[SCSI] mpt2sas: WarpDrive Infinite command retries due to wrong scsi command entry...
authorKashyap, Desai <kashyap.desai@lsi.com>
Tue, 5 Jul 2011 07:10:23 +0000 (12:40 +0530)
committerJames Bottomley <JBottomley@Parallels.com>
Tue, 26 Jul 2011 09:56:27 +0000 (13:56 +0400)
Issue:

This issue is seen on LSI H/W WarpDrive SSS6200 When filed direct I/O
is tried as volume I/O the scmd field in internal lookup table get
cleared and because of that the retried volume I/O never gets reported
as completed to SML.

Result:

I/O timeout and Error handling thread will kicking off

Fix:

Setting back the scmd in the lookup table before retrying the failed
direct i/o

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
drivers/scsi/mpt2sas/mpt2sas_scsih.c

index 939f283d0c28c3e83fa6d5cfc3e2bab00505a0f3..6abd2fcc43e2dba024ff28c8e7f993b2c82c3d28 100644 (file)
@@ -4258,6 +4258,7 @@ _scsih_io_done(struct MPT2SAS_ADAPTER *ioc, u16 smid, u8 msix_index, u32 reply)
        u32 log_info;
        struct MPT2SAS_DEVICE *sas_device_priv_data;
        u32 response_code = 0;
+       unsigned long flags;
 
        mpi_reply = mpt2sas_base_get_reply_virt_addr(ioc, reply);
        scmd = _scsih_scsi_lookup_get_clear(ioc, smid);
@@ -4282,6 +4283,9 @@ _scsih_io_done(struct MPT2SAS_ADAPTER *ioc, u16 smid, u8 msix_index, u32 reply)
         * the failed direct I/O should be redirected to volume
         */
        if (_scsih_scsi_direct_io_get(ioc, smid)) {
+               spin_lock_irqsave(&ioc->scsi_lookup_lock, flags);
+               ioc->scsi_lookup[smid - 1].scmd = scmd;
+               spin_unlock_irqrestore(&ioc->scsi_lookup_lock, flags);
                _scsih_scsi_direct_io_set(ioc, smid, 0);
                memcpy(mpi_request->CDB.CDB32, scmd->cmnd, scmd->cmd_len);
                mpi_request->DevHandle =