Skip to contentSkip to conversation

scsi: core: Cap initial sdev queue depth at shost.can_queue

PATCH In review John Garry 2 participants 6 messages Apr 19, 2021 – Apr 23, 2021 SCSI

lore
Share & tools
lore

John Garry

author
#0 lore raw

PATCH 1/1 scsi: core: Cap initial sdev queue depth at shost.can_queue

Function sdev_store_queue_depth() enforces that the sdev queue depth cannot exceed shost.can_queue.

However, the LLDD may still set cmd_per_lun > can_queue, which leads to an initial sdev queue depth greater than can_queue.

Stop this happened by capping initial sdev queue depth at can_queue.

---
Topic originally discussed at:
https://lore.kernel.org/linux-scsi/[email protected]/T/#m5663d0cac657d843b93d0c9a2374f98fc04384b9

Last idea there was to error/warn in scsi_add_host() for cmd_per_lun > can_queue. However, such a shost driver could still configure the sdev queue depth to be sound value at .slave_configure callback, so now thinking the orig patch better.

drivers/scsi/scsi_scan.c +5 -1
index 9f1b7f3c650a..8de2f830bcdc 100644
@@ -277,7 +277,11 @@ static struct scsi_device *scsi_alloc_sdev(struct scsi_target *starget,
 	WARN_ON_ONCE(!blk_get_queue(sdev->request_queue));
 	sdev->request_queue->queuedata = sdev;
 
-	depth = sdev->host->cmd_per_lun ?: 1;
+	if (sdev->host->cmd_per_lun)
+		depth = min_t(unsigned int, sdev->host->cmd_per_lun,
+			      sdev->host->can_queue);
+	else
+		depth = 1;
 
 	/*
 	 * Use .can_queue as budget map's depth because we have to
Signature · 1 line
2.26.2
#1 lore raw
Quoted 6 lines · John Garry

Function sdev_store_queue_depth() enforces that the sdev queue depth cannot exceed shost.can_queue.

However, the LLDD may still set cmd_per_lun > can_queue, which leads to an initial sdev queue depth greater than can_queue.

Quoted 8 lines

Stop this happened by capping initial sdev queue depth at can_queue.

---
Topic originally discussed at:
https://lore.kernel.org/linux-scsi/[email protected]/T/#m5663d0cac657d843b93d0c9a2374f98fc04384b9

Last idea there was to error/warn in scsi_add_host() for cmd_per_lun >

No, that isn't my suggestion.

Quoted 3 lines

can_queue. However, such a shost driver could still configure the sdev queue depth to be sound value at .slave_configure callback, so now thinking the orig patch better.

As I mentioned last time, why can't we fix ->cmd_per_lun in scsi_add_host() using .can_queue?

Signature · 1 line
Ming
author
#2 lore raw
Quoted 7 lines · Ming Lei
Quoted 6 lines · John Garry

Function sdev_store_queue_depth() enforces that the sdev queue depth cannot exceed shost.can_queue.

However, the LLDD may still set cmd_per_lun > can_queue, which leads to an initial sdev queue depth greater than can_queue.

Quoted 9 lines
Quoted 8 lines

Stop this happened by capping initial sdev queue depth at can_queue.

---
Topic originally discussed at:
https://lore.kernel.org/linux-scsi/[email protected]/T/#m5663d0cac657d843b93d0c9a2374f98fc04384b9

Last idea there was to error/warn in scsi_add_host() for cmd_per_lun >

Hi Ming,

Quoted 1 line

No, that isn't my suggestion.

Right, it was what I mentioned.

Quoted 8 lines
Quoted 4 lines

can_queue. However, such a shost driver could still configure the sdev queue depth to be sound value at .slave_configure callback, so now thinking the orig patch better.

As I mentioned last time, why can't we fix ->cmd_per_lun in scsi_add_host() using .can_queue?

I would rather not change the values which are provided from the driver. I would rather take the original values and try to use them in a sane way.

I have not seen other places where driver shost config values are modified by the core code.

Thanks,
John

#3 lore raw
Quoted 33 lines · John Garry
Quoted 7 lines · Ming Lei
Quoted 6 lines · John Garry

Function sdev_store_queue_depth() enforces that the sdev queue depth cannot exceed shost.can_queue.

However, the LLDD may still set cmd_per_lun > can_queue, which leads to an initial sdev queue depth greater than can_queue.

Quoted 9 lines
Quoted 8 lines

Stop this happened by capping initial sdev queue depth at can_queue.

---
Topic originally discussed at:
https://lore.kernel.org/linux-scsi/[email protected]/T/#m5663d0cac657d843b93d0c9a2374f98fc04384b9

Last idea there was to error/warn in scsi_add_host() for cmd_per_lun >

Hi Ming,

Quoted 1 line

No, that isn't my suggestion.

Right, it was what I mentioned.

Quoted 8 lines
Quoted 4 lines

can_queue. However, such a shost driver could still configure the sdev queue depth to be sound value at .slave_configure callback, so now thinking the orig patch better.

As I mentioned last time, why can't we fix ->cmd_per_lun in scsi_add_host() using .can_queue?

Quoted 5 lines

I would rather not change the values which are provided from the driver. I would rather take the original values and try to use them in a sane way.

I have not seen other places where driver shost config values are modified by the core code.

Wrt. .cmd_per_lun, I think it is safe to modify it into one correct depth because almost all drivers are just producer of .cmd_per_lun. And except for debug purpose, there are only three consumers of .cmd_per_lun in scsi, and all are for scsi_change_queue_depth():

process_message()
scsi_alloc_sdev()
virtscsi_change_queue_depth()

Thanks,
Ming

author
#4 lore raw
Quoted 5 lines · Ming Lei
Quoted 5 lines

I would rather not change the values which are provided from the driver. I would rather take the original values and try to use them in a sane way.

I have not seen other places where driver shost config values are modified by the core code.

Hi Ming,

Quoted 8 lines

Wrt. .cmd_per_lun, I think it is safe to modify it into one correct depth because almost all drivers are just producer of .cmd_per_lun. And except for debug purpose, there are only three consumers of .cmd_per_lun in scsi, and all are for scsi_change_queue_depth():

process_message()
scsi_alloc_sdev()
virtscsi_change_queue_depth()

sg_ioctl_common() also looks to read it, but I can't imagine we could break that interface with either suggested change.

So I still prefer not to modify shost.cmd_per_lun, but if you feel strongly enough then I can look to make that change.

Thanks,
John

#5 lore raw
Quoted 14 lines · John Garry
Quoted 5 lines · Ming Lei
Quoted 5 lines

I would rather not change the values which are provided from the driver. I would rather take the original values and try to use them in a sane way.

I have not seen other places where driver shost config values are modified by the core code.

Hi Ming,

Quoted 5 lines

Wrt. .cmd_per_lun, I think it is safe to modify it into one correct depth because almost all drivers are just producer of .cmd_per_lun. And except for debug purpose, there are only three consumers of .cmd_per_lun in scsi, and all are for scsi_change_queue_depth():

Quoted 6 lines
Quoted 3 lines

process_message()
scsi_alloc_sdev()
virtscsi_change_queue_depth()

sg_ioctl_common() also looks to read it, but I can't imagine we could break that interface with either suggested change.

Then one bad .cmd_per_lun can be passed to userspace, as your patch doesn't cover this case.

Quoted 3 lines

So I still prefer not to modify shost.cmd_per_lun, but if you feel strongly enough then I can look to make that change.

I still suggest to make .cmd_per_lun correct since the beginning, otherwise you may have to cover anywhere .cmd_per_lun is used.

Thanks,
Ming