On Sun 10-07-22 12:21:59, Yu Kuai wrote:
> From: Yu Kuai <yukuai3@xxxxxxxxxx>
>
> For example, 2 * wake_batch tags are put, while only wake_batch threads
> are woken:
>
> __sbq_wake_up
> atomic_cmpxchg -> reset wait_cnt
> __sbq_wake_up -> decrease wait_cnt
> ...
> __sbq_wake_up -> wait_cnt is decreased to 0 again
> atomic_cmpxchg
> sbq_index_atomic_inc -> increase wake_index
> wake_up_nr -> wake up and waitqueue might be empty
> sbq_index_atomic_inc -> increase again, one waitqueue is skipped
> wake_up_nr -> invalid wake up because old wakequeue might be empty
>
> To fix the problem, increasing 'wake_index' before resetting 'wait_cnt'.
>
> Fixes: 88459642cba4 ("blk-mq: abstract tag allocation out into sbitmap library")
> Signed-off-by: Yu Kuai <yukuai3@xxxxxxxxxx>
This patch and the following one look sane to me but please merge them to a
single patch. They fix the same race of two concurrent wakers just with a
slightly different timing so there isn't a point in having two patches for
this (in particular changes in this patch are difficult to reason about
when we know the result is still buggy).
Honza
> ---
> lib/sbitmap.c | 45 +++++++++++++++++++++++----------------------
> 1 file changed, 23 insertions(+), 22 deletions(-)
>
> diff --git a/lib/sbitmap.c b/lib/sbitmap.c
> index b46fce1beb3a..57095dd88a33 100644
> --- a/lib/sbitmap.c
> +++ b/lib/sbitmap.c
> @@ -616,32 +616,33 @@ static bool __sbq_wake_up(struct sbitmap_queue *sbq)
> return false;
>
> wait_cnt = atomic_dec_return(&ws->wait_cnt);
> - if (wait_cnt <= 0) {
> - int ret;
> + if (wait_cnt > 0)
> + return false;
>
> - wake_batch = READ_ONCE(sbq->wake_batch);
> + /*
> + * For concurrent callers of this, callers should call this function
> + * again to wakeup a new batch on a different 'ws'.
> + */
> + if (wait_cnt < 0)
> + return true;
>
> - /*
> - * Pairs with the memory barrier in sbitmap_queue_resize() to
> - * ensure that we see the batch size update before the wait
> - * count is reset.
> - */
> - smp_mb__before_atomic();
> + wake_batch = READ_ONCE(sbq->wake_batch);
>
> - /*
> - * For concurrent callers of this, the one that failed the
> - * atomic_cmpxhcg() race should call this function again
> - * to wakeup a new batch on a different 'ws'.
> - */
> - ret = atomic_cmpxchg(&ws->wait_cnt, wait_cnt, wake_batch);
> - if (ret == wait_cnt) {
> - sbq_index_atomic_inc(&sbq->wake_index);
> - wake_up_nr(&ws->wait, wake_batch);
> - return false;
> - }
> + /*
> + * Pairs with the memory barrier in sbitmap_queue_resize() to
> + * ensure that we see the batch size update before the wait
> + * count is reset.
> + */
> + smp_mb__before_atomic();
>
> - return true;
> - }
> + /*
> + * Increase wake_index before updating wait_cnt, otherwise concurrent
> + * callers can see valid wait_cnt in old waitqueue, which can cause
> + * invalid wakeup on the old waitqueue.
> + */
> + sbq_index_atomic_inc(&sbq->wake_index);
> + atomic_set(&ws->wait_cnt, wake_batch);
> + wake_up_nr(&ws->wait, wake_batch);
>
> return false;
> }
> --
> 2.31.1
>
--
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR
