= counter;
+ TRACE(TRACE_CMDSN, "Still missing %hu CmdSN(s),"
+ " continuing out of order operation.\n",
+ conn->sess->ooo_cmdsn_count);
+ }
+ spin_unlock(&conn->sess->cmdsn_lock);
+ return CMDSN_NORMAL_OPERATION;
+ }
+
+ooo_cmdsn:
+ ret = iscsit_handle_ooo_cmdsn(conn->sess, cmd, cmdsn);
+ spin_unlock(&conn->sess->cmdsn_lock);
+ return ret;
+}
+
+int iscsit_check_unsolicited_dataout(struct iscsi_cmd *cmd, unsigned char *buf)
+{
+ struct iscsi_conn *conn = cmd->conn;
+ struct se_cmd *se_cmd = SE_CMD(cmd);
+ struct iscsi_data *hdr = (struct iscsi_data *) buf;
+ u32 payload_length = ntoh24(hdr->dlength);
+
+ if (conn->sess->sess_ops->InitialR2T) {
+ printk(KERN_ERR "Received unexpected unsolicited data"
+ " while InitialR2T=Yes, protocol error.\n");
+ transport_send_check_condition_and_sense(se_cmd,
+ TCM_UNEXPECTED_UNSOLICITED_DATA, 0);
+ return -1;
+ }
+
+ if ((cmd->first_burst_len + payload_length) >
+ conn->sess->sess_ops->FirstBurstLength) {
+ printk(KERN_ERR "Total %u bytes exceeds FirstBurstLength: %u"
+ " for this Unsolicited DataOut Burst.\n",
+ (cmd->first_burst_len + payload_length),
+ conn->sess->sess_ops->FirstBurstLength);
+ transport_send_check_condition_and_sense(se_cmd,
+ TCM_INCORRECT_AMOUNT_OF_DATA, 0);
+ return -1;
+ }
+
+ if (!(hdr->flags & ISCSI_FLAG_CMD_FINAL))
+ return 0;
+
+ if (((cmd->first_burst_len + payload_length) != cmd->data_length) &&
+ ((cmd->first_burst_len + payload_length) !=
+ conn->sess->sess_ops->FirstBurstLength)) {
+ printk(KERN_ERR "Unsolicited non-immediate data received %u"
+ " does not equal FirstBurstLength: %u, and does"
+ " not equal ExpXferLen %u.\n",
+ (cmd->first_burst_len + payload_length),
+ conn->sess->sess_ops->FirstBurstLength, cmd->data_length);
+ transport_send_check_condition_and_sense(se_cmd,
+ TCM_INCORRECT_AMOUNT_OF_DATA, 0);
+ return -1;
+ }
+ return 0;
+}
+
+struct iscsi_cmd *iscsit_find_cmd_from_itt(
+ struct iscsi_conn *conn,
+ u32 init_task_tag)
+{
+ struct iscsi_cmd *cmd;
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry(cmd, &conn->conn_cmd_list, i_list) {
+ if (cmd->init_task_tag == init_task_tag)
+ break;
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (!cmd) {
+ printk(KERN_ERR "Unable to locate ITT: 0x%08x on CID: %hu",
+ init_task_tag, conn->cid);
+ return NULL;
+ }
+
+ return cmd;
+}
+
+struct iscsi_cmd *iscsit_find_cmd_from_itt_or_dump(
+ struct iscsi_conn *conn,
+ u32 init_task_tag,
+ u32 length)
+{
+ struct iscsi_cmd *cmd;
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry(cmd, &conn->conn_cmd_list, i_list) {
+ if (cmd->init_task_tag == init_task_tag)
+ break;
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (!cmd) {
+ printk(KERN_ERR "Unable to locate ITT: 0x%08x on CID: %hu,"
+ " dumping payload\n", init_task_tag, conn->cid);
+ if (length)
+ iscsit_dump_data_payload(conn, length, 1);
+ return NULL;
+ }
+
+ return cmd;
+}
+
+struct iscsi_cmd *iscsit_find_cmd_from_ttt(
+ struct iscsi_conn *conn,
+ u32 targ_xfer_tag)
+{
+ struct iscsi_cmd *cmd = NULL;
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_for_each_entry(cmd, &conn->conn_cmd_list, i_list) {
+ if (cmd->targ_xfer_tag == targ_xfer_tag)
+ break;
+ }
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (!cmd) {
+ printk(KERN_ERR "Unable to locate TTT: 0x%08x on CID: %hu\n",
+ targ_xfer_tag, conn->cid);
+ return NULL;
+ }
+
+ return cmd;
+}
+
+int iscsit_find_cmd_for_recovery(
+ struct iscsi_session *sess,
+ struct iscsi_cmd **cmd_ptr,
+ struct iscsi_conn_recovery **cr_ptr,
+ u32 init_task_tag)
+{
+ int found_itt = 0;
+ struct iscsi_cmd *cmd = NULL;
+ struct iscsi_conn_recovery *cr;
+
+ /*
+ * Scan through the inactive connection recovery list's command list.
+ * If init_task_tag matches the command is still alligent.
+ */
+ spin_lock(&sess->cr_i_lock);
+ list_for_each_entry(cr, &sess->cr_inactive_list, cr_list) {
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ list_for_each_entry(cmd, &cr->conn_recovery_cmd_list, i_list) {
+ if (cmd->init_task_tag == init_task_tag) {
+ found_itt = 1;
+ break;
+ }
+ }
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+ if (found_itt)
+ break;
+ }
+ spin_unlock(&sess->cr_i_lock);
+
+ if (cmd) {
+ *cr_ptr = cr;
+ *cmd_ptr = cmd;
+ return -2;
+ }
+
+ found_itt = 0;
+
+ /*
+ * Scan through the active connection recovery list's command list.
+ * If init_task_tag matches the command is ready to be reassigned.
+ */
+ spin_lock(&sess->cr_a_lock);
+ list_for_each_entry(cr, &sess->cr_active_list, cr_list) {
+ spin_lock(&cr->conn_recovery_cmd_lock);
+ list_for_each_entry(cmd, &cr->conn_recovery_cmd_list, i_list) {
+ if (cmd->init_task_tag == init_task_tag) {
+ found_itt = 1;
+ break;
+ }
+ }
+ spin_unlock(&cr->conn_recovery_cmd_lock);
+ if (found_itt)
+ break;
+ }
+ spin_unlock(&sess->cr_a_lock);
+
+ if (!cmd || !cr)
+ return -1;
+
+ *cr_ptr = cr;
+ *cmd_ptr = cmd;
+
+ return 0;
+}
+
+void iscsit_add_cmd_to_immediate_queue(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn,
+ u8 state)
+{
+ struct iscsi_queue_req *qr;
+
+ qr = kmem_cache_zalloc(lio_qr_cache, GFP_ATOMIC);
+ if (!qr) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_queue_req\n");
+ return;
+ }
+ INIT_LIST_HEAD(&qr->qr_list);
+ qr->cmd = cmd;
+ qr->state = state;
+
+ spin_lock_bh(&conn->immed_queue_lock);
+ list_add_tail(&qr->qr_list, &conn->immed_queue_list);
+ atomic_inc(&cmd->immed_queue_count);
+ atomic_set(&conn->check_immediate_queue, 1);
+ spin_unlock_bh(&conn->immed_queue_lock);
+
+ wake_up_process(conn->thread_set->tx_thread);
+}
+
+struct iscsi_queue_req *iscsit_get_cmd_from_immediate_queue(struct iscsi_conn *conn)
+{
+ struct iscsi_queue_req *qr;
+
+ spin_lock_bh(&conn->immed_queue_lock);
+ if (list_empty(&conn->immed_queue_list)) {
+ spin_unlock_bh(&conn->immed_queue_lock);
+ return NULL;
+ }
+ list_for_each_entry(qr, &conn->immed_queue_list, qr_list)
+ break;
+
+ list_del(&qr->qr_list);
+ if (qr->cmd)
+ atomic_dec(&qr->cmd->immed_queue_count);
+ spin_unlock_bh(&conn->immed_queue_lock);
+
+ return qr;
+}
+
+static void iscsit_remove_cmd_from_immediate_queue(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ struct iscsi_queue_req *qr, *qr_tmp;
+
+ spin_lock_bh(&conn->immed_queue_lock);
+ if (!atomic_read(&cmd->immed_queue_count)) {
+ spin_unlock_bh(&conn->immed_queue_lock);
+ return;
+ }
+
+ list_for_each_entry_safe(qr, qr_tmp, &conn->immed_queue_list, qr_list) {
+ if (qr->cmd != cmd)
+ continue;
+
+ atomic_dec(&qr->cmd->immed_queue_count);
+ list_del(&qr->qr_list);
+ kmem_cache_free(lio_qr_cache, qr);
+ }
+ spin_unlock_bh(&conn->immed_queue_lock);
+
+ if (atomic_read(&cmd->immed_queue_count)) {
+ printk(KERN_ERR "ITT: 0x%08x immed_queue_count: %d\n",
+ cmd->init_task_tag,
+ atomic_read(&cmd->immed_queue_count));
+ }
+}
+
+void iscsit_add_cmd_to_response_queue(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn,
+ u8 state)
+{
+ struct iscsi_queue_req *qr;
+
+ qr = kmem_cache_zalloc(lio_qr_cache, GFP_ATOMIC);
+ if (!qr) {
+ printk(KERN_ERR "Unable to allocate memory for"
+ " struct iscsi_queue_req\n");
+ return;
+ }
+ INIT_LIST_HEAD(&qr->qr_list);
+ qr->cmd = cmd;
+ qr->state = state;
+
+ spin_lock_bh(&conn->response_queue_lock);
+ list_add_tail(&qr->qr_list, &conn->response_queue_list);
+ atomic_inc(&cmd->response_queue_count);
+ spin_unlock_bh(&conn->response_queue_lock);
+
+ wake_up_process(conn->thread_set->tx_thread);
+}
+
+struct iscsi_queue_req *iscsit_get_cmd_from_response_queue(struct iscsi_conn *conn)
+{
+ struct iscsi_queue_req *qr;
+
+ spin_lock_bh(&conn->response_queue_lock);
+ if (list_empty(&conn->response_queue_list)) {
+ spin_unlock_bh(&conn->response_queue_lock);
+ return NULL;
+ }
+
+ list_for_each_entry(qr, &conn->response_queue_list, qr_list)
+ break;
+
+ list_del(&qr->qr_list);
+ if (qr->cmd)
+ atomic_dec(&qr->cmd->response_queue_count);
+ spin_unlock_bh(&conn->response_queue_lock);
+
+ return qr;
+}
+
+static void iscsit_remove_cmd_from_response_queue(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn)
+{
+ struct iscsi_queue_req *qr, *qr_tmp;
+
+ spin_lock_bh(&conn->response_queue_lock);
+ if (!(atomic_read(&cmd->response_queue_count))) {
+ spin_unlock_bh(&conn->response_queue_lock);
+ return;
+ }
+
+ list_for_each_entry_safe(qr, qr_tmp, &conn->response_queue_list,
+ qr_list) {
+ if (qr->cmd != cmd)
+ continue;
+
+ atomic_dec(&qr->cmd->response_queue_count);
+ list_del(&qr->qr_list);
+ kmem_cache_free(lio_qr_cache, qr);
+ }
+ spin_unlock_bh(&conn->response_queue_lock);
+
+ if (atomic_read(&cmd->response_queue_count)) {
+ printk(KERN_ERR "ITT: 0x%08x response_queue_count: %d\n",
+ cmd->init_task_tag,
+ atomic_read(&cmd->response_queue_count));
+ }
+}
+
+void iscsit_free_queue_reqs_for_conn(struct iscsi_conn *conn)
+{
+ struct iscsi_queue_req *qr, *qr_tmp;
+
+ spin_lock_bh(&conn->immed_queue_lock);
+ list_for_each_entry_safe(qr, qr_tmp, &conn->immed_queue_list, qr_list) {
+ list_del(&qr->qr_list);
+ if (qr->cmd)
+ atomic_dec(&qr->cmd->immed_queue_count);
+
+ kmem_cache_free(lio_qr_cache, qr);
+ }
+ spin_unlock_bh(&conn->immed_queue_lock);
+
+ spin_lock_bh(&conn->response_queue_lock);
+ list_for_each_entry_safe(qr, qr_tmp, &conn->response_queue_list,
+ qr_list) {
+ list_del(&qr->qr_list);
+ if (qr->cmd)
+ atomic_dec(&qr->cmd->response_queue_count);
+
+ kmem_cache_free(lio_qr_cache, qr);
+ }
+ spin_unlock_bh(&conn->response_queue_lock);
+}
+
+void iscsit_release_cmd(struct iscsi_cmd *cmd)
+{
+ struct iscsi_conn *conn = cmd->conn;
+
+ iscsit_free_r2ts_from_list(cmd);
+ iscsit_free_all_datain_reqs(cmd);
+
+ kfree(cmd->buf_ptr);
+ kfree(cmd->pdu_list);
+ kfree(cmd->seq_list);
+ kfree(cmd->tmr_req);
+ kfree(cmd->iov_data);
+
+ if (conn) {
+ iscsit_remove_cmd_from_immediate_queue(cmd, conn);
+ iscsit_remove_cmd_from_response_queue(cmd, conn);
+ }
+
+ kmem_cache_free(lio_cmd_cache, cmd);
+}
+
+int iscsit_check_session_usage_count(struct iscsi_session *sess)
+{
+ spin_lock_bh(&sess->session_usage_lock);
+ if (sess->session_usage_count != 0) {
+ sess->session_waiting_on_uc = 1;
+ spin_unlock_bh(&sess->session_usage_lock);
+ if (in_interrupt())
+ return 2;
+
+ wait_for_completion(&sess->session_waiting_on_uc_comp);
+ return 1;
+ }
+ spin_unlock_bh(&sess->session_usage_lock);
+
+ return 0;
+}
+
+void iscsit_dec_session_usage_count(struct iscsi_session *sess)
+{
+ spin_lock_bh(&sess->session_usage_lock);
+ sess->session_usage_count--;
+
+ if (!sess->session_usage_count && sess->session_waiting_on_uc)
+ complete(&sess->session_waiting_on_uc_comp);
+
+ spin_unlock_bh(&sess->session_usage_lock);
+}
+
+void iscsit_inc_session_usage_count(struct iscsi_session *sess)
+{
+ spin_lock_bh(&sess->session_usage_lock);
+ sess->session_usage_count--;
+ spin_unlock_bh(&sess->session_usage_lock);
+}
+
+/*
+ * Used before iscsi_do[rx,tx]_data() to determine iov and [rx,tx]_marker
+ * array counts needed for sync and steering.
+ */
+static inline int iscsit_determine_sync_and_steering_counts(
+ struct iscsi_conn *conn,
+ struct iscsi_data_count *count)
+{
+ u32 length = count->data_length;
+ u32 marker, markint;
+
+ count->sync_and_steering = 1;
+
+ marker = (count->type == ISCSI_RX_DATA) ?
+ conn->of_marker : conn->if_marker;
+ markint = (count->type == ISCSI_RX_DATA) ?
+ (conn->conn_ops->OFMarkInt * 4) :
+ (conn->conn_ops->IFMarkInt * 4);
+ count->ss_iov_count = count->iov_count;
+
+ while (length > 0) {
+ if (length >= marker) {
+ count->ss_iov_count += 3;
+ count->ss_marker_count += 2;
+
+ length -= marker;
+ marker = markint;
+ } else
+ length = 0;
+ }
+
+ return 0;
+}
+
+/*
+ * Setup conn->if_marker and conn->of_marker values based upon
+ * the initial marker-less interval. (see iSCSI v19 A.2)
+ */
+int iscsit_set_sync_and_steering_values(struct iscsi_conn *conn)
+{
+ int login_ifmarker_count = 0, login_ofmarker_count = 0, next_marker = 0;
+ /*
+ * IFMarkInt and OFMarkInt are negotiated as 32-bit words.
+ */
+ u32 IFMarkInt = (conn->conn_ops->IFMarkInt * 4);
+ u32 OFMarkInt = (conn->conn_ops->OFMarkInt * 4);
+
+ if (conn->conn_ops->OFMarker) {
+ /*
+ * Account for the first Login Command received not
+ * via iscsi_recv_msg().
