* [PATCH v2 01/15] megaraid_sas: Do not allow PCI access during OCR
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 02/15] megaraid_sas: MFI IO timeout handling Sumit Saxena
` (14 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
This patch will do synhronization between OCR function and AEN function using "reset_mutex" lock.
reset_mutex will be acquire only in first half of the AEN function which issue DCMD. Second half
of the function calls SCSI API (scsi_add_device/scsi_remove_device) should be out of reset_mutex
to avoid deadlock between scsi_eh thread and Driver.
During chip reset(inside OCR function), there should not be any PCI access and AEN function
(which is called in delayed context) may be firirng DCMDs(doing PCI writes) when chip reset is
happening in parallel which will cause FW fault. This patch will solve the problem by making
AEN thread and OCR thread mutually exclusive.
There are no changes in this patch from last time sent patch.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 2 +
drivers/scsi/megaraid/megaraid_sas_base.c | 254 ++++++++++--------------------
2 files changed, 82 insertions(+), 174 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index c0f7c8c..ef4ff03 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -1083,6 +1083,8 @@ struct megasas_ctrl_info {
#define VD_EXT_DEBUG 0
+#define SCAN_PD_CHANNEL 0x1
+#define SCAN_VD_CHANNEL 0x2
enum MR_SCSI_CMD_TYPE {
READ_WRITE_LDIO = 0,
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 97a1c1c..9650487 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -5476,7 +5476,6 @@ static int megasas_probe_one(struct pci_dev *pdev,
spin_lock_init(&instance->hba_lock);
spin_lock_init(&instance->completion_lock);
- mutex_init(&instance->aen_mutex);
mutex_init(&instance->reset_mutex);
/*
@@ -6442,10 +6441,10 @@ static int megasas_mgmt_ioctl_aen(struct file *file, unsigned long arg)
}
spin_unlock_irqrestore(&instance->hba_lock, flags);
- mutex_lock(&instance->aen_mutex);
+ mutex_lock(&instance->reset_mutex);
error = megasas_register_aen(instance, aen.seq_num,
aen.class_locale_word);
- mutex_unlock(&instance->aen_mutex);
+ mutex_unlock(&instance->reset_mutex);
return error;
}
@@ -6647,6 +6646,7 @@ megasas_aen_polling(struct work_struct *work)
int i, j, doscan = 0;
u32 seq_num, wait_time = MEGASAS_RESET_WAIT_TIME;
int error;
+ u8 dcmd_ret = 0;
if (!instance) {
printk(KERN_ERR "invalid instance!\n");
@@ -6659,16 +6659,7 @@ megasas_aen_polling(struct work_struct *work)
wait_time = MEGASAS_ROUTINE_WAIT_TIME_VF;
/* Don't run the event workqueue thread if OCR is running */
- for (i = 0; i < wait_time; i++) {
- if (instance->adprecovery == MEGASAS_HBA_OPERATIONAL)
- break;
- if (!(i % MEGASAS_RESET_NOTICE_INTERVAL)) {
- dev_notice(&instance->pdev->dev, "%s waiting for "
- "controller reset to finish for scsi%d\n",
- __func__, instance->host->host_no);
- }
- msleep(1000);
- }
+ mutex_lock(&instance->reset_mutex);
instance->ev = NULL;
host = instance->host;
@@ -6676,212 +6667,127 @@ megasas_aen_polling(struct work_struct *work)
megasas_decode_evt(instance);
switch (le32_to_cpu(instance->evt_detail->code)) {
- case MR_EVT_PD_INSERTED:
- if (megasas_get_pd_list(instance) == 0) {
- for (i = 0; i < MEGASAS_MAX_PD_CHANNELS; i++) {
- for (j = 0;
- j < MEGASAS_MAX_DEV_PER_CHANNEL;
- j++) {
-
- pd_index =
- (i * MEGASAS_MAX_DEV_PER_CHANNEL) + j;
-
- sdev1 = scsi_device_lookup(host, i, j, 0);
-
- if (instance->pd_list[pd_index].driveState
- == MR_PD_STATE_SYSTEM) {
- if (!sdev1)
- scsi_add_device(host, i, j, 0);
-
- if (sdev1)
- scsi_device_put(sdev1);
- }
- }
- }
- }
- doscan = 0;
- break;
+ case MR_EVT_PD_INSERTED:
case MR_EVT_PD_REMOVED:
- if (megasas_get_pd_list(instance) == 0) {
- for (i = 0; i < MEGASAS_MAX_PD_CHANNELS; i++) {
- for (j = 0;
- j < MEGASAS_MAX_DEV_PER_CHANNEL;
- j++) {
-
- pd_index =
- (i * MEGASAS_MAX_DEV_PER_CHANNEL) + j;
-
- sdev1 = scsi_device_lookup(host, i, j, 0);
-
- if (instance->pd_list[pd_index].driveState
- == MR_PD_STATE_SYSTEM) {
- if (sdev1)
- scsi_device_put(sdev1);
- } else {
- if (sdev1) {
- scsi_remove_device(sdev1);
- scsi_device_put(sdev1);
- }
- }
- }
- }
- }
- doscan = 0;
+ dcmd_ret = megasas_get_pd_list(instance);
+ if (dcmd_ret == 0)
+ doscan = SCAN_PD_CHANNEL;
break;
case MR_EVT_LD_OFFLINE:
case MR_EVT_CFG_CLEARED:
case MR_EVT_LD_DELETED:
- if (!instance->requestorId ||
- megasas_get_ld_vf_affiliation(instance, 0)) {
- if (megasas_ld_list_query(instance,
- MR_LD_QUERY_TYPE_EXPOSED_TO_HOST))
- megasas_get_ld_list(instance);
- for (i = 0; i < MEGASAS_MAX_LD_CHANNELS; i++) {
- for (j = 0;
- j < MEGASAS_MAX_DEV_PER_CHANNEL;
- j++) {
-
- ld_index =
- (i * MEGASAS_MAX_DEV_PER_CHANNEL) + j;
-
- sdev1 = scsi_device_lookup(host, MEGASAS_MAX_PD_CHANNELS + i, j, 0);
-
- if (instance->ld_ids[ld_index]
- != 0xff) {
- if (sdev1)
- scsi_device_put(sdev1);
- } else {
- if (sdev1) {
- scsi_remove_device(sdev1);
- scsi_device_put(sdev1);
- }
- }
- }
- }
- doscan = 0;
- }
- break;
case MR_EVT_LD_CREATED:
if (!instance->requestorId ||
- megasas_get_ld_vf_affiliation(instance, 0)) {
- if (megasas_ld_list_query(instance,
- MR_LD_QUERY_TYPE_EXPOSED_TO_HOST))
- megasas_get_ld_list(instance);
- for (i = 0; i < MEGASAS_MAX_LD_CHANNELS; i++) {
- for (j = 0;
- j < MEGASAS_MAX_DEV_PER_CHANNEL;
- j++) {
- ld_index =
- (i * MEGASAS_MAX_DEV_PER_CHANNEL) + j;
-
- sdev1 = scsi_device_lookup(host, MEGASAS_MAX_PD_CHANNELS + i, j, 0);
-
- if (instance->ld_ids[ld_index]
- != 0xff) {
- if (!sdev1)
- scsi_add_device(host, MEGASAS_MAX_PD_CHANNELS + i, j, 0);
- }
- if (sdev1)
- scsi_device_put(sdev1);
- }
- }
- doscan = 0;
- }
+ (instance->requestorId && megasas_get_ld_vf_affiliation(instance, 0)))
+ dcmd_ret = megasas_ld_list_query(instance, MR_LD_QUERY_TYPE_EXPOSED_TO_HOST);
+
+ if (dcmd_ret == 0)
+ doscan = SCAN_VD_CHANNEL;
+
break;
+
case MR_EVT_CTRL_HOST_BUS_SCAN_REQUESTED:
case MR_EVT_FOREIGN_CFG_IMPORTED:
case MR_EVT_LD_STATE_CHANGE:
- doscan = 1;
+ dcmd_ret = megasas_get_pd_list(instance);
+
+ if (dcmd_ret != 0)
+ break;
+
+ if (!instance->requestorId ||
+ (instance->requestorId && megasas_get_ld_vf_affiliation(instance, 0)))
+ dcmd_ret = megasas_ld_list_query(instance, MR_LD_QUERY_TYPE_EXPOSED_TO_HOST);
+
+ if (dcmd_ret != 0)
+ break;
+
+ doscan = SCAN_VD_CHANNEL | SCAN_PD_CHANNEL;
+ dev_info(&instance->pdev->dev, "scanning for scsi%d...\n",
+ instance->host->host_no);
break;
+
case MR_EVT_CTRL_PROP_CHANGED:
- megasas_get_ctrl_info(instance);
- break;
+ dcmd_ret = megasas_get_ctrl_info(instance);
+ break;
default:
doscan = 0;
break;
}
} else {
dev_err(&instance->pdev->dev, "invalid evt_detail!\n");
+ mutex_unlock(&instance->reset_mutex);
kfree(ev);
return;
}
- if (doscan) {
- dev_info(&instance->pdev->dev, "scanning for scsi%d...\n",
- instance->host->host_no);
- if (megasas_get_pd_list(instance) == 0) {
- for (i = 0; i < MEGASAS_MAX_PD_CHANNELS; i++) {
- for (j = 0; j < MEGASAS_MAX_DEV_PER_CHANNEL; j++) {
- pd_index = i*MEGASAS_MAX_DEV_PER_CHANNEL + j;
- sdev1 = scsi_device_lookup(host, i, j, 0);
- if (instance->pd_list[pd_index].driveState ==
- MR_PD_STATE_SYSTEM) {
- if (!sdev1) {
- scsi_add_device(host, i, j, 0);
- }
- if (sdev1)
- scsi_device_put(sdev1);
- } else {
- if (sdev1) {
- scsi_remove_device(sdev1);
- scsi_device_put(sdev1);
- }
+ mutex_unlock(&instance->reset_mutex);
+
+ if (doscan & SCAN_PD_CHANNEL) {
+ for (i = 0; i < MEGASAS_MAX_PD_CHANNELS; i++) {
+ for (j = 0; j < MEGASAS_MAX_DEV_PER_CHANNEL; j++) {
+ pd_index = i*MEGASAS_MAX_DEV_PER_CHANNEL + j;
+ sdev1 = scsi_device_lookup(host, i, j, 0);
+ if (instance->pd_list[pd_index].driveState ==
+ MR_PD_STATE_SYSTEM) {
+ if (!sdev1)
+ scsi_add_device(host, i, j, 0);
+ else
+ scsi_device_put(sdev1);
+ } else {
+ if (sdev1) {
+ scsi_remove_device(sdev1);
+ scsi_device_put(sdev1);
}
}
}
}
+ }
- if (!instance->requestorId ||
- megasas_get_ld_vf_affiliation(instance, 0)) {
- if (megasas_ld_list_query(instance,
- MR_LD_QUERY_TYPE_EXPOSED_TO_HOST))
- megasas_get_ld_list(instance);
- for (i = 0; i < MEGASAS_MAX_LD_CHANNELS; i++) {
- for (j = 0; j < MEGASAS_MAX_DEV_PER_CHANNEL;
- j++) {
- ld_index =
- (i * MEGASAS_MAX_DEV_PER_CHANNEL) + j;
-
- sdev1 = scsi_device_lookup(host,
- MEGASAS_MAX_PD_CHANNELS + i, j, 0);
- if (instance->ld_ids[ld_index]
- != 0xff) {
- if (!sdev1)
- scsi_add_device(host, MEGASAS_MAX_PD_CHANNELS + i, j, 0);
- else
- scsi_device_put(sdev1);
- } else {
- if (sdev1) {
- scsi_remove_device(sdev1);
- scsi_device_put(sdev1);
- }
+ if (doscan & SCAN_VD_CHANNEL) {
+ for (i = 0; i < MEGASAS_MAX_LD_CHANNELS; i++) {
+ for (j = 0; j < MEGASAS_MAX_DEV_PER_CHANNEL; j++) {
+ ld_index = (i * MEGASAS_MAX_DEV_PER_CHANNEL) + j;
+ sdev1 = scsi_device_lookup(host, MEGASAS_MAX_PD_CHANNELS + i, j, 0);
+ if (instance->ld_ids[ld_index] != 0xff) {
+ if (!sdev1)
+ scsi_add_device(host, MEGASAS_MAX_PD_CHANNELS + i, j, 0);
+ else
+ scsi_device_put(sdev1);
+ } else {
+ if (sdev1) {
+ scsi_remove_device(sdev1);
+ scsi_device_put(sdev1);
}
}
}
}
}
- if (instance->aen_cmd != NULL) {
- kfree(ev);
- return ;
- }
-
- seq_num = le32_to_cpu(instance->evt_detail->seq_num) + 1;
+ if (dcmd_ret == 0)
+ seq_num = le32_to_cpu(instance->evt_detail->seq_num) + 1;
+ else
+ seq_num = instance->last_seq_num;
/* Register AEN with FW for latest sequence number plus 1 */
class_locale.members.reserved = 0;
class_locale.members.locale = MR_EVT_LOCALE_ALL;
class_locale.members.class = MR_EVT_CLASS_DEBUG;
- mutex_lock(&instance->aen_mutex);
+
+ if (instance->aen_cmd != NULL) {
+ kfree(ev);
+ return;
+ }
+
+ mutex_lock(&instance->reset_mutex);
error = megasas_register_aen(instance, seq_num,
class_locale.word);
- mutex_unlock(&instance->aen_mutex);
-
if (error)
- dev_err(&instance->pdev->dev, "register aen failed error %x\n", error);
+ dev_err(&instance->pdev->dev,
+ "register aen failed error %x\n", error);
+ mutex_unlock(&instance->reset_mutex);
kfree(ev);
}
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 02/15] megaraid_sas: MFI IO timeout handling
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 01/15] megaraid_sas: Do not allow PCI access during OCR Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 03/15] megaraid_sas: Syncing request flags macro names with firmware Sumit Saxena
` (13 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
This patch will do proper error handling for DCMD timeout and failed cases for fusion adapters.
Below are few key design points-
1. For MFI adapters, in case of DCMD timeout(DCMD which must return SUCCESS) driver will call kill adapter.
2. What action needs to be taken in case of DCMD timeout is decided by function dcmd_timeout_ocr_possible().
DCMD timeout causing OCR is applicable to below DCMDs-
MR_DCMD_PD_LIST_QUERY
MR_DCMD_LD_GET_LIST
MR_DCMD_LD_LIST_QUERY
MR_DCMD_CTRL_SET_CRASH_DUMP_PARAMS
MR_DCMD_SYSTEM_PD_MAP_GET_INFO(for non pended DCMD)
MR_DCMD_LD_MAP_GET_INFO(for non pended DCMD)
3. If DCMD fails from Driver init path, there are certain DCMD which is must to be return SUCCESS. If those DCMD fails,
driver bail out load. For optional DCMD like pd_info etc, driver continue without executing certain functionality.
There are no changes in this patch from last time sent patch.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 22 +-
drivers/scsi/megaraid/megaraid_sas_base.c | 372 +++++++++++++++++++++-------
drivers/scsi/megaraid/megaraid_sas_fusion.c | 54 ++--
3 files changed, 338 insertions(+), 110 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index ef4ff03..dcc6ff8 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -170,6 +170,7 @@
/* Driver internal */
#define DRV_DCMD_POLLED_MODE 0x1
+#define DRV_DCMD_SKIP_REFIRE 0x2
/*
* Definition for cmd_status
@@ -1093,6 +1094,11 @@ enum MR_SCSI_CMD_TYPE {
NON_READ_WRITE_SYSPDIO = 3,
};
+enum DCMD_TIMEOUT_ACTION {
+ INITIATE_OCR = 0,
+ KILL_ADAPTER = 1,
+ IGNORE_TIMEOUT = 2,
+};
/* Frame Type */
#define IO_FRAME 0
#define PTHRU_FRAME 1
@@ -1139,6 +1145,7 @@ enum MR_SCSI_CMD_TYPE {
#define MFI_OB_INTR_STATUS_MASK 0x00000002
#define MFI_POLL_TIMEOUT_SECS 60
+#define MFI_IO_TIMEOUT_SECS 180
#define MEGASAS_SRIOV_HEARTBEAT_INTERVAL_VF (5 * HZ)
#define MEGASAS_OCR_SETTLE_TIME_VF (1000 * 30)
#define MEGASAS_ROUTINE_WAIT_TIME_VF 300
@@ -1918,7 +1925,7 @@ struct megasas_instance_template {
u32 (*init_adapter)(struct megasas_instance *);
u32 (*build_and_issue_cmd) (struct megasas_instance *,
struct scsi_cmnd *);
- void (*issue_dcmd) (struct megasas_instance *instance,
+ int (*issue_dcmd)(struct megasas_instance *instance,
struct megasas_cmd *cmd);
};
@@ -2016,6 +2023,19 @@ struct megasas_mgmt_info {
int max_index;
};
+enum MEGASAS_OCR_CAUSE {
+ FW_FAULT_OCR = 0,
+ SCSIIO_TIMEOUT_OCR = 1,
+ MFI_IO_TIMEOUT_OCR = 2,
+};
+
+enum DCMD_RETURN_STATUS {
+ DCMD_SUCCESS = 0,
+ DCMD_TIMEOUT = 1,
+ DCMD_FAILED = 2,
+ DCMD_NOT_FIRED = 3,
+};
+
u8
MR_BuildRaidContext(struct megasas_instance *instance,
struct IO_REQUEST_INFO *io_info,
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 9650487..380c627 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -196,11 +196,12 @@ static int megasas_get_ld_vf_affiliation(struct megasas_instance *instance,
int megasas_check_mpio_paths(struct megasas_instance *instance,
struct scsi_cmnd *scmd);
-void
+int
megasas_issue_dcmd(struct megasas_instance *instance, struct megasas_cmd *cmd)
{
instance->instancet->fire_cmd(instance,
cmd->frame_phys_addr, 0, instance->reg_set);
+ return 0;
}
/**
@@ -983,25 +984,20 @@ extern struct megasas_instance_template megasas_instance_template_fusion;
int
megasas_issue_polled(struct megasas_instance *instance, struct megasas_cmd *cmd)
{
- int seconds;
struct megasas_header *frame_hdr = &cmd->frame->hdr;
- frame_hdr->cmd_status = MFI_CMD_STATUS_POLL_MODE;
+ frame_hdr->cmd_status = MFI_STAT_INVALID_STATUS;
frame_hdr->flags |= cpu_to_le16(MFI_FRAME_DONT_POST_IN_REPLY_QUEUE);
- /*
- * Issue the frame using inbound queue port
- */
- instance->instancet->issue_dcmd(instance, cmd);
+ if ((instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) ||
+ (instance->instancet->issue_dcmd(instance, cmd))) {
+ dev_err(&instance->pdev->dev, "Failed from %s %d\n",
+ __func__, __LINE__);
+ return DCMD_NOT_FIRED;
+ }
- /*
- * Wait for cmd_status to change
- */
- if (instance->requestorId)
- seconds = MEGASAS_ROUTINE_WAIT_TIME_VF;
- else
- seconds = MFI_POLL_TIMEOUT_SECS;
- return wait_and_poll(instance, cmd, seconds);
+ return wait_and_poll(instance, cmd, instance->requestorId ?
+ MEGASAS_ROUTINE_WAIT_TIME_VF : MFI_IO_TIMEOUT_SECS);
}
/**
@@ -1019,21 +1015,29 @@ megasas_issue_blocked_cmd(struct megasas_instance *instance,
struct megasas_cmd *cmd, int timeout)
{
int ret = 0;
-
cmd->cmd_status_drv = MFI_STAT_INVALID_STATUS;
- instance->instancet->issue_dcmd(instance, cmd);
+ if ((instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) ||
+ (instance->instancet->issue_dcmd(instance, cmd))) {
+ dev_err(&instance->pdev->dev, "Failed from %s %d\n",
+ __func__, __LINE__);
+ return DCMD_NOT_FIRED;
+ }
+
if (timeout) {
ret = wait_event_timeout(instance->int_cmd_wait_q,
cmd->cmd_status_drv != MFI_STAT_INVALID_STATUS, timeout * HZ);
- if (!ret)
- return 1;
+ if (!ret) {
+ dev_err(&instance->pdev->dev, "Failed from %s %d DCMD Timed out\n",
+ __func__, __LINE__);
+ return DCMD_TIMEOUT;
+ }
} else
wait_event(instance->int_cmd_wait_q,
cmd->cmd_status_drv != MFI_STAT_INVALID_STATUS);
return (cmd->cmd_status_drv == MFI_STAT_OK) ?
- 0 : 1;
+ DCMD_SUCCESS : DCMD_FAILED;
}
/**
@@ -1077,15 +1081,20 @@ megasas_issue_blocked_abort_cmd(struct megasas_instance *instance,
cmd->sync_cmd = 1;
cmd->cmd_status_drv = MFI_STAT_INVALID_STATUS;
- instance->instancet->issue_dcmd(instance, cmd);
+ if ((instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) ||
+ (instance->instancet->issue_dcmd(instance, cmd))) {
+ dev_err(&instance->pdev->dev, "Failed from %s %d\n",
+ __func__, __LINE__);
+ return DCMD_NOT_FIRED;
+ }
if (timeout) {
ret = wait_event_timeout(instance->abort_cmd_wait_q,
cmd->cmd_status_drv != MFI_STAT_INVALID_STATUS, timeout * HZ);
if (!ret) {
- dev_err(&instance->pdev->dev, "Command timedout"
- "from %s\n", __func__);
- return 1;
+ dev_err(&instance->pdev->dev, "Failed from %s %d Abort Timed out\n",
+ __func__, __LINE__);
+ return DCMD_TIMEOUT;
}
} else
wait_event(instance->abort_cmd_wait_q,
@@ -1094,7 +1103,8 @@ megasas_issue_blocked_abort_cmd(struct megasas_instance *instance,
cmd->sync_cmd = 0;
megasas_return_cmd(instance, cmd);
- return 0;
+ return (cmd->cmd_status_drv == MFI_STAT_OK) ?
+ DCMD_SUCCESS : DCMD_FAILED;
}
/**
@@ -2054,9 +2064,7 @@ static int megasas_get_ld_vf_affiliation_111(struct megasas_instance *instance,
dev_warn(&instance->pdev->dev, "SR-IOV: Getting LD/VF affiliation for "
"scsi%d\n", instance->host->host_no);
- megasas_issue_blocked_cmd(instance, cmd, 0);
-
- if (dcmd->cmd_status) {
+ if (megasas_issue_blocked_cmd(instance, cmd, 0) != DCMD_SUCCESS) {
dev_warn(&instance->pdev->dev, "SR-IOV: LD/VF affiliation DCMD"
" failed with status 0x%x for scsi%d\n",
dcmd->cmd_status, instance->host->host_no);
@@ -2166,9 +2174,8 @@ static int megasas_get_ld_vf_affiliation_12(struct megasas_instance *instance,
dev_warn(&instance->pdev->dev, "SR-IOV: Getting LD/VF affiliation for "
"scsi%d\n", instance->host->host_no);
- megasas_issue_blocked_cmd(instance, cmd, 0);
- if (dcmd->cmd_status) {
+ if (megasas_issue_blocked_cmd(instance, cmd, 0) != DCMD_SUCCESS) {
dev_warn(&instance->pdev->dev, "SR-IOV: LD/VF affiliation DCMD"
" failed with status 0x%x for scsi%d\n",
dcmd->cmd_status, instance->host->host_no);
@@ -3852,6 +3859,25 @@ int megasas_alloc_cmds(struct megasas_instance *instance)
}
/*
+ * dcmd_timeout_ocr_possible - Check if OCR is possible based on Driver/FW state.
+ * @instance: Adapter soft state
+ *
+ * Return 0 for only Fusion adapter, if driver load/unload is not in progress
+ * or FW is not under OCR.
+ */
+inline int
+dcmd_timeout_ocr_possible(struct megasas_instance *instance) {
+
+ if (!instance->ctrl_context)
+ return KILL_ADAPTER;
+ else if (instance->unload ||
+ test_bit(MEGASAS_FUSION_IN_RESET, &instance->reset_flags))
+ return IGNORE_TIMEOUT;
+ else
+ return INITIATE_OCR;
+}
+
+/*
* megasas_get_pd_list_info - Returns FW's pd_list structure
* @instance: Adapter soft state
* @pd_list: pd_list structure
@@ -3906,42 +3932,72 @@ megasas_get_pd_list(struct megasas_instance *instance)
if (instance->ctrl_context && !instance->mask_interrupts)
ret = megasas_issue_blocked_cmd(instance, cmd,
- MEGASAS_BLOCKED_CMD_TIMEOUT);
+ MFI_IO_TIMEOUT_SECS);
else
ret = megasas_issue_polled(instance, cmd);
- /*
- * the following function will get the instance PD LIST.
- */
+ switch (ret) {
+ case DCMD_FAILED:
+ megaraid_sas_kill_hba(instance);
+ break;
+ case DCMD_TIMEOUT:
- pd_addr = ci->addr;
+ switch (dcmd_timeout_ocr_possible(instance)) {
+ case INITIATE_OCR:
+ cmd->flags |= DRV_DCMD_SKIP_REFIRE;
+ /*
+ * DCMD failed from AEN path.
+ * AEN path already hold reset_mutex to avoid PCI access
+ * while OCR is in progress.
