* Re: [2.6 patch] document that 8139TOO supports 8129/8130
From: Jeff Garzik @ 2005-05-15 21:59 UTC (permalink / raw)
To: Adrian Bunk; +Cc: netdev, linux-kernel
In-Reply-To: <20050513040445.GD3603@stusta.de>
Adrian Bunk wrote:
> + the RTL 8129/81308139 chips. If you have one of those, say Y and
typo: "81308139"
^ permalink raw reply
* Re: [IPV4/IPV6] Ensure all frag_list members have NULL sk
From: Herbert Xu @ 2005-05-15 21:29 UTC (permalink / raw)
To: Evgeniy Polyakov; +Cc: netdev, davem
In-Reply-To: <20050515213341.4011c00e@zanzibar.2ka.mipt.ru>
On Sun, May 15, 2005 at 09:33:41PM +0400, Evgeniy Polyakov wrote:
>
> > So let's go the other way and make this an invariant:
> >
> > For any skb on a frag_list, skb->sk must be NULL.
>
> This requires skb_set_owner_* to check if it is called
> for head skb or one from fragment and does nothing if
> it is from frag_list.
> Or to check the whole tree for ownering calls...
Not really. The frag_list skb's owned by sk's are generated in
one place only, and that place is ip*_push_pending_frames.
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* Re: git repository for net drivers available
From: Jeff Garzik @ 2005-05-15 20:20 UTC (permalink / raw)
To: Petr Baudis; +Cc: James Ketrenos, Netdev, Linux Kernel, git, Andrew Morton
In-Reply-To: <20050515200514.GA31414@pasky.ji.cz>
Petr Baudis wrote:
> Dear diary, on Fri, May 13, 2005 at 05:29:30PM CEST, I got a letter
> where Jeff Garzik <jgarzik@pobox.com> told me that...
>
>>Looks like cogito is using $repo/heads/$branch, whereas my git repo is
>>using $repo/branches/$branch.
>
>
> Would it be a big problem to use refs/heads/$branch? That's the
> currently commonly agreed convention about location for storing branch
> heads, not just some weird Cogito-specific invention. And it'd be very
> nice to have those locations consistent across git repositories.
Sure, that's doable.
I've pushed out this change to kernel.org.
Jeff
^ permalink raw reply
* Re: git repository for net drivers available
From: Petr Baudis @ 2005-05-15 20:05 UTC (permalink / raw)
To: Jeff Garzik; +Cc: James Ketrenos, Netdev, Linux Kernel, git
In-Reply-To: <4284C7DA.1020707@pobox.com>
Dear diary, on Fri, May 13, 2005 at 05:29:30PM CEST, I got a letter
where Jeff Garzik <jgarzik@pobox.com> told me that...
> Looks like cogito is using $repo/heads/$branch, whereas my git repo is
> using $repo/branches/$branch.
Would it be a big problem to use refs/heads/$branch? That's the
currently commonly agreed convention about location for storing branch
heads, not just some weird Cogito-specific invention. And it'd be very
nice to have those locations consistent across git repositories.
Thanks,
--
Petr "Pasky" Baudis
Stuff: http://pasky.or.cz/
C++: an octopus made by nailing extra legs onto a dog. -- Steve Taylor
^ permalink raw reply
* [PATCH 3/3] add open iscsi netlink interface to iscsi transport class
From: Mike Christie @ 2005-05-15 19:30 UTC (permalink / raw)
To: open-iscsi, netdev, linux-scsi, linux-kernel, inux-iscsi-devel
incorporate the open-iscsi/linux-iscsi netlink interface
into the iscsi transport class.
See here for details on the sysfs layout
http://marc.theaimsgroup.com/?l=linux-kernel&m=111526269831567&w=2
diff -aurp linux-2.6.12-rc4/drivers/scsi/scsi_transport_iscsi.c linux-2.6.12-rc4.work/drivers/scsi/scsi_transport_iscsi.c
--- linux-2.6.12-rc4/drivers/scsi/scsi_transport_iscsi.c 2005-05-06 22:20:31.000000000 -0700
+++ linux-2.6.12-rc4.work/drivers/scsi/scsi_transport_iscsi.c 2005-05-14 12:20:22.000000000 -0700
@@ -1,8 +1,10 @@
-/*
+/*
* iSCSI transport class definitions
*
* Copyright (C) IBM Corporation, 2004
- * Copyright (C) Mike Christie, 2004
+ * Copyright (C) Mike Christie, 2004 - 2005
+ * Copyright (C) Dmitry Yusupov, 2004 - 2005
+ * Copyright (C) Alex Aizman, 2004 - 2005
*
* 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
@@ -19,370 +21,1189 @@
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*/
#include <linux/module.h>
+#include <linux/mempool.h>
+#include <net/tcp.h>
#include <scsi/scsi.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_transport.h>
#include <scsi/scsi_transport_iscsi.h>
+#include <scsi/iscsi_ifev.h>
-#define ISCSI_SESSION_ATTRS 20
-#define ISCSI_HOST_ATTRS 2
+#define ISCSI_SESSION_ATTRS 8
+#define ISCSI_CONN_ATTRS 6
struct iscsi_internal {
struct scsi_transport_template t;
- struct iscsi_function_template *fnt;
+ struct iscsi_transport *iscsi_transport;
+ struct list_head list;
+ /*
+ * List of sessions for this transport
+ */
+ struct list_head sessions;
+ /*
+ * lock to serialize access to the sessions list which must
+ * be taken after the rx_queue_sema
+ */
+ spinlock_t session_lock;
+ /*
+ * based on transport capabilities, at register time we sets these
+ * bits to tell the transport class it wants the attributes
+ * displayed in sysfs.
+ */
+ uint32_t param_mask;
+
/*
* We do not have any private or other attrs.
*/
+ struct transport_container conn_cont;
+ struct class_device_attribute *conn_attrs[ISCSI_CONN_ATTRS + 1];
+ struct transport_container session_cont;
struct class_device_attribute *session_attrs[ISCSI_SESSION_ATTRS + 1];
- struct class_device_attribute *host_attrs[ISCSI_HOST_ATTRS + 1];
};
+/*
+ * list of registered transports and lock that must
+ * be held while accessing list. The iscsi_transport_lock must
+ * be acquired after the rx_queue_sema.
+ */
+static LIST_HEAD(iscsi_transports);
+static DEFINE_SPINLOCK(iscsi_transport_lock);
+
#define to_iscsi_internal(tmpl) container_of(tmpl, struct iscsi_internal, t)
-static DECLARE_TRANSPORT_CLASS(iscsi_transport_class,
- "iscsi_transport",
+static DECLARE_TRANSPORT_CLASS(iscsi_session_class,
+ "iscsi_session",
NULL,
NULL,
NULL);
-static DECLARE_TRANSPORT_CLASS(iscsi_host_class,
- "iscsi_host",
+static DECLARE_TRANSPORT_CLASS(iscsi_connection_class,
+ "iscsi_connection",
NULL,
NULL,
NULL);
+
+static struct sock *nls;
+static int daemon_pid;
+static DECLARE_MUTEX(rx_queue_sema);
+
+struct mempool_zone {
+ mempool_t *pool;
+ atomic_t allocated;
+ int size;
+ int hiwat;
+ struct list_head freequeue;
+ spinlock_t freelock;
+};
+
+static struct mempool_zone z_reply;
+
/*
- * iSCSI target and session attrs
+ * Z_MAX_* - actual mempool size allocated at the mempool_zone_init() time
+ * Z_HIWAT_* - zone's high watermark when if_error bit will be set to -ENOMEM
+ * so daemon will notice OOM on NETLINK tranposrt level and will
+ * be able to predict or change operational behavior
*/
-#define iscsi_session_show_fn(field, format) \
- \
-static ssize_t \
-show_session_##field(struct class_device *cdev, char *buf) \
-{ \
- struct scsi_target *starget = transport_class_to_starget(cdev); \
- struct Scsi_Host *shost = dev_to_shost(starget->dev.parent); \
- struct iscsi_internal *i = to_iscsi_internal(shost->transportt); \
- \
- if (i->fnt->get_##field) \
- i->fnt->get_##field(starget); \
- return snprintf(buf, 20, format"\n", iscsi_##field(starget)); \
-}
-
-#define iscsi_session_rd_attr(field, format) \
- iscsi_session_show_fn(field, format) \
-static CLASS_DEVICE_ATTR(field, S_IRUGO, show_session_##field, NULL);
-
-iscsi_session_rd_attr(tpgt, "%hu");
-iscsi_session_rd_attr(tsih, "%2x");
-iscsi_session_rd_attr(max_recv_data_segment_len, "%u");
-iscsi_session_rd_attr(max_burst_len, "%u");
-iscsi_session_rd_attr(first_burst_len, "%u");
-iscsi_session_rd_attr(def_time2wait, "%hu");
-iscsi_session_rd_attr(def_time2retain, "%hu");
-iscsi_session_rd_attr(max_outstanding_r2t, "%hu");
-iscsi_session_rd_attr(erl, "%d");
+#define Z_MAX_REPLY 8
+#define Z_HIWAT_REPLY 6
+#define Z_MAX_PDU 8
+#define Z_HIWAT_PDU 6
+#define Z_MAX_ERROR 16
+#define Z_HIWAT_ERROR 12
+
+struct iscsi_if_conn {
+ struct list_head conn_list; /* item in connlist */
+ struct list_head session_list; /* item in session->connections */
+ iscsi_connh_t connh;
+ int active; /* must be accessed with the connlock */
+ struct Scsi_Host *host; /* originated shost */
+ struct device dev; /* sysfs transport/container device */
+ struct iscsi_transport *transport;
+ struct mempool_zone z_error;
+ struct mempool_zone z_pdu;
+ struct list_head freequeue;
+};
+#define iscsi_dev_to_if_conn(_dev) \
+ container_of(_dev, struct iscsi_if_conn, dev)
-#define iscsi_session_show_bool_fn(field) \
- \
-static ssize_t \
-show_session_bool_##field(struct class_device *cdev, char *buf) \
-{ \
- struct scsi_target *starget = transport_class_to_starget(cdev); \
- struct Scsi_Host *shost = dev_to_shost(starget->dev.parent); \
- struct iscsi_internal *i = to_iscsi_internal(shost->transportt); \
- \
- if (i->fnt->get_##field) \
- i->fnt->get_##field(starget); \
- \
- if (iscsi_##field(starget)) \
- return sprintf(buf, "Yes\n"); \
- return sprintf(buf, "No\n"); \
+#define iscsi_cdev_to_if_conn(_cdev) \
+ iscsi_dev_to_if_conn(_cdev->dev)
+
+static LIST_HEAD(connlist);
+static DEFINE_SPINLOCK(connlock);
+
+struct iscsi_if_session {
+ struct list_head list; /* item in session_list */
+ struct list_head connections;
+ iscsi_sessionh_t sessionh;
+ struct iscsi_transport *transport;
+ struct device dev; /* sysfs transport/container device */
+};
+
+#define iscsi_dev_to_if_session(_dev) \
+ container_of(_dev, struct iscsi_if_session, dev)
+
+#define iscsi_cdev_to_if_session(_cdev) \
+ iscsi_dev_to_if_session(_cdev->dev)
+
+#define iscsi_if_session_to_shost(_session) \
+ dev_to_shost(_session->dev.parent)
+
+static struct iscsi_if_conn*
+iscsi_if_find_conn(uint64_t key)
+{
+ unsigned long flags;
+ struct iscsi_if_conn *conn;
+
+ spin_lock_irqsave(&connlock, flags);
+ list_for_each_entry(conn, &connlist, conn_list)
+ if (conn->connh == key) {
+ spin_unlock_irqrestore(&connlock, flags);
+ return conn;
+ }
+ spin_unlock_irqrestore(&connlock, flags);
+ return NULL;
}
-#define iscsi_session_rd_bool_attr(field) \
- iscsi_session_show_bool_fn(field) \
-static CLASS_DEVICE_ATTR(field, S_IRUGO, show_session_bool_##field, NULL);
+static struct iscsi_internal *
+iscsi_if_transport_lookup(struct iscsi_transport *tt)
+{
+ struct iscsi_internal *priv;
+ unsigned long flags;
-iscsi_session_rd_bool_attr(initial_r2t);
-iscsi_session_rd_bool_attr(immediate_data);
-iscsi_session_rd_bool_attr(data_pdu_in_order);
-iscsi_session_rd_bool_attr(data_sequence_in_order);
+ spin_lock_irqsave(&iscsi_transport_lock, flags);
+ list_for_each_entry(priv, &iscsi_transports, list)
+ if (tt == priv->iscsi_transport) {
+ spin_unlock_irqrestore(&iscsi_transport_lock, flags);
+ return priv;
+ }
+ spin_unlock_irqrestore(&iscsi_transport_lock, flags);
+ return NULL;
+}
-#define iscsi_session_show_digest_fn(field) \
- \
-static ssize_t \
-show_##field(struct class_device *cdev, char *buf) \
-{ \
- struct scsi_target *starget = transport_class_to_starget(cdev); \
- struct Scsi_Host *shost = dev_to_shost(starget->dev.parent); \
- struct iscsi_internal *i = to_iscsi_internal(shost->transportt); \
- \
- if (i->fnt->get_##field) \
- i->fnt->get_##field(starget); \
- \
- if (iscsi_##field(starget)) \
- return sprintf(buf, "CRC32C\n"); \
- return sprintf(buf, "None\n"); \
+static inline struct list_head *skb_to_lh(struct sk_buff *skb)
+{
+ return (struct list_head *)&skb->cb;
}
-#define iscsi_session_rd_digest_attr(field) \
- iscsi_session_show_digest_fn(field) \
-static CLASS_DEVICE_ATTR(field, S_IRUGO, show_##field, NULL);
+static void*
+mempool_zone_alloc_skb(unsigned int gfp_mask, void *pool_data)
+{
+ struct mempool_zone *zone = pool_data;
-iscsi_session_rd_digest_attr(header_digest);
-iscsi_session_rd_digest_attr(data_digest);
+ return alloc_skb(zone->size, gfp_mask);
+}
-static ssize_t
-show_port(struct class_device *cdev, char *buf)
+static void
+mempool_zone_free_skb(void *element, void *pool_data)
{
- struct scsi_target *starget = transport_class_to_starget(cdev);
- struct Scsi_Host *shost = dev_to_shost(starget->dev.parent);
- struct iscsi_internal *i = to_iscsi_internal(shost->transportt);
+ kfree_skb(element);
+}
- if (i->fnt->get_port)
- i->fnt->get_port(starget);
+static void
+mempool_zone_complete(struct mempool_zone *zone)
+{
+ unsigned long flags;
+ struct list_head *lh, *n;
- return snprintf(buf, 20, "%hu\n", ntohs(iscsi_port(starget)));
+ spin_lock_irqsave(&zone->freelock, flags);
+ list_for_each_safe(lh, n, &zone->freequeue) {
+ struct sk_buff *skb = (struct sk_buff *)((char *)lh -
+ offsetof(struct sk_buff, cb));
+ if (!skb_shared(skb)) {
+ list_del(skb_to_lh(skb));
+ mempool_free(skb, zone->pool);
+ atomic_dec(&zone->allocated);
+ }
+ }
+ spin_unlock_irqrestore(&zone->freelock, flags);
}
-static CLASS_DEVICE_ATTR(port, S_IRUGO, show_port, NULL);
-static ssize_t
-show_ip_address(struct class_device *cdev, char *buf)
+static int
+mempool_zone_init(struct mempool_zone *zp, unsigned max, unsigned size,
+ unsigned hiwat)
{
- struct scsi_target *starget = transport_class_to_starget(cdev);
- struct Scsi_Host *shost = dev_to_shost(starget->dev.parent);
- struct iscsi_internal *i = to_iscsi_internal(shost->transportt);
+ zp->pool = mempool_create(max, mempool_zone_alloc_skb,
+ mempool_zone_free_skb, zp);
+ if (!zp->pool)
+ return -ENOMEM;
+
+ zp->size = size;
+ zp->hiwat = hiwat;
+
+ INIT_LIST_HEAD(&zp->freequeue);
+ spin_lock_init(&zp->freelock);
+ atomic_set(&zp->allocated, 0);
+
+ return 0;
+}
- if (i->fnt->get_ip_address)
- i->fnt->get_ip_address(starget);
- if (iscsi_addr_type(starget) == AF_INET)
- return sprintf(buf, "%u.%u.%u.%u\n",
- NIPQUAD(iscsi_sin_addr(starget)));
- else if(iscsi_addr_type(starget) == AF_INET6)
- return sprintf(buf, "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x\n",
- NIP6(iscsi_sin6_addr(starget)));
- return -EINVAL;
+static struct sk_buff*
+mempool_zone_get_skb(struct mempool_zone *zone)
+{
+ struct sk_buff *skb;
+
+ skb = mempool_alloc(zone->pool, GFP_ATOMIC);
+ if (skb)
+ atomic_inc(&zone->allocated);
+ return skb;
}
-static CLASS_DEVICE_ATTR(ip_address, S_IRUGO, show_ip_address, NULL);
-static ssize_t
-show_isid(struct class_device *cdev, char *buf)
+static int
+iscsi_unicast_skb(struct mempool_zone *zone, struct sk_buff *skb)
{
- struct scsi_target *starget = transport_class_to_starget(cdev);
- struct Scsi_Host *shost = dev_to_shost(starget->dev.parent);
- struct iscsi_internal *i = to_iscsi_internal(shost->transportt);
+ unsigned long flags;
+ int rc;
+
+ skb_get(skb);
+ rc = netlink_unicast(nls, skb, daemon_pid, MSG_DONTWAIT);
+ if (rc < 0) {
+ mempool_free(skb, zone->pool);
+ printk(KERN_ERR "iscsi: can not unicast skb (%d)\n", rc);
+ return rc;
+ }
- if (i->fnt->get_isid)
- i->fnt->get_isid(starget);
+ spin_lock_irqsave(&zone->freelock, flags);
+ list_add(skb_to_lh(skb), &zone->freequeue);
+ spin_unlock_irqrestore(&zone->freelock, flags);
- return sprintf(buf, "%02x%02x%02x%02x%02x%02x\n",
- iscsi_isid(starget)[0], iscsi_isid(starget)[1],
- iscsi_isid(starget)[2], iscsi_isid(starget)[3],
- iscsi_isid(starget)[4], iscsi_isid(starget)[5]);
+ return 0;
}
-static CLASS_DEVICE_ATTR(isid, S_IRUGO, show_isid, NULL);
-/*
- * This is used for iSCSI names. Normally, we follow
- * the transport class convention of having the lld
- * set the field, but in these cases the value is
- * too large.
