| Message ID | 20151014022857.34443.65126.stgit@dwillia2-desk3.jf.intel.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
>>>>> "Dan" == Dan Williams <dan.j.williams@intel.com> writes:
Dan> Since they lack requests to pin the request_queue active,
Dan> synchronous bio-based drivers may have in-flight integrity work
Dan> from bio_integrity_endio() that is not flushed by
Dan> blk_freeze_queue(). Flush that work to prevent races to free the
Dan> queue and the final usage of the blk_integrity profile.
Does this work when BLK_DEV_INTEGRITY is undefined?
On Wed, Oct 14, 2015 at 3:41 PM, Martin K. Petersen <martin.petersen@oracle.com> wrote: >>>>>> "Dan" == Dan Williams <dan.j.williams@intel.com> writes: > > Dan> Since they lack requests to pin the request_queue active, > Dan> synchronous bio-based drivers may have in-flight integrity work > Dan> from bio_integrity_endio() that is not flushed by > Dan> blk_freeze_queue(). Flush that work to prevent races to free the > Dan> queue and the final usage of the blk_integrity profile. > > Does this work when BLK_DEV_INTEGRITY is undefined? > Nope. Fixed up this and patch 10.
diff --git a/block/bio-integrity.c b/block/bio-integrity.c index 6a90eca9cea1..f6325d573c10 100644 --- a/block/bio-integrity.c +++ b/block/bio-integrity.c @@ -32,6 +32,11 @@ static struct kmem_cache *bip_slab; static struct workqueue_struct *kintegrityd_wq; +void blk_flush_integrity(void) +{ + flush_workqueue(kintegrityd_wq); +} + /** * bio_integrity_alloc - Allocate integrity payload and attach it to bio * @bio: bio to attach integrity metadata to diff --git a/block/blk-core.c b/block/blk-core.c index 9b4d735cb5b8..6ebe33ed5154 100644 --- a/block/blk-core.c +++ b/block/blk-core.c @@ -561,6 +561,9 @@ void blk_cleanup_queue(struct request_queue *q) queue_flag_set(QUEUE_FLAG_DEAD, q); spin_unlock_irq(lock); + /* for synchronous bio-based driver finish in-flight integrity i/o */ + blk_flush_integrity(); + /* @q won't process any more request, flush async actions */ del_timer_sync(&q->backing_dev_info.laptop_mode_wb_timer); blk_sync_queue(q); diff --git a/block/blk.h b/block/blk.h index 5b2cd393afbe..cda2e80ea30e 100644 --- a/block/blk.h +++ b/block/blk.h @@ -87,6 +87,8 @@ static inline void blk_queue_enter_live(struct request_queue *q) percpu_ref_get(&q->q_usage_counter); } +void blk_flush_integrity(void); + void blk_rq_timed_out_timer(unsigned long data); unsigned long blk_rq_timeout(unsigned long timeout); void blk_add_timer(struct request *req);
Since they lack requests to pin the request_queue active, synchronous bio-based drivers may have in-flight integrity work from bio_integrity_endio() that is not flushed by blk_freeze_queue(). Flush that work to prevent races to free the queue and the final usage of the blk_integrity profile. This is temporary unless/until bio-based drivers start to generically take a q_usage_counter reference while a bio is in-flight. Signed-off-by: Dan Williams <dan.j.williams@intel.com> --- block/bio-integrity.c | 5 +++++ block/blk-core.c | 3 +++ block/blk.h | 2 ++ 3 files changed, 10 insertions(+)