xemu/block
Vladimir Sementsov-Ogievskiy d12ade5732 mirror: fix dead-lock
Let start from the beginning:

Commit b9e413dd37 (in 2.9)
"block: explicitly acquire aiocontext in aio callbacks that need it"
added pairs of aio_context_acquire/release to mirror_write_complete and
mirror_read_complete, when they were aio callbacks for blk_aio_* calls.

Then, commit 2e1990b26e (in 3.0) "block/mirror: Convert to coroutines"
dropped these blk_aio_* calls, than mirror_write_complete and
mirror_read_complete are not callbacks more, and don't need additional
aiocontext acquiring. Furthermore, mirror_read_complete calls
blk_co_pwritev inside these pair of aio_context_acquire/release, which
leads to the following dead-lock with mirror:

 (gdb) info thr
   Id   Target Id         Frame
   3    Thread (LWP 145412) "qemu-system-x86" syscall ()
   2    Thread (LWP 145416) "qemu-system-x86" __lll_lock_wait ()
 * 1    Thread (LWP 145411) "qemu-system-x86" __lll_lock_wait ()

 (gdb) bt
 #0  __lll_lock_wait ()
 #1  _L_lock_812 ()
 #2  __GI___pthread_mutex_lock
 #3  qemu_mutex_lock_impl (mutex=0x561032dce420 <qemu_global_mutex>,
     file=0x5610327d8654 "util/main-loop.c", line=236) at
     util/qemu-thread-posix.c:66
 #4  qemu_mutex_lock_iothread_impl
 #5  os_host_main_loop_wait (timeout=480116000) at util/main-loop.c:236
 #6  main_loop_wait (nonblocking=0) at util/main-loop.c:497
 #7  main_loop () at vl.c:1892
 #8  main

Printing contents of qemu_global_mutex, I see that "__owner = 145416",
so, thr1 is main loop, and now it wants BQL, which is owned by thr2.

 (gdb) thr 2
 (gdb) bt
 #0  __lll_lock_wait ()
 #1  _L_lock_870 ()
 #2  __GI___pthread_mutex_lock
 #3  qemu_mutex_lock_impl (mutex=0x561034d25dc0, ...
 #4  aio_context_acquire (ctx=0x561034d25d60)
 #5  dma_blk_cb
 #6  dma_blk_io
 #7  dma_blk_read
 #8  ide_dma_cb
 #9  bmdma_cmd_writeb
 #10 bmdma_write
 #11 memory_region_write_accessor
 #12 access_with_adjusted_size
 #15 flatview_write
 #16 address_space_write
 #17 address_space_rw
 #18 kvm_handle_io
 #19 kvm_cpu_exec
 #20 qemu_kvm_cpu_thread_fn
 #21 qemu_thread_start
 #22 start_thread
 #23 clone ()

Printing mutex in fr 2, I see "__owner = 145411", so thr2 wants aio
context mutex, which is owned by thr1. Classic dead-lock.

Then, let's check that aio context is hold by mirror coroutine: just
print coroutine stack of first tracked request in mirror job target:

 (gdb) [...]
 (gdb) qemu coroutine 0x561035dd0860
 #0  qemu_coroutine_switch
 #1  qemu_coroutine_yield
 #2  qemu_co_mutex_lock_slowpath
 #3  qemu_co_mutex_lock
 #4  qcow2_co_pwritev
 #5  bdrv_driver_pwritev
 #6  bdrv_aligned_pwritev
 #7  bdrv_co_pwritev
 #8  blk_co_pwritev
 #9  mirror_read_complete () at block/mirror.c:232
 #10 mirror_co_read () at block/mirror.c:370
 #11 coroutine_trampoline
 #12 __start_context