- */
-#define iscsi_session_show_str_fn(field) \
- \
-static ssize_t \
-show_session_str_##field(struct class_device *cdev, char *buf) \
-{ \
- ssize_t ret = 0; \
- struct scsi_target *starget = transport_class_to_starget(cdev); \
- struct Scsi_Host *shost = dev_to_shost(starget->dev.parent); \
- struct iscsi_internal *i = to_iscsi_internal(shost->transportt); \
- \
- if (i->fnt->get_##field) \
- ret = i->fnt->get_##field(starget, buf, PAGE_SIZE); \
- return ret; \
+int iscsi_recv_pdu(iscsi_connh_t connh, struct iscsi_hdr *hdr,
+ char *data, uint32_t data_size)
+{
+ struct nlmsghdr *nlh;
+ struct sk_buff *skb;
+ struct iscsi_uevent *ev;
+ struct iscsi_if_conn *conn;
+ char *pdu;
+ int len = NLMSG_SPACE(sizeof(*ev) + sizeof(struct iscsi_hdr) +
+ data_size);
+
+ conn = iscsi_if_find_conn(connh);
+ BUG_ON(!conn);
+
+ mempool_zone_complete(&conn->z_pdu);
+
+ skb = mempool_zone_get_skb(&conn->z_pdu);
+ if (!skb) {
+ iscsi_conn_error(connh, ISCSI_ERR_CONN_FAILED);
+ printk(KERN_ERR "iscsi%d: can not deliver control PDU: OOM\n",
+ conn->host->host_no);
+ return -ENOMEM;
+ }
+
+ nlh = __nlmsg_put(skb, daemon_pid, 0, 0, (len - sizeof(*nlh)));
+ ev = NLMSG_DATA(nlh);
+ memset(ev, 0, sizeof(*ev));
+ ev->transport_handle = iscsi_handle(conn->transport);
+ ev->type = ISCSI_KEVENT_RECV_PDU;
+ if (atomic_read(&conn->z_pdu.allocated) >= conn->z_pdu.hiwat)
+ ev->iferror = -ENOMEM;
+ ev->r.recv_req.conn_handle = connh;
+ pdu = (char*)ev + sizeof(*ev);
+ memcpy(pdu, hdr, sizeof(struct iscsi_hdr));
+ memcpy(pdu + sizeof(struct iscsi_hdr), data, data_size);
+
+ return iscsi_unicast_skb(&conn->z_pdu, skb);
}
+EXPORT_SYMBOL_GPL(iscsi_recv_pdu);
-#define iscsi_session_rd_str_attr(field) \
- iscsi_session_show_str_fn(field) \
-static CLASS_DEVICE_ATTR(field, S_IRUGO, show_session_str_##field, NULL);
+void iscsi_conn_error(iscsi_connh_t connh, enum iscsi_err error)
+{
+ struct nlmsghdr *nlh;
+ struct sk_buff *skb;
+ struct iscsi_uevent *ev;
+ struct iscsi_if_conn *conn;
+ int len = NLMSG_SPACE(sizeof(*ev));
+
+ conn = iscsi_if_find_conn(connh);
+ BUG_ON(!conn);
+
+ mempool_zone_complete(&conn->z_error);
+
+ skb = mempool_zone_get_skb(&conn->z_error);
+ if (!skb) {
+ printk(KERN_ERR "iscsi%d: gracefully ignored conn error (%d)\n",
+ conn->host->host_no, error);
+ return;
+ }
+
+ nlh = __nlmsg_put(skb, daemon_pid, 0, 0, (len - sizeof(*nlh)));
+ ev = NLMSG_DATA(nlh);
+ ev->transport_handle = iscsi_handle(conn->transport);
+ ev->type = ISCSI_KEVENT_CONN_ERROR;
+ if (atomic_read(&conn->z_error.allocated) >= conn->z_error.hiwat)
+ ev->iferror = -ENOMEM;
+ ev->r.connerror.error = error;
+ ev->r.connerror.conn_handle = connh;
-iscsi_session_rd_str_attr(target_name);
-iscsi_session_rd_str_attr(target_alias);
+ iscsi_unicast_skb(&conn->z_error, skb);
+
+ printk(KERN_INFO "iscsi%d: detected conn error (%d)\n",
+ conn->host->host_no, error);
+}
+EXPORT_SYMBOL_GPL(iscsi_conn_error);
+
+static int
+iscsi_if_send_reply(int pid, int seq, int type, int done, int multi,
+ void *payload, int size)
+{
+ struct sk_buff *skb;
+ struct nlmsghdr *nlh;
+ int len = NLMSG_SPACE(size);
+ int flags = multi ? NLM_F_MULTI : 0;
+ int t = done ? NLMSG_DONE : type;
+
+ mempool_zone_complete(&z_reply);
+
+ skb = mempool_zone_get_skb(&z_reply);
+ /*
+ * FIXME:
+ * user is supposed to react on iferror == -ENOMEM;
+ * see iscsi_if_rx().
+ */
+ BUG_ON(!skb);
+
+ nlh = __nlmsg_put(skb, pid, seq, t, (len - sizeof(*nlh)));
+ nlh->nlmsg_flags = flags;
+ memcpy(NLMSG_DATA(nlh), payload, size);
+ return iscsi_unicast_skb(&z_reply, skb);
+}
/*
- * iSCSI host attrs
+ * iSCSI Session's hostdata organization:
+ *
+ * *------------------* <== host->hostdata
+ * | transport |
+ * |------------------| <== iscsi_hostdata(host->hostdata)
+ * | transport's data |
+ * |------------------| <== hostdata_session(host->hostdata)
+ * | interface's data |
+ * *------------------*
*/
+#define hostdata_privsize(_t) (sizeof(unsigned long) + _t->hostdata_size + \
+ _t->hostdata_size % sizeof(unsigned long) + \
+ sizeof(struct iscsi_if_session))
+
+#define hostdata_session(_hostdata) ((void*)_hostdata + sizeof(unsigned long) + \
+ ((struct iscsi_transport *) \
+ iscsi_ptr(*(uint64_t *)_hostdata))->hostdata_size)
+
+static void iscsi_if_session_dev_release(struct device *dev)
+{
+ struct iscsi_if_session *session = iscsi_dev_to_if_session(dev);
+ struct iscsi_transport *transport = session->transport;
+ struct Scsi_Host *shost = iscsi_if_session_to_shost(session);
+ struct iscsi_if_conn *conn, *tmp;
+ unsigned long flags;
+
+ /* now free connections */
+ spin_lock_irqsave(&connlock, flags);
+ list_for_each_entry_safe(conn, tmp, &session->connections,
+ session_list) {
+ list_del(&conn->session_list);
+ mempool_destroy(conn->z_pdu.pool);
+ mempool_destroy(conn->z_error.pool);
+ kfree(conn);
+ }
+ spin_unlock_irqrestore(&connlock, flags);
+ scsi_host_put(shost);
+ module_put(transport->owner);
+}
+
+static int
+iscsi_if_create_session(struct iscsi_internal *priv, struct iscsi_uevent *ev)
+{
+ struct iscsi_transport *transport = priv->iscsi_transport;
+ struct iscsi_if_session *session;
+ struct Scsi_Host *shost;
+ unsigned long flags;
+ int error;
+
+ if (!try_module_get(transport->owner))
+ return -EPERM;
+
+ shost = scsi_host_alloc(transport->host_template,
+ hostdata_privsize(transport));
+ if (!shost) {
+ ev->r.c_session_ret.session_handle = iscsi_handle(NULL);
+ printk(KERN_ERR "iscsi: can not allocate SCSI host for "
+ "session\n");
+ error = -ENOMEM;
+ goto out_module_put;
+ }
+ shost->max_id = 1;
+ shost->max_channel = 0;
+ shost->max_lun = transport->max_lun;
+ shost->max_cmd_len = transport->max_cmd_len;
+ shost->transportt = &priv->t;
+
+ /* store struct iscsi_transport in hostdata */
+ *(uint64_t*)shost->hostdata = ev->transport_handle;
+
+ ev->r.c_session_ret.session_handle = transport->create_session(
+ ev->u.c_session.initial_cmdsn, shost);
+ if (ev->r.c_session_ret.session_handle == iscsi_handle(NULL)) {
+ error = 0;
+ goto out_host_put;
+ }
+
+ /* host_no becomes assigned SID */
+ ev->r.c_session_ret.sid = shost->host_no;
+ /* initialize session */
+ session = hostdata_session(shost->hostdata);
+ INIT_LIST_HEAD(&session->connections);
+ INIT_LIST_HEAD(&session->list);
+ session->sessionh = ev->r.c_session_ret.session_handle;
+ session->transport = transport;
+
+ error = scsi_add_host(shost, NULL);
+ if (error)
+ goto out_destroy_session;
+
+ /*
+ * this is released in the dev's release function)
+ */
+ scsi_host_get(shost);
+ snprintf(session->dev.bus_id, BUS_ID_SIZE, "session%u", shost->host_no);
+ session->dev.parent = &shost->shost_gendev;
+ session->dev.release = iscsi_if_session_dev_release;
+ error = device_register(&session->dev);
+ if (error) {
+ printk(KERN_ERR "iscsi: could not register session%d's dev\n",
+ shost->host_no);
+ goto out_remove_host;
+ }
+ transport_register_device(&session->dev);
+
+ /* add this session to the list of active sessions */
+ spin_lock_irqsave(&priv->session_lock, flags);
+ list_add(&session->list, &priv->sessions);
+ spin_unlock_irqrestore(&priv->session_lock, flags);
+
+ return 0;
+
+ out_remove_host:
+ scsi_remove_host(shost);
+ out_destroy_session:
+ transport->destroy_session(ev->r.c_session_ret.session_handle);
+ ev->r.c_session_ret.session_handle = iscsi_handle(NULL);
+ out_host_put:
+ scsi_host_put(shost);
+ out_module_put:
+ module_put(transport->owner);
+ return error;
+}
+
+static int
+iscsi_if_destroy_session(struct iscsi_internal *priv, struct iscsi_uevent *ev)
+{
+ struct iscsi_transport *transport = priv->iscsi_transport;
+ struct Scsi_Host *shost;
+ struct iscsi_if_session *session;
+ unsigned long flags;
+ struct iscsi_if_conn *conn;
+ int error = 0;
+
+ shost = scsi_host_lookup(ev->u.d_session.sid);
+ if (shost == ERR_PTR(-ENXIO))
+ return -EEXIST;
+ session = hostdata_session(shost->hostdata);
+
+ /* check if we have active connections */
+ spin_lock_irqsave(&connlock, flags);
+ list_for_each_entry(conn, &session->connections, session_list) {
+ if (conn->active) {
+ printk(KERN_ERR "iscsi%d: can not destroy session: "
+ "has active connection (%p)\n",
+ shost->host_no, iscsi_ptr(conn->connh));
+ spin_unlock_irqrestore(&connlock, flags);
+ error = EIO;
+ goto out_release_ref;
+ }
+ }
+ spin_unlock_irqrestore(&connlock, flags);
+
+ scsi_remove_host(shost);
+ transport->destroy_session(ev->u.d_session.session_handle);
+ transport_unregister_device(&session->dev);
+ device_unregister(&session->dev);
+
+ /* remove this session from the list of active sessions */
+ spin_lock_irqsave(&priv->session_lock, flags);
+ list_del(&session->list);
+ spin_unlock_irqrestore(&priv->session_lock, flags);
+
+ /* ref from host alloc */
+ scsi_host_put(shost);
+ out_release_ref:
+ /* ref from host lookup */
+ scsi_host_put(shost);
+ return error;
+}
+
+static void iscsi_if_conn_dev_release(struct device *dev)
+{
+ struct iscsi_if_conn *conn = iscsi_dev_to_if_conn(dev);
+ struct Scsi_Host *shost = conn->host;
+
+ scsi_host_put(shost);
+}
+
+static int
+iscsi_if_create_conn(struct iscsi_transport *transport, struct iscsi_uevent *ev)
+{
+ struct iscsi_if_session *session;
+ struct Scsi_Host *shost;
+ struct iscsi_if_conn *conn;
+ unsigned long flags;
+ int error;
+
+ shost = scsi_host_lookup(ev->u.c_conn.sid);
+ if (shost == ERR_PTR(-ENXIO))
+ return -EEXIST;
+ session = hostdata_session(shost->hostdata);
+
+ conn = kmalloc(sizeof(struct iscsi_if_conn), GFP_KERNEL);
+ if (!conn) {
+ error = -ENOMEM;
+ goto out_release_ref;
+ }
+ memset(conn, 0, sizeof(struct iscsi_if_conn));
+ INIT_LIST_HEAD(&conn->session_list);
+ INIT_LIST_HEAD(&conn->conn_list);
+ conn->host = shost;
+ conn->transport = transport;
+
+ error = mempool_zone_init(&conn->z_pdu, Z_MAX_PDU,
+ NLMSG_SPACE(sizeof(struct iscsi_uevent) +
+ sizeof(struct iscsi_hdr) +
+ DEFAULT_MAX_RECV_DATA_SEGMENT_LENGTH),
+ Z_HIWAT_PDU);
+ if (error) {
+ printk(KERN_ERR "iscsi%d: can not allocate pdu zone for new "
+ "conn\n", shost->host_no);
+ goto out_free_conn;
+ }
+ error = mempool_zone_init(&conn->z_error, Z_MAX_ERROR,
+ NLMSG_SPACE(sizeof(struct iscsi_uevent)),
+ Z_HIWAT_ERROR);
+ if (error) {
+ printk(KERN_ERR "iscsi%d: can not allocate error zone for "
+ "new conn\n", shost->host_no);
+ goto out_free_pdu_pool;
+ }
+
+ ev->r.handle = transport->create_conn(ev->u.c_conn.session_handle,
+ ev->u.c_conn.cid);
+ if (!ev->r.handle) {
+ error = -ENODEV;
+ goto out_free_error_pool;
+ }
+
+ conn->connh = ev->r.handle;
+
+ /*
+ * this is released in the dev's release function
+ */
+ if (!scsi_host_get(shost))
+ goto out_destroy_conn;
+ snprintf(conn->dev.bus_id, BUS_ID_SIZE, "connection%d:%u",
+ shost->host_no, ev->u.c_conn.cid);
+ conn->dev.parent = &session->dev;
+ conn->dev.release = iscsi_if_conn_dev_release;
+ error = device_register(&conn->dev);
+ if (error) {
+ printk(KERN_ERR "iscsi%d: could not register connections%u "
+ "dev\n", shost->host_no, ev->u.c_conn.cid);
+ goto out_release_parent_ref;
+ }
+ transport_register_device(&conn->dev);
+
+ spin_lock_irqsave(&connlock, flags);
+ list_add(&conn->conn_list, &connlist);
+ list_add(&conn->session_list, &session->connections);
+ conn->active = 1;
+ spin_unlock_irqrestore(&connlock, flags);
+
+ scsi_host_put(shost);
+ return 0;
+
+ out_release_parent_ref:
+ scsi_host_put(shost);
+ out_destroy_conn:
+ transport->destroy_conn(ev->r.handle);
+ out_free_error_pool:
+ mempool_destroy(conn->z_error.pool);
+ out_free_pdu_pool:
+ mempool_destroy(conn->z_pdu.pool);
+ out_free_conn:
+ kfree(conn);
+ out_release_ref:
+ scsi_host_put(shost);
+ return error;
+}
+
+static int
+iscsi_if_destroy_conn(struct iscsi_transport *transport, struct iscsi_uevent *ev)
+{
+ unsigned long flags;
+ struct iscsi_if_conn *conn;
+
+ conn = iscsi_if_find_conn(ev->u.d_conn.conn_handle);
+ if (!conn)
+ return -EEXIST;
+
+ transport->destroy_conn(ev->u.d_conn.conn_handle);
+
+ spin_lock_irqsave(&connlock, flags);
+ conn->active = 0;
+ list_del(&conn->conn_list);
+ spin_unlock_irqrestore(&connlock, flags);
+
+ transport_unregister_device(&conn->dev);
+ device_unregister(&conn->dev);
+ return 0;
+}
+
+static int
+iscsi_if_get_stats(struct iscsi_transport *transport, struct sk_buff *skb,
+ struct nlmsghdr *nlh)
+{
+ struct iscsi_uevent *ev = NLMSG_DATA(nlh);
+ struct iscsi_stats *stats;
+ struct sk_buff *skbstat;
+ struct iscsi_if_conn *conn;
+ struct nlmsghdr *nlhstat;
+ struct iscsi_uevent *evstat;
+ int len = NLMSG_SPACE(sizeof(*ev) +
+ sizeof(struct iscsi_stats) +
+ sizeof(struct iscsi_stats_custom) *
+ ISCSI_STATS_CUSTOM_MAX);
+ int err = 0;
+
+ conn = iscsi_if_find_conn(ev->u.get_stats.conn_handle);
+ if (!conn)
+ return -EEXIST;
+
+ do {
+ int actual_size;
+
+ mempool_zone_complete(&conn->z_pdu);
+
+ skbstat = mempool_zone_get_skb(&conn->z_pdu);
+ if (!skbstat) {
+ printk(KERN_ERR "iscsi%d: can not deliver stats: OOM\n",
+ conn->host->host_no);
+ return -ENOMEM;
+ }
+
+ nlhstat = __nlmsg_put(skbstat, daemon_pid, 0, 0,
+ (len - sizeof(*nlhstat)));
+ evstat = NLMSG_DATA(nlhstat);
+ memset(evstat, 0, sizeof(*evstat));
+ evstat->transport_handle = iscsi_handle(conn->transport);
+ evstat->type = nlh->nlmsg_type;
+ if (atomic_read(&conn->z_pdu.allocated) >= conn->z_pdu.hiwat)
+ evstat->iferror = -ENOMEM;
+ evstat->u.get_stats.conn_handle =
+ ev->u.get_stats.conn_handle;
+ stats = (struct iscsi_stats *)
+ ((char*)evstat + sizeof(*evstat));
+ memset(stats, 0, sizeof(*stats));
+
+ transport->get_stats(ev->u.get_stats.conn_handle, stats);
+ actual_size = NLMSG_SPACE(sizeof(struct iscsi_uevent) +
+ sizeof(struct iscsi_stats) +
+ sizeof(struct iscsi_stats_custom) *
+ stats->custom_length);
+ actual_size -= sizeof(*nlhstat);
+ actual_size = NLMSG_LENGTH(actual_size);
+ skb_trim(skb, NLMSG_ALIGN(actual_size));
+ nlhstat->nlmsg_len = actual_size;
+
+ err = iscsi_unicast_skb(&conn->z_pdu, skbstat);
+ } while (err < 0 && err != -ECONNREFUSED);
+
+ return err;
+}
+
+static int
+iscsi_if_recv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
+{
+ int err = 0;
+ struct iscsi_uevent *ev = NLMSG_DATA(nlh);
+ struct iscsi_transport *transport = NULL;
+ struct iscsi_internal *priv;
+
+ if (NETLINK_CREDS(skb)->uid)
+ return -EPERM;
+
+ priv = iscsi_if_transport_lookup(iscsi_ptr(ev->transport_handle));
+ if (priv)
+ transport = priv->iscsi_transport;
+ else if (nlh->nlmsg_type != ISCSI_UEVENT_TRANS_LIST)
+ return -EEXIST;
+
+ daemon_pid = NETLINK_CREDS(skb)->pid;
+
+ switch (nlh->nlmsg_type) {
+ case ISCSI_UEVENT_TRANS_LIST: {
+ unsigned long flags;
+
+ /*
+ * this will always succeed for now since there
+ * is only one transport. We can kill this event
+ * and just export the info in sysfs when we settle
+ * the iscsi sysfs layout debate
+ */
+ spin_lock_irqsave(&iscsi_transport_lock, flags);
+ list_for_each_entry(priv, &iscsi_transports, list) {
+ ev->r.t_list.elements[0].trans_handle =
+ iscsi_handle(priv->iscsi_transport);
+ strncpy(ev->r.t_list.elements[0].name,
+ priv->iscsi_transport->name,
+ ISCSI_TRANSPORT_NAME_MAXLEN);
+ }
+ spin_unlock_irqrestore(&iscsi_transport_lock, flags);
+
+ } break;
+ case ISCSI_UEVENT_CREATE_SESSION:
+ err = iscsi_if_create_session(priv, ev);
+ break;
+ case ISCSI_UEVENT_DESTROY_SESSION:
+ err = iscsi_if_destroy_session(priv, ev);
+ break;
+ case ISCSI_UEVENT_CREATE_CONN:
+ err = iscsi_if_create_conn(transport, ev);
+ break;
+ case ISCSI_UEVENT_DESTROY_CONN:
+ err = iscsi_if_destroy_conn(transport, ev);
+ break;
+ case ISCSI_UEVENT_BIND_CONN:
+ if (!iscsi_if_find_conn(ev->u.b_conn.conn_handle))
+ return -EEXIST;
+ ev->r.retcode = transport->bind_conn(
+ ev->u.b_conn.session_handle,
+ ev->u.b_conn.conn_handle,
+ ev->u.b_conn.transport_fd,
+ ev->u.b_conn.is_leading);
+ break;
+ case ISCSI_UEVENT_SET_PARAM:
+ if (!iscsi_if_find_conn(ev->u.set_param.conn_handle))
+ return -EEXIST;
+ ev->r.retcode = transport->set_param(
+ ev->u.set_param.conn_handle,
+ ev->u.set_param.param, ev->u.set_param.value);
+ break;
+ case ISCSI_UEVENT_START_CONN:
+ if (!iscsi_if_find_conn(ev->u.start_conn.conn_handle))
+ return -EEXIST;
+ ev->r.retcode = transport->start_conn(
+ ev->u.start_conn.conn_handle);
+ break;
+ case ISCSI_UEVENT_STOP_CONN:
+ if (!iscsi_if_find_conn(ev->u.stop_conn.conn_handle))
+ return -EEXIST;
+ transport->stop_conn(ev->u.stop_conn.conn_handle,
+ ev->u.stop_conn.flag);
+ break;
+ case ISCSI_UEVENT_SEND_PDU:
+ if (!iscsi_if_find_conn(ev->u.send_pdu.conn_handle))
+ return -EEXIST;
+ ev->r.retcode = transport->send_pdu(
+ ev->u.send_pdu.conn_handle,
+ (struct iscsi_hdr*)((char*)ev + sizeof(*ev)),
+ (char*)ev + sizeof(*ev) + ev->u.send_pdu.hdr_size,
+ ev->u.send_pdu.data_size);
+ break;
+ case ISCSI_UEVENT_GET_STATS:
+ err = iscsi_if_get_stats(transport, skb, nlh);
+ break;
+ default:
+ err = -EINVAL;
+ break;
+ }
+
+ return err;
+}
+
+/* Get message from skb (based on rtnetlink_rcv_skb). Each message is
+ * processed by iscsi_if_recv_msg. Malformed skbs with wrong length are
+ * discarded silently. */
+static void
+iscsi_if_rx(struct sock *sk, int len)
+{
+ struct sk_buff *skb;
+
+ down(&rx_queue_sema);
+ while ((skb = skb_dequeue(&sk->sk_receive_queue)) != NULL) {
+ while (skb->len >= NLMSG_SPACE(0)) {
+ int err;
+ uint32_t rlen;
+ struct nlmsghdr *nlh;
+ struct iscsi_uevent *ev;
+
+ nlh = (struct nlmsghdr *)skb->data;
+ if (nlh->nlmsg_len < sizeof(*nlh) ||
+ skb->len < nlh->nlmsg_len) {
+ break;
+ }
+ ev = NLMSG_DATA(nlh);
+ rlen = NLMSG_ALIGN(nlh->nlmsg_len);
+ if (rlen > skb->len)
+ rlen = skb->len;
+ err = iscsi_if_recv_msg(skb, nlh);
+ if (err) {
+ ev->type = ISCSI_KEVENT_IF_ERROR;
+ ev->iferror = err;
+ }
+ do {
+ /*
+ * special case for GET_STATS:
+ * on success - sending reply and stats from
+ * inside of if_recv_msg(),
+ * on error - fall through.
