diff mbox

megaraid_sas: Don't issue kill adapter for MFI controllers in case of PD list DCMD failure

Message ID 1457604877-5280-1-git-send-email-sumit.saxena@broadcom.com (mailing list archive)
State Accepted, archived
Headers show

Commit Message

Sumit Saxena March 10, 2016, 10:14 a.m. UTC
There are few MFI adapters which do not support MR_DCMD_PD_LIST_QUERY so
if MFI adapters fail this DCMD, it should not be considered as FATAL and
driver should not issue kill adapter and set per controller's instance
variable- pd_list_not_supported so that same variable can be used inside
functions- slave_alloc and slave_configure to allow firmware scan.

Killing adapter because of DCMD failure when this DCMD is not supported
causes driver's probe getting failed. This issue got introduced because
of below commit when MFI IO timeout handling was introduced-

6d40afb megaraid_sas: MFI IO timeout handling

Killing adapter in case of this DCMD failure should be limited to Fusion
adapters only. Per controller's instance variable allow_fw_scan is removed
as pd_list_not_supported better reflect the purpose.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
---
 drivers/scsi/megaraid/megaraid_sas.h      |  2 +-
 drivers/scsi/megaraid/megaraid_sas_base.c | 14 ++++++++++----
 2 files changed, 11 insertions(+), 5 deletions(-)

Comments

Johannes Thumshirn March 10, 2016, 10:40 a.m. UTC | #1
On Thu, Mar 10, 2016 at 02:14:37AM -0800, Sumit Saxena wrote:
> There are few MFI adapters which do not support MR_DCMD_PD_LIST_QUERY so
> if MFI adapters fail this DCMD, it should not be considered as FATAL and
> driver should not issue kill adapter and set per controller's instance
> variable- pd_list_not_supported so that same variable can be used inside
> functions- slave_alloc and slave_configure to allow firmware scan.
> 
> Killing adapter because of DCMD failure when this DCMD is not supported
> causes driver's probe getting failed. This issue got introduced because
> of below commit when MFI IO timeout handling was introduced-
> 
> 6d40afb megaraid_sas: MFI IO timeout handling
> 
> Killing adapter in case of this DCMD failure should be limited to Fusion
> adapters only. Per controller's instance variable allow_fw_scan is removed
> as pd_list_not_supported better reflect the purpose.
> 
> Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>

Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Hannes Reinecke March 10, 2016, 1:06 p.m. UTC | #2
On 03/10/2016 11:14 AM, Sumit Saxena wrote:
> There are few MFI adapters which do not support MR_DCMD_PD_LIST_QUERY so
> if MFI adapters fail this DCMD, it should not be considered as FATAL and
> driver should not issue kill adapter and set per controller's instance
> variable- pd_list_not_supported so that same variable can be used inside
> functions- slave_alloc and slave_configure to allow firmware scan.
> 
> Killing adapter because of DCMD failure when this DCMD is not supported
> causes driver's probe getting failed. This issue got introduced because
> of below commit when MFI IO timeout handling was introduced-
> 
> 6d40afb megaraid_sas: MFI IO timeout handling
> 
> Killing adapter in case of this DCMD failure should be limited to Fusion
> adapters only. Per controller's instance variable allow_fw_scan is removed
> as pd_list_not_supported better reflect the purpose.
> 
> Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
> ---
>  drivers/scsi/megaraid/megaraid_sas.h      |  2 +-
>  drivers/scsi/megaraid/megaraid_sas_base.c | 14 ++++++++++----
>  2 files changed, 11 insertions(+), 5 deletions(-)
> 
Took you a while to figure that one out :-)
This was actually the main issue why I started with the DCMD timeout
stuff.

