[PATCH 5/8][SCSI] mpt2sas: Infinite loop can occur if MPI2_IOCSTATUS_CONFIG_INVALID_PAGE is not returned

[PATCH 5/8][SCSI] mpt2sas: Infinite loop can occur if MPI2_IOCSTATUS_CONFIG_INVALID_PAGE is not returned

[Sreekanth Reddy]

Infinite loop can occur if IOCStatus is not equal to
MPI2_IOCSTATUS_CONFIG_INVALID_PAGE value in the while loops in functions
_scsih_search_responding_sas_devices,
_scsih_search_responding_raid_devices and
_scsih_search_responding_expanders

So, Instead of checking for MPI2_IOCSTATUS_CONFIG_INVALID_PAGE value,
in this patch code is modified to check for IOCStatus not equals to
MPI2_IOCSTATUS_SUCCESS to break the while loop.

Signed-off-by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com>
Cc: stable@vger.kernel.org
---
 drivers/scsi/mpt2sas/mpt2sas_scsih.c |   16 ++++------------
 1 files changed, 4 insertions(+), 12 deletions(-)

diff --git a/drivers/scsi/mpt2sas/mpt2sas_scsih.c b/drivers/scsi/mpt2sas/mpt2sas_scsih.c
index 6c86355..b7f8229 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_scsih.c
+++ b/drivers/scsi/mpt2sas/mpt2sas_scsih.c
@@ -6753,7 +6753,7 @@ _scsih_search_responding_sas_devices(struct MPT2SAS_ADAPTER *ioc)
 	    handle))) {
 		ioc_status = le16_to_cpu(mpi_reply.IOCStatus) &
 		    MPI2_IOCSTATUS_MASK;
-		if (ioc_status == MPI2_IOCSTATUS_CONFIG_INVALID_PAGE)
+		if (ioc_status != MPI2_IOCSTATUS_SUCCESS)
 			break;
 		handle = le16_to_cpu(sas_device_pg0.DevHandle);
 		device_info = le32_to_cpu(sas_device_pg0.DeviceInfo);
@@ -6862,7 +6862,7 @@ _scsih_search_responding_raid_devices(struct MPT2SAS_ADAPTER *ioc)
 	    &volume_pg1, MPI2_RAID_VOLUME_PGAD_FORM_GET_NEXT_HANDLE, handle))) {
 		ioc_status = le16_to_cpu(mpi_reply.IOCStatus) &
 		    MPI2_IOCSTATUS_MASK;
-		if (ioc_status == MPI2_IOCSTATUS_CONFIG_INVALID_PAGE)
+		if (ioc_status != MPI2_IOCSTATUS_SUCCESS)
 			break;
 		handle = le16_to_cpu(volume_pg1.DevHandle);
 
@@ -6887,7 +6887,7 @@ _scsih_search_responding_raid_devices(struct MPT2SAS_ADAPTER *ioc)
 		    phys_disk_num))) {
 			ioc_status = le16_to_cpu(mpi_reply.IOCStatus) &
 			    MPI2_IOCSTATUS_MASK;
-			if (ioc_status == MPI2_IOCSTATUS_CONFIG_INVALID_PAGE)
+			if (ioc_status != MPI2_IOCSTATUS_SUCCESS)
 				break;
 			phys_disk_num = pd_pg0.PhysDiskNum;
 			handle = le16_to_cpu(pd_pg0.DevHandle);
@@ -6967,7 +6967,7 @@ _scsih_search_responding_expanders(struct MPT2SAS_ADAPTER *ioc)
 
 		ioc_status = le16_to_cpu(mpi_reply.IOCStatus) &
 		    MPI2_IOCSTATUS_MASK;
-		if (ioc_status == MPI2_IOCSTATUS_CONFIG_INVALID_PAGE)
+		if (ioc_status != MPI2_IOCSTATUS_SUCCESS)
 			break;
 