+ */
+ if (ev->type == ISCSI_UEVENT_GET_STATS && !err)
+ break;
+ err = iscsi_if_send_reply(
+ NETLINK_CREDS(skb)->pid, nlh->nlmsg_seq,
+ nlh->nlmsg_type, 0, 0, ev, sizeof(*ev));
+ if (atomic_read(&z_reply.allocated) >=
+ z_reply.hiwat)
+ ev->iferror = -ENOMEM;
+ } while (err < 0 && err != -ECONNREFUSED);
+ skb_pull(skb, rlen);
+ }
+ kfree_skb(skb);
+ }
+ up(&rx_queue_sema);
+}
+
/*
- * Again, this is used for iSCSI names. Normally, we follow
- * the transport class convention of having the lld set
- * the field, but in these cases the value is too large.
+ * iSCSI connection attrs
*/
-#define iscsi_host_show_str_fn(field) \
- \
+#define iscsi_conn_int_attr_show(param, format) \
static ssize_t \
-show_host_str_##field(struct class_device *cdev, char *buf) \
+show_conn_int_param_##param(struct class_device *cdev, char *buf) \
{ \
- int ret = 0; \
- struct Scsi_Host *shost = transport_class_to_shost(cdev); \
- struct iscsi_internal *i = to_iscsi_internal(shost->transportt); \
+ uint32_t value = 0; \
+ struct iscsi_if_conn *conn = iscsi_cdev_to_if_conn(cdev); \
+ struct iscsi_internal *priv; \
\
- if (i->fnt->get_##field) \
- ret = i->fnt->get_##field(shost, buf, PAGE_SIZE); \
- return ret; \
+ priv = to_iscsi_internal(conn->host->transportt); \
+ if (priv->param_mask & (1 << param)) \
+ priv->iscsi_transport->get_param(conn->connh, param, &value); \
+ return snprintf(buf, 20, format"\n", value); \
}
-#define iscsi_host_rd_str_attr(field) \
- iscsi_host_show_str_fn(field) \
-static CLASS_DEVICE_ATTR(field, S_IRUGO, show_host_str_##field, NULL);
+#define iscsi_conn_int_attr(field, param, format) \
+ iscsi_conn_int_attr_show(param, format) \
+static CLASS_DEVICE_ATTR(field, S_IRUGO, show_conn_int_param_##param, NULL);
+
+iscsi_conn_int_attr(max_recv_dlength, ISCSI_PARAM_MAX_RECV_DLENGTH, "%u");
+iscsi_conn_int_attr(max_xmit_dlength, ISCSI_PARAM_MAX_XMIT_DLENGTH, "%u");
+iscsi_conn_int_attr(header_digest, ISCSI_PARAM_HDRDGST_EN, "%d");
+iscsi_conn_int_attr(data_digest, ISCSI_PARAM_DATADGST_EN, "%d");
+iscsi_conn_int_attr(ifmarker, ISCSI_PARAM_IFMARKER_EN, "%d");
+iscsi_conn_int_attr(ofmarker, ISCSI_PARAM_OFMARKER_EN, "%d");
-iscsi_host_rd_str_attr(initiator_name);
-iscsi_host_rd_str_attr(initiator_alias);
+/*
+ * iSCSI session attrs
+ */
+#define iscsi_session_int_attr_show(param, format) \
+static ssize_t \
+show_session_int_param_##param(struct class_device *cdev, char *buf) \
+{ \
+ uint32_t value = 0; \
+ struct iscsi_if_session *session = iscsi_cdev_to_if_session(cdev); \
+ struct Scsi_Host *shost = iscsi_if_session_to_shost(session); \
+ struct iscsi_internal *priv = to_iscsi_internal(shost->transportt); \
+ struct iscsi_if_conn *conn = NULL; \
+ unsigned long flags; \
+ \
+ spin_lock_irqsave(&connlock, flags); \
+ if (!list_empty(&session->connections)) { \
+ conn = list_entry(session->connections.next, \
+ struct iscsi_if_conn, session_list); \
+ spin_unlock_irqrestore(&connlock, flags); \
+ \
+ if (conn && (priv->param_mask & (1 << param))) \
+ priv->iscsi_transport->get_param(conn->connh, \
+ param, &value);\
+ } \
+ return snprintf(buf, 20, format"\n", value); \
+}
-#define SETUP_SESSION_RD_ATTR(field) \
- if (i->fnt->show_##field) { \
- i->session_attrs[count] = &class_device_attr_##field; \
+#define iscsi_session_int_attr(field, param, format) \
+ iscsi_session_int_attr_show(param, format) \
+static CLASS_DEVICE_ATTR(field, S_IRUGO, show_session_int_param_##param, NULL);
+
+iscsi_session_int_attr(initial_r2t, ISCSI_PARAM_INITIAL_R2T_EN, "%d");
+iscsi_session_int_attr(max_outstanding_r2t, ISCSI_PARAM_MAX_R2T, "%hu");
+iscsi_session_int_attr(immediate_data, ISCSI_PARAM_IMM_DATA_EN, "%d");
+iscsi_session_int_attr(first_burst_len, ISCSI_PARAM_FIRST_BURST, "%u");
+iscsi_session_int_attr(max_burst_len, ISCSI_PARAM_MAX_BURST, "%u");
+iscsi_session_int_attr(data_pdu_in_order, ISCSI_PARAM_PDU_INORDER_EN, "%d");
+iscsi_session_int_attr(data_seq_in_order, ISCSI_PARAM_DATASEQ_INORDER_EN, "%d");
+iscsi_session_int_attr(erl, ISCSI_PARAM_ERL, "%d");
+
+#define SETUP_SESSION_RD_ATTR(field, param) \
+ if (priv->param_mask & (1 << param)) { \
+ priv->session_attrs[count] = &class_device_attr_##field;\
count++; \
}
-#define SETUP_HOST_RD_ATTR(field) \
- if (i->fnt->show_##field) { \
- i->host_attrs[count] = &class_device_attr_##field; \
+#define SETUP_CONN_RD_ATTR(field, param) \
+ if (priv->param_mask & (1 << param)) { \
+ priv->conn_attrs[count] = &class_device_attr_##field; \
count++; \
}
-static int iscsi_host_match(struct attribute_container *cont,
- struct device *dev)
+static int iscsi_is_session_dev(const struct device *dev)
{
+ return dev->release == iscsi_if_session_dev_release;
+}
+
+static int iscsi_session_match(struct attribute_container *cont,
+ struct device *dev)
+{
+ struct iscsi_if_session *session;
struct Scsi_Host *shost;
- struct iscsi_internal *i;
+ struct iscsi_internal *priv;
+
+ if (!iscsi_is_session_dev(dev))
+ return 0;
- if (!scsi_is_host_device(dev))
+ session = iscsi_dev_to_if_session(dev);
+ shost = iscsi_if_session_to_shost(session);
+ if (!shost->transportt)
return 0;
- shost = dev_to_shost(dev);
- if (!shost->transportt || shost->transportt->host_attrs.ac.class
- != &iscsi_host_class.class)
+ priv = to_iscsi_internal(shost->transportt);
+ if (priv->session_cont.ac.class != &iscsi_session_class.class)
return 0;
- i = to_iscsi_internal(shost->transportt);
-
- return &i->t.host_attrs.ac == cont;
+ return &priv->session_cont.ac == cont;
}
-static int iscsi_target_match(struct attribute_container *cont,
- struct device *dev)
+static int iscsi_is_conn_dev(const struct device *dev)
{
+ return dev->release == iscsi_if_conn_dev_release;
+}
+
+static int iscsi_conn_match(struct attribute_container *cont,
+ struct device *dev)
+{
+ struct iscsi_if_conn *conn;
struct Scsi_Host *shost;
- struct iscsi_internal *i;
+ struct iscsi_internal *priv;
+
+ if (!iscsi_is_conn_dev(dev))
+ return 0;
- if (!scsi_is_target_device(dev))
+ conn = iscsi_dev_to_if_conn(dev);
+ shost = conn->host;
+ if (!shost->transportt)
return 0;
- shost = dev_to_shost(dev->parent);
- if (!shost->transportt || shost->transportt->host_attrs.ac.class
- != &iscsi_host_class.class)
+ priv = to_iscsi_internal(shost->transportt);
+ if (priv->conn_cont.ac.class != &iscsi_connection_class.class)
return 0;
- i = to_iscsi_internal(shost->transportt);
-
- return &i->t.target_attrs.ac == cont;
+ return &priv->conn_cont.ac == cont;
}
-struct scsi_transport_template *
-iscsi_attach_transport(struct iscsi_function_template *fnt)
+int iscsi_register_transport(struct iscsi_transport *tt)
{
- struct iscsi_internal *i = kmalloc(sizeof(struct iscsi_internal),
- GFP_KERNEL);
+ struct iscsi_internal *priv;
+ unsigned long flags;
int count = 0;
- if (unlikely(!i))
- return NULL;
+ BUG_ON(!tt);
- memset(i, 0, sizeof(struct iscsi_internal));
- i->fnt = fnt;
+ priv = iscsi_if_transport_lookup(tt);
+ if (priv)
+ return -EEXIST;
+
+ priv = kmalloc(sizeof(*priv), GFP_KERNEL);
+ if (!priv)
+ return -ENOMEM;
+ memset(priv, 0, sizeof(*priv));
+ INIT_LIST_HEAD(&priv->list);
+ INIT_LIST_HEAD(&priv->sessions);
+ spin_lock_init(&priv->session_lock);
+ priv->iscsi_transport = tt;
+
+ /* setup parameters mask */
+ priv->param_mask = 0xFFFFFFFF;
+ if (!(tt->caps & CAP_MULTI_R2T))
+ priv->param_mask &= ~(1 << ISCSI_PARAM_MAX_R2T);
+ if (!(tt->caps & CAP_HDRDGST))
+ priv->param_mask &= ~(1 << ISCSI_PARAM_HDRDGST_EN);
+ if (!(tt->caps & CAP_DATADGST))
+ priv->param_mask &= ~(1 << ISCSI_PARAM_DATADGST_EN);
+ if (!(tt->caps & CAP_MARKERS)) {
+ priv->param_mask &= ~(1 << ISCSI_PARAM_IFMARKER_EN);
+ priv->param_mask &= ~(1 << ISCSI_PARAM_OFMARKER_EN);
+ }
- i->t.target_attrs.ac.attrs = &i->session_attrs[0];
- i->t.target_attrs.ac.class = &iscsi_transport_class.class;
- i->t.target_attrs.ac.match = iscsi_target_match;
- transport_container_register(&i->t.target_attrs);
- i->t.target_size = sizeof(struct iscsi_class_session);
-
- SETUP_SESSION_RD_ATTR(tsih);
- SETUP_SESSION_RD_ATTR(isid);
- SETUP_SESSION_RD_ATTR(header_digest);
- SETUP_SESSION_RD_ATTR(data_digest);
- SETUP_SESSION_RD_ATTR(target_name);
- SETUP_SESSION_RD_ATTR(target_alias);
- SETUP_SESSION_RD_ATTR(port);
- SETUP_SESSION_RD_ATTR(tpgt);
- SETUP_SESSION_RD_ATTR(ip_address);
- SETUP_SESSION_RD_ATTR(initial_r2t);
- SETUP_SESSION_RD_ATTR(immediate_data);
- SETUP_SESSION_RD_ATTR(max_recv_data_segment_len);
- SETUP_SESSION_RD_ATTR(max_burst_len);
- SETUP_SESSION_RD_ATTR(first_burst_len);
- SETUP_SESSION_RD_ATTR(def_time2wait);
- SETUP_SESSION_RD_ATTR(def_time2retain);
- SETUP_SESSION_RD_ATTR(max_outstanding_r2t);
- SETUP_SESSION_RD_ATTR(data_pdu_in_order);
- SETUP_SESSION_RD_ATTR(data_sequence_in_order);
- SETUP_SESSION_RD_ATTR(erl);
+ /* connection parameters */
+ priv->conn_cont.ac.attrs = &priv->conn_attrs[0];
+ priv->conn_cont.ac.class = &iscsi_connection_class.class;
+ priv->conn_cont.ac.match = iscsi_conn_match;
+ transport_container_register(&priv->conn_cont);
+
+ SETUP_CONN_RD_ATTR(max_recv_dlength, ISCSI_PARAM_MAX_RECV_DLENGTH);
+ SETUP_CONN_RD_ATTR(max_xmit_dlength, ISCSI_PARAM_MAX_XMIT_DLENGTH);
+ SETUP_CONN_RD_ATTR(header_digest, ISCSI_PARAM_HDRDGST_EN);
+ SETUP_CONN_RD_ATTR(data_digest, ISCSI_PARAM_DATADGST_EN);
+ SETUP_CONN_RD_ATTR(ifmarker, ISCSI_PARAM_IFMARKER_EN);
+ SETUP_CONN_RD_ATTR(ofmarker, ISCSI_PARAM_OFMARKER_EN);
- BUG_ON(count > ISCSI_SESSION_ATTRS);
- i->session_attrs[count] = NULL;
+ BUG_ON(count > ISCSI_CONN_ATTRS);
+ priv->conn_attrs[count] = NULL;
+ count = 0;
- i->t.host_attrs.ac.attrs = &i->host_attrs[0];
- i->t.host_attrs.ac.class = &iscsi_host_class.class;
- i->t.host_attrs.ac.match = iscsi_host_match;
- transport_container_register(&i->t.host_attrs);
- i->t.host_size = 0;
+ /* session parameters */
+ priv->session_cont.ac.attrs = &priv->session_attrs[0];
+ priv->session_cont.ac.class = &iscsi_session_class.class;
+ priv->session_cont.ac.match = iscsi_session_match;
+ transport_container_register(&priv->session_cont);
+
+ SETUP_SESSION_RD_ATTR(initial_r2t, ISCSI_PARAM_INITIAL_R2T_EN);
+ SETUP_SESSION_RD_ATTR(max_outstanding_r2t, ISCSI_PARAM_MAX_R2T);
+ SETUP_SESSION_RD_ATTR(immediate_data, ISCSI_PARAM_IMM_DATA_EN);
+ SETUP_SESSION_RD_ATTR(first_burst_len, ISCSI_PARAM_FIRST_BURST);
+ SETUP_SESSION_RD_ATTR(max_burst_len, ISCSI_PARAM_MAX_BURST);
+ SETUP_SESSION_RD_ATTR(data_pdu_in_order, ISCSI_PARAM_PDU_INORDER_EN);
+ SETUP_SESSION_RD_ATTR(data_seq_in_order,ISCSI_PARAM_DATASEQ_INORDER_EN)
+ SETUP_SESSION_RD_ATTR(erl, ISCSI_PARAM_ERL);
- count = 0;
- SETUP_HOST_RD_ATTR(initiator_name);
- SETUP_HOST_RD_ATTR(initiator_alias);
+ BUG_ON(count > ISCSI_SESSION_ATTRS);
+ priv->session_attrs[count] = NULL;
- BUG_ON(count > ISCSI_HOST_ATTRS);
- i->host_attrs[count] = NULL;
+ spin_lock_irqsave(&iscsi_transport_lock, flags);
+ list_add(&priv->list, &iscsi_transports);
+ spin_unlock_irqrestore(&iscsi_transport_lock, flags);
- return &i->t;
+ printk(KERN_NOTICE "iscsi: registered transport (%s)\n", tt->name);
+ return 0;
}
+EXPORT_SYMBOL_GPL(iscsi_register_transport);
+
+int iscsi_unregister_transport(struct iscsi_transport *tt)
+{
+ struct iscsi_internal *priv;
+ unsigned long flags;
+
+ BUG_ON(!tt);
+
+ down(&rx_queue_sema);
+
+ priv = iscsi_if_transport_lookup(tt);
+ BUG_ON (!priv);
+
+ spin_lock_irqsave(&priv->session_lock, flags);
+ if (!list_empty(&priv->sessions)) {
+ spin_unlock_irqrestore(&priv->session_lock, flags);
+ up(&rx_queue_sema);
+ return -EPERM;
+ }
+ spin_unlock_irqrestore(&priv->session_lock, flags);
+
+ spin_lock_irqsave(&iscsi_transport_lock, flags);
+ list_del(&priv->list);
+ spin_unlock_irqrestore(&iscsi_transport_lock, flags);
+
+ transport_container_unregister(&priv->conn_cont);
+ transport_container_unregister(&priv->session_cont);
+ kfree(priv);
+ up(&rx_queue_sema);
-EXPORT_SYMBOL(iscsi_attach_transport);
+ return 0;
+}
+EXPORT_SYMBOL_GPL(iscsi_unregister_transport);
-void iscsi_release_transport(struct scsi_transport_template *t)
+static int
+iscsi_rcv_nl_event(struct notifier_block *this, unsigned long event, void *ptr)
{
- struct iscsi_internal *i = to_iscsi_internal(t);
+ struct netlink_notify *n = ptr;
+
+ if (event == NETLINK_URELEASE &&
+ n->protocol == NETLINK_ISCSI && n->pid) {
+ struct iscsi_if_conn *conn;
+ unsigned long flags;
+
+ mempool_zone_complete(&z_reply);
+ spin_lock_irqsave(&connlock, flags);
+ list_for_each_entry(conn, &connlist, conn_list) {
+ mempool_zone_complete(&conn->z_error);
+ mempool_zone_complete(&conn->z_pdu);
+ }
+ spin_unlock_irqrestore(&connlock, flags);
+ }
- transport_container_unregister(&i->t.target_attrs);
- transport_container_unregister(&i->t.host_attrs);
-
- kfree(i);
+ return NOTIFY_DONE;
}
-EXPORT_SYMBOL(iscsi_release_transport);
+static struct notifier_block iscsi_nl_notifier = {
+ .notifier_call = iscsi_rcv_nl_event,
+};
static __init int iscsi_transport_init(void)
{
- int err = transport_class_register(&iscsi_transport_class);
+ int err;
+ err = transport_class_register(&iscsi_connection_class);
if (err)
return err;
- return transport_class_register(&iscsi_host_class);
+
+ err = transport_class_register(&iscsi_session_class);
+ if (err)
+ goto unregister_conn_class;
+
+ netlink_register_notifier(&iscsi_nl_notifier);
+ nls = netlink_kernel_create(NETLINK_ISCSI, iscsi_if_rx);
+ if (!nls) {
+ err = -ENOBUFS;
+ goto unregister_notifier;
+ }
+
+ err = mempool_zone_init(&z_reply, Z_MAX_REPLY,
+ NLMSG_SPACE(sizeof(struct iscsi_uevent)), Z_HIWAT_REPLY);
+ if (!err)
+ return 0;
+
+ sock_release(nls->sk_socket);
+ unregister_notifier:
+ netlink_unregister_notifier(&iscsi_nl_notifier);
+ unregister_conn_class:
+ transport_class_unregister(&iscsi_connection_class);
+ return err;
}
static void __exit iscsi_transport_exit(void)
{
- transport_class_unregister(&iscsi_host_class);
- transport_class_unregister(&iscsi_transport_class);
+ mempool_destroy(z_reply.pool);
+ sock_release(nls->sk_socket);
+ netlink_unregister_notifier(&iscsi_nl_notifier);
+ transport_class_unregister(&iscsi_connection_class);
+ transport_class_unregister(&iscsi_session_class);
}
module_init(iscsi_transport_init);
module_exit(iscsi_transport_exit);
-MODULE_AUTHOR("Mike Christie");
-MODULE_DESCRIPTION("iSCSI Transport Attributes");
+MODULE_AUTHOR("Mike Christie <michaelc@cs.wisc.edu>, "
+ "Dmitry Yusupov <dmitry_yus@yahoo.com>, "
+ "Alex Aizman <itn780@yahoo.com>");
+MODULE_DESCRIPTION("iSCSI Transport Interface");
MODULE_LICENSE("GPL");
diff -aurp linux-2.6.12-rc4/include/scsi/scsi_transport_iscsi.h linux-2.6.12-rc4.work/include/scsi/scsi_transport_iscsi.h
--- linux-2.6.12-rc4/include/scsi/scsi_transport_iscsi.h 2005-05-06 22:20:31.000000000 -0700
+++ linux-2.6.12-rc4.work/include/scsi/scsi_transport_iscsi.h 2005-05-14 12:19:35.000000000 -0700
@@ -1,8 +1,8 @@
-/*
+/*
* iSCSI transport class definitions
*
* Copyright (C) IBM Corporation, 2004
- * Copyright (C) Mike Christie, 2004
+ * Copyright (C) Mike Christie, Dmitry Yusupov, Alex Aizman, 2004 - 2005
*
* 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
@@ -21,158 +21,68 @@
#ifndef SCSI_TRANSPORT_ISCSI_H
#define SCSI_TRANSPORT_ISCSI_H
-#include <linux/config.h>
-#include <linux/in6.h>
-#include <linux/in.h>
-
-struct scsi_transport_template;
+#include <scsi/iscsi_if.h>
-struct iscsi_class_session {
- uint8_t isid[6];
- uint16_t tsih;
- int header_digest; /* 1 CRC32, 0 None */
- int data_digest; /* 1 CRC32, 0 None */
- uint16_t tpgt;
- union {
- struct in6_addr sin6_addr;
- struct in_addr sin_addr;
- } u;
- sa_family_t addr_type; /* must be AF_INET or AF_INET6 */
- uint16_t port; /* must be in network byte order */
- int initial_r2t; /* 1 Yes, 0 No */
- int immediate_data; /* 1 Yes, 0 No */
- uint32_t max_recv_data_segment_len;
- uint32_t max_burst_len;
- uint32_t first_burst_len;
- uint16_t def_time2wait;
- uint16_t def_time2retain;
- uint16_t max_outstanding_r2t;
- int data_pdu_in_order; /* 1 Yes, 0 No */
- int data_sequence_in_order; /* 1 Yes, 0 No */
- int erl;
+/**
+ * struct iscsi_transport - iSCSI Transport template
+ *
+ * @name: transport name
+ * @caps: iSCSI Data-Path capabilities
+ * @create_session: create new iSCSI session object
+ * @destroy_session: destroy existing iSCSI session object
+ * @create_conn: create new iSCSI connection
+ * @bind_conn: associate this connection with existing iSCSI session
+ * and specified transport descriptor
+ * @destroy_conn: destroy inactive iSCSI connection
+ * @set_param: set iSCSI Data-Path operational parameter
+ * @start_conn: set connection to be operational
+ * @stop_conn: suspend/recover/terminate connection
+ * @send_pdu: send iSCSI PDU, Login, Logout, NOP-Out, Reject, Text.