+ */
+ mutex_unlock(&instance->reset_mutex);
+ megasas_reset_fusion(instance->host,
+ MFI_IO_TIMEOUT_OCR);
+ mutex_lock(&instance->reset_mutex);
+ break;
+ case KILL_ADAPTER:
+ megaraid_sas_kill_hba(instance);
+ break;
+ case IGNORE_TIMEOUT:
+ dev_info(&instance->pdev->dev, "Ignore DCMD timeout: %s %d \n",
+ __func__, __LINE__);
+ break;
+ }
- if (ret == 0 &&
- (le32_to_cpu(ci->count) <
- (MEGASAS_MAX_PD_CHANNELS * MEGASAS_MAX_DEV_PER_CHANNEL))) {
+ break;
+
+ case DCMD_SUCCESS:
+ pd_addr = ci->addr;
+
+ if ((le32_to_cpu(ci->count) >
+ (MEGASAS_MAX_PD_CHANNELS * MEGASAS_MAX_DEV_PER_CHANNEL)))
+ break;
memset(instance->local_pd_list, 0,
- MEGASAS_MAX_PD * sizeof(struct megasas_pd_list));
+ MEGASAS_MAX_PD * sizeof(struct megasas_pd_list));
for (pd_index = 0; pd_index < le32_to_cpu(ci->count); pd_index++) {
-
instance->local_pd_list[le16_to_cpu(pd_addr->deviceId)].tid =
- le16_to_cpu(pd_addr->deviceId);
+ le16_to_cpu(pd_addr->deviceId);
instance->local_pd_list[le16_to_cpu(pd_addr->deviceId)].driveType =
- pd_addr->scsiDevType;
+ pd_addr->scsiDevType;
instance->local_pd_list[le16_to_cpu(pd_addr->deviceId)].driveState =
- MR_PD_STATE_SYSTEM;
+ MR_PD_STATE_SYSTEM;
pd_addr++;
}
+
memcpy(instance->pd_list, instance->local_pd_list,
sizeof(instance->pd_list));
+ break;
+
}
pci_free_consistent(instance->pdev,
MEGASAS_MAX_PD * sizeof(struct MR_PD_LIST),
ci, ci_h);
- megasas_return_cmd(instance, cmd);
+ if (ret != DCMD_TIMEOUT)
+ megasas_return_cmd(instance, cmd);
return ret;
}
@@ -4002,33 +4058,63 @@ megasas_get_ld_list(struct megasas_instance *instance)
if (instance->ctrl_context && !instance->mask_interrupts)
ret = megasas_issue_blocked_cmd(instance, cmd,
- MEGASAS_BLOCKED_CMD_TIMEOUT);
+ MFI_IO_TIMEOUT_SECS);
else
ret = megasas_issue_polled(instance, cmd);
-
ld_count = le32_to_cpu(ci->ldCount);
- /* the following function will get the instance PD LIST */
+ switch (ret) {
+ case DCMD_FAILED:
+ megaraid_sas_kill_hba(instance);
+ break;
+ case DCMD_TIMEOUT:
+
+ switch (dcmd_timeout_ocr_possible(instance)) {
+ case INITIATE_OCR:
+ cmd->flags |= DRV_DCMD_SKIP_REFIRE;
+ /*
+ * DCMD failed from AEN path.
+ * AEN path already hold reset_mutex to avoid PCI access
+ * while OCR is in progress.
+ */
+ mutex_unlock(&instance->reset_mutex);
+ megasas_reset_fusion(instance->host,
+ MFI_IO_TIMEOUT_OCR);
+ mutex_lock(&instance->reset_mutex);
+ break;
+ case KILL_ADAPTER:
+ megaraid_sas_kill_hba(instance);
+ break;
+ case IGNORE_TIMEOUT:
+ dev_info(&instance->pdev->dev, "Ignore DCMD timeout: %s %d\n",
+ __func__, __LINE__);
+ break;
+ }
+
+ break;
+
+ case DCMD_SUCCESS:
+ if (ld_count > instance->fw_supported_vd_count)
+ break;
- if ((ret == 0) && (ld_count <= instance->fw_supported_vd_count)) {
memset(instance->ld_ids, 0xff, MAX_LOGICAL_DRIVES_EXT);
for (ld_index = 0; ld_index < ld_count; ld_index++) {
if (ci->ldList[ld_index].state != 0) {
ids = ci->ldList[ld_index].ref.targetId;
- instance->ld_ids[ids] =
- ci->ldList[ld_index].ref.targetId;
+ instance->ld_ids[ids] = ci->ldList[ld_index].ref.targetId;
}
}
+
+ break;
}
- pci_free_consistent(instance->pdev,
- sizeof(struct MR_LD_LIST),
- ci,
- ci_h);
+ pci_free_consistent(instance->pdev, sizeof(struct MR_LD_LIST), ci, ci_h);
+
+ if (ret != DCMD_TIMEOUT)
+ megasas_return_cmd(instance, cmd);
- megasas_return_cmd(instance, cmd);
return ret;
}
@@ -4090,26 +4176,61 @@ megasas_ld_list_query(struct megasas_instance *instance, u8 query_type)
dcmd->pad_0 = 0;
if (instance->ctrl_context && !instance->mask_interrupts)
- ret = megasas_issue_blocked_cmd(instance, cmd,
- MEGASAS_BLOCKED_CMD_TIMEOUT);
+ ret = megasas_issue_blocked_cmd(instance, cmd, MFI_IO_TIMEOUT_SECS);
else
ret = megasas_issue_polled(instance, cmd);
- tgtid_count = le32_to_cpu(ci->count);
+ switch (ret) {
+ case DCMD_FAILED:
+ dev_info(&instance->pdev->dev,
+ "DCMD not supported by firmware - %s %d\n",
+ __func__, __LINE__);
+ ret = megasas_get_ld_list(instance);
+ break;
+ case DCMD_TIMEOUT:
+ switch (dcmd_timeout_ocr_possible(instance)) {
+ case INITIATE_OCR:
+ cmd->flags |= DRV_DCMD_SKIP_REFIRE;
+ /*
+ * DCMD failed from AEN path.
+ * AEN path already hold reset_mutex to avoid PCI access
+ * while OCR is in progress.
+ */
+ mutex_unlock(&instance->reset_mutex);
+ megasas_reset_fusion(instance->host,
+ MFI_IO_TIMEOUT_OCR);
+ mutex_lock(&instance->reset_mutex);
+ break;
+ case KILL_ADAPTER:
+ megaraid_sas_kill_hba(instance);
+ break;
+ case IGNORE_TIMEOUT:
+ dev_info(&instance->pdev->dev, "Ignore DCMD timeout: %s %d\n",
+ __func__, __LINE__);
+ break;
+ }
+
+ break;
+ case DCMD_SUCCESS:
+ tgtid_count = le32_to_cpu(ci->count);
+
+ if ((tgtid_count > (instance->fw_supported_vd_count)))
+ break;
- if ((ret == 0) && (tgtid_count <= (instance->fw_supported_vd_count))) {
memset(instance->ld_ids, 0xff, MEGASAS_MAX_LD_IDS);
for (ld_index = 0; ld_index < tgtid_count; ld_index++) {
ids = ci->targetId[ld_index];
instance->ld_ids[ids] = ci->targetId[ld_index];
}
+ break;
}
pci_free_consistent(instance->pdev, sizeof(struct MR_LD_TARGETID_LIST),
- ci, ci_h);
+ ci, ci_h);
- megasas_return_cmd(instance, cmd);
+ if (ret != DCMD_TIMEOUT)
+ megasas_return_cmd(instance, cmd);
return ret;
}
@@ -4223,38 +4344,73 @@ megasas_get_ctrl_info(struct megasas_instance *instance)
dcmd->mbox.b[0] = 1;
if (instance->ctrl_context && !instance->mask_interrupts)
- ret = megasas_issue_blocked_cmd(instance, cmd,
- MEGASAS_BLOCKED_CMD_TIMEOUT);
+ ret = megasas_issue_blocked_cmd(instance, cmd, MFI_IO_TIMEOUT_SECS);
else
ret = megasas_issue_polled(instance, cmd);
- if (!ret) {
+ switch (ret) {
+ case DCMD_SUCCESS:
memcpy(ctrl_info, ci, sizeof(struct megasas_ctrl_info));
+ /* Save required controller information in
+ * CPU endianness format.
+ */
le32_to_cpus((u32 *)&ctrl_info->properties.OnOffProperties);
le32_to_cpus((u32 *)&ctrl_info->adapterOperations2);
le32_to_cpus((u32 *)&ctrl_info->adapterOperations3);
+
+ /* Update the latest Ext VD info.
+ * From Init path, store current firmware details.
+ * From OCR path, detect any firmware properties changes.
+ * in case of Firmware upgrade without system reboot.
+ */
megasas_update_ext_vd_details(instance);
instance->use_seqnum_jbod_fp =
ctrl_info->adapterOperations3.useSeqNumJbodFP;
+
+ /*Check whether controller is iMR or MR */
instance->is_imr = (ctrl_info->memory_size ? 0 : 1);
dev_info(&instance->pdev->dev,
- "controller type\t: %s(%dMB)\n",
- instance->is_imr ? "iMR" : "MR",
- le16_to_cpu(ctrl_info->memory_size));
+ "controller type\t: %s(%dMB)\n",
+ instance->is_imr ? "iMR" : "MR",
+ le16_to_cpu(ctrl_info->memory_size));
+
instance->disableOnlineCtrlReset =
ctrl_info->properties.OnOffProperties.disableOnlineCtrlReset;
- dev_info(&instance->pdev->dev, "Online Controller Reset(OCR)\t: %s\n",
- instance->disableOnlineCtrlReset ? "Disabled" : "Enabled");
instance->secure_jbod_support =
ctrl_info->adapterOperations3.supportSecurityonJBOD;
+ dev_info(&instance->pdev->dev, "Online Controller Reset(OCR)\t: %s\n",
+ instance->disableOnlineCtrlReset ? "Disabled" : "Enabled");
dev_info(&instance->pdev->dev, "Secure JBOD support\t: %s\n",
instance->secure_jbod_support ? "Yes" : "No");
+ break;
+
+ case DCMD_TIMEOUT:
+ switch (dcmd_timeout_ocr_possible(instance)) {
+ case INITIATE_OCR:
+ cmd->flags |= DRV_DCMD_SKIP_REFIRE;
+ megasas_reset_fusion(instance->host,
+ MFI_IO_TIMEOUT_OCR);
+ break;
+ case KILL_ADAPTER:
+ megaraid_sas_kill_hba(instance);
+ break;
+ case IGNORE_TIMEOUT:
+ dev_info(&instance->pdev->dev, "Ignore DCMD timeout: %s %d\n",
+ __func__, __LINE__);
+ break;
+ }
+ case DCMD_FAILED:
+ megaraid_sas_kill_hba(instance);
+ break;
+
}
pci_free_consistent(instance->pdev, sizeof(struct megasas_ctrl_info),
ci, ci_h);
megasas_return_cmd(instance, cmd);
+
+
return ret;
}
@@ -4304,12 +4460,28 @@ int megasas_set_crash_dump_params(struct megasas_instance *instance,
dcmd->sgl.sge32[0].length = cpu_to_le32(CRASH_DMA_BUF_SIZE);
if (instance->ctrl_context && !instance->mask_interrupts)
- ret = megasas_issue_blocked_cmd(instance, cmd,
- MEGASAS_BLOCKED_CMD_TIMEOUT);
+ ret = megasas_issue_blocked_cmd(instance, cmd, MFI_IO_TIMEOUT_SECS);
else
ret = megasas_issue_polled(instance, cmd);
- megasas_return_cmd(instance, cmd);
+ if (ret == DCMD_TIMEOUT) {
+ switch (dcmd_timeout_ocr_possible(instance)) {
+ case INITIATE_OCR:
+ cmd->flags |= DRV_DCMD_SKIP_REFIRE;
+ megasas_reset_fusion(instance->host,
+ MFI_IO_TIMEOUT_OCR);
+ break;
+ case KILL_ADAPTER:
+ megaraid_sas_kill_hba(instance);
+ break;
+ case IGNORE_TIMEOUT:
+ dev_info(&instance->pdev->dev, "Ignore DCMD timeout: %s %d\n",
+ __func__, __LINE__);
+ break;
+ }
+ } else
+ megasas_return_cmd(instance, cmd);
+
return ret;
}
@@ -5035,10 +5207,8 @@ megasas_get_seq_num(struct megasas_instance *instance,
dcmd->sgl.sge32[0].phys_addr = cpu_to_le32(el_info_h);
dcmd->sgl.sge32[0].length = cpu_to_le32(sizeof(struct megasas_evt_log_info));
- if (megasas_issue_blocked_cmd(instance, cmd, 30))
- dev_err(&instance->pdev->dev, "Command timedout"
- "from %s\n", __func__);
- else {
+ if (megasas_issue_blocked_cmd(instance, cmd, MFI_IO_TIMEOUT_SECS) ==
+ DCMD_SUCCESS) {
/*
* Copy the data back into callers buffer
*/
@@ -5047,7 +5217,9 @@ megasas_get_seq_num(struct megasas_instance *instance,
eli->clear_seq_num = el_info->clear_seq_num;
eli->shutdown_seq_num = el_info->shutdown_seq_num;
eli->boot_seq_num = el_info->boot_seq_num;
- }
+ } else
+ dev_err(&instance->pdev->dev, "DCMD failed "
+ "from %s\n", __func__);
pci_free_consistent(instance->pdev, sizeof(struct megasas_evt_log_info),
el_info, el_info_h);
@@ -5637,9 +5809,12 @@ static void megasas_flush_cache(struct megasas_instance *instance)
dcmd->opcode = cpu_to_le32(MR_DCMD_CTRL_CACHE_FLUSH);
dcmd->mbox.b[0] = MR_FLUSH_CTRL_CACHE | MR_FLUSH_DISK_CACHE;
- if (megasas_issue_blocked_cmd(instance, cmd, 30))
- dev_err(&instance->pdev->dev, "Command timedout"
- " from %s\n", __func__);
+ if (megasas_issue_blocked_cmd(instance, cmd, MFI_IO_TIMEOUT_SECS)
+ != DCMD_SUCCESS) {
+ dev_err(&instance->pdev->dev,
+ "return from %s %d\n", __func__, __LINE__);
+ return;
+ }
megasas_return_cmd(instance, cmd);
}
@@ -5665,13 +5840,13 @@ static void megasas_shutdown_controller(struct megasas_instance *instance,
if (instance->aen_cmd)
megasas_issue_blocked_abort_cmd(instance,
- instance->aen_cmd, MEGASAS_BLOCKED_CMD_TIMEOUT);
+ instance->aen_cmd, MFI_IO_TIMEOUT_SECS);
if (instance->map_update_cmd)
megasas_issue_blocked_abort_cmd(instance,
- instance->map_update_cmd, MEGASAS_BLOCKED_CMD_TIMEOUT);
+ instance->map_update_cmd, MFI_IO_TIMEOUT_SECS);
if (instance->jbod_seq_cmd)
megasas_issue_blocked_abort_cmd(instance,
- instance->jbod_seq_cmd, MEGASAS_BLOCKED_CMD_TIMEOUT);
+ instance->jbod_seq_cmd, MFI_IO_TIMEOUT_SECS);
dcmd = &cmd->frame->dcmd;
@@ -5686,9 +5861,12 @@ static void megasas_shutdown_controller(struct megasas_instance *instance,
dcmd->data_xfer_len = 0;
dcmd->opcode = cpu_to_le32(opcode);
- if (megasas_issue_blocked_cmd(instance, cmd, 30))
- dev_err(&instance->pdev->dev, "Command timedout"
- "from %s\n", __func__);
+ if (megasas_issue_blocked_cmd(instance, cmd, MFI_IO_TIMEOUT_SECS)
+ != DCMD_SUCCESS) {
+ dev_err(&instance->pdev->dev,
+ "return from %s %d\n", __func__, __LINE__);
+ return;
+ }
megasas_return_cmd(instance, cmd);
}
@@ -6226,7 +6404,15 @@ megasas_mgmt_fw_ioctl(struct megasas_instance *instance,
* cmd to the SCSI mid-layer
*/
cmd->sync_cmd = 1;
- megasas_issue_blocked_cmd(instance, cmd, 0);
+ if (megasas_issue_blocked_cmd(instance, cmd, 0) == DCMD_NOT_FIRED) {
+ cmd->sync_cmd = 0;
+ dev_err(&instance->pdev->dev,
+ "return -EBUSY from %s %d opcode 0x%x cmd->cmd_status_drv 0x%x\n",
+ __func__, __LINE__, cmd->frame->dcmd.opcode,
+ cmd->cmd_status_drv);
+ return -EBUSY;
+ }
+
cmd->sync_cmd = 0;
if (instance->unload == 1) {
@@ -6646,7 +6832,7 @@ megasas_aen_polling(struct work_struct *work)
int i, j, doscan = 0;
u32 seq_num, wait_time = MEGASAS_RESET_WAIT_TIME;
int error;
- u8 dcmd_ret = 0;
+ u8 dcmd_ret = DCMD_SUCCESS;
if (!instance) {
printk(KERN_ERR "invalid instance!\n");
@@ -6671,7 +6857,7 @@ megasas_aen_polling(struct work_struct *work)
case MR_EVT_PD_INSERTED:
case MR_EVT_PD_REMOVED:
dcmd_ret = megasas_get_pd_list(instance);
- if (dcmd_ret == 0)
+ if (dcmd_ret == DCMD_SUCCESS)
doscan = SCAN_PD_CHANNEL;
break;
@@ -6683,7 +6869,7 @@ megasas_aen_polling(struct work_struct *work)
(instance->requestorId && megasas_get_ld_vf_affiliation(instance, 0)))
dcmd_ret = megasas_ld_list_query(instance, MR_LD_QUERY_TYPE_EXPOSED_TO_HOST);
- if (dcmd_ret == 0)
+ if (dcmd_ret == DCMD_SUCCESS)
doscan = SCAN_VD_CHANNEL;
break;
@@ -6693,14 +6879,14 @@ megasas_aen_polling(struct work_struct *work)
case MR_EVT_LD_STATE_CHANGE:
dcmd_ret = megasas_get_pd_list(instance);
- if (dcmd_ret != 0)
+ if (dcmd_ret != DCMD_SUCCESS)
break;
if (!instance->requestorId ||
(instance->requestorId && megasas_get_ld_vf_affiliation(instance, 0)))
dcmd_ret = megasas_ld_list_query(instance, MR_LD_QUERY_TYPE_EXPOSED_TO_HOST);
- if (dcmd_ret != 0)
+ if (dcmd_ret != DCMD_SUCCESS)
break;
doscan = SCAN_VD_CHANNEL | SCAN_PD_CHANNEL;
@@ -6765,7 +6951,7 @@ megasas_aen_polling(struct work_struct *work)
}
}
- if (dcmd_ret == 0)
+ if (dcmd_ret == DCMD_SUCCESS)
seq_num = le32_to_cpu(instance->evt_detail->seq_num) + 1;
else
seq_num = instance->last_seq_num;
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 8d630a5..6e48707 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -576,11 +576,12 @@ wait_and_poll(struct megasas_instance *instance, struct megasas_cmd *cmd,
msleep(20);
}
- if (frame_hdr->cmd_status == 0xff)
- return -ETIME;
-
- return (frame_hdr->cmd_status == MFI_STAT_OK) ?
- 0 : 1;
+ if (frame_hdr->cmd_status == MFI_STAT_INVALID_STATUS)
+ return DCMD_TIMEOUT;
+ else if (frame_hdr->cmd_status == MFI_STAT_OK)
+ return DCMD_SUCCESS;
+ else
+ return DCMD_FAILED;
}
/**
@@ -784,7 +785,8 @@ megasas_sync_pd_seq_num(struct megasas_instance *instance, bool pend) {
/* Below code is only for non pended DCMD */
if (instance->ctrl_context && !instance->mask_interrupts)
- ret = megasas_issue_blocked_cmd(instance, cmd, 60);
+ ret = megasas_issue_blocked_cmd(instance, cmd,
+ MFI_IO_TIMEOUT_SECS);
else
ret = megasas_issue_polled(instance, cmd);
@@ -795,7 +797,10 @@ megasas_sync_pd_seq_num(struct megasas_instance *instance, bool pend) {
ret = -EINVAL;
}
- if (!ret)
+ if (ret == DCMD_TIMEOUT && instance->ctrl_context)
+ megaraid_sas_kill_hba(instance);
+
+ if (ret == DCMD_SUCCESS)
instance->pd_seq_map_id++;
megasas_return_cmd(instance, cmd);
@@ -875,10 +880,13 @@ megasas_get_ld_map_info(struct megasas_instance *instance)
if (instance->ctrl_context && !instance->mask_interrupts)
ret = megasas_issue_blocked_cmd(instance, cmd,
- MEGASAS_BLOCKED_CMD_TIMEOUT);
+ MFI_IO_TIMEOUT_SECS);
else
ret = megasas_issue_polled(instance, cmd);
+ if (ret == DCMD_TIMEOUT && instance->ctrl_context)
+ megaraid_sas_kill_hba(instance);
+
megasas_return_cmd(instance, cmd);
return ret;
@@ -2411,7 +2419,7 @@ build_mpt_cmd(struct megasas_instance *instance, struct megasas_cmd *cmd)
* @cmd: mfi cmd pointer
*
*/
-void
+int
megasas_issue_dcmd_fusion(struct megasas_instance *instance,
struct megasas_cmd *cmd)
{
@@ -2419,10 +2427,13 @@ megasas_issue_dcmd_fusion(struct megasas_instance *instance,
req_desc = build_mpt_cmd(instance, cmd);
if (!req_desc) {
- dev_err(&instance->pdev->dev, "Couldn't issue MFI pass thru cmd\n");
- return;
+ dev_info(&instance->pdev->dev, "Failed from %s %d\n",
+ __func__, __LINE__);
+ return DCMD_NOT_FIRED;
}
+
megasas_fire_cmd_fusion(instance, req_desc);
+ return DCMD_SUCCESS;
}
/**
@@ -2583,7 +2594,7 @@ megasas_check_reset_fusion(struct megasas_instance *instance,
/* This function waits for outstanding commands on fusion to complete */
int megasas_wait_for_outstanding_fusion(struct megasas_instance *instance,
- int iotimeout, int *convert)
+ int reason, int *convert)
{
int i, outstanding, retval = 0, hb_seconds_missed = 0;
u32 fw_state;
@@ -2599,14 +2610,22 @@ int megasas_wait_for_outstanding_fusion(struct megasas_instance *instance,
retval = 1;
goto out;
}
+
+ if (reason == MFI_IO_TIMEOUT_OCR) {
+ dev_info(&instance->pdev->dev,
+ "MFI IO is timed out, initiating OCR\n");
+ retval = 1;
+ goto out;
+ }
+
/* If SR-IOV VF mode & heartbeat timeout, don't wait */
- if (instance->requestorId && !iotimeout) {
+ if (instance->requestorId && !reason) {
retval = 1;
goto out;
}
/* If SR-IOV VF mode & I/O timeout, check for HB timeout */
- if (instance->requestorId && iotimeout) {
+ if (instance->requestorId && reason) {
if (instance->hb_host_mem->HB.fwCounter !=
instance->hb_host_mem->HB.driverCounter) {
instance->hb_host_mem->HB.driverCounter =
@@ -2680,6 +2699,7 @@ void megasas_refire_mgmt_cmd(struct megasas_instance *instance)
struct megasas_cmd *cmd_mfi;
union MEGASAS_REQUEST_DESCRIPTOR_UNION *req_desc;
u16 smid;
+ bool refire_cmd = 0;
fusion = instance->ctrl_context;
@@ -2695,10 +2715,12 @@ void megasas_refire_mgmt_cmd(struct megasas_instance *instance)
continue;
req_desc = megasas_get_request_descriptor
(instance, smid - 1);
- if (req_desc && ((cmd_mfi->frame->dcmd.opcode !=
+ refire_cmd = req_desc && ((cmd_mfi->frame->dcmd.opcode !=
cpu_to_le32(MR_DCMD_LD_MAP_GET_INFO)) &&
(cmd_mfi->frame->dcmd.opcode !=
- cpu_to_le32(MR_DCMD_SYSTEM_PD_MAP_GET_INFO))))
+ cpu_to_le32(MR_DCMD_SYSTEM_PD_MAP_GET_INFO)))
+ && !(cmd_mfi->flags & DRV_DCMD_SKIP_REFIRE);
+ if (refire_cmd)
megasas_fire_cmd_fusion(instance, req_desc);
else
megasas_return_cmd(instance, cmd_mfi);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 03/15] megaraid_sas: Syncing request flags macro names with firmware
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 01/15] megaraid_sas: Do not allow PCI access during OCR Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 02/15] megaraid_sas: MFI IO timeout handling Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 04/15] megaraid_sas: Task management support Sumit Saxena
` (12 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
There are no changes in this patch from last time sent patch.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
---
drivers/scsi/megaraid/megaraid_sas_fusion.c | 6 +++---
drivers/scsi/megaraid/megaraid_sas_fusion.h | 3 ++-
2 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 6e48707..1dc4537 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -1666,7 +1666,7 @@ megasas_build_ldio_fusion(struct megasas_instance *instance,
local_map_ptr, start_lba_lo);
io_request->Function = MPI2_FUNCTION_SCSI_IO_REQUEST;
cmd->request_desc->SCSIIO.RequestFlags =
- (MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY
+ (MPI2_REQ_DESCRIPT_FLAGS_FP_IO
<< MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
if (fusion->adapter_type == INVADER_SERIES) {
if (io_request->RaidContext.regLockFlags ==
@@ -1799,7 +1799,7 @@ static void megasas_build_ld_nonrw_fusion(struct megasas_instance *instance,
/* build request descriptor */
cmd->request_desc->SCSIIO.RequestFlags =
- (MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY <<
+ (MPI2_REQ_DESCRIPT_FLAGS_FP_IO <<
MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
cmd->request_desc->SCSIIO.DevHandle = devHandle;
@@ -1905,7 +1905,7 @@ megasas_build_syspd_fusion(struct megasas_instance *instance,
cpu_to_le16(MPI25_SAS_DEVICE0_FLAGS_ENABLED_FAST_PATH);
}
cmd->request_desc->SCSIIO.RequestFlags =
- (MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY <<
+ (MPI2_REQ_DESCRIPT_FLAGS_FP_IO <<
MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
}
}
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.h b/drivers/scsi/megaraid/megaraid_sas_fusion.h
index 473005c..a9e10c4 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.h
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.h
@@ -176,7 +176,8 @@ enum REGION_TYPE {
#define MPI2_SCSIIO_EEDPFLAGS_CHECK_GUARD (0x0100)
#define MPI2_SCSIIO_EEDPFLAGS_INSERT_OP (0x0004)
#define MPI2_FUNCTION_SCSI_IO_REQUEST (0x00) /* SCSI IO */
-#define MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY (0x06)
+#define MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY (0x03)
+#define MPI2_REQ_DESCRIPT_FLAGS_FP_IO (0x06)
#define MPI2_REQ_DESCRIPT_FLAGS_SCSI_IO (0x00)
#define MPI2_SGE_FLAGS_64_BIT_ADDRESSING (0x02)
#define MPI2_SCSIIO_CONTROL_WRITE (0x01000000)
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 04/15] megaraid_sas: Task management support
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (2 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 03/15] megaraid_sas: Syncing request flags macro names with firmware Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 05/15] megaraid_sas: Update device Queue depth based on interface type Sumit Saxena
` (11 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
This patch will add task management(TM) support for SCSI commands in megaraid_sas driver.
