[Qemu-devel] [PATCH for 2.7 0/2] block: fixes for deadlock in flush code

[Denis V. Lunev]

We have suffered from the following deadlock

Thread 2 (Thread 0x7f1b7edf9700 (LWP 240293)):
#0  0x00007f1bd1f0675f in ppoll () from /lib64/libc.so.6
#1  0x00007f1bd8c1d78b in ppoll (__ss=0x0, __timeout=0x0, __nfds=<optimized 
out>, __fds=<optimized out>) at /usr/include/bits/poll2.h:77
#2  qemu_poll_ns (fds=<optimized out>, nfds=<optimized out>, address@hidden) at 
qemu-timer.c:310
#3  0x00007f1bd8c1e8bf in aio_poll (ctx=0x7f1bda091780, address@hidden) at 
aio-posix.c:451
#4  0x00007f1bd8c119cf in bdrv_drain_one (address@hidden) at block.c:2055
#5  0x00007f1bd8c13244 in bdrv_drain_all () at block.c:2115
#6  0x00007f1bd8a2c5e3 in vm_stop (state=<optimized out>) at 
/usr/src/debug/qemu-2.3.0/cpus.c:685
#7  0x00007f1bd8a2c636 in vm_stop_force_state (state=<optimized out>) at 
/usr/src/debug/qemu-2.3.0/cpus.c:1383
#8  0x00007f1bd8bc798f in migration_completion (start_time=<synthetic pointer>, 
old_vm_running=<synthetic pointer>, current_active_state=<optimized out>, 
s=0x7f1bd90e3c20 <current_migration.37255>)
        at migration/migration.c:1213
#9  migration_thread (opaque=0x7f1bd90e3c20 <current_migration.37255>) at 
migration/migration.c:1314
#10 0x00007f1bd21e3dc5 in start_thread () from /lib64/libpthread.so.0
#11 0x00007f1bd1f10ced in clone () from /lib64/libc.so.6

The problem was narrowed down to the commit
    commit 3ff2f67a7c24183fcbcfe1332e5223ac6f96438c
    Author: Evgeny Yakovlev <address@hidden>
    Date:   Mon Jul 18 22:39:52 2016 +0300
    block: ignore flush requests when storage is clean

This patches contains fixes for the situation. The probability of the
problem is not that big. Our regression testing faces it ~1 time
a week or less.

Signed-off-by: Evgeny Yakovlev <address@hidden>
Signed-off-by: Denis V. Lunev <address@hidden>
CC: Stefan Hajnoczi <address@hidden>
CC: Fam Zheng <address@hidden>
CC: Kevin Wolf <address@hidden>
CC: Max Reitz <address@hidden>