Yes it is mirror_read_complete calling blk_co_pwritev after acquiring
aio context.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2018-12-03 16:50:58 +01:00
..
accounting.c block/accounting: introduce latency histogram 2018-03-19 14:58:37 -05:00
backup.c block/backup: make function variables consistently named 2018-08-31 16:28:33 +02:00
blkdebug.c qstring: Move qstring_from_substr()'s @end one to the right 2018-07-28 09:09:58 +02:00
blklogwrites.c block: Use BdrvChild to discard 2018-07-10 16:01:52 +02:00
blkreplay.c trivial: Make bios files and source files non-executable 2018-09-25 17:26:18 +02:00
blkverify.c qstring: Move qstring_from_substr()'s @end one to the right 2018-07-28 09:09:58 +02:00
block-backend.c block: Null pointer dereference in blk_root_get_parent_desc() 2018-11-12 17:49:21 +01:00
bochs.c block: Require auto-read-only for existing fallbacks 2018-11-05 15:09:55 +01:00
cloop.c block: Require auto-read-only for existing fallbacks 2018-11-05 15:09:55 +01:00
commit.c block/commit: refactor commit to use job callbacks 2018-09-25 15:31:15 +02:00
copy-on-read.c block: drop empty .bdrv_close handlers 2018-08-15 12:50:39 +02:00
create.c jobs: utilize job_exit shim 2018-08-31 16:28:33 +02:00
crypto.c luks: Allow share-rw=on 2018-08-15 12:50:39 +02:00
crypto.h block/crypto: Simplify block_crypto_{open,create}_opts_init() 2018-06-29 14:20:56 +02:00
curl.c curl: Support auto-read-only option 2018-11-05 15:09:55 +01:00
dirty-bitmap.c dirty-bitmaps: clean-up bitmaps loading and migration logic 2018-10-29 16:23:17 -04:00
dmg-bz2.c dmg: Move libbz2 code to dmg-bz2.so 2016-10-07 14:14:06 +02:00
dmg.c block: Require auto-read-only for existing fallbacks 2018-11-05 15:09:55 +01:00
dmg.h block: remove "qemu/osdep.h" from header file 2017-12-18 17:07:02 +03:00
file-posix.c file-posix: Fix shared locks on reopen commit 2018-11-19 14:32:01 +01:00
file-win32.c block: Prefix file driver trace points with "file_" 2018-07-10 16:01:51 +02:00
gluster.c gluster: Support auto-read-only option 2018-11-05 15:09:55 +01:00
io.c block: Use a single global AioWait 2018-09-25 15:50:15 +02:00
iscsi-opts.c Move include qemu/option.h from qemu-common.h to actual users 2018-02-09 13:52:16 +01:00
iscsi.c iscsi: Support auto-read-only option 2018-11-05 15:09:55 +01:00
linux-aio.c block/linux-aio: acquire AioContext before qemu_laio_process_completions 2018-09-25 15:50:15 +02:00
Makefile.objs block: Make more block drivers compile-time configurable 2018-11-12 17:49:21 +01:00
mirror.c mirror: fix dead-lock 2018-12-03 16:50:58 +01:00
nbd-client.c nbd: Support auto-read-only option 2018-11-05 15:09:55 +01:00
nbd-client.h nbd/client: Add x-dirty-bitmap to query bitmap from server 2018-07-02 15:27:38 -05:00
nbd.c qstring: Move qstring_from_substr()'s @end one to the right 2018-07-28 09:09:58 +02:00
nfs.c block: Convert .bdrv_truncate callback to coroutine_fn 2018-06-29 14:20:56 +02:00
null.c block: drop empty .bdrv_close handlers 2018-08-15 12:50:39 +02:00
nvme.c nvme: correct locking around completion 2018-10-12 09:46:14 +08:00
parallels.c parallels: Switch to byte-based calls 2018-06-29 14:20:56 +02:00
parallels.h Clean up includes 2018-02-09 05:05:11 +01:00
