Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
 help / color / mirror / Atom feed
From: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
To: ffmpeg-devel@ffmpeg.org
Subject: Re: [FFmpeg-devel] [PATCH 08/12] lavc/frame_thread_encoder: avoid assigning a whole AVCodecContext
Date: Fri, 22 Mar 2024 21:42:11 +0100
Message-ID: <GV1P250MB073779A405CD87E3716807A78F312@GV1P250MB0737.EURP250.PROD.OUTLOOK.COM> (raw)
In-Reply-To: <20240322202841.31730-8-anton@khirnov.net>

Anton Khirnov:
> It is highly unsafe, as AVCodecContext contains many allocated fields.
> Copying information via AVCodecParameters and with av_opt_copy() should
> handle everything needed by thread workers.
> ---
>  libavcodec/frame_thread_encoder.c | 26 ++++++++++++++++++++------
>  1 file changed, 20 insertions(+), 6 deletions(-)
> 
> diff --git a/libavcodec/frame_thread_encoder.c b/libavcodec/frame_thread_encoder.c
> index cda5158117..744062b776 100644
> --- a/libavcodec/frame_thread_encoder.c
> +++ b/libavcodec/frame_thread_encoder.c
> @@ -28,6 +28,7 @@
>  #include "libavutil/thread.h"
>  #include "avcodec.h"
>  #include "avcodec_internal.h"
> +#include "codec_par.h"
>  #include "encode.h"
>  #include "internal.h"
>  #include "pthread_internal.h"
> @@ -121,6 +122,7 @@ av_cold int ff_frame_thread_encoder_init(AVCodecContext *avctx)
>      int i=0;
>      ThreadContext *c;
>      AVCodecContext *thread_avctx = NULL;
> +    AVCodecParameters *par = NULL;
>      int ret;
>  
>      if(   !(avctx->thread_type & FF_THREAD_FRAME)
> @@ -194,18 +196,27 @@ av_cold int ff_frame_thread_encoder_init(AVCodecContext *avctx)
>          }
>      }
>  
> +    par = avcodec_parameters_alloc();
> +    if (!par) {
> +        ret = AVERROR(ENOMEM);
> +        goto fail;
> +    }
> +
> +    ret = avcodec_parameters_from_context(par, avctx);
> +    if (ret < 0)
> +        goto fail;
> +
>      for(i=0; i<avctx->thread_count ; i++){
> -        void *tmpv;
>          thread_avctx = avcodec_alloc_context3(avctx->codec);
>          if (!thread_avctx) {
>              ret = AVERROR(ENOMEM);
>              goto fail;
>          }
> -        tmpv = thread_avctx->priv_data;
> -        *thread_avctx = *avctx;
> -        thread_avctx->priv_data = tmpv;
> -        thread_avctx->internal = NULL;
> -        thread_avctx->hw_frames_ctx = NULL;
> +
> +        ret = avcodec_parameters_to_context(thread_avctx, par);
> +        if (ret < 0)
> +            goto fail;
> +
>          ret = av_opt_copy(thread_avctx, avctx);
>          if (ret < 0)
>              goto fail;
> @@ -227,10 +238,13 @@ av_cold int ff_frame_thread_encoder_init(AVCodecContext *avctx)
>          }
>      }
>  
> +    avcodec_parameters_free(&par);
> +
>      avctx->active_thread_type = FF_THREAD_FRAME;
>  
>      return 0;
>  fail:
> +    avcodec_parameters_free(&par);
>      ff_codec_close(thread_avctx);
>      av_freep(&thread_avctx);
>      avctx->thread_count = i;

IIRC the mjpeg encoders use intra_matrix and this will now no longer be
copied to the worker threads.

- 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".

  reply	other threads:[~2024-03-22 20:42 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-03-22 20:28 [FFmpeg-devel] [PATCH 01/12] tests/fate/ffmpeg: evaluate thread count in fate-run.sh rather than make Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 02/12] fftools/ffmpeg_demux: only call filter_codec_opts() when we have a decoder Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 03/12] fftools/cmdutils: do not use a random codec's private options Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 04/12] fftools/ffmpeg_dec: apply decoder options manually Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 05/12] fftools/ffmpeg_{demux, dec}: pass -bitexact through DecoderFlags Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 06/12] fftools/ffmpeg_dec: apply cropping manually Anton Khirnov
2024-03-23  2:53   ` James Almer
2024-03-23  7:20     ` Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 07/12] fftools/ffmpeg_filter: remove display matrix if we have applied it Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 08/12] lavc/frame_thread_encoder: avoid assigning a whole AVCodecContext Anton Khirnov
2024-03-22 20:42   ` Andreas Rheinhardt [this message]
2024-03-23 14:00     ` [FFmpeg-devel] [PATCH v2 " Anton Khirnov
2024-03-23 14:11       ` Andreas Rheinhardt
2024-03-24  9:27         ` Anton Khirnov
2024-03-24 10:19           ` Andreas Rheinhardt
2024-03-24 10:33             ` Anton Khirnov
2024-03-24 11:10               ` Andreas Rheinhardt
2024-03-24 11:23                 ` Anton Khirnov
2024-03-24  0:11   ` [FFmpeg-devel] [PATCH " Michael Niedermayer
2024-03-24  9:29     ` [FFmpeg-devel] [PATCH v3 " Anton Khirnov
2024-03-24 11:25       ` Andreas Rheinhardt
2024-03-24 11:36         ` Anton Khirnov
2024-03-27 10:27   ` [FFmpeg-devel] [PATCH v4 " Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 09/12] lavc/decode: move sd_global_map to avcodec Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 10/12] lavc/encode: map AVCodecContext.decoded_side_data to coded_side_data Anton Khirnov
2024-03-23  2:33   ` James Almer
2024-03-23  7:20     ` Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 11/12] fftools/ffmpeg_enc: stop copying demuxer side data to the muxer Anton Khirnov
2024-03-22 20:28 ` [FFmpeg-devel] [PATCH 12/12] fftools/ffmpeg_demux: make InputStream.autorotate private Anton Khirnov

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=GV1P250MB073779A405CD87E3716807A78F312@GV1P250MB0737.EURP250.PROD.OUTLOOK.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