mpt2sas: fix for driver fails EEH recovery from injected pci bus error

This patch stops the driver to invoke kthread (which remove the dead ioc) for some time while EEH recovery has started.

Signed-off-by Sreekanth Reddy <[email protected]>

---

drivers/scsi/mpt2sas/mpt2sas_base.c +16 -1

index ffd85c5..2349531 100755
@@ -155,7 +155,7 @@ _base_fault_reset_work(struct work_struct *work)
 	struct task_struct *p;
 
 	spin_lock_irqsave(&ioc->ioc_reset_in_progress_lock, flags);
-	if (ioc->shost_recovery)
+	if (ioc->shost_recovery || ioc->pci_error_recovery)
 		goto rearm_timer;
 	spin_unlock_irqrestore(&ioc->ioc_reset_in_progress_lock, flags);
 

@@ -164,6 +164,20 @@ _base_fault_reset_work(struct work_struct *work)
 		printk(MPT2SAS_INFO_FMT "%s : SAS host is non-operational !!!!\n",
 			ioc->name, __func__);
 
+		/* It may be possible that EEH recovery can resolve some of
+		 * pci bus failure issues rather removing the dead ioc function
+		 * by considering controller is in a non-operational state. So
+		 * here priority is given to the EEH recovery. If it doesn't
+		 * not resolve this issue, mpt2sas driver will consider this
+		 * controller to non-operational state and remove the dead ioc
+		 * function.
+		 */
+		if (ioc->non_operational_loop++ < 5) {
+			spin_lock_irqsave(&ioc->ioc_reset_in_progress_lock,
+							 flags);
+			goto rearm_timer;
+		}
+
 		/*
 		 * Call _scsih_flush_pending_cmds callback so that we flush all
 		 * pending commands back to OS. This call is required to aovid

@@ -4386,6 +4400,7 @@ mpt2sas_base_attach(struct MPT2SAS_ADAPTER *ioc)
 	if (missing_delay[0] != -1 && missing_delay[1] != -1)
 		_base_update_missing_delay(ioc, missing_delay[0],
 		    missing_delay[1]);
+	ioc->non_operational_loop = 0;
 
 	return 0;
 

drivers/scsi/mpt2sas/mpt2sas_base.h +1 -1

index 543d8d6..c6ee7aa 100755
@@ -835,6 +835,7 @@ struct MPT2SAS_ADAPTER {
 	u16		cpu_msix_table_sz;
 	u32		ioc_reset_count;
 	MPT2SAS_FLUSH_RUNNING_CMDS schedule_dead_ioc_flush_running_cmds;
+	u32             non_operational_loop;
 
 	/* internal commands, callback index */
 	u8		scsi_io_cb_idx;
--

To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

This patch stops the driver to invoke kthread (which remove the dead ioc) for some time while EEH recovery has started.

Signed-off-by Sreekanth Reddy <[email protected]>

---

drivers/scsi/mpt2sas/mpt2sas_base.c +16 -2

index ffd85c5..2349531 100755
@@ -155,7 +155,7 @@ _base_fault_reset_work(struct work_struct *work)
 	struct task_struct *p;
 
 	spin_lock_irqsave(&ioc->ioc_reset_in_progress_lock, flags);
-	if (ioc->shost_recovery)
+	if (ioc->shost_recovery || ioc->pci_error_recovery)
 		goto rearm_timer;
 	spin_unlock_irqrestore(&ioc->ioc_reset_in_progress_lock, flags);
 

@@ -164,6 +164,20 @@ _base_fault_reset_work(struct work_struct *work)
 		printk(MPT2SAS_INFO_FMT "%s : SAS host is non-operational !!!!\n",
 			ioc->name, __func__);
 
+		/* It may be possible that EEH recovery can resolve some of
+		 * pci bus failure issues rather removing the dead ioc function
+		 * by considering controller is in a non-operational state. So
+		 * here priority is given to the EEH recovery. If it doesn't
+		 * not resolve this issue, mpt2sas driver will consider this
+		 * controller to non-operational state and remove the dead ioc
+		 * function.
+		 */
+		if (ioc->non_operational_loop++ < 5) {
+			spin_lock_irqsave(&ioc->ioc_reset_in_progress_lock,
+							 flags);
+			goto rearm_timer;
+		}
+
 		/*
 		 * Call _scsih_flush_pending_cmds callback so that we flush all
 		 * pending commands back to OS. This call is required to aovid @@ 
-4386,6 +4400,7 @@ mpt2sas_base_attach(struct MPT2SAS_ADAPTER *ioc)
 	if (missing_delay[0] != -1 && missing_delay[1] != -1)
 		_base_update_missing_delay(ioc, missing_delay[0],
 		    missing_delay[1]);
+	ioc->non_operational_loop = 0;
 
 	return 0;

drivers/scsi/mpt2sas/mpt2sas_base.h +1 -1

index 543d8d6..c6ee7aa 100755
@@ -835,6 +835,7 @@ struct MPT2SAS_ADAPTER {
 	u16		cpu_msix_table_sz;
 	u32		ioc_reset_count;
 	MPT2SAS_FLUSH_RUNNING_CMDS schedule_dead_ioc_flush_running_cmds;
+	u32             non_operational_loop;
 
