I stumbled across this while testing surprise device removal, particularly
during device initialization.
The test added an LSI Logic / Symbios Logic SAS2008 PCI-Express Fusion-MPT
SAS-2 with a few disks attached. During an async SCSI scan, the device is
removed via PCI hotplug. Cleanup of the scsi host took a long time
(roughly five minutes) which I traced back to the timeout value in the
mpt2sas _scsih_scan_finished() function.
Since the driver has already set the adapter's "remove_host" value in its
.remove callback, it seems like the driver should use that information to
give up on the scan and cleanup a bit faster.
I don't have mpt3sas HW, so that part of the patch is untested (though the
code looks 99% the same here.)
Regards,
-- Joe
>From 9900a0a30464ef1fff67f28780ec036d3d7d0cbe Mon Sep 17 00:00:00 2001
From: Joe Lawrence <joe.lawrence@xxxxxxxxxxx>
Date: Wed, 13 Feb 2013 16:30:31 -0500
Subject: [PATCH] mpt2sas/mpt3sas: set async scan finished on adapter removal
If a scsi host is removed during an asynchronous SCSI scan, the driver's
.scan_finished callback function may take up to 5 minutes to timeout and
notify the SCSI layer that the operation has failed. Once the driver
determines that an adapter is gone, stop waiting for the scan to complete.
Signed-off-by: Joe Lawrence <joe.lawrence@xxxxxxxxxxx>
---
drivers/scsi/mpt2sas/mpt2sas_scsih.c | 10 ++++++++--
drivers/scsi/mpt3sas/mpt3sas_scsih.c | 12 +++++++++---
2 files changed, 17 insertions(+), 5 deletions(-)
diff --git a/drivers/scsi/mpt2sas/mpt2sas_scsih.c b/drivers/scsi/mpt2sas/mpt2sas_scsih.c
index c6bdc92..d1e24e7 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_scsih.c
+++ b/drivers/scsi/mpt2sas/mpt2sas_scsih.c
@@ -7951,14 +7951,20 @@ _scsih_scan_start(struct Scsi_Host *shost)
* @time: elapsed time of the scan in jiffies
*
* This function will be called periodically until it returns 1 with the
- * scsi_host and the elapsed time of the scan in jiffies. In our implemention,
- * we wait for firmware discovery to complete, then return 1.
+ * scsi_host and the elapsed time of the scan in jiffies. In our
+ * implementation, we wait for firmware discovery to complete, then return 1.
*/
static int
_scsih_scan_finished(struct Scsi_Host *shost, unsigned long time)
{
struct MPT2SAS_ADAPTER *ioc = shost_priv(shost);
+ if (ioc->remove_host) {
+ printk(MPT2SAS_INFO_FMT "port enable: FAILED host removed\n",
+ ioc->name);
+ return 1;
+ }
+
if (disable_discovery > 0) {
ioc->is_driver_loading = 0;
ioc->wait_for_discovery_to_complete = 0;
diff --git a/drivers/scsi/mpt3sas/mpt3sas_scsih.c b/drivers/scsi/mpt3sas/mpt3sas_scsih.c
index 6421a06..170f485 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_scsih.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_scsih.c
@@ -7665,15 +7665,21 @@ _scsih_scan_start(struct Scsi_Host *shost)
* @shost: SCSI host pointer
* @time: elapsed time of the scan in jiffies
*
- * This function will be called periodicallyn until it returns 1 with the
- * scsi_host and the elapsed time of the scan in jiffies. In our implemention,
- * we wait for firmware discovery to complete, then return 1.
+ * This function will be called periodically until it returns 1 with the
+ * scsi_host and the elapsed time of the scan in jiffies. In our
+ * implementation, we wait for firmware discovery to complete, then return 1.
*/
static int
_scsih_scan_finished(struct Scsi_Host *shost, unsigned long time)
{
struct MPT3SAS_ADAPTER *ioc = shost_priv(shost);
+ if (ioc->remove_host) {
+ pr_info(MPT3SAS_FMT "port enable: FAILED host removed\n",
+ ioc->name);
+ return 1;
+ }
+
if (disable_discovery > 0) {
ioc->is_driver_loading = 0;
ioc->wait_for_discovery_to_complete = 0;
--
1.8.1.2
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