qapi.c block/qapi: Fix memory leak in qmp_query_blockstats() 2018-08-15 12:50:39 +02:00
qcow2-bitmap.c block/qcow2-bitmap: Don't take address of fields in packed structs 2018-11-05 15:09:54 +01:00
qcow2-cache.c qcow2: Allow configuring the L2 slice size 2018-02-13 17:00:00 +01:00
qcow2-cluster.c qcow2: Free allocated clusters on write error 2018-06-29 14:20:56 +02:00
qcow2-refcount.c qcow2: Don't allow overflow during cluster allocation 2018-11-19 12:51:40 +01:00
qcow2-snapshot.c block: use local path for local headers 2018-05-31 04:16:06 +03:00
qcow2.c qcow2: Get the request alignment for encrypted images from QCryptoBlock 2018-11-05 15:09:54 +01:00
qcow2.h qcow2: Don't allow overflow during cluster allocation 2018-11-19 12:51:40 +01:00
qcow.c block/qcow: Don't take address of fields in packed structs 2018-11-05 15:09:54 +01:00
qed-check.c block: convert bdrv_check callback to coroutine_fn 2018-03-09 15:17:47 +01:00
qed-cluster.c qed: protect table cache with CoMutex 2017-07-17 11:34:11 +08:00
qed-l2-cache.c qed: protect table cache with CoMutex 2017-07-17 11:34:11 +08:00
qed-table.c block: convert bdrv_check callback to coroutine_fn 2018-03-09 15:17:47 +01:00
qed.c error: Fix use of error_prepend() with &error_fatal, &error_abort 2018-10-19 14:51:34 +02:00
qed.h qed: protect table cache with CoMutex 2017-07-17 11:34:11 +08:00
quorum.c quorum: Forbid adding children in blkverify mode 2018-11-05 15:09:54 +01:00
raw-format.c block: drop empty .bdrv_close handlers 2018-08-15 12:50:39 +02:00
rbd.c block: Require auto-read-only for existing fallbacks 2018-11-05 15:09:55 +01:00
replication.c replication: Switch to byte-based calls 2018-06-29 14:20:56 +02:00
sheepdog.c block: Use warn_report() & friends to report warnings 2018-10-19 14:51:34 +02:00
snapshot.c block: make .bdrv_close optional 2018-08-15 12:50:39 +02:00
ssh.c block: Convert .bdrv_truncate callback to coroutine_fn 2018-06-29 14:20:56 +02:00
stream.c block/stream: refactor stream to use job callbacks 2018-09-25 15:31:15 +02:00
throttle-groups.c throttle-groups: Don't allow timers without throttled requests 2018-08-15 12:50:39 +02:00
throttle.c block: Use BdrvChild to discard 2018-07-10 16:01:52 +02:00
trace-events block: Add copy offloading trace points 2018-07-10 16:01:52 +02:00
vdi.c vdi: Use a literal number of bytes for DEFAULT_CLUSTER_SIZE 2018-11-05 15:28:48 +01:00
vhdx-endian.c block/vhdx: Don't take address of fields in packed structs 2018-11-05 15:09:54 +01:00
vhdx-log.c block/vhdx: Don't take address of fields in packed structs 2018-11-05 15:09:54 +01:00
vhdx.c block/vhdx: Don't take address of fields in packed structs 2018-11-05 15:09:54 +01:00
vhdx.h block/vhdx: Don't take address of fields in packed structs 2018-11-05 15:09:54 +01:00
vmdk.c vmdk: align end of file to a sector boundary 2018-09-26 10:47:18 +08:00
vpc.c vpc: Don't leak opts in vpc_open() 2018-11-05 15:09:54 +01:00
vvfat.c vvfat: Fix memory leak 2018-11-19 12:51:40 +01:00
vxhs.c block: Add block-specific QDict header 2018-06-15 14:49:44 +02:00
win32-aio.c file-win32: Switch to byte-based callbacks 2018-05-15 16:11:41 +02:00
write-threshold.c qapi: Drop qapi_event_send_FOO()'s Error ** argument 2018-08-28 18:21:38 +02:00