+ *
+ * Template API provided by iSCSI Transport
+ */
+struct iscsi_transport {
+ struct module *owner;
+ char *name;
+ unsigned int caps;
+ struct scsi_host_template *host_template;
+ int hostdata_size;
+ int max_lun;
+ unsigned int max_conn;
+ unsigned int max_cmd_len;
+ iscsi_sessionh_t (*create_session) (uint32_t initial_cmdsn,
+ struct Scsi_Host *shost);
+ void (*destroy_session) (iscsi_sessionh_t session);
+ iscsi_connh_t (*create_conn) (iscsi_sessionh_t session, uint32_t cid);
+ int (*bind_conn) (iscsi_sessionh_t session, iscsi_connh_t conn,
+ uint32_t transport_fd, int is_leading);
+ int (*start_conn) (iscsi_connh_t conn);
+ void (*stop_conn) (iscsi_connh_t conn, int flag);
+ void (*destroy_conn) (iscsi_connh_t conn);
+ int (*set_param) (iscsi_connh_t conn, enum iscsi_param param,
+ uint32_t value);
+ int (*get_param) (iscsi_connh_t conn, enum iscsi_param param,
+ uint32_t *value);
+ int (*send_pdu) (iscsi_connh_t conn, struct iscsi_hdr *hdr,
+ char *data, uint32_t data_size);
+ void (*get_stats) (iscsi_connh_t conn, struct iscsi_stats *stats);
+ /*
+ * transport class private data
+ */
+ struct scsi_transport_template *scsi_transport;
};
/*
- * accessor macros
+ * transport registration upcalls
*/
-#define iscsi_isid(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->isid)
-#define iscsi_tsih(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->tsih)
-#define iscsi_header_digest(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->header_digest)
-#define iscsi_data_digest(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->data_digest)
-#define iscsi_port(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->port)
-#define iscsi_addr_type(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->addr_type)
-#define iscsi_sin_addr(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->u.sin_addr)
-#define iscsi_sin6_addr(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->u.sin6_addr)
-#define iscsi_tpgt(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->tpgt)
-#define iscsi_initial_r2t(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->initial_r2t)
-#define iscsi_immediate_data(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->immediate_data)
-#define iscsi_max_recv_data_segment_len(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->max_recv_data_segment_len)
-#define iscsi_max_burst_len(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->max_burst_len)
-#define iscsi_first_burst_len(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->first_burst_len)
-#define iscsi_def_time2wait(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->def_time2wait)
-#define iscsi_def_time2retain(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->def_time2retain)
-#define iscsi_max_outstanding_r2t(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->max_outstanding_r2t)
-#define iscsi_data_pdu_in_order(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->data_pdu_in_order)
-#define iscsi_data_sequence_in_order(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->data_sequence_in_order)
-#define iscsi_erl(x) \
- (((struct iscsi_class_session *)&(x)->starget_data)->erl)
+extern int iscsi_register_transport(struct iscsi_transport *tt);
+extern int iscsi_unregister_transport(struct iscsi_transport *tt);
/*
- * The functions by which the transport class and the driver communicate
+ * control plane upcalls
*/
-struct iscsi_function_template {
- /*
- * target attrs
- */
- void (*get_isid)(struct scsi_target *);
- void (*get_tsih)(struct scsi_target *);
- void (*get_header_digest)(struct scsi_target *);
- void (*get_data_digest)(struct scsi_target *);
- void (*get_port)(struct scsi_target *);
- void (*get_tpgt)(struct scsi_target *);
- /*
- * In get_ip_address the lld must set the address and
- * the address type
- */
- void (*get_ip_address)(struct scsi_target *);
- /*
- * The lld should snprintf the name or alias to the buffer
- */
- ssize_t (*get_target_name)(struct scsi_target *, char *, ssize_t);
- ssize_t (*get_target_alias)(struct scsi_target *, char *, ssize_t);
- void (*get_initial_r2t)(struct scsi_target *);
- void (*get_immediate_data)(struct scsi_target *);
- void (*get_max_recv_data_segment_len)(struct scsi_target *);
- void (*get_max_burst_len)(struct scsi_target *);
- void (*get_first_burst_len)(struct scsi_target *);
- void (*get_def_time2wait)(struct scsi_target *);
- void (*get_def_time2retain)(struct scsi_target *);
- void (*get_max_outstanding_r2t)(struct scsi_target *);
- void (*get_data_pdu_in_order)(struct scsi_target *);
- void (*get_data_sequence_in_order)(struct scsi_target *);
- void (*get_erl)(struct scsi_target *);
-
- /*
- * host atts
- */
-
- /*
- * The lld should snprintf the name or alias to the buffer
- */
- ssize_t (*get_initiator_alias)(struct Scsi_Host *, char *, ssize_t);
- ssize_t (*get_initiator_name)(struct Scsi_Host *, char *, ssize_t);
- /*
- * The driver sets these to tell the transport class it
- * wants the attributes displayed in sysfs. If the show_ flag
- * is not set, the attribute will be private to the transport
- * class. We could probably just test if a get_ fn was set
- * since we only use the values for sysfs but this is how
- * fc does it too.
- */
- unsigned long show_isid:1;
- unsigned long show_tsih:1;
- unsigned long show_header_digest:1;
- unsigned long show_data_digest:1;
- unsigned long show_port:1;
- unsigned long show_tpgt:1;
- unsigned long show_ip_address:1;
- unsigned long show_target_name:1;
- unsigned long show_target_alias:1;
- unsigned long show_initial_r2t:1;
- unsigned long show_immediate_data:1;
- unsigned long show_max_recv_data_segment_len:1;
- unsigned long show_max_burst_len:1;
- unsigned long show_first_burst_len:1;
- unsigned long show_def_time2wait:1;
- unsigned long show_def_time2retain:1;
- unsigned long show_max_outstanding_r2t:1;
- unsigned long show_data_pdu_in_order:1;
- unsigned long show_data_sequence_in_order:1;
- unsigned long show_erl:1;
- unsigned long show_initiator_name:1;
- unsigned long show_initiator_alias:1;
-};
-
-struct scsi_transport_template *iscsi_attach_transport(struct iscsi_function_template *);
-void iscsi_release_transport(struct scsi_transport_template *);
+extern void iscsi_conn_error(iscsi_connh_t conn, enum iscsi_err error);
+extern int iscsi_recv_pdu(iscsi_connh_t conn, struct iscsi_hdr *hdr,
+ char *data, uint32_t data_size);
#endif
^ permalink raw reply
* [PATCH 2/3] add open iscsi netlink interface to iscsi transport class
From: Mike Christie @ 2005-05-15 19:30 UTC (permalink / raw)
To: open-iscsi, netdev, linux-scsi, linux-kernel, inux-iscsi-devel
Common iSCSI header files:
- iscsi_if.h (user/kernel #defines);
- iscsi_proto.h (RFC3720 #defines and types);
- iscsi_ifev.h (user/kernel events).
Thanks,
Linux-iscsi Team
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
diff -Naurp linux-2.6.12-rc4/include/scsi/iscsi_ifev.h linux-2.6.12-rc4.work/include/scsi/iscsi_ifev.h
--- linux-2.6.12-rc4/include/scsi/iscsi_ifev.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-2.6.12-rc4.work/include/scsi/iscsi_ifev.h 2005-05-14 13:19:45.000000000 -0700
@@ -0,0 +1,121 @@
+/*
+ * iSCSI Kernel/User Interface Events
+ *
+ * Copyright (C) 2005 Dmitry Yusupov, Alex Aizman
+ * maintained by open-iscsi@googlegroups.com
+ *
+ * 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.
+ *
+ * See the file COPYING included with this distribution for more details.
+ */
+
+#ifndef ISCSI_IFEV_H
+#define ISCSI_IFEV_H
+
+enum iscsi_uevent_e {
+ ISCSI_UEVENT_UNKNOWN = 0,
+
+ /* down events */
+ ISCSI_UEVENT_CREATE_SESSION = UEVENT_BASE + 1,
+ ISCSI_UEVENT_DESTROY_SESSION = UEVENT_BASE + 2,
+ ISCSI_UEVENT_CREATE_CONN = UEVENT_BASE + 3,
+ ISCSI_UEVENT_DESTROY_CONN = UEVENT_BASE + 4,
+ ISCSI_UEVENT_BIND_CONN = UEVENT_BASE + 5,
+ ISCSI_UEVENT_SET_PARAM = UEVENT_BASE + 6,
+ ISCSI_UEVENT_START_CONN = UEVENT_BASE + 7,
+ ISCSI_UEVENT_STOP_CONN = UEVENT_BASE + 8,
+ ISCSI_UEVENT_SEND_PDU = UEVENT_BASE + 9,
+ ISCSI_UEVENT_TRANS_LIST = UEVENT_BASE + 10,
+ ISCSI_UEVENT_GET_STATS = UEVENT_BASE + 11,
+ ISCSI_UEVENT_GET_PARAM = UEVENT_BASE + 12,
+
+ /* up events */
+ ISCSI_KEVENT_RECV_PDU = KEVENT_BASE + 1,
+ ISCSI_KEVENT_CONN_ERROR = KEVENT_BASE + 2,
+ ISCSI_KEVENT_IF_ERROR = KEVENT_BASE + 3,
+};
+
+struct iscsi_uevent {
+ uint32_t type; /* k/u events type */
+ uint32_t iferror; /* carries interface or resource errors */
+ uint64_t transport_handle;
+
+ union {
+ /* messages u -> k */
+ struct msg_create_session {
+ uint32_t initial_cmdsn;
+ } c_session;
+ struct msg_destroy_session {
+ uint64_t session_handle;
+ uint32_t sid;
+ } d_session;
+ struct msg_create_conn {
+ uint64_t session_handle;
+ uint32_t cid;
+ uint32_t sid;
+ } c_conn;
+ struct msg_bind_conn {
+ uint64_t session_handle;
+ uint64_t conn_handle;
+ uint32_t transport_fd;
+ uint32_t is_leading;
+ } b_conn;
+ struct msg_destroy_conn {
+ uint64_t conn_handle;
+ uint32_t cid;
+ } d_conn;
+ struct msg_send_pdu {
+ uint32_t hdr_size;
+ uint32_t data_size;
+ uint64_t conn_handle;
+ } send_pdu;
+ struct msg_set_param {
+ uint64_t conn_handle;
+ uint32_t param; /* enum iscsi_param */
+ uint32_t value;
+ } set_param;
+ struct msg_start_conn {
+ uint64_t conn_handle;
+ } start_conn;
+ struct msg_stop_conn {
+ uint64_t conn_handle;
+ uint32_t flag;
+ } stop_conn;
+ struct msg_get_stats {
+ uint64_t conn_handle;
+ } get_stats;
+ } u;
+ union {
+ /* messages k -> u */
+ uint64_t handle;
+ int retcode;
+ struct msg_create_session_ret {
+ uint64_t session_handle;
+ uint32_t sid;
+ } c_session_ret;
+ struct msg_recv_req {
+ uint64_t recv_handle;
+ uint64_t conn_handle;
+ } recv_req;
+ struct msg_conn_error {
+ uint64_t conn_handle;
+ uint32_t error; /* enum iscsi_err */
+ } connerror;
+ struct msg_trans_list {
+ struct {
+ uint64_t trans_handle;
+ char name[ISCSI_TRANSPORT_NAME_MAXLEN];
+ } elements[ISCSI_TRANSPORT_MAX];
+ } t_list;
+ } r;
+} __attribute__ ((aligned (sizeof(uint64_t))));
+
+#endif /* ISCSI_IFEV_H */
diff -Naurp linux-2.6.12-rc4/include/scsi/iscsi_if.h linux-2.6.12-rc4.work/include/scsi/iscsi_if.h
--- linux-2.6.12-rc4/include/scsi/iscsi_if.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-2.6.12-rc4.work/include/scsi/iscsi_if.h 2005-05-14 13:20:22.000000000 -0700
@@ -0,0 +1,156 @@
+/*
+ * iSCSI User/Kernel Shares (Defines, Constants, Protocol definitions, etc)
+ *
+ * Copyright (C) 2005 Dmitry Yusupov, Alex Aizman
+ * maintained by open-iscsi@googlegroups.com
+ *
+ * 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.
+ *
+ * See the file COPYING included with this distribution for more details.
+ */
+
+#ifndef ISCSI_IF_H
+#define ISCSI_IF_H
+
+#include <scsi/iscsi_proto.h>
+
+#define ISCSI_TRANSPORT_NAME_MAXLEN 16
+#define ISCSI_TRANSPORT_MAX 1
+#define UEVENT_BASE 10
+#define KEVENT_BASE 100
+#define ISCSI_ERR_BASE 1000
+
+/*
+ * Common error codes
+ */
+enum iscsi_err {
+ ISCSI_OK = 0,
+
+ ISCSI_ERR_DATASN = ISCSI_ERR_BASE + 1,
+ ISCSI_ERR_DATA_OFFSET = ISCSI_ERR_BASE + 2,
+ ISCSI_ERR_MAX_CMDSN = ISCSI_ERR_BASE + 3,
+ ISCSI_ERR_EXP_CMDSN = ISCSI_ERR_BASE + 4,
+ ISCSI_ERR_BAD_OPCODE = ISCSI_ERR_BASE + 5,
+ ISCSI_ERR_DATALEN = ISCSI_ERR_BASE + 6,
+ ISCSI_ERR_AHSLEN = ISCSI_ERR_BASE + 7,
+ ISCSI_ERR_PROTO = ISCSI_ERR_BASE + 8,
+ ISCSI_ERR_LUN = ISCSI_ERR_BASE + 9,
+ ISCSI_ERR_BAD_ITT = ISCSI_ERR_BASE + 10,
+ ISCSI_ERR_CONN_FAILED = ISCSI_ERR_BASE + 11,
+ ISCSI_ERR_R2TSN = ISCSI_ERR_BASE + 12,
+ ISCSI_ERR_SESSION_FAILED = ISCSI_ERR_BASE + 13,
+ ISCSI_ERR_HDR_DGST = ISCSI_ERR_BASE + 14,
+ ISCSI_ERR_DATA_DGST = ISCSI_ERR_BASE + 15,
+ ISCSI_ERR_PDU_GATHER_FAILED = ISCSI_ERR_BASE + 16,
+ ISCSI_ERR_PARAM_NOT_FOUND = ISCSI_ERR_BASE + 17
+};
+
+/*
+ * iSCSI Parameters (RFC3720)
+ */
+enum iscsi_param {
+ ISCSI_PARAM_MAX_RECV_DLENGTH = 0,
+ ISCSI_PARAM_MAX_XMIT_DLENGTH = 1,
+ ISCSI_PARAM_HDRDGST_EN = 2,
+ ISCSI_PARAM_DATADGST_EN = 3,
+ ISCSI_PARAM_INITIAL_R2T_EN = 4,
+ ISCSI_PARAM_MAX_R2T = 5,
+ ISCSI_PARAM_IMM_DATA_EN = 6,
+ ISCSI_PARAM_FIRST_BURST = 7,
+ ISCSI_PARAM_MAX_BURST = 8,
+ ISCSI_PARAM_PDU_INORDER_EN = 9,
+ ISCSI_PARAM_DATASEQ_INORDER_EN = 10,
+ ISCSI_PARAM_ERL = 11,
+ ISCSI_PARAM_IFMARKER_EN = 12,
+ ISCSI_PARAM_OFMARKER_EN = 13,
+};
+#define ISCSI_PARAM_MAX 14
+
+typedef uint64_t iscsi_sessionh_t; /* iSCSI Data-Path session handle */
+typedef uint64_t iscsi_connh_t; /* iSCSI Data-Path connection handle */
+
+#define iscsi_ptr(_handle) ((void*)(unsigned long)_handle)
+#define iscsi_handle(_ptr) ((uint64_t)(unsigned long)_ptr)
+#define iscsi_hostdata(_hostdata) ((void*)_hostdata + sizeof(unsigned long))
+
+/*
+ * These flags presents iSCSI Data-Path capabilities.
+ */
+#define CAP_RECOVERY_L0 0x1
+#define CAP_RECOVERY_L1 0x2
+#define CAP_RECOVERY_L2 0x4
+#define CAP_MULTI_R2T 0x8
+#define CAP_HDRDGST 0x10
+#define CAP_DATADGST 0x20
+#define CAP_MULTI_CONN 0x40
+#define CAP_TEXT_NEGO 0x80
+#define CAP_MARKERS 0x100
+
+/*
+ * These flags describes reason of stop_conn() call
+ */
+#define STOP_CONN_TERM 0x1
+#define STOP_CONN_SUSPEND 0x2
+#define STOP_CONN_RECOVER 0x3
+
+#define ISCSI_STATS_CUSTOM_MAX 32
+#define ISCSI_STATS_CUSTOM_DESC_MAX 64
+struct iscsi_stats_custom {
+ char desc[ISCSI_STATS_CUSTOM_DESC_MAX];
+ uint64_t value;
+};
+
+/*
+ * struct iscsi_stats - iSCSI Statistics (iSCSI MIB)
+ *
+ * Note: this structure contains counters collected on per-connection basis.
+ */
+struct iscsi_stats {
+ /* octets */
+ uint64_t txdata_octets;
+ uint64_t rxdata_octets;
+
+ /* xmit pdus */
+ uint32_t noptx_pdus;
+ uint32_t scsicmd_pdus;
+ uint32_t tmfcmd_pdus;
+ uint32_t login_pdus;
+ uint32_t text_pdus;
+ uint32_t dataout_pdus;
+ uint32_t logout_pdus;
+ uint32_t snack_pdus;
+
+ /* recv pdus */
+ uint32_t noprx_pdus;
+ uint32_t scsirsp_pdus;
+ uint32_t tmfrsp_pdus;
+ uint32_t textrsp_pdus;
+ uint32_t datain_pdus;
+ uint32_t logoutrsp_pdus;
+ uint32_t r2t_pdus;
+ uint32_t async_pdus;
+ uint32_t rjt_pdus;
+
+ /* errors */
+ uint32_t digest_err;
+ uint32_t timeout_err;
+
+ /*
+ * iSCSI Custom Statistics support, i.e. Transport could
+ * extend existing MIB statistics with its own specific statistics
+ * up to ISCSI_STATS_CUSTOM_MAX
+ */
+ uint32_t custom_length;
+ struct iscsi_stats_custom custom[0]
+ __attribute__ ((aligned (sizeof(uint64_t))));
+};
+
+#endif
diff -Naurp linux-2.6.12-rc4/include/scsi/iscsi_proto.h linux-2.6.12-rc4.work/include/scsi/iscsi_proto.h
--- linux-2.6.12-rc4/include/scsi/iscsi_proto.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-2.6.12-rc4.work/include/scsi/iscsi_proto.h 2005-05-14 13:19:45.000000000 -0700
@@ -0,0 +1,565 @@
+/*
+ * RFC 3720 (iSCSI) protocol data types
+ *
+ * Copyright (C) 2005 Dmitry Yusupov, Alex Aizman
+ * maintained by open-iscsi@googlegroups.com
+ *
+ * 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.
+ *
+ * See the file COPYING included with this distribution for more details.
+ */
+
+#ifndef ISCSI_PROTO_H
+#define ISCSI_PROTO_H
+
+#define ISCSI_VERSION_STR "0.3"
+#define ISCSI_DATE_STR "22-Apr-2005"
+#define ISCSI_DRAFT20_VERSION 0x00
+
+/* default iSCSI listen port for incoming connections */
+#define ISCSI_LISTEN_PORT 3260
+
+/* Padding word length */
+#define PAD_WORD_LEN 4
+
+/*
+ * useful common(control and data pathes) macro
+ */
+#define ntoh24(p) (((p)[0] << 16) | ((p)[1] << 8) | ((p)[2]))
+#define hton24(p, v) { \
+ p[0] = (((v) >> 16) & 0xFF); \
+ p[1] = (((v) >> 8) & 0xFF); \
+ p[2] = ((v) & 0xFF); \
+}
+#define zero_data(p) {p[0]=0;p[1]=0;p[2]=0;}
+
+/*
+ * iSCSI Template Message Header
+ */
+struct iscsi_hdr {
+ uint8_t opcode;
+ uint8_t flags; /* Final bit */
+ uint8_t rsvd2[2];
+ uint8_t hlength; /* AHSs total length */
+ uint8_t dlength[3]; /* Data length */
+ uint8_t lun[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 ttt; /* Target Task Tag */
+ __be32 statsn;
+ __be32 exp_statsn;
+ uint8_t other[16];
+};
+
+/************************* RFC 3720 Begin *****************************/
+
+#define ISCSI_RESERVED_TAG 0xffffffff
+
+/* Opcode encoding bits */
+#define ISCSI_OP_RETRY 0x80
+#define ISCSI_OP_IMMEDIATE 0x40
+#define ISCSI_OPCODE_MASK 0x3F
+
+/* Initiator Opcode values */
+#define ISCSI_OP_NOOP_OUT 0x00
+#define ISCSI_OP_SCSI_CMD 0x01
+#define ISCSI_OP_SCSI_TMFUNC 0x02
+#define ISCSI_OP_LOGIN 0x03
+#define ISCSI_OP_TEXT 0x04
+#define ISCSI_OP_SCSI_DATA_OUT 0x05
+#define ISCSI_OP_LOGOUT 0x06
+#define ISCSI_OP_SNACK 0x10
+
+/* Target Opcode values */
+#define ISCSI_OP_NOOP_IN 0x20
+#define ISCSI_OP_SCSI_CMD_RSP 0x21
+#define ISCSI_OP_SCSI_TMFUNC_RSP 0x22
+#define ISCSI_OP_LOGIN_RSP 0x23
+#define ISCSI_OP_TEXT_RSP 0x24
+#define ISCSI_OP_SCSI_DATA_IN 0x25
+#define ISCSI_OP_LOGOUT_RSP 0x26
+#define ISCSI_OP_R2T 0x31
+#define ISCSI_OP_ASYNC_EVENT 0x32
+#define ISCSI_OP_REJECT 0x3f
+
+/* SCSI Command Header */
+struct iscsi_cmd {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd2;
+ uint8_t cmdrn;
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t lun[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 data_length;
+ __be32 cmdsn;
+ __be32 exp_statsn;
+ uint8_t cdb[16]; /* SCSI Command Block */
+ /* Additional Data (Command Dependent) */
+};
+
+/* Command PDU flags */
+#define ISCSI_FLAG_CMD_FINAL 0x80
+#define ISCSI_FLAG_CMD_READ 0x40
+#define ISCSI_FLAG_CMD_WRITE 0x20
+#define ISCSI_FLAG_CMD_ATTR_MASK 0x07 /* 3 bits */
+
+/* SCSI Command Attribute values */
+#define ISCSI_ATTR_UNTAGGED 0
+#define ISCSI_ATTR_SIMPLE 1
+#define ISCSI_ATTR_ORDERED 2
+#define ISCSI_ATTR_HEAD_OF_QUEUE 3
+#define ISCSI_ATTR_ACA 4
+
+/* SCSI Response Header */
+struct iscsi_cmd_rsp {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t response;
+ uint8_t cmd_status;
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t rsvd[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 rsvd1;
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ __be32 exp_datasn;
+ __be32 bi_residual_count;
+ __be32 residual_count;
+ /* Response or Sense Data (optional) */
+};
+
+/* Command Response PDU flags */
+#define ISCSI_FLAG_CMD_BIDI_OVERFLOW 0x10
+#define ISCSI_FLAG_CMD_BIDI_UNDERFLOW 0x08
+#define ISCSI_FLAG_CMD_OVERFLOW 0x04
+#define ISCSI_FLAG_CMD_UNDERFLOW 0x02
+
+/* iSCSI Status values. Valid if Rsp Selector bit is not set */
+#define ISCSI_STATUS_CMD_COMPLETED 0
+#define ISCSI_STATUS_TARGET_FAILURE 1
+#define ISCSI_STATUS_SUBSYS_FAILURE 2
+
+/* Asynchronous Event Header */
+struct iscsi_async {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd2[2];
+ uint8_t rsvd3;
+ uint8_t dlength[3];
+ uint8_t lun[8];
+ uint8_t rsvd4[8];
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ uint8_t async_event;
+ uint8_t async_vcode;
+ __be16 param1;
+ __be16 param2;
+ __be16 param3;
+ uint8_t rsvd5[4];
+};
+
+/* iSCSI Event Codes */
+#define ISCSI_ASYNC_MSG_SCSI_EVENT 0
+#define ISCSI_ASYNC_MSG_REQUEST_LOGOUT 1
+#define ISCSI_ASYNC_MSG_DROPPING_CONNECTION 2
+#define ISCSI_ASYNC_MSG_DROPPING_ALL_CONNECTIONS 3
+#define ISCSI_ASYNC_MSG_PARAM_NEGOTIATION 4
+#define ISCSI_ASYNC_MSG_VENDOR_SPECIFIC 255
+
+/* NOP-Out Message */
+struct iscsi_nopout {
+ uint8_t opcode;
+ uint8_t flags;
+ __be16 rsvd2;
+ uint8_t rsvd3;
+ uint8_t dlength[3];
+ uint8_t lun[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 ttt; /* Target Transfer Tag */
+ __be32 cmdsn;
+ __be32 exp_statsn;
+ uint8_t rsvd4[16];
+};
+
+/* NOP-In Message */
+struct iscsi_nopin {
+ uint8_t opcode;
+ uint8_t flags;
+ __be16 rsvd2;
+ uint8_t rsvd3;
+ uint8_t dlength[3];
+ uint8_t lun[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 ttt; /* Target Transfer Tag */
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ uint8_t rsvd4[12];
+};
+
+/* SCSI Task Management Message Header */
+struct iscsi_tm {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd1[2];
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t lun[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 rtt; /* Reference Task Tag */
+ __be32 cmdsn;
+ __be32 exp_statsn;
+ __be32 refcmdsn;
+ __be32 exp_datasn;
+ uint8_t rsvd2[8];
+};
+
+#define ISCSI_FLAG_TASK_MGMT_FUNCTION_MASK 0x7F
+
+/* Function values */
+#define ISCSI_TM_FUNC_ABORT_TASK 1
+#define ISCSI_TM_FUNC_ABORT_TASK_SET 2
+#define ISCSI_TM_FUNC_CLEAR_ACA 3
+#define ISCSI_TM_FUNC_CLEAR_TASK_SET 4
+#define ISCSI_TM_FUNC_LOGICAL_UNIT_RESET 5
+#define ISCSI_TM_FUNC_TARGET_WARM_RESET 6
+#define ISCSI_TM_FUNC_TARGET_COLD_RESET 7
+#define ISCSI_TM_FUNC_TASK_REASSIGN 8
+
+/* SCSI Task Management Response Header */
+struct iscsi_tm_rsp {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t response; /* see Response values below */
+ uint8_t qualifier;
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t rsvd2[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 rtt; /* Reference Task Tag */
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ uint8_t rsvd3[12];
+};
+
+/* Response values */
+#define SCSI_TCP_TM_RESP_COMPLETE 0x00
+#define SCSI_TCP_TM_RESP_NO_TASK 0x01
+#define SCSI_TCP_TM_RESP_NO_LUN 0x02
+#define SCSI_TCP_TM_RESP_TASK_ALLEGIANT 0x03
+#define SCSI_TCP_TM_RESP_NO_FAILOVER 0x04
+#define SCSI_TCP_TM_RESP_NOT_SUPPORTED 0x05
+#define SCSI_TCP_TM_RESP_AUTH_FAILED 0x06
+#define SCSI_TCP_TM_RESP_REJECTED 0xff
+
+/* Ready To Transfer Header */
+struct iscsi_r2t_rsp {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd2[2];
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t lun[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 ttt; /* Target Transfer Tag */
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ __be32 r2tsn;
+ __be32 data_offset;
+ __be32 data_length;
+};
+
+/* SCSI Data Hdr */
+struct iscsi_data {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd2[2];
+ uint8_t rsvd3;
+ uint8_t dlength[3];
+ uint8_t lun[8];
+ __be32 itt;
+ __be32 ttt;
+ __be32 rsvd4;
+ __be32 exp_statsn;
+ __be32 rsvd5;
+ __be32 datasn;
+ __be32 offset;
+ __be32 rsvd6;
+ /* Payload */
+};
+
+/* SCSI Data Response Hdr */
+struct iscsi_data_rsp {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd2;
+ uint8_t cmd_status;
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t lun[8];
+ __be32 itt;
+ __be32 ttt;
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ __be32 datasn;
+ __be32 offset;
+ __be32 residual_count;
+};
+
+/* Data Response PDU flags */
+#define ISCSI_FLAG_DATA_ACK 0x40
+#define ISCSI_FLAG_DATA_OVERFLOW 0x04
+#define ISCSI_FLAG_DATA_UNDERFLOW 0x02
+#define ISCSI_FLAG_DATA_STATUS 0x01
+
+/* Text Header */
+struct iscsi_text {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd2[2];
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t rsvd4[8];
+ __be32 itt;
+ __be32 ttt;
+ __be32 cmdsn;
+ __be32 exp_statsn;
+ uint8_t rsvd5[16];
+ /* Text - key=value pairs */
+};
+
+#define ISCSI_FLAG_TEXT_CONTINUE 0x40
+
+/* Text Response Header */
+struct iscsi_text_rsp {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd2[2];
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t rsvd4[8];
+ __be32 itt;
+ __be32 ttt;
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ uint8_t rsvd5[12];
+ /* Text Response - key:value pairs */
+};
+
+/* Login Header */
+struct iscsi_login {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t max_version; /* Max. version supported */
+ uint8_t min_version; /* Min. version supported */
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t isid[6]; /* Initiator Session ID */
+ __be16 tsih; /* Target Session Handle */
+ __be32 itt; /* Initiator Task Tag */
+ __be16 cid;
+ __be16 rsvd3;
+ __be32 cmdsn;
+ __be32 exp_statsn;
+ uint8_t rsvd5[16];
+};
+
+/* Login PDU flags */
+#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_NEXT_STAGE_MASK 0x03 /* 2 bits */
+
+#define ISCSI_LOGIN_CURRENT_STAGE(flags) \
+ ((flags & ISCSI_FLAG_LOGIN_CURRENT_STAGE_MASK) >> 2)
+#define ISCSI_LOGIN_NEXT_STAGE(flags) \
+ (flags & ISCSI_FLAG_LOGIN_NEXT_STAGE_MASK)
+
+/* Login Response Header */
+struct iscsi_login_rsp {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t max_version; /* Max. version supported */
+ uint8_t active_version; /* Active version */
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t isid[6]; /* Initiator Session ID */
+ __be16 tsih; /* Target Session Handle */
+ __be32 itt; /* Initiator Task Tag */
+ __be32 rsvd3;
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ uint8_t status_class; /* see Login RSP ststus classes below */
+ uint8_t status_detail; /* see Login RSP Status details below */
+ uint8_t rsvd4[10];
+};
+
+/* Login stage (phase) codes for CSG, NSG */
+#define ISCSI_INITIAL_LOGIN_STAGE -1
+#define ISCSI_SECURITY_NEGOTIATION_STAGE 0
+#define ISCSI_OP_PARMS_NEGOTIATION_STAGE 1
+#define ISCSI_FULL_FEATURE_PHASE 3
+
+/* Login Status response classes */
+#define ISCSI_STATUS_CLS_SUCCESS 0x00
+#define ISCSI_STATUS_CLS_REDIRECT 0x01
+#define ISCSI_STATUS_CLS_INITIATOR_ERR 0x02
+#define ISCSI_STATUS_CLS_TARGET_ERR 0x03
+
+/* Login Status response detail codes */
+/* Class-0 (Success) */
+#define ISCSI_LOGIN_STATUS_ACCEPT 0x00
+
+/* Class-1 (Redirection) */
+#define ISCSI_LOGIN_STATUS_TGT_MOVED_TEMP 0x01
+#define ISCSI_LOGIN_STATUS_TGT_MOVED_PERM 0x02
+
+/* Class-2 (Initiator Error) */
+#define ISCSI_LOGIN_STATUS_INIT_ERR 0x00
+#define ISCSI_LOGIN_STATUS_AUTH_FAILED 0x01
+#define ISCSI_LOGIN_STATUS_TGT_FORBIDDEN 0x02
+#define ISCSI_LOGIN_STATUS_TGT_NOT_FOUND 0x03
+#define ISCSI_LOGIN_STATUS_TGT_REMOVED 0x04
+#define ISCSI_LOGIN_STATUS_NO_VERSION 0x05
+#define ISCSI_LOGIN_STATUS_ISID_ERROR 0x06
+#define ISCSI_LOGIN_STATUS_MISSING_FIELDS 0x07
+#define ISCSI_LOGIN_STATUS_CONN_ADD_FAILED 0x08
+#define ISCSI_LOGIN_STATUS_NO_SESSION_TYPE 0x09
+#define ISCSI_LOGIN_STATUS_NO_SESSION 0x0a
+#define ISCSI_LOGIN_STATUS_INVALID_REQUEST 0x0b
+
+/* Class-3 (Target Error) */
+#define ISCSI_LOGIN_STATUS_TARGET_ERROR 0x00
+#define ISCSI_LOGIN_STATUS_SVC_UNAVAILABLE 0x01
+#define ISCSI_LOGIN_STATUS_NO_RESOURCES 0x02
+
+/* Logout Header */
+struct iscsi_logout {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd1[2];
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t rsvd2[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be16 cid;
+ uint8_t rsvd3[2];
+ __be32 cmdsn;
+ __be32 exp_statsn;
+ uint8_t rsvd4[16];
+};
+
+/* Logout PDU flags */
+#define ISCSI_FLAG_LOGOUT_REASON_MASK 0x7F
+
+/* logout reason_code values */
+
+#define ISCSI_LOGOUT_REASON_CLOSE_SESSION 0
+#define ISCSI_LOGOUT_REASON_CLOSE_CONNECTION 1
+#define ISCSI_LOGOUT_REASON_RECOVERY 2
+#define ISCSI_LOGOUT_REASON_AEN_REQUEST 3
+
+/* Logout Response Header */
+struct iscsi_logout_rsp {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t response; /* see Logout response values below */
+ uint8_t rsvd2;
+ uint8_t hlength;
+ uint8_t dlength[3];
+ uint8_t rsvd3[8];
+ __be32 itt; /* Initiator Task Tag */
+ __be32 rsvd4;
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ __be32 rsvd5;
+ __be16 t2wait;
+ __be16 t2retain;
+ __be32 rsvd6;
+};
+
+/* logout response status values */
+
+#define ISCSI_LOGOUT_SUCCESS 0
+#define ISCSI_LOGOUT_CID_NOT_FOUND 1
+#define ISCSI_LOGOUT_RECOVERY_UNSUPPORTED 2
+#define ISCSI_LOGOUT_CLEANUP_FAILED 3
+
+/* SNACK Header */
+struct iscsi_snack {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t rsvd2[14];
+ __be32 itt;
+ __be32 begrun;
+ __be32 runlength;
+ __be32 exp_statsn;
+ __be32 rsvd3;
+ __be32 exp_datasn;
+ uint8_t rsvd6[8];
+};
+
+/* SNACK PDU flags */
+#define ISCSI_FLAG_SNACK_TYPE_MASK 0x0F /* 4 bits */
+
+/* Reject Message Header */
+struct iscsi_reject {
+ uint8_t opcode;
+ uint8_t flags;
+ uint8_t reason;
+ uint8_t rsvd2;
+ uint8_t rsvd3;
+ uint8_t dlength[3];
+ uint8_t rsvd4[16];
+ __be32 statsn;
+ __be32 exp_cmdsn;
+ __be32 max_cmdsn;
+ __be32 datasn;
+ uint8_t rsvd5[8];
+ /* Text - Rejected hdr */
+};
+
+/* Reason for Reject */
+#define CMD_BEFORE_LOGIN 1
+#define DATA_DIGEST_ERROR 2
+#define DATA_SNACK_REJECT 3
+#define ISCSI_PROTOCOL_ERROR 4
+#define CMD_NOT_SUPPORTED 5
+#define IMM_CMD_REJECT 6
+#define TASK_IN_PROGRESS 7
+#define INVALID_SNACK 8
+#define BOOKMARK_REJECTED 9
+#define BOOKMARK_NO_RESOURCES 10
+#define NEGOTIATION_RESET 11
+
+/* Max. number of Key=Value pairs in a text message */
+#define MAX_KEY_VALUE_PAIRS 8192
+
+/* maximum length for text keys/values */
+#define KEY_MAXLEN 64
+#define VALUE_MAXLEN 255
+#define TARGET_NAME_MAXLEN VALUE_MAXLEN
+
+#define DEFAULT_MAX_RECV_DATA_SEGMENT_LENGTH 8192
+
+/************************* RFC 3720 End *****************************/
+
+#endif /* ISCSI_PROTO_H */
^ permalink raw reply
* [PATCH 1/3] add open iscsi netlink interface to iscsi transport class
From: Mike Christie @ 2005-05-15 19:29 UTC (permalink / raw)
To: open-iscsi, netdev, linux-scsi, linux-kernel, inux-iscsi-devel
include/linux/netlink.h changes (added new protocol NETLINK_ISCSI)
Thanks,
Linux-iscsi Team
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
--- linux-2.6.12-rc3.orig/include/linux/netlink.h 2005-04-20 17:03:16.000000000 -0700
+++ linux-2.6.12-rc3/include/linux/netlink.h 2005-05-04 18:06:44.000000000 -0700
@@ -14,6 +14,7 @@
#define NETLINK_SELINUX 7 /* SELinux event notifications */
#define NETLINK_ARPD 8
#define NETLINK_AUDIT 9 /* auditing */
+#define NETLINK_ISCSI 10 /* iSCSI Open Interface */
#define NETLINK_ROUTE6 11 /* af_inet6 route comm channel */
#define NETLINK_IP6_FW 13
#define NETLINK_DNRTMSG 14 /* DECnet routing messages */
^ permalink raw reply
* [PATCH 0/3] add open iscsi netlink interface to iscsi transport class
From: Mike Christie @ 2005-05-15 19:29 UTC (permalink / raw)
To: open-iscsi, netdev, linux-scsi, linux-kernel, linux-iscsi-devel
The following patches add the linux-iscsi-5.X/open-iscsi project's
netlink interface to the iscsi transport class.
