Messages in this thread | | | Date | Mon, 8 Dec 2014 21:39:22 -0700 | From | Jens Axboe <> | Subject | Re: [PATCH] blk-mq: prevent unmapped hw queue from being scheduled |
| |
On 12/08/2014 05:41 PM, Ming Lei wrote: > On Wed, Dec 3, 2014 at 7:38 PM, Ming Lei <ming.lei@canonical.com> wrote: >> When one hardware queue has no mapped software queues, it >> shouldn't have been scheduled. Otherwise WARNING or OOPS >> can triggered. >> >> blk_mq_hw_queue_mapped() helper is introduce for fixing >> the problem. >> >> Signed-off-by: Ming Lei <ming.lei@canonical.com> >> --- >> block/blk-mq.c | 8 ++++++-- >> block/blk-mq.h | 5 +++++ >> 2 files changed, 11 insertions(+), 2 deletions(-) >> >> diff --git a/block/blk-mq.c b/block/blk-mq.c >> index c95abc6..c916ad0 100644 >> --- a/block/blk-mq.c >> +++ b/block/blk-mq.c >> @@ -599,7 +599,7 @@ static void blk_mq_rq_timer(unsigned long priv) >> * If not software queues are currently mapped to this >> * hardware queue, there's nothing to check >> */ >> - if (!hctx->nr_ctx || !hctx->tags) >> + if (!blk_mq_hw_queue_mapped(hctx)) >> continue; >> >> blk_mq_tag_busy_iter(hctx, blk_mq_check_expired, &data); >> @@ -819,7 +819,8 @@ static int blk_mq_hctx_next_cpu(struct blk_mq_hw_ctx *hctx) >> >> void blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async) >> { >> - if (unlikely(test_bit(BLK_MQ_S_STOPPED, &hctx->state))) >> + if (unlikely(test_bit(BLK_MQ_S_STOPPED, &hctx->state) || >> + !blk_mq_hw_queue_mapped(hctx))) >> return; >> >> if (!async) { >> @@ -926,6 +927,9 @@ static void blk_mq_delay_work_fn(struct work_struct *work) >> >> void blk_mq_delay_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs) >> { >> + if (unlikely(!blk_mq_hw_queue_mapped(hctx))) >> + return; >> + >> kblockd_schedule_delayed_work_on(blk_mq_hctx_next_cpu(hctx), >> &hctx->delay_work, msecs_to_jiffies(msecs)); >> } >> diff --git a/block/blk-mq.h b/block/blk-mq.h >> index d567d52..206230e 100644 >> --- a/block/blk-mq.h >> +++ b/block/blk-mq.h >> @@ -115,4 +115,9 @@ static inline void blk_mq_set_alloc_data(struct blk_mq_alloc_data *data, >> data->hctx = hctx; >> } >> >> +static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx) >> +{ >> + return hctx->nr_ctx && hctx->tags; >> +} >> + >> #endif > > Gentle ping...
Picked up for 3.19, sorry for the delay. I'm curious how this queue gets scheduled, though. My worry here would be that we are masking a bug that should be fixed separately.
-- Jens Axboe
| |