+ */
+ conn->of_marker += ISCSI_HDR_LEN;
+ if (conn->of_marker <= OFMarkInt) {
+ conn->of_marker = (OFMarkInt - conn->of_marker);
+ } else {
+ login_ofmarker_count = (conn->of_marker / OFMarkInt);
+ next_marker = (OFMarkInt * (login_ofmarker_count + 1)) +
+ (login_ofmarker_count * MARKER_SIZE);
+ conn->of_marker = (next_marker - conn->of_marker);
+ }
+ conn->of_marker_offset = 0;
+ printk(KERN_INFO "Setting OFMarker value to %u based on Initial"
+ " Markerless Interval.\n", conn->of_marker);
+ }
+
+ if (conn->conn_ops->IFMarker) {
+ if (conn->if_marker <= IFMarkInt) {
+ conn->if_marker = (IFMarkInt - conn->if_marker);
+ } else {
+ login_ifmarker_count = (conn->if_marker / IFMarkInt);
+ next_marker = (IFMarkInt * (login_ifmarker_count + 1)) +
+ (login_ifmarker_count * MARKER_SIZE);
+ conn->if_marker = (next_marker - conn->if_marker);
+ }
+ printk(KERN_INFO "Setting IFMarker value to %u based on Initial"
+ " Markerless Interval.\n", conn->if_marker);
+ }
+
+ return 0;
+}
+
+struct iscsi_conn *iscsit_get_conn_from_cid(struct iscsi_session *sess, u16 cid)
+{
+ struct iscsi_conn *conn;
+
+ spin_lock_bh(&sess->conn_lock);
+ list_for_each_entry(conn, &sess->sess_conn_list, conn_list) {
+ if ((conn->cid == cid) &&
+ (conn->conn_state == TARG_CONN_STATE_LOGGED_IN)) {
+ iscsit_inc_conn_usage_count(conn);
+ spin_unlock_bh(&sess->conn_lock);
+ return conn;
+ }
+ }
+ spin_unlock_bh(&sess->conn_lock);
+
+ return NULL;
+}
+
+struct iscsi_conn *iscsit_get_conn_from_cid_rcfr(struct iscsi_session *sess, u16 cid)
+{
+ struct iscsi_conn *conn;
+
+ spin_lock_bh(&sess->conn_lock);
+ list_for_each_entry(conn, &sess->sess_conn_list, conn_list) {
+ if (conn->cid == cid) {
+ iscsit_inc_conn_usage_count(conn);
+ spin_lock(&conn->state_lock);
+ atomic_set(&conn->connection_wait_rcfr, 1);
+ spin_unlock(&conn->state_lock);
+ spin_unlock_bh(&sess->conn_lock);
+ return conn;
+ }
+ }
+ spin_unlock_bh(&sess->conn_lock);
+
+ return NULL;
+}
+
+void iscsit_check_conn_usage_count(struct iscsi_conn *conn)
+{
+ spin_lock_bh(&conn->conn_usage_lock);
+ if (conn->conn_usage_count != 0) {
+ conn->conn_waiting_on_uc = 1;
+ spin_unlock_bh(&conn->conn_usage_lock);
+
+ wait_for_completion(&conn->conn_waiting_on_uc_comp);
+ return;
+ }
+ spin_unlock_bh(&conn->conn_usage_lock);
+}
+
+void iscsit_dec_conn_usage_count(struct iscsi_conn *conn)
+{
+ spin_lock_bh(&conn->conn_usage_lock);
+ conn->conn_usage_count--;
+
+ if (!conn->conn_usage_count && conn->conn_waiting_on_uc)
+ complete(&conn->conn_waiting_on_uc_comp);
+
+ spin_unlock_bh(&conn->conn_usage_lock);
+}
+
+void iscsit_inc_conn_usage_count(struct iscsi_conn *conn)
+{
+ spin_lock_bh(&conn->conn_usage_lock);
+ conn->conn_usage_count++;
+ spin_unlock_bh(&conn->conn_usage_lock);
+}
+
+static int iscsit_add_nopin(struct iscsi_conn *conn, int want_response)
+{
+ u8 state;
+ struct iscsi_cmd *cmd;
+
+ cmd = iscsit_allocate_cmd(conn);
+ if (!cmd)
+ return -1;
+
+ cmd->iscsi_opcode = ISCSI_OP_NOOP_IN;
+ state = (want_response) ? ISTATE_SEND_NOPIN_WANT_RESPONSE :
+ ISTATE_SEND_NOPIN_NO_RESPONSE;
+ cmd->init_task_tag = 0xFFFFFFFF;
+ spin_lock_bh(&conn->sess->ttt_lock);
+ cmd->targ_xfer_tag = (want_response) ? conn->sess->targ_xfer_tag++ :
+ 0xFFFFFFFF;
+ if (want_response && (cmd->targ_xfer_tag == 0xFFFFFFFF))
+ cmd->targ_xfer_tag = conn->sess->targ_xfer_tag++;
+ spin_unlock_bh(&conn->sess->ttt_lock);
+
+ spin_lock_bh(&conn->cmd_lock);
+ list_add_tail(&cmd->i_list, &conn->conn_cmd_list);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (want_response)
+ iscsit_start_nopin_response_timer(conn);
+ iscsit_add_cmd_to_immediate_queue(cmd, conn, state);
+
+ return 0;
+}
+
+static void iscsit_handle_nopin_response_timeout(unsigned long data)
+{
+ struct iscsi_conn *conn = (struct iscsi_conn *) data;
+
+ iscsit_inc_conn_usage_count(conn);
+
+ spin_lock_bh(&conn->nopin_timer_lock);
+ if (conn->nopin_response_timer_flags & ISCSI_TF_STOP) {
+ spin_unlock_bh(&conn->nopin_timer_lock);
+ iscsit_dec_conn_usage_count(conn);
+ return;
+ }
+
+ TRACE(TRACE_TIMER, "Did not receive response to NOPIN on CID: %hu on"
+ " SID: %u, failing connection.\n", conn->cid,
+ conn->sess->sid);
+ conn->nopin_response_timer_flags &= ~ISCSI_TF_RUNNING;
+ spin_unlock_bh(&conn->nopin_timer_lock);
+
+ {
+ struct iscsi_portal_group *tpg = conn->sess->tpg;
+ struct iscsi_tiqn *tiqn = tpg->tpg_tiqn;
+
+ if (tiqn) {
+ spin_lock_bh(&tiqn->sess_err_stats.lock);
+ strcpy(tiqn->sess_err_stats.last_sess_fail_rem_name,
+ (void *)conn->sess->sess_ops->InitiatorName);
+ tiqn->sess_err_stats.last_sess_failure_type =
+ ISCSI_SESS_ERR_CXN_TIMEOUT;
+ tiqn->sess_err_stats.cxn_timeout_errors++;
+ conn->sess->conn_timeout_errors++;
+ spin_unlock_bh(&tiqn->sess_err_stats.lock);
+ }
+ }
+
+ iscsit_cause_connection_reinstatement(conn, 0);
+ iscsit_dec_conn_usage_count(conn);
+}
+
+void iscsit_mod_nopin_response_timer(struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na = iscsit_tpg_get_node_attrib(sess);
+
+ spin_lock_bh(&conn->nopin_timer_lock);
+ if (!(conn->nopin_response_timer_flags & ISCSI_TF_RUNNING)) {
+ spin_unlock_bh(&conn->nopin_timer_lock);
+ return;
+ }
+
+ mod_timer(&conn->nopin_response_timer,
+ (get_jiffies_64() + na->nopin_response_timeout * HZ));
+ spin_unlock_bh(&conn->nopin_timer_lock);
+}
+
+/*
+ * Called with conn->nopin_timer_lock held.
+ */
+void iscsit_start_nopin_response_timer(struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na = iscsit_tpg_get_node_attrib(sess);
+
+ spin_lock_bh(&conn->nopin_timer_lock);
+ if (conn->nopin_response_timer_flags & ISCSI_TF_RUNNING) {
+ spin_unlock_bh(&conn->nopin_timer_lock);
+ return;
+ }
+
+ init_timer(&conn->nopin_response_timer);
+ conn->nopin_response_timer.expires =
+ (get_jiffies_64() + na->nopin_response_timeout * HZ);
+ conn->nopin_response_timer.data = (unsigned long)conn;
+ conn->nopin_response_timer.function = iscsit_handle_nopin_response_timeout;
+ conn->nopin_response_timer_flags &= ~ISCSI_TF_STOP;
+ conn->nopin_response_timer_flags |= ISCSI_TF_RUNNING;
+ add_timer(&conn->nopin_response_timer);
+
+ TRACE(TRACE_TIMER, "Started NOPIN Response Timer on CID: %d to %u"
+ " seconds\n", conn->cid, na->nopin_response_timeout);
+ spin_unlock_bh(&conn->nopin_timer_lock);
+}
+
+void iscsit_stop_nopin_response_timer(struct iscsi_conn *conn)
+{
+ spin_lock_bh(&conn->nopin_timer_lock);
+ if (!(conn->nopin_response_timer_flags & ISCSI_TF_RUNNING)) {
+ spin_unlock_bh(&conn->nopin_timer_lock);
+ return;
+ }
+ conn->nopin_response_timer_flags |= ISCSI_TF_STOP;
+ spin_unlock_bh(&conn->nopin_timer_lock);
+
+ del_timer_sync(&conn->nopin_response_timer);
+
+ spin_lock_bh(&conn->nopin_timer_lock);
+ conn->nopin_response_timer_flags &= ~ISCSI_TF_RUNNING;
+ spin_unlock_bh(&conn->nopin_timer_lock);
+}
+
+static void iscsit_handle_nopin_timeout(unsigned long data)
+{
+ struct iscsi_conn *conn = (struct iscsi_conn *) data;
+
+ iscsit_inc_conn_usage_count(conn);
+
+ spin_lock_bh(&conn->nopin_timer_lock);
+ if (conn->nopin_timer_flags & ISCSI_TF_STOP) {
+ spin_unlock_bh(&conn->nopin_timer_lock);
+ iscsit_dec_conn_usage_count(conn);
+ return;
+ }
+ conn->nopin_timer_flags &= ~ISCSI_TF_RUNNING;
+ spin_unlock_bh(&conn->nopin_timer_lock);
+
+ iscsit_add_nopin(conn, 1);
+ iscsit_dec_conn_usage_count(conn);
+}
+
+/*
+ * Called with conn->nopin_timer_lock held.
+ */
+void __iscsit_start_nopin_timer(struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na = iscsit_tpg_get_node_attrib(sess);
+ /*
+ * NOPIN timeout is disabled.
+ */
+ if (!na->nopin_timeout)
+ return;
+
+ if (conn->nopin_timer_flags & ISCSI_TF_RUNNING)
+ return;
+
+ init_timer(&conn->nopin_timer);
+ conn->nopin_timer.expires = (get_jiffies_64() + na->nopin_timeout * HZ);
+ conn->nopin_timer.data = (unsigned long)conn;
+ conn->nopin_timer.function = iscsit_handle_nopin_timeout;
+ conn->nopin_timer_flags &= ~ISCSI_TF_STOP;
+ conn->nopin_timer_flags |= ISCSI_TF_RUNNING;
+ add_timer(&conn->nopin_timer);
+
+ TRACE(TRACE_TIMER, "Started NOPIN Timer on CID: %d at %u second"
+ " interval\n", conn->cid, na->nopin_timeout);
+}
+
+void iscsit_start_nopin_timer(struct iscsi_conn *conn)
+{
+ struct iscsi_session *sess = conn->sess;
+ struct iscsi_node_attrib *na = iscsit_tpg_get_node_attrib(sess);
+ /*
+ * NOPIN timeout is disabled..
+ */
+ if (!na->nopin_timeout)
+ return;
+
+ spin_lock_bh(&conn->nopin_timer_lock);
+ if (conn->nopin_timer_flags & ISCSI_TF_RUNNING) {
+ spin_unlock_bh(&conn->nopin_timer_lock);
+ return;
+ }
+
+ init_timer(&conn->nopin_timer);
+ conn->nopin_timer.expires = (get_jiffies_64() + na->nopin_timeout * HZ);
+ conn->nopin_timer.data = (unsigned long)conn;
+ conn->nopin_timer.function = iscsit_handle_nopin_timeout;
+ conn->nopin_timer_flags &= ~ISCSI_TF_STOP;
+ conn->nopin_timer_flags |= ISCSI_TF_RUNNING;
+ add_timer(&conn->nopin_timer);
+
+ TRACE(TRACE_TIMER, "Started NOPIN Timer on CID: %d at %u second"
+ " interval\n", conn->cid, na->nopin_timeout);
+ spin_unlock_bh(&conn->nopin_timer_lock);
+}
+
+void iscsit_stop_nopin_timer(struct iscsi_conn *conn)
+{
+ spin_lock_bh(&conn->nopin_timer_lock);
+ if (!(conn->nopin_timer_flags & ISCSI_TF_RUNNING)) {
+ spin_unlock_bh(&conn->nopin_timer_lock);
+ return;
+ }
+ conn->nopin_timer_flags |= ISCSI_TF_STOP;
+ spin_unlock_bh(&conn->nopin_timer_lock);
+
+ del_timer_sync(&conn->nopin_timer);
+
+ spin_lock_bh(&conn->nopin_timer_lock);
+ conn->nopin_timer_flags &= ~ISCSI_TF_RUNNING;
+ spin_unlock_bh(&conn->nopin_timer_lock);
+}
+
+int iscsit_send_tx_data(
+ struct iscsi_cmd *cmd,
+ struct iscsi_conn *conn,
+ int use_misc)
+{
+ int tx_sent, tx_size;
+ u32 iov_count;
+ struct kvec *iov;
+
+send_data:
+ tx_size = cmd->tx_size;
+
+ if (!use_misc) {
+ iov = &cmd->iov_data[0];
+ iov_count = cmd->iov_data_count;
+ } else {
+ iov = &cmd->iov_misc[0];
+ iov_count = cmd->iov_misc_count;
+ }
+
+ tx_sent = tx_data(conn, &iov[0], iov_count, tx_size);
+ if (tx_size != tx_sent) {
+ if (tx_sent == -EAGAIN) {
+ printk(KERN_ERR "tx_data() returned -EAGAIN\n");
+ goto send_data;
+ } else
+ return -1;
+ }
+ cmd->tx_size = 0;
+
+ return 0;
+}
+
+int iscsit_fe_sendpage_sg(
+ struct se_unmap_sg *u_sg,
+ struct iscsi_conn *conn)
+{
+ int tx_sent;
+ struct iscsi_cmd *cmd = (struct iscsi_cmd *)u_sg->fabric_cmd;
+ struct se_cmd *se_cmd = SE_CMD(cmd);
+ u32 len = cmd->tx_size, pg_len, se_len, se_off, tx_size;
+ struct kvec *iov = &cmd->iov_data[0];
+ struct page *page;
+ struct se_mem *se_mem = u_sg->cur_se_mem;
+
+send_hdr:
+ tx_size = (conn->conn_ops->HeaderDigest) ? ISCSI_HDR_LEN + CRC_LEN :
+ ISCSI_HDR_LEN;
+ tx_sent = tx_data(conn, iov, 1, tx_size);
+ if (tx_size != tx_sent) {
+ if (tx_sent == -EAGAIN) {
+ printk(KERN_ERR "tx_data() returned -EAGAIN\n");
+ goto send_hdr;
+ }
+ return -1;
+ }
+
+ len -= tx_size;
+ len -= u_sg->padding;
+ if (conn->conn_ops->DataDigest)
+ len -= CRC_LEN;
+ /*
+ * Start calculating from the first page of current struct se_mem.
+ */
+ page = se_mem->se_page;
+ pg_len = (PAGE_SIZE - se_mem->se_off);
+ se_len = se_mem->se_len;
+ if (se_len < pg_len)
+ pg_len = se_len;
+ se_off = se_mem->se_off;
+ /*
+ * Calucate new se_len and se_off based upon u_sg->t_offset into
+ * the current struct se_mem and possibily a different page.