Last time I messed up and did not inline the patches so the reason for
the allignment in some structures was not clear. There was also a bug in
that unsigned long was being used instead of u64/unit64_t so that
complicated matters. We have been discussing this internally but I am
not sure I know the best route - I had hoped to just be able to remove
them all like was suggested, so I am reposting the patches inlined with
hopefully at least Chris Wright and almost all of Christoph Hellwig's
comments addressed.
The one suggestion we were not able to complete was the one to move the
scsi_remove_host and scsi_host_add calls. It is trivial to do, but
seemed odd since every iscsi_transport would need to add these calls. I
can understand in the future when HW iSCSI is supported all our
scsi_host lifetime code will need to be reworked because they allocate a
host per pci device and we allocate a host per session (maybe a common
place to meet would be a host per initiator port), but with only
software iSCSI supported today maybe a little more info is needed. Is
the suggestion to move the add/remove host calls to the iscsi_transport
to allow the transport to control when they want to begin queueing comamnds?
Maybe the removal of ISCSI_TRANSPORT_MAX is also not completely met too.
I am hoping to finish fixing that by exposing the same information
through sysfs, but I am still trying to figure out how the layout should
be implmented using the driver model transport classes and struct
devices or kobjects or whatever makes people happy. We could simply do
this using driver model transport classes by just adding a
iscsi_transport class that represents the iSCSI transport like
iscsi_tcp, iscsi_iser and making that the parent of the session, but
maybe that is better done in a seperate patch so we can discuss the
abstraction for the parent of the session seperately.
More info about the open-iscsi/linux-iscsi-5 project can found here
http://open-iscsi.org/.
Patches
1. add-iscsi-netlink-def.patch - include/linux/netlink.h changes (added
new protocol NETLINK_ISCSI)
2. common-iscsi-headers.patch - Common header files:
- iscsi_if.h (user/kernel #defines);
- iscsi_proto.h (RFC3720 #defines and types);
- iscsi_ifev.h (user/kernel events).
3. integrate-iscsi-netlink.patch - incorporate the
open-iscsi/linux-iscsi netlink interface into the iscsi transport class.
Thanks,
Linux-iscsi Team
^ permalink raw reply
* Re: [IPV4/IPV6] Ensure all frag_list members have NULL sk
From: Evgeniy Polyakov @ 2005-05-15 17:33 UTC (permalink / raw)
To: Herbert Xu; +Cc: netdev, davem
In-Reply-To: <20050515122256.GA22251@gondor.apana.org.au>
On Sun, 15 May 2005 22:22:56 +1000
Herbert Xu <herbert@gondor.apana.org.au> wrote:
> On Sun, May 15, 2005 at 09:41:21PM +1000, herbert wrote:
> >
> > I'll post a new patch soon. However, since this is a pretty major change
> > and the bugs it fixes aren't that important it should probably be delayed
> > until 2.6.13.
>
> Here it is:
>
>
> Having frag_list members which holds wmem of an sk leads to nightmares
> with partially cloned frag skb's. The reason is that once you unleash
> a skb with a frag_list that has individual sk ownerships into the stack
> you can never undo those ownerships safely as they may have been cloned
> by things like netfilter. Since we have to undo them in order to make
> skb_linearize happy this approach leads to a dead-end.
>
> So let's go the other way and make this an invariant:
>
> For any skb on a frag_list, skb->sk must be NULL.
This requires skb_set_owner_* to check if it is called
for head skb or one from fragment and does nothing if
it is from frag_list.
Or to check the whole tree for ownering calls...
> That is, the socket ownership always belongs to the head skb.
> It turns out that the implementation is actually pretty simple.
>
> The above invariant is actually violated in the following patch
> for a short duration inside ip_fragment. This is OK because the
> offending frag_list member is either destroyed at the end of the
> slow path without being sent anywhere, or it is detached from
> the frag_list before being sent.
>
> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
>
> Cheers,
> --
> Visit Openswan at http://www.openswan.org/
> Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
> Home Page: http://gondor.apana.org.au/~herbert/
> PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
Evgeniy Polyakov
Only failure makes us experts. -- Theo de Raadt
^ permalink raw reply
* ip_forward difference in kernel 2.4 and 2.6
From: cranium2003 @ 2005-05-15 13:39 UTC (permalink / raw)
To: net dev
hello,
I check kernel 2.4.28 and 2.6.11 on lxr and
found that there is condition to check fragmnetation
required for packet in 2.4.28 ain ip_forward.c as
if (skb->len > mtu && (ntohs(iph->frag_off) &
IP_DF))
goto frag_needed;
But same will not be seen in 2.6.11 why? Then how 2.6
kenrel check this condition? Why its removed? Is it
addedd somewhere else?
Thanks in advance.
regards,
cranium
__________________________________
Do you Yahoo!?
Read only the mail you want - Yahoo! Mail SpamGuard.
http://promotions.yahoo.com/new_mail
^ permalink raw reply
* [2.6 patch] remove net/ipv4/ipvs/ip_vs_proto_icmp.c?
From: Adrian Bunk @ 2005-05-15 13:29 UTC (permalink / raw)
To: Julian Anastasov; +Cc: Wensong Zhang, netdev, linux-kernel
In-Reply-To: <Pine.LNX.4.58.0505141013520.1568@u.domain.uli>
On Sat, May 14, 2005 at 10:17:53AM +0300, Julian Anastasov wrote:
>
> Hello,
>
> On Fri, 13 May 2005, Adrian Bunk wrote:
>
> > Will it be made working in the forseeable future or is it a candidate
> > for removal?
>
> IMO, it can be removed as it was never finished. We can always
> add it later.
Thanks for the confirmation.
A patch to remove it is below.
> Regards
> Julian Anastasov <ja@ssi.bg>
cu
Adrian
<-- snip -->
ip_vs_proto_icmp.c was never finished.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
---
net/ipv4/ipvs/Makefile | 2
net/ipv4/ipvs/ip_vs_proto.c | 3
net/ipv4/ipvs/ip_vs_proto_icmp.c | 182 -------------------------------
3 files changed, 1 insertion(+), 186 deletions(-)
--- linux-2.6.12-rc4-mm1-full/net/ipv4/ipvs/Makefile.old 2005-05-15 13:14:02.000000000 +0200
+++ linux-2.6.12-rc4-mm1-full/net/ipv4/ipvs/Makefile 2005-05-15 13:14:17.000000000 +0200
@@ -11,7 +11,7 @@
ip_vs-objs := ip_vs_conn.o ip_vs_core.o ip_vs_ctl.o ip_vs_sched.o \
ip_vs_xmit.o ip_vs_app.o ip_vs_sync.o \
- ip_vs_est.o ip_vs_proto.o ip_vs_proto_icmp.o \
+ ip_vs_est.o ip_vs_proto.o \
$(ip_vs_proto-objs-y)
--- linux-2.6.12-rc4-mm1-full/net/ipv4/ipvs/ip_vs_proto.c.old 2005-05-15 13:18:55.000000000 +0200
+++ linux-2.6.12-rc4-mm1-full/net/ipv4/ipvs/ip_vs_proto.c 2005-05-15 13:15:03.000000000 +0200
@@ -216,9 +216,6 @@
#ifdef CONFIG_IP_VS_PROTO_UDP
REGISTER_PROTOCOL(&ip_vs_protocol_udp);
#endif
-#ifdef CONFIG_IP_VS_PROTO_ICMP
- REGISTER_PROTOCOL(&ip_vs_protocol_icmp);
-#endif
#ifdef CONFIG_IP_VS_PROTO_AH
REGISTER_PROTOCOL(&ip_vs_protocol_ah);
#endif
--- linux-2.6.12-rc4-mm1-full/net/ipv4/ipvs/ip_vs_proto_icmp.c 2005-03-02 08:37:31.000000000 +0100
+++ /dev/null 2005-04-28 03:52:17.000000000 +0200
@@ -1,182 +0,0 @@
-/*
- * ip_vs_proto_icmp.c: ICMP load balancing support for IP Virtual Server
- *
- * Authors: Julian Anastasov <ja@ssi.bg>, March 2002
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License
- * version 2 as published by the Free Software Foundation;
- *
- */
-
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/icmp.h>
-#include <linux/netfilter.h>
-#include <linux/netfilter_ipv4.h>
-
-#include <net/ip_vs.h>
-
-
-static int icmp_timeouts[1] = { 1*60*HZ };
-
-static char * icmp_state_name_table[1] = { "ICMP" };
-
-static struct ip_vs_conn *
-icmp_conn_in_get(const struct sk_buff *skb,
- struct ip_vs_protocol *pp,
- const struct iphdr *iph,
- unsigned int proto_off,
- int inverse)
-{
-#if 0
- struct ip_vs_conn *cp;
-
- if (likely(!inverse)) {
- cp = ip_vs_conn_in_get(iph->protocol,
- iph->saddr, 0,
- iph->daddr, 0);
- } else {
- cp = ip_vs_conn_in_get(iph->protocol,
- iph->daddr, 0,
- iph->saddr, 0);
- }
-
- return cp;
-
-#else
- return NULL;
-#endif
-}
-
-static struct ip_vs_conn *
-icmp_conn_out_get(const struct sk_buff *skb,
- struct ip_vs_protocol *pp,
- const struct iphdr *iph,
- unsigned int proto_off,
- int inverse)
-{
-#if 0
- struct ip_vs_conn *cp;
-
- if (likely(!inverse)) {
- cp = ip_vs_conn_out_get(iph->protocol,
- iph->saddr, 0,
- iph->daddr, 0);
- } else {
- cp = ip_vs_conn_out_get(IPPROTO_UDP,
- iph->daddr, 0,
- iph->saddr, 0);
- }
-
- return cp;
-#else
- return NULL;
-#endif
-}
-
-static int
-icmp_conn_schedule(struct sk_buff *skb, struct ip_vs_protocol *pp,
- int *verdict, struct ip_vs_conn **cpp)
-{
- *verdict = NF_ACCEPT;
- return 0;
-}
-
-static int
-icmp_csum_check(struct sk_buff *skb, struct ip_vs_protocol *pp)
-{
- if (!(skb->nh.iph->frag_off & __constant_htons(IP_OFFSET))) {
- if (skb->ip_summed != CHECKSUM_UNNECESSARY) {
- if (ip_vs_checksum_complete(skb, skb->nh.iph->ihl * 4)) {
- IP_VS_DBG_RL_PKT(0, pp, skb, 0, "Failed checksum for");
- return 0;
- }
- }
- }
- return 1;
-}
-
-static void
-icmp_debug_packet(struct ip_vs_protocol *pp,
- const struct sk_buff *skb,
- int offset,
- const char *msg)
-{
- char buf[256];
- struct iphdr _iph, *ih;
-
- ih = skb_header_pointer(skb, offset, sizeof(_iph), &_iph);
- if (ih == NULL)
- sprintf(buf, "%s TRUNCATED", pp->name);
- else if (ih->frag_off & __constant_htons(IP_OFFSET))
- sprintf(buf, "%s %u.%u.%u.%u->%u.%u.%u.%u frag",
- pp->name, NIPQUAD(ih->saddr),
- NIPQUAD(ih->daddr));
- else {
- struct icmphdr _icmph, *ic;
-
- ic = skb_header_pointer(skb, offset + ih->ihl*4,
- sizeof(_icmph), &_icmph);
- if (ic == NULL)
- sprintf(buf, "%s TRUNCATED to %u bytes\n",
- pp->name, skb->len - offset);
- else
- sprintf(buf, "%s %u.%u.%u.%u->%u.%u.%u.%u T:%d C:%d",
- pp->name, NIPQUAD(ih->saddr),
- NIPQUAD(ih->daddr),
- ic->type, ic->code);
- }
- printk(KERN_DEBUG "IPVS: %s: %s\n", msg, buf);
-}
-
-static int
-icmp_state_transition(struct ip_vs_conn *cp, int direction,
- const struct sk_buff *skb,
- struct ip_vs_protocol *pp)
-{
- cp->timeout = pp->timeout_table[IP_VS_ICMP_S_NORMAL];
- return 1;
-}
-
-static int
-icmp_set_state_timeout(struct ip_vs_protocol *pp, char *sname, int to)
-{
- int num;
- char **names;
-
- num = IP_VS_ICMP_S_LAST;
- names = icmp_state_name_table;
- return ip_vs_set_state_timeout(pp->timeout_table, num, names, sname, to);
-}
-
-
-static void icmp_init(struct ip_vs_protocol *pp)
-{
- pp->timeout_table = icmp_timeouts;
-}
-
-static void icmp_exit(struct ip_vs_protocol *pp)
-{
-}
-
-struct ip_vs_protocol ip_vs_protocol_icmp = {
- .name = "ICMP",
- .protocol = IPPROTO_ICMP,
- .dont_defrag = 0,
- .init = icmp_init,
- .exit = icmp_exit,
- .conn_schedule = icmp_conn_schedule,
- .conn_in_get = icmp_conn_in_get,
- .conn_out_get = icmp_conn_out_get,
- .snat_handler = NULL,
- .dnat_handler = NULL,
- .csum_check = icmp_csum_check,
- .state_transition = icmp_state_transition,
- .register_app = NULL,
- .unregister_app = NULL,
- .app_conn_bind = NULL,
- .debug_packet = icmp_debug_packet,
- .timeout_change = NULL,
- .set_state_timeout = icmp_set_state_timeout,
-};
^ permalink raw reply
* [IPV4/IPV6] Ensure all frag_list members have NULL sk
From: Herbert Xu @ 2005-05-15 12:22 UTC (permalink / raw)
To: Evgeniy Polyakov; +Cc: netdev, davem
In-Reply-To: <20050515114121.GA4830@gondor.apana.org.au>
[-- Attachment #1: Type: text/plain, Size: 1455 bytes --]
On Sun, May 15, 2005 at 09:41:21PM +1000, herbert wrote:
>
> I'll post a new patch soon. However, since this is a pretty major change
> and the bugs it fixes aren't that important it should probably be delayed
> until 2.6.13.
Here it is:
Having frag_list members which holds wmem of an sk leads to nightmares
with partially cloned frag skb's. The reason is that once you unleash
a skb with a frag_list that has individual sk ownerships into the stack
you can never undo those ownerships safely as they may have been cloned
by things like netfilter. Since we have to undo them in order to make
skb_linearize happy this approach leads to a dead-end.
So let's go the other way and make this an invariant:
For any skb on a frag_list, skb->sk must be NULL.
That is, the socket ownership always belongs to the head skb.
It turns out that the implementation is actually pretty simple.
The above invariant is actually violated in the following patch
for a short duration inside ip_fragment. This is OK because the
offending frag_list member is either destroyed at the end of the
slow path without being sent anywhere, or it is detached from
the frag_list before being sent.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
[-- Attachment #2: p --]
[-- Type: text/plain, Size: 1368 bytes --]
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -490,6 +490,14 @@ int ip_fragment(struct sk_buff *skb, int
/* Partially cloned skb? */
if (skb_shared(frag))
goto slow_path;
+
+ BUG_ON(frag->sk);
+ if (skb->sk) {
+ sock_hold(skb->sk);
+ frag->sk = skb->sk;
+ frag->destructor = sock_wfree;
+ skb->truesize -= frag->truesize;
+ }
}
/* Everything is OK. Generate! */
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -552,13 +552,17 @@ static int ip6_fragment(struct sk_buff *
skb_headroom(frag) < hlen)
goto slow_path;
- /* Correct socket ownership. */
- if (frag->sk == NULL)
- goto slow_path;
-
/* Partially cloned skb? */
if (skb_shared(frag))
goto slow_path;
+
+ BUG_ON(frag->sk);
+ if (skb->sk) {
+ sock_hold(skb->sk);
+ frag->sk = skb->sk;
+ frag->destructor = sock_wfree;
+ skb->truesize -= frag->truesize;
+ }
}
err = 0;
@@ -1116,12 +1120,10 @@ int ip6_push_pending_frames(struct sock
tail_skb = &(tmp_skb->next);
skb->len += tmp_skb->len;
skb->data_len += tmp_skb->len;
-#if 0 /* Logically correct, but useless work, ip_fragment() will have to undo */
skb->truesize += tmp_skb->truesize;
__sock_put(tmp_skb->sk);
tmp_skb->destructor = NULL;
tmp_skb->sk = NULL;
-#endif
}
ipv6_addr_copy(final_dst, &fl->fl6_dst);
^ permalink raw reply
* Re: [RFC: 2.6 patch] make MII no longer user visible
From: Adrian Bunk @ 2005-05-15 12:12 UTC (permalink / raw)
To: Jeff Garzik; +Cc: netdev, linux-kernel
In-Reply-To: <4284DD16.8090405@pobox.com>
On Fri, May 13, 2005 at 01:00:06PM -0400, Jeff Garzik wrote:
> Adrian Bunk wrote:
> >MII is a classical example of a helper option no user should ever see.
