scsi: core: Always send batch on reset or error handling command

commit 066c5b4 upstream.

In commit 8930a6c ("scsi: core: add support for request batching") the
block layer bd->last flag was mapped to SCMD_LAST and used as an indicator
to send the batch for the drivers that implement this feature. However, the
error handling code was not updated accordingly.

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 a result the device goes
offline after several failed resets. This was observed on virtio_scsi
during a 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: 8930a6c ("scsi: core: add support for request batching")
Cc: <stable@vger.kernel.org>
Signed-off-by: Alexander Atanasov <alexander.atanasov@virtuozzo.com>
Link: https://lore.kernel.org/r/20231215121008.2881653-1-alexander.atanasov@virtuozzo.com
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/scsi/scsi_error.c

@@ -1069,6 +1069,7 @@ static int scsi_send_eh_cmnd(struct scsi_cmnd *scmd, unsigned char *cmnd,
 
 	scsi_log_send(scmd);
 	scmd->scsi_done = scsi_eh_done;
+	scmd->flags |= SCMD_LAST;
 
 	/*
 	 * Lock sdev->state_mutex to avoid that scsi_device_quiesce() can
@@ -2361,6 +2362,7 @@ scsi_ioctl_reset(struct scsi_device *dev, int __user *arg)
 	scsi_init_command(dev, scmd);
 	scmd->request = rq;
 	scmd->cmnd = scsi_req(rq)->cmd;
+	scmd->flags |= SCMD_LAST;
 
 	scmd->scsi_done		= scsi_reset_provider_done_command;
 	memset(&scmd->sdb, 0, sizeof(scmd->sdb));