[FFmpeg-user] compiler warnings current snapshot

Carl Eugen Hoyos ceffmpeg at gmail.com
Mon Jan 15 04:58:11 EET 2018


2018-01-15 3:53 GMT+01:00 Reindl Harald <h.reindl at thelounge.net>:
>
>
> Am 15.01.2018 um 03:43 schrieb Carl Eugen Hoyos:
>>
>> 2018-01-15 3:41 GMT+01:00 Reindl Harald <h.reindl at thelounge.net>:
>>>
>>>
>>> Am 15.01.2018 um 03:38 schrieb Carl Eugen Hoyos:
>>>>
>>>> 2018-01-15 3:10 GMT+01:00 Reindl Harald <h.reindl at thelounge.net>:
>>>>
>>>>> src/libavformat/tls_openssl.c: In function 'ff_openssl_deinit':
>>>>> src/libavformat/tls_openssl.c:105:43: warning: the comparison will
>>>>> always
>>>>> evaluate as 'false' for the address of 'openssl_lock' will never be
>>>>> NULL
>>>>> [-Waddress]
>>>>>            if (CRYPTO_get_locking_callback() == openssl_lock) {
>>>>
>>>> I suspect this is related to a particular build environment.
>>>> (Does not happen everywhere, likely cannot be fixed.)
>>>
>>> openssl-1.1.0g-1.fc26.x86_64
>>
>> And?
>> Do you suggest an "#if" to disable to comparison?
>
> in doubt yes when it's useless on newer versions

I am not convinced this is a "newer version" issue but
please feel free to send such a patch. Make sure it is
not too ugly...

> as said the goal should be
> zero warnings and after that every involvd developer can enable all warnings
> and face little mistakes at the time when they happen

I believe you (still) misunderstand:
All except one warnings you quoted are compiler bugs, the one
warning is shown for a non-default build and is definitely
not a security issue. (For a second one, I am not sure if the
issue is in FFmpeg or gcc: It is not trivial to understand and
since it warns about a non-optimization in not speed-relevant
code, it is hardly important.)
Note that you forgot to quote the only grave issue that
gcc indicates...
(Another gcc bug)

Carl Eugen


More information about the ffmpeg-user mailing list