>
> Incorrect.
>
> It's the classic example of an option that distributors may want to
> build as a module, even if no shipped modules need it, to enable net
> driver development and use in their kernel.
Every distibutor will have more than one net driver select'ing MII
enabled.
And if you are doing net driver development, you can always enable it in
your kernel.
> Jeff
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
^ permalink raw reply
* Re: [IPV4/IPV6] Keep wmem accounting separate in ip*_push_pending_frames
From: Herbert Xu @ 2005-05-15 11:41 UTC (permalink / raw)
To: Evgeniy Polyakov; +Cc: netdev, davem
In-Reply-To: <20050515104016.GA24344@gondor.apana.org.au>
On Sun, May 15, 2005 at 08:40:16PM +1000, herbert wrote:
Please discard this patch.
> Since frag_list skb's constructed by ip*_push_pending_frames are
> usually going to be fragmented in ip*_fragment and sent separately,
> it makes no sense to account them all in the head skb. Doing so
> means that ip*_fragment would have to undo it all in order to keep
> the correct wmem accounting.
Unfortunately this leads to nightmares with partially cloned frag skb's.
The reason is that once you unleash a skb with a frag_list that has
individual sk ownerships into the stack you can never undo those
ownerships safely as they may have been cloned by things like netfilter.
Since we have to undo them in order to make skb_linearize happy this
approach leads to a dead-end.
So let's go the other way and make this an invariant:
For any skb on a frag_list, skb->sk must be NULL.
That is, the socket ownership always belongs to the head skb.
It turns out that the implementation is actually pretty simple.
> On the slow path it double-charges the skb's which may unnecssarily
> delay new data from being sent.
Especially because I was wrong about this. The slow paths in ip*_fragment
is actually correct since it frees the original skb after constructing
the new fragments.
I'll post a new patch soon. However, since this is a pretty major change
and the bugs it fixes aren't that important it should probably be delayed
until 2.6.13.
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* [IPV4/IPV6] Keep wmem accounting separate in ip*_push_pending_frames
From: Herbert Xu @ 2005-05-15 10:40 UTC (permalink / raw)
To: Evgeniy Polyakov; +Cc: netdev, davem
In-Reply-To: <E1DXE3h-0002jR-00@gondolin.me.apana.org.au>
[-- Attachment #1: Type: text/plain, Size: 1513 bytes --]
On Sun, May 15, 2005 at 06:01:09PM +1000, Herbert Xu wrote:
>
> In fact you've uncovered a couple of bugs elsewhere too. IPv4's
> ip_push_pending_frames attributes all the data to the first skb
> which breaks when the packet is sent through ip_fragment since
> the latter does not undo the truesize attribution. skb_linearize
> is broken as well since it ignores truesize/owner altogether.
Here is a patch that addresses the first issue.
Since frag_list skb's constructed by ip*_push_pending_frames are
usually going to be fragmented in ip*_fragment and sent separately,
it makes no sense to account them all in the head skb. Doing so
means that ip*_fragment would have to undo it all in order to keep
the correct wmem accounting.
In fact as it is IPv4 is attributing all the memory to the head skb
in ip_push_pending_frames and not undoing the attribution correctly
in ip_fragment. On the fast path it will not undo the work at all
which means that as soon as the head skb is transmitted the quota
which may still be used by the other skb's can be used by new data.
On the slow path it double-charges the skb's which may unnecssarily
delay new data from being sent.
This patch simply leaves the wmem accounting with each skb.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
[-- Attachment #2: p --]
[-- Type: text/plain, Size: 888 bytes --]
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -1138,10 +1138,6 @@ int ip_push_pending_frames(struct sock *
tail_skb = &(tmp_skb->next);
skb->len += tmp_skb->len;
skb->data_len += tmp_skb->len;
- skb->truesize += tmp_skb->truesize;
- __sock_put(tmp_skb->sk);
- tmp_skb->destructor = NULL;
- tmp_skb->sk = NULL;
}
/* Unless user demanded real pmtu discovery (IP_PMTUDISC_DO), we allow
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -1116,12 +1116,6 @@ int ip6_push_pending_frames(struct sock
tail_skb = &(tmp_skb->next);
skb->len += tmp_skb->len;
skb->data_len += tmp_skb->len;
-#if 0 /* Logically correct, but useless work, ip_fragment() will have to undo */
- skb->truesize += tmp_skb->truesize;
- __sock_put(tmp_skb->sk);
- tmp_skb->destructor = NULL;
- tmp_skb->sk = NULL;
-#endif
}
ipv6_addr_copy(final_dst, &fl->fl6_dst);
^ permalink raw reply
* Re: Status of net/ipv4/ipvs/ip_vs_proto_icmp.c?
From: Adrian Bunk @ 2005-05-15 8:27 UTC (permalink / raw)
To: Andres Salomon; +Cc: linux-kernel, netdev
In-Reply-To: <pan.2005.05.13.19.09.00.598647@athenacr.com>
On Fri, May 13, 2005 at 03:09:01PM -0400, Andres Salomon wrote:
> On Fri, 13 May 2005 06:16:22 +0200, Adrian Bunk wrote:
>
> > Hi,
> >
> > can anyone explain the status of?
> >
> > This file is always included in the kernel if CONFIG_IP_VS=y, but it's
> > completely unused.
> >
> > Will it be made working in the forseeable future or is it a candidate
> > for removal?
> >
> > TIA
> > Adrian
>
> The people/places to ask would probably be:
>
> IPVS
> P: Wensong Zhang
> M: wensong@linux-vs.org
> P: Julian Anastasov
> M: ja@ssi.bg
> S: Maintained
>...
And these are exactly the people in the To-header of the email asking
this question.
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
^ permalink raw reply
* Re: [1/1] xfrm: skb_cow_data() does not set proper owner for new skbs.
From: Herbert Xu @ 2005-05-15 8:01 UTC (permalink / raw)
To: Evgeniy Polyakov; +Cc: netdev, davem
In-Reply-To: <20050514134834.GA2698@uganda.factory.vocord.ru>
Evgeniy Polyakov <johnpol@2ka.mipt.ru> wrote:
>
> As far as I can see, current code sets destructor and socket
> for the first one skb only and uses truesize of the first skb
> only to increment sk_wmem_alloc value.
>
> If above "analysis" is correct then attached patch fixes that.
Yes you're absolutely right.
> Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
In fact you've uncovered a couple of bugs elsewhere too. IPv4's
ip_push_pending_frames attributes all the data to the first skb
which breaks when the packet is sent through ip_fragment since
the latter does not undo the truesize attribution. skb_linearize
is broken as well since it ignores truesize/owner altogether.
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* Re: [PATCH 11/12] orinoco: monitor mode support
From: Francois Romieu @ 2005-05-14 17:39 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: jgarzik, proski, hermes, netdev
In-Reply-To: <20050514153100.GL3643@lst.de>
Christoph Hellwig <hch@lst.de> :
[...]
> + skb = dev_alloc_skb(hdrlen + datalen);
> + if (!skb) {
> + printk(KERN_WARNING "%s: Cannot allocate skb for monitor frame\n",
> + dev->name);
> + err = -ENOMEM;
> + goto drop;
> + }
> +
> + /* Copy the 802.11 header to the skb */
> + memcpy(skb_put(skb, hdrlen), &(desc->frame_ctl), hdrlen);
> + skb->mac.raw = skb->data;
> +
> + /* If any, copy the data from the card to the skb */
> + if (datalen > 0) {
> + err = hermes_bap_pread(hw, IRQ_BAP, skb_put(skb, datalen),
> + ALIGN(datalen, 2), rxfid,
> + HERMES_802_2_OFFSET);
> + if (err) {
> + printk(KERN_ERR "%s: error %d reading monitor frame\n",
> + dev->name, err);
> + goto drop;
[...]
> + drop:
> + stats->rx_errors++;
> + stats->rx_dropped++;
> +}
-> leak (skb).
--
Ueimor
^ permalink raw reply
* [PATCH 12/12] orinoco: update changelog and version
From: Christoph Hellwig @ 2005-05-14 15:31 UTC (permalink / raw)
To: jgarzik, proski, hermes; +Cc: netdev
diff -urp -X /home/hch/dontdiff linux-2.6/drivers/net/wireless/orinoco.c orinoco/orinoco.c
--- linux-2.6/drivers/net/wireless/orinoco.c 2005-05-14 13:49:02.000000000 +0200
+++ orinoco/orinoco.c 2005-05-14 13:05:46.000000000 +0200
@@ -416,12 +416,50 @@
* port. This is supposed to allow 802.1x to work sanely, but
* doesn't seem to yet.
*
+ * v0.14alpha2 -> v0.15rc1 - 19 Apr 2004 - Pavel Roskin & David Gibson
+ * o Fix bug which prevented setting 32 character ESSIDs from
+ * iwconfig (Thomas Schulz).
+ * o Fix for incorrect CIS access in orinoco_plx (Pavel Roskin).
+ * o Fix setting WEP key if __orinoco_fastkeychange() is not
+ * supported (Pavel Roskin).
+ * o New wireless extensions API and scanning support (patch from
+ * Moustafa Youssef, updated by Jim Carter and Pavel Roskin).
+ * o Add minimal ethtool support (Pavel Roskin).
+ * o Replace CardServices() calls for compatibility with Linux
+ * 2.6.2 and above (Pavel Roskin).
+ * o Fix recognition of Intersil x.x.1 firmware (Pavel Roskin).
+ * o Replace dump_recs with more flexible get_rid ioctl (Pavel
+ * Roskin).
+ * o RF monitor mode support (Pavel Roskin).
+ * o Lots of bugfixes.
+ *
+ * v0.15rc1 -> v0.15rc2 - 28 Jul 2004 - Pavel Roskin & David Gibson
+ * o orinoco_pci saves PCI registers on suspend (Simon Huggins).
+ * o Monitor mode disabled on Agere 8.xx firmware - it's broken.
+ * o BAP timeout increased - needed for Intersil firmware.
+ * o Tx power is no longer reported - it's unreliable.
+ * o Use 802.11 header in rx path. Hide packets with ToDS flag
+ * from programs that don't need promiscous mode (John Denker).
+ * o Manual roaming implemented for Symbol and Intersil firmware.
+ * o Use netdev_priv() instead of directly dereferencing dev->priv.
+ * o Some simplification of pcmcia init code in orinoco_cs and
+ * spectrum_cs.
+ * o Numerous trivial cleanups, mainly arising from long-overdue
+ * merge with mainline.
+ *
+ * v0.15rc2 -> ???? - ???? - David Gibson
+ * o Use ssleep() or msleep() instead of hardcoded
+ * schedule_timeout()s (Nishanth Aravamudan via kernel-janitors
+ * list).
+ * o Several cleanups and bugfixes in pci/plx/tmd/nortel drivers.
+ * o Fix memory leak in orinoco_join_ap().
+ * o Change io handling to avoid sparse and gcc warnings.
+ * o Use C99 array initializers and ARRAY_SIZE() for iw_handler
+ * tables.
+ *
* TODO
- * o New wireless extensions API (patch from Moustafa
- * Youssef, updated by Jim Carter and Pavel Roskin).
* o Handle de-encapsulation within network layer, provide 802.11
* headers (patch from Thomas 'Dent' Mirlacher)
- * o RF monitor mode support
* o Fix possible races in SPY handling.
* o Disconnect wireless extensions from fundamental configuration.
* o (maybe) Software WEP support (patch from Stano Meduna).
diff -urp -X /home/hch/dontdiff linux-2.6/drivers/net/wireless/orinoco.h orinoco/orinoco.h
--- linux-2.6/drivers/net/wireless/orinoco.h 2005-05-14 13:48:44.000000000 +0200
+++ orinoco/orinoco.h 2005-03-18 16:03:24.000000000 +0100
@@ -7,7 +7,7 @@
#ifndef _ORINOCO_H
#define _ORINOCO_H
-#define DRIVER_VERSION "0.14alpha2"
+#define DRIVER_VERSION "0.15rc2"
#include <linux/types.h>
#include <linux/spinlock.h>
^ permalink raw reply
* [PATCH 11/12] orinoco: monitor mode support
From: Christoph Hellwig @ 2005-05-14 15:31 UTC (permalink / raw)
To: jgarzik, proski, hermes; +Cc: netdev
Patch from Pavel Roskin
Index: linux-2.6/drivers/net/wireless/orinoco.c
===================================================================
--- linux-2.6.orig/drivers/net/wireless/orinoco.c 2005-05-14 17:07:45.000000000 +0200
+++ linux-2.6/drivers/net/wireless/orinoco.c 2005-05-14 17:08:39.000000000 +0200
@@ -499,6 +499,10 @@
module_param(ignore_disconnect, int, 0644);
MODULE_PARM_DESC(ignore_disconnect, "Don't report lost link to the network layer");
+static int force_monitor; /* = 0 */
+module_param(force_monitor, int, 0644);
+MODULE_PARM_DESC(force_monitor, "Allow monitor mode for all firmware versions");
+
/********************************************************************/
/* Compile time configuration and compatibility stuff */
/********************************************************************/
@@ -670,6 +674,10 @@
priv->createibss = 1;
}
break;
+ case IW_MODE_MONITOR:
+ priv->port_type = 3;
+ priv->createibss = 0;
+ break;
default:
printk(KERN_ERR "%s: Invalid priv->iw_mode in set_port_type()\n",
priv->ndev->name);
@@ -856,7 +864,7 @@
return 1;
}
- if (! netif_carrier_ok(dev)) {
+ if (! netif_carrier_ok(dev) || (priv->iw_mode == IW_MODE_MONITOR)) {
/* Oops, the firmware hasn't established a connection,
silently drop the packet (this seems to be the
safest approach). */
@@ -1118,6 +1126,114 @@
}
}
+/*
+ * orinoco_rx_monitor - handle received monitor frames.
+ *
+ * Arguments:
+ * dev network device
+ * rxfid received FID
+ * desc rx descriptor of the frame
+ *
+ * Call context: interrupt
+ */
+static void orinoco_rx_monitor(struct net_device *dev, u16 rxfid,
+ struct hermes_rx_descriptor *desc)
+{
+ u32 hdrlen = 30; /* return full header by default */
+ u32 datalen = 0;
+ u16 fc;
+ int err;
+ int len;
+ struct sk_buff *skb;
+ struct orinoco_private *priv = netdev_priv(dev);
+ struct net_device_stats *stats = &priv->stats;
+ hermes_t *hw = &priv->hw;
+
+ len = le16_to_cpu(desc->data_len);
+
+ /* Determine the size of the header and the data */
+ fc = le16_to_cpu(desc->frame_ctl);
+ switch (fc & IEEE80211_FCTL_FTYPE) {
+ case IEEE80211_FTYPE_DATA:
+ if ((fc & IEEE80211_FCTL_TODS)
+ && (fc & IEEE80211_FCTL_FROMDS))
+ hdrlen = 30;
+ else
+ hdrlen = 24;
+ datalen = len;
+ break;
+ case IEEE80211_FTYPE_MGMT:
+ hdrlen = 24;
+ datalen = len;
+ break;
+ case IEEE80211_FTYPE_CTL:
+ switch (fc & IEEE80211_FCTL_STYPE) {
+ case IEEE80211_STYPE_PSPOLL:
+ case IEEE80211_STYPE_RTS:
+ case IEEE80211_STYPE_CFEND:
+ case IEEE80211_STYPE_CFENDACK:
+ hdrlen = 16;
+ break;
+ case IEEE80211_STYPE_CTS:
+ case IEEE80211_STYPE_ACK:
+ hdrlen = 10;
+ break;
+ }
+ break;
+ default:
+ /* Unknown frame type */
+ break;
+ }
+
+ /* sanity check the length */
+ if (datalen > IEEE80211_DATA_LEN + 12) {
+ printk(KERN_DEBUG "%s: oversized monitor frame, "
+ "data length = %d\n", dev->name, datalen);
+ err = -EIO;
+ goto drop;
+ }
+
+ skb = dev_alloc_skb(hdrlen + datalen);
+ if (!skb) {
+ printk(KERN_WARNING "%s: Cannot allocate skb for monitor frame\n",
+ dev->name);
+ err = -ENOMEM;
+ goto drop;
+ }
+
+ /* Copy the 802.11 header to the skb */
+ memcpy(skb_put(skb, hdrlen), &(desc->frame_ctl), hdrlen);
+ skb->mac.raw = skb->data;
+
+ /* If any, copy the data from the card to the skb */
+ if (datalen > 0) {
+ err = hermes_bap_pread(hw, IRQ_BAP, skb_put(skb, datalen),
+ ALIGN(datalen, 2), rxfid,
+ HERMES_802_2_OFFSET);
+ if (err) {
+ printk(KERN_ERR "%s: error %d reading monitor frame\n",
+ dev->name, err);
+ goto drop;
+ }
+ }
+
+ skb->dev = dev;
+ skb->ip_summed = CHECKSUM_NONE;
+ skb->pkt_type = PACKET_OTHERHOST;
+ skb->protocol = __constant_htons(ETH_P_802_2);
+
+ dev->last_rx = jiffies;
+ stats->rx_packets++;
+ stats->rx_bytes += skb->len;
+
+ netif_rx(skb);
+ return;
+
+ drop:
+ stats->rx_errors++;
+ stats->rx_dropped++;
+}
+
static void __orinoco_ev_rx(struct net_device *dev, hermes_t *hw)
{
struct orinoco_private *priv = netdev_priv(dev);
@@ -1143,16 +1259,24 @@
status = le16_to_cpu(desc.status);
- if (status & HERMES_RXSTAT_ERR) {
- if (status & HERMES_RXSTAT_UNDECRYPTABLE) {
- wstats->discard.code++;
- DEBUG(1, "%s: Undecryptable frame on Rx. Frame dropped.\n",
- dev->name);
- } else {
- stats->rx_crc_errors++;
- DEBUG(1, "%s: Bad CRC on Rx. Frame dropped.\n", dev->name);
- }
+ if (status & HERMES_RXSTAT_BADCRC) {
+ DEBUG(1, "%s: Bad CRC on Rx. Frame dropped.\n",
+ dev->name);
+ stats->rx_crc_errors++;
+ stats->rx_errors++;
+ goto drop;
+ }
+ /* Handle frames in monitor mode */
+ if (priv->iw_mode == IW_MODE_MONITOR) {
+ orinoco_rx_monitor(dev, rxfid, &desc);
+ return;
+ }
+
+ if (status & HERMES_RXSTAT_UNDECRYPTABLE) {
+ DEBUG(1, "%s: Undecryptable frame on Rx. Frame dropped.\n",
+ dev->name);
+ wstats->discard.code++;
stats->rx_errors++;
goto drop;
}
@@ -2065,6 +2189,20 @@
}
}
+ if (priv->iw_mode == IW_MODE_MONITOR) {
+ /* Enable monitor mode */
+ dev->type = ARPHRD_IEEE80211;
+ err = hermes_docmd_wait(hw, HERMES_CMD_TEST |
+ HERMES_TEST_MONITOR, 0, NULL);
+ } else {
+ /* Disable monitor mode */
+ dev->type = ARPHRD_ETHER;
+ err = hermes_docmd_wait(hw, HERMES_CMD_TEST |
+ HERMES_TEST_STOP, 0, NULL);
+ }
+ if (err)
+ return err;
+
/* Set promiscuity / multicast*/
priv->promiscuous = 0;
priv->mc_count = 0;
@@ -2413,6 +2551,7 @@
priv->has_pm = (firmver >= 0x40020); /* Don't work in 7.52 ? */
priv->ibss_port = 1;
priv->has_hostscan = (firmver >= 0x8000a);
+ priv->broken_monitor = (firmver >= 0x80000);
/* Tested with Agere firmware :
* 1.16 ; 4.08 ; 4.52 ; 6.04 ; 6.16 ; 7.28 => Jean II
@@ -2980,6 +3119,15 @@
case IW_MODE_INFRA:
break;
+ case IW_MODE_MONITOR:
+ if (priv->broken_monitor && !force_monitor) {
+ printk(KERN_WARNING "%s: Monitor mode support is "
+ "buggy in this firmware, not enabling\n",
+ dev->name);
+ err = -EOPNOTSUPP;
+ }
+ break;
+
default:
err = -EOPNOTSUPP;
break;
@@ -3355,11 +3503,9 @@
unsigned long flags;
int err = -EINPROGRESS; /* Call commit handler */
- /* We can only use this in Ad-Hoc demo mode to set the operating
- * frequency, or in IBSS mode to set the frequency where the IBSS
- * will be created - Jean II */
- if (priv->iw_mode != IW_MODE_ADHOC)
- return -EOPNOTSUPP;
+ /* In infrastructure mode the AP sets the channel */
+ if (priv->iw_mode == IW_MODE_INFRA)
+ return -EBUSY;
if ( (frq->e == 0) && (frq->m <= 1000) ) {
/* Setting by channel number */
@@ -3383,7 +3529,15 @@
if (orinoco_lock(priv, &flags) != 0)
return -EBUSY;
+
priv->channel = chan;
+ if (priv->iw_mode == IW_MODE_MONITOR) {
+ /* Fast channel change - no commit if successful */
+ hermes_t *hw = &priv->hw;
+ err = hermes_docmd_wait(hw, HERMES_CMD_TEST |
+ HERMES_TEST_SET_CHANNEL,
+ chan, NULL);
+ }
orinoco_unlock(priv, &flags);
return err;
Index: linux-2.6/drivers/net/wireless/orinoco.h
===================================================================
--- linux-2.6.orig/drivers/net/wireless/orinoco.h 2005-05-14 17:05:42.000000000 +0200
+++ linux-2.6/drivers/net/wireless/orinoco.h 2005-05-14 17:07:57.000000000 +0200
@@ -94,6 +94,7 @@
unsigned int has_sensitivity:1;
unsigned int has_hostscan:1;
unsigned int broken_disableport:1;
+ unsigned int broken_monitor:1;
/* Configuration paramaters */
u32 iw_mode;
^ permalink raw reply
* [PATCH 10/12] orinoco: always use 802.11 header for rx processing
From: Christoph Hellwig @ 2005-05-14 15:30 UTC (permalink / raw)
To: jgarzik, proski, hermes; +Cc: netdev
If the frame has ToDS flag set, mark it by setting skb->pkt_type to
PACKET_OTHERHOST, so that applications unaware of promiscous mode won't get
uplink (STA->AP) packets for STA->STA transmissions relayed by the AP.