+ */
+ while (u_sg->t_offset) {
+ if (u_sg->t_offset >= pg_len) {
+ u_sg->t_offset -= pg_len;
+ se_len -= pg_len;
+ se_off = 0;
+ pg_len = PAGE_SIZE;
+ page++;
+ } else {
+ se_off += u_sg->t_offset;
+ se_len -= u_sg->t_offset;
+ u_sg->t_offset = 0;
+ }
+ }
+ /*
+ * Perform sendpage() for each page in the struct se_mem
+ */
+ while (len) {
+ if (se_len > len)
+ se_len = len;
+send_pg:
+ tx_sent = conn->sock->ops->sendpage(conn->sock,
+ page, se_off, se_len, 0);
+ if (tx_sent != se_len) {
+ if (tx_sent == -EAGAIN) {
+ printk(KERN_ERR "tcp_sendpage() returned"
+ " -EAGAIN\n");
+ goto send_pg;
+ }
+
+ printk(KERN_ERR "tcp_sendpage() failure: %d\n",
+ tx_sent);
+ return -1;
+ }
+
+ len -= se_len;
+ if (!(len))
+ break;
+
+ se_len -= tx_sent;
+ if (!(se_len)) {
+ list_for_each_entry_continue(se_mem,
+ T_TASK(se_cmd)->t_mem_list, se_list)
+ break;
+
+ if (!se_mem) {
+ printk(KERN_ERR "Unable to locate next struct se_mem\n");
+ return -1;
+ }
+
+ se_len = se_mem->se_len;
+ se_off = se_mem->se_off;
+ page = se_mem->se_page;
+ } else {
+ se_len = PAGE_SIZE;
+ se_off = 0;
+ page++;
+ }
+ }
+
+send_padding:
+ if (u_sg->padding) {
+ struct kvec *iov_p =
+ &cmd->iov_data[cmd->iov_data_count-2];
+
+ tx_sent = tx_data(conn, iov_p, 1, u_sg->padding);
+ if (u_sg->padding != tx_sent) {
+ if (tx_sent == -EAGAIN) {
+ printk(KERN_ERR "tx_data() returned -EAGAIN\n");
+ goto send_padding;
+ }
+ return -1;
+ }
+ }
+
+send_datacrc:
+ if (conn->conn_ops->DataDigest) {
+ struct kvec *iov_d =
+ &cmd->iov_data[cmd->iov_data_count-1];
+
+ tx_sent = tx_data(conn, iov_d, 1, CRC_LEN);
+ if (CRC_LEN != tx_sent) {
+ if (tx_sent == -EAGAIN) {
+ printk(KERN_ERR "tx_data() returned -EAGAIN\n");
+ goto send_datacrc;
+ }
+ return -1;
+ }
+ }
+
+ return 0;
+}
+
+/*
+ * This function is used for mainly sending a ISCSI_TARG_LOGIN_RSP PDU
+ * back to the Initiator when an expection condition occurs with the
+ * errors set in status_class and status_detail.
+ *
+ * Parameters: iSCSI Connection, Status Class, Status Detail.
+ * Returns: 0 on success, -1 on error.
+ */
+int iscsit_tx_login_rsp(struct iscsi_conn *conn, u8 status_class, u8 status_detail)
+{
+ u8 iscsi_hdr[ISCSI_HDR_LEN];
+ int err;
+ struct kvec iov;
+ struct iscsi_login_rsp *hdr;
+
+ iscsit_collect_login_stats(conn, status_class, status_detail);
+
+ memset(&iov, 0, sizeof(struct kvec));
+ memset(&iscsi_hdr, 0x0, ISCSI_HDR_LEN);
+
+ hdr = (struct iscsi_login_rsp *)&iscsi_hdr;
+ hdr->opcode = ISCSI_OP_LOGIN_RSP;
+ hdr->status_class = status_class;
+ hdr->status_detail = status_detail;
+ hdr->itt = cpu_to_be32(conn->login_itt);
+
+ iov.iov_base = &iscsi_hdr;
+ iov.iov_len = ISCSI_HDR_LEN;
+
+ PRINT_BUFF(iscsi_hdr, ISCSI_HDR_LEN);
+
+ err = tx_data(conn, &iov, 1, ISCSI_HDR_LEN);
+ if (err != ISCSI_HDR_LEN) {
+ printk(KERN_ERR "tx_data returned less than expected\n");
+ return -1;
+ }
+
+ return 0;
+}
+
+void iscsit_print_session_params(struct iscsi_session *sess)
+{
+ struct iscsi_conn *conn;
+
+ printk(KERN_INFO "-----------------------------[Session Params for"
+ " SID: %u]-----------------------------\n", sess->sid);
+ spin_lock_bh(&sess->conn_lock);
+ list_for_each_entry(conn, &sess->sess_conn_list, conn_list)
+ iscsi_dump_conn_ops(conn->conn_ops);
+ spin_unlock_bh(&sess->conn_lock);
+
+ iscsi_dump_sess_ops(sess->sess_ops);
+}
+
+static inline int iscsit_do_rx_data(
+ struct iscsi_conn *conn,
+ struct iscsi_data_count *count)
+{
+ int data = count->data_length, rx_loop = 0, total_rx = 0, iov_len;
+ u32 rx_marker_val[count->ss_marker_count], rx_marker_iov = 0;
+ struct kvec iov[count->ss_iov_count], *iov_p;
+ struct msghdr msg;
+
+ if (!conn || !conn->sock || !conn->conn_ops)
+ return -1;
+
+ memset(&msg, 0, sizeof(struct msghdr));
+
+ if (count->sync_and_steering) {
+ int size = 0;
+ u32 i, orig_iov_count = 0;
+ u32 orig_iov_len = 0, orig_iov_loc = 0;
+ u32 iov_count = 0, per_iov_bytes = 0;
+ u32 *rx_marker, old_rx_marker = 0;
+ struct kvec *iov_record;
+
+ memset(&rx_marker_val, 0,
+ count->ss_marker_count * sizeof(u32));
+ memset(&iov, 0, count->ss_iov_count * sizeof(struct kvec));
+
+ iov_record = count->iov;
+ orig_iov_count = count->iov_count;
+ rx_marker = &conn->of_marker;
+
+ i = 0;
+ size = data;
+ orig_iov_len = iov_record[orig_iov_loc].iov_len;
+ while (size > 0) {
+ TRACE(TRACE_SSLR, "rx_data: #1 orig_iov_len %u,"
+ " orig_iov_loc %u\n", orig_iov_len, orig_iov_loc);
+ TRACE(TRACE_SSLR, "rx_data: #2 rx_marker %u, size"
+ " %u\n", *rx_marker, size);
+
+ if (orig_iov_len >= *rx_marker) {
+ iov[iov_count].iov_len = *rx_marker;
+ iov[iov_count++].iov_base =
+ (iov_record[orig_iov_loc].iov_base +
+ per_iov_bytes);
+
+ iov[iov_count].iov_len = (MARKER_SIZE / 2);
+ iov[iov_count++].iov_base =
+ &rx_marker_val[rx_marker_iov++];
+ iov[iov_count].iov_len = (MARKER_SIZE / 2);
+ iov[iov_count++].iov_base =
+ &rx_marker_val[rx_marker_iov++];
+ old_rx_marker = *rx_marker;
+
+ /*
+ * OFMarkInt is in 32-bit words.
+ */
+ *rx_marker = (conn->conn_ops->OFMarkInt * 4);
+ size -= old_rx_marker;
+ orig_iov_len -= old_rx_marker;
+ per_iov_bytes += old_rx_marker;
+
+ TRACE(TRACE_SSLR, "rx_data: #3 new_rx_marker"
+ " %u, size %u\n", *rx_marker, size);
+ } else {
+ iov[iov_count].iov_len = orig_iov_len;
+ iov[iov_count++].iov_base =
+ (iov_record[orig_iov_loc].iov_base +
+ per_iov_bytes);
+
+ per_iov_bytes = 0;
+ *rx_marker -= orig_iov_len;
+ size -= orig_iov_len;
+
+ if (size)
+ orig_iov_len =
+ iov_record[++orig_iov_loc].iov_len;
+
+ TRACE(TRACE_SSLR, "rx_data: #4 new_rx_marker"
+ " %u, size %u\n", *rx_marker, size);
+ }
+ }
+ data += (rx_marker_iov * (MARKER_SIZE / 2));
+
+ iov_p = &iov[0];
+ iov_len = iov_count;
+
+ if (iov_count > count->ss_iov_count) {
+ printk(KERN_ERR "iov_count: %d, count->ss_iov_count:"
+ " %d\n", iov_count, count->ss_iov_count);
+ return -1;
+ }
+ if (rx_marker_iov > count->ss_marker_count) {
+ printk(KERN_ERR "rx_marker_iov: %d, count->ss_marker"
+ "_count: %d\n", rx_marker_iov,
+ count->ss_marker_count);
+ return -1;
+ }
+ } else {
+ iov_p = count->iov;
+ iov_len = count->iov_count;
+ }
+
+ while (total_rx < data) {
+ rx_loop = kernel_recvmsg(conn->sock, &msg, iov_p, iov_len,
+ (data - total_rx), MSG_WAITALL);
+ if (rx_loop <= 0) {
+ TRACE(TRACE_NET, "rx_loop: %d total_rx: %d\n",
+ rx_loop, total_rx);
+ return rx_loop;
+ }
+ total_rx += rx_loop;
+ TRACE(TRACE_NET, "rx_loop: %d, total_rx: %d, data: %d\n",
+ rx_loop, total_rx, data);
+ }
+
+ if (count->sync_and_steering) {
+ int j;
+ for (j = 0; j < rx_marker_iov; j++) {
+ TRACE(TRACE_SSLR, "rx_data: #5 j: %d, offset: %d\n",
+ j, rx_marker_val[j]);
+ conn->of_marker_offset = rx_marker_val[j];
+ }
+ total_rx -= (rx_marker_iov * (MARKER_SIZE / 2));
+ }
+
+ return total_rx;
+}
+
+static inline int iscsit_do_tx_data(
+ struct iscsi_conn *conn,
+ struct iscsi_data_count *count)
+{
+ int data = count->data_length, total_tx = 0, tx_loop = 0, iov_len;
+ u32 tx_marker_val[count->ss_marker_count], tx_marker_iov = 0;
+ struct kvec iov[count->ss_iov_count], *iov_p;
+ struct msghdr msg;
+
+ if (!conn || !conn->sock || !conn->conn_ops)
+ return -1;
+
+ if (data <= 0) {
+ printk(KERN_ERR "Data length is: %d\n", data);
+ return -1;
+ }
+
+ memset(&msg, 0, sizeof(struct msghdr));
+
+ if (count->sync_and_steering) {
+ int size = 0;
+ u32 i, orig_iov_count = 0;
+ u32 orig_iov_len = 0, orig_iov_loc = 0;
+ u32 iov_count = 0, per_iov_bytes = 0;
+ u32 *tx_marker, old_tx_marker = 0;
+ struct kvec *iov_record;
+
+ memset(&tx_marker_val, 0,
+ count->ss_marker_count * sizeof(u32));
+ memset(&iov, 0, count->ss_iov_count * sizeof(struct kvec));
+
+ iov_record = count->iov;
+ orig_iov_count = count->iov_count;
+ tx_marker = &conn->if_marker;
+
+ i = 0;
+ size = data;
+ orig_iov_len = iov_record[orig_iov_loc].iov_len;
+ while (size > 0) {
+ TRACE(TRACE_SSLT, "tx_data: #1 orig_iov_len %u,"
+ " orig_iov_loc %u\n", orig_iov_len, orig_iov_loc);
+ TRACE(TRACE_SSLT, "tx_data: #2 tx_marker %u, size"
+ " %u\n", *tx_marker, size);
+
+ if (orig_iov_len >= *tx_marker) {
+ iov[iov_count].iov_len = *tx_marker;
+ iov[iov_count++].iov_base =
+ (iov_record[orig_iov_loc].iov_base +
+ per_iov_bytes);
+
+ tx_marker_val[tx_marker_iov] =
+ (size - *tx_marker);
+ iov[iov_count].iov_len = (MARKER_SIZE / 2);
+ iov[iov_count++].iov_base =
+ &tx_marker_val[tx_marker_iov++];
+ iov[iov_count].iov_len = (MARKER_SIZE / 2);
+ iov[iov_count++].iov_base =
+ &tx_marker_val[tx_marker_iov++];
+ old_tx_marker = *tx_marker;
+
+ /*
+ * IFMarkInt is in 32-bit words.
+ */
+ *tx_marker = (conn->conn_ops->IFMarkInt * 4);
+ size -= old_tx_marker;
+ orig_iov_len -= old_tx_marker;
+ per_iov_bytes += old_tx_marker;
+
+ TRACE(TRACE_SSLT, "tx_data: #3 new_tx_marker"
+ " %u, size %u\n", *tx_marker, size);
+ TRACE(TRACE_SSLT, "tx_data: #4 offset %u\n",
+ tx_marker_val[tx_marker_iov-1]);
+ } else {
+ iov[iov_count].iov_len = orig_iov_len;
+ iov[iov_count++].iov_base
+ = (iov_record[orig_iov_loc].iov_base +
+ per_iov_bytes);
+
+ per_iov_bytes = 0;
+ *tx_marker -= orig_iov_len;
+ size -= orig_iov_len;
+
+ if (size)
+ orig_iov_len =
+ iov_record[++orig_iov_loc].iov_len;
+
+ TRACE(TRACE_SSLT, "tx_data: #5 new_tx_marker"
+ " %u, size %u\n", *tx_marker, size);
+ }
+ }
+
+ data += (tx_marker_iov * (MARKER_SIZE / 2));
+
+ iov_p = &iov[0];
+ iov_len = iov_count;
+
+ if (iov_count > count->ss_iov_count) {
+ printk(KERN_ERR "iov_count: %d, count->ss_iov_count:"
+ " %d\n", iov_count, count->ss_iov_count);
+ return -1;
+ }
+ if (tx_marker_iov > count->ss_marker_count) {
+ printk(KERN_ERR "tx_marker_iov: %d, count->ss_marker"
+ "_count: %d\n", tx_marker_iov,
+ count->ss_marker_count);
+ return -1;
+ }
+ } else {
+ iov_p = count->iov;
+ iov_len = count->iov_count;
+ }
+
+ while (total_tx < data) {
+ tx_loop = kernel_sendmsg(conn->sock, &msg, iov_p, iov_len,
+ (data - total_tx));
+ if (tx_loop <= 0) {
+ TRACE(TRACE_NET, "tx_loop: %d total_tx %d\n",
+ tx_loop, total_tx);
+ return tx_loop;
+ }
+ total_tx += tx_loop;
+ TRACE(TRACE_NET, "tx_loop: %d, total_tx: %d, data: %d\n",
+ tx_loop, total_tx, data);
+ }
+
+ if (count->sync_and_steering)
+ total_tx -= (tx_marker_iov * (MARKER_SIZE / 2));
+
+ return total_tx;
+}
+
+int rx_data(
+ struct iscsi_conn *conn,
+ struct kvec *iov,
+ int iov_count,
+ int data)
+{
+ struct iscsi_data_count c;
+
+ if (!conn || !conn->sock || !conn->conn_ops)
+ return -1;
+
+ memset(&c, 0, sizeof(struct iscsi_data_count));
+ c.iov = iov;
+ c.iov_count = iov_count;
+ c.data_length = data;
+ c.type = ISCSI_RX_DATA;
+
+ if (conn->conn_ops->OFMarker &&
+ (conn->conn_state >= TARG_CONN_STATE_LOGGED_IN)) {
+ if (iscsit_determine_sync_and_steering_counts(conn, &c) < 0)
+ return -1;
+ }
+
+ return iscsit_do_rx_data(conn, &c);
+}
+
+int tx_data(
+ struct iscsi_conn *conn,
+ struct kvec *iov,
+ int iov_count,
+ int data)
+{
+ struct iscsi_data_count c;
+
+ if (!conn || !conn->sock || !conn->conn_ops)
+ return -1;
+
+ memset(&c, 0, sizeof(struct iscsi_data_count));
+ c.iov = iov;
+ c.iov_count = iov_count;
+ c.data_length = data;
+ c.type = ISCSI_TX_DATA;
+
+ if (conn->conn_ops->IFMarker &&
+ (conn->conn_state >= TARG_CONN_STATE_LOGGED_IN)) {
+ if (iscsit_determine_sync_and_steering_counts(conn, &c) < 0)
+ return -1;
+ }
+
+ return iscsit_do_tx_data(conn, &c);
+}
+
+void iscsit_collect_login_stats(
+ struct iscsi_conn *conn,
+ u8 status_class,
+ u8 status_detail)
+{
+ struct iscsi_param *intrname = NULL;
+ struct iscsi_tiqn *tiqn;
+ struct iscsi_login_stats *ls;
+
+ tiqn = iscsit_snmp_get_tiqn(conn);
+ if (!(tiqn))
+ return;
+
+ ls = &tiqn->login_stats;
+
+ spin_lock(&ls->lock);
+ if (!strcmp(conn->login_ip, ls->last_intr_fail_ip_addr) &&
+ ((get_jiffies_64() - ls->last_fail_time) < 10)) {
+ /* We already have the failure info for this login */
+ spin_unlock(&ls->lock);
+ return;
+ }
+
+ if (status_class == ISCSI_STATUS_CLS_SUCCESS)
+ ls->accepts++;
+ else if (status_class == ISCSI_STATUS_CLS_REDIRECT) {
+ ls->redirects++;
+ ls->last_fail_type = ISCSI_LOGIN_FAIL_REDIRECT;
+ } else if ((status_class == ISCSI_STATUS_CLS_INITIATOR_ERR) &&
+ (status_detail == ISCSI_LOGIN_STATUS_AUTH_FAILED)) {
+ ls->authenticate_fails++;
+ ls->last_fail_type = ISCSI_LOGIN_FAIL_AUTHENTICATE;
+ } else if ((status_class == ISCSI_STATUS_CLS_INITIATOR_ERR) &&
+ (status_detail == ISCSI_LOGIN_STATUS_TGT_FORBIDDEN)) {
+ ls->authorize_fails++;
+ ls->last_fail_type = ISCSI_LOGIN_FAIL_AUTHORIZE;
+ } else if ((status_class == ISCSI_STATUS_CLS_INITIATOR_ERR) &&
+ (status_detail == ISCSI_LOGIN_STATUS_INIT_ERR)) {
+ ls->negotiate_fails++;
+ ls->last_fail_type = ISCSI_LOGIN_FAIL_NEGOTIATE;
+ } else {
+ ls->other_fails++;
+ ls->last_fail_type = ISCSI_LOGIN_FAIL_OTHER;
+ }
+
+ /* Save initiator name, ip address and time, if it is a failed login */
+ if (status_class != ISCSI_STATUS_CLS_SUCCESS) {
+ if (conn->param_list)
+ intrname = iscsi_find_param_from_key(INITIATORNAME,
+ conn->param_list);
+ strcpy(ls->last_intr_fail_name,
+ (intrname ? intrname->value : "Unknown"));
+
+ ls->last_intr_fail_ip_family = conn->sock->sk->sk_family;
+ snprintf(ls->last_intr_fail_ip_addr, IPV6_ADDRESS_SPACE,
+ "%s", conn->login_ip);
+ ls->last_fail_time = get_jiffies_64();
+ }
+
+ spin_unlock(&ls->lock);
+}
+
+struct iscsi_tiqn *iscsit_snmp_get_tiqn(struct iscsi_conn *conn)
+{
+ struct iscsi_portal_group *tpg;
+
+ if (!conn || !conn->sess)
+ return NULL;
+
+ tpg = conn->sess->tpg;
+ if (!tpg)
+ return NULL;
+
+ if (!tpg->tpg_tiqn)
+ return NULL;
+
+ return tpg->tpg_tiqn;
+}
+
+extern int iscsit_build_sendtargets_response(struct iscsi_cmd *cmd)
+{
+ char *payload = NULL;
+ struct iscsi_conn *conn = cmd->conn;
+ struct iscsi_portal_group *tpg;
+ struct iscsi_tiqn *tiqn;
+ struct iscsi_tpg_np *tpg_np;
+ int buffer_len, end_of_buf = 0, len = 0, payload_len = 0;
+ unsigned char buf[256];
+
+ buffer_len = (conn->conn_ops->MaxRecvDataSegmentLength > 32768) ?
