From: mchristi@redhat.com <mchristi@redhat.com>
Date: Thu, 2 Oct 2008 13:58:01 -0500
Subject: [scsi] scsi_error: retry cmd handling of transport error
Message-id: 1222973881-4441-2-git-send-email-mchristi@redhat.com
O-Subject: [PATCH 2/2] scsi_error: Fix retry command handling for transport errors
Bugzilla: 463206
RH-Acked-by: Doug Ledford <dledford@redhat.com>
RH-Acked-by: Tomas Henzl <thenzl@redhat.com>

From: Mike Christie <mchristi@redhat.com>

This is for BZ 463206.

Drivers want to be able to return DID_TRANSPORT_DISRUPTED and
have it do the right thing for commands like tape and passthrouh
as far as retries go. The LLDs previously used DID_BUS_BUSY or DID_ERROR
which followed the cmd->retries limit, but DID_TRANSPORT_DISRUPTED
was skipping that check so it could have caused a problem with tape
commands.

This patch has DID_TRANSPORT_DISRUPTED check the cmd->retries/cmd->allowed.
It also removes the target blocking behavior for now because it is not needed
as drivers are checking for this scenario and may change behavior in ways
out of tree drivers were not expecting.

I tested for regressions with iscsi by pulling cables and making sure
IO is still not fastfailed or is fast failed when dm-multipath is used
Then I tested with sg io to make sure that those commands are not retried.

It is not upstream. I just sent it the other day.

diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
index 341a82a..e6bc2de 100644
--- a/drivers/scsi/scsi_error.c
+++ b/drivers/scsi/scsi_error.c
@@ -1242,9 +1242,10 @@ int scsi_decide_disposition(struct scsi_cmnd *scmd)
 		 * LLD/transport was disrupted during processing of the IO.
 		 * The transport class is now blocked/blocking,
 		 * and the transport will decide what to do with the IO
-		 * based on its timers and recovery capablilities.
+		 * based on its timers and recovery capablilities if
+		 * cmd->retries permits.
 		 */
-		return SCSI_MLQUEUE_TARGET_BUSY2;
+		return SCSI_MLQUEUE_TARGET_COM_ERR;
 	case DID_TRANSPORT_FAILFAST:
 		/*
 		 * The transport decided to failfast the IO (most likely
diff --git a/include/scsi/scsi.h b/include/scsi/scsi.h
index 4dd674e..660cfc9 100644
--- a/include/scsi/scsi.h
+++ b/include/scsi/scsi.h
@@ -403,6 +403,8 @@ enum {
 	SCSI_MLQUEUE_IMM_RETRY		= (0x08 << SCSI_MLQUEUE_BUSY_SHIFT) |
 		SCSI_MLQUEUE_DIS_RETRY | SCSI_IGN_BLK_FAILFAST |
 		SCSI_IGN_ALLOWED,
+	SCSI_MLQUEUE_TARGET_COM_ERR	= (0x10 << SCSI_MLQUEUE_BUSY_SHIFT) |
+		SCSI_MLQUEUE_DIS_RETRY | SCSI_IGN_BLK_FAILFAST,
 };
 
 #define scsi_disposition_finish(dis) (dis & SCSI_MLQUEUE_DIS_FINISH)