Re: [PATCH] [SCSI] mpt2sas: fix for driver fails EEH recovery from injected pci bus error

[Tomas Henzl <thenzl@redhat.com>]

On 12/17/2012 10:58 PM, Sreekanth Reddy wrote:
> 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
 
diff --git a/drivers/scsi/mpt2sas/mpt2sas_base.c b/drivers/scsi/mpt2sas/mpt2sas_base.c
index fd3b3d7..480111c 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_base.c
+++ b/drivers/scsi/mpt2sas/mpt2sas_base.c
@@ -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);



>
> Signed-off-by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com>
> ---
>
> diff --git a/drivers/scsi/mpt2sas/mpt2sas_base.c b/drivers/scsi/mpt2sas/mpt2sas_base.c
> index ffd85c5..2349531 100755
> --- a/drivers/scsi/mpt2sas/mpt2sas_base.c
> +++ b/drivers/scsi/mpt2sas/mpt2sas_base.c
> @@ -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;
>  
> diff --git a/drivers/scsi/mpt2sas/mpt2sas_base.h b/drivers/scsi/mpt2sas/mpt2sas_base.h
> index 543d8d6..c6ee7aa 100755
> --- a/drivers/scsi/mpt2sas/mpt2sas_base.h
> +++ b/drivers/scsi/mpt2sas/mpt2sas_base.h
> @@ -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 majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html