On 01/18/2013 11:26 PM, Alan Stern wrote:
On Fri, 18 Jan 2013, Aaron Lu wrote:
Either you have to make sure the q->nr_pending is always correct, even
when runtime PM isn't being used, or else the caller has to make sure
that no I/O takes place before blk_pm_runtime_init is called.
I think we can say:
blk_pm_runtime_init can't be called after any requests have been
submitted but not finished.
Sounds more accurate?
Okay. I think you can add (in parentheses) that in most cases drivers
should call the routine before any I/O has taken place.
The reason I put it that way is, in patch 4, the blk_pm_runtime_init is
called after a request is executed(scsi_probe_lun). I do not want people
get confused by the comments for blk_pm_runtime_init and the example
code in patch 4 where we didn't follow it :-)
Considering ODD's use case, I was thinking of moving the
blk_pm_runtime_init call to sd.c, as sr will not use request based auto
suspend. Probably right before we decrease usage count for the device in
sd_probe_async. What do you think?
Thanks,
Aaron
--
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
