On Sat, Sep 30, 2017 at 10:57:31PM +0800, weiping zhang wrote:
> On Fri, Sep 22, 2017 at 11:36:28PM +0800, weiping zhang wrote:
> > if blk-mq use "none" io scheduler, nr_request get a wrong value when
> > input a number > tag_set->queue_depth. blk_mq_tag_update_depth will get
> > the smaller one min(nr, set->queue_depth), and then q->nr_request get a
> > wrong value.
> >
> > Reproduce:
> >
> > echo none > /sys/block/nvme0n1/queue/scheduler
> > echo 1000000 > /sys/block/nvme0n1/queue/nr_requests
> > cat /sys/block/nvme0n1/queue/nr_requests
> > 1000000
> >
> > Signed-off-by: weiping zhang <zhangweiping@xxxxxxxxxxxxxxx>
> > ---
> >
> > Changes since v4:
> > * fix typo in commit message(queue/ioscheduler => queue/scheduler)
> >
> > Changes since v3:
> > * remove compare nr with tags->qdepth, pass nr to blk_mq_tag_update_depth
> > directly
> >
> > * remove return EINVAL when user modify nr_request less than BLKDEV_MIN_RQ
> >
> > Changes since v2:
> > * add return EINVAL when user modify nr_request less than BLKDEV_MIN_RQ
> > * remove pr_warn, and return EINVAL, if input number is too large
> >
> > block/blk-mq.c | 3 +--
> > 1 file changed, 1 insertion(+), 2 deletions(-)
> >
> > diff --git a/block/blk-mq.c b/block/blk-mq.c
> > index 98a1860..491e336 100644
> > --- a/block/blk-mq.c
> > +++ b/block/blk-mq.c
> > @@ -2642,8 +2642,7 @@ int blk_mq_update_nr_requests(struct request_queue *q, unsigned int nr)
> > * queue depth. This is similar to what the old code would do.
> > */
> > if (!hctx->sched_tags) {
> > - ret = blk_mq_tag_update_depth(hctx, &hctx->tags,
> > - min(nr, set->queue_depth),
> > + ret = blk_mq_tag_update_depth(hctx, &hctx->tags, nr,
> > false);
> > } else {
> > ret = blk_mq_tag_update_depth(hctx, &hctx->sched_tags,
> > --
> > 2.9.4
> >
>
> Hi Jens,
>
> As you say before:
> > blk_mq_tag_update_depth() should already return
> > -EINVAL for the case where we can't grow the tags. Looks like this patch
> > should simply remove the min(nr, set->queue_depth) and just pass in 'nr'.
> I also find that hctx->tags->nr_tags equal to tag_set->queue_depth no matter use
> hctx->sched_tags or not. So I think it's safe to verify 'nr' by
> hctx->tags->nr_tags.
>
Hello Jens,
ping
