Skip to contentSkip to conversation

aio-wait: avoid AioContext lock in aio_wait_bh_oneshot()

PATCH Applied Stefan Hajnoczi 4 participants 4 messages Apr 4, 2023 – Apr 27, 2023

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 aio-wait: avoid AioContext lock in aio_wait_bh_oneshot()

There is no need for the AioContext lock in aio_wait_bh_oneshot(). It's easy to remove the lock from existing callers and then switch from AIO_WAIT_WHILE() to AIO_WAIT_WHILE_UNLOCKED() in aio_wait_bh_oneshot().

Document that the AioContext lock should not be held across aio_wait_bh_oneshot(). Holding a lock across aio_poll() can cause deadlock so we don't want callers to do that.

This is a step towards getting rid of the AioContext lock.

---
 include/block/aio-wait.h        | 2 +-
 hw/block/dataplane/virtio-blk.c | 3 ++-
 hw/scsi/virtio-scsi-dataplane.c | 2 --
 util/aio-wait.c                 | 2 +-
 4 files changed, 4 insertions(+), 5 deletions(-)
include/block/aio-wait.h +1 -1
index da13357bb8..fdd14294ec 100644
@@ -131,7 +131,7 @@ void aio_wait_kick(void);
  *
  * Run a BH in @ctx and wait for it to complete.
  *
- * Must be called from the main loop thread with @ctx acquired exactly once.
+ * Must be called from the main loop thread without @ctx acquired.
  * Note that main loop event processing may occur.
  */
 void aio_wait_bh_oneshot(AioContext *ctx, QEMUBHFunc *cb, void *opaque);
hw/block/dataplane/virtio-blk.c +2 -1
index b28d81737e..e0111efd6d 100644
@@ -314,9 +314,10 @@ void virtio_blk_data_plane_stop(VirtIODevice *vdev)
     s->stopping = true;
     trace_virtio_blk_data_plane_stop(s);
 
-    aio_context_acquire(s->ctx);
     aio_wait_bh_oneshot(s->ctx, virtio_blk_data_plane_stop_bh, s);
 
+    aio_context_acquire(s->ctx);
+
     /* Wait for virtio_blk_dma_restart_bh() and in flight I/O to complete */
     blk_drain(s->conf->conf.blk);
 
hw/scsi/virtio-scsi-dataplane.c +0 -2
index 20bb91766e..f3214e1c57 100644
@@ -197,9 +197,7 @@ void virtio_scsi_dataplane_stop(VirtIODevice *vdev)
     }
     s->dataplane_stopping = true;
 
-    aio_context_acquire(s->ctx);
     aio_wait_bh_oneshot(s->ctx, virtio_scsi_dataplane_stop_bh, s);
-    aio_context_release(s->ctx);
 
     blk_drain_all(); /* ensure there are no in-flight requests */
 
util/aio-wait.c +1 -1
index 98c5accd29..b5336cf5fd 100644
@@ -82,5 +82,5 @@ void aio_wait_bh_oneshot(AioContext *ctx, QEMUBHFunc *cb, void *opaque)
     assert(qemu_get_current_aio_context() == qemu_get_aio_context());
 
     aio_bh_schedule_oneshot(ctx, aio_wait_bh, &data);
-    AIO_WAIT_WHILE(ctx, !data.done);
+    AIO_WAIT_WHILE_UNLOCKED(NULL, !data.done);
 }
Signature · 1 line
2.39.2
Reviewed-by on v1 1/1
#1 lore raw
Quoted 9 lines · Stefan Hajnoczi

There is no need for the AioContext lock in aio_wait_bh_oneshot(). It's easy to remove the lock from existing callers and then switch from AIO_WAIT_WHILE() to AIO_WAIT_WHILE_UNLOCKED() in aio_wait_bh_oneshot().

Document that the AioContext lock should not be held across aio_wait_bh_oneshot(). Holding a lock across aio_poll() can cause deadlock so we don't want callers to do that.

This is a step towards getting rid of the AioContext lock.

Paolo

Reviewed-by on v1 1/1
#2 lore raw
In reply to Stefan Hajnoczi

Am 04/04/2023 um 17:33 schrieb Stefan Hajnoczi:

Quoted 4 lines

There is no need for the AioContext lock in aio_wait_bh_oneshot(). It's easy to remove the lock from existing callers and then switch from AIO_WAIT_WHILE() to AIO_WAIT_WHILE_UNLOCKED() in aio_wait_bh_oneshot().

Quoted 9 lines

Document that the AioContext lock should not be held across aio_wait_bh_oneshot(). Holding a lock across aio_poll() can cause deadlock so we don't want callers to do that.

This is a step towards getting rid of the AioContext lock.

Applied
#3 lore raw
In reply to Stefan Hajnoczi

Am 04.04.2023 um 17:33 hat Stefan Hajnoczi geschrieben:

Quoted 4 lines

There is no need for the AioContext lock in aio_wait_bh_oneshot(). It's easy to remove the lock from existing callers and then switch from AIO_WAIT_WHILE() to AIO_WAIT_WHILE_UNLOCKED() in aio_wait_bh_oneshot().

Quoted 8 lines

Document that the AioContext lock should not be held across aio_wait_bh_oneshot(). Holding a lock across aio_poll() can cause deadlock so we don't want callers to do that.

This is a step towards getting rid of the AioContext lock.

Thanks, applied to the block branch.

Kevin