From a77fd4bb2988c05953fdc9f1524085870ec1c939 Mon Sep 17 00:00:00 2001 From: Fam Zheng Date: Tue, 5 Apr 2016 19:20:52 +0800 Subject: block: Fix bdrv_drain in coroutine Using the nested aio_poll() in coroutine is a bad idea. This patch replaces the aio_poll loop in bdrv_drain with a BH, if called in coroutine. For example, the bdrv_drain() in mirror.c can hang when a guest issued request is pending on it in qemu_co_mutex_lock(). Mirror coroutine in this case has just finished a request, and the block job is about to complete. It calls bdrv_drain() which waits for the other coroutine to complete. The other coroutine is a scsi-disk request. The deadlock happens when the latter is in turn pending on the former to yield/terminate, in qemu_co_mutex_lock(). The state flow is as below (assuming a qcow2 image): mirror coroutine scsi-disk coroutine ------------------------------------------------------------- do last write qcow2:qemu_co_mutex_lock() ... scsi disk read tracked request begin qcow2:qemu_co_mutex_lock.enter qcow2:qemu_co_mutex_unlock() bdrv_drain while (has tracked request) aio_poll() In the scsi-disk coroutine, the qemu_co_mutex_lock() will never return because the mirror coroutine is blocked in the aio_poll(blocking=true). With this patch, the added qemu_coroutine_yield() allows the scsi-disk coroutine to make progress as expected: mirror coroutine scsi-disk coroutine ------------------------------------------------------------- do last write qcow2:qemu_co_mutex_lock() ... scsi disk read tracked request begin qcow2:qemu_co_mutex_lock.enter qcow2:qemu_co_mutex_unlock() bdrv_drain.enter > schedule BH > qemu_coroutine_yield() > qcow2:qemu_co_mutex_lock.return > ... tracked request end ... (resumed from BH callback) bdrv_drain.return ... Reported-by: Laurent Vivier Signed-off-by: Fam Zheng Message-id: 1459855253-5378-2-git-send-email-famz@redhat.com Suggested-by: Paolo Bonzini Signed-off-by: Fam Zheng Signed-off-by: Stefan Hajnoczi --- block/io.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ include/block/block.h | 1 + 2 files changed, 46 insertions(+) diff --git a/block/io.c b/block/io.c index c4869b96c5..a7dbf85b19 100644 --- a/block/io.c +++ b/block/io.c @@ -253,6 +253,47 @@ static void bdrv_drain_recurse(BlockDriverState *bs) } } +typedef struct { + Coroutine *co; + BlockDriverState *bs; + QEMUBH *bh; + bool done; +} BdrvCoDrainData; + +static void bdrv_co_drain_bh_cb(void *opaque) +{ + BdrvCoDrainData *data = opaque; + Coroutine *co = data->co; + + qemu_bh_delete(data->bh); + bdrv_drain(data->bs); + data->done = true; + qemu_coroutine_enter(co, NULL); +} + +void coroutine_fn bdrv_co_drain(BlockDriverState *bs) +{ + BdrvCoDrainData data; + + /* Calling bdrv_drain() from a BH ensures the current coroutine yields and + * other coroutines run if they were queued from + * qemu_co_queue_run_restart(). */ + + assert(qemu_in_coroutine()); + data = (BdrvCoDrainData) { + .co = qemu_coroutine_self(), + .bs = bs, + .done = false, + .bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_co_drain_bh_cb, &data), + }; + qemu_bh_schedule(data.bh); + + qemu_coroutine_yield(); + /* If we are resumed from some other event (such as an aio completion or a + * timer callback), it is a bug in the caller that should be fixed. */ + assert(data.done); +} + /* * Wait for pending requests to complete on a single BlockDriverState subtree, * and suspend block driver's internal I/O until next request arrives. @@ -269,6 +310,10 @@ void bdrv_drain(BlockDriverState *bs) bool busy = true; bdrv_drain_recurse(bs); + if (qemu_in_coroutine()) { + bdrv_co_drain(bs); + return; + } while (busy) { /* Keep iterating */ bdrv_flush_io_queue(bs); diff --git a/include/block/block.h b/include/block/block.h index 6a39f946f5..3a731377db 100644 --- a/include/block/block.h +++ b/include/block/block.h @@ -372,6 +372,7 @@ int bdrv_flush(BlockDriverState *bs); int coroutine_fn bdrv_co_flush(BlockDriverState *bs); void bdrv_close_all(void); void bdrv_drain(BlockDriverState *bs); +void coroutine_fn bdrv_co_drain(BlockDriverState *bs); void bdrv_drain_all(void); int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors); -- cgit v1.2.1 From 39bf92dd703efb04fc516aad1376986aea7d96bf Mon Sep 17 00:00:00 2001 From: Fam Zheng Date: Tue, 5 Apr 2016 19:20:53 +0800 Subject: mirror: Replace bdrv_drain(bs) with bdrv_co_drain(bs) Suggested-by: Stefan Hajnoczi Signed-off-by: Fam Zheng Message-id: 1459855253-5378-3-git-send-email-famz@redhat.com Signed-off-by: Stefan Hajnoczi --- block/mirror.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/block/mirror.c b/block/mirror.c index f64db1a69b..c2cfc1a5ef 100644 --- a/block/mirror.c +++ b/block/mirror.c @@ -650,7 +650,7 @@ static void coroutine_fn mirror_run(void *opaque) * mirror_populate runs. */ trace_mirror_before_drain(s, cnt); - bdrv_drain(bs); + bdrv_co_drain(bs); cnt = bdrv_get_dirty_count(s->dirty_bitmap); } -- cgit v1.2.1 From 9ca3003df3c9882925c023c1411ab3b2041d7db5 Mon Sep 17 00:00:00 2001 From: Fam Zheng Date: Tue, 5 Apr 2016 17:38:25 +0800 Subject: MAINTAINERS: Add Fam Zheng as a co-maintainer of block I/O path As agreed with Stefan, I'm listing myself a co-maintainer of block I/O path and assist with the maintainership. Signed-off-by: Fam Zheng Message-id: 1459849105-7767-1-git-send-email-famz@redhat.com Signed-off-by: Stefan Hajnoczi --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index 9277fbf128..d26141277d 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -998,6 +998,7 @@ T: git git://repo.or.cz/qemu/kevin.git block Block I/O path M: Stefan Hajnoczi +M: Fam Zheng L: qemu-block@nongnu.org S: Supported F: async.c -- cgit v1.2.1