[FFmpeg-devel] [PATCH] configure: remove false positives with valgrind-memcheck

[James Almer]

On 6/13/2024 2:00 PM, Pierre-Anthony Lemieux wrote:
> On Sun, Jun 9, 2024 at 9:00 AM Pierre-Anthony Lemieux <pal at sandflow.com> wrote:
>>
>> On Sun, Jun 9, 2024 at 6:41 AM James Almer <jamrial at gmail.com> wrote:
>>>
>>> On 6/9/2024 2:56 AM, Pierre-Anthony Lemieux wrote:
>>>> On Sat, Jun 8, 2024 at 6:13 PM James Almer <jamrial at gmail.com> wrote:
>>>>>
>>>>> On 6/8/2024 9:59 PM, pal at sandflow.com wrote:
>>>>>> From: Pierre-Anthony Lemieux <pal at palemieux.com>
>>>>>>
>>>>>> If  `--toolchain=valgrind-memcheck` is selected, fate-lavf-asf fails on a false positive.
>>>>>
>>>>> I can't reproduce this. Tried with both --disable-optimizations and
>>>>> without it, on Ubuntu 22.04 x86_64.
>>>>> How are you getting these false positives?
>>>>
>>>> Command and log at https://pastebin.com/3TKk0RF4
>>>
>>> But that command is missing all the extra args configure adds to the
>>> invocation, before and after your patch.
> 
> Quick note to check if I have addressed your comment.
> 
>>
>> Full story.
>>
>> Before the proposed patch, "./configure --toolchain=valgrind-memcheck;
>> make fate-lavf-asf" results in "Conditional jump or move depends on
>> uninitialised value(s)" being emitted on my system (gcc (Ubuntu
>> 9.4.0-1ubuntu1~20.04.1) 9.4.0).
>>
>> The proposed patch only adds  "--expensive-definedness-checks=yes" to
>> the "valgrind-memcheck" command line -- this was the result of a
>> discussion on the mailing list.
>>
>> After the proposed patch, "./configure --toolchain=valgrind-memcheck;
>> make fate-lavf-asf" does not result in any error being emitted.
>>
>> The pastebin is intended to isolate the issue/solution. On my system,
>> adding/removing "--expensive-definedness-checks=yes" to the command
>> line removes/triggers the issue.
>>
>> Let me know if you need additional information.
>>
>> Thanks for reviewing.

I still can't reproduce this issue. Valgrind does not complain about 
this test at all for me no matter what settings do i use.

The name --expensive-definedness-checks=yes implies that Valgrind will 
do extra work that will slow the process even more than it already is. 
I'd like someone else's opinion about it (Andreas maybe, since he 
suggested this change).

>>
>>>
>>>>
>>>>>
>>>>>> ---
>>>>>>     configure | 2 +-
>>>>>>     1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>>
>>>>>> diff --git a/configure b/configure
>>>>>> index 6c5b8aab9a..c708f2e38c 100755
>>>>>> --- a/configure
>>>>>> +++ b/configure
>>>>>> @@ -4571,7 +4571,7 @@ case "$toolchain" in
>>>>>>                     target_exec_args="--tool=massif --alloc-fn=av_malloc --alloc-fn=av_mallocz --alloc-fn=av_calloc --alloc-fn=av_fast_padded_malloc --alloc-fn=av_fast_malloc --alloc-fn=av_realloc_f --alloc-fn=av_fast_realloc --alloc-fn=av_realloc"
>>>>>>                     ;;
>>>>>>                 valgrind-memcheck)
>>>>>> -                target_exec_args="--error-exitcode=1 --malloc-fill=0x2a --track-origins=yes --leak-check=full --gen-suppressions=all --suppressions=$source_path/tests/fate-valgrind.supp"
>>>>>> +                target_exec_args="--error-exitcode=1 --expensive-definedness-checks=yes --malloc-fill=0x2a --track-origins=yes --leak-check=full --gen-suppressions=all --suppressions=$source_path/tests/fate-valgrind.supp"
>>>>>>                     ;;
>>>>>>             esac
>>>>>>         ;;
>>>>> _______________________________________________
>>>>> ffmpeg-devel mailing list
>>>>> ffmpeg-devel at ffmpeg.org
>>>>> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>>>>>
>>>>> To unsubscribe, visit link above, or email
>>>>> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
>>>> _______________________________________________
>>>> ffmpeg-devel mailing list
>>>> ffmpeg-devel at ffmpeg.org
>>>> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>>>>
>>>> To unsubscribe, visit link above, or email
>>>> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
>>> _______________________________________________
>>> ffmpeg-devel mailing list
>>> ffmpeg-devel at ffmpeg.org
>>> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>>>
>>> To unsubscribe, visit link above, or email
>>> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".