From: Eric Moore <eric.moore@lsi.com>
To: linux-scsi@vger.kernel.org, James.Bottomley@SteelEye.com,
DL-MPTFusionLinux@lsi.com
Subject: [PATCH 2/2] mpt fusion: fix for mounted raid volume filesytem that goes read-only
Date: Fri, 15 Jun 2007 17:27:21 -0600 [thread overview]
Message-ID: <20070615232721.GA12851@lsil.com> (raw)
If there is IO going to the volume while a hidden disk is being torn down,
there is a case where we would return a DID_NO_CONNECT for IO sent to the
volume. The end result is the volume goes read-only. This problem is due to the
fact the firmware mapped target ids saved in per device object is phys_disk_num
for hidden raid components, and target_id for the volume. There is a single
case when both phys_disk_num and target_id are equal, so enters this issue. We
fix this issue by checking the tflags when the device is torned down,
insuring the IO being completed is meant for hidden raid component, not the
volume. In addition to this fix, there are a couple other cases to address
hidden raid components. For instance task_abort and device reset are not
supported by mpt fw for hidden raid components, a bus reset would be required
or target reset to volume.
Signed-off-by: Eric Moore <Eric.Moore@lsi.com>
diff -uarpN b/drivers/message/fusion/mptscsih.c a/drivers/message/fusion/mptscsih.c
--- b/drivers/message/fusion/mptscsih.c 2007-06-15 17:05:12.000000000 -0600
+++ a/drivers/message/fusion/mptscsih.c 2007-06-15 17:16:48.000000000 -0600
@@ -447,7 +447,12 @@ mptscsih_issue_sep_command(MPT_ADAPTER *
MPT_FRAME_HDR *mf;
SEPRequest_t *SEPMsg;
- if (ioc->bus_type == FC)
+ if (ioc->bus_type != SAS)
+ return;
+
+ /* Not supported for hidden raid components
+ */
+ if (vtarget->tflags & MPT_TARGET_FLAGS_RAID_COMPONENT)
return;
if ((mf = mpt_get_msg_frame(ioc->InternalCtx, ioc)) == NULL) {
@@ -991,14 +996,19 @@ mptscsih_search_running_cmds(MPT_SCSI_HO
mf = (SCSIIORequest_t *)MPT_INDEX_2_MFPTR(hd->ioc, ii);
if (mf == NULL)
continue;
+ /* If the device is a hidden raid component, then its
+ * expected that the mf->function will be RAID_SCSI_IO
+ */
+ if (vdevice->vtarget->tflags &
+ MPT_TARGET_FLAGS_RAID_COMPONENT && mf->Function !=
+ MPI_FUNCTION_RAID_SCSI_IO_PASSTHROUGH)
+ continue;
+
int_to_scsilun(vdevice->lun, &lun);
if ((mf->Bus != vdevice->vtarget->channel) ||
(mf->TargetID != vdevice->vtarget->id) ||
memcmp(lun.scsi_lun, mf->LUN, 8))
continue;
- dsprintk(( "search_running: found (sc=%p, mf = %p) "
- "channel %d id %d, lun %d \n", hd->ScsiLookup[ii],
- mf, mf->Bus, mf->TargetID, vdevice->lun));
/* Cleanup
*/
@@ -1008,9 +1018,11 @@ mptscsih_search_running_cmds(MPT_SCSI_HO
if ((unsigned char *)mf != sc->host_scribble)
continue;
scsi_dma_unmap(sc);
-
sc->host_scribble = NULL;
sc->result = DID_NO_CONNECT << 16;
+ dsprintk(( "search_running: found (sc=%p, mf = %p) "
+ "channel %d id %d, lun %d \n", sc, mf,
+ vdevice->vtarget->channel, vdevice->vtarget->id, vdevice->lun));
sc->scsi_done(sc);
}
}
@@ -1756,6 +1768,16 @@ mptscsih_abort(struct scsi_cmnd * SCpnt)
goto out;
}
+ /* Task aborts are not supported for hidden raid components.
+ */
+ if (vdevice->vtarget->tflags & MPT_TARGET_FLAGS_RAID_COMPONENT) {
+ dtmprintk((MYIOC_s_DEBUG_FMT "task abort: hidden raid "
+ "component (sc=%p)\n", ioc->name, SCpnt));
+ SCpnt->result = DID_RESET << 16;
+ retval = FAILED;
+ goto out;
+ }
+
/* Find this command
*/
if ((scpnt_idx = SCPNT_TO_LOOKUP_IDX(SCpnt)) < 0) {
@@ -1849,6 +1871,13 @@ mptscsih_dev_reset(struct scsi_cmnd * SC
goto out;
}
+ /* Target reset to hidden raid component is not supported
+ */
+ if (vdevice->vtarget->tflags & MPT_TARGET_FLAGS_RAID_COMPONENT) {
+ retval = FAILED;
+ goto out;
+ }
+
retval = mptscsih_TMHandler(hd, MPI_SCSITASKMGMT_TASKTYPE_TARGET_RESET,
vdevice->vtarget->channel, vdevice->vtarget->id, 0, 0,
mptscsih_get_tm_timeout(ioc));
@@ -3131,6 +3160,16 @@ mptscsih_synchronize_cache(MPT_SCSI_HOST
{
INTERNAL_CMD iocmd;
+ /* Ignore hidden raid components, this is handled when the command
+ * is sent to the volume
+ */
+ if (vdevice->vtarget->tflags & MPT_TARGET_FLAGS_RAID_COMPONENT)
+ return;
+
+ if (vdevice->vtarget->type != TYPE_DISK || vdevice->vtarget->deleted ||
+ !vdevice->configured_lun)
+ return;
+
/* Following parameters will not change
* in this routine.
*/
@@ -3145,9 +3184,7 @@ mptscsih_synchronize_cache(MPT_SCSI_HOST
iocmd.id = vdevice->vtarget->id;
iocmd.lun = vdevice->lun;
- if ((vdevice->vtarget->type == TYPE_DISK) &&
- (vdevice->configured_lun))
- mptscsih_do_cmd(hd, &iocmd);
+ mptscsih_do_cmd(hd, &iocmd);
}
EXPORT_SYMBOL(mptscsih_remove);
reply other threads:[~2007-06-15 23:31 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20070615232721.GA12851@lsil.com \
--to=eric.moore@lsi.com \
--cc=DL-MPTFusionLinux@lsi.com \
--cc=James.Bottomley@SteelEye.com \
--cc=linux-scsi@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.