Added TM functions are below-
1)Task abort
2)Target reset
Below are few key points-
1. Currently, megaraid_sas driver performs Controller reset when any IO times out.
With these TM support added in driver, in case of IO timeout task abort and target
reset will be tried to recover timed out IO. If both fails to recover IO, then
Controller reset will be called. If the TM request times out, fail the TM and escalate
to the next level(Controller reset).
2. mr_device_priv_data will be allocated for all generation of controller, but is_tm_capable
flag will never be set for older controllers (prior to Invader series) as firmware support is not
available for T.M functionality.
3. whichever firmware is capable for TM will set is_tm_capable flag in firmware API, which will be used
by Driver to pass TM frame to firmware or return back to OS as Failure to escalate next level of Error handling.
Additionally, Tomas Henzl's feedback has been accomodated in this patch- return MFI frame used for task management(TM)
in case of TM timeout and error cases.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 13 +
drivers/scsi/megaraid/megaraid_sas_base.c | 60 +++-
drivers/scsi/megaraid/megaraid_sas_fusion.c | 474 +++++++++++++++++++++++++++-
drivers/scsi/megaraid/megaraid_sas_fusion.h | 117 ++++++-
4 files changed, 646 insertions(+), 18 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index dcc6ff8..0fcb156 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -1520,6 +1520,15 @@ union megasas_frame {
u8 raw_bytes[64];
};
+/**
+ * struct MR_PRIV_DEVICE - sdev private hostdata
+ * @is_tm_capable: firmware managed tm_capable flag
+ * @tm_busy: TM request is in progress
+ */
+struct MR_PRIV_DEVICE {
+ bool is_tm_capable;
+ bool tm_busy;
+};
struct megasas_cmd;
union megasas_evt_class_locale {
@@ -2073,4 +2082,8 @@ void megasas_return_mfi_mpt_pthr(struct megasas_instance *instance,
int megasas_cmd_type(struct scsi_cmnd *cmd);
void megasas_setup_jbod_map(struct megasas_instance *instance);
+void megasas_update_sdev_properties(struct scsi_device *sdev);
+int megasas_reset_fusion(struct Scsi_Host *shost, int reason);
+int megasas_task_abort_fusion(struct scsi_cmnd *scmd);
+int megasas_reset_target_fusion(struct scsi_cmnd *scmd);
#endif /*LSI_MEGARAID_SAS_H */
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 380c627..57cf4e3 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -189,7 +189,6 @@ int
wait_and_poll(struct megasas_instance *instance, struct megasas_cmd *cmd,
int seconds);
void megasas_reset_reply_desc(struct megasas_instance *instance);
-int megasas_reset_fusion(struct Scsi_Host *shost, int iotimeout);
void megasas_fusion_ocr_wq(struct work_struct *work);
static int megasas_get_ld_vf_affiliation(struct megasas_instance *instance,
int initial);
@@ -1645,6 +1644,7 @@ megasas_queue_command(struct Scsi_Host *shost, struct scsi_cmnd *scmd)
{
struct megasas_instance *instance;
unsigned long flags;
+ struct MR_PRIV_DEVICE *mr_device_priv_data;
instance = (struct megasas_instance *)
scmd->device->host->hostdata;
@@ -1681,11 +1681,24 @@ megasas_queue_command(struct Scsi_Host *shost, struct scsi_cmnd *scmd)
return 0;
}
+ mr_device_priv_data = scmd->device->hostdata;
+ if (!mr_device_priv_data) {
+ spin_unlock_irqrestore(&instance->hba_lock, flags);
+ scmd->result = DID_NO_CONNECT << 16;
+ scmd->scsi_done(scmd);
+ return 0;
+ }
+
if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
spin_unlock_irqrestore(&instance->hba_lock, flags);
return SCSI_MLQUEUE_HOST_BUSY;
}
+ if (mr_device_priv_data->tm_busy) {
+ spin_unlock_irqrestore(&instance->hba_lock, flags);
+ return SCSI_MLQUEUE_DEVICE_BUSY;
+ }
+
spin_unlock_irqrestore(&instance->hba_lock, flags);
scmd->result = 0;
@@ -1736,27 +1749,39 @@ static struct megasas_instance *megasas_lookup_instance(u16 host_no)
}
/*
-* megasas_set_dma_alignment - Set DMA alignment for PI enabled VD
+* megasas_update_sdev_properties - Update sdev structure based on controller's FW capabilities
*
* @sdev: OS provided scsi device
*
* Returns void
*/
-static void megasas_set_dma_alignment(struct scsi_device *sdev)
+void megasas_update_sdev_properties(struct scsi_device *sdev)
{
+ u16 pd_index = 0;
u32 device_id, ld;
struct megasas_instance *instance;
struct fusion_context *fusion;
+ struct MR_PRIV_DEVICE *mr_device_priv_data;
+ struct MR_PD_CFG_SEQ_NUM_SYNC *pd_sync;
struct MR_LD_RAID *raid;
struct MR_DRV_RAID_MAP_ALL *local_map_ptr;
instance = megasas_lookup_instance(sdev->host->host_no);
fusion = instance->ctrl_context;
+ mr_device_priv_data = sdev->hostdata;
if (!fusion)
return;
- if (sdev->channel >= MEGASAS_MAX_PD_CHANNELS) {
+ if (sdev->channel < MEGASAS_MAX_PD_CHANNELS &&
+ instance->use_seqnum_jbod_fp) {
+ pd_index = (sdev->channel * MEGASAS_MAX_DEV_PER_CHANNEL) +
+ sdev->id;
+ pd_sync = (void *)fusion->pd_seq_sync
+ [(instance->pd_seq_map_id - 1) & 1];
+ mr_device_priv_data->is_tm_capable =
+ pd_sync->seq[pd_index].capability.tmCapable;
+ } else {
device_id = ((sdev->channel % 2) * MEGASAS_MAX_DEV_PER_CHANNEL)
+ sdev->id;
local_map_ptr = fusion->ld_drv_map[(instance->map_id & 1)];
@@ -1764,10 +1789,13 @@ static void megasas_set_dma_alignment(struct scsi_device *sdev)
raid = MR_LdRaidGet(ld, local_map_ptr);
if (raid->capability.ldPiMode == MR_PROT_INFO_TYPE_CONTROLLER)
- blk_queue_update_dma_alignment(sdev->request_queue, 0x7);
+ blk_queue_update_dma_alignment(sdev->request_queue, 0x7);
+ mr_device_priv_data->is_tm_capable =
+ raid->capability.tmCapable;
}
}
+
static int megasas_slave_configure(struct scsi_device *sdev)
{
u16 pd_index = 0;
@@ -1784,7 +1812,8 @@ static int megasas_slave_configure(struct scsi_device *sdev)
return -ENXIO;
}
}
- megasas_set_dma_alignment(sdev);
+ megasas_update_sdev_properties(sdev);
+
/*
* The RAID firmware may require extended timeouts.
*/
@@ -1798,6 +1827,7 @@ static int megasas_slave_alloc(struct scsi_device *sdev)
{
u16 pd_index = 0;
struct megasas_instance *instance ;
+ struct MR_PRIV_DEVICE *mr_device_priv_data;
instance = megasas_lookup_instance(sdev->host->host_no);
if (sdev->channel < MEGASAS_MAX_PD_CHANNELS) {
@@ -1809,13 +1839,26 @@ static int megasas_slave_alloc(struct scsi_device *sdev)
sdev->id;
if ((instance->allow_fw_scan || instance->pd_list[pd_index].driveState ==
MR_PD_STATE_SYSTEM)) {
- return 0;
+ goto scan_target;
}
return -ENXIO;
}
+
+scan_target:
+ mr_device_priv_data = kzalloc(sizeof(*mr_device_priv_data),
+ GFP_KERNEL);
+ if (!mr_device_priv_data)
+ return -ENOMEM;
+ sdev->hostdata = mr_device_priv_data;
return 0;
}
+static void megasas_slave_destroy(struct scsi_device *sdev)
+{
+ kfree(sdev->hostdata);
+ sdev->hostdata = NULL;
+}
+
/*
* megasas_complete_outstanding_ioctls - Complete outstanding ioctls after a
* kill adapter
@@ -2885,6 +2928,7 @@ static struct scsi_host_template megasas_template = {
.proc_name = "megaraid_sas",
.slave_configure = megasas_slave_configure,
.slave_alloc = megasas_slave_alloc,
+ .slave_destroy = megasas_slave_destroy,
.queuecommand = megasas_queue_command,
.eh_device_reset_handler = megasas_reset_device,
.eh_bus_reset_handler = megasas_reset_bus_host,
@@ -5434,6 +5478,8 @@ static int megasas_io_attach(struct megasas_instance *instance)
if (instance->ctrl_context) {
host->hostt->eh_device_reset_handler = NULL;
host->hostt->eh_bus_reset_handler = NULL;
+ host->hostt->eh_target_reset_handler = megasas_reset_target_fusion;
+ host->hostt->eh_abort_handler = megasas_task_abort_fusion;
}
/*
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 1dc4537..4b0c86c 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -2100,6 +2100,8 @@ complete_cmd_fusion(struct megasas_instance *instance, u32 MSIxIndex)
struct LD_LOAD_BALANCE_INFO *lbinfo;
int threshold_reply_count = 0;
struct scsi_cmnd *scmd_local = NULL;
+ struct MR_TASK_MANAGE_REQUEST *mr_tm_req;
+ struct MPI2_SCSI_TASK_MANAGE_REQUEST *mpi_tm_req;
fusion = instance->ctrl_context;
@@ -2141,6 +2143,16 @@ complete_cmd_fusion(struct megasas_instance *instance, u32 MSIxIndex)
extStatus = scsi_io_req->RaidContext.exStatus;
switch (scsi_io_req->Function) {
+ case MPI2_FUNCTION_SCSI_TASK_MGMT:
+ mr_tm_req = (struct MR_TASK_MANAGE_REQUEST *)
+ cmd_fusion->io_request;
+ mpi_tm_req = (struct MPI2_SCSI_TASK_MANAGE_REQUEST *)
+ &mr_tm_req->TmRequest;
+ dev_dbg(&instance->pdev->dev, "TM completion:"
+ "type: 0x%x TaskMID: 0x%x\n",
+ mpi_tm_req->TaskType, mpi_tm_req->TaskMID);
+ complete(&cmd_fusion->done);
+ break;
case MPI2_FUNCTION_SCSI_IO_REQUEST: /*Fast Path IO.*/
/* Update load balancing info */
device_id = MEGASAS_DEV_INDEX(scmd_local);
@@ -2727,6 +2739,452 @@ void megasas_refire_mgmt_cmd(struct megasas_instance *instance)
}
}
+/*
+ * megasas_track_scsiio : Track SCSI IOs outstanding to a SCSI device
+ * @instance: per adapter struct
+ * @channel: the channel assigned by the OS
+ * @id: the id assigned by the OS
+ *
+ * Returns SUCCESS if no IOs pending to SCSI device, else return FAILED
+ */
+
+static int megasas_track_scsiio(struct megasas_instance *instance,
+ int id, int channel)
+{
+ int i, found = 0;
+ struct megasas_cmd_fusion *cmd_fusion;
+ struct fusion_context *fusion;
+ fusion = instance->ctrl_context;
+
+ for (i = 0 ; i < instance->max_scsi_cmds; i++) {
+ cmd_fusion = fusion->cmd_list[i];
+ if (cmd_fusion->scmd &&
+ (cmd_fusion->scmd->device->id == id &&
+ cmd_fusion->scmd->device->channel == channel)) {
+ dev_info(&instance->pdev->dev,
+ "SCSI commands pending to target"
+ "channel %d id %d \tSMID: 0x%x\n",
+ channel, id, cmd_fusion->index);
+ scsi_print_command(cmd_fusion->scmd);
+ found = 1;
+ break;
+ }
+ }
+
+ return found ? FAILED : SUCCESS;
+}
+
+/**
+ * megasas_tm_response_code - translation of device response code
+ * @ioc: per adapter object
+ * @mpi_reply: MPI reply returned by firmware
+ *
+ * Return nothing.
+ */
+static void
+megasas_tm_response_code(struct megasas_instance *instance,
+ struct MPI2_SCSI_TASK_MANAGE_REPLY *mpi_reply)
+{
+ char *desc;
+
+ switch (mpi_reply->ResponseCode) {
+ case MPI2_SCSITASKMGMT_RSP_TM_COMPLETE:
+ desc = "task management request completed";
+ break;
+ case MPI2_SCSITASKMGMT_RSP_INVALID_FRAME:
+ desc = "invalid frame";
+ break;
+ case MPI2_SCSITASKMGMT_RSP_TM_NOT_SUPPORTED:
+ desc = "task management request not supported";
+ break;
+ case MPI2_SCSITASKMGMT_RSP_TM_FAILED:
+ desc = "task management request failed";
+ break;
+ case MPI2_SCSITASKMGMT_RSP_TM_SUCCEEDED:
+ desc = "task management request succeeded";
+ break;
+ case MPI2_SCSITASKMGMT_RSP_TM_INVALID_LUN:
+ desc = "invalid lun";
+ break;
+ case 0xA:
+ desc = "overlapped tag attempted";
+ break;
+ case MPI2_SCSITASKMGMT_RSP_IO_QUEUED_ON_IOC:
+ desc = "task queued, however not sent to target";
+ break;
+ default:
+ desc = "unknown";
+ break;
+ }
+ dev_dbg(&instance->pdev->dev, "response_code(%01x): %s\n",
+ mpi_reply->ResponseCode, desc);
+ dev_dbg(&instance->pdev->dev,
+ "TerminationCount/DevHandle/Function/TaskType/IOCStat/IOCLoginfo"
+ " 0x%x/0x%x/0x%x/0x%x/0x%x/0x%x\n",
+ mpi_reply->TerminationCount, mpi_reply->DevHandle,
+ mpi_reply->Function, mpi_reply->TaskType,
+ mpi_reply->IOCStatus, mpi_reply->IOCLogInfo);
+}
+
+/**
+ * megasas_issue_tm - main routine for sending tm requests
+ * @instance: per adapter struct
+ * @device_handle: device handle
+ * @channel: the channel assigned by the OS
+ * @id: the id assigned by the OS
+ * @type: MPI2_SCSITASKMGMT_TASKTYPE__XXX (defined in megaraid_sas_fusion.c)
+ * @smid_task: smid assigned to the task
+ * @m_type: TM_MUTEX_ON or TM_MUTEX_OFF
+ * Context: user
+ *
+ * MegaRaid use MPT interface for Task Magement request.
+ * A generic API for sending task management requests to firmware.
+ *
+ * Return SUCCESS or FAILED.
+ */
+static int
+megasas_issue_tm(struct megasas_instance *instance, u16 device_handle,
+ uint channel, uint id, u16 smid_task, u8 type)
+{
+ struct MR_TASK_MANAGE_REQUEST *mr_request;
+ struct MPI2_SCSI_TASK_MANAGE_REQUEST *mpi_request;
+ unsigned long timeleft;
+ struct megasas_cmd_fusion *cmd_fusion;
+ struct megasas_cmd *cmd_mfi;
+ union MEGASAS_REQUEST_DESCRIPTOR_UNION *req_desc;
+ struct fusion_context *fusion;
+ struct megasas_cmd_fusion *scsi_lookup;
+ int rc;
+ struct MPI2_SCSI_TASK_MANAGE_REPLY *mpi_reply;
+
+ fusion = instance->ctrl_context;
+
+ cmd_mfi = megasas_get_cmd(instance);
+
+ if (!cmd_mfi) {
+ dev_err(&instance->pdev->dev, "Failed from %s %d\n",
+ __func__, __LINE__);
+ return -ENOMEM;
+ }
+
+ cmd_fusion = megasas_get_cmd_fusion(instance,
+ instance->max_scsi_cmds + cmd_mfi->index);
+
+ /* Save the smid. To be used for returning the cmd */
+ cmd_mfi->context.smid = cmd_fusion->index;
+
+ req_desc = megasas_get_request_descriptor(instance,
+ (cmd_fusion->index - 1));
+ if (!req_desc) {
+ dev_err(&instance->pdev->dev, "Failed from %s %d\n",
+ __func__, __LINE__);
+ megasas_return_cmd(instance, cmd_mfi);
+ return -ENOMEM;
+ }
+
+ cmd_fusion->request_desc = req_desc;
+ req_desc->Words = 0;
+
+ scsi_lookup = fusion->cmd_list[smid_task - 1];
+
+ mr_request = (struct MR_TASK_MANAGE_REQUEST *) cmd_fusion->io_request;
+ memset(mr_request, 0, sizeof(struct MR_TASK_MANAGE_REQUEST));
+ mpi_request = (struct MPI2_SCSI_TASK_MANAGE_REQUEST *) &mr_request->TmRequest;
+ mpi_request->Function = MPI2_FUNCTION_SCSI_TASK_MGMT;
+ mpi_request->DevHandle = cpu_to_le16(device_handle);
+ mpi_request->TaskType = type;
+ mpi_request->TaskMID = cpu_to_le16(smid_task);
+ mpi_request->LUN[1] = 0;
+
+
+ req_desc = cmd_fusion->request_desc;
+ req_desc->HighPriority.SMID = cpu_to_le16(cmd_fusion->index);
+ req_desc->HighPriority.RequestFlags =
+ (MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY <<
+ MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
+ req_desc->HighPriority.MSIxIndex = 0;
+ req_desc->HighPriority.LMID = 0;
+ req_desc->HighPriority.Reserved1 = 0;
+
+ if (channel < MEGASAS_MAX_PD_CHANNELS)
+ mr_request->tmReqFlags.isTMForPD = 1;
+ else
+ mr_request->tmReqFlags.isTMForLD = 1;
+
+ init_completion(&cmd_fusion->done);
+ megasas_fire_cmd_fusion(instance, req_desc);
+
+ timeleft = wait_for_completion_timeout(&cmd_fusion->done, 50 * HZ);
+
+ if (!timeleft) {
+ dev_err(&instance->pdev->dev,
+ "task mgmt type 0x%x timed out\n", type);
+ cmd_mfi->flags |= DRV_DCMD_SKIP_REFIRE;
+ mutex_unlock(&instance->reset_mutex);
+ rc = megasas_reset_fusion(instance->host, MFI_IO_TIMEOUT_OCR);
+ mutex_lock(&instance->reset_mutex);
+ return rc;
+ }
+
+ mpi_reply = (struct MPI2_SCSI_TASK_MANAGE_REPLY *) &mr_request->TMReply;
+ megasas_tm_response_code(instance, mpi_reply);
+
+ megasas_return_cmd(instance, cmd_mfi);
+ rc = SUCCESS;
+ switch (type) {
+ case MPI2_SCSITASKMGMT_TASKTYPE_ABORT_TASK:
+ if (scsi_lookup->scmd == NULL)
+ break;
+ else {
+ instance->instancet->disable_intr(instance);
+ msleep(1000);
+ megasas_complete_cmd_dpc_fusion
+ ((unsigned long)instance);
+ instance->instancet->enable_intr(instance);
+ if (scsi_lookup->scmd == NULL)
+ break;
+ }
+ rc = FAILED;
+ break;
+
+ case MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET:
+ if ((channel == 0xFFFFFFFF) && (id == 0xFFFFFFFF))
+ break;
+ instance->instancet->disable_intr(instance);
+ msleep(1000);
+ megasas_complete_cmd_dpc_fusion
+ ((unsigned long)instance);
+ rc = megasas_track_scsiio(instance, id, channel);
+ instance->instancet->enable_intr(instance);
+
+ break;
+ case MPI2_SCSITASKMGMT_TASKTYPE_ABRT_TASK_SET:
+ case MPI2_SCSITASKMGMT_TASKTYPE_QUERY_TASK:
+ break;
+ default:
+ rc = FAILED;
+ break;
+ }
+
+ return rc;
+
+}
+
+/*
+ * megasas_fusion_smid_lookup : Look for fusion command correpspodning to SCSI
+ * @instance: per adapter struct
+ *
+ * Return Non Zero index, if SMID found in outstanding commands
+ */
+static u16 megasas_fusion_smid_lookup(struct scsi_cmnd *scmd)
+{
+ int i, ret = 0;
+ struct megasas_instance *instance;
+ struct megasas_cmd_fusion *cmd_fusion;
+ struct fusion_context *fusion;
+
+ instance = (struct megasas_instance *)scmd->device->host->hostdata;
+
+ fusion = instance->ctrl_context;
+
+ for (i = 0; i < instance->max_scsi_cmds; i++) {
+ cmd_fusion = fusion->cmd_list[i];
+ if (cmd_fusion->scmd && (cmd_fusion->scmd == scmd)) {
+ scmd_printk(KERN_NOTICE, scmd, "Abort request is for"
+ " SMID: %d\n", cmd_fusion->index);
+ ret = cmd_fusion->index;
+ break;
+ }
+ }
+
+ return ret;
+}
+
+/*
+* megasas_get_tm_devhandle - Get devhandle for TM request
+* @sdev- OS provided scsi device
+*
+* Returns- devhandle/targetID of SCSI device
+*/
+static u16 megasas_get_tm_devhandle(struct scsi_device *sdev)
+{
+ u16 pd_index = 0;
+ u32 device_id;
+ struct megasas_instance *instance;
+ struct fusion_context *fusion;
+ struct MR_PD_CFG_SEQ_NUM_SYNC *pd_sync;
+ u16 devhandle = (u16)ULONG_MAX;
+
+ instance = (struct megasas_instance *)sdev->host->hostdata;
+ fusion = instance->ctrl_context;
+
+ if (sdev->channel < MEGASAS_MAX_PD_CHANNELS) {
+ if (instance->use_seqnum_jbod_fp) {
+ pd_index = (sdev->channel * MEGASAS_MAX_DEV_PER_CHANNEL) +
+ sdev->id;
+ pd_sync = (void *)fusion->pd_seq_sync
+ [(instance->pd_seq_map_id - 1) & 1];
+ devhandle = pd_sync->seq[pd_index].devHandle;
+ } else
+ sdev_printk(KERN_ERR, sdev, "Firmware expose tmCapable"
+ " without JBOD MAP support from %s %d\n", __func__, __LINE__);
+ } else {
+ device_id = ((sdev->channel % 2) * MEGASAS_MAX_DEV_PER_CHANNEL)
+ + sdev->id;
+ devhandle = device_id;
+ }
+
+ return devhandle;
+}
+
+/*
+ * megasas_task_abort_fusion : SCSI task abort function for fusion adapters
+ * @scmd : pointer to scsi command object
+ *
+ * Return SUCCESS, if command aborted else FAILED
+ */
+
+int megasas_task_abort_fusion(struct scsi_cmnd *scmd)
+{
+ struct megasas_instance *instance;
+ u16 smid, devhandle;
+ struct fusion_context *fusion;
+ int ret;
+ struct MR_PRIV_DEVICE *mr_device_priv_data;
+ mr_device_priv_data = scmd->device->hostdata;
+
+
+ instance = (struct megasas_instance *)scmd->device->host->hostdata;
+ fusion = instance->ctrl_context;
+
+ if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ dev_err(&instance->pdev->dev, "Controller is not OPERATIONAL,"
+ "SCSI host:%d\n", instance->host->host_no);
+ ret = FAILED;
+ return ret;
+ }
+
+ if (!mr_device_priv_data) {
+ sdev_printk(KERN_INFO, scmd->device, "device been deleted! "
+ "scmd(%p)\n", scmd);
+ scmd->result = DID_NO_CONNECT << 16;
+ ret = SUCCESS;
+ goto out;
+ }
+
+
+ if (!mr_device_priv_data->is_tm_capable) {
+ ret = FAILED;
+ goto out;
+ }
+
+ mutex_lock(&instance->reset_mutex);
+
+ smid = megasas_fusion_smid_lookup(scmd);
+
+ if (!smid) {
+ ret = SUCCESS;
+ scmd_printk(KERN_NOTICE, scmd, "Command for which abort is"
+ " issued is not found in oustanding commands\n");
+ mutex_unlock(&instance->reset_mutex);
+ goto out;
+ }
+
+ devhandle = megasas_get_tm_devhandle(scmd->device);
+
+ if (devhandle == (u16)ULONG_MAX) {
+ ret = SUCCESS;
+ sdev_printk(KERN_INFO, scmd->device,
+ "task abort issued for invalid devhandle\n");
+ mutex_unlock(&instance->reset_mutex);
+ goto out;
+ }
+ sdev_printk(KERN_INFO, scmd->device,
+ "attempting task abort! scmd(%p) tm_dev_handle 0x%x\n",
+ scmd, devhandle);
+
+ mr_device_priv_data->tm_busy = 1;
+ ret = megasas_issue_tm(instance, devhandle,
+ scmd->device->channel, scmd->device->id, smid,
+ MPI2_SCSITASKMGMT_TASKTYPE_ABORT_TASK);
+ mr_device_priv_data->tm_busy = 0;
+
+ mutex_unlock(&instance->reset_mutex);
+out:
+ sdev_printk(KERN_INFO, scmd->device, "task abort: %s scmd(%p)\n",
+ ((ret == SUCCESS) ? "SUCCESS" : "FAILED"), scmd);
+
+ return ret;
+}
+
+/*
+ * megasas_reset_target_fusion : target reset function for fusion adapters
+ * scmd: SCSI command pointer
+ *
+ * Returns SUCCESS if all commands associated with target aborted else FAILED
+ */
+
+int megasas_reset_target_fusion(struct scsi_cmnd *scmd)
+{
+
+ struct megasas_instance *instance;
+ int ret = FAILED;
+ u16 devhandle;
+ struct fusion_context *fusion;
+ struct MR_PRIV_DEVICE *mr_device_priv_data;
+ mr_device_priv_data = scmd->device->hostdata;
+
+ instance = (struct megasas_instance *)scmd->device->host->hostdata;
+ fusion = instance->ctrl_context;
+
+ if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ dev_err(&instance->pdev->dev, "Controller is not OPERATIONAL,"
+ "SCSI host:%d\n", instance->host->host_no);
+ ret = FAILED;
+ return ret;
+ }
+
+ if (!mr_device_priv_data) {
+ sdev_printk(KERN_INFO, scmd->device, "device been deleted! "
+ "scmd(%p)\n", scmd);
+ scmd->result = DID_NO_CONNECT << 16;
+ ret = SUCCESS;
+ goto out;
+ }
+
+
+ if (!mr_device_priv_data->is_tm_capable) {
+ ret = FAILED;
+ goto out;
+ }
+
+ mutex_lock(&instance->reset_mutex);
+ devhandle = megasas_get_tm_devhandle(scmd->device);
+
+ if (devhandle == (u16)ULONG_MAX) {