+ 32768 : conn->conn_ops->MaxRecvDataSegmentLength;
+
+ payload = kzalloc(buffer_len, GFP_KERNEL);
+ if (!payload) {
+ printk(KERN_ERR "Unable to allocate memory for sendtargets"
+ " response.\n");
+ return -1;
+ }
+
+ spin_lock(&tiqn_lock);
+ list_for_each_entry(tiqn, &g_tiqn_list, tiqn_list) {
+ memset(buf, 0, 256);
+
+ len = sprintf(buf, "TargetName=%s", tiqn->tiqn);
+ len += 1;
+
+ if ((len + payload_len) > buffer_len) {
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+ end_of_buf = 1;
+ goto eob;
+ }
+ memcpy((void *)payload + payload_len, buf, len);
+ payload_len += len;
+
+ spin_lock(&tiqn->tiqn_tpg_lock);
+ list_for_each_entry(tpg, &tiqn->tiqn_tpg_list, tpg_list) {
+
+ spin_lock(&tpg->tpg_state_lock);
+ if ((tpg->tpg_state == TPG_STATE_FREE) ||
+ (tpg->tpg_state == TPG_STATE_INACTIVE)) {
+ spin_unlock(&tpg->tpg_state_lock);
+ continue;
+ }
+ spin_unlock(&tpg->tpg_state_lock);
+
+ spin_lock(&tpg->tpg_np_lock);
+ list_for_each_entry(tpg_np, &tpg->tpg_gnp_list,
+ tpg_np_list) {
+ memset(buf, 0, 256);
+
+ len = sprintf(buf, "TargetAddress="
+ "%s%s%s:%hu,%hu",
+ (tpg_np->tpg_np->np_sockaddr.ss_family == AF_INET6) ?
+ "[" : "", tpg_np->tpg_np->np_ip,
+ (tpg_np->tpg_np->np_sockaddr.ss_family == AF_INET6) ?
+ "]" : "", tpg_np->tpg_np->np_port,
+ tpg->tpgt);
+ len += 1;
+
+ if ((len + payload_len) > buffer_len) {
+ spin_unlock(&tpg->tpg_np_lock);
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+ end_of_buf = 1;
+ goto eob;
+ }
+
+ memcpy((void *)payload + payload_len, buf, len);
+ payload_len += len;
+ }
+ spin_unlock(&tpg->tpg_np_lock);
+ }
+ spin_unlock(&tiqn->tiqn_tpg_lock);
+eob:
+ if (end_of_buf)
+ break;
+ }
+ spin_unlock(&tiqn_lock);
+
+ cmd->buf_ptr = payload;
+
+ return payload_len;
+}
diff --git a/drivers/target/iscsi/iscsi_target_util.h b/drivers/target/iscsi/iscsi_target_util.h
new file mode 100644
index 0000000..18aa8dd
--- /dev/null
+++ b/drivers/target/iscsi/iscsi_target_util.h
@@ -0,0 +1,105 @@
+#ifndef ISCSI_TARGET_UTIL_H
+#define ISCSI_TARGET_UTIL_H
+
+#define MARKER_SIZE 8
+
+struct se_cmd;
+
+struct se_offset_map {
+ int map_reset;
+ u32 iovec_length;
+ u32 iscsi_offset;
+ u32 current_offset;
+ u32 orig_offset;
+ u32 sg_count;
+ u32 sg_current;
+ u32 sg_length;
+ struct page *sg_page;
+ struct se_mem *map_se_mem;
+ struct se_mem *map_orig_se_mem;
+ void *iovec_base;
+};
+
+struct se_map_sg {
+ int sg_kmap_active:1;
+ u32 data_length;
+ u32 data_offset;
+ void *fabric_cmd;
+ struct se_cmd *se_cmd;
+ struct kvec *iov;
+};
+
+struct se_unmap_sg {
+ u32 data_length;
+ u32 sg_count;
+ u32 sg_offset;
+ u32 padding;
+ u32 t_offset;
+ void *fabric_cmd;
+ struct se_cmd *se_cmd;
+ struct se_offset_map lmap;
+ struct se_mem *cur_se_mem;
+};
+
+extern int iscsit_add_r2t_to_list(struct iscsi_cmd *, u32, u32, int, u32);
+extern struct iscsi_r2t *iscsit_get_r2t_for_eos(struct iscsi_cmd *, u32, u32);
+extern struct iscsi_r2t *iscsit_get_r2t_from_list(struct iscsi_cmd *);
+extern void iscsit_free_r2t(struct iscsi_r2t *, struct iscsi_cmd *);
+extern void iscsit_free_r2ts_from_list(struct iscsi_cmd *);
+extern struct iscsi_cmd *iscsit_allocate_cmd(struct iscsi_conn *);
+extern struct iscsi_cmd *iscsit_allocate_se_cmd(struct iscsi_conn *, u32, int, int);
+extern struct iscsi_cmd *iscsit_allocate_se_cmd_for_tmr(struct iscsi_conn *, u8);
+extern int iscsit_decide_list_to_build(struct iscsi_cmd *, u32);
+extern struct iscsi_seq *iscsit_get_seq_holder_for_datain(struct iscsi_cmd *, u32);
+extern struct iscsi_seq *iscsit_get_seq_holder_for_r2t(struct iscsi_cmd *);
+extern struct iscsi_r2t *iscsit_get_holder_for_r2tsn(struct iscsi_cmd *, u32);
+extern int iscsit_check_received_cmdsn(struct iscsi_conn *, struct iscsi_cmd *, u32);
+extern int iscsit_check_unsolicited_dataout(struct iscsi_cmd *, unsigned char *);
+extern struct iscsi_cmd *iscsit_find_cmd_from_itt(struct iscsi_conn *, u32);
+extern struct iscsi_cmd *iscsit_find_cmd_from_itt_or_dump(struct iscsi_conn *,
+ u32, u32);
+extern struct iscsi_cmd *iscsit_find_cmd_from_ttt(struct iscsi_conn *, u32);
+extern int iscsit_find_cmd_for_recovery(struct iscsi_session *, struct iscsi_cmd **,
+ struct iscsi_conn_recovery **, u32);
+extern void iscsit_add_cmd_to_immediate_queue(struct iscsi_cmd *, struct iscsi_conn *, u8);
+extern struct iscsi_queue_req *iscsit_get_cmd_from_immediate_queue(struct iscsi_conn *);
+extern void iscsit_add_cmd_to_response_queue(struct iscsi_cmd *, struct iscsi_conn *, u8);
+extern struct iscsi_queue_req *iscsit_get_cmd_from_response_queue(struct iscsi_conn *);
+extern void iscsit_remove_cmd_from_tx_queues(struct iscsi_cmd *, struct iscsi_conn *);
+extern void iscsit_free_queue_reqs_for_conn(struct iscsi_conn *);
+extern void iscsit_release_cmd(struct iscsi_cmd *);
+extern int iscsit_check_session_usage_count(struct iscsi_session *);
+extern void iscsit_dec_session_usage_count(struct iscsi_session *);
+extern void iscsit_inc_session_usage_count(struct iscsi_session *);
+extern int iscsit_set_sync_and_steering_values(struct iscsi_conn *);
+extern struct iscsi_conn *iscsit_get_conn_from_cid(struct iscsi_session *, u16);
+extern struct iscsi_conn *iscsit_get_conn_from_cid_rcfr(struct iscsi_session *, u16);
+extern void iscsit_check_conn_usage_count(struct iscsi_conn *);
+extern void iscsit_dec_conn_usage_count(struct iscsi_conn *);
+extern void iscsit_inc_conn_usage_count(struct iscsi_conn *);
+extern void iscsit_mod_nopin_response_timer(struct iscsi_conn *);
+extern void iscsit_start_nopin_response_timer(struct iscsi_conn *);
+extern void iscsit_stop_nopin_response_timer(struct iscsi_conn *);
+extern void __iscsit_start_nopin_timer(struct iscsi_conn *);
+extern void iscsit_start_nopin_timer(struct iscsi_conn *);
+extern void iscsit_stop_nopin_timer(struct iscsi_conn *);
+extern int iscsit_send_tx_data(struct iscsi_cmd *, struct iscsi_conn *, int);
+extern int iscsit_fe_sendpage_sg(struct se_unmap_sg *, struct iscsi_conn *);
+extern int iscsit_tx_login_rsp(struct iscsi_conn *, u8, u8);
+extern void iscsit_print_session_params(struct iscsi_session *);
+extern int iscsit_print_dev_to_proc(char *, char **, off_t, int);
+extern int iscsit_print_sessions_to_proc(char *, char **, off_t, int);
+extern int iscsit_print_tpg_to_proc(char *, char **, off_t, int);
+extern int rx_data(struct iscsi_conn *, struct kvec *, int, int);
+extern int tx_data(struct iscsi_conn *, struct kvec *, int, int);
+extern void iscsit_collect_login_stats(struct iscsi_conn *, u8, u8);
+extern struct iscsi_tiqn *iscsit_snmp_get_tiqn(struct iscsi_conn *);
+extern int iscsit_build_sendtargets_response(struct iscsi_cmd *);
+
+extern struct target_fabric_configfs *lio_target_fabric_configfs;
+extern struct iscsi_global *iscsi_global;
+extern struct kmem_cache *lio_cmd_cache;
+extern struct kmem_cache *lio_qr_cache;
+extern struct kmem_cache *lio_r2t_cache;
+
+#endif /*** ISCSI_TARGET_UTIL_H ***/
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 40+ messages in thread* [PATCH-v2 14/14] iscsi-target: Add Makefile/Kconfig and update TCM top level
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (13 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 13/14] iscsi-target: Add misc utility and debug logic Nicholas A. Bellinger
@ 2011-03-23 3:06 ` Nicholas A. Bellinger
2011-03-23 8:54 ` [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge FUJITA Tomonori
15 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 3:06 UTC (permalink / raw)
To: linux-scsi, James Bottomley
Cc: Christoph Hellwig, Mike Christie, Hannes Reinecke,
FUJITA Tomonori, Boaz Harrosh, Nicholas Bellinger
From: Nicholas Bellinger <nab@linux-iscsi.org>
Add Makefile/Kconfig and update drivers/target/[Makefile,Kconfig]
to include the fabric module.
Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
---
drivers/target/Kconfig | 2 ++
drivers/target/Makefile | 3 +++
drivers/target/iscsi/Kconfig | 17 +++++++++++++++++
drivers/target/iscsi/Makefile | 20 ++++++++++++++++++++
4 files changed, 42 insertions(+), 0 deletions(-)
create mode 100644 drivers/target/iscsi/Kconfig
create mode 100644 drivers/target/iscsi/Makefile
diff --git a/drivers/target/Kconfig b/drivers/target/Kconfig
index 2fac3be..03a821c 100644
--- a/drivers/target/Kconfig
+++ b/drivers/target/Kconfig
@@ -29,4 +29,6 @@ config TCM_PSCSI
Say Y here to enable the TCM/pSCSI subsystem plugin for non-buffered
passthrough access to Linux/SCSI device
+source "drivers/target/iscsi/Kconfig"
+
endif
diff --git a/drivers/target/Makefile b/drivers/target/Makefile
index 4f3cb78..ff9d62b 100644
--- a/drivers/target/Makefile
+++ b/drivers/target/Makefile
@@ -21,3 +21,6 @@ obj-$(CONFIG_TARGET_CORE) += target_core_mod.o
obj-$(CONFIG_TCM_IBLOCK) += target_core_iblock.o
obj-$(CONFIG_TCM_FILEIO) += target_core_file.o
obj-$(CONFIG_TCM_PSCSI) += target_core_pscsi.o
+
+# Fabric modules
+obj-$(CONFIG_ISCSI_TARGET) += iscsi/
diff --git a/drivers/target/iscsi/Kconfig b/drivers/target/iscsi/Kconfig
new file mode 100644
index 0000000..cc34243
--- /dev/null
+++ b/drivers/target/iscsi/Kconfig
@@ -0,0 +1,17 @@
+config ISCSI_TARGET
+ tristate "Linux-iSCSI.org iSCSI Target Mode Stack"
+ select CRYPTO
+ select CRYPTO_CRC32C
+ select CRYPTO_CRC32C_INTEL if X86
+ help
+ Say Y here to enable the ConfigFS enabled Linux-iSCSI.org iSCSI
+ Target Mode Stack.
+
+if ISCSI_TARGET
+
+config ISCSI_TARGET_DEBUG
+ bool "LIO-Target iscsi_debug.h ring buffer messages"
+ help
+ Say Y here to enable the legacy DEBUG_ISCSI macros in iscsi_debug.h
+
+endif
diff --git a/drivers/target/iscsi/Makefile b/drivers/target/iscsi/Makefile
new file mode 100644
index 0000000..5b9a2cf
--- /dev/null
+++ b/drivers/target/iscsi/Makefile
@@ -0,0 +1,20 @@
+iscsi_target_mod-y += iscsi_target_parameters.o \
+ iscsi_target_seq_pdu_list.o \
+ iscsi_target_tq.o \
+ iscsi_target_auth.o \
+ iscsi_target_datain_values.o \
+ iscsi_target_device.o \
+ iscsi_target_erl0.o \
+ iscsi_target_erl1.o \
+ iscsi_target_erl2.o \
+ iscsi_target_login.o \
+ iscsi_target_nego.o \
+ iscsi_target_nodeattrib.o \
+ iscsi_target_tmr.o \
+ iscsi_target_tpg.o \
+ iscsi_target_util.o \
+ iscsi_target.o \
+ iscsi_target_configfs.o \
+ iscsi_target_stat.o
+
+obj-$(CONFIG_ISCSI_TARGET) += iscsi_target_mod.o
--
1.7.4.1
^ permalink raw reply related [flat|nested] 40+ messages in thread* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-23 3:06 [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge Nicholas A. Bellinger
` (14 preceding siblings ...)
