Re: [Qemu-devel] [Qemu-block] [PATCH] blk: postpone request execution on a context protected with "drained section"

[Denis Plotnikov]

On 24.06.2019 12:46, Denis Plotnikov wrote:
> 
> 
> On 21.06.2019 12:59, Vladimir Sementsov-Ogievskiy wrote:
>> 21.06.2019 12:16, Kevin Wolf wrote:
>>> Am 09.04.2019 um 12:01 hat Kevin Wolf geschrieben:
>>>> Am 02.04.2019 um 10:35 hat Denis Plotnikov geschrieben:
>>>>> On 13.03.2019 19:04, Kevin Wolf wrote:
>>>>>> Am 14.12.2018 um 12:54 hat Denis Plotnikov geschrieben:
>>>>>>> On 13.12.2018 15:20, Kevin Wolf wrote:
>>>>>>>> Am 13.12.2018 um 12:07 hat Denis Plotnikov geschrieben:
>>>>>>>>> Sounds it should be so, but it doesn't work that way and that's why:
>>>>>>>>> when doing mirror we may resume postponed coroutines too early when 
>>>>>>>>> the
>>>>>>>>> underlying bs is protected from writing at and thus we encounter the
>>>>>>>>> assert on a write request execution at bdrv_co_write_req_prepare when
>>>>>>>>> resuming the postponed coroutines.
>>>>>>>>>
>>>>>>>>> The thing is that the bs is protected for writing before execution of
>>>>>>>>> bdrv_replace_node at mirror_exit_common and bdrv_replace_node calls
>>>>>>>>> bdrv_replace_child_noperm which, in turn, calls 
>>>>>>>>> child->role->drained_end
>>>>>>>>> where one of the callbacks is blk_root_drained_end which check
>>>>>>>>> if(--blk->quiesce_counter == 0) and runs the postponed requests
>>>>>>>>> (coroutines) if the coundition is true.
>>>>>>>>
>>>>>>>> Hm, so something is messed up with the drain sections in the mirror
>>>>>>>> driver. We have:
>>>>>>>>
>>>>>>>>          bdrv_drained_begin(target_bs);
>>>>>>>>          bdrv_replace_node(to_replace, target_bs, &local_err);
>>>>>>>>          bdrv_drained_end(target_bs);
>>>>>>>>
>>>>>>>> Obviously, the intention was to keep the BlockBackend drained during
>>>>>>>> bdrv_replace_node(). So how could blk->quiesce_counter ever get to 0
>>>>>>>> inside bdrv_replace_node() when target_bs is drained?
>>>>>>>>
>>>>>>>> Looking at bdrv_replace_child_noperm(), it seems that the function has
>>>>>>>> a bug: Even if old_bs and new_bs are both drained, the quiesce_counter
>>>>>>>> for the parent reaches 0 for a moment because we call .drained_end for
>>>>>>>> the old child first and .drained_begin for the new one later.
>>>>>>>>
>>>>>>>> So it seems the fix would be to reverse the order and first call
>>>>>>>> .drained_begin for the new child and then .drained_end for the old
>>>>>>>> child. Sounds like a good new testcase for tests/test-bdrv-drain.c, 
>>>>>>>> too.
>>>>>>> Yes, it's true, but it's not enough...
>>>>>>
>>>>>> Did you ever implement the changes suggested so far, so that we could
>>>>>> continue from there? Or should I try and come up with something myself?
>>>>>
>>>>> Sorry for the late reply...
>>>>> Yes, I did ...
>>>>
>>>> If there are more question or problems, can you post the patches in
>>>> their current shape (as an RFC) or a git URL so I can play with it a
>>>> bit? If you could include a failing test case, too, that would be ideal.
>>>
>>> Denis? Please?
>>>
>>> We really should get this fixed and I would be willing to lend a hand,
>>> but if you keep your patches secret, I can't really do so and would have
>>> to duplicate your work.
>>>
>>> Also, please see my old answer from April below for the last problem you
>>> had with implementing the correct approach.
>>>
>>> Kevin
> 
> Hi Kevin,
> I'm sorry for not replying for so long. Please, give me some time (a day
> or two) so I could refresh everything and send the current state of the
> patches as well as the test case checking the issue

Hi Kevin,
The current state of the patches is available at 
https://github.com/denis-plotnikov/qemu/tree/postponed-request

I didn't manage to create an automatic reproducer but one of the patches 
contains a step-by-step description of how to reproduce the problem.

Please take a look. I'm ready to discuss the ways to improve it and will 
reply as fast as I can.

Thanks!

Denis


> 
> Denis
>>
>> He is not at work today, I think he'll be able to answer on Monday.
>>
>>
> 

-- 
Best,
Denis