+ ret = SUCCESS;
+ sdev_printk(KERN_INFO, scmd->device,
+ "target reset issued for invalid devhandle\n");
+ mutex_unlock(&instance->reset_mutex);
+ goto out;
+ }
+
+ sdev_printk(KERN_INFO, scmd->device,
+ "attempting target reset! scmd(%p) tm_dev_handle 0x%x\n",
+ scmd, devhandle);
+ mr_device_priv_data->tm_busy = 1;
+ ret = megasas_issue_tm(instance, devhandle,
+ scmd->device->channel, scmd->device->id, 0,
+ MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET);
+ mr_device_priv_data->tm_busy = 0;
+ mutex_unlock(&instance->reset_mutex);
+out:
+ scmd_printk(KERN_NOTICE, scmd, "megasas: target reset %s!!\n",
+ (ret == SUCCESS) ? "SUCCESS" : "FAILED");
+
+ return ret;
+}
+
/* Check for a second path that is currently UP */
int megasas_check_mpio_paths(struct megasas_instance *instance,
struct scsi_cmnd *scmd)
@@ -2752,7 +3210,7 @@ out:
}
/* Core fusion reset function */
-int megasas_reset_fusion(struct Scsi_Host *shost, int iotimeout)
+int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
{
int retval = SUCCESS, i, convert = 0;
struct megasas_instance *instance;
@@ -2761,6 +3219,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int iotimeout)
u32 abs_state, status_reg, reset_adapter;
u32 io_timeout_in_crash_mode = 0;
struct scsi_cmnd *scmd_local = NULL;
+ struct scsi_device *sdev;
instance = (struct megasas_instance *)shost->hostdata;
fusion = instance->ctrl_context;
@@ -2779,8 +3238,8 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int iotimeout)
/* IO timeout detected, forcibly put FW in FAULT state */
if (abs_state != MFI_STATE_FAULT && instance->crash_dump_buf &&
- instance->crash_dump_app_support && iotimeout) {
- dev_info(&instance->pdev->dev, "IO timeout is detected, "
+ instance->crash_dump_app_support && reason) {
+ dev_info(&instance->pdev->dev, "IO/DCMD timeout is detected, "
"forcibly FAULT Firmware\n");
instance->adprecovery = MEGASAS_ADPRESET_SM_INFAULT;
status_reg = readl(&instance->reg_set->doorbell);
@@ -2819,13 +3278,13 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int iotimeout)
msleep(1000);
/* First try waiting for commands to complete */
- if (megasas_wait_for_outstanding_fusion(instance, iotimeout,
+ if (megasas_wait_for_outstanding_fusion(instance, reason,
&convert)) {
instance->adprecovery = MEGASAS_ADPRESET_SM_INFAULT;
dev_warn(&instance->pdev->dev, "resetting fusion "
"adapter scsi%d.\n", instance->host->host_no);
if (convert)
- iotimeout = 0;
+ reason = 0;
/* Now return commands back to the OS */
for (i = 0 ; i < instance->max_scsi_cmds; i++) {
@@ -2859,7 +3318,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int iotimeout)
}
/* Let SR-IOV VF & PF sync up if there was a HB failure */
- if (instance->requestorId && !iotimeout) {
+ if (instance->requestorId && !reason) {
msleep(MEGASAS_OCR_SETTLE_TIME_VF);
/* Look for a late HB update after VF settle time */
if (abs_state == MFI_STATE_OPERATIONAL &&
@@ -2954,6 +3413,9 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int iotimeout)
megasas_setup_jbod_map(instance);
+ shost_for_each_device(sdev, shost)
+ megasas_update_sdev_properties(sdev);
+
clear_bit(MEGASAS_FUSION_IN_RESET,
&instance->reset_flags);
instance->instancet->enable_intr(instance);
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.h b/drivers/scsi/megaraid/megaraid_sas_fusion.h
index a9e10c4..a1f1c0b 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.h
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.h
@@ -176,6 +176,7 @@ enum REGION_TYPE {
#define MPI2_SCSIIO_EEDPFLAGS_CHECK_GUARD (0x0100)
#define MPI2_SCSIIO_EEDPFLAGS_INSERT_OP (0x0004)
#define MPI2_FUNCTION_SCSI_IO_REQUEST (0x00) /* SCSI IO */
+#define MPI2_FUNCTION_SCSI_TASK_MGMT (0x01)
#define MPI2_REQ_DESCRIPT_FLAGS_HIGH_PRIORITY (0x03)
#define MPI2_REQ_DESCRIPT_FLAGS_FP_IO (0x06)
#define MPI2_REQ_DESCRIPT_FLAGS_SCSI_IO (0x00)
@@ -278,6 +279,100 @@ union MPI2_SCSI_IO_CDB_UNION {
struct MPI2_SGE_SIMPLE_UNION SGE;
};
+/****************************************************************************
+* SCSI Task Management messages
+****************************************************************************/
+
+/*SCSI Task Management Request Message */
+struct MPI2_SCSI_TASK_MANAGE_REQUEST {
+ u16 DevHandle; /*0x00 */
+ u8 ChainOffset; /*0x02 */
+ u8 Function; /*0x03 */
+ u8 Reserved1; /*0x04 */
+ u8 TaskType; /*0x05 */
+ u8 Reserved2; /*0x06 */
+ u8 MsgFlags; /*0x07 */
+ u8 VP_ID; /*0x08 */
+ u8 VF_ID; /*0x09 */
+ u16 Reserved3; /*0x0A */
+ u8 LUN[8]; /*0x0C */
+ u32 Reserved4[7]; /*0x14 */
+ u16 TaskMID; /*0x30 */
+ u16 Reserved5; /*0x32 */
+};
+
+
+/*SCSI Task Management Reply Message */
+struct MPI2_SCSI_TASK_MANAGE_REPLY {
+ u16 DevHandle; /*0x00 */
+ u8 MsgLength; /*0x02 */
+ u8 Function; /*0x03 */
+ u8 ResponseCode; /*0x04 */
+ u8 TaskType; /*0x05 */
+ u8 Reserved1; /*0x06 */
+ u8 MsgFlags; /*0x07 */
+ u8 VP_ID; /*0x08 */
+ u8 VF_ID; /*0x09 */
+ u16 Reserved2; /*0x0A */
+ u16 Reserved3; /*0x0C */
+ u16 IOCStatus; /*0x0E */
+ u32 IOCLogInfo; /*0x10 */
+ u32 TerminationCount; /*0x14 */
+ u32 ResponseInfo; /*0x18 */
+};
+
+struct MR_TM_REQUEST {
+ char request[128];
+};
+
+struct MR_TM_REPLY {
+ char reply[128];
+};
+
+/* SCSI Task Management Request Message */
+struct MR_TASK_MANAGE_REQUEST {
+ /*To be type casted to struct MPI2_SCSI_TASK_MANAGE_REQUEST */
+ struct MR_TM_REQUEST TmRequest;
+ union {
+ struct {
+#if defined(__BIG_ENDIAN_BITFIELD)
+ u32 reserved1:30;
+ u32 isTMForPD:1;
+ u32 isTMForLD:1;
+#else
+ u32 isTMForLD:1;
+ u32 isTMForPD:1;
+ u32 reserved1:30;
+#endif
+ u32 reserved2;
+ } tmReqFlags;
+ struct MR_TM_REPLY TMReply;
+ };
+};
+
+/* TaskType values */
+
+#define MPI2_SCSITASKMGMT_TASKTYPE_ABORT_TASK (0x01)
+#define MPI2_SCSITASKMGMT_TASKTYPE_ABRT_TASK_SET (0x02)
+#define MPI2_SCSITASKMGMT_TASKTYPE_TARGET_RESET (0x03)
+#define MPI2_SCSITASKMGMT_TASKTYPE_LOGICAL_UNIT_RESET (0x05)
+#define MPI2_SCSITASKMGMT_TASKTYPE_CLEAR_TASK_SET (0x06)
+#define MPI2_SCSITASKMGMT_TASKTYPE_QUERY_TASK (0x07)
+#define MPI2_SCSITASKMGMT_TASKTYPE_CLR_ACA (0x08)
+#define MPI2_SCSITASKMGMT_TASKTYPE_QRY_TASK_SET (0x09)
+#define MPI2_SCSITASKMGMT_TASKTYPE_QRY_ASYNC_EVENT (0x0A)
+
+/* ResponseCode values */
+
+#define MPI2_SCSITASKMGMT_RSP_TM_COMPLETE (0x00)
+#define MPI2_SCSITASKMGMT_RSP_INVALID_FRAME (0x02)
+#define MPI2_SCSITASKMGMT_RSP_TM_NOT_SUPPORTED (0x04)
+#define MPI2_SCSITASKMGMT_RSP_TM_FAILED (0x05)
+#define MPI2_SCSITASKMGMT_RSP_TM_SUCCEEDED (0x08)
+#define MPI2_SCSITASKMGMT_RSP_TM_INVALID_LUN (0x09)
+#define MPI2_SCSITASKMGMT_RSP_TM_OVERLAPPED_TAG (0x0A)
+#define MPI2_SCSITASKMGMT_RSP_IO_QUEUED_ON_IOC (0x80)
+
/*
* RAID SCSI IO Request Message
* Total SGE count will be one less than _MPI2_SCSI_IO_REQUEST
@@ -548,7 +643,8 @@ struct MR_SPAN_BLOCK_INFO {
struct MR_LD_RAID {
struct {
#if defined(__BIG_ENDIAN_BITFIELD)
- u32 reserved4:7;
+ u32 reserved4:6;
+ u32 tmCapable:1;
u32 fpNonRWCapable:1;
u32 fpReadAcrossStripe:1;
u32 fpWriteAcrossStripe:1;
@@ -570,7 +666,8 @@ struct MR_LD_RAID {
u32 fpWriteAcrossStripe:1;
u32 fpReadAcrossStripe:1;
u32 fpNonRWCapable:1;
- u32 reserved4:7;
+ u32 tmCapable:1;
+ u32 reserved4:6;
#endif
} capability;
__le32 reserved6;
@@ -695,6 +792,7 @@ struct megasas_cmd_fusion {
u32 sync_cmd_idx;
u32 index;
u8 pd_r1_lb;
+ struct completion done;
};
struct LD_LOAD_BALANCE_INFO {
@@ -808,9 +906,18 @@ struct MR_FW_RAID_MAP_EXT {
* * define MR_PD_CFG_SEQ structure for system PDs
* */
struct MR_PD_CFG_SEQ {
- __le16 seqNum;
- __le16 devHandle;
- u8 reserved[4];
+ u16 seqNum;
+ u16 devHandle;
+ struct {
+#if defined(__BIG_ENDIAN_BITFIELD)
+ u8 reserved:7;
+ u8 tmCapable:1;
+#else
+ u8 tmCapable:1;
+ u8 reserved:7;
+#endif
+ } capability;
+ u8 reserved[3];
} __packed;
struct MR_PD_CFG_SEQ_NUM_SYNC {
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 05/15] megaraid_sas: Update device Queue depth based on interface type
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (3 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 04/15] megaraid_sas: Task management support Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 06/15] megaraid_sas: Fastpath region lock bypass Sumit Saxena
` (10 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
This patch will update device Queue depth based on interface type(SAS, SATA..) for sysPDs.
For Virtual disks(VDs), there will be no change in queue depth(will remain 256).
To fetch interface type(SAS or SATA or FC..) of syspD, driver will send DCMD MR_DCMD_PD_GET_INFO per sysPD.
There are no changes in this patch from last time sent patch. There was some feedback to use dma_alloc_coherent
instead of pci_alloc_consistent which will be handled in follow up along with several other cases where
dma_alloc_coherent can be used.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 270 +++++++++++++++++++++++++++++-
drivers/scsi/megaraid/megaraid_sas_base.c | 127 ++++++++++++++
2 files changed, 396 insertions(+), 1 deletion(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index 0fcb156..773fc54 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -215,6 +215,7 @@
#define MR_DCMD_CTRL_SET_CRASH_DUMP_PARAMS 0x01190100
#define MR_DRIVER_SET_APP_CRASHDUMP_MODE (0xF0010000 | 0x0600)
+#define MR_DCMD_PD_GET_INFO 0x02020000
/*
* Global functions
@@ -435,6 +436,257 @@ enum MR_PD_STATE {
MR_PD_STATE_SYSTEM = 0x40,
};
+union MR_PD_REF {
+ struct {
+ u16 deviceId;
+ u16 seqNum;
+ } mrPdRef;
+ u32 ref;
+};
+
+/*
+ * define the DDF Type bit structure
+ */
+union MR_PD_DDF_TYPE {
+ struct {
+ union {
+ struct {
+#ifndef __BIG_ENDIAN_BITFIELD
+ u16 forcedPDGUID:1;
+ u16 inVD:1;
+ u16 isGlobalSpare:1;
+ u16 isSpare:1;
+ u16 isForeign:1;
+ u16 reserved:7;
+ u16 intf:4;
+#else
+ u16 intf:4;
+ u16 reserved:7;
+ u16 isForeign:1;
+ u16 isSpare:1;
+ u16 isGlobalSpare:1;
+ u16 inVD:1;
+ u16 forcedPDGUID:1;
+#endif
+ } pdType;
+ u16 type;
+ };
+ u16 reserved;
+ } ddf;
+ struct {
+ u32 reserved;
+ } nonDisk;
+ u32 type;
+} __packed;
+
+/*
+ * defines the progress structure
+ */
+union MR_PROGRESS {
+ struct {
+ u16 progress;
+ union {
+ u16 elapsedSecs;
+ u16 elapsedSecsForLastPercent;
+ };
+ } mrProgress;
+ u32 w;
+} __packed;
+
+/*
+ * defines the physical drive progress structure
+ */
+struct MR_PD_PROGRESS {
+ struct {
+#ifndef MFI_BIG_ENDIAN
+ u32 rbld:1;
+ u32 patrol:1;
+ u32 clear:1;
+ u32 copyBack:1;
+ u32 erase:1;
+ u32 locate:1;
+ u32 reserved:26;
+#else
+ u32 reserved:26;
+ u32 locate:1;
+ u32 erase:1;
+ u32 copyBack:1;
+ u32 clear:1;
+ u32 patrol:1;
+ u32 rbld:1;
+#endif
+ } active;
+ union MR_PROGRESS rbld;
+ union MR_PROGRESS patrol;
+ union {
+ union MR_PROGRESS clear;
+ union MR_PROGRESS erase;
+ };
+
+ struct {
+#ifndef MFI_BIG_ENDIAN
+ u32 rbld:1;
+ u32 patrol:1;
+ u32 clear:1;
+ u32 copyBack:1;
+ u32 erase:1;
+ u32 reserved:27;
+#else
+ u32 reserved:27;
+ u32 erase:1;
+ u32 copyBack:1;
+ u32 clear:1;
+ u32 patrol:1;
+ u32 rbld:1;
+#endif
+ } pause;
+
+ union MR_PROGRESS reserved[3];
+} __packed;
+
+struct MR_PD_INFO {
+ union MR_PD_REF ref;
+ u8 inquiryData[96];
+ u8 vpdPage83[64];
+ u8 notSupported;
+ u8 scsiDevType;
+
+ union {
+ u8 connectedPortBitmap;
+ u8 connectedPortNumbers;
+ };
+
+ u8 deviceSpeed;
+ u32 mediaErrCount;
+ u32 otherErrCount;
+ u32 predFailCount;
+ u32 lastPredFailEventSeqNum;
+
+ u16 fwState;
+ u8 disabledForRemoval;
+ u8 linkSpeed;
+ union MR_PD_DDF_TYPE state;
+
+ struct {
+ u8 count;
+#ifndef __BIG_ENDIAN_BITFIELD
+ u8 isPathBroken:4;
+ u8 reserved3:3;
+ u8 widePortCapable:1;
+#else
+ u8 widePortCapable:1;
+ u8 reserved3:3;
+ u8 isPathBroken:4;
+#endif
+
+ u8 connectorIndex[2];
+ u8 reserved[4];
+ u64 sasAddr[2];
+ u8 reserved2[16];
+ } pathInfo;
+
+ u64 rawSize;
+ u64 nonCoercedSize;
+ u64 coercedSize;
+ u16 enclDeviceId;
+ u8 enclIndex;
+
+ union {
+ u8 slotNumber;
+ u8 enclConnectorIndex;
+ };
+
+ struct MR_PD_PROGRESS progInfo;
+ u8 badBlockTableFull;
+ u8 unusableInCurrentConfig;
+ u8 vpdPage83Ext[64];
+ u8 powerState;
+ u8 enclPosition;
+ u32 allowedOps;
+ u16 copyBackPartnerId;
+ u16 enclPartnerDeviceId;
+ struct {
+#ifndef __BIG_ENDIAN_BITFIELD
+ u16 fdeCapable:1;
+ u16 fdeEnabled:1;
+ u16 secured:1;
+ u16 locked:1;
+ u16 foreign:1;
+ u16 needsEKM:1;
+ u16 reserved:10;
+#else
+ u16 reserved:10;
+ u16 needsEKM:1;
+ u16 foreign:1;
+ u16 locked:1;
+ u16 secured:1;
+ u16 fdeEnabled:1;
+ u16 fdeCapable:1;
+#endif
+ } security;
+ u8 mediaType;
+ u8 notCertified;
+ u8 bridgeVendor[8];
+ u8 bridgeProductIdentification[16];
+ u8 bridgeProductRevisionLevel[4];
+ u8 satBridgeExists;
+
+ u8 interfaceType;
+ u8 temperature;
+ u8 emulatedBlockSize;
+ u16 userDataBlockSize;
+ u16 reserved2;
+
+ struct {
+#ifndef __BIG_ENDIAN_BITFIELD
+ u32 piType:3;
+ u32 piFormatted:1;
+ u32 piEligible:1;
+ u32 NCQ:1;
+ u32 WCE:1;
+ u32 commissionedSpare:1;
+ u32 emergencySpare:1;
+ u32 ineligibleForSSCD:1;
+ u32 ineligibleForLd:1;
+ u32 useSSEraseType:1;
+ u32 wceUnchanged:1;
+ u32 supportScsiUnmap:1;
+ u32 reserved:18;
+#else
+ u32 reserved:18;
+ u32 supportScsiUnmap:1;
+ u32 wceUnchanged:1;
+ u32 useSSEraseType:1;
+ u32 ineligibleForLd:1;
+ u32 ineligibleForSSCD:1;
+ u32 emergencySpare:1;
+ u32 commissionedSpare:1;
+ u32 WCE:1;
+ u32 NCQ:1;
+ u32 piEligible:1;
+ u32 piFormatted:1;
+ u32 piType:3;
+#endif
+ } properties;
+
+ u64 shieldDiagCompletionTime;
+ u8 shieldCounter;
+
+ u8 linkSpeedOther;
+ u8 reserved4[2];
+
+ struct {
+#ifndef __BIG_ENDIAN_BITFIELD
+ u32 bbmErrCountSupported:1;
+ u32 bbmErrCount:31;
+#else
+ u32 bbmErrCount:31;
+ u32 bbmErrCountSupported:1;
+#endif
+ } bbmErr;
+
+ u8 reserved1[512-428];
+} __packed;
/*
* defines the physical drive address structure
@@ -474,6 +726,7 @@ struct megasas_pd_list {
u16 tid;
u8 driveType;
u8 driveState;
+ u8 interface;
} __packed;
/*
@@ -1718,6 +1971,19 @@ struct MR_DRV_SYSTEM_INFO {
u8 reserved[1980];
};
+enum MR_PD_TYPE {
+ UNKNOWN_DRIVE = 0,
+ PARALLEL_SCSI = 1,
+ SAS_PD = 2,
+ SATA_PD = 3,
+ FC_PD = 4,
+};
+
+/* JBOD Queue depth definitions */
+#define MEGASAS_SATA_QD 32
+#define MEGASAS_SAS_QD 64
+#define MEGASAS_DEFAULT_PD_QD 64
+
struct megasas_instance {
__le32 *producer;
@@ -1732,6 +1998,8 @@ struct megasas_instance {
dma_addr_t vf_affiliation_111_h;
struct MR_CTRL_HB_HOST_MEM *hb_host_mem;
dma_addr_t hb_host_mem_h;
+ struct MR_PD_INFO *pd_info;
+ dma_addr_t pd_info_h;
__le32 *reply_queue;
dma_addr_t reply_queue_h;
@@ -1780,7 +2048,7 @@ struct megasas_instance {
struct megasas_evt_detail *evt_detail;
dma_addr_t evt_detail_h;
struct megasas_cmd *aen_cmd;
- struct mutex aen_mutex;
+ struct mutex hba_mutex;
struct semaphore ioctl_sem;
struct Scsi_Host *host;
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 57cf4e3..ea3994b 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -104,6 +104,8 @@ static int megasas_ld_list_query(struct megasas_instance *instance,
static int megasas_issue_init_mfi(struct megasas_instance *instance);
static int megasas_register_aen(struct megasas_instance *instance,
u32 seq_num, u32 class_locale_word);
+static int
+megasas_get_pd_info(struct megasas_instance *instance, u16 device_id);
/*
* PCI ID table for all supported controllers
*/
@@ -1795,6 +1797,44 @@ void megasas_update_sdev_properties(struct scsi_device *sdev)
}
}
+static void megasas_set_device_queue_depth(struct scsi_device *sdev)
+{
+ u16 pd_index = 0;
+ int ret = DCMD_FAILED;
+ struct megasas_instance *instance;
+
+ instance = megasas_lookup_instance(sdev->host->host_no);
+
+ if (sdev->channel < MEGASAS_MAX_PD_CHANNELS) {
+ pd_index = (sdev->channel * MEGASAS_MAX_DEV_PER_CHANNEL) + sdev->id;
+
+ if (instance->pd_info) {
+ mutex_lock(&instance->hba_mutex);
+ ret = megasas_get_pd_info(instance, pd_index);
+ mutex_unlock(&instance->hba_mutex);
+ }
+
+ if (ret != DCMD_SUCCESS)
+ return;
+
+ if (instance->pd_list[pd_index].driveState == MR_PD_STATE_SYSTEM) {
+
+ switch (instance->pd_list[pd_index].interface) {
+ case SAS_PD:
+ scsi_change_queue_depth(sdev, MEGASAS_SAS_QD);
+ break;
+
+ case SATA_PD:
+ scsi_change_queue_depth(sdev, MEGASAS_SATA_QD);
+ break;
+
+ default:
+ scsi_change_queue_depth(sdev, MEGASAS_DEFAULT_PD_QD);
+ }
+ }
+ }
+}
+
static int megasas_slave_configure(struct scsi_device *sdev)
{
@@ -1812,6 +1852,7 @@ static int megasas_slave_configure(struct scsi_device *sdev)
return -ENXIO;
}
}
+ megasas_set_device_queue_depth(sdev);
megasas_update_sdev_properties(sdev);
/*
@@ -3921,6 +3962,73 @@ dcmd_timeout_ocr_possible(struct megasas_instance *instance) {
return INITIATE_OCR;
}
+static int
+megasas_get_pd_info(struct megasas_instance *instance, u16 device_id)
+{
+ int ret;
+ struct megasas_cmd *cmd;
+ struct megasas_dcmd_frame *dcmd;
+
+ cmd = megasas_get_cmd(instance);
+
+ if (!cmd) {
+ dev_err(&instance->pdev->dev, "Failed to get cmd %s\n", __func__);
+ return -ENOMEM;
+ }
+
+ dcmd = &cmd->frame->dcmd;
+
+ memset(instance->pd_info, 0, sizeof(*instance->pd_info));
+ memset(dcmd->mbox.b, 0, MFI_MBOX_SIZE);
+
+ dcmd->mbox.s[0] = cpu_to_le16(device_id);
+ dcmd->cmd = MFI_CMD_DCMD;
+ dcmd->cmd_status = 0xFF;
+ dcmd->sge_count = 1;
+ dcmd->flags = cpu_to_le16(MFI_FRAME_DIR_READ);
+ dcmd->timeout = 0;
+ dcmd->pad_0 = 0;
+ dcmd->data_xfer_len = cpu_to_le32(sizeof(struct MR_PD_INFO));
+ dcmd->opcode = cpu_to_le32(MR_DCMD_PD_GET_INFO);
+ dcmd->sgl.sge32[0].phys_addr = cpu_to_le32(instance->pd_info_h);
+ dcmd->sgl.sge32[0].length = cpu_to_le32(sizeof(struct MR_PD_INFO));
+
+ if (instance->ctrl_context && !instance->mask_interrupts)
+ ret = megasas_issue_blocked_cmd(instance, cmd, MFI_IO_TIMEOUT_SECS);
+ else
+ ret = megasas_issue_polled(instance, cmd);
+
+ switch (ret) {
+ case DCMD_SUCCESS:
+ instance->pd_list[device_id].interface =
+ instance->pd_info->state.ddf.pdType.intf;
+ break;
+
+ case DCMD_TIMEOUT:
+
+ switch (dcmd_timeout_ocr_possible(instance)) {
+ case INITIATE_OCR:
+ cmd->flags |= DRV_DCMD_SKIP_REFIRE;
+ megasas_reset_fusion(instance->host,
+ MFI_IO_TIMEOUT_OCR);
+ break;
+ case KILL_ADAPTER:
+ megaraid_sas_kill_hba(instance);
+ break;
+ case IGNORE_TIMEOUT:
+ dev_info(&instance->pdev->dev, "Ignore DCMD timeout: %s %d\n",
+ __func__, __LINE__);
+ break;
+ }
+
+ break;
+ }
+
+ if (ret != DCMD_TIMEOUT)
+ megasas_return_cmd(instance, cmd);
+
+ return ret;
+}
/*
* megasas_get_pd_list_info - Returns FW's pd_list structure
* @instance: Adapter soft state
@@ -5679,6 +5787,12 @@ static int megasas_probe_one(struct pci_dev *pdev,
goto fail_alloc_dma_buf;
}
+ instance->pd_info = pci_alloc_consistent(pdev,
+ sizeof(struct MR_PD_INFO), &instance->pd_info_h);
+
+ if (!instance->pd_info)
+ dev_err(&instance->pdev->dev, "Failed to alloc mem for pd_info\n");
+
/*
* Initialize locks and queues
*/
@@ -5695,6 +5809,7 @@ static int megasas_probe_one(struct pci_dev *pdev,
spin_lock_init(&instance->completion_lock);
mutex_init(&instance->reset_mutex);
+ mutex_init(&instance->hba_mutex);
/*
* Initialize PCI related and misc parameters
@@ -5809,6 +5924,10 @@ fail_alloc_dma_buf:
instance->evt_detail,
instance->evt_detail_h);
+ if (instance->pd_info)
+ pci_free_consistent(pdev, sizeof(struct MR_PD_INFO),
+ instance->pd_info,
+ instance->pd_info_h);
if (instance->producer)
pci_free_consistent(pdev, sizeof(u32), instance->producer,
instance->producer_h);
@@ -6070,6 +6189,10 @@ fail_init_mfi:
instance->evt_detail,
instance->evt_detail_h);
+ if (instance->pd_info)
+ pci_free_consistent(pdev, sizeof(struct MR_PD_INFO),
+ instance->pd_info,
+ instance->pd_info_h);
if (instance->producer)
pci_free_consistent(pdev, sizeof(u32), instance->producer,
instance->producer_h);
@@ -6188,6 +6311,10 @@ static void megasas_detach_one(struct pci_dev *pdev)
pci_free_consistent(pdev, sizeof(struct megasas_evt_detail),
instance->evt_detail, instance->evt_detail_h);
+ if (instance->pd_info)
+ pci_free_consistent(pdev, sizeof(struct MR_PD_INFO),
+ instance->pd_info,
+ instance->pd_info_h);
if (instance->vf_affiliation)
pci_free_consistent(pdev, (MAX_LOGICAL_DRIVES + 1) *
sizeof(struct MR_LD_VF_AFFILIATION),
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 06/15] megaraid_sas: Fastpath region lock bypass
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (4 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 05/15] megaraid_sas: Update device Queue depth based on interface type Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 07/15] megaraid_sas: Reply Descriptor Post Queue(RDPQ) support Sumit Saxena
` (9 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
Firmware will fill per LD data to tell driver that particular LD supports Region lock bypass or not. If yes, then
Driver will send non FP LDIO to region lock bypass FIFO. With this change in driver, firmware will optimize certain
code to improve performance.