2011-03-23 3:06 ` [PATCH-v2 14/14] iscsi-target: Add Makefile/Kconfig and update TCM top level Nicholas A. Bellinger
@ 2011-03-23 8:54 ` FUJITA Tomonori
2011-03-23 10:22 ` Nicholas A. Bellinger
15 siblings, 1 reply; 40+ messages in thread
From: FUJITA Tomonori @ 2011-03-23 8:54 UTC (permalink / raw)
To: nab
Cc: linux-scsi, James.Bottomley, hch, michaelc, hare, fujita.tomonori,
bharrosh
I've not looked at the code yet but...
On Tue, 22 Mar 2011 20:06:31 -0700
"Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> Nicholas Bellinger (14):
> iscsi: Resolve iscsi_proto.h naming conflicts with
> drivers/target/iscsi
> iscsi: Add Serial Number Arithmetic LT and GT into iscsi_proto.h
> target: Convert REPORT_LUNs to use int_to_scsilun
> iscsi-target: Add iSCSI fabric support for target v4
> iscsi-target: Add TCM v4 compatiable ConfigFS control plane
> iscsi-target: Add configfs fabric dependent statistics
> iscsi-target: Add TPG and Device logic
> iscsi-target: Add iSCSI Login Negotiation and Parameter logic
> iscsi-target: Add CHAP Authentication support using libcrypto
> iscsi-target: Add Sequence/PDU list + DataIN response logic
> iscsi-target: Add iSCSI Error Recovery Hierarchy support
> iscsi-target: Add support for task management operations
> iscsi-target: Add misc utility and debug logic
> iscsi-target: Add Makefile/Kconfig and update TCM top level
>
> drivers/infiniband/ulp/iser/iser_initiator.c | 2 +-
> drivers/scsi/be2iscsi/be_main.h | 4 +-
> drivers/scsi/bnx2i/bnx2i_hwi.c | 8 +-
> drivers/scsi/bnx2i/bnx2i_iscsi.c | 2 +-
> drivers/scsi/libiscsi.c | 22 +-
> drivers/target/Kconfig | 2 +
> drivers/target/Makefile | 3 +
> drivers/target/iscsi/Kconfig | 17 +
> drivers/target/iscsi/Makefile | 20 +
> drivers/target/iscsi/iscsi_target.c | 5017 +++++++++++++++++++++
> drivers/target/iscsi/iscsi_target.h | 32 +
> drivers/target/iscsi/iscsi_target_auth.c | 496 ++
> drivers/target/iscsi/iscsi_target_auth.h | 32 +
> drivers/target/iscsi/iscsi_target_configfs.c | 1914 ++++++++
> drivers/target/iscsi/iscsi_target_configfs.h | 7 +
> drivers/target/iscsi/iscsi_target_core.h | 881 ++++
> drivers/target/iscsi/iscsi_target_datain_values.c | 531 +++
> drivers/target/iscsi/iscsi_target_datain_values.h | 15 +
> drivers/target/iscsi/iscsi_target_debug.h | 113 +
> drivers/target/iscsi/iscsi_target_device.c | 108 +
> drivers/target/iscsi/iscsi_target_device.h | 9 +
> drivers/target/iscsi/iscsi_target_erl0.c | 1005 +++++
> drivers/target/iscsi/iscsi_target_erl0.h | 17 +
> drivers/target/iscsi/iscsi_target_erl1.c | 1305 ++++++
> drivers/target/iscsi/iscsi_target_erl1.h | 28 +
> drivers/target/iscsi/iscsi_target_erl2.c | 483 ++
> drivers/target/iscsi/iscsi_target_erl2.h | 18 +
> drivers/target/iscsi/iscsi_target_login.c | 1245 +++++
> drivers/target/iscsi/iscsi_target_login.h | 13 +
> drivers/target/iscsi/iscsi_target_nego.c | 1068 +++++
> drivers/target/iscsi/iscsi_target_nego.h | 19 +
Why do we need to do the pre-nexus operations such as login,
authentication, negotiation, etc in kernel?
^ permalink raw reply [flat|nested] 40+ messages in thread* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-23 8:54 ` [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge FUJITA Tomonori
@ 2011-03-23 10:22 ` Nicholas A. Bellinger
2011-03-23 11:07 ` FUJITA Tomonori
0 siblings, 1 reply; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 10:22 UTC (permalink / raw)
To: FUJITA Tomonori
Cc: linux-scsi, James.Bottomley, hch, michaelc, hare, bharrosh
On Wed, 2011-03-23 at 17:54 +0900, FUJITA Tomonori wrote:
> I've not looked at the code yet but...
>
> On Tue, 22 Mar 2011 20:06:31 -0700
> "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
>
> > Nicholas Bellinger (14):
> > iscsi: Resolve iscsi_proto.h naming conflicts with
> > drivers/target/iscsi
> > iscsi: Add Serial Number Arithmetic LT and GT into iscsi_proto.h
> > target: Convert REPORT_LUNs to use int_to_scsilun
> > iscsi-target: Add iSCSI fabric support for target v4
> > iscsi-target: Add TCM v4 compatiable ConfigFS control plane
> > iscsi-target: Add configfs fabric dependent statistics
> > iscsi-target: Add TPG and Device logic
> > iscsi-target: Add iSCSI Login Negotiation and Parameter logic
> > iscsi-target: Add CHAP Authentication support using libcrypto
> > iscsi-target: Add Sequence/PDU list + DataIN response logic
> > iscsi-target: Add iSCSI Error Recovery Hierarchy support
> > iscsi-target: Add support for task management operations
> > iscsi-target: Add misc utility and debug logic
> > iscsi-target: Add Makefile/Kconfig and update TCM top level
> >
> > drivers/infiniband/ulp/iser/iser_initiator.c | 2 +-
> > drivers/scsi/be2iscsi/be_main.h | 4 +-
> > drivers/scsi/bnx2i/bnx2i_hwi.c | 8 +-
> > drivers/scsi/bnx2i/bnx2i_iscsi.c | 2 +-
> > drivers/scsi/libiscsi.c | 22 +-
> > drivers/target/Kconfig | 2 +
> > drivers/target/Makefile | 3 +
> > drivers/target/iscsi/Kconfig | 17 +
> > drivers/target/iscsi/Makefile | 20 +
> > drivers/target/iscsi/iscsi_target.c | 5017 +++++++++++++++++++++
> > drivers/target/iscsi/iscsi_target.h | 32 +
> > drivers/target/iscsi/iscsi_target_auth.c | 496 ++
> > drivers/target/iscsi/iscsi_target_auth.h | 32 +
> > drivers/target/iscsi/iscsi_target_configfs.c | 1914 ++++++++
> > drivers/target/iscsi/iscsi_target_configfs.h | 7 +
> > drivers/target/iscsi/iscsi_target_core.h | 881 ++++
> > drivers/target/iscsi/iscsi_target_datain_values.c | 531 +++
> > drivers/target/iscsi/iscsi_target_datain_values.h | 15 +
> > drivers/target/iscsi/iscsi_target_debug.h | 113 +
> > drivers/target/iscsi/iscsi_target_device.c | 108 +
> > drivers/target/iscsi/iscsi_target_device.h | 9 +
> > drivers/target/iscsi/iscsi_target_erl0.c | 1005 +++++
> > drivers/target/iscsi/iscsi_target_erl0.h | 17 +
> > drivers/target/iscsi/iscsi_target_erl1.c | 1305 ++++++
> > drivers/target/iscsi/iscsi_target_erl1.h | 28 +
> > drivers/target/iscsi/iscsi_target_erl2.c | 483 ++
> > drivers/target/iscsi/iscsi_target_erl2.h | 18 +
> > drivers/target/iscsi/iscsi_target_login.c | 1245 +++++
> > drivers/target/iscsi/iscsi_target_login.h | 13 +
> > drivers/target/iscsi/iscsi_target_nego.c | 1068 +++++
> > drivers/target/iscsi/iscsi_target_nego.h | 19 +
>
> Why do we need to do the pre-nexus operations such as login,
> authentication, negotiation, etc in kernel?
>
Because defining an API for this type of userspace code is an
unnecessary overhead that iscsi_target_mod does not require. I really
don't want to add any type of C level interface/API and C userspace code
for fabric level login logic that is intended to be transparent to the
actual iscsi-target configfs control plane. It's just unnecessary
complexity and C userspace code for a kernel-level implementation with
no obvious benefit.
The area where some form of userspace passthrough specific to iSCSI
login negotiation in CURRENT_STAGE=0 would be useful is for userspace
authentication support beyond the built-in kernel-level CHAP code for
extending AuthMethod to support Kerberos, SRP, etc. This should be
limited to the AuthMethod state and keys, and the rest of fabric login
should still reside in-kernel.
--nab
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-23 10:22 ` Nicholas A. Bellinger
@ 2011-03-23 11:07 ` FUJITA Tomonori
2011-03-23 21:37 ` Nicholas A. Bellinger
0 siblings, 1 reply; 40+ messages in thread
From: FUJITA Tomonori @ 2011-03-23 11:07 UTC (permalink / raw)
To: nab
Cc: fujita.tomonori, linux-scsi, James.Bottomley, hch, michaelc, hare,
bharrosh
On Wed, 23 Mar 2011 03:22:06 -0700
"Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> On Wed, 2011-03-23 at 17:54 +0900, FUJITA Tomonori wrote:
> > I've not looked at the code yet but...
> >
> > On Tue, 22 Mar 2011 20:06:31 -0700
> > "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> >
> > > Nicholas Bellinger (14):
> > > iscsi: Resolve iscsi_proto.h naming conflicts with
> > > drivers/target/iscsi
> > > iscsi: Add Serial Number Arithmetic LT and GT into iscsi_proto.h
> > > target: Convert REPORT_LUNs to use int_to_scsilun
> > > iscsi-target: Add iSCSI fabric support for target v4
> > > iscsi-target: Add TCM v4 compatiable ConfigFS control plane
> > > iscsi-target: Add configfs fabric dependent statistics
> > > iscsi-target: Add TPG and Device logic
> > > iscsi-target: Add iSCSI Login Negotiation and Parameter logic
> > > iscsi-target: Add CHAP Authentication support using libcrypto
> > > iscsi-target: Add Sequence/PDU list + DataIN response logic
> > > iscsi-target: Add iSCSI Error Recovery Hierarchy support
> > > iscsi-target: Add support for task management operations
> > > iscsi-target: Add misc utility and debug logic
> > > iscsi-target: Add Makefile/Kconfig and update TCM top level
> > >
> > > drivers/infiniband/ulp/iser/iser_initiator.c | 2 +-
> > > drivers/scsi/be2iscsi/be_main.h | 4 +-
> > > drivers/scsi/bnx2i/bnx2i_hwi.c | 8 +-
> > > drivers/scsi/bnx2i/bnx2i_iscsi.c | 2 +-
> > > drivers/scsi/libiscsi.c | 22 +-
> > > drivers/target/Kconfig | 2 +
> > > drivers/target/Makefile | 3 +
> > > drivers/target/iscsi/Kconfig | 17 +
> > > drivers/target/iscsi/Makefile | 20 +
> > > drivers/target/iscsi/iscsi_target.c | 5017 +++++++++++++++++++++
> > > drivers/target/iscsi/iscsi_target.h | 32 +
> > > drivers/target/iscsi/iscsi_target_auth.c | 496 ++
> > > drivers/target/iscsi/iscsi_target_auth.h | 32 +
> > > drivers/target/iscsi/iscsi_target_configfs.c | 1914 ++++++++
> > > drivers/target/iscsi/iscsi_target_configfs.h | 7 +
> > > drivers/target/iscsi/iscsi_target_core.h | 881 ++++
> > > drivers/target/iscsi/iscsi_target_datain_values.c | 531 +++
> > > drivers/target/iscsi/iscsi_target_datain_values.h | 15 +
> > > drivers/target/iscsi/iscsi_target_debug.h | 113 +
> > > drivers/target/iscsi/iscsi_target_device.c | 108 +
> > > drivers/target/iscsi/iscsi_target_device.h | 9 +
> > > drivers/target/iscsi/iscsi_target_erl0.c | 1005 +++++
> > > drivers/target/iscsi/iscsi_target_erl0.h | 17 +
> > > drivers/target/iscsi/iscsi_target_erl1.c | 1305 ++++++
> > > drivers/target/iscsi/iscsi_target_erl1.h | 28 +
> > > drivers/target/iscsi/iscsi_target_erl2.c | 483 ++
> > > drivers/target/iscsi/iscsi_target_erl2.h | 18 +
> > > drivers/target/iscsi/iscsi_target_login.c | 1245 +++++
> > > drivers/target/iscsi/iscsi_target_login.h | 13 +
> > > drivers/target/iscsi/iscsi_target_nego.c | 1068 +++++
> > > drivers/target/iscsi/iscsi_target_nego.h | 19 +
> >
> > Why do we need to do the pre-nexus operations such as login,
> > authentication, negotiation, etc in kernel?
> >
>
> Because defining an API for this type of userspace code is an
> unnecessary overhead that iscsi_target_mod does not require. I really
> don't want to add any type of C level interface/API and C userspace code
> for fabric level login logic that is intended to be transparent to the
> actual iscsi-target configfs control plane. It's just unnecessary
> complexity and C userspace code for a kernel-level implementation with
> no obvious benefit.
Removing several thousand lines from kernel isn't benefit?
> The area where some form of userspace passthrough specific to iSCSI
> login negotiation in CURRENT_STAGE=0 would be useful is for userspace
> authentication support beyond the built-in kernel-level CHAP code for
> extending AuthMethod to support Kerberos, SRP, etc. This should be
> limited to the AuthMethod state and keys, and the rest of fabric login
> should still reside in-kernel.
I don't see what you mean. Why the negotiation should be in kernel?
In other words, what the approach to do the pre-nexus operations like
IET (or SCST) can't do?
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-23 11:07 ` FUJITA Tomonori
@ 2011-03-23 21:37 ` Nicholas A. Bellinger
2011-03-23 22:46 ` FUJITA Tomonori
0 siblings, 1 reply; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 21:37 UTC (permalink / raw)
To: FUJITA Tomonori
Cc: linux-scsi, James.Bottomley, hch, michaelc, hare, bharrosh
On Wed, 2011-03-23 at 20:07 +0900, FUJITA Tomonori wrote:
> On Wed, 23 Mar 2011 03:22:06 -0700
> "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
>
> > On Wed, 2011-03-23 at 17:54 +0900, FUJITA Tomonori wrote:
> > > I've not looked at the code yet but...
