Re: [PATCH v2] migration/colo: Fix bdrv_graph_rdlock_main_loop: Assertion `!qemu_in_coroutine()' failed.

["Zhijian Li (Fujitsu)" via <qemu-devel@nongnu.org>]

On 17/04/2024 14:13, Philippe Mathieu-Daudé wrote:
> On 17/4/24 04:56, Li Zhijian via wrote:
>> bdrv_activate_all() should not be called from the coroutine context, move
>> it to the QEMU thread colo_process_incoming_thread() with the bql_lock
>> protected.
>>
>> The backtrace is as follows:
>>   #4  0x0000561af7948362 in bdrv_graph_rdlock_main_loop () at ../block/graph-lock.c:260
>>   #5  0x0000561af7907a68 in graph_lockable_auto_lock_mainloop (x=0x7fd29810be7b) at /patch/to/qemu/include/block/graph-lock.h:259
>>   #6  0x0000561af79167d1 in bdrv_activate_all (errp=0x7fd29810bed0) at ../block.c:6906
>>   #7  0x0000561af762b4af in colo_incoming_co () at ../migration/colo.c:935
>>   #8  0x0000561af7607e57 in process_incoming_migration_co (opaque=0x0) at ../migration/migration.c:793
>>   #9  0x0000561af7adbeeb in coroutine_trampoline (i0=-106876144, i1=22042) at ../util/coroutine-ucontext.c:175
>>   #10 0x00007fd2a5cf21c0 in  () at /lib64/libc.so.6
>>
>> CC: Fabiano Rosas <farosas@suse.de>
> 
> Cc: qemu-stable@nongnu.org
> 
>> Closes: https://gitlab.com/qemu-project/qemu/-/issues/2277
>> Fixes: 2b3912f135 ("block: Mark bdrv_first_blk() and bdrv_is_root_node() GRAPH_RDLOCK")
>> Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
>> ---
>> V2: fix missing bql_unlock() in error path.
>> ---
>>   migration/colo.c | 18 ++++++++++--------
>>   1 file changed, 10 insertions(+), 8 deletions(-)
>>
>> diff --git a/migration/colo.c b/migration/colo.c
>> index 84632a603e..5600a43d78 100644
>> --- a/migration/colo.c
>> +++ b/migration/colo.c
>> @@ -835,6 +835,16 @@ static void *colo_process_incoming_thread(void *opaque)
>>           return NULL;
>>       }
>> +    /* Make sure all file formats throw away their mutable metadata */
>> +    bql_lock();
> 
> Note there is also the convenient BQL_LOCK_GUARD() macro.

Thanks for your reminder, Philippe.

IIUC, BQL_LOCK_GUARD() will lock all the rest code till the subroutine ends.
So it's not suitable for this case where we only need a partial critical section.


Thanks
Zhijian


> 
>> +    bdrv_activate_all(&local_err);
>> +    if (local_err) {
>> +        bql_unlock();
>> +        error_report_err(local_err);
>> +        return NULL;
>> +    }
>> +    bql_unlock();
>> +
>>       failover_init_state();
>