Re: [PATCH 06/10] btrfs: add helper to kill background process running _btrfs_stress_remount_compress

[Anand Jain <anand.jain@oracle.com>]

On 3/28/24 20:03, Filipe Manana wrote:
> On Thu, Mar 28, 2024 at 11:46 AM Anand Jain <anand.jain@oracle.com> wrote:
>>
>>
>>>> diff --git a/tests/btrfs/071 b/tests/btrfs/071
>>>> index 6ebbd8cc..7ba15390 100755
>>>> --- a/tests/btrfs/071
>>>> +++ b/tests/btrfs/071
>>>> @@ -58,17 +58,15 @@ run_test()
>>>>        echo "$remount_pid" >>$seqres.full
>>
>>
>>
>>>>        echo "Wait for fsstress to exit and kill all background workers"
>>>>>> $seqres.full
>>>> -    wait $fsstress_pid
>>>> -    kill $replace_pid $remount_pid
>>>> -    wait
>>>> +    kill $replace_pid
>>>> +    wait $fsstress_pid $replace_pid
>>

>> The change first kills the replace and then wait for fsstress. Was this
>> intentional?
> 

> It wasn't.

Ok. Thanks for confirming. I will fix it.

-Anand


> But after all patches applied, everything's correct, we
> wait for fsstress to finish and then kill replace.
> Do you want me to fix that?
> 
> The sequence will have to be:
> 
> wait $fsstress_pid
> kill $replace_pid
> wait $replace_pid
> 
> Thanks.
> 
>>
>> This patch is causing a regression. The replace never gets killed, as
>> the echo-comment specifically states to wait for fsstress and then kill
>> the replace. Following it fixes the issue.
>>
>> Which the patch 7/10 reversed the order to fix. But why?
>>
>> Thanks, Anand
>>
>>
>>>> -    # wait for the remount and replace operations to finish
>>>> +    # wait for the replace operationss to finish
>>
>>
>>
>>
>>>>        while ps aux | grep "replace start" | grep -qv grep; do
>>>>            sleep 1
>>>>        done
>>>> -    while ps aux | grep "mount.*$SCRATCH_MNT" | grep -qv grep; do
>>>> -        sleep 1
>>>> -    done
>>>> +
>>>> +    _btrfs_kill_stress_remount_compress_pid $remount_pid $SCRATCH_MNT
>>>>        echo "Scrub the filesystem" >>$seqres.full
>>>>        $BTRFS_UTIL_PROG scrub start -B $SCRATCH_MNT >>$seqres.full 2>&1