> > >
> > > On Tue, 22 Mar 2011 20:06:31 -0700
> > > "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> > >
> > > > Nicholas Bellinger (14):
> > > > iscsi: Resolve iscsi_proto.h naming conflicts with
> > > > drivers/target/iscsi
> > > > iscsi: Add Serial Number Arithmetic LT and GT into iscsi_proto.h
> > > > target: Convert REPORT_LUNs to use int_to_scsilun
> > > > iscsi-target: Add iSCSI fabric support for target v4
> > > > iscsi-target: Add TCM v4 compatiable ConfigFS control plane
> > > > iscsi-target: Add configfs fabric dependent statistics
> > > > iscsi-target: Add TPG and Device logic
> > > > iscsi-target: Add iSCSI Login Negotiation and Parameter logic
> > > > iscsi-target: Add CHAP Authentication support using libcrypto
> > > > iscsi-target: Add Sequence/PDU list + DataIN response logic
> > > > iscsi-target: Add iSCSI Error Recovery Hierarchy support
> > > > iscsi-target: Add support for task management operations
> > > > iscsi-target: Add misc utility and debug logic
> > > > iscsi-target: Add Makefile/Kconfig and update TCM top level
> > > >
> > > > drivers/infiniband/ulp/iser/iser_initiator.c | 2 +-
> > > > drivers/scsi/be2iscsi/be_main.h | 4 +-
> > > > drivers/scsi/bnx2i/bnx2i_hwi.c | 8 +-
> > > > drivers/scsi/bnx2i/bnx2i_iscsi.c | 2 +-
> > > > drivers/scsi/libiscsi.c | 22 +-
> > > > drivers/target/Kconfig | 2 +
> > > > drivers/target/Makefile | 3 +
> > > > drivers/target/iscsi/Kconfig | 17 +
> > > > drivers/target/iscsi/Makefile | 20 +
> > > > drivers/target/iscsi/iscsi_target.c | 5017 +++++++++++++++++++++
> > > > drivers/target/iscsi/iscsi_target.h | 32 +
> > > > drivers/target/iscsi/iscsi_target_auth.c | 496 ++
> > > > drivers/target/iscsi/iscsi_target_auth.h | 32 +
> > > > drivers/target/iscsi/iscsi_target_configfs.c | 1914 ++++++++
> > > > drivers/target/iscsi/iscsi_target_configfs.h | 7 +
> > > > drivers/target/iscsi/iscsi_target_core.h | 881 ++++
> > > > drivers/target/iscsi/iscsi_target_datain_values.c | 531 +++
> > > > drivers/target/iscsi/iscsi_target_datain_values.h | 15 +
> > > > drivers/target/iscsi/iscsi_target_debug.h | 113 +
> > > > drivers/target/iscsi/iscsi_target_device.c | 108 +
> > > > drivers/target/iscsi/iscsi_target_device.h | 9 +
> > > > drivers/target/iscsi/iscsi_target_erl0.c | 1005 +++++
> > > > drivers/target/iscsi/iscsi_target_erl0.h | 17 +
> > > > drivers/target/iscsi/iscsi_target_erl1.c | 1305 ++++++
> > > > drivers/target/iscsi/iscsi_target_erl1.h | 28 +
> > > > drivers/target/iscsi/iscsi_target_erl2.c | 483 ++
> > > > drivers/target/iscsi/iscsi_target_erl2.h | 18 +
> > > > drivers/target/iscsi/iscsi_target_login.c | 1245 +++++
> > > > drivers/target/iscsi/iscsi_target_login.h | 13 +
> > > > drivers/target/iscsi/iscsi_target_nego.c | 1068 +++++
> > > > drivers/target/iscsi/iscsi_target_nego.h | 19 +
> > >
> > > Why do we need to do the pre-nexus operations such as login,
> > > authentication, negotiation, etc in kernel?
> > >
> >
> > Because defining an API for this type of userspace code is an
> > unnecessary overhead that iscsi_target_mod does not require. I really
> > don't want to add any type of C level interface/API and C userspace code
> > for fabric level login logic that is intended to be transparent to the
> > actual iscsi-target configfs control plane. It's just unnecessary
> > complexity and C userspace code for a kernel-level implementation with
> > no obvious benefit.
>
> Removing several thousand lines from kernel isn't benefit?
>
>
No, not when it adds complexity without benefit for the default case and
introduces unnecessary userspace C code.
At this point in lio-utils.git for target v4 code, we require zero C
userspace code for the default operation of iscsi-target, and I want to
keep it this way.
Maintaining backwards compatibility with interpreted script code for the
fabric control plane makes life so much eaiser than any kernel <-> user
C API that I have ever encountered.
> > The area where some form of userspace passthrough specific to iSCSI
> > login negotiation in CURRENT_STAGE=0 would be useful is for userspace
> > authentication support beyond the built-in kernel-level CHAP code for
> > extending AuthMethod to support Kerberos, SRP, etc. This should be
> > limited to the AuthMethod state and keys, and the rest of fabric login
> > should still reside in-kernel.
>
> I don't see what you mean. Why the negotiation should be in kernel?
>
> In other words, what the approach to do the pre-nexus operations like
> IET (or SCST) can't do?
Because it does not makes sense for the default case, and adds
unnecessary complexity to the iscsi-target login process.
Using mainline libcrypto for CHAP works as expected and allows us to
provide one-way and mutual authentication for iSCSI discovery and
explict initiator NodeACLs via configfs attributes without any userspace
C dependcies.
Like I said, I don't have an issue with adding some form of userspace
passthrough in iscsi_target_auth.c:iscsi_handle_authentication() in
order to support other types of iSCSI authentication methods. But
changing iscsi-target to require that every single iSCSI login to every
iSCSI target endpoint needs to go through a userspace daemon, we when
can currently configure 10K -> 100K -> and upwards of individual
/sys/kernel/config/target/iscsi/$TARGETNAME/ endpoints, this seems like
an unnecessary bottleneck to me.
--nab
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-23 21:37 ` Nicholas A. Bellinger
@ 2011-03-23 22:46 ` FUJITA Tomonori
2011-03-23 23:28 ` Nicholas A. Bellinger
0 siblings, 1 reply; 40+ messages in thread
From: FUJITA Tomonori @ 2011-03-23 22:46 UTC (permalink / raw)
To: nab
Cc: fujita.tomonori, linux-scsi, James.Bottomley, hch, michaelc, hare,
bharrosh
On Wed, 23 Mar 2011 14:37:34 -0700
"Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> > Removing several thousand lines from kernel isn't benefit?
> >
> >
>
> No, not when it adds complexity without benefit for the default case and
> introduces unnecessary userspace C code.
>
> At this point in lio-utils.git for target v4 code, we require zero C
> userspace code for the default operation of iscsi-target, and I want to
> keep it this way.
Hmm, you prefer 'zero C userspace code'? On the other hand, you insist
that requiring user Python userspace code to set up ibmvscsis is a
better approach even when the kernel space can set up everything for
ibmvscsis?
I can't understand your logic.
And iSCSI setup usually needs userspace code such as iSNS
anyway. You'll add iSNS to kernel too.
> Maintaining backwards compatibility with interpreted script code for the
> fabric control plane makes life so much eaiser than any kernel <-> user
> C API that I have ever encountered.
I think that the target can use open-iscsi's kernel <-> user
interface. All the userspace code needs is passing negotiated iSCSI
parameters. Once a nexus establishes, the user space will never do
anything for it. So no complicity like open-iscsi.
> > In other words, what the approach to do the pre-nexus operations like
> > IET (or SCST) can't do?
>
> Because it does not makes sense for the default case, and adds
> unnecessary complexity to the iscsi-target login process.
Well, I don't think so.
> Using mainline libcrypto for CHAP works as expected and allows us to
> provide one-way and mutual authentication for iSCSI discovery and
> explict initiator NodeACLs via configfs attributes without any userspace
> C dependcies.
The user space daemon can read configfs easily to do the same thing.
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-23 22:46 ` FUJITA Tomonori
@ 2011-03-23 23:28 ` Nicholas A. Bellinger
2011-03-24 1:29 ` FUJITA Tomonori
0 siblings, 1 reply; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-23 23:28 UTC (permalink / raw)
To: FUJITA Tomonori
Cc: linux-scsi, James.Bottomley, hch, michaelc, hare, bharrosh
On Thu, 2011-03-24 at 07:46 +0900, FUJITA Tomonori wrote:
> On Wed, 23 Mar 2011 14:37:34 -0700
> "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
>
> > > Removing several thousand lines from kernel isn't benefit?
> > >
> > >
> >
> > No, not when it adds complexity without benefit for the default case and
> > introduces unnecessary userspace C code.
> >
> > At this point in lio-utils.git for target v4 code, we require zero C
> > userspace code for the default operation of iscsi-target, and I want to
> > keep it this way.
>
> Hmm, you prefer 'zero C userspace code'? On the other hand, you insist
> that requiring user Python userspace code to set up ibmvscsis is a
> better approach even when the kernel space can set up everything for
> ibmvscsis?
>
Yes, because we expect userspace to drive creation of the current
configfs group layout.
The configfs maintainer explictly requested to drop the ability to drive
the config_group creation from kernel-space that I had implemented
originally, and which has not been included in the mainline target v4.0
control plane.
It's folks like Joel and Greg-KH that need to be convinced in order for
me to consider this type of logic for for mainline target code. I do
agree some modules like ibmviscsis are a very good canidate to revisit
the discussion for something like this. However I still think we can
achieve the same results with more flexiable python library code that
provides an API to programmers to drive the underlying target configfs
control plane, than trying to add this logic direct in kernel code,
regardless of fabric module type.
> I can't understand your logic.
>
> And iSCSI setup usually needs userspace code such as iSNS
> anyway. You'll add iSNS to kernel too.
>
iSNS should be walking the /sys/kernel/config/target/iscsi layout, and
should not require any kernel code at all.
>
> > Maintaining backwards compatibility with interpreted script code for the
> > fabric control plane makes life so much eaiser than any kernel <-> user
> > C API that I have ever encountered.
>
> I think that the target can use open-iscsi's kernel <-> user
> interface. All the userspace code needs is passing negotiated iSCSI
> parameters. Once a nexus establishes, the user space will never do
> anything for it. So no complicity like open-iscsi.
>
What about for the boot1.kernel.org cases where we can expect 1000s of
R/O clients in the future to a proper 10 Gb/sec uplink.
Why do all of these types of logins need to talk with a userspace login
queue interface? Why do we need to worry about an interface in the
first place for the standard iSCSI login case..? What happens if this
daemon is unexpectedly killed..? Do I now have to worry about
implementing a multi-threaded userspace daemon to handle the high login
volume public iSCSI target case..?
Can we please just use iSCSI $TARGET_WWN endpoint specific kernel-level
login for high volume bko type operation, and allow other TPG contexts
to use an userspace passthrough for the other optional authentication
cases from RFC-3720 that we are not going to support in-kernel..?
>
> > > In other words, what the approach to do the pre-nexus operations like
> > > IET (or SCST) can't do?
> >
> > Because it does not makes sense for the default case, and adds
> > unnecessary complexity to the iscsi-target login process.
>
> Well, I don't think so.
>
>
> > Using mainline libcrypto for CHAP works as expected and allows us to
> > provide one-way and mutual authentication for iSCSI discovery and
> > explict initiator NodeACLs via configfs attributes without any userspace
> > C dependcies.
>
> The user space daemon can read configfs easily to do the same thing.
But reading the of the configfs layout is not the problem, yes..?
It's the synchronization between a kernel and userspace daemon for every
login attempt that does not require standard or disabled authentication,
and having to use another interface for this and worry about all of the
extra things that come with this type of design.
iscsi_target_mod has never required this to function, and I still don't
see a reason why we should enforce a hard-requirement to do this for all
iSCSI login cases.
--nab
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-23 23:28 ` Nicholas A. Bellinger
@ 2011-03-24 1:29 ` FUJITA Tomonori
2011-03-24 6:59 ` Nicholas A. Bellinger
0 siblings, 1 reply; 40+ messages in thread
From: FUJITA Tomonori @ 2011-03-24 1:29 UTC (permalink / raw)
To: nab
Cc: fujita.tomonori, linux-scsi, James.Bottomley, hch, michaelc, hare,
bharrosh
On Wed, 23 Mar 2011 16:28:37 -0700
"Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> > Hmm, you prefer 'zero C userspace code'? On the other hand, you insist
> > that requiring user Python userspace code to set up ibmvscsis is a
> > better approach even when the kernel space can set up everything for
> > ibmvscsis?
> >
>
> Yes, because we expect userspace to drive creation of the current
> configfs group layout.
>
> The configfs maintainer explictly requested to drop the ability to drive
> the config_group creation from kernel-space that I had implemented
> originally, and which has not been included in the mainline target v4.0
> control plane.
>
> It's folks like Joel and Greg-KH that need to be convinced in order for
> me to consider this type of logic for for mainline target code. I do
Can you stop insisting that configfs can't do that so the target core
can't do that?
We should think about what is the best design for the target code
first.
> > And iSCSI setup usually needs userspace code such as iSNS
> > anyway. You'll add iSNS to kernel too.
> >
>
> iSNS should be walking the /sys/kernel/config/target/iscsi layout, and
> should not require any kernel code at all.
How you draw the line? Why iSNS can live in user space?
For example, you say that iSNS should live in user space. The similar
feature to sending the list of target lives in kernel space.
> What about for the boot1.kernel.org cases where we can expect 1000s of
> R/O clients in the future to a proper 10 Gb/sec uplink.
>
> Why do all of these types of logins need to talk with a userspace login
> queue interface?
I prefer less kernel space code.
> Why do we need to worry about an interface in the
> first place for the standard iSCSI login case..? What happens if this
> daemon is unexpectedly killed..? Do I now have to worry about
Can you stop such pointless argument? What happens if the iscsi target
kernel module crashed?
Linux systems already depend on some essential user space daemons. We
know how to deal with them.
> implementing a multi-threaded userspace daemon to handle the high login
> volume public iSCSI target case..?
I'm pretty sure that even single-threaded userspace deamon can handle
iSCSI login operations.
> > The user space daemon can read configfs easily to do the same thing.
>
> But reading the of the configfs layout is not the problem, yes..?
I don't think so. The user space doesn't cache the configfs info. So
the rule is pretty simple.
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-24 1:29 ` FUJITA Tomonori
@ 2011-03-24 6:59 ` Nicholas A. Bellinger
2011-03-25 0:18 ` James Bottomley
2011-03-27 10:02 ` FUJITA Tomonori
0 siblings, 2 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-24 6:59 UTC (permalink / raw)
To: FUJITA Tomonori
Cc: linux-scsi, James.Bottomley, hch, michaelc, hare, bharrosh
On Thu, 2011-03-24 at 10:29 +0900, FUJITA Tomonori wrote:
> On Wed, 23 Mar 2011 16:28:37 -0700
> "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
>
> > > Hmm, you prefer 'zero C userspace code'? On the other hand, you insist
> > > that requiring user Python userspace code to set up ibmvscsis is a
> > > better approach even when the kernel space can set up everything for
> > > ibmvscsis?
> > >
> >
> > Yes, because we expect userspace to drive creation of the current
> > configfs group layout.
> >
> > The configfs maintainer explictly requested to drop the ability to drive
> > the config_group creation from kernel-space that I had implemented
> > originally, and which has not been included in the mainline target v4.0
> > control plane.
> >
> > It's folks like Joel and Greg-KH that need to be convinced in order for
> > me to consider this type of logic for for mainline target code. I do
>
> Can you stop insisting that configfs can't do that so the target core
> can't do that?
>
That's certainly not what I am insisting upon. I am trying to explain
that we currently have a configfs control plane that is native for
target mode.
In the last three years to get to this point, I have developed prototype
code and asked upstream review for:
*) Creating logic to drive configuration from kernel-space for configfs
*) Tried at creating sysfs -> configfs symlinks for referencing target
core backend devices
So far both of these attempts at patches have been firmly rejected by
the configfs and sysfs maintainers.
I am not saying that configfs is the end-all for all generic target mode
control plane, but what I am saying is that I have yet to see the code
for a different control plane that makes me want to move the away from a
'native configfs' to 'a configfs/sysfs hyrbid', or something else all
together for mainline target code.
> We should think about what is the best design for the target code
> first.
>
> > > And iSCSI setup usually needs userspace code such as iSNS
> > > anyway. You'll add iSNS to kernel too.
> > >
> >
> > iSNS should be walking the /sys/kernel/config/target/iscsi layout, and
> > should not require any kernel code at all.
>
> How you draw the line? Why iSNS can live in user space?
>
> For example, you say that iSNS should live in user space. The similar
> feature to sending the list of target lives in kernel space.
>
>
iSNS is a seperate protocol, and there is no hard-requirement of iSNS
client logic to be in place in order for basic iSCSI login to function.
> > What about for the boot1.kernel.org cases where we can expect 1000s of
> > R/O clients in the future to a proper 10 Gb/sec uplink.
> >
> > Why do all of these types of logins need to talk with a userspace login
> > queue interface?
>
> I prefer less kernel space code.
>
>
> > Why do we need to worry about an interface in the
> > first place for the standard iSCSI login case..? What happens if this
> > daemon is unexpectedly killed..? Do I now have to worry about
>
> Can you stop such pointless argument? What happens if the iscsi target
> kernel module crashed?
>
> Linux systems already depend on some essential user space daemons. We
> know how to deal with them.
>
>
Sorry, but I have no interest in adding the extra complexity to handle
this for the default case.
> > implementing a multi-threaded userspace daemon to handle the high login
> > volume public iSCSI target case..?
>
> I'm pretty sure that even single-threaded userspace deamon can handle
> iSCSI login operations.