There are no changes in this patch from last time sent patch.
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 8 ++++++--
drivers/scsi/megaraid/megaraid_sas_fp.c | 2 ++
drivers/scsi/megaraid/megaraid_sas_fusion.c | 6 ++++--
drivers/scsi/megaraid/megaraid_sas_fusion.h | 8 +++++---
4 files changed, 17 insertions(+), 7 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index 773fc54..01135be 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -1528,7 +1528,9 @@ union megasas_sgl_frame {
typedef union _MFI_CAPABILITIES {
struct {
#if defined(__BIG_ENDIAN_BITFIELD)
- u32 reserved:23;
+ u32 reserved:21;
+ u32 support_fp_rlbypass:1;
+ u32 support_vfid_in_ioframe:1;
u32 support_ext_io_size:1;
u32 support_ext_queue_depth:1;
u32 security_protocol_cmds_fw:1;
@@ -1548,7 +1550,9 @@ typedef union _MFI_CAPABILITIES {
u32 security_protocol_cmds_fw:1;
u32 support_ext_queue_depth:1;
u32 support_ext_io_size:1;
- u32 reserved:23;
+ u32 support_vfid_in_ioframe:1;
+ u32 support_fp_rlbypass:1;
+ u32 reserved:21;
#endif
} mfi_capabilities;
__le32 reg;
diff --git a/drivers/scsi/megaraid/megaraid_sas_fp.c b/drivers/scsi/megaraid/megaraid_sas_fp.c
index 741509b..e413113 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fp.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fp.c
@@ -1020,6 +1020,8 @@ MR_BuildRaidContext(struct megasas_instance *instance,
/* assume this IO needs the full row - we'll adjust if not true */
regSize = stripSize;
+ io_info->do_fp_rlbypass = raid->capability.fpBypassRegionLock;
+
/* Check if we can send this I/O via FastPath */
if (raid->capability.fpCapable) {
if (isRead)
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 4b0c86c..518b488 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -666,6 +666,8 @@ megasas_ioc_init_fusion(struct megasas_instance *instance)
if (instance->max_chain_frame_sz > MEGASAS_CHAIN_FRAME_SZ_MIN)
drv_ops->mfi_capabilities.support_ext_io_size = 1;
+ drv_ops->mfi_capabilities.support_fp_rlbypass = 1;
+
/* Convert capability to LE32 */
cpu_to_le32s((u32 *)&init_frame->driver_operations.mfi_capabilities);
@@ -1710,8 +1712,8 @@ megasas_build_ldio_fusion(struct megasas_instance *instance,
(MEGASAS_REQ_DESCRIPT_FLAGS_LD_IO
<< MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
if (fusion->adapter_type == INVADER_SERIES) {
- if (io_request->RaidContext.regLockFlags ==
- REGION_TYPE_UNUSED)
+ if (io_info.do_fp_rlbypass ||
+ (io_request->RaidContext.regLockFlags == REGION_TYPE_UNUSED))
cmd->request_desc->SCSIIO.RequestFlags =
(MEGASAS_REQ_DESCRIPT_FLAGS_NO_LOCK <<
MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.h b/drivers/scsi/megaraid/megaraid_sas_fusion.h
index a1f1c0b..db0978d 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.h
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.h
@@ -643,7 +643,8 @@ struct MR_SPAN_BLOCK_INFO {
struct MR_LD_RAID {
struct {
#if defined(__BIG_ENDIAN_BITFIELD)
- u32 reserved4:6;
+ u32 reserved4:5;
+ u32 fpBypassRegionLock:1;
u32 tmCapable:1;
u32 fpNonRWCapable:1;
u32 fpReadAcrossStripe:1;
@@ -667,7 +668,8 @@ struct MR_LD_RAID {
u32 fpReadAcrossStripe:1;
u32 fpNonRWCapable:1;
u32 tmCapable:1;
- u32 reserved4:6;
+ u32 fpBypassRegionLock:1;
+ u32 reserved4:5;
#endif
} capability;
__le32 reserved6;
@@ -737,7 +739,7 @@ struct IO_REQUEST_INFO {
u8 fpOkForIo;
u8 IoforUnevenSpan;
u8 start_span;
- u8 reserved;
+ u8 do_fp_rlbypass;
u64 start_row;
u8 span_arm; /* span[7:5], arm[4:0] */
u8 pd_after_lb;
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 07/15] megaraid_sas: Reply Descriptor Post Queue(RDPQ) support
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (5 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 06/15] megaraid_sas: Fastpath region lock bypass Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-29 15:19 ` Tomas Henzl
2016-01-28 15:34 ` [PATCH v2 08/15] megaraid_sas: Code optimization build_and_issue_cmd return-type Sumit Saxena
` (8 subsequent siblings)
15 siblings, 1 reply; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
This patch will create reply queue pool for each MSI-x index and will provide array of all reply queue base address
instead of single base address of legacy mode. Using this new interface Driver can support higher Queue depth allocating
more reply queue as scattered DMA pool.
If array mode is not supported driver will fall back to legacy method of allocation reply pool.
This method fall back controller queue depth to 1K max. To enable more than 1K QD, driver expect FW to support Array mode
and scratch_pad3 should provide new queue depth value.
Using this method, Firmware should not allow downgrade (OFU) if latest driver and latest FW report 4K QD and Array mode reply queue.
This type of FW upgrade may cause firmware fault and it should not be supported. Upgrade of FW will work,
but queue depth of the controller will be unchanged until reboot/driver reload.
I have accomodated Tomas' comments in this patch- fix few error handling cases inside functions-
megasas_alloc_cmdlist_fusion() and megasas_alloc_cmds_fusion(), fix typo- disable(old- disbale)
in rdpq_enable module parameter's description, modify print to reflect RDPQ support statement correctly.
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 6 +-
drivers/scsi/megaraid/megaraid_sas_base.c | 9 +
drivers/scsi/megaraid/megaraid_sas_fusion.c | 543 ++++++++++++++++------------
drivers/scsi/megaraid/megaraid_sas_fusion.h | 12 +-
4 files changed, 331 insertions(+), 239 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index 01135be..3b1ed2d 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -152,6 +152,7 @@
#define MFI_RESET_FLAGS MFI_INIT_READY| \
MFI_INIT_MFIMODE| \
MFI_INIT_ABORT
+#define MPI2_IOCINIT_MSGFLAG_RDPQ_ARRAY_MODE (0x01)
/*
* MFI frame flags
@@ -1416,6 +1417,7 @@ enum DCMD_TIMEOUT_ACTION {
#define MR_MAX_REPLY_QUEUES_EXT_OFFSET 0X003FC000
#define MR_MAX_REPLY_QUEUES_EXT_OFFSET_SHIFT 14
#define MR_MAX_MSIX_REG_ARRAY 16
+#define MR_RDPQ_MODE_OFFSET 0X00800000
/*
* register set for both 1068 and 1078 controllers
* structure extended for 1078 registers
@@ -1455,8 +1457,9 @@ struct megasas_register_set {
u32 outbound_scratch_pad ; /*00B0h*/
u32 outbound_scratch_pad_2; /*00B4h*/
+ u32 outbound_scratch_pad_3; /*00B8h*/
- u32 reserved_4[2]; /*00B8h*/
+ u32 reserved_4; /*00BCh*/
u32 inbound_low_queue_port ; /*00C0h*/
@@ -2117,6 +2120,7 @@ struct megasas_instance {
u8 mask_interrupts;
u16 max_chain_frame_sz;
u8 is_imr;
+ u8 is_rdpq;
bool dev_handle;
};
struct MR_LD_VF_MAP {
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index ea3994b..8df58c2 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -92,6 +92,10 @@ int smp_affinity_enable = 1;
module_param(smp_affinity_enable, int, S_IRUGO);
MODULE_PARM_DESC(smp_affinity_enable, "SMP affinity feature enable/disbale Default: enable(1)");
+int rdpq_enable = 1;
+module_param(rdpq_enable, int, S_IRUGO);
+MODULE_PARM_DESC(rdpq_enable, " Allocate reply queue in chunks for large queue depth enable/disable Default: disable(0)");
+
MODULE_LICENSE("GPL");
MODULE_VERSION(MEGASAS_VERSION);
MODULE_AUTHOR("megaraidlinux.pdl@avagotech.com");
@@ -5080,6 +5084,9 @@ static int megasas_init_fw(struct megasas_instance *instance)
instance->msix_vectors = ((scratch_pad_2
& MR_MAX_REPLY_QUEUES_EXT_OFFSET)
>> MR_MAX_REPLY_QUEUES_EXT_OFFSET_SHIFT) + 1;
+ if (rdpq_enable)
+ instance->is_rdpq = (scratch_pad_2 & MR_RDPQ_MODE_OFFSET) ?
+ 1 : 0;
fw_msix_count = instance->msix_vectors;
/* Save 1-15 reply post index address to local memory
* Index 0 is already saved from reg offset
@@ -5116,6 +5123,8 @@ static int megasas_init_fw(struct megasas_instance *instance)
dev_info(&instance->pdev->dev,
"current msix/online cpus\t: (%d/%d)\n",
instance->msix_vectors, (unsigned int)num_online_cpus());
+ dev_info(&instance->pdev->dev,
+ "RDPQ mode\t: (%s)\n", instance->is_rdpq ? "enabled" : "disabled");
tasklet_init(&instance->isr_tasklet, instance->instancet->tasklet,
(unsigned long)instance);
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 518b488..1351cae 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -92,6 +92,8 @@ void megasas_start_timer(struct megasas_instance *instance,
void *fn, unsigned long interval);
extern struct megasas_mgmt_info megasas_mgmt_info;
extern int resetwaittime;
+static void megasas_free_rdpq_fusion(struct megasas_instance *instance);
+static void megasas_free_reply_fusion(struct megasas_instance *instance);
@@ -205,112 +207,74 @@ megasas_fire_cmd_fusion(struct megasas_instance *instance,
#endif
}
-
/**
- * megasas_teardown_frame_pool_fusion - Destroy the cmd frame DMA pool
- * @instance: Adapter soft state
+ * megasas_free_cmds_fusion - Free all the cmds in the free cmd pool
+ * @instance: Adapter soft state
*/
-static void megasas_teardown_frame_pool_fusion(
- struct megasas_instance *instance)
+void
+megasas_free_cmds_fusion(struct megasas_instance *instance)
{
int i;
struct fusion_context *fusion = instance->ctrl_context;
-
- u16 max_cmd = instance->max_fw_cmds;
-
struct megasas_cmd_fusion *cmd;
- if (!fusion->sg_dma_pool || !fusion->sense_dma_pool) {
- dev_err(&instance->pdev->dev, "dma pool is null. SG Pool %p, "
- "sense pool : %p\n", fusion->sg_dma_pool,
- fusion->sense_dma_pool);
- return;
- }
-
- /*
- * Return all frames to pool
- */
- for (i = 0; i < max_cmd; i++) {
-
+ /* SG, Sense */
+ for (i = 0; i < instance->max_fw_cmds; i++) {
cmd = fusion->cmd_list[i];
-
- if (cmd->sg_frame)
- pci_pool_free(fusion->sg_dma_pool, cmd->sg_frame,
+ if (cmd) {
+ if (cmd->sg_frame)
+ pci_pool_free(fusion->sg_dma_pool, cmd->sg_frame,
cmd->sg_frame_phys_addr);
-
- if (cmd->sense)
- pci_pool_free(fusion->sense_dma_pool, cmd->sense,
+ if (cmd->sense)
+ pci_pool_free(fusion->sense_dma_pool, cmd->sense,
cmd->sense_phys_addr);
+ }
}
- /*
- * Now destroy the pool itself
- */
- pci_pool_destroy(fusion->sg_dma_pool);
- pci_pool_destroy(fusion->sense_dma_pool);
-
- fusion->sg_dma_pool = NULL;
- fusion->sense_dma_pool = NULL;
-}
-
-/**
- * megasas_free_cmds_fusion - Free all the cmds in the free cmd pool
- * @instance: Adapter soft state
- */
-void
-megasas_free_cmds_fusion(struct megasas_instance *instance)
-{
- int i;
- struct fusion_context *fusion = instance->ctrl_context;
-
- u32 max_cmds, req_sz, reply_sz, io_frames_sz;
-
+ if (fusion->sg_dma_pool) {
+ pci_pool_destroy(fusion->sg_dma_pool);
+ fusion->sg_dma_pool = NULL;
+ }
+ if (fusion->sense_dma_pool) {
+ pci_pool_destroy(fusion->sense_dma_pool);
+ fusion->sense_dma_pool = NULL;
+ }
- req_sz = fusion->request_alloc_sz;
- reply_sz = fusion->reply_alloc_sz;
- io_frames_sz = fusion->io_frames_alloc_sz;
- max_cmds = instance->max_fw_cmds;
+ /* Reply Frame, Desc*/
+ if (instance->is_rdpq)
+ megasas_free_rdpq_fusion(instance);
+ else
+ megasas_free_reply_fusion(instance);
- /* Free descriptors and request Frames memory */
+ /* Request Frame, Desc*/
if (fusion->req_frames_desc)
- dma_free_coherent(&instance->pdev->dev, req_sz,
- fusion->req_frames_desc,
- fusion->req_frames_desc_phys);
-
- if (fusion->reply_frames_desc) {
- pci_pool_free(fusion->reply_frames_desc_pool,
- fusion->reply_frames_desc,
- fusion->reply_frames_desc_phys);
- pci_pool_destroy(fusion->reply_frames_desc_pool);
- }
-
- if (fusion->io_request_frames) {
+ dma_free_coherent(&instance->pdev->dev,
+ fusion->request_alloc_sz, fusion->req_frames_desc,
+ fusion->req_frames_desc_phys);
+ if (fusion->io_request_frames)
pci_pool_free(fusion->io_request_frames_pool,
- fusion->io_request_frames,
- fusion->io_request_frames_phys);
+ fusion->io_request_frames,
+ fusion->io_request_frames_phys);
+ if (fusion->io_request_frames_pool) {
pci_pool_destroy(fusion->io_request_frames_pool);
+ fusion->io_request_frames_pool = NULL;
}
- /* Free the Fusion frame pool */
- megasas_teardown_frame_pool_fusion(instance);
- /* Free all the commands in the cmd_list */
- for (i = 0; i < max_cmds; i++)
+ /* cmd_list */
+ for (i = 0; i < instance->max_fw_cmds; i++)
kfree(fusion->cmd_list[i]);
- /* Free the cmd_list buffer itself */
kfree(fusion->cmd_list);
- fusion->cmd_list = NULL;
-
}
/**
- * megasas_create_frame_pool_fusion - Creates DMA pool for cmd frames
+ * megasas_create_sg_sense_fusion - Creates DMA pool for cmd frames
* @instance: Adapter soft state
*
*/
-static int megasas_create_frame_pool_fusion(struct megasas_instance *instance)
+static int megasas_create_sg_sense_fusion(struct megasas_instance *instance)
{
int i;
u32 max_cmd;
@@ -321,25 +285,17 @@ static int megasas_create_frame_pool_fusion(struct megasas_instance *instance)
max_cmd = instance->max_fw_cmds;
- /*
- * Use DMA pool facility provided by PCI layer
- */
-
- fusion->sg_dma_pool = pci_pool_create("sg_pool_fusion", instance->pdev,
- instance->max_chain_frame_sz,
- 4, 0);
- if (!fusion->sg_dma_pool) {
- dev_printk(KERN_DEBUG, &instance->pdev->dev, "failed to setup request pool fusion\n");
- return -ENOMEM;
- }
- fusion->sense_dma_pool = pci_pool_create("sense pool fusion",
- instance->pdev,
- SCSI_SENSE_BUFFERSIZE, 64, 0);
+ fusion->sg_dma_pool =
+ pci_pool_create("mr_sg", instance->pdev,
+ instance->max_chain_frame_sz, 4, 0);
+ /* SCSI_SENSE_BUFFERSIZE = 96 bytes */
+ fusion->sense_dma_pool =
+ pci_pool_create("mr_sense", instance->pdev,
+ SCSI_SENSE_BUFFERSIZE, 64, 0);
- if (!fusion->sense_dma_pool) {
- dev_printk(KERN_DEBUG, &instance->pdev->dev, "failed to setup sense pool fusion\n");
- pci_pool_destroy(fusion->sg_dma_pool);
- fusion->sg_dma_pool = NULL;
+ if (!fusion->sense_dma_pool || !fusion->sg_dma_pool) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
return -ENOMEM;
}
@@ -347,160 +303,280 @@ static int megasas_create_frame_pool_fusion(struct megasas_instance *instance)
* Allocate and attach a frame to each of the commands in cmd_list
*/
for (i = 0; i < max_cmd; i++) {
-
cmd = fusion->cmd_list[i];
-
cmd->sg_frame = pci_pool_alloc(fusion->sg_dma_pool,
- GFP_KERNEL,
- &cmd->sg_frame_phys_addr);
+ GFP_KERNEL, &cmd->sg_frame_phys_addr);
cmd->sense = pci_pool_alloc(fusion->sense_dma_pool,
- GFP_KERNEL, &cmd->sense_phys_addr);
- /*
- * megasas_teardown_frame_pool_fusion() takes care of freeing
- * whatever has been allocated
- */
+ GFP_KERNEL, &cmd->sense_phys_addr);
if (!cmd->sg_frame || !cmd->sense) {
- dev_printk(KERN_DEBUG, &instance->pdev->dev, "pci_pool_alloc failed\n");
- megasas_teardown_frame_pool_fusion(instance);
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
return -ENOMEM;
}
}
return 0;
}
-/**
- * megasas_alloc_cmds_fusion - Allocates the command packets
- * @instance: Adapter soft state
- *
- *
- * Each frame has a 32-bit field called context. This context is used to get
- * back the megasas_cmd_fusion from the frame when a frame gets completed
- * In this driver, the 32 bit values are the indices into an array cmd_list.
- * This array is used only to look up the megasas_cmd_fusion given the context.
- * The free commands themselves are maintained in a linked list called cmd_pool.
- *
- * cmds are formed in the io_request and sg_frame members of the
- * megasas_cmd_fusion. The context field is used to get a request descriptor
- * and is used as SMID of the cmd.
- * SMID value range is from 1 to max_fw_cmds.
- */
int
-megasas_alloc_cmds_fusion(struct megasas_instance *instance)
+megasas_alloc_cmdlist_fusion(struct megasas_instance *instance)
{
- int i, j, count;
- u32 max_cmd, io_frames_sz;
+ u32 max_cmd, i;
struct fusion_context *fusion;
- struct megasas_cmd_fusion *cmd;
- union MPI2_REPLY_DESCRIPTORS_UNION *reply_desc;
- u32 offset;
- dma_addr_t io_req_base_phys;
- u8 *io_req_base;
fusion = instance->ctrl_context;
max_cmd = instance->max_fw_cmds;
+ /*
+ * fusion->cmd_list is an array of struct megasas_cmd_fusion pointers.
+ * Allocate the dynamic array first and then allocate individual
+ * commands.
+ */
+ fusion->cmd_list = kzalloc(sizeof(struct megasas_cmd_fusion *) * max_cmd,
+ GFP_KERNEL);
+ if (!fusion->cmd_list) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
+ }
+
+ for (i = 0; i < max_cmd; i++) {
+ fusion->cmd_list[i] = kzalloc(sizeof(struct megasas_cmd_fusion),
+ GFP_KERNEL);
+ if (!fusion->cmd_list[i]) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
+ }
+ }
+ return 0;
+}
+int
+megasas_alloc_request_fusion(struct megasas_instance *instance)
+{
+ struct fusion_context *fusion;
+
+ fusion = instance->ctrl_context;
+
fusion->req_frames_desc =
dma_alloc_coherent(&instance->pdev->dev,
- fusion->request_alloc_sz,
- &fusion->req_frames_desc_phys, GFP_KERNEL);
-
+ fusion->request_alloc_sz,
+ &fusion->req_frames_desc_phys, GFP_KERNEL);
if (!fusion->req_frames_desc) {
- dev_err(&instance->pdev->dev, "Could not allocate memory for "
- "request_frames\n");
- goto fail_req_desc;
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
}
+ fusion->io_request_frames_pool =
+ pci_pool_create("mr_ioreq", instance->pdev,
+ fusion->io_frames_alloc_sz, 16, 0);
+
+ if (!fusion->io_request_frames_pool) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
+ }
+
+ fusion->io_request_frames =
+ pci_pool_alloc(fusion->io_request_frames_pool,
+ GFP_KERNEL, &fusion->io_request_frames_phys);
+ if (!fusion->io_request_frames) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
+ }
+ return 0;
+}
+
+int
+megasas_alloc_reply_fusion(struct megasas_instance *instance)
+{
+ int i, count;
+ struct fusion_context *fusion;
+ union MPI2_REPLY_DESCRIPTORS_UNION *reply_desc;
+ fusion = instance->ctrl_context;
+
count = instance->msix_vectors > 0 ? instance->msix_vectors : 1;
fusion->reply_frames_desc_pool =
- pci_pool_create("reply_frames pool", instance->pdev,
+ pci_pool_create("mr_reply", instance->pdev,
fusion->reply_alloc_sz * count, 16, 0);
if (!fusion->reply_frames_desc_pool) {
- dev_err(&instance->pdev->dev, "Could not allocate memory for "
- "reply_frame pool\n");
- goto fail_reply_desc;
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
}
- fusion->reply_frames_desc =
- pci_pool_alloc(fusion->reply_frames_desc_pool, GFP_KERNEL,
- &fusion->reply_frames_desc_phys);
- if (!fusion->reply_frames_desc) {
- dev_err(&instance->pdev->dev, "Could not allocate memory for "
- "reply_frame pool\n");
- pci_pool_destroy(fusion->reply_frames_desc_pool);
- goto fail_reply_desc;
+ fusion->reply_frames_desc[0] =
+ pci_pool_alloc(fusion->reply_frames_desc_pool,
+ GFP_KERNEL, &fusion->reply_frames_desc_phys[0]);
+ if (!fusion->reply_frames_desc[0]) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
}
-
- reply_desc = fusion->reply_frames_desc;
+ reply_desc = fusion->reply_frames_desc[0];
for (i = 0; i < fusion->reply_q_depth * count; i++, reply_desc++)
reply_desc->Words = cpu_to_le64(ULLONG_MAX);
- io_frames_sz = fusion->io_frames_alloc_sz;
+ /* This is not a rdpq mode, but driver still populate
+ * reply_frame_desc array to use same msix index in ISR path.
