LSI Logic MPT driver mapping of scsi device busy to scsi host+device busy leads to read-only ext3 fs remounts on VMware ESX Server.

[Ed Goggin]

The function mptscsih_io_done() in drivers/message/fusion/mptscsih.c maps scsi device busy to scsi host+device busy for the case where the MPI IOC status is MPI_IOCSTATUS_SUCCESS. This action results in limiting Linux mid-layer SCSI retransmissions for this case to 5 without any delay between successive retransmissions (see DID_BUS_BUSY handling in scsi_decide_disposition in drivers/scsi/scsi_error.c). While this change was made in order to fail-over to Linux host resident MPIO code more quickly in the event of such scsi device busy failures, this also leads to scsi device busy I/O failures being reported to SCSI client code more readily in VMware ESX Server configurations. When such an I/O failure occurs on a SCSI WRITE command to a journal of an ext3 file system, the ext3 file system is remounted in read-only mode.

A fix for this issue is present in drivers/message/fusion/mptscsih.c:mptscsih_io_done() in upstream 2.6.22 code. I've included below the patch from Eric Moore from LSI Logic which contained the fix.

This address the issue of VMWare guest OS being remounted as read-only becuase the underlying device was held busy too long, , and at the same time address Engenio MPP driver concerns over infinite retries. This patch removes the code that snoops the SAM STATUS on busy, which would be returning DID_BUS_BUSY, instead we return the status as is. Retry hanlding seems to be properly handled in scsi_softirq_done, where a busy sam status would only occurr for the time specified by (cmd->allowed +1) * cmd->timeout_per_command.

Signed-off-by: Eric Moore <email address hidden>

diff -uarpN b/drivers/message/fusion/mptscsih.c a/drivers/message/fusion/mptscsih.c
--- b/drivers/message/fusion/mptscsih.c 2007-03-15 18:20:01.000000000 -0600
+++ a/drivers/message/fusion/mptscsih.c 2007-03-18 12:07:26.000000000 -0600
@@ -819,10 +819,7 @@ mptscsih_io_done(MPT_ADAPTER *ioc, MPT_F
    sc->resid=0;
   case MPI_IOCSTATUS_SCSI_RECOVERED_ERROR: /* 0x0040 */
   case MPI_IOCSTATUS_SUCCESS: /* 0x0000 */
- if (scsi_status == MPI_SCSI_STATUS_BUSY)
- sc->result = (DID_BUS_BUSY << 16) | scsi_status;
- else
- sc->result = (DID_OK << 16) | scsi_status;
+ sc->result = (DID_OK << 16) | scsi_status;
    if (scsi_state == 0) {
     ;
    } else if (scsi_state & MPI_SCSI_STATE_AUTOSENSE_VALID) {
-

[Leann Ogasawara (leannogasawara)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. This patch is in the current gutsy kernel release, 2.6.22-11.32. Thanks!

[Ed Goggin (egoggin)]

Leann,

Thanks for the prompt reply. That is good news. We are also seeking to have this patch installed into some maintenance release of Ubuntu-7.0.4. Is this possible?

Ed

[Leann Ogasawara (leannogasawara)]

Ed, I'm re-assigning this to the kernel team for their consideration to backport to 2.6.20. Just note that updates to a stable release are only done under certain circumstances as outlined here:

https://wiki.ubuntu.com/StableReleaseUpdates

Thanks.

[Leann Ogasawara (leannogasawara)]

It unfortunately looks like this never went in as an SRU for Feisty. Currently, the 18 month support period for Feisty Fawn 7.04 has reached it's end of life - http://www.ubuntu.com/news/ubuntu-7.04-end-of-life . As a result we are closing the linux-source-2.6.20 task. Thanks.

[Launchpad Janitor (janitor)]

Per a decision made by the Ubuntu Kernel Team, bugs will longer be assigned to the ubuntu-kernel-team in Launchpad as part of the bug triage process. The ubuntu-kernel-team is being unassigned from this bug report. Refer to https://wiki.ubuntu.com/KernelTeamBugPolicies for more information. Thanks.