>
No, I don't want a single threaded userspace interface for handling
iSCSI login logic for default operation. This is an unnecessary
bottleneck.
>
> > > The user space daemon can read configfs easily to do the same thing.
> >
> > But reading the of the configfs layout is not the problem, yes..?
>
> I don't think so. The user space doesn't cache the configfs info. So
> the rule is pretty simple.
It's more complicated than that. We have to then start referencing
things like struct se_node_acl usage to userspace when locating an
explict NodeACL during login. What happens if the explict NodeACL is
removed from configfs context while userspace code is running..? What
happens if the entire TPG is removed during an active userspace login..?
It all has to be sychronized with a userspace daemon through a single
thread that needs to know about the network portals, TPG layout and
explict NodeACls in order to complete a iSCSI login.
I do not want to deal with any global userspace sychronization for the
default iSCSI login case, especially for high volume public demo-mode
access. This only adds unnecessary complexity to the current
iscsi-target code, and I have no interest to make this type of
fundemental change to iscsi-target design at this point without a strong
practical benefit for existing code. I need a signficant reason to drop
everyhing that has been functioning for years and re-invent the wheel,
again, in userspace. I have not yet to heard a convincing case for
doing any of this.
--nab
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-24 6:59 ` Nicholas A. Bellinger
@ 2011-03-25 0:18 ` James Bottomley
2011-03-25 11:31 ` Bart Van Assche
2011-03-25 21:14 ` Nicholas A. Bellinger
2011-03-27 10:02 ` FUJITA Tomonori
1 sibling, 2 replies; 40+ messages in thread
From: James Bottomley @ 2011-03-25 0:18 UTC (permalink / raw)
To: Nicholas A. Bellinger
Cc: FUJITA Tomonori, linux-scsi, hch, michaelc, hare, bharrosh
On Wed, 2011-03-23 at 23:59 -0700, Nicholas A. Bellinger wrote:
> On Thu, 2011-03-24 at 10:29 +0900, FUJITA Tomonori wrote:
> > On Wed, 23 Mar 2011 16:28:37 -0700
> > "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> >
> > > > Hmm, you prefer 'zero C userspace code'? On the other hand, you insist
> > > > that requiring user Python userspace code to set up ibmvscsis is a
> > > > better approach even when the kernel space can set up everything for
> > > > ibmvscsis?
> > > >
> > >
> > > Yes, because we expect userspace to drive creation of the current
> > > configfs group layout.
> > >
> > > The configfs maintainer explictly requested to drop the ability to drive
> > > the config_group creation from kernel-space that I had implemented
> > > originally, and which has not been included in the mainline target v4.0
> > > control plane.
> > >
> > > It's folks like Joel and Greg-KH that need to be convinced in order for
> > > me to consider this type of logic for for mainline target code. I do
> >
> > Can you stop insisting that configfs can't do that so the target core
> > can't do that?
> >
>
> That's certainly not what I am insisting upon. I am trying to explain
> that we currently have a configfs control plane that is native for
> target mode.
>
> In the last three years to get to this point, I have developed prototype
> code and asked upstream review for:
>
> *) Creating logic to drive configuration from kernel-space for configfs
> *) Tried at creating sysfs -> configfs symlinks for referencing target
> core backend devices
>
> So far both of these attempts at patches have been firmly rejected by
> the configfs and sysfs maintainers.
>
> I am not saying that configfs is the end-all for all generic target mode
> control plane, but what I am saying is that I have yet to see the code
> for a different control plane that makes me want to move the away from a
> 'native configfs' to 'a configfs/sysfs hyrbid', or something else all
> together for mainline target code.
OK, so what about an upcall to userspace to create the necessary
directories? That could be driven by the kernel and still not require
any implementation in configfs.
> > We should think about what is the best design for the target code
> > first.
> >
> > > > And iSCSI setup usually needs userspace code such as iSNS
> > > > anyway. You'll add iSNS to kernel too.
> > > >
> > >
> > > iSNS should be walking the /sys/kernel/config/target/iscsi layout, and
> > > should not require any kernel code at all.
> >
> > How you draw the line? Why iSNS can live in user space?
> >
> > For example, you say that iSNS should live in user space. The similar
> > feature to sending the list of target lives in kernel space.
> >
> >
>
> iSNS is a seperate protocol, and there is no hard-requirement of iSNS
> client logic to be in place in order for basic iSCSI login to function.
>
> > > What about for the boot1.kernel.org cases where we can expect 1000s of
> > > R/O clients in the future to a proper 10 Gb/sec uplink.
> > >
> > > Why do all of these types of logins need to talk with a userspace login
> > > queue interface?
> >
> > I prefer less kernel space code.
> >
> >
> > > Why do we need to worry about an interface in the
> > > first place for the standard iSCSI login case..? What happens if this
> > > daemon is unexpectedly killed..? Do I now have to worry about
> >
> > Can you stop such pointless argument? What happens if the iscsi target
> > kernel module crashed?
> >
> > Linux systems already depend on some essential user space daemons. We
> > know how to deal with them.
> >
> >
>
> Sorry, but I have no interest in adding the extra complexity to handle
> this for the default case.
Look at it this way: there's no way an implementation of all the RFC3720
authentication methods (let alone the extensions) is going into the
kernel, so you need a user space interface anyway. Once you have one,
creating two paths: one in-kernel and one via upcall just doubles the
maintenance load and the amount of work that has to be done fixing bugs.
As far as I can see, authentication isn't in the fast path, so there's
no real need for any of it to be in-kernel in the first place.
James
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 0:18 ` James Bottomley
@ 2011-03-25 11:31 ` Bart Van Assche
2011-03-25 15:06 ` James Bottomley
2011-03-25 21:14 ` Nicholas A. Bellinger
1 sibling, 1 reply; 40+ messages in thread
From: Bart Van Assche @ 2011-03-25 11:31 UTC (permalink / raw)
To: James Bottomley
Cc: Nicholas A. Bellinger, FUJITA Tomonori, linux-scsi, hch, michaelc,
hare, bharrosh
On Fri, Mar 25, 2011 at 1:18 AM, James Bottomley
<James.Bottomley@suse.de> wrote:
> On Wed, 2011-03-23 at 23:59 -0700, Nicholas A. Bellinger wrote:
>> On Thu, 2011-03-24 at 10:29 +0900, FUJITA Tomonori wrote:
>> > On Wed, 23 Mar 2011 16:28:37 -0700
>> > "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
>> >
>> > > > Hmm, you prefer 'zero C userspace code'? On the other hand, you insist
>> > > > that requiring user Python userspace code to set up ibmvscsis is a
>> > > > better approach even when the kernel space can set up everything for
>> > > > ibmvscsis?
>> > > >
>> > >
>> > > Yes, because we expect userspace to drive creation of the current
>> > > configfs group layout.
>> > >
>> > > The configfs maintainer explictly requested to drop the ability to drive
>> > > the config_group creation from kernel-space that I had implemented
>> > > originally, and which has not been included in the mainline target v4.0
>> > > control plane.
>> > >
>> > > It's folks like Joel and Greg-KH that need to be convinced in order for
>> > > me to consider this type of logic for for mainline target code. I do
>> >
>> > Can you stop insisting that configfs can't do that so the target core
>> > can't do that?
>> >
>>
>> That's certainly not what I am insisting upon. I am trying to explain
>> that we currently have a configfs control plane that is native for
>> target mode.
>>
>> In the last three years to get to this point, I have developed prototype
>> code and asked upstream review for:
>>
>> *) Creating logic to drive configuration from kernel-space for configfs
>> *) Tried at creating sysfs -> configfs symlinks for referencing target
>> core backend devices
>>
>> So far both of these attempts at patches have been firmly rejected by
>> the configfs and sysfs maintainers.
>>
>> I am not saying that configfs is the end-all for all generic target mode
>> control plane, but what I am saying is that I have yet to see the code
>> for a different control plane that makes me want to move the away from a
>> 'native configfs' to 'a configfs/sysfs hyrbid', or something else all
>> together for mainline target code.
>
> OK, so what about an upcall to userspace to create the necessary
> directories? That could be driven by the kernel and still not require
> any implementation in configfs.
Hi James,
I might have missed something, but which upcall mechanism are you
referring to ? Personally I'm not fond of the upcall concept because
as far as I can see any synchronous upcall mechanism can potentially
be used to trigger lock inversions not detectable by the PROVE_LOCKING
mechanism.
Regarding kernel-space driven directory creation in configfs: I have
been wondering whether it is possible to implement any configuration
filesystem such that directories can be created synchronously from
kernel space without triggering lock inversion. I don't see this as a
configfs limitation but as an inherent limitation of a configuration
filesystem. In a similar way a self-declarative kernel interface like
sysfs has the limitation that it is not possible to add configfs-style
configuration functionality without triggering lock inversion. Both
sysfs and configfs have important advantages over mechanisms like
ioctl() and netlink but there are some disadvantages too.
Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 11:31 ` Bart Van Assche
@ 2011-03-25 15:06 ` James Bottomley
2011-03-25 18:44 ` Bart Van Assche
0 siblings, 1 reply; 40+ messages in thread
From: James Bottomley @ 2011-03-25 15:06 UTC (permalink / raw)
To: Bart Van Assche
Cc: Nicholas A. Bellinger, FUJITA Tomonori, linux-scsi, hch, michaelc,
hare, bharrosh
On Fri, 2011-03-25 at 12:31 +0100, Bart Van Assche wrote:
> On Fri, Mar 25, 2011 at 1:18 AM, James Bottomley
> <James.Bottomley@suse.de> wrote:
> >> OK, so what about an upcall to userspace to create the necessary
> > directories? That could be driven by the kernel and still not require
> > any implementation in configfs.
>
> Hi James,
>
> I might have missed something, but which upcall mechanism are you
> referring to ? Personally I'm not fond of the upcall concept because
> as far as I can see any synchronous upcall mechanism can potentially
> be used to trigger lock inversions not detectable by the PROVE_LOCKING
> mechanism.
Any ... we have relay, netlink, uevent etc.
I don't see how there could be a lock inversion. Upcalls are by
definition asynchronous.
The use case for ibmvscsi, which seems the most pressing, is simply that
on load it unpacks the config information, does the upcall and exits.
The daemon creates the necessary directories with the information and
the vscsi interface is functional once everything is set up. There's no
locking problems in that use case.
> Regarding kernel-space driven directory creation in configfs: I have
> been wondering whether it is possible to implement any configuration
> filesystem such that directories can be created synchronously from
> kernel space without triggering lock inversion. I don't see this as a
> configfs limitation but as an inherent limitation of a configuration
> filesystem. In a similar way a self-declarative kernel interface like
> sysfs has the limitation that it is not possible to add configfs-style
> configuration functionality without triggering lock inversion. Both
> sysfs and configfs have important advantages over mechanisms like
> ioctl() and netlink but there are some disadvantages too.
Explain first what lock inversion problems you see ... those usually
only happen if you have the in the kernel upcall thread waiting for
completion.
James
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 15:06 ` James Bottomley
@ 2011-03-25 18:44 ` Bart Van Assche
2011-03-25 19:34 ` James Bottomley
0 siblings, 1 reply; 40+ messages in thread
From: Bart Van Assche @ 2011-03-25 18:44 UTC (permalink / raw)
To: James Bottomley
Cc: Nicholas A. Bellinger, FUJITA Tomonori, linux-scsi, hch, michaelc,
hare, bharrosh
On Fri, Mar 25, 2011 at 4:06 PM, James Bottomley
<James.Bottomley@suse.de> wrote:
> Explain first what lock inversion problems you see ... those usually
> only happen if you have the in the kernel upcall thread waiting for
> completion.
In this context asynchronous upcalls have their problems too: the
in-kernel and the user space visible state may become inconsistent.
What will happen if processing of such notifications is delayed and
concurrently attempts are made from user space to act on these objects
? I'm afraid that could result in horrible race conditions.
Bart.
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 18:44 ` Bart Van Assche
@ 2011-03-25 19:34 ` James Bottomley
2011-03-25 19:49 ` Bart Van Assche
0 siblings, 1 reply; 40+ messages in thread
From: James Bottomley @ 2011-03-25 19:34 UTC (permalink / raw)
To: Bart Van Assche
Cc: Nicholas A. Bellinger, FUJITA Tomonori, linux-scsi, hch, michaelc,
hare, bharrosh
On Fri, 2011-03-25 at 19:44 +0100, Bart Van Assche wrote:
> On Fri, Mar 25, 2011 at 4:06 PM, James Bottomley
> <James.Bottomley@suse.de> wrote:
> > Explain first what lock inversion problems you see ... those usually
> > only happen if you have the in the kernel upcall thread waiting for
> > completion.
>
> In this context asynchronous upcalls have their problems too: the
> in-kernel and the user space visible state may become inconsistent.
> What will happen if processing of such notifications is delayed and
> concurrently attempts are made from user space to act on these objects
> ? I'm afraid that could result in horrible race conditions.
What race conditions? The current exemplar is udev which uses the
uevent mechanism. The usual way of coping with use before init is to
key visibility off completion ... at least that's the way we do it in
sysfs.
James
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 19:34 ` James Bottomley
@ 2011-03-25 19:49 ` Bart Van Assche
2011-03-25 19:54 ` James Bottomley
0 siblings, 1 reply; 40+ messages in thread
From: Bart Van Assche @ 2011-03-25 19:49 UTC (permalink / raw)
To: James Bottomley
Cc: Nicholas A. Bellinger, FUJITA Tomonori, linux-scsi, hch, michaelc,
hare, bharrosh
On Fri, Mar 25, 2011 at 8:34 PM, James Bottomley
<James.Bottomley@suse.de> wrote:
> On Fri, 2011-03-25 at 19:44 +0100, Bart Van Assche wrote:
>> On Fri, Mar 25, 2011 at 4:06 PM, James Bottomley
>> <James.Bottomley@suse.de> wrote:
>> > Explain first what lock inversion problems you see ... those usually
>> > only happen if you have the in the kernel upcall thread waiting for
>> > completion.
>>
>> In this context asynchronous upcalls have their problems too: the
>> in-kernel and the user space visible state may become inconsistent.
>> What will happen if processing of such notifications is delayed and
>> concurrently attempts are made from user space to act on these objects
>> ? I'm afraid that could result in horrible race conditions.
>
> What race conditions? The current exemplar is udev which uses the
> uevent mechanism. The usual way of coping with use before init is to
> key visibility off completion ... at least that's the way we do it in
> sysfs.
A storage target subsystem is more demanding than e.g. udev. As an
example, for the session reassignment logic in SCST it is necessary to
remove and create user space visible objects synchronously and under
lock. Both are necessary - the locking and the synchronous
manipulation of user space objects. I'm not sure it is possible to
convert such an approach to something based on asynchronous userspace
upcalls.
Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 19:49 ` Bart Van Assche
@ 2011-03-25 19:54 ` James Bottomley
2011-03-25 20:05 ` Bart Van Assche
0 siblings, 1 reply; 40+ messages in thread
From: James Bottomley @ 2011-03-25 19:54 UTC (permalink / raw)
To: Bart Van Assche
Cc: Nicholas A. Bellinger, FUJITA Tomonori, linux-scsi, hch, michaelc,
hare, bharrosh
On Fri, 2011-03-25 at 20:49 +0100, Bart Van Assche wrote:
> On Fri, Mar 25, 2011 at 8:34 PM, James Bottomley
> <James.Bottomley@suse.de> wrote:
> > On Fri, 2011-03-25 at 19:44 +0100, Bart Van Assche wrote:
> >> On Fri, Mar 25, 2011 at 4:06 PM, James Bottomley
> >> <James.Bottomley@suse.de> wrote:
> >> > Explain first what lock inversion problems you see ... those usually
> >> > only happen if you have the in the kernel upcall thread waiting for
> >> > completion.
> >>
> >> In this context asynchronous upcalls have their problems too: the
> >> in-kernel and the user space visible state may become inconsistent.
> >> What will happen if processing of such notifications is delayed and
> >> concurrently attempts are made from user space to act on these objects
> >> ? I'm afraid that could result in horrible race conditions.
> >
> > What race conditions? The current exemplar is udev which uses the
> > uevent mechanism. The usual way of coping with use before init is to
> > key visibility off completion ... at least that's the way we do it in
> > sysfs.