+ */
+ for (i = 0; i < (count - 1); i++)
+ fusion->reply_frames_desc[i + 1] =
+ fusion->reply_frames_desc[i] +
+ (fusion->reply_alloc_sz)/sizeof(union MPI2_REPLY_DESCRIPTORS_UNION);
- fusion->io_request_frames_pool =
- pci_pool_create("io_request_frames pool", instance->pdev,
- fusion->io_frames_alloc_sz, 16, 0);
+ return 0;
+}
- if (!fusion->io_request_frames_pool) {
- dev_err(&instance->pdev->dev, "Could not allocate memory for "
- "io_request_frame pool\n");
- goto fail_io_frames;
+int
+megasas_alloc_rdpq_fusion(struct megasas_instance *instance)
+{
+ int i, j, count;
+ struct fusion_context *fusion;
+ union MPI2_REPLY_DESCRIPTORS_UNION *reply_desc;
+
+ fusion = instance->ctrl_context;
+
+ fusion->rdpq_virt = pci_alloc_consistent(instance->pdev,
+ sizeof(struct MPI2_IOC_INIT_RDPQ_ARRAY_ENTRY) * MAX_MSIX_QUEUES_FUSION,
+ &fusion->rdpq_phys);
+ if (!fusion->rdpq_virt) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
}
- fusion->io_request_frames =
- pci_pool_alloc(fusion->io_request_frames_pool, GFP_KERNEL,
- &fusion->io_request_frames_phys);
- if (!fusion->io_request_frames) {
- dev_err(&instance->pdev->dev, "Could not allocate memory for "
- "io_request_frames frames\n");
- pci_pool_destroy(fusion->io_request_frames_pool);
- goto fail_io_frames;
+ memset(fusion->rdpq_virt, 0,
+ sizeof(struct MPI2_IOC_INIT_RDPQ_ARRAY_ENTRY) * MAX_MSIX_QUEUES_FUSION);
+ count = instance->msix_vectors > 0 ? instance->msix_vectors : 1;
+ fusion->reply_frames_desc_pool = pci_pool_create("mr_rdpq",
+ instance->pdev, fusion->reply_alloc_sz, 16, 0);
+
+ if (!fusion->reply_frames_desc_pool) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
}
- /*
- * fusion->cmd_list is an array of struct megasas_cmd_fusion pointers.
- * Allocate the dynamic array first and then allocate individual
- * commands.
- */
- fusion->cmd_list = kzalloc(sizeof(struct megasas_cmd_fusion *)
- * max_cmd, GFP_KERNEL);
+ for (i = 0; i < count; i++) {
+ fusion->reply_frames_desc[i] =
+ pci_pool_alloc(fusion->reply_frames_desc_pool,
+ GFP_KERNEL, &fusion->reply_frames_desc_phys[i]);
+ if (!fusion->reply_frames_desc[i]) {
+ dev_err(&instance->pdev->dev,
+ "Failed from %s %d\n", __func__, __LINE__);
+ return -ENOMEM;
+ }
- if (!fusion->cmd_list) {
- dev_printk(KERN_DEBUG, &instance->pdev->dev, "out of memory. Could not alloc "
- "memory for cmd_list_fusion\n");
- goto fail_cmd_list;
+ fusion->rdpq_virt[i].RDPQBaseAddress =
+ fusion->reply_frames_desc_phys[i];
+
+ reply_desc = fusion->reply_frames_desc[i];
+ for (j = 0; j < fusion->reply_q_depth; j++, reply_desc++)
+ reply_desc->Words = cpu_to_le64(ULLONG_MAX);
}
+ return 0;
+}
- max_cmd = instance->max_fw_cmds;
- for (i = 0; i < max_cmd; i++) {
- fusion->cmd_list[i] = kmalloc(sizeof(struct megasas_cmd_fusion),
- GFP_KERNEL);
- if (!fusion->cmd_list[i]) {
- dev_err(&instance->pdev->dev, "Could not alloc cmd list fusion\n");
+static void
+megasas_free_rdpq_fusion(struct megasas_instance *instance) {
- for (j = 0; j < i; j++)
- kfree(fusion->cmd_list[j]);
+ int i;
+ struct fusion_context *fusion;
- kfree(fusion->cmd_list);
- fusion->cmd_list = NULL;
- goto fail_cmd_list;
- }
+ fusion = instance->ctrl_context;
+
+ for (i = 0; i < MAX_MSIX_QUEUES_FUSION; i++) {
+ if (fusion->reply_frames_desc[i])
+ pci_pool_free(fusion->reply_frames_desc_pool,
+ fusion->reply_frames_desc[i],
+ fusion->reply_frames_desc_phys[i]);
}
- /* The first 256 bytes (SMID 0) is not used. Don't add to cmd list */
- io_req_base = fusion->io_request_frames +
- MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE;
- io_req_base_phys = fusion->io_request_frames_phys +
- MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE;
+ if (fusion->reply_frames_desc_pool)
+ pci_pool_destroy(fusion->reply_frames_desc_pool);
+
+ if (fusion->rdpq_virt)
+ pci_free_consistent(instance->pdev,
+ sizeof(struct MPI2_IOC_INIT_RDPQ_ARRAY_ENTRY) * MAX_MSIX_QUEUES_FUSION,
+ fusion->rdpq_virt, fusion->rdpq_phys);
+}
+
+static void
+megasas_free_reply_fusion(struct megasas_instance *instance) {
+
+ struct fusion_context *fusion;
+
+ fusion = instance->ctrl_context;
+
+ if (fusion->reply_frames_desc[0])
+ pci_pool_free(fusion->reply_frames_desc_pool,
+ fusion->reply_frames_desc[0],
+ fusion->reply_frames_desc_phys[0]);
+
+ if (fusion->reply_frames_desc_pool)
+ pci_pool_destroy(fusion->reply_frames_desc_pool);
+
+}
+
+
+/**
+ * megasas_alloc_cmds_fusion - Allocates the command packets
+ * @instance: Adapter soft state
+ *
+ *
+ * Each frame has a 32-bit field called context. This context is used to get
+ * back the megasas_cmd_fusion from the frame when a frame gets completed
+ * In this driver, the 32 bit values are the indices into an array cmd_list.
+ * This array is used only to look up the megasas_cmd_fusion given the context.
+ * The free commands themselves are maintained in a linked list called cmd_pool.
+ *
+ * cmds are formed in the io_request and sg_frame members of the
+ * megasas_cmd_fusion. The context field is used to get a request descriptor
+ * and is used as SMID of the cmd.
+ * SMID value range is from 1 to max_fw_cmds.
+ */
+int
+megasas_alloc_cmds_fusion(struct megasas_instance *instance)
+{
+ int i;
+ struct fusion_context *fusion;
+ struct megasas_cmd_fusion *cmd;
+ u32 offset;
+ dma_addr_t io_req_base_phys;
+ u8 *io_req_base;
+
+
+ fusion = instance->ctrl_context;
+
+ if (megasas_alloc_cmdlist_fusion(instance))
+ goto fail_exit;
+
+ if (megasas_alloc_request_fusion(instance))
+ goto fail_exit;
+
+ if (instance->is_rdpq) {
+ if (megasas_alloc_rdpq_fusion(instance))
+ goto fail_exit;
+ } else
+ if (megasas_alloc_reply_fusion(instance))
+ goto fail_exit;
+
+
+ /* The first 256 bytes (SMID 0) is not used. Don't add to the cmd list */
+ io_req_base = fusion->io_request_frames + MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE;
+ io_req_base_phys = fusion->io_request_frames_phys + MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE;
/*
* Add all the commands to command pool (fusion->cmd_pool)
*/
/* SMID 0 is reserved. Set SMID/index from 1 */
- for (i = 0; i < max_cmd; i++) {
+ for (i = 0; i < instance->max_fw_cmds; i++) {
cmd = fusion->cmd_list[i];
offset = MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE * i;
memset(cmd, 0, sizeof(struct megasas_cmd_fusion));
@@ -518,35 +594,13 @@ megasas_alloc_cmds_fusion(struct megasas_instance *instance)
cmd->io_request_phys_addr = io_req_base_phys + offset;
}
- /*
- * Create a frame pool and assign one frame to each cmd
- */
- if (megasas_create_frame_pool_fusion(instance)) {
- dev_printk(KERN_DEBUG, &instance->pdev->dev, "Error creating frame DMA pool\n");
- megasas_free_cmds_fusion(instance);
- goto fail_req_desc;
- }
+ if (megasas_create_sg_sense_fusion(instance))
+ goto fail_exit;
return 0;
-fail_cmd_list:
- pci_pool_free(fusion->io_request_frames_pool, fusion->io_request_frames,
- fusion->io_request_frames_phys);
- pci_pool_destroy(fusion->io_request_frames_pool);
-fail_io_frames:
- dma_free_coherent(&instance->pdev->dev, fusion->request_alloc_sz,
- fusion->reply_frames_desc,
- fusion->reply_frames_desc_phys);
- pci_pool_free(fusion->reply_frames_desc_pool,
- fusion->reply_frames_desc,
- fusion->reply_frames_desc_phys);
- pci_pool_destroy(fusion->reply_frames_desc_pool);
-
-fail_reply_desc:
- dma_free_coherent(&instance->pdev->dev, fusion->request_alloc_sz,
- fusion->req_frames_desc,
- fusion->req_frames_desc_phys);
-fail_req_desc:
+fail_exit:
+ megasas_free_cmds_fusion(instance);
return -ENOMEM;
}
@@ -594,16 +648,17 @@ int
megasas_ioc_init_fusion(struct megasas_instance *instance)
{
struct megasas_init_frame *init_frame;
- struct MPI2_IOC_INIT_REQUEST *IOCInitMessage;
+ struct MPI2_IOC_INIT_REQUEST *IOCInitMessage = NULL;
dma_addr_t ioc_init_handle;
struct megasas_cmd *cmd;
- u8 ret;
+ u8 ret, cur_rdpq_mode;
struct fusion_context *fusion;
union MEGASAS_REQUEST_DESCRIPTOR_UNION req_desc;
int i;
struct megasas_header *frame_hdr;
const char *sys_info;
MFI_CAPABILITIES *drv_ops;
+ u32 scratch_pad_2;
fusion = instance->ctrl_context;
@@ -615,6 +670,18 @@ megasas_ioc_init_fusion(struct megasas_instance *instance)
goto fail_get_cmd;
}
+ scratch_pad_2 = readl
+ (&instance->reg_set->outbound_scratch_pad_2);
+
+ cur_rdpq_mode = (scratch_pad_2 & MR_RDPQ_MODE_OFFSET) ? 1 : 0;
+
+ if (instance->is_rdpq && !cur_rdpq_mode) {
+ dev_err(&instance->pdev->dev, "Firmware downgrade *NOT SUPPORTED*"
+ " from RDPQ mode to non RDPQ mode\n");
+ ret = 1;
+ goto fail_fw_init;
+ }
+
IOCInitMessage =
dma_alloc_coherent(&instance->pdev->dev,
sizeof(struct MPI2_IOC_INIT_REQUEST),
@@ -636,7 +703,11 @@ megasas_ioc_init_fusion(struct megasas_instance *instance)
IOCInitMessage->SystemRequestFrameSize = cpu_to_le16(MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE / 4);
IOCInitMessage->ReplyDescriptorPostQueueDepth = cpu_to_le16(fusion->reply_q_depth);
- IOCInitMessage->ReplyDescriptorPostQueueAddress = cpu_to_le64(fusion->reply_frames_desc_phys);
+ IOCInitMessage->ReplyDescriptorPostQueueAddress = instance->is_rdpq ?
+ cpu_to_le64(fusion->rdpq_phys) :
+ cpu_to_le64(fusion->reply_frames_desc_phys[0]);
+ IOCInitMessage->MsgFlags = instance->is_rdpq ?
+ MPI2_IOCINIT_MSGFLAG_RDPQ_ARRAY_MODE : 0;
IOCInitMessage->SystemRequestFrameBaseAddress = cpu_to_le64(fusion->io_request_frames_phys);
IOCInitMessage->HostMSIxVectors = instance->msix_vectors;
init_frame = (struct megasas_init_frame *)cmd->frame;
@@ -1087,7 +1158,10 @@ megasas_init_adapter_fusion(struct megasas_instance *instance)
*/
instance->max_fw_cmds =
instance->instancet->read_fw_status_reg(reg_set) & 0x00FFFF;
- instance->max_fw_cmds = min(instance->max_fw_cmds, (u16)1008);
+ dev_info(&instance->pdev->dev,
+ "firmware support max fw cmd\t: (%d)\n", instance->max_fw_cmds);
+ if (!instance->is_rdpq)
+ instance->max_fw_cmds = min_t(u16, instance->max_fw_cmds, 1024);
/*
* Reduce the max supported cmds by 1. This is to ensure that the
@@ -2110,10 +2184,8 @@ complete_cmd_fusion(struct megasas_instance *instance, u32 MSIxIndex)
if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR)
return IRQ_HANDLED;
- desc = fusion->reply_frames_desc;
- desc += ((MSIxIndex * fusion->reply_alloc_sz)/
- sizeof(union MPI2_REPLY_DESCRIPTORS_UNION)) +
- fusion->last_reply_idx[MSIxIndex];
+ desc = fusion->reply_frames_desc[MSIxIndex] +
+ fusion->last_reply_idx[MSIxIndex];
reply_desc = (struct MPI2_SCSI_IO_SUCCESS_REPLY_DESCRIPTOR *)desc;
@@ -2208,9 +2280,7 @@ complete_cmd_fusion(struct megasas_instance *instance, u32 MSIxIndex)
/* Get the next reply descriptor */
if (!fusion->last_reply_idx[MSIxIndex])
- desc = fusion->reply_frames_desc +
- ((MSIxIndex * fusion->reply_alloc_sz)/
- sizeof(union MPI2_REPLY_DESCRIPTORS_UNION));
+ desc = fusion->reply_frames_desc[MSIxIndex];
else
desc++;
@@ -2688,17 +2758,18 @@ out:
void megasas_reset_reply_desc(struct megasas_instance *instance)
{
- int i, count;
+ int i, j, count;
struct fusion_context *fusion;
union MPI2_REPLY_DESCRIPTORS_UNION *reply_desc;
fusion = instance->ctrl_context;
count = instance->msix_vectors > 0 ? instance->msix_vectors : 1;
- for (i = 0 ; i < count ; i++)
+ for (i = 0 ; i < count ; i++) {
fusion->last_reply_idx[i] = 0;
- reply_desc = fusion->reply_frames_desc;
- for (i = 0 ; i < fusion->reply_q_depth * count; i++, reply_desc++)
- reply_desc->Words = cpu_to_le64(ULLONG_MAX);
+ reply_desc = fusion->reply_frames_desc[i];
+ for (j = 0 ; j < fusion->reply_q_depth; j++, reply_desc++)
+ reply_desc->Words = cpu_to_le64(ULLONG_MAX);
+ }
}
/*
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.h b/drivers/scsi/megaraid/megaraid_sas_fusion.h
index db0978d..80eaee2 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.h
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.h
@@ -928,6 +928,12 @@ struct MR_PD_CFG_SEQ_NUM_SYNC {
struct MR_PD_CFG_SEQ seq[1];
} __packed;
+struct MPI2_IOC_INIT_RDPQ_ARRAY_ENTRY {
+ u64 RDPQBaseAddress;
+ u32 Reserved1;
+ u32 Reserved2;
+};
+
struct fusion_context {
struct megasas_cmd_fusion **cmd_list;
dma_addr_t req_frames_desc_phys;
@@ -940,8 +946,8 @@ struct fusion_context {
struct dma_pool *sg_dma_pool;
struct dma_pool *sense_dma_pool;
- dma_addr_t reply_frames_desc_phys;
- union MPI2_REPLY_DESCRIPTORS_UNION *reply_frames_desc;
+ dma_addr_t reply_frames_desc_phys[MAX_MSIX_QUEUES_FUSION];
+ union MPI2_REPLY_DESCRIPTORS_UNION *reply_frames_desc[MAX_MSIX_QUEUES_FUSION];
struct dma_pool *reply_frames_desc_pool;
u16 last_reply_idx[MAX_MSIX_QUEUES_FUSION];
@@ -951,6 +957,8 @@ struct fusion_context {
u32 reply_alloc_sz;
u32 io_frames_alloc_sz;
+ struct MPI2_IOC_INIT_RDPQ_ARRAY_ENTRY *rdpq_virt;
+ dma_addr_t rdpq_phys;
u16 max_sge_in_main_msg;
u16 max_sge_in_chain;
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* Re: [PATCH v2 07/15] megaraid_sas: Reply Descriptor Post Queue(RDPQ) support
2016-01-28 15:34 ` [PATCH v2 07/15] megaraid_sas: Reply Descriptor Post Queue(RDPQ) support Sumit Saxena
@ 2016-01-29 15:19 ` Tomas Henzl
0 siblings, 0 replies; 20+ messages in thread
From: Tomas Henzl @ 2016-01-29 15:19 UTC (permalink / raw)
To: Sumit Saxena, jbottomley, hch, martin.petersen; +Cc: linux-scsi, kashyap.desai
On 28.1.2016 16:34, Sumit Saxena wrote:
> This patch will create reply queue pool for each MSI-x index and will provide array of all reply queue base address
> instead of single base address of legacy mode. Using this new interface Driver can support higher Queue depth allocating
> more reply queue as scattered DMA pool.
>
> If array mode is not supported driver will fall back to legacy method of allocation reply pool.
> This method fall back controller queue depth to 1K max. To enable more than 1K QD, driver expect FW to support Array mode
> and scratch_pad3 should provide new queue depth value.
>
> Using this method, Firmware should not allow downgrade (OFU) if latest driver and latest FW report 4K QD and Array mode reply queue.
> This type of FW upgrade may cause firmware fault and it should not be supported. Upgrade of FW will work,
> but queue depth of the controller will be unchanged until reboot/driver reload.
>
> I have accomodated Tomas' comments in this patch- fix few error handling cases inside functions-
> megasas_alloc_cmdlist_fusion() and megasas_alloc_cmds_fusion(), fix typo- disable(old- disbale)
> in rdpq_enable module parameter's description, modify print to reflect RDPQ support statement correctly.
>
>
> Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
> Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
> ---
> drivers/scsi/megaraid/megaraid_sas.h | 6 +-
> drivers/scsi/megaraid/megaraid_sas_base.c | 9 +
> drivers/scsi/megaraid/megaraid_sas_fusion.c | 543 ++++++++++++++++------------
> drivers/scsi/megaraid/megaraid_sas_fusion.h | 12 +-
> 4 files changed, 331 insertions(+), 239 deletions(-)
>
> diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
> index 01135be..3b1ed2d 100644
> --- a/drivers/scsi/megaraid/megaraid_sas.h
> +++ b/drivers/scsi/megaraid/megaraid_sas.h
> @@ -152,6 +152,7 @@
> #define MFI_RESET_FLAGS MFI_INIT_READY| \
> MFI_INIT_MFIMODE| \
> MFI_INIT_ABORT
> +#define MPI2_IOCINIT_MSGFLAG_RDPQ_ARRAY_MODE (0x01)
>
> /*
> * MFI frame flags
> @@ -1416,6 +1417,7 @@ enum DCMD_TIMEOUT_ACTION {
> #define MR_MAX_REPLY_QUEUES_EXT_OFFSET 0X003FC000
> #define MR_MAX_REPLY_QUEUES_EXT_OFFSET_SHIFT 14
> #define MR_MAX_MSIX_REG_ARRAY 16
> +#define MR_RDPQ_MODE_OFFSET 0X00800000
> /*
> * register set for both 1068 and 1078 controllers
> * structure extended for 1078 registers
> @@ -1455,8 +1457,9 @@ struct megasas_register_set {
>
> u32 outbound_scratch_pad ; /*00B0h*/
> u32 outbound_scratch_pad_2; /*00B4h*/
> + u32 outbound_scratch_pad_3; /*00B8h*/
>
> - u32 reserved_4[2]; /*00B8h*/
> + u32 reserved_4; /*00BCh*/
>
> u32 inbound_low_queue_port ; /*00C0h*/
>
> @@ -2117,6 +2120,7 @@ struct megasas_instance {
> u8 mask_interrupts;
> u16 max_chain_frame_sz;
> u8 is_imr;
> + u8 is_rdpq;
> bool dev_handle;
> };
> struct MR_LD_VF_MAP {
> diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
> index ea3994b..8df58c2 100644
> --- a/drivers/scsi/megaraid/megaraid_sas_base.c
> +++ b/drivers/scsi/megaraid/megaraid_sas_base.c
> @@ -92,6 +92,10 @@ int smp_affinity_enable = 1;
> module_param(smp_affinity_enable, int, S_IRUGO);
> MODULE_PARM_DESC(smp_affinity_enable, "SMP affinity feature enable/disbale Default: enable(1)");
>
> +int rdpq_enable = 1;
> +module_param(rdpq_enable, int, S_IRUGO);
> +MODULE_PARM_DESC(rdpq_enable, " Allocate reply queue in chunks for large queue depth enable/disable Default: disable(0)");
Thanks for fixing the typo, but the message still says "Default: disable(0)" - it should be "Default: enabled(1)"
A fix in next series if fine for me. Martin has also already added this series to scsi-queue
so my review tag is not needed.
Cheers,
Tomas
^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH v2 08/15] megaraid_sas: Code optimization build_and_issue_cmd return-type
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (6 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 07/15] megaraid_sas: Reply Descriptor Post Queue(RDPQ) support Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 09/15] megaraid_sas: Dual Queue depth support Sumit Saxena
` (7 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
build_and_issue_cmd should return SCSI_MLQUEUE_HOST_BUSY for few error case instead of returning 1.
There are no changes in this patch from last time sent patch.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
---
drivers/scsi/megaraid/megaraid_sas_base.c | 9 ++-------
drivers/scsi/megaraid/megaraid_sas_fusion.c | 4 ++--
2 files changed, 4 insertions(+), 9 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 8df58c2..edf8911 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -1636,7 +1636,7 @@ megasas_build_and_issue_cmd(struct megasas_instance *instance,
return 0;
out_return_cmd:
megasas_return_cmd(instance, cmd);
- return 1;
+ return SCSI_MLQUEUE_HOST_BUSY;
}
@@ -1728,12 +1728,7 @@ megasas_queue_command(struct Scsi_Host *shost, struct scsi_cmnd *scmd)
break;
}
- if (instance->instancet->build_and_issue_cmd(instance, scmd)) {
- dev_err(&instance->pdev->dev, "Err returned from build_and_issue_cmd\n");
- return SCSI_MLQUEUE_HOST_BUSY;
- }
-
- return 0;
+ return instance->instancet->build_and_issue_cmd(instance, scmd);
out_done:
scmd->scsi_done(scmd);
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 1351cae..f553830 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -2125,7 +2125,7 @@ megasas_build_and_issue_cmd_fusion(struct megasas_instance *instance,
req_desc = megasas_get_request_descriptor(instance, index-1);
if (!req_desc)
- return 1;
+ return SCSI_MLQUEUE_HOST_BUSY;
req_desc->Words = 0;
cmd->request_desc = req_desc;
@@ -2134,7 +2134,7 @@ megasas_build_and_issue_cmd_fusion(struct megasas_instance *instance,
megasas_return_cmd_fusion(instance, cmd);
dev_err(&instance->pdev->dev, "Error building command\n");
cmd->request_desc = NULL;
- return 1;
+ return SCSI_MLQUEUE_HOST_BUSY;
}
req_desc = cmd->request_desc;
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 09/15] megaraid_sas: Dual Queue depth support
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (7 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 08/15] megaraid_sas: Code optimization build_and_issue_cmd return-type Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 10/15] megaraid_sas: IO throttling support Sumit Saxena
` (6 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
This patch will add support for Dual Queue depth reported by firmware.
Below are key points-
1. For iMR controllers, firmware will report two queue depths- 1. Controller wide Queue depth 2. LDIO Queue depth(240).
Ofcourse, Controller wide Queue depth will be greater among two. Using this new method, iMR can provide larger Queue depth(QD)
for JBOD and limited QD for Virtual Disk(VD). This feature gives benefit for iMR product which will be used for deployment
with large number of JBOD and limited number of VD on setup.
2. megaraid_sas driver will throttle Read write LDIOs based when RW LDIOs reaches "LDIO Queue Depth".
3. This feature of dual queue depth can enabled/disabled via module parameter. Default behavior is: Dual Queue depth is enabled.
4. Added sysfs parameter "ldio_outstanding" for user to read LDIO outstanding at run time.
5. There are different flavors of firmware using same driver and for specific firmware this will be
turned on(by default) where it's really needed. For rest of firmware flavors, this will be switched off(not supported).
I have accomodated Tomas' comments provided on last time sent patch- use atomic_inc_return instead of using atomic_read() and
atomic_inc() separately inside function- megasas_build_and_issue_cmd_fusion().
