[FFmpeg-devel] [PATCH] ffprobe: restore reporting error code for failed inputs

[Marton Balint]

On Wed, 17 Aug 2022, Gyan Doshi wrote:

>
>
> On 2022-08-17 04:36 am, Stefano Sabatini wrote:
>>  On date Tuesday 2022-08-16 00:04:10 +0530, Gyan Doshi wrote:
>>>  c11fb46731 led to a regression whereby the return code for missing
>>>  input or input probe is overridden by writer close return code and
>>>  hence not conveyed in the exit code.
>>>  ---
>>>    fftools/ffprobe.c | 6 +++++-
>>>    1 file changed, 5 insertions(+), 1 deletion(-)
>>>
>>>  Affects 5.1 so will need to be backported there.
>>>
>>>  diff --git a/fftools/ffprobe.c b/fftools/ffprobe.c
>>>  index ad633ccc44..8983dc28cc 100644
>>>  --- a/fftools/ffprobe.c
>>>  +++ b/fftools/ffprobe.c
>>> @@  -4032,7 +4032,7 @@ int main(int argc, char **argv)
>>>        WriterContext *wctx;
>>>        char *buf;
>>>        char *w_name = NULL, *w_args = NULL;
>>>  -    int ret, i;
>>>  +    int ret, input_ret, i;
>>>
>>>        init_dynload();
>>> 
>>> @@  -4156,10 +4156,14 @@ int main(int argc, char **argv)
>>>                    show_error(wctx, ret);
>>>            }
>>>    +        input_ret = ret;
>>>  +
>>>            writer_print_section_footer(wctx);
>>>            ret = writer_close(&wctx);
>>>            if (ret < 0)
>>>                av_log(NULL, AV_LOG_ERROR, "Writing output failed: %s\n",
>>>  av_err2str(ret));
>>>  +
>>>  +        ret = FFMIN(ret, input_ret);
>>  maybe we should give priority to input_ret in case they are both
>>  negative?
>>
>>  return input_ret < 0 ? input_ret : ret;
>
> Scripts usually depend on exit code being 0 or something else. Also, error is 
> logged for both input failure and writer_close failure, so it doesn't matter.
> Finally, input_ret is not initialized if writer_open fails, so we shouldn't 
> be referencing it outside.

I would do something like

ret2 = writer_close();
ret = FFMIN(ret2, ret);

But fine either way, please push any version you prefer, this has been 
broken for too long. Sorry about that.

Thanks,
Marton