 		handle = le16_to_cpu(expander_pg0.DevHandle);
@@ -7109,8 +7109,6 @@ _scsih_scan_for_devices_after_reset(struct MPT2SAS_ADAPTER *ioc)
 	    MPI2_SAS_EXPAND_PGAD_FORM_GET_NEXT_HNDL, handle))) {
 		ioc_status = le16_to_cpu(mpi_reply.IOCStatus) &
 		    MPI2_IOCSTATUS_MASK;
-		if (ioc_status == MPI2_IOCSTATUS_CONFIG_INVALID_PAGE)
-			break;
 		if (ioc_status != MPI2_IOCSTATUS_SUCCESS) {
 			printk(MPT2SAS_INFO_FMT "\tbreak from expander scan: "
 				"ioc_status(0x%04x), loginfo(0x%08x)\n",
@@ -7153,8 +7151,6 @@ _scsih_scan_for_devices_after_reset(struct MPT2SAS_ADAPTER *ioc)
 	    phys_disk_num))) {
 		ioc_status = le16_to_cpu(mpi_reply.IOCStatus) &
 		    MPI2_IOCSTATUS_MASK;
-		if (ioc_status == MPI2_IOCSTATUS_CONFIG_INVALID_PAGE)
-			break;
 		if (ioc_status != MPI2_IOCSTATUS_SUCCESS) {
 			printk(MPT2SAS_INFO_FMT "\tbreak from phys disk scan:"
 				"ioc_status(0x%04x), loginfo(0x%08x)\n",
@@ -7219,8 +7215,6 @@ _scsih_scan_for_devices_after_reset(struct MPT2SAS_ADAPTER *ioc)
 	    &volume_pg1, MPI2_RAID_VOLUME_PGAD_FORM_GET_NEXT_HANDLE, handle))) {
 		ioc_status = le16_to_cpu(mpi_reply.IOCStatus) &
 		    MPI2_IOCSTATUS_MASK;
-		if (ioc_status == MPI2_IOCSTATUS_CONFIG_INVALID_PAGE)
-			break;
 		if (ioc_status != MPI2_IOCSTATUS_SUCCESS) {
 			printk(MPT2SAS_INFO_FMT "\tbreak from volume scan: "
 				"ioc_status(0x%04x), loginfo(0x%08x)\n",
@@ -7278,8 +7272,6 @@ _scsih_scan_for_devices_after_reset(struct MPT2SAS_ADAPTER *ioc)
 	    handle))) {
 		ioc_status = le16_to_cpu(mpi_reply.IOCStatus) &
 		    MPI2_IOCSTATUS_MASK;
-		if (ioc_status == MPI2_IOCSTATUS_CONFIG_INVALID_PAGE)
-			break;
 		if (ioc_status != MPI2_IOCSTATUS_SUCCESS) {
 			printk(MPT2SAS_INFO_FMT "\tbreak from end device scan:"
 				" ioc_status(0x%04x), loginfo(0x%08x)\n",
-- 
1.7.3

Re: [PATCH 5/8][SCSI] mpt2sas: Infinite loop can occur if MPI2_IOCSTATUS_CONFIG_INVALID_PAGE is not returned

[James Bottomley]

On Thu, 2013-07-25 at 11:29 +0530, Sreekanth Reddy wrote:
> Infinite loop can occur if IOCStatus is not equal to
> MPI2_IOCSTATUS_CONFIG_INVALID_PAGE value in the while loops in functions
> _scsih_search_responding_sas_devices,
> _scsih_search_responding_raid_devices and
> _scsih_search_responding_expanders
> 
> So, Instead of checking for MPI2_IOCSTATUS_CONFIG_INVALID_PAGE value,
> in this patch code is modified to check for IOCStatus not equals to
> MPI2_IOCSTATUS_SUCCESS to break the while loop.
> 
> Signed-off-by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com>
> Cc: stable@vger.kernel.org

Patches like this aren't acceptable any more: Greg will find this and
reject it from Stable.

If there's a bug in the driver, it needs to be in a separate patch set
for the fixes branch of SCSI to go upstream now rather than waiting for
the merge window.  Only patches for the fixes branch can be tagged for
stable.

James

Re: [PATCH 5/8][SCSI] mpt2sas: Infinite loop can occur if MPI2_IOCSTATUS_CONFIG_INVALID_PAGE is not returned

[James Bottomley]

On Thu, 2013-07-25 at 07:00 -0700, James Bottomley wrote:
> On Thu, 2013-07-25 at 11:29 +0530, Sreekanth Reddy wrote:
> > Infinite loop can occur if IOCStatus is not equal to
> > MPI2_IOCSTATUS_CONFIG_INVALID_PAGE value in the while loops in functions
> > _scsih_search_responding_sas_devices,
> > _scsih_search_responding_raid_devices and
> > _scsih_search_responding_expanders
> > 
> > So, Instead of checking for MPI2_IOCSTATUS_CONFIG_INVALID_PAGE value,
> > in this patch code is modified to check for IOCStatus not equals to
> > MPI2_IOCSTATUS_SUCCESS to break the while loop.
> > 
> > Signed-off-by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com>
> > Cc: stable@vger.kernel.org
> 
> Patches like this aren't acceptable any more: Greg will find this and
> reject it from Stable.
> 
> If there's a bug in the driver, it needs to be in a separate patch set
> for the fixes branch of SCSI to go upstream now rather than waiting for
> the merge window.  Only patches for the fixes branch can be tagged for
> stable.

Ping on this, please.  What do you want me to do?  I can strip the cc:
stable from the series and send it to the misc branch or you can
resubmit as two series: one for fixes with the stable tags and the rest
for misc.

James

RE: [PATCH 5/8][SCSI] mpt2sas: Infinite loop can occur if MPI2_IOCSTATUS_CONFIG_INVALID_PAGE is not returned

[Reddy, Sreekanth]

Hi James,

Sorry  for the delay. 

we support the first method i.e. "strip the cc: stable from this patch and send it to the misc branch".

Thanks,
Sreekanth.

>-----Original Message-----
>From: James Bottomley [mailto:James.Bottomley@HansenPartnership.com]
>Sent: Monday, August 26, 2013 2:11 PM
>To: Reddy, Sreekanth
>Cc: linux-scsi; Prakash, Sathya; Nandigama, Nagalakshmi
>Subject: Re: [PATCH 5/8][SCSI] mpt2sas: Infinite loop can occur if
>MPI2_IOCSTATUS_CONFIG_INVALID_PAGE is not returned
>
>On Thu, 2013-07-25 at 07:00 -0700, James Bottomley wrote:
>> On Thu, 2013-07-25 at 11:29 +0530, Sreekanth Reddy wrote:
>> > Infinite loop can occur if IOCStatus is not equal to
>> > MPI2_IOCSTATUS_CONFIG_INVALID_PAGE value in the while loops in
>> > functions _scsih_search_responding_sas_devices,
>> > _scsih_search_responding_raid_devices and
>> > _scsih_search_responding_expanders
>> >
>> > So, Instead of checking for MPI2_IOCSTATUS_CONFIG_INVALID_PAGE
>> > value, in this patch code is modified to check for IOCStatus not
>> > equals to MPI2_IOCSTATUS_SUCCESS to break the while loop.
>> >
>> > Signed-off-by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com>
>> > Cc: stable@vger.kernel.org
>>
>> Patches like this aren't acceptable any more: Greg will find this and
>> reject it from Stable.
>>
>> If there's a bug in the driver, it needs to be in a separate patch set
>> for the fixes branch of SCSI to go upstream now rather than waiting
>> for the merge window.  Only patches for the fixes branch can be tagged
>> for stable.
>
>Ping on this, please.  What do you want me to do?  I can strip the cc:
>stable from the series and send it to the misc branch or you can resubmit as
>two series: one for fixes with the stable tags and the rest for misc.
>
>James
>
>