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 9 +++
drivers/scsi/megaraid/megaraid_sas_base.c | 20 ++++++-
drivers/scsi/megaraid/megaraid_sas_fusion.c | 88 ++++++++++++++++++++++++++---
3 files changed, 107 insertions(+), 10 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index 3b1ed2d..2a2f491 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -1353,6 +1353,12 @@ enum DCMD_TIMEOUT_ACTION {
KILL_ADAPTER = 1,
IGNORE_TIMEOUT = 2,
};
+
+enum FW_BOOT_CONTEXT {
+ PROBE_CONTEXT = 0,
+ OCR_CONTEXT = 1,
+};
+
/* Frame Type */
#define IO_FRAME 0
#define PTHRU_FRAME 1
@@ -2038,6 +2044,8 @@ struct megasas_instance {
u16 max_fw_cmds;
u16 max_mfi_cmds;
u16 max_scsi_cmds;
+ u16 ldio_threshold;
+ u16 cur_can_queue;
u32 max_sectors_per_req;
struct megasas_aen_event *ev;
@@ -2068,6 +2076,7 @@ struct megasas_instance {
u32 fw_support_ieee;
atomic_t fw_outstanding;
+ atomic_t ldio_outstanding;
atomic_t fw_reset_no_pci_access;
struct megasas_instance_template *instancet;
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index edf8911..961c024 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -96,6 +96,10 @@ int rdpq_enable = 1;
module_param(rdpq_enable, int, S_IRUGO);
MODULE_PARM_DESC(rdpq_enable, " Allocate reply queue in chunks for large queue depth enable/disable Default: disable(0)");
+unsigned int dual_qdepth_disable;
+module_param(dual_qdepth_disable, int, S_IRUGO);
+MODULE_PARM_DESC(dual_qdepth_disable, "Disable dual queue depth feature. Default: 0");
+
MODULE_LICENSE("GPL");
MODULE_VERSION(MEGASAS_VERSION);
MODULE_AUTHOR("megaraidlinux.pdl@avagotech.com");
@@ -1976,7 +1980,7 @@ megasas_check_and_restore_queue_depth(struct megasas_instance *instance)
spin_lock_irqsave(instance->host->host_lock, flags);
instance->flag &= ~MEGASAS_FW_BUSY;
- instance->host->can_queue = instance->max_scsi_cmds;
+ instance->host->can_queue = instance->cur_can_queue;
spin_unlock_irqrestore(instance->host->host_lock, flags);
}
}
@@ -2941,6 +2945,16 @@ megasas_page_size_show(struct device *cdev,
return snprintf(buf, PAGE_SIZE, "%ld\n", (unsigned long)PAGE_SIZE - 1);
}
+static ssize_t
+megasas_ldio_outstanding_show(struct device *cdev, struct device_attribute *attr,
+ char *buf)
+{
+ struct Scsi_Host *shost = class_to_shost(cdev);
+ struct megasas_instance *instance = (struct megasas_instance *)shost->hostdata;
+
+ return snprintf(buf, PAGE_SIZE, "%d\n", atomic_read(&instance->ldio_outstanding));
+}
+
static DEVICE_ATTR(fw_crash_buffer, S_IRUGO | S_IWUSR,
megasas_fw_crash_buffer_show, megasas_fw_crash_buffer_store);
static DEVICE_ATTR(fw_crash_buffer_size, S_IRUGO,
@@ -2949,12 +2963,15 @@ static DEVICE_ATTR(fw_crash_state, S_IRUGO | S_IWUSR,
megasas_fw_crash_state_show, megasas_fw_crash_state_store);
static DEVICE_ATTR(page_size, S_IRUGO,
megasas_page_size_show, NULL);
+static DEVICE_ATTR(ldio_outstanding, S_IRUGO,
+ megasas_ldio_outstanding_show, NULL);
struct device_attribute *megaraid_host_attrs[] = {
&dev_attr_fw_crash_buffer_size,
&dev_attr_fw_crash_buffer,
&dev_attr_fw_crash_state,
&dev_attr_page_size,
+ &dev_attr_ldio_outstanding,
NULL,
};
@@ -4749,6 +4766,7 @@ megasas_init_adapter_mfi(struct megasas_instance *instance)
sema_init(&instance->ioctl_sem, (MEGASAS_MFI_IOCTL_CMDS));
}
+ instance->cur_can_queue = instance->max_scsi_cmds;
/*
* Create a pool of commands
*/
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index f553830..6b8547c 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -92,6 +92,7 @@ void megasas_start_timer(struct megasas_instance *instance,
void *fn, unsigned long interval);
extern struct megasas_mgmt_info megasas_mgmt_info;
extern int resetwaittime;
+extern unsigned int dual_qdepth_disable;
static void megasas_free_rdpq_fusion(struct megasas_instance *instance);
static void megasas_free_reply_fusion(struct megasas_instance *instance);
@@ -208,6 +209,67 @@ megasas_fire_cmd_fusion(struct megasas_instance *instance,
}
/**
+ * megasas_fusion_update_can_queue - Do all Adapter Queue depth related calculations here
+ * @instance: Adapter soft state
+ * fw_boot_context: Whether this function called during probe or after OCR
+ *
+ * This function is only for fusion controllers.
+ * Update host can queue, if firmware downgrade max supported firmware commands.
+ * Firmware upgrade case will be skiped because underlying firmware has
+ * more resource than exposed to the OS.
+ *
+ */
+static void
+megasas_fusion_update_can_queue(struct megasas_instance *instance, int fw_boot_context)
+{
+ u16 cur_max_fw_cmds = 0;
+ u16 ldio_threshold = 0;
+ struct megasas_register_set __iomem *reg_set;
+
+ reg_set = instance->reg_set;
+
+ cur_max_fw_cmds = readl(&instance->reg_set->outbound_scratch_pad_3) & 0x00FFFF;
+
+ if (dual_qdepth_disable || !cur_max_fw_cmds)
+ cur_max_fw_cmds = instance->instancet->read_fw_status_reg(reg_set) & 0x00FFFF;
+ else
+ ldio_threshold =
+ (instance->instancet->read_fw_status_reg(reg_set) & 0x00FFFF) - MEGASAS_FUSION_IOCTL_CMDS;
+
+ dev_info(&instance->pdev->dev,
+ "Current firmware maximum commands: %d\t LDIO threshold: %d\n",
+ cur_max_fw_cmds, ldio_threshold);
+
+ if (fw_boot_context == OCR_CONTEXT) {
+ cur_max_fw_cmds = cur_max_fw_cmds - 1;
+ if (cur_max_fw_cmds <= instance->max_fw_cmds) {
+ instance->cur_can_queue =
+ cur_max_fw_cmds - (MEGASAS_FUSION_INTERNAL_CMDS +
+ MEGASAS_FUSION_IOCTL_CMDS);
+ instance->host->can_queue = instance->cur_can_queue;
+ instance->ldio_threshold = ldio_threshold;
+ }
+ } else {
+ instance->max_fw_cmds = cur_max_fw_cmds;
+ instance->ldio_threshold = ldio_threshold;
+
+ if (!instance->is_rdpq)
+ instance->max_fw_cmds = min_t(u16, instance->max_fw_cmds, 1024);
+
+ /*
+ * Reduce the max supported cmds by 1. This is to ensure that the
+ * reply_q_sz (1 more than the max cmd that driver may send)
+ * does not exceed max cmds that the FW can support
+ */
+ instance->max_fw_cmds = instance->max_fw_cmds-1;
+
+ instance->max_scsi_cmds = instance->max_fw_cmds -
+ (MEGASAS_FUSION_INTERNAL_CMDS +
+ MEGASAS_FUSION_IOCTL_CMDS);
+ instance->cur_can_queue = instance->max_scsi_cmds;
+ }
+}
+/**
* megasas_free_cmds_fusion - Free all the cmds in the free cmd pool
* @instance: Adapter soft state
*/
@@ -738,6 +800,8 @@ megasas_ioc_init_fusion(struct megasas_instance *instance)
drv_ops->mfi_capabilities.support_ext_io_size = 1;
drv_ops->mfi_capabilities.support_fp_rlbypass = 1;
+ if (!dual_qdepth_disable)
+ drv_ops->mfi_capabilities.support_ext_queue_depth = 1;
/* Convert capability to LE32 */
cpu_to_le32s((u32 *)&init_frame->driver_operations.mfi_capabilities);
@@ -1153,15 +1217,7 @@ megasas_init_adapter_fusion(struct megasas_instance *instance)
reg_set = instance->reg_set;
- /*
- * Get various operational parameters from status register
- */
- instance->max_fw_cmds =
- instance->instancet->read_fw_status_reg(reg_set) & 0x00FFFF;
- dev_info(&instance->pdev->dev,
- "firmware support max fw cmd\t: (%d)\n", instance->max_fw_cmds);
- if (!instance->is_rdpq)
- instance->max_fw_cmds = min_t(u16, instance->max_fw_cmds, 1024);
+ megasas_fusion_update_can_queue(instance, PROBE_CONTEXT);
/*
* Reduce the max supported cmds by 1. This is to ensure that the
@@ -2119,6 +2175,14 @@ megasas_build_and_issue_cmd_fusion(struct megasas_instance *instance,
fusion = instance->ctrl_context;
+ if ((megasas_cmd_type(scmd) == READ_WRITE_LDIO) &&
+ instance->ldio_threshold &&
+ (atomic_inc_return(&instance->ldio_outstanding) >
+ instance->ldio_threshold)) {
+ atomic_dec(&instance->ldio_outstanding);
+ return SCSI_MLQUEUE_DEVICE_BUSY;
+ }
+
cmd = megasas_get_cmd_fusion(instance, scmd->request->tag);
index = cmd->index;
@@ -2249,6 +2313,8 @@ complete_cmd_fusion(struct megasas_instance *instance, u32 MSIxIndex)
map_cmd_status(cmd_fusion, status, extStatus);
scsi_io_req->RaidContext.status = 0;
scsi_io_req->RaidContext.exStatus = 0;
+ if (megasas_cmd_type(scmd_local) == READ_WRITE_LDIO)
+ atomic_dec(&instance->ldio_outstanding);
megasas_return_cmd_fusion(instance, cmd_fusion);
scsi_dma_unmap(scmd_local);
scmd_local->scsi_done(scmd_local);
@@ -3367,6 +3433,8 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
scmd_local->result =
megasas_check_mpio_paths(instance,
scmd_local);
+ if (megasas_cmd_type(scmd_local) == READ_WRITE_LDIO)
+ atomic_dec(&instance->ldio_outstanding);
megasas_return_cmd_fusion(instance, cmd_fusion);
scsi_dma_unmap(scmd_local);
scmd_local->scsi_done(scmd_local);
@@ -3459,6 +3527,8 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
}
megasas_reset_reply_desc(instance);
+ megasas_fusion_update_can_queue(instance, OCR_CONTEXT);
+
if (megasas_ioc_init_fusion(instance)) {
dev_warn(&instance->pdev->dev,
"megasas_ioc_init_fusion() failed!"
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 10/15] megaraid_sas: IO throttling support
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (8 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 09/15] megaraid_sas: Dual Queue depth support Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 11/15] megaraid_sas: Make adprecovery variable atomic Sumit Saxena
` (5 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
This patch will add capability in driver to tell firmware that it can throttle IOs in case Controller's Queue depth is downgraded post OFU
(Online firmware upgrade). This feature will ensure firmware can be downgraded from higher queue depth to lower queue depth without needing system reboot.
Added throttling code in IO path of driver, in case OS tries to send more IOs than post OFU firmware's queue depth.
I have accomodated Tomas' comments provided on last time sent patch- removed the un-necessary code of throttling of IOs against can_queue inside
function- megasas_build_and_issue_cmd_fusion() as SCSI mid layer will anyways does this.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 6 ++++--
drivers/scsi/megaraid/megaraid_sas_fusion.c | 1 +
2 files changed, 5 insertions(+), 2 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index 2a2f491..c8d25a7 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -1537,7 +1537,8 @@ union megasas_sgl_frame {
typedef union _MFI_CAPABILITIES {
struct {
#if defined(__BIG_ENDIAN_BITFIELD)
- u32 reserved:21;
+ u32 reserved:20;
+ u32 support_qd_throttling:1;
u32 support_fp_rlbypass:1;
u32 support_vfid_in_ioframe:1;
u32 support_ext_io_size:1;
@@ -1561,7 +1562,8 @@ typedef union _MFI_CAPABILITIES {
u32 support_ext_io_size:1;
u32 support_vfid_in_ioframe:1;
u32 support_fp_rlbypass:1;
- u32 reserved:21;
+ u32 support_qd_throttling:1;
+ u32 reserved:20;
#endif
} mfi_capabilities;
__le32 reg;
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 6b8547c..2c4912f 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -803,6 +803,7 @@ megasas_ioc_init_fusion(struct megasas_instance *instance)
if (!dual_qdepth_disable)
drv_ops->mfi_capabilities.support_ext_queue_depth = 1;
+ drv_ops->mfi_capabilities.support_qd_throttling = 1;
/* Convert capability to LE32 */
cpu_to_le32s((u32 *)&init_frame->driver_operations.mfi_capabilities);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 11/15] megaraid_sas: Make adprecovery variable atomic
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (9 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 10/15] megaraid_sas: IO throttling support Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-29 15:52 ` Tomas Henzl
2016-01-28 15:34 ` [PATCH v2 12/15] megaraid_sas: MFI adapter's OCR changes Sumit Saxena
` (4 subsequent siblings)
15 siblings, 1 reply; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
Make instance->adprecovery variable atomic and removes hba_lock spinlock while accessing instance->adprecovery.
Tomas commented on last time sent patch asking to use u8 instead of atomic for adprecovery. I agree that atomic_t is not required
here but this is done for not to touch legacy code of MFI adapters and replace hba_lock with atomic_t so there are no changes
in this patch on top of last time sent patch.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 2 +-
drivers/scsi/megaraid/megaraid_sas_base.c | 95 +++++++++++------------------
drivers/scsi/megaraid/megaraid_sas_fusion.c | 27 ++++----
3 files changed, 50 insertions(+), 74 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index c8d25a7..3e92f20 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -2101,7 +2101,7 @@ struct megasas_instance {
u16 drv_supported_vd_count;
u16 drv_supported_pd_count;
- u8 adprecovery;
+ atomic_t adprecovery;
unsigned long last_time;
u32 mfiStatus;
u32 last_seq_num;
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 961c024..1bd5da4 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -483,7 +483,7 @@ static int
megasas_check_reset_xscale(struct megasas_instance *instance,
struct megasas_register_set __iomem *regs)
{
- if ((instance->adprecovery != MEGASAS_HBA_OPERATIONAL) &&
+ if ((atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) &&
(le32_to_cpu(*instance->consumer) ==
MEGASAS_ADPRESET_INPROG_SIGN))
return 1;
@@ -619,7 +619,7 @@ static int
megasas_check_reset_ppc(struct megasas_instance *instance,
struct megasas_register_set __iomem *regs)
{
- if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL)
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL)
return 1;
return 0;
@@ -756,7 +756,7 @@ static int
megasas_check_reset_skinny(struct megasas_instance *instance,
struct megasas_register_set __iomem *regs)
{
- if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL)
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL)
return 1;
return 0;
@@ -950,9 +950,8 @@ static int
megasas_check_reset_gen2(struct megasas_instance *instance,
struct megasas_register_set __iomem *regs)
{
- if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL)
return 1;
- }
return 0;
}
@@ -998,7 +997,7 @@ megasas_issue_polled(struct megasas_instance *instance, struct megasas_cmd *cmd)
frame_hdr->cmd_status = MFI_STAT_INVALID_STATUS;
frame_hdr->flags |= cpu_to_le16(MFI_FRAME_DONT_POST_IN_REPLY_QUEUE);
- if ((instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) ||
+ if ((atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) ||
(instance->instancet->issue_dcmd(instance, cmd))) {
dev_err(&instance->pdev->dev, "Failed from %s %d\n",
__func__, __LINE__);
@@ -1026,7 +1025,7 @@ megasas_issue_blocked_cmd(struct megasas_instance *instance,
int ret = 0;
cmd->cmd_status_drv = MFI_STAT_INVALID_STATUS;
- if ((instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) ||
+ if ((atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) ||
(instance->instancet->issue_dcmd(instance, cmd))) {
dev_err(&instance->pdev->dev, "Failed from %s %d\n",
__func__, __LINE__);
@@ -1090,7 +1089,7 @@ megasas_issue_blocked_abort_cmd(struct megasas_instance *instance,
cmd->sync_cmd = 1;
cmd->cmd_status_drv = MFI_STAT_INVALID_STATUS;
- if ((instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) ||
+ if ((atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) ||
(instance->instancet->issue_dcmd(instance, cmd))) {
dev_err(&instance->pdev->dev, "Failed from %s %d\n",
__func__, __LINE__);
@@ -1653,7 +1652,6 @@ static int
megasas_queue_command(struct Scsi_Host *shost, struct scsi_cmnd *scmd)
{
struct megasas_instance *instance;
- unsigned long flags;
struct MR_PRIV_DEVICE *mr_device_priv_data;
instance = (struct megasas_instance *)
@@ -1668,24 +1666,20 @@ megasas_queue_command(struct Scsi_Host *shost, struct scsi_cmnd *scmd)
if (instance->issuepend_done == 0)
return SCSI_MLQUEUE_HOST_BUSY;
- spin_lock_irqsave(&instance->hba_lock, flags);
/* Check for an mpio path and adjust behavior */
- if (instance->adprecovery == MEGASAS_ADPRESET_SM_INFAULT) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_ADPRESET_SM_INFAULT) {
if (megasas_check_mpio_paths(instance, scmd) ==
(DID_RESET << 16)) {
- spin_unlock_irqrestore(&instance->hba_lock, flags);
return SCSI_MLQUEUE_HOST_BUSY;
} else {
- spin_unlock_irqrestore(&instance->hba_lock, flags);
scmd->result = DID_NO_CONNECT << 16;
scmd->scsi_done(scmd);
return 0;
}
}
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) {
- spin_unlock_irqrestore(&instance->hba_lock, flags);
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) {
scmd->result = DID_NO_CONNECT << 16;
scmd->scsi_done(scmd);
return 0;
@@ -1693,23 +1687,17 @@ megasas_queue_command(struct Scsi_Host *shost, struct scsi_cmnd *scmd)
mr_device_priv_data = scmd->device->hostdata;
if (!mr_device_priv_data) {
- spin_unlock_irqrestore(&instance->hba_lock, flags);
scmd->result = DID_NO_CONNECT << 16;
scmd->scsi_done(scmd);
return 0;
}
- if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
- spin_unlock_irqrestore(&instance->hba_lock, flags);
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL)
return SCSI_MLQUEUE_HOST_BUSY;
- }
- if (mr_device_priv_data->tm_busy) {
- spin_unlock_irqrestore(&instance->hba_lock, flags);
+ if (mr_device_priv_data->tm_busy)
return SCSI_MLQUEUE_DEVICE_BUSY;
- }
- spin_unlock_irqrestore(&instance->hba_lock, flags);
scmd->result = 0;
@@ -1942,7 +1930,7 @@ static void megasas_complete_outstanding_ioctls(struct megasas_instance *instanc
void megaraid_sas_kill_hba(struct megasas_instance *instance)
{
/* Set critical error to block I/O & ioctls in case caller didn't */
- instance->adprecovery = MEGASAS_HW_CRITICAL_ERROR;
+ atomic_set(&instance->adprecovery, MEGASAS_HW_CRITICAL_ERROR);
/* Wait 1 second to ensure IO or ioctls in build have posted */
msleep(1000);
if ((instance->pdev->device == PCI_DEVICE_ID_LSI_SAS0073SKINNY) ||
@@ -2002,7 +1990,7 @@ static void megasas_complete_cmd_dpc(unsigned long instance_addr)
unsigned long flags;
/* If we have already declared adapter dead, donot complete cmds */
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR)
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR)
return;
spin_lock_irqsave(&instance->completion_lock, flags);
@@ -2071,7 +2059,7 @@ void megasas_do_ocr(struct megasas_instance *instance)
*instance->consumer = cpu_to_le32(MEGASAS_ADPRESET_INPROG_SIGN);
}
instance->instancet->disable_intr(instance);
- instance->adprecovery = MEGASAS_ADPRESET_SM_INFAULT;
+ atomic_set(&instance->adprecovery, MEGASAS_ADPRESET_SM_INFAULT);
instance->issuepend_done = 0;
atomic_set(&instance->fw_outstanding, 0);
@@ -2470,18 +2458,14 @@ static int megasas_wait_for_outstanding(struct megasas_instance *instance)
int i;
u32 reset_index;
u32 wait_time = MEGASAS_RESET_WAIT_TIME;
- u8 adprecovery;
unsigned long flags;
struct list_head clist_local;
struct megasas_cmd *reset_cmd;
u32 fw_state;
u8 kill_adapter_flag;
- spin_lock_irqsave(&instance->hba_lock, flags);
- adprecovery = instance->adprecovery;
- spin_unlock_irqrestore(&instance->hba_lock, flags);
- if (adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) {
INIT_LIST_HEAD(&clist_local);
spin_lock_irqsave(&instance->hba_lock, flags);
@@ -2492,18 +2476,13 @@ static int megasas_wait_for_outstanding(struct megasas_instance *instance)
dev_notice(&instance->pdev->dev, "HBA reset wait ...\n");
for (i = 0; i < wait_time; i++) {
msleep(1000);
- spin_lock_irqsave(&instance->hba_lock, flags);
- adprecovery = instance->adprecovery;
- spin_unlock_irqrestore(&instance->hba_lock, flags);
- if (adprecovery == MEGASAS_HBA_OPERATIONAL)
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HBA_OPERATIONAL)
break;
}
- if (adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) {
dev_notice(&instance->pdev->dev, "reset: Stopping HBA.\n");
- spin_lock_irqsave(&instance->hba_lock, flags);
- instance->adprecovery = MEGASAS_HW_CRITICAL_ERROR;
- spin_unlock_irqrestore(&instance->hba_lock, flags);
+ atomic_set(&instance->adprecovery, MEGASAS_HW_CRITICAL_ERROR);
return FAILED;
}
@@ -2612,9 +2591,7 @@ static int megasas_wait_for_outstanding(struct megasas_instance *instance)
&instance->reg_set->inbound_doorbell);
}
megasas_dump_pending_frames(instance);
- spin_lock_irqsave(&instance->hba_lock, flags);
- instance->adprecovery = MEGASAS_HW_CRITICAL_ERROR;
- spin_unlock_irqrestore(&instance->hba_lock, flags);
+ atomic_set(&instance->adprecovery, MEGASAS_HW_CRITICAL_ERROR);
return FAILED;
}
@@ -2641,7 +2618,7 @@ static int megasas_generic_reset(struct scsi_cmnd *scmd)
scmd_printk(KERN_NOTICE, scmd, "megasas: RESET cmd=%x retries=%x\n",
scmd->cmnd[0], scmd->retries);
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) {
dev_err(&instance->pdev->dev, "cannot recover from previous reset failures\n");
return FAILED;
}
@@ -3385,13 +3362,13 @@ process_fw_state_change_wq(struct work_struct *work)
u32 wait;
unsigned long flags;
- if (instance->adprecovery != MEGASAS_ADPRESET_SM_INFAULT) {
+ if (atomic_read(&instance->adprecovery) != MEGASAS_ADPRESET_SM_INFAULT) {
dev_notice(&instance->pdev->dev, "error, recovery st %x\n",
- instance->adprecovery);
+ atomic_read(&instance->adprecovery));
return ;
}
- if (instance->adprecovery == MEGASAS_ADPRESET_SM_INFAULT) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_ADPRESET_SM_INFAULT) {
dev_notice(&instance->pdev->dev, "FW detected to be in fault"
"state, restarting it...\n");
@@ -3434,7 +3411,7 @@ process_fw_state_change_wq(struct work_struct *work)
megasas_issue_init_mfi(instance);
spin_lock_irqsave(&instance->hba_lock, flags);
- instance->adprecovery = MEGASAS_HBA_OPERATIONAL;
+ atomic_set(&instance->adprecovery, MEGASAS_HBA_OPERATIONAL);
spin_unlock_irqrestore(&instance->hba_lock, flags);
instance->instancet->enable_intr(instance);
@@ -3499,14 +3476,14 @@ megasas_deplete_reply_queue(struct megasas_instance *instance,
instance->instancet->disable_intr(instance);
- instance->adprecovery = MEGASAS_ADPRESET_SM_INFAULT;
+ atomic_set(&instance->adprecovery, MEGASAS_ADPRESET_SM_INFAULT);
instance->issuepend_done = 0;
atomic_set(&instance->fw_outstanding, 0);
megasas_internal_reset_defer_cmds(instance);
dev_notice(&instance->pdev->dev, "fwState=%x, stage:%d\n",
- fw_state, instance->adprecovery);
+ fw_state, atomic_read(&instance->adprecovery));
schedule_work(&instance->work_init);
return IRQ_HANDLED;
@@ -5795,7 +5772,7 @@ static int megasas_probe_one(struct pci_dev *pdev,
instance->flag_ieee = 0;
instance->ev = NULL;
instance->issuepend_done = 1;
- instance->adprecovery = MEGASAS_HBA_OPERATIONAL;
+ atomic_set(&instance->adprecovery, MEGASAS_HBA_OPERATIONAL);
instance->is_imr = 0;
instance->evt_detail = pci_alloc_consistent(pdev,
@@ -5974,7 +5951,7 @@ static void megasas_flush_cache(struct megasas_instance *instance)
struct megasas_cmd *cmd;
struct megasas_dcmd_frame *dcmd;
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR)
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR)
return;
cmd = megasas_get_cmd(instance);
@@ -6017,7 +5994,7 @@ static void megasas_shutdown_controller(struct megasas_instance *instance,
struct megasas_cmd *cmd;
struct megasas_dcmd_frame *dcmd;
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR)
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR)
return;
cmd = megasas_get_cmd(instance);
@@ -6462,7 +6439,7 @@ static int megasas_set_crash_dump_params_ioctl(struct megasas_cmd *cmd)
for (i = 0; i < megasas_mgmt_info.max_index; i++) {
local_instance = megasas_mgmt_info.instance[i];
if (local_instance && local_instance->crash_dump_drv_support) {