Thanks to John Denker and David Gibson for finding the problem and the
solution.
Patch from Pavel Roskin
Index: linux-2.6/drivers/net/wireless/orinoco.c
===================================================================
--- linux-2.6.orig/drivers/net/wireless/orinoco.c 2005-05-14 17:07:00.000000000 +0200
+++ linux-2.6/drivers/net/wireless/orinoco.c 2005-05-14 17:07:45.000000000 +0200
@@ -619,7 +619,9 @@
u16 ethertype;
} __attribute__ ((packed));
+/* Rx frame header except compatibility 802.3 header */
struct hermes_rx_descriptor {
+ /* Control */
u16 status;
u32 time;
u8 silence;
@@ -627,6 +629,18 @@
u8 rate;
u8 rxflow;
u32 reserved;
+
+ /* 802.11 header */
+ u16 frame_ctl;
+ u16 duration_id;
+ u8 addr1[ETH_ALEN];
+ u8 addr2[ETH_ALEN];
+ u8 addr3[ETH_ALEN];
+ u16 seq_ctl;
+ u8 addr4[ETH_ALEN];
+
+ /* Data length */
+ u16 data_len;
} __attribute__ ((packed));
/********************************************************************/
@@ -1110,12 +1124,10 @@
struct net_device_stats *stats = &priv->stats;
struct iw_statistics *wstats = &priv->wstats;
struct sk_buff *skb = NULL;
- u16 rxfid, status;
- int length, data_len, data_off;
- char *p;
+ u16 rxfid, status, fc;
+ int length;
struct hermes_rx_descriptor desc;
- struct header_struct hdr;
- struct ethhdr *eh;
+ struct ethhdr *hdr;
int err;
rxfid = hermes_read_regn(hw, RXFID);
@@ -1140,24 +1152,14 @@
stats->rx_crc_errors++;
DEBUG(1, "%s: Bad CRC on Rx. Frame dropped.\n", dev->name);
}
- stats->rx_errors++;
- goto drop;
- }
-
- /* For now we ignore the 802.11 header completely, assuming
- that the card's firmware has handled anything vital */
- err = hermes_bap_pread(hw, IRQ_BAP, &hdr, sizeof(hdr),
- rxfid, HERMES_802_3_OFFSET);
- if (err) {
- printk(KERN_ERR "%s: error %d reading frame header. "
- "Frame dropped.\n", dev->name, err);
stats->rx_errors++;
goto drop;
}
- length = ntohs(hdr.len);
-
+ length = le16_to_cpu(desc.data_len);
+ fc = le16_to_cpu(desc.frame_ctl);
+
/* Sanity checks */
if (length < 3) { /* No for even an 802.2 LLC header */
/* At least on Symbol firmware with PCF we get quite a
@@ -1186,57 +1188,51 @@
goto drop;
}
- skb_reserve(skb, 2); /* This way the IP header is aligned */
+ /* We'll prepend the header, so reserve space for it. The worst
+ case is no decapsulation, when 802.3 header is prepended and
+ nothing is removed. 2 is for aligning the IP header. */
+ skb_reserve(skb, ETH_HLEN + 2);
+
+ err = hermes_bap_pread(hw, IRQ_BAP, skb_put(skb, length),
+ ALIGN(length, 2), rxfid,
+ HERMES_802_2_OFFSET);
+ if (err) {
+ printk(KERN_ERR "%s: error %d reading frame. "
+ "Frame dropped.\n", dev->name, err);
+ stats->rx_errors++;
+ goto drop;
+ }
/* Handle decapsulation
* In most cases, the firmware tell us about SNAP frames.
* For some reason, the SNAP frames sent by LinkSys APs
* are not properly recognised by most firmwares.
* So, check ourselves */
- if (((status & HERMES_RXSTAT_MSGTYPE) == HERMES_RXSTAT_1042) ||
- ((status & HERMES_RXSTAT_MSGTYPE) == HERMES_RXSTAT_TUNNEL) ||
- is_ethersnap(&hdr)) {
+ if (length >= ENCAPS_OVERHEAD &&
+ (((status & HERMES_RXSTAT_MSGTYPE) == HERMES_RXSTAT_1042) ||
+ ((status & HERMES_RXSTAT_MSGTYPE) == HERMES_RXSTAT_TUNNEL) ||
+ is_ethersnap(skb->data))) {
/* These indicate a SNAP within 802.2 LLC within
802.11 frame which we'll need to de-encapsulate to
the original EthernetII frame. */
-
- if (length < ENCAPS_OVERHEAD) { /* No room for full LLC+SNAP */
- stats->rx_length_errors++;
- goto drop;
- }
-
- /* Remove SNAP header, reconstruct EthernetII frame */
- data_len = length - ENCAPS_OVERHEAD;
- data_off = HERMES_802_3_OFFSET + sizeof(hdr);
-
- eh = (struct ethhdr *)skb_put(skb, ETH_HLEN);
-
- memcpy(eh, &hdr, 2 * ETH_ALEN);
- eh->h_proto = hdr.ethertype;
+ hdr = (struct ethhdr *)skb_push(skb, ETH_HLEN - ENCAPS_OVERHEAD);
} else {
- /* All other cases indicate a genuine 802.3 frame. No
- decapsulation needed. We just throw the whole
- thing in, and hope the protocol layer can deal with
- it as 802.3 */
- data_len = length;
- data_off = HERMES_802_3_OFFSET;
- /* FIXME: we re-read from the card data we already read here */
- }
-
- p = skb_put(skb, data_len);
- err = hermes_bap_pread(hw, IRQ_BAP, p, ALIGN(data_len, 2),
- rxfid, data_off);
- if (err) {
- printk(KERN_ERR "%s: error %d reading frame. "
- "Frame dropped.\n", dev->name, err);
- stats->rx_errors++;
- goto drop;
- }
+ /* 802.3 frame - prepend 802.3 header as is */
+ hdr = (struct ethhdr *)skb_push(skb, ETH_HLEN);
+ hdr->h_proto = htons(length);
+ }
+ memcpy(hdr->h_dest, desc.addr1, ETH_ALEN);
+ if (fc & IEEE80211_FCTL_FROMDS)
+ memcpy(hdr->h_source, desc.addr3, ETH_ALEN);
+ else
+ memcpy(hdr->h_source, desc.addr2, ETH_ALEN);
dev->last_rx = jiffies;
skb->dev = dev;
skb->protocol = eth_type_trans(skb, dev);
skb->ip_summed = CHECKSUM_NONE;
+ if (fc & IEEE80211_FCTL_TODS)
+ skb->pkt_type = PACKET_OTHERHOST;
/* Process the wireless stats if needed */
orinoco_stat_gather(dev, skb, &desc);
@@ -1457,6 +1453,9 @@
u16 newstatus;
int connected;
+ if (priv->iw_mode == IW_MODE_MONITOR)
+ break;
+
if (len != sizeof(linkstatus)) {
printk(KERN_WARNING "%s: Unexpected size for linkstatus frame (%d bytes)\n",
dev->name, len);
^ permalink raw reply
* Re: git repository for net drivers available
From: Jeff Garzik @ 2005-05-14 14:02 UTC (permalink / raw)
To: Krzysztof Halasa; +Cc: Netdev, Linux Kernel
In-Reply-To: <m3oebe2hy8.fsf@defiant.localdomain>
Krzysztof Halasa wrote:
> Hi,
>
> Jeff Garzik <jgarzik@pobox.com> writes:
>
>
>>Although I have over 200 net driver patches to go through in my
>>'Pending' folder, I have fully converted the existing netdev-2.6
>>repository from BitKeeper to git. This includes the wireless-2.6
>>repository.
>
>
> Do you prefer plain patches or GIT repository on kernel.org to merge
> for future (and "current") updates?
> I haven't set a public repo up yet, but it's a matter of seconds.
For now, plain patches.
FWIW, I do still have your WAN update patch.
Jeff
^ permalink raw reply
* [1/1] xfrm: skb_cow_data() does not set proper owner for new skbs.
From: Evgeniy Polyakov @ 2005-05-14 13:48 UTC (permalink / raw)
To: netdev; +Cc: David S. Miller
It looks like skb_cow_data() does not set
proper owner for newly created skb.
If we have several fragments for skb and some of them
are shared(?) or cloned (like in async IPsec) there
might be a situation when we require recreating skb and
thus using skb_copy() for it.
Newly created skb has neither a destructor nor a socket
assotiated with it, which must be copied from the old skb.
As far as I can see, current code sets destructor and socket
for the first one skb only and uses truesize of the first skb
only to increment sk_wmem_alloc value.
If above "analysis" is correct then attached patch fixes that.
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
--- ./net/xfrm/xfrm_algo.c~ 2005-04-27 12:08:59.000000000 +0400
+++ ./net/xfrm/xfrm_algo.c 2005-05-14 17:36:52.000000000 +0400
@@ -698,7 +698,7 @@
return -ENOMEM;
if (skb1->sk)
- skb_set_owner_w(skb, skb1->sk);
+ skb_set_owner_w(skb2, skb1->sk);
/* Looking around. Are we still alive?
* OK, link new skb, drop old one */
--
Evgeniy Polyakov
Crash is better than data corruption. -- Artur Grabowski
^ permalink raw reply
* Re: sis900 and vlan support
From: Axel Reinhold @ 2005-05-14 12:59 UTC (permalink / raw)
To: Daniele Venzano; +Cc: romieu, netdev
In-Reply-To: <a36d6b20f493b0e80b82f68e5f157606@brownhat.org>
According to romieu@fr.zoreil.com:
Please keep netdev@oss.sgi.com in the Cc: : my mailboxes are not indexed
by Google nor accessible to the usual contributors.
According to Daniele Venzano:
> Thanks for the pointers, I'll see if there is something I can do, but
> bear in mind that I'm not the 2.4 maintainer.
>
If you could provde a patch - i'm happy.
> On 12/mag/05, at 20:00, Axel Reinhold wrote:
> > i tried the actual driver in a 2.4.30 tree:
> >
> > /lib/modules/2.4.30-ARXc3/kernel/drivers/net/sis900.o: unresolved
> > symbol pci_get_device
> > /lib/modules/2.4.30-ARXc3/kernel/drivers/net/sis900.o: unresolved
> > symbol pci_dev_put
> You should try a whole 2.6 kernel, not only the driver, there have been
> big core changes and, as you have seen, things are no more compatible.
> If I'm sure the problem exists also on 2.6 I can try to do something
> about it.
>
Can not use full 2.6 on my server at the moment. There are still
things not working in 2.6 (i need lvm-snapshots, umsdos, iBCS
emulation, ...) - so stuck into 2.4 for the next year at least.
> If I understand right to reproduce I have to load the vlan module,
> create a vlan device (how??) and send packets with a size near to the
> MTU. Right ?
>
my setup:
modprobe sis900
modprobe 8021q
ifconfig eth0 192.168.207.68
vconfig add eth0 7
vconfig add eth0 8
ifconfig eth0.7 192.168.7.68
ifconfig eth0.8 192.168.8.68
the switchport must be configured as a trunc and allow tagged
vlan membership for at least VLAN 7 and 8.
You need another device in one of the VLANs to ping to.
Here's the output of a little test-script - it pings with sizes
from 1466 to 1474 - from 1470 on the ping failes:
from dmesg:
sis900.c: v1.08.07 11/02/2003
PCI: Found IRQ 12 for device 00:01.1
eth0: SiS 900 Internal MII PHY transceiver found at address 1.
eth0: Using transceiver found at address 1 as default
eth0: SiS 900 PCI Fast Ethernet at 0xd400, IRQ 12, 00:e0:18:3a:db:1a.
802.1Q VLAN Support v1.8 Ben Greear <greearb@candelatech.com>
All bugs added by David S. Miller <davem@redhat.com>
eth0.7: add 01:00:5e:00:00:01 mcast address to master interface
eth0.8: add 01:00:5e:00:00:01 mcast address to master interface
+ cat /proc/net/vlan/config
VLAN Dev name | VLAN ID
Name-Type: VLAN_NAME_TYPE_RAW_PLUS_VID_NO_PAD
eth0.7 | 7 | eth0
eth0.8 | 8 | eth0
+ cat /proc/net/vlan/eth0.7
eth0.7 VID: 7 REORDER_HDR: 1 dev->priv_flags: 1
total frames received: 192
total bytes received: 13276
Broadcast/Multicast Rcvd: 0
total frames transmitted: 231
total bytes transmitted: 41025
total headroom inc: 0
total encap on xmit: 231
Device: eth0
INGRESS priority mappings: 0:0 1:0 2:0 3:0 4:0 5:0 6:0 7:0
EGRESSS priority Mappings:
+ cat /proc/net/vlan/eth0.8
eth0.8 VID: 8 REORDER_HDR: 1 dev->priv_flags: 1
total frames received: 99
total bytes received: 16978
Broadcast/Multicast Rcvd: 0
total frames transmitted: 137
total bytes transmitted: 14456
total headroom inc: 0
total encap on xmit: 137
Device: eth0
INGRESS priority mappings: 0:0 1:0 2:0 3:0 4:0 5:0 6:0 7:0
EGRESSS priority Mappings:
+ /sbin/ifconfig -s -a
Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK TX-ERR TX-DRP TX-OVR Flg
eth0 1500 0 293 0 0 0 374 0 0 0 BMRU
eth0.7 1500 0 194 0 0 0 234 0 0 0 BMRU
eth0.8 1500 0 99 0 0 0 137 0 0 0 BMRU
ippp2 1500 0 0 0 0 0 0 0 0 0 OPRU
ippp6 1500 0 0 0 0 0 0 0 0 0 OPRU
lo 16436 0 689 0 0 0 689 0 0 0 LRU
+ ping -c4 -s1466 joes
PING joes.freakout.de (192.168.8.66) from 192.168.8.68 : 1466(1494) bytes of data.
1474 bytes from joes.freakout.de (192.168.8.66): icmp_seq=1 ttl=64 time=2.32 ms
1474 bytes from joes.freakout.de (192.168.8.66): icmp_seq=2 ttl=64 time=2.06 ms
1474 bytes from joes.freakout.de (192.168.8.66): icmp_seq=3 ttl=64 time=2.06 ms
1474 bytes from joes.freakout.de (192.168.8.66): icmp_seq=4 ttl=64 time=0.934 ms
--- joes.freakout.de ping statistics ---
4 packets transmitted, 4 received, 0% loss, time 3024ms
rtt min/avg/max/mdev = 0.934/1.848/2.326/0.540 ms
+ ping -c4 -s1468 joes
PING joes.freakout.de (192.168.8.66) from 192.168.8.68 : 1468(1496) bytes of data.
1476 bytes from joes.freakout.de (192.168.8.66): icmp_seq=1 ttl=64 time=2.10 ms
1476 bytes from joes.freakout.de (192.168.8.66): icmp_seq=2 ttl=64 time=0.921 ms
1476 bytes from joes.freakout.de (192.168.8.66): icmp_seq=3 ttl=64 time=2.06 ms
1476 bytes from joes.freakout.de (192.168.8.66): icmp_seq=4 ttl=64 time=0.905 ms
--- joes.freakout.de ping statistics ---
4 packets transmitted, 4 received, 0% loss, time 3029ms
rtt min/avg/max/mdev = 0.905/1.498/2.103/0.587 ms
+ ping -c4 -s1470 joes
PING joes.freakout.de (192.168.8.66) from 192.168.8.68 : 1470(1498) bytes of data.
--- joes.freakout.de ping statistics ---
4 packets transmitted, 0 received, 100% loss, time 2999ms
+ ping -c4 -s1472 joes
PING joes.freakout.de (192.168.8.66) from 192.168.8.68 : 1472(1500) bytes of data.
--- joes.freakout.de ping statistics ---
2 packets transmitted, 0 received, 100% loss, time 1018ms
+ ping -c4 -s1474 joes
PING joes.freakout.de (192.168.8.66) from 192.168.8.68 : 1474(1502) bytes of data.
--- joes.freakout.de ping statistics ---
1 packets transmitted, 0 received, 100% loss, time 0ms
--
|------------------------+--------------------------------------|
| Axel Reinhold | Fax: +49-9287-8244 |
| Franz-Heinrich-Str. 20 | eMail: axel@freakout.de |
| 95100 Selb | http://www.freakout.de |
| Germany | Please do not send more than 100 kilobytes |
|------------------+--------------------------------------------|
| Fingerprint: 8D EF 9F 22 DF 9A 9B 68 E5 8C 12 C7 8D 6A 97 4E |
|---------------------------------------------------------------|
| Legal Warning: Do NOT send unsolicited commercial email to me |
^ permalink raw reply
* Re: 2.4.30-hf1 do_IRQ stack overflows
From: Herbert Xu @ 2005-05-14 11:05 UTC (permalink / raw)
To: Manfred Schwarb; +Cc: marcelo.tosatti, linux-kernel, davem, netdev
In-Reply-To: <21780.1115887477@www69.gmx.net>
On Thu, May 12, 2005 at 10:44:37AM +0200, Manfred Schwarb wrote:
>
> Trace; f8b531fc <[reiserfs]reiserfs_insert_item+14c/150>
> Trace; c02387be <__kfree_skb+fe/160>
> Trace; c02387be <__kfree_skb+fe/160>
> Trace; f90dd5f4 <[8139too]rtl8139_start_xmit+84/180>
Do you have any funky netfilter/iptables modules loaded?
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox