In commit 8930a6c20791 ("scsi: core: add support for request batching")
blk-mq last flags was mapped to SCMD_LAST and used as an indicator to
send the batch for the drivers that implement it but the error handling
code was not updated.
scsi_send_eh_cmnd(...) is used to send error handling commands and
request sense. The problem is that request sense comes as a single
command that gets into the batch queue and times out. As result
device goes offline after several failed resets. This was observed
on virtio_scsi device resize operation.
[ 496.316946] sd 0:0:4:0: [sdd] tag#117 scsi_eh_0: requesting sense
[ 506.786356] sd 0:0:4:0: [sdd] tag#117 scsi_send_eh_cmnd timeleft: 0
[ 506.787981] sd 0:0:4:0: [sdd] tag#117 abort
To fix this always set SCMD_LAST flag in scsi_send_eh_cmnd and
scsi_reset_ioctl(...).
Fixes: 8930a6c20791 ("scsi: core: add support for request batching")
Signed-off-by: Alexander Atanasov <alexander.atanasov@xxxxxxxxxxxxx>
---
drivers/scsi/scsi_error.c | 2 ++
1 file changed, 2 insertions(+)
v1->v2: fix it globally not only for virtio_scsi, as suggested by
Paolo Bonzini, to avoid reintroducing the same bug.
diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
index c67cdcdc3ba8..1223d34c04da 100644
--- a/drivers/scsi/scsi_error.c
+++ b/drivers/scsi/scsi_error.c
@@ -1152,6 +1152,7 @@ static enum scsi_disposition scsi_send_eh_cmnd(struct scsi_cmnd *scmd,
scsi_log_send(scmd);
scmd->submitter = SUBMITTED_BY_SCSI_ERROR_HANDLER;
+ scmd->flags |= SCMD_LAST;
/*
* Lock sdev->state_mutex to avoid that scsi_device_quiesce() can
@@ -2459,6 +2460,7 @@ scsi_ioctl_reset(struct scsi_device *dev, int __user *arg)
scsi_init_command(dev, scmd);
scmd->submitter = SUBMITTED_BY_SCSI_RESET_IOCTL;
+ scmd->flags |= SCMD_LAST;
memset(&scmd->sdb, 0, sizeof(scmd->sdb));
scmd->cmd_len = 0;
--
2.39.3
