From: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
To: ffmpeg-devel@ffmpeg.org
Subject: Re: [FFmpeg-devel] [PATCH] avcodec/avcodec: don't free AVOption settable fields in avcodec_close()
Date: Mon, 21 Mar 2022 00:34:06 +0100
Message-ID: <AS1PR01MB9564B116D5A469BB7B77C9D58F159@AS1PR01MB9564.eurprd01.prod.exchangelabs.com> (raw)
In-Reply-To: <66b6e472-d8fd-f544-8256-e3b7aadb6bb3@gmail.com>
James Almer:
>
>
> On 3/20/2022 8:26 PM, Andreas Rheinhardt wrote:
>> James Almer:
>>> It can uninitialize fields that may still be used after the context
>>> was closed,
>>> so do it instead in avcodec_free_context().
>>>
>>> Signed-off-by: James Almer <jamrial@gmail.com>
>>> ---
>>> libavcodec/avcodec.c | 1 -
>>> libavcodec/options.c | 2 +-
>>> 2 files changed, 1 insertion(+), 2 deletions(-)
>>>
>>> diff --git a/libavcodec/avcodec.c b/libavcodec/avcodec.c
>>> index 38bdaad4fa..122d09b63a 100644
>>> --- a/libavcodec/avcodec.c
>>> +++ b/libavcodec/avcodec.c
>>> @@ -524,7 +524,6 @@ av_cold int avcodec_close(AVCodecContext *avctx)
>>> if (avctx->priv_data && avctx->codec &&
>>> avctx->codec->priv_class)
>>> av_opt_free(avctx->priv_data);
>>> - av_opt_free(avctx);
>>> av_freep(&avctx->priv_data);
>>> if (av_codec_is_encoder(avctx->codec)) {
>>> av_freep(&avctx->extradata);
>>> diff --git a/libavcodec/options.c b/libavcodec/options.c
>>> index 33f11480a7..91335415c1 100644
>>> --- a/libavcodec/options.c
>>> +++ b/libavcodec/options.c
>>> @@ -172,7 +172,7 @@ void avcodec_free_context(AVCodecContext **pavctx)
>>> av_freep(&avctx->intra_matrix);
>>> av_freep(&avctx->inter_matrix);
>>> av_freep(&avctx->rc_override);
>>> - av_channel_layout_uninit(&avctx->ch_layout);
>>> + av_opt_free(avctx);
>>> av_freep(pavctx);
>>> }
>>
>> This will lead to memleaks for users that use avcodec_close(avctx) +
>> av_free(avctx) to free an AVCodecContext (e.g. our frame-threaded
>> encoders do this). Notice that avcodec_free_context() violates the
>> documentation of AVCodecContext.extradata (documented to not be freed
>> for decoders) and AVCodecContext.subtitle_header and
>> AVCodecContext.rc_override (documented to not be freed by lavc for
>> encoders), so there is a reason for using it instead of
>> avcodec_free_context() (even when not reusing the context).
>
> That's an absolute mess of a situation. av_free(avctx) should not be an
> allowed or supported scenario when avcodec_free_context() exists. And
> why is the latter violating its own documentation?
>
It is not violating its own documentation, but the documentation of the
relevant AVCodecContext fields. IIRC Anton wanted a function that just
frees the whole context, even if this meant that fields which are
documented as being owned by the user are freed. Even documenting the
current state of affairs in avcodec.h doesn't change the fact that there
is a valid reason to use avcodec_close()+av_free(), so we can't pretend
it doesn't happen.
- Andreas
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
next prev parent reply other threads:[~2022-03-20 23:34 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-03-20 23:18 James Almer
2022-03-20 23:26 ` Andreas Rheinhardt
2022-03-20 23:29 ` James Almer
2022-03-20 23:34 ` Andreas Rheinhardt [this message]
2022-03-20 23:41 ` James Almer
2022-03-21 0:04 ` Marton Balint
2022-03-21 0:05 ` Marton Balint
2022-03-21 0:15 ` James Almer
2022-03-21 0:21 ` Marton Balint
2022-03-21 0:23 ` James Almer
2022-03-21 0:16 ` Andreas Rheinhardt
2022-03-21 0:46 ` Andreas Rheinhardt
2022-03-21 11:17 ` James Almer
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=AS1PR01MB9564B116D5A469BB7B77C9D58F159@AS1PR01MB9564.eurprd01.prod.exchangelabs.com \
--to=andreas.rheinhardt@outlook.com \
--cc=ffmpeg-devel@ffmpeg.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
This inbox may be cloned and mirrored by anyone:
git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git
# If you have public-inbox 1.1+ installed, you may
# initialize and index your mirror using the following commands:
public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \
ffmpegdev@gitmailbox.com
public-inbox-index ffmpegdev
Example config snippet for mirrors.
AGPL code for this site: git clone https://public-inbox.org/public-inbox.git