Message ID | 20211203130907.3667775-1-ming.lei@redhat.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | [V2] null_blk: allow zero poll queues | expand |
On Dec 03, 2021 / 21:09, Ming Lei wrote: > There isn't any reason to not allow zero poll queues from user > viewpoint. > > Also sometimes we need to compare io poll between poll mode and irq > mode, so not allowing poll queues is bad. > > Fixes: 15dfc662ef31 ("null_blk: Fix handling of submit_queues and poll_queues attributes") > Cc: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> > Signed-off-by: Ming Lei <ming.lei@redhat.com> Thank you for the fix. Looks good to me. Reviewed-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
On Dec 03, 2021 / 21:09, Ming Lei wrote: > There isn't any reason to not allow zero poll queues from user > viewpoint. > > Also sometimes we need to compare io poll between poll mode and irq > mode, so not allowing poll queues is bad. > > Fixes: 15dfc662ef31 ("null_blk: Fix handling of submit_queues and poll_queues attributes") > Cc: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> > Signed-off-by: Ming Lei <ming.lei@redhat.com> > --- > V2: > - fix divide zero fault as reported by Shin'ichiro Jens, Ming, I find the V1 was applied to the for/5.17-drivers branch, and the additional fix in the V2 does not look applied to the branch. Do we need another small patch to cover the delta between V1 and V2?
On 12/23/21 5:52 PM, Shinichiro Kawasaki wrote: > On Dec 03, 2021 / 21:09, Ming Lei wrote: >> There isn't any reason to not allow zero poll queues from user >> viewpoint. >> >> Also sometimes we need to compare io poll between poll mode and irq >> mode, so not allowing poll queues is bad. >> >> Fixes: 15dfc662ef31 ("null_blk: Fix handling of submit_queues and poll_queues attributes") >> Cc: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> >> Signed-off-by: Ming Lei <ming.lei@redhat.com> >> --- >> V2: >> - fix divide zero fault as reported by Shin'ichiro > > Jens, Ming, > > I find the V1 was applied to the for/5.17-drivers branch, and the additional fix > in the V2 does not look applied to the branch. Do we need another small patch to > cover the delta between V1 and V2? Yes, please send an incremental, thanks.
On Dec 23, 2021 / 17:55, Jens Axboe wrote: > On 12/23/21 5:52 PM, Shinichiro Kawasaki wrote: > > On Dec 03, 2021 / 21:09, Ming Lei wrote: > >> There isn't any reason to not allow zero poll queues from user > >> viewpoint. > >> > >> Also sometimes we need to compare io poll between poll mode and irq > >> mode, so not allowing poll queues is bad. > >> > >> Fixes: 15dfc662ef31 ("null_blk: Fix handling of submit_queues and poll_queues attributes") > >> Cc: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> > >> Signed-off-by: Ming Lei <ming.lei@redhat.com> > >> --- > >> V2: > >> - fix divide zero fault as reported by Shin'ichiro > > > > Jens, Ming, > > > > I find the V1 was applied to the for/5.17-drivers branch, and the additional fix > > in the V2 does not look applied to the branch. Do we need another small patch to > > cover the delta between V1 and V2? > > Yes, please send an incremental, thanks. Thanks, and Ming already provided the small patch (So quick!). I've confirmed that the patch avoids the issue. Merge to for/5.17-drivers will be appreciated.
diff --git a/drivers/block/null_blk/main.c b/drivers/block/null_blk/main.c index 54f7d490f8eb..c73e69ab77ed 100644 --- a/drivers/block/null_blk/main.c +++ b/drivers/block/null_blk/main.c @@ -340,9 +340,9 @@ static int nullb_update_nr_hw_queues(struct nullb_device *dev, return 0; /* - * Make sure at least one queue exists for each of submit and poll. + * Make sure at least one submit queue exists. */ - if (!submit_queues || !poll_queues) + if (!submit_queues) return -EINVAL; /* @@ -1878,8 +1878,7 @@ static int null_init_tag_set(struct nullb *nullb, struct blk_mq_tag_set *set) set->nr_hw_queues = nullb ? nullb->dev->submit_queues : g_submit_queues; poll_queues = nullb ? nullb->dev->poll_queues : g_poll_queues; - if (poll_queues) - set->nr_hw_queues += poll_queues; + set->nr_hw_queues += poll_queues; set->queue_depth = nullb ? nullb->dev->hw_queue_depth : g_hw_queue_depth; set->numa_node = nullb ? nullb->dev->home_node : g_home_node; @@ -1890,7 +1889,7 @@ static int null_init_tag_set(struct nullb *nullb, struct blk_mq_tag_set *set) if (g_shared_tag_bitmap) set->flags |= BLK_MQ_F_TAG_HCTX_SHARED; set->driver_data = nullb; - if (g_poll_queues) + if (poll_queues) set->nr_maps = 3; else set->nr_maps = 1; @@ -1917,8 +1916,6 @@ static int null_validate_conf(struct nullb_device *dev) if (dev->poll_queues > g_poll_queues) dev->poll_queues = g_poll_queues; - else if (dev->poll_queues == 0) - dev->poll_queues = 1; dev->prev_poll_queues = dev->poll_queues; dev->queue_mode = min_t(unsigned int, dev->queue_mode, NULL_Q_MQ);
There isn't any reason to not allow zero poll queues from user viewpoint. Also sometimes we need to compare io poll between poll mode and irq mode, so not allowing poll queues is bad. Fixes: 15dfc662ef31 ("null_blk: Fix handling of submit_queues and poll_queues attributes") Cc: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> Signed-off-by: Ming Lei <ming.lei@redhat.com> --- V2: - fix divide zero fault as reported by Shin'ichiro drivers/block/null_blk/main.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-)