Reviewed-by: Hannes Reinecke <hare@suse.com>

Cheers,

Hannes
Ewan Milne March 10, 2016, 3:28 p.m. UTC | #3
On Thu, 2016-03-10 at 02:14 -0800, Sumit Saxena wrote:
> There are few MFI adapters which do not support MR_DCMD_PD_LIST_QUERY so
> if MFI adapters fail this DCMD, it should not be considered as FATAL and
> driver should not issue kill adapter and set per controller's instance
> variable- pd_list_not_supported so that same variable can be used inside
> functions- slave_alloc and slave_configure to allow firmware scan.
> 
> Killing adapter because of DCMD failure when this DCMD is not supported
> causes driver's probe getting failed. This issue got introduced because
> of below commit when MFI IO timeout handling was introduced-
> 
> 6d40afb megaraid_sas: MFI IO timeout handling
> 
> Killing adapter in case of this DCMD failure should be limited to Fusion
> adapters only. Per controller's instance variable allow_fw_scan is removed
> as pd_list_not_supported better reflect the purpose.
> 
> Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
> ---
>  drivers/scsi/megaraid/megaraid_sas.h      |  2 +-
>  drivers/scsi/megaraid/megaraid_sas_base.c | 14 ++++++++++----
>  2 files changed, 11 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
> index 4484e63..fce414a 100644
> --- a/drivers/scsi/megaraid/megaraid_sas.h
> +++ b/drivers/scsi/megaraid/megaraid_sas.h
> @@ -2097,7 +2097,7 @@ struct megasas_instance {
>  	u8 UnevenSpanSupport;
>  
>  	u8 supportmax256vd;
> -	u8 allow_fw_scan;
> +	u8 pd_list_not_supported;
>  	u16 fw_supported_vd_count;
>  	u16 fw_supported_pd_count;
>  
> diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
> index 5c08568..69d375b 100644
> --- a/drivers/scsi/megaraid/megaraid_sas_base.c
> +++ b/drivers/scsi/megaraid/megaraid_sas_base.c
> @@ -1838,7 +1838,7 @@ static int megasas_slave_configure(struct scsi_device *sdev)
>  	struct megasas_instance *instance;
>  
>  	instance = megasas_lookup_instance(sdev->host->host_no);
> -	if (instance->allow_fw_scan) {
> +	if (instance->pd_list_not_supported) {
>  		if (sdev->channel < MEGASAS_MAX_PD_CHANNELS &&
>  			sdev->type == TYPE_DISK) {
>  			pd_index = (sdev->channel * MEGASAS_MAX_DEV_PER_CHANNEL) +
> @@ -1874,7 +1874,8 @@ static int megasas_slave_alloc(struct scsi_device *sdev)
>  		pd_index =
>  			(sdev->channel * MEGASAS_MAX_DEV_PER_CHANNEL) +
>  			sdev->id;
> -		if ((instance->allow_fw_scan || instance->pd_list[pd_index].driveState ==
> +		if ((instance->pd_list_not_supported ||
> +			instance->pd_list[pd_index].driveState ==
>  			MR_PD_STATE_SYSTEM)) {
>  			goto scan_target;
>  		}
> @@ -4087,7 +4088,13 @@ megasas_get_pd_list(struct megasas_instance *instance)
>  
>  	switch (ret) {
>  	case DCMD_FAILED:
> -		megaraid_sas_kill_hba(instance);
> +		dev_info(&instance->pdev->dev, "MR_DCMD_PD_LIST_QUERY "
> +			"failed/not supported by firmware\n");
> +
> +		if (instance->ctrl_context)
> +			megaraid_sas_kill_hba(instance);
> +		else
> +			instance->pd_list_not_supported = 1;
>  		break;
>  	case DCMD_TIMEOUT:
>  
> @@ -5034,7 +5041,6 @@ static int megasas_init_fw(struct megasas_instance *instance)
>  	case PCI_DEVICE_ID_DELL_PERC5:
>  	default:
>  		instance->instancet = &megasas_instance_template_xscale;
> -		instance->allow_fw_scan = 1;
>  		break;
>  	}
>  

Note the comment "fixes 6d40afb megaraid_sas: MFI IO timeout handling"