 	/* internal commands, callback index */
 	u8		scsi_io_cb_idx;
--

To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

Yes Thomas, we need to reset the non_operational_loop to zero after the transient event.

Thanks,
Sreekanth.

-----Original Message-----
From: Tomas Henzl [mailto:[email protected]]
Sent: Monday, December 17, 2012 6:43 PM
To: Reddy, Sreekanth
Cc: [email protected]; Nandigama, Nagalakshmi; [email protected]; [email protected]; Prakash, Sathya
Subject: Re: [PATCH] [SCSI] mpt2sas: fix for driver fails EEH recovery from injected pci bus error

Quoted 2 lines · Sreekanth Reddy

This patch stops the driver to invoke kthread (which remove the dead ioc) for some time while EEH recovery has started.

Thank you for posting this, the issue we have seen is resolved now. Shouldn't be an additional initialization added?
So after a transient event the non_operational_loop is reset again?

Tomas

drivers/scsi/mpt2sas/mpt2sas_base.c +2 -0

index fd3b3d7..480111c 100644
@@ -208,6 +208,8 @@ _base_fault_reset_work(struct work_struct *work)
 		return; /* don't rearm timer */
 	}
 
+	ioc->non_operational_loop = 0;
+
 	if ((doorbell & MPI2_IOC_STATE_MASK) == MPI2_IOC_STATE_FAULT) {
 		rc = mpt2sas_base_hard_reset_handler(ioc, CAN_SLEEP,
 		    FORCE_BIG_HAMMER);

Quoted 63 lines

Signed-off-by Sreekanth Reddy <[email protected]>

---

drivers/scsi/mpt2sas/mpt2sas_base.c +16 -2

index ffd85c5..2349531 100755
@@ -155,7 +155,7 @@ _base_fault_reset_work(struct work_struct *work)
 	struct task_struct *p;
 
 	spin_lock_irqsave(&ioc->ioc_reset_in_progress_lock, flags);
-	if (ioc->shost_recovery)
+	if (ioc->shost_recovery || ioc->pci_error_recovery)
 		goto rearm_timer;
 	spin_unlock_irqrestore(&ioc->ioc_reset_in_progress_lock, flags);
 

@@ -164,6 +164,20 @@ _base_fault_reset_work(struct work_struct *work)
 		printk(MPT2SAS_INFO_FMT "%s : SAS host is non-operational !!!!\n",
 			ioc->name, __func__);
 
+		/* It may be possible that EEH recovery can resolve some of
+		 * pci bus failure issues rather removing the dead ioc function
+		 * by considering controller is in a non-operational state. So
+		 * here priority is given to the EEH recovery. If it doesn't
+		 * not resolve this issue, mpt2sas driver will consider this
+		 * controller to non-operational state and remove the dead ioc
+		 * function.
+		 */
+		if (ioc->non_operational_loop++ < 5) {
+			spin_lock_irqsave(&ioc->ioc_reset_in_progress_lock,
+							 flags);
+			goto rearm_timer;
+		}
+
 		/*
 		 * Call _scsih_flush_pending_cmds callback so that we flush all
 		 * pending commands back to OS. This call is required to aovid @@ 
-4386,6 +4400,7 @@ mpt2sas_base_attach(struct MPT2SAS_ADAPTER *ioc)
 	if (missing_delay[0] != -1 && missing_delay[1] != -1)
 		_base_update_missing_delay(ioc, missing_delay[0],
 		    missing_delay[1]);
+	ioc->non_operational_loop = 0;
 
 	return 0;

drivers/scsi/mpt2sas/mpt2sas_base.h +1 -1

index 543d8d6..c6ee7aa 100755
@@ -835,6 +835,7 @@ struct MPT2SAS_ADAPTER {
 	u16		cpu_msix_table_sz;
 	u32		ioc_reset_count;
 	MPT2SAS_FLUSH_RUNNING_CMDS schedule_dead_ioc_flush_running_cmds;
+	u32             non_operational_loop;
 
 	/* internal commands, callback index */
 	u8		scsi_io_cb_idx;
--

Quoted 3 lines

To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html