- if ((local_instance->adprecovery ==
+ if ((atomic_read(&local_instance->adprecovery) ==
MEGASAS_HBA_OPERATIONAL) &&
!megasas_set_crash_dump_params(local_instance,
crash_support)) {
@@ -6710,7 +6687,7 @@ static int megasas_mgmt_ioctl_fw(struct file *file, unsigned long arg)
goto out_kfree_ioc;
}
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) {
dev_err(&instance->pdev->dev, "Controller in crit error\n");
error = -ENODEV;
goto out_kfree_ioc;
@@ -6729,7 +6706,7 @@ static int megasas_mgmt_ioctl_fw(struct file *file, unsigned long arg)
for (i = 0; i < wait_time; i++) {
spin_lock_irqsave(&instance->hba_lock, flags);
- if (instance->adprecovery == MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HBA_OPERATIONAL) {
spin_unlock_irqrestore(&instance->hba_lock, flags);
break;
}
@@ -6744,7 +6721,7 @@ static int megasas_mgmt_ioctl_fw(struct file *file, unsigned long arg)
}
spin_lock_irqsave(&instance->hba_lock, flags);
- if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) {
spin_unlock_irqrestore(&instance->hba_lock, flags);
dev_err(&instance->pdev->dev, "timed out while"
@@ -6786,7 +6763,7 @@ static int megasas_mgmt_ioctl_aen(struct file *file, unsigned long arg)
if (!instance)
return -ENODEV;
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) {
return -ENODEV;
}
@@ -6797,7 +6774,7 @@ static int megasas_mgmt_ioctl_aen(struct file *file, unsigned long arg)
for (i = 0; i < wait_time; i++) {
spin_lock_irqsave(&instance->hba_lock, flags);
- if (instance->adprecovery == MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HBA_OPERATIONAL) {
spin_unlock_irqrestore(&instance->hba_lock,
flags);
break;
@@ -6814,7 +6791,7 @@ static int megasas_mgmt_ioctl_aen(struct file *file, unsigned long arg)
}
spin_lock_irqsave(&instance->hba_lock, flags);
- if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) {
spin_unlock_irqrestore(&instance->hba_lock, flags);
dev_err(&instance->pdev->dev, "timed out while waiting"
"for HBA to recover\n");
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 2c4912f..64926f7 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -2246,7 +2246,7 @@ complete_cmd_fusion(struct megasas_instance *instance, u32 MSIxIndex)
fusion = instance->ctrl_context;
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR)
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR)
return IRQ_HANDLED;
desc = fusion->reply_frames_desc[MSIxIndex] +
@@ -2413,7 +2413,7 @@ megasas_complete_cmd_dpc_fusion(unsigned long instance_addr)
/* If we have already declared adapter dead, donot complete cmds */
spin_lock_irqsave(&instance->hba_lock, flags);
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) {
spin_unlock_irqrestore(&instance->hba_lock, flags);
return;
}
@@ -3197,7 +3197,7 @@ int megasas_task_abort_fusion(struct scsi_cmnd *scmd)
instance = (struct megasas_instance *)scmd->device->host->hostdata;
fusion = instance->ctrl_context;
- if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) {
dev_err(&instance->pdev->dev, "Controller is not OPERATIONAL,"
"SCSI host:%d\n", instance->host->host_no);
ret = FAILED;
@@ -3277,7 +3277,7 @@ int megasas_reset_target_fusion(struct scsi_cmnd *scmd)
instance = (struct megasas_instance *)scmd->device->host->hostdata;
fusion = instance->ctrl_context;
- if (instance->adprecovery != MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) {
dev_err(&instance->pdev->dev, "Controller is not OPERATIONAL,"
"SCSI host:%d\n", instance->host->host_no);
ret = FAILED;
@@ -3366,7 +3366,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
mutex_lock(&instance->reset_mutex);
- if (instance->adprecovery == MEGASAS_HW_CRITICAL_ERROR) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) {
dev_warn(&instance->pdev->dev, "Hardware critical error, "
"returning FAILED for scsi%d.\n",
instance->host->host_no);
@@ -3381,7 +3381,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
instance->crash_dump_app_support && reason) {
dev_info(&instance->pdev->dev, "IO/DCMD timeout is detected, "
"forcibly FAULT Firmware\n");
- instance->adprecovery = MEGASAS_ADPRESET_SM_INFAULT;
+ atomic_set(&instance->adprecovery, MEGASAS_ADPRESET_SM_INFAULT);
status_reg = readl(&instance->reg_set->doorbell);
writel(status_reg | MFI_STATE_FORCE_OCR,
&instance->reg_set->doorbell);
@@ -3393,10 +3393,10 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
dev_dbg(&instance->pdev->dev, "waiting for [%d] "
"seconds for crash dump collection and OCR "
"to be done\n", (io_timeout_in_crash_mode * 3));
- } while ((instance->adprecovery != MEGASAS_HBA_OPERATIONAL) &&
+ } while ((atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) &&
(io_timeout_in_crash_mode < 80));
- if (instance->adprecovery == MEGASAS_HBA_OPERATIONAL) {
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HBA_OPERATIONAL) {
dev_info(&instance->pdev->dev, "OCR done for IO "
"timeout case\n");
retval = SUCCESS;
@@ -3413,14 +3413,14 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
if (instance->requestorId && !instance->skip_heartbeat_timer_del)
del_timer_sync(&instance->sriov_heartbeat_timer);
set_bit(MEGASAS_FUSION_IN_RESET, &instance->reset_flags);
- instance->adprecovery = MEGASAS_ADPRESET_SM_POLLING;
+ atomic_set(&instance->adprecovery, MEGASAS_ADPRESET_SM_POLLING);
instance->instancet->disable_intr(instance);
msleep(1000);
/* First try waiting for commands to complete */
if (megasas_wait_for_outstanding_fusion(instance, reason,
&convert)) {
- instance->adprecovery = MEGASAS_ADPRESET_SM_INFAULT;
+ atomic_set(&instance->adprecovery, MEGASAS_ADPRESET_SM_INFAULT);
dev_warn(&instance->pdev->dev, "resetting fusion "
"adapter scsi%d.\n", instance->host->host_no);
if (convert)
@@ -3503,8 +3503,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
status_reg);
megaraid_sas_kill_hba(instance);
instance->skip_heartbeat_timer_del = 1;
- instance->adprecovery =
- MEGASAS_HW_CRITICAL_ERROR;
+ atomic_set(&instance->adprecovery, MEGASAS_HW_CRITICAL_ERROR);
retval = FAILED;
goto out;
}
@@ -3563,7 +3562,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
clear_bit(MEGASAS_FUSION_IN_RESET,
&instance->reset_flags);
instance->instancet->enable_intr(instance);
- instance->adprecovery = MEGASAS_HBA_OPERATIONAL;
+ atomic_set(&instance->adprecovery, MEGASAS_HBA_OPERATIONAL);
/* Restart SR-IOV heartbeat */
if (instance->requestorId) {
@@ -3608,7 +3607,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
}
clear_bit(MEGASAS_FUSION_IN_RESET, &instance->reset_flags);
instance->instancet->enable_intr(instance);
- instance->adprecovery = MEGASAS_HBA_OPERATIONAL;
+ atomic_set(&instance->adprecovery, MEGASAS_HBA_OPERATIONAL);
}
out:
clear_bit(MEGASAS_FUSION_IN_RESET, &instance->reset_flags);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* Re: [PATCH v2 11/15] megaraid_sas: Make adprecovery variable atomic
2016-01-28 15:34 ` [PATCH v2 11/15] megaraid_sas: Make adprecovery variable atomic Sumit Saxena
@ 2016-01-29 15:52 ` Tomas Henzl
0 siblings, 0 replies; 20+ messages in thread
From: Tomas Henzl @ 2016-01-29 15:52 UTC (permalink / raw)
To: Sumit Saxena, jbottomley, hch, martin.petersen; +Cc: linux-scsi, kashyap.desai
On 28.1.2016 16:34, Sumit Saxena wrote:
> Make instance->adprecovery variable atomic and removes hba_lock spinlock while accessing instance->adprecovery.
>
> Tomas commented on last time sent patch asking to use u8 instead of atomic for adprecovery. I agree that atomic_t is not required
> here but this is done for not to touch legacy code of MFI adapters and replace hba_lock with atomic_t so there are no changes
> in this patch on top of last time sent patch.
The way how both, the atomic variable now and the hba_lock we had before, work
is not equal - the sw will now behave differently, so you can't replace it
and expect nothing would change
The protection they provide was/is weak anyway, I think that it does not
protect you from anything.
btw. you forgot to remove the hba_lock in megasas_complete_cmd_dpc_fusion
OK, old code, you don't want to touch/test, and the maintainer has accepted it
so I accept it too.
--tm
^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH v2 12/15] megaraid_sas: MFI adapter's OCR changes
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (10 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 11/15] megaraid_sas: Make adprecovery variable atomic Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 13/15] megaraid_sas: Introduce module parameter for SCSI command-timeout Sumit Saxena
` (3 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
Optimized MFI adapters' OCR path, particularly megasas_wait_for_outstanding() function.
Accomodated Tomas' comments provided on last time sent patch- remove redundant checks when label- kill_hba_and_failed is being called.
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
---
drivers/scsi/megaraid/megaraid_sas_base.c | 101 +++++++++++++++---------------
1 file changed, 50 insertions(+), 51 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 1bd5da4..d2ea977 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -2455,15 +2455,19 @@ void megasas_sriov_heartbeat_handler(unsigned long instance_addr)
*/
static int megasas_wait_for_outstanding(struct megasas_instance *instance)
{
- int i;
+ int i, sl, outstanding;
u32 reset_index;
u32 wait_time = MEGASAS_RESET_WAIT_TIME;
unsigned long flags;
struct list_head clist_local;
struct megasas_cmd *reset_cmd;
u32 fw_state;
- u8 kill_adapter_flag;
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) {
+ dev_info(&instance->pdev->dev, "%s:%d HBA is killed.\n",
+ __func__, __LINE__);
+ return FAILED;
+ }
if (atomic_read(&instance->adprecovery) != MEGASAS_HBA_OPERATIONAL) {
@@ -2520,7 +2524,7 @@ static int megasas_wait_for_outstanding(struct megasas_instance *instance)
}
for (i = 0; i < resetwaittime; i++) {
- int outstanding = atomic_read(&instance->fw_outstanding);
+ outstanding = atomic_read(&instance->fw_outstanding);
if (!outstanding)
break;
@@ -2539,65 +2543,60 @@ static int megasas_wait_for_outstanding(struct megasas_instance *instance)
}
i = 0;
- kill_adapter_flag = 0;
+ outstanding = atomic_read(&instance->fw_outstanding);
+ fw_state = instance->instancet->read_fw_status_reg(instance->reg_set) & MFI_STATE_MASK;
+
+ if ((!outstanding && (fw_state == MFI_STATE_OPERATIONAL)))
+ goto no_outstanding;
+
+ if (instance->disableOnlineCtrlReset)
+ goto kill_hba_and_failed;
do {
- fw_state = instance->instancet->read_fw_status_reg(
- instance->reg_set) & MFI_STATE_MASK;
- if ((fw_state == MFI_STATE_FAULT) &&
- (instance->disableOnlineCtrlReset == 0)) {
- if (i == 3) {
- kill_adapter_flag = 2;
- break;
- }
+ if ((fw_state == MFI_STATE_FAULT) || atomic_read(&instance->fw_outstanding)) {
+ dev_info(&instance->pdev->dev,
+ "%s:%d waiting_for_outstanding: before issue OCR. FW state = 0x%x, oustanding 0x%x\n",
+ __func__, __LINE__, fw_state, atomic_read(&instance->fw_outstanding));
+ if (i == 3)
+ goto kill_hba_and_failed;
megasas_do_ocr(instance);
- kill_adapter_flag = 1;
- /* wait for 1 secs to let FW finish the pending cmds */
- msleep(1000);
+ if (atomic_read(&instance->adprecovery) == MEGASAS_HW_CRITICAL_ERROR) {
+ dev_info(&instance->pdev->dev, "%s:%d OCR failed and HBA is killed.\n",
+ __func__, __LINE__);
+ return FAILED;
+ }
+ dev_info(&instance->pdev->dev, "%s:%d waiting_for_outstanding: after issue OCR.\n",
+ __func__, __LINE__);
+
+ for (sl = 0; sl < 10; sl++)
+ msleep(500);
+
+ outstanding = atomic_read(&instance->fw_outstanding);
+
+ fw_state = instance->instancet->read_fw_status_reg(instance->reg_set) & MFI_STATE_MASK;
+ if ((!outstanding && (fw_state == MFI_STATE_OPERATIONAL)))
+ goto no_outstanding;
}
i++;
} while (i <= 3);
- if (atomic_read(&instance->fw_outstanding) && !kill_adapter_flag) {
- if (instance->disableOnlineCtrlReset == 0) {
- megasas_do_ocr(instance);
+no_outstanding:
- /* wait for 5 secs to let FW finish the pending cmds */
- for (i = 0; i < wait_time; i++) {
- int outstanding =
- atomic_read(&instance->fw_outstanding);
- if (!outstanding)
- return SUCCESS;
- msleep(1000);
- }
- }
- }
+ dev_info(&instance->pdev->dev, "%s:%d no more pending commands remain after reset handling.\n",
+ __func__, __LINE__);
+ return SUCCESS;
- if (atomic_read(&instance->fw_outstanding) ||
- (kill_adapter_flag == 2)) {
- dev_notice(&instance->pdev->dev, "pending cmds after reset\n");
- /*
- * Send signal to FW to stop processing any pending cmds.
- * The controller will be taken offline by the OS now.
- */
- if ((instance->pdev->device ==
- PCI_DEVICE_ID_LSI_SAS0073SKINNY) ||
- (instance->pdev->device ==
- PCI_DEVICE_ID_LSI_SAS0071SKINNY)) {
- writel(MFI_STOP_ADP,
- &instance->reg_set->doorbell);
- } else {
- writel(MFI_STOP_ADP,
- &instance->reg_set->inbound_doorbell);
- }
- megasas_dump_pending_frames(instance);
- atomic_set(&instance->adprecovery, MEGASAS_HW_CRITICAL_ERROR);
- return FAILED;
- }
+kill_hba_and_failed:
- dev_notice(&instance->pdev->dev, "no pending cmds after reset\n");
+ /* Reset not supported, kill adapter */
+ dev_info(&instance->pdev->dev, "%s:%d killing adapter scsi%d"
+ " disableOnlineCtrlReset %d fw_outstanding %d \n",
+ __func__, __LINE__, instance->host->host_no, instance->disableOnlineCtrlReset,
+ atomic_read(&instance->fw_outstanding));
+ megasas_dump_pending_frames(instance);
+ megaraid_sas_kill_hba(instance);
- return SUCCESS;
+ return FAILED;
}
/**
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 13/15] megaraid_sas: Introduce module parameter for SCSI command-timeout
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (11 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 12/15] megaraid_sas: MFI adapter's OCR changes Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 14/15] megaraid_sas: SPERC OCR changes Sumit Saxena
` (2 subsequent siblings)
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
This patch will introduce module-parameter for SCSI command timeout value and fix setting of resetwaitime beyond a value.
There was some comment from Tomas on last time sent patch regarding why should driver provided scmd_timeout module parameter.
This is done for special cases where user wants to tune scmd_timeout value during driver load time. Default value is same as before-90 secs.
There are no changes in this patch on top of last time sent patch.
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
---
drivers/scsi/megaraid/megaraid_sas_base.c | 15 ++++++++++++---
drivers/scsi/megaraid/megaraid_sas_fusion.c | 2 +-
2 files changed, 13 insertions(+), 4 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index d2ea977..54922e5 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -83,7 +83,7 @@ module_param(throttlequeuedepth, int, S_IRUGO);
MODULE_PARM_DESC(throttlequeuedepth,
"Adapter queue depth when throttled due to I/O timeout. Default: 16");
-int resetwaittime = MEGASAS_RESET_WAIT_TIME;
+unsigned int resetwaittime = MEGASAS_RESET_WAIT_TIME;
module_param(resetwaittime, int, S_IRUGO);
MODULE_PARM_DESC(resetwaittime, "Wait time in seconds after I/O timeout "
"before resetting adapter. Default: 180");
@@ -100,6 +100,10 @@ unsigned int dual_qdepth_disable;
module_param(dual_qdepth_disable, int, S_IRUGO);
MODULE_PARM_DESC(dual_qdepth_disable, "Disable dual queue depth feature. Default: 0");
+unsigned int scmd_timeout = MEGASAS_DEFAULT_CMD_TIMEOUT;
+module_param(scmd_timeout, int, S_IRUGO);
+MODULE_PARM_DESC(scmd_timeout, "scsi command timeout (10-90s), default 90s. See megasas_reset_timer.");
+
MODULE_LICENSE("GPL");
MODULE_VERSION(MEGASAS_VERSION);
MODULE_AUTHOR("megaraidlinux.pdl@avagotech.com");
@@ -1850,7 +1854,7 @@ static int megasas_slave_configure(struct scsi_device *sdev)
* The RAID firmware may require extended timeouts.
*/
blk_queue_rq_timeout(sdev->request_queue,
- MEGASAS_DEFAULT_CMD_TIMEOUT * HZ);
+ scmd_timeout * HZ);
return 0;
}
@@ -2645,7 +2649,7 @@ blk_eh_timer_return megasas_reset_timer(struct scsi_cmnd *scmd)
unsigned long flags;
if (time_after(jiffies, scmd->jiffies_at_alloc +
- (MEGASAS_DEFAULT_CMD_TIMEOUT * 2) * HZ)) {
+ (scmd_timeout * 2) * HZ)) {
return BLK_EH_NOT_HANDLED;
}
@@ -5254,6 +5258,11 @@ static int megasas_init_fw(struct megasas_instance *instance)
instance->throttlequeuedepth =
MEGASAS_THROTTLE_QUEUE_DEPTH;
+ if (resetwaittime > MEGASAS_RESET_WAIT_TIME)
+ resetwaittime = MEGASAS_RESET_WAIT_TIME;
+
+ if ((scmd_timeout < 10) || (scmd_timeout > MEGASAS_DEFAULT_CMD_TIMEOUT))
+ scmd_timeout = MEGASAS_DEFAULT_CMD_TIMEOUT;
/* Launch SR-IOV heartbeat timer */
if (instance->requestorId) {
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index 64926f7..e740e26 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -91,7 +91,7 @@ void megasas_start_timer(struct megasas_instance *instance,
struct timer_list *timer,
void *fn, unsigned long interval);
extern struct megasas_mgmt_info megasas_mgmt_info;
-extern int resetwaittime;
+extern unsigned int resetwaittime;
extern unsigned int dual_qdepth_disable;
static void megasas_free_rdpq_fusion(struct megasas_instance *instance);
static void megasas_free_reply_fusion(struct megasas_instance *instance);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 14/15] megaraid_sas: SPERC OCR changes
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (12 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 13/15] megaraid_sas: Introduce module parameter for SCSI command-timeout Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-28 15:34 ` [PATCH v2 15/15] megaraid_sas: driver version upgrade Sumit Saxena
2016-01-29 2:54 ` [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Martin K. Petersen
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena, Uday Lingala
This patch will do some fixes in OCR path of SRIOV enabled series of Avago controllers.
1)Removing late detection HB.
2)Change in the behavior if the FW found in READY/OPERAETIONAL state.
There are no changes in this patch from last time sent patch.
Signed-off-by: Uday Lingala <uday.lingala@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
---
drivers/scsi/megaraid/megaraid_sas_fusion.c | 73 +++++++----------------------
1 file changed, 16 insertions(+), 57 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c
index e740e26..be9c3f1 100644
--- a/drivers/scsi/megaraid/megaraid_sas_fusion.c
+++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c
@@ -3462,52 +3462,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
/* Let SR-IOV VF & PF sync up if there was a HB failure */
if (instance->requestorId && !reason) {
msleep(MEGASAS_OCR_SETTLE_TIME_VF);
- /* Look for a late HB update after VF settle time */
- if (abs_state == MFI_STATE_OPERATIONAL &&
- (instance->hb_host_mem->HB.fwCounter !=
- instance->hb_host_mem->HB.driverCounter)) {
- instance->hb_host_mem->HB.driverCounter =
- instance->hb_host_mem->HB.fwCounter;
- dev_warn(&instance->pdev->dev, "SR-IOV:"
- "Late FW heartbeat update for "
- "scsi%d.\n",
- instance->host->host_no);
- } else {
- /* In VF mode, first poll for FW ready */
- for (i = 0;
- i < (MEGASAS_RESET_WAIT_TIME * 1000);
- i += 20) {
- status_reg =
- instance->instancet->
- read_fw_status_reg(
- instance->reg_set);
- abs_state = status_reg &
- MFI_STATE_MASK;
- if (abs_state == MFI_STATE_READY) {
- dev_warn(&instance->pdev->dev,
- "SR-IOV: FW was found"
- "to be in ready state "
- "for scsi%d.\n",
- instance->host->host_no);
- break;
- }
- msleep(20);
- }
- if (abs_state != MFI_STATE_READY) {
- dev_warn(&instance->pdev->dev, "SR-IOV: "
- "FW not in ready state after %d"
- " seconds for scsi%d, status_reg = "
- "0x%x.\n",
- MEGASAS_RESET_WAIT_TIME,
- instance->host->host_no,
- status_reg);
- megaraid_sas_kill_hba(instance);
- instance->skip_heartbeat_timer_del = 1;
- atomic_set(&instance->adprecovery, MEGASAS_HW_CRITICAL_ERROR);
- retval = FAILED;
- goto out;
- }
- }
+ goto transition_to_ready;
}
/* Now try to reset the chip */
@@ -3516,25 +3471,28 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
if (instance->instancet->adp_reset
(instance, instance->reg_set))
continue;
-
+transition_to_ready:
/* Wait for FW to become ready */
if (megasas_transition_to_ready(instance, 1)) {
- dev_warn(&instance->pdev->dev, "Failed to "
- "transition controller to ready "
- "for scsi%d.\n",
- instance->host->host_no);
- continue;
+ dev_warn(&instance->pdev->dev,
+ "Failed to transition controller to ready for "
+ "scsi%d.\n", instance->host->host_no);
+ if (instance->requestorId && !reason)
+ goto fail_kill_adapter;
+ else
+ continue;
}
-
megasas_reset_reply_desc(instance);
megasas_fusion_update_can_queue(instance, OCR_CONTEXT);
if (megasas_ioc_init_fusion(instance)) {
dev_warn(&instance->pdev->dev,
- "megasas_ioc_init_fusion() failed!"
- " for scsi%d\n",
- instance->host->host_no);
- continue;
+ "megasas_ioc_init_fusion() failed! for "
+ "scsi%d\n", instance->host->host_no);
+ if (instance->requestorId && !reason)
+ goto fail_kill_adapter;
+ else
+ continue;
}
megasas_refire_mgmt_cmd(instance);
@@ -3591,6 +3549,7 @@ int megasas_reset_fusion(struct Scsi_Host *shost, int reason)
retval = SUCCESS;
goto out;
}
+fail_kill_adapter:
/* Reset failed, kill the adapter */
dev_warn(&instance->pdev->dev, "Reset failed, killing "
"adapter scsi%d.\n", instance->host->host_no);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [PATCH v2 15/15] megaraid_sas: driver version upgrade
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (13 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 14/15] megaraid_sas: SPERC OCR changes Sumit Saxena
@ 2016-01-28 15:34 ` Sumit Saxena
2016-01-29 2:54 ` [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Martin K. Petersen
15 siblings, 0 replies; 20+ messages in thread
From: Sumit Saxena @ 2016-01-28 15:34 UTC (permalink / raw)
To: jbottomley, hch, martin.petersen, thenzl
Cc: linux-scsi, kashyap.desai, sumit.saxena
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
---
drivers/scsi/megaraid/megaraid_sas.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index 3e92f20..b6fdb48 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -35,8 +35,8 @@
/*
* MegaRAID SAS Driver meta data
*/
-#define MEGASAS_VERSION "06.808.16.00-rc1"
-#define MEGASAS_RELDATE "Oct. 8, 2015"
+#define MEGASAS_VERSION "06.810.09.00-rc1"
+#define MEGASAS_RELDATE "Jan. 28, 2016"
/*
* Device IDs
--
1.8.3.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* Re: [PATCH v2 00/15] megaraid_sas: Updates for scsi-next
2016-01-28 15:34 [PATCH v2 00/15] megaraid_sas: Updates for scsi-next Sumit Saxena
` (14 preceding siblings ...)
2016-01-28 15:34 ` [PATCH v2 15/15] megaraid_sas: driver version upgrade Sumit Saxena
@ 2016-01-29 2:54 ` Martin K. Petersen
15 siblings, 0 replies; 20+ messages in thread
From: Martin K. Petersen @ 2016-01-29 2:54 UTC (permalink / raw)
To: Sumit Saxena
Cc: jbottomley, hch, martin.petersen, thenzl, linux-scsi,
kashyap.desai
>>>>> "Sumit" == Sumit Saxena <sumit.saxena@avagotech.com> writes:
Sumit,
Applied to 4.6/scsi-queue.
Please make sure you put patch revision and feedback comments under the
--- separator so they don't become part of the commit message.
--
Martin K. Petersen Oracle Linux Engineering
^ permalink raw reply [flat|nested] 20+ messages in thread