Reviewed-by: Ewan D. Milne <emilne@redhat.com>


--
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
Tomas Henzl March 10, 2016, 9:35 p.m. UTC | #4
On 10.3.2016 11:14, Sumit Saxena wrote:
> There are few MFI adapters which do not support MR_DCMD_PD_LIST_QUERY so
> if MFI adapters fail this DCMD, it should not be considered as FATAL and
> driver should not issue kill adapter and set per controller's instance
> variable- pd_list_not_supported so that same variable can be used inside
> functions- slave_alloc and slave_configure to allow firmware scan.
>
> Killing adapter because of DCMD failure when this DCMD is not supported
> causes driver's probe getting failed. This issue got introduced because
> of below commit when MFI IO timeout handling was introduced-
>
> 6d40afb megaraid_sas: MFI IO timeout handling
>
> Killing adapter in case of this DCMD failure should be limited to Fusion
> adapters only. Per controller's instance variable allow_fw_scan is removed
> as pd_list_not_supported better reflect the purpose.
>
> Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>

Reviewed-by: Tomas Henzl <thenzl@redhat.com>

--
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
Martin K. Petersen March 11, 2016, 1:29 a.m. UTC | #5
>>>>> "Sumit" == Sumit Saxena <sumit.saxena@broadcom.com> writes:

Sumit> There are few MFI adapters which do not support
Sumit> MR_DCMD_PD_LIST_QUERY so if MFI adapters fail this DCMD, it
Sumit> should not be considered as FATAL and driver should not issue
Sumit> kill adapter and set per controller's instance variable-
Sumit> pd_list_not_supported so that same variable can be used inside
Sumit> functions- slave_alloc and slave_configure to allow firmware
Sumit> scan.

Applied to 4.6/scsi-queue.
diff mbox

Patch

diff --git a/drivers/scsi/megaraid/megaraid_sas.h b/drivers/scsi/megaraid/megaraid_sas.h
index 4484e63..fce414a 100644
--- a/drivers/scsi/megaraid/megaraid_sas.h
+++ b/drivers/scsi/megaraid/megaraid_sas.h
@@ -2097,7 +2097,7 @@  struct megasas_instance {
 	u8 UnevenSpanSupport;
 
 	u8 supportmax256vd;
-	u8 allow_fw_scan;
+	u8 pd_list_not_supported;
 	u16 fw_supported_vd_count;
 	u16 fw_supported_pd_count;
 
diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 5c08568..69d375b 100644
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -1838,7 +1838,7 @@  static int megasas_slave_configure(struct scsi_device *sdev)
 	struct megasas_instance *instance;
 
 	instance = megasas_lookup_instance(sdev->host->host_no);
-	if (instance->allow_fw_scan) {
+	if (instance->pd_list_not_supported) {
 		if (sdev->channel < MEGASAS_MAX_PD_CHANNELS &&
 			sdev->type == TYPE_DISK) {
 			pd_index = (sdev->channel * MEGASAS_MAX_DEV_PER_CHANNEL) +
@@ -1874,7 +1874,8 @@  static int megasas_slave_alloc(struct scsi_device *sdev)
 		pd_index =
 			(sdev->channel * MEGASAS_MAX_DEV_PER_CHANNEL) +
 			sdev->id;
-		if ((instance->allow_fw_scan || instance->pd_list[pd_index].driveState ==
+		if ((instance->pd_list_not_supported ||
+			instance->pd_list[pd_index].driveState ==
 			MR_PD_STATE_SYSTEM)) {
 			goto scan_target;
 		}
@@ -4087,7 +4088,13 @@  megasas_get_pd_list(struct megasas_instance *instance)
 
 	switch (ret) {
 	case DCMD_FAILED:
-		megaraid_sas_kill_hba(instance);
+		dev_info(&instance->pdev->dev, "MR_DCMD_PD_LIST_QUERY "
+			"failed/not supported by firmware\n");
+
+		if (instance->ctrl_context)
+			megaraid_sas_kill_hba(instance);
+		else
+			instance->pd_list_not_supported = 1;
 		break;
 	case DCMD_TIMEOUT:
 
@@ -5034,7 +5041,6 @@  static int megasas_init_fw(struct megasas_instance *instance)
 	case PCI_DEVICE_ID_DELL_PERC5:
 	default:
 		instance->instancet = &megasas_instance_template_xscale;
-		instance->allow_fw_scan = 1;
 		break;
 	}