>
> A storage target subsystem is more demanding than e.g. udev. As an
> example, for the session reassignment logic in SCST it is necessary to
> remove and create user space visible objects synchronously and under
> lock. Both are necessary - the locking and the synchronous
> manipulation of user space objects. I'm not sure it is possible to
> convert such an approach to something based on asynchronous userspace
> upcalls.
Really, no ... this is an example of an a-priori design fault. What it
implies is that the user and kernel share locks for state access, which
always leads to deadlocks. A correctly separated event driven model
shouldn't need state locks across event subsystems.
Just because SCST couldn't get it right, doesn't mean that no-one else
can.
James
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 19:54 ` James Bottomley
@ 2011-03-25 20:05 ` Bart Van Assche
0 siblings, 0 replies; 40+ messages in thread
From: Bart Van Assche @ 2011-03-25 20:05 UTC (permalink / raw)
To: James Bottomley
Cc: Nicholas A. Bellinger, FUJITA Tomonori, linux-scsi, hch, michaelc,
hare, bharrosh
On Fri, Mar 25, 2011 at 8:54 PM, James Bottomley
<James.Bottomley@suse.de> wrote:
> On Fri, 2011-03-25 at 20:49 +0100, Bart Van Assche wrote:
>> A storage target subsystem is more demanding than e.g. udev. As an
>> example, for the session reassignment logic in SCST it is necessary to
>> remove and create user space visible objects synchronously and under
>> lock. Both are necessary - the locking and the synchronous
>> manipulation of user space objects. I'm not sure it is possible to
>> convert such an approach to something based on asynchronous userspace
>> upcalls.
>
> Really, no ... this is an example of an a-priori design fault. What it
> implies is that the user and kernel share locks for state access, which
> always leads to deadlocks. A correctly separated event driven model
> shouldn't need state locks across event subsystems.
>
> Just because SCST couldn't get it right, doesn't mean that no-one else
> can.
Hi James,
There seems to be a misunderstanding. There is no design fault at all
in SCST in this area - since SCST uses sysfs the session reassignment
code doesn't trigger any lock inversion at all. Such lock inversion
would only be triggered if someone would try to convert that code to
configfs.
Also, the reason I mentioned that example was not to invite criticism
on that specific example but to illustrate that in a storage target
some algorithms can only be implemented correctly by synchronously
updating the internal state and the state of a self-describing
interface like sysfs or configfs. It's not because udev doesn't need
this that no other subsystem needs this.
Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 0:18 ` James Bottomley
2011-03-25 11:31 ` Bart Van Assche
@ 2011-03-25 21:14 ` Nicholas A. Bellinger
2011-03-27 10:02 ` FUJITA Tomonori
1 sibling, 1 reply; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-25 21:14 UTC (permalink / raw)
To: James Bottomley
Cc: FUJITA Tomonori, linux-scsi, hch, michaelc, hare, bharrosh
On Thu, 2011-03-24 at 19:18 -0500, James Bottomley wrote:
> On Wed, 2011-03-23 at 23:59 -0700, Nicholas A. Bellinger wrote:
> > On Thu, 2011-03-24 at 10:29 +0900, FUJITA Tomonori wrote:
> > > On Wed, 23 Mar 2011 16:28:37 -0700
> > > "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> > >
> > > > > Hmm, you prefer 'zero C userspace code'? On the other hand, you insist
> > > > > that requiring user Python userspace code to set up ibmvscsis is a
> > > > > better approach even when the kernel space can set up everything for
> > > > > ibmvscsis?
> > > > >
> > > >
> > > > Yes, because we expect userspace to drive creation of the current
> > > > configfs group layout.
> > > >
> > > > The configfs maintainer explictly requested to drop the ability to drive
> > > > the config_group creation from kernel-space that I had implemented
> > > > originally, and which has not been included in the mainline target v4.0
> > > > control plane.
> > > >
> > > > It's folks like Joel and Greg-KH that need to be convinced in order for
> > > > me to consider this type of logic for for mainline target code. I do
> > >
> > > Can you stop insisting that configfs can't do that so the target core
> > > can't do that?
> > >
> >
> > That's certainly not what I am insisting upon. I am trying to explain
> > that we currently have a configfs control plane that is native for
> > target mode.
> >
> > In the last three years to get to this point, I have developed prototype
> > code and asked upstream review for:
> >
> > *) Creating logic to drive configuration from kernel-space for configfs
> > *) Tried at creating sysfs -> configfs symlinks for referencing target
> > core backend devices
> >
> > So far both of these attempts at patches have been firmly rejected by
> > the configfs and sysfs maintainers.
> >
> > I am not saying that configfs is the end-all for all generic target mode
> > control plane, but what I am saying is that I have yet to see the code
> > for a different control plane that makes me want to move the away from a
> > 'native configfs' to 'a configfs/sysfs hyrbid', or something else all
> > together for mainline target code.
>
> OK, so what about an upcall to userspace to create the necessary
> directories? That could be driven by the kernel and still not require
> any implementation in configfs.
>
I think finding a way to do this is a good idea for the target fabric
endpoints as long as we can ensure that symlinks from target core
backports in target_core_mod to fabric Port/LUNs are still driven by
interpreted userspace code.
Allowing certain modules to drive target struct config_group endpoint
creation/deletion from kernel-space is fine with me, but I think for
something like this to really work they need to be done interchangebly.
At this point I would be leaning for uevent for something like this
(this would work on a 64-bit kernel / 32-bit user environment, right).
Do you have any kernel code pointers handy that I can grok as a future
target v4.1 feature..?
> > > We should think about what is the best design for the target code
> > > first.
> > >
> > > > > And iSCSI setup usually needs userspace code such as iSNS
> > > > > anyway. You'll add iSNS to kernel too.
> > > > >
> > > >
> > > > iSNS should be walking the /sys/kernel/config/target/iscsi layout, and
> > > > should not require any kernel code at all.
> > >
> > > How you draw the line? Why iSNS can live in user space?
> > >
> > > For example, you say that iSNS should live in user space. The similar
> > > feature to sending the list of target lives in kernel space.
> > >
> > >
> >
> > iSNS is a seperate protocol, and there is no hard-requirement of iSNS
> > client logic to be in place in order for basic iSCSI login to function.
> >
> > > > What about for the boot1.kernel.org cases where we can expect 1000s of
> > > > R/O clients in the future to a proper 10 Gb/sec uplink.
> > > >
> > > > Why do all of these types of logins need to talk with a userspace login
> > > > queue interface?
> > >
> > > I prefer less kernel space code.
> > >
> > >
> > > > Why do we need to worry about an interface in the
> > > > first place for the standard iSCSI login case..? What happens if this
> > > > daemon is unexpectedly killed..? Do I now have to worry about
> > >
> > > Can you stop such pointless argument? What happens if the iscsi target
> > > kernel module crashed?
> > >
> > > Linux systems already depend on some essential user space daemons. We
> > > know how to deal with them.
> > >
> > >
> >
> > Sorry, but I have no interest in adding the extra complexity to handle
> > this for the default case.
>
> Look at it this way: there's no way an implementation of all the RFC3720
> authentication methods (let alone the extensions) is going into the
> kernel, so you need a user space interface anyway. Once you have one,
> creating two paths: one in-kernel and one via upcall just doubles the
> maintenance load and the amount of work that has to be done fixing bugs.
> As far as I can see, authentication isn't in the fast path, so there's
> no real need for any of it to be in-kernel in the first place.
>
To clarify my stance here. I was not implying that we use two seperate
in-kernel and userspace level iSCSI login codebases, or that maintenance
load has anything to do with my choice of iSCSI login implementation for
this particular point.
What I am saying is that only the optional to implement iSCSI
authentication *payloads* should have a request/response interface with
userspace daemons for those types of auth libs that we always expect to
reside in userspace. This is what iscsi-target v2 had originally been
doing for CHAP and SRP (using a loopback socket to a userspace daemon)
until I decided to make add RFC-3720 required default CHAP support into
iscsi-target v3 using libcrypto md5.
The reasons why I really like a design where only the 'optional to
implement' iSCSI authentication payloads in iSCSI login stage CSG=0
interact with userspace is:
*) In future code, it allows 'optional to implement' userspace
authentication daemons to function seperately per iSCSI TargetName
+TargetPortalGroupTag context. eg: Different authentication daemon
processes can function independently on each target endpoint context.
*) In current code, it allows incoming iSCSI TargetName
+TargetPortalGroup reference to struct iscsi_portal_group <-> struct
iscsi_np, and InitiatorName+ISID to target core struct se_node_acl in
kernel-space without global userspace sychronization. eg: iscsi_np
kthreads are independent of userspace
*) In current code, it allows for the AuthMethod=None iSCSI login case
to skip the iSCSI login authentication phase all-together (by-passing
login stage CSG=0) and perform the login in kernel-space without global
userspace sychronization.
*) In current code, it allows the RFC-3720 required CHAP support to
'just-work' out of the box when you load iscsi_target_mod.ko
*) The vast majority of iSCSI initiators on Linux and non Linux
platforms only support CHAP.
So that said, having to do any type of global or non-global userspace
sychronization for the second and third point above is what I strongly
consider to be a functional step backwards from what exists in
iscsi-target v4.1 and has been proposed in PATCH-v2.
Having to sychronize all iSCSI login state with userspace code is a
awkward approach for an in-kernel iSCSI target, especially for one who's
control plane is built from the ground up to be real-time configurable
from configfs.
--nab
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-25 21:14 ` Nicholas A. Bellinger
@ 2011-03-27 10:02 ` FUJITA Tomonori
2011-03-28 8:30 ` Nicholas A. Bellinger
0 siblings, 1 reply; 40+ messages in thread
From: FUJITA Tomonori @ 2011-03-27 10:02 UTC (permalink / raw)
To: nab
Cc: James.Bottomley, fujita.tomonori, linux-scsi, hch, michaelc, hare,
bharrosh
On Fri, 25 Mar 2011 14:14:50 -0700
"Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> What I am saying is that only the optional to implement iSCSI
> authentication *payloads* should have a request/response interface with
> userspace daemons for those types of auth libs that we always expect to
We don't need such interface (authentication *payloads* between kernel
and user spaces). All operations (including authentication) before
establishing a nexus should be done in userspace.
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-27 10:02 ` FUJITA Tomonori
@ 2011-03-28 8:30 ` Nicholas A. Bellinger
2011-03-28 8:33 ` FUJITA Tomonori
0 siblings, 1 reply; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-28 8:30 UTC (permalink / raw)
To: FUJITA Tomonori
Cc: James.Bottomley, linux-scsi, hch, michaelc, hare, bharrosh
On Sun, 2011-03-27 at 19:02 +0900, FUJITA Tomonori wrote:
> On Fri, 25 Mar 2011 14:14:50 -0700
> "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
>
> > What I am saying is that only the optional to implement iSCSI
> > authentication *payloads* should have a request/response interface with
> > userspace daemons for those types of auth libs that we always expect to
>
> We don't need such interface (authentication *payloads* between kernel
> and user spaces). All operations (including authentication) before
> establishing a nexus should be done in userspace.
> --
No, the interface I speak of is most definately not specific to iSCSI
authentication payloads.
What I meant here is that iscsi-target should be sending RFC-3720
optional authentication type specific key=value payloads into userspace
daemon(s) using some form of acceptable mainline kernel <-> user
interface in mainline kernel code.
This interface also really needs to be 64-bit kernel / 32-bit user safe
as well.
--nab
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-28 8:30 ` Nicholas A. Bellinger
@ 2011-03-28 8:33 ` FUJITA Tomonori
0 siblings, 0 replies; 40+ messages in thread
From: FUJITA Tomonori @ 2011-03-28 8:33 UTC (permalink / raw)
To: nab
Cc: fujita.tomonori, James.Bottomley, linux-scsi, hch, michaelc, hare,
bharrosh
On Mon, 28 Mar 2011 01:30:03 -0700
"Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> On Sun, 2011-03-27 at 19:02 +0900, FUJITA Tomonori wrote:
> > On Fri, 25 Mar 2011 14:14:50 -0700
> > "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> >
> > > What I am saying is that only the optional to implement iSCSI
> > > authentication *payloads* should have a request/response interface with
> > > userspace daemons for those types of auth libs that we always expect to
> >
> > We don't need such interface (authentication *payloads* between kernel
> > and user spaces). All operations (including authentication) before
> > establishing a nexus should be done in userspace.
> > --
>
> No, the interface I speak of is most definately not specific to iSCSI
> authentication payloads.
>
> What I meant here is that iscsi-target should be sending RFC-3720
> optional authentication type specific key=value payloads into userspace
> daemon(s) using some form of acceptable mainline kernel <-> user
> interface in mainline kernel code.
The deamon can read configfs files and build the payloads.
> This interface also really needs to be 64-bit kernel / 32-bit user safe
> as well.
I'm sure that we (kernel developers) are good enough to handle it
nicely.
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-24 6:59 ` Nicholas A. Bellinger
2011-03-25 0:18 ` James Bottomley
@ 2011-03-27 10:02 ` FUJITA Tomonori
2011-03-28 8:53 ` Nicholas A. Bellinger
1 sibling, 1 reply; 40+ messages in thread
From: FUJITA Tomonori @ 2011-03-27 10:02 UTC (permalink / raw)
To: nab
Cc: fujita.tomonori, linux-scsi, James.Bottomley, hch, michaelc, hare,
bharrosh
On Wed, 23 Mar 2011 23:59:40 -0700
"Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
> > > But reading the of the configfs layout is not the problem, yes..?
> >
> > I don't think so. The user space doesn't cache the configfs info. So
> > the rule is pretty simple.
>
> It's more complicated than that. We have to then start referencing
> things like struct se_node_acl usage to userspace when locating an
> explict NodeACL during login. What happens if the explict NodeACL is
> removed from configfs context while userspace code is running..? What
> happens if the entire TPG is removed during an active userspace login..?
> It all has to be sychronized with a userspace daemon through a single
> thread that needs to know about the network portals, TPG layout and
> explict NodeACls in order to complete a iSCSI login.
Such can happen even if you do in kernel. During the login process,
the configuration may change. You can't lock the configuration during
the login process even if you do all in kernel. We need no
synchronization at all with the user space approach.
^ permalink raw reply [flat|nested] 40+ messages in thread
* Re: [PATCH-v2 00/14] iscsi-target: iSCSI target v4.1.0-rc1 series initial merge
2011-03-27 10:02 ` FUJITA Tomonori
@ 2011-03-28 8:53 ` Nicholas A. Bellinger
0 siblings, 0 replies; 40+ messages in thread
From: Nicholas A. Bellinger @ 2011-03-28 8:53 UTC (permalink / raw)
To: FUJITA Tomonori
Cc: linux-scsi, James.Bottomley, hch, michaelc, hare, bharrosh
On Sun, 2011-03-27 at 19:02 +0900, FUJITA Tomonori wrote:
> On Wed, 23 Mar 2011 23:59:40 -0700
> "Nicholas A. Bellinger" <nab@linux-iscsi.org> wrote:
>
> > > > But reading the of the configfs layout is not the problem, yes..?
> > >
> > > I don't think so. The user space doesn't cache the configfs info. So
> > > the rule is pretty simple.
> >
> > It's more complicated than that. We have to then start referencing
> > things like struct se_node_acl usage to userspace when locating an
> > explict NodeACL during login. What happens if the explict NodeACL is
> > removed from configfs context while userspace code is running..? What
> > happens if the entire TPG is removed during an active userspace login..?
> > It all has to be sychronized with a userspace daemon through a single
> > thread that needs to know about the network portals, TPG layout and
> > explict NodeACls in order to complete a iSCSI login.
>
> Such can happen even if you do in kernel. During the login process,
> the configuration may change. You can't lock the configuration during
> the login process even if you do all in kernel. We need no
> synchronization at all with the user space approach.
> --
Not sure what you mean here by saying 'You can't lock the configuration
during the login process'.
The PATCH-v2 of iscsi-target code already provides the ability to
add/remove iSCSI network portals <-> iSCSI target portal group
assocations independently of other iSCSI TargetName+TargetPortalGroupTag
endpoints in a real-time M:N configuration via configfs regardless of
individual IPv6/IPv4 network portal or iSCSI login state.
Having to perform a level synchronization with an userspace iSCSI login
daemon for real-time configurable control plane of M:N iSCSI network
portals <-> iSCSI target portal group really does not make any sense
here. This type of design change only adds extra unnecessary complexity
to existing code for in-kernel iscsi-target who's control plane is
already driven exclusively by userspace using mainline generic
target_core_fabric_configfs.c code.
--nab
^ permalink raw reply [flat|nested] 40+ messages in thread