* [FFmpeg-devel] [PATCH] avcodec/pthread_frame: sync decoded_side_data arrays with all worker threads
@ 2025-04-26 23:41 James Almer
2025-04-27 1:41 ` [FFmpeg-devel] [PATCH 2/2] avcodec/decode: sync side_data_pref_mask across " James Almer
0 siblings, 1 reply; 4+ messages in thread
From: James Almer @ 2025-04-26 23:41 UTC (permalink / raw)
To: ffmpeg-devel
After being filled by FFCodec->init() from the first worker thread, it needs to
be copied to every other thread context. Otherwise, its entries will not be
present when ff_decode_frame_props() populates output frames.
Signed-off-by: James Almer <jamrial@gmail.com>
---
libavcodec/pthread_frame.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/libavcodec/pthread_frame.c b/libavcodec/pthread_frame.c
index 4ea5dd3698..7019108af6 100644
--- a/libavcodec/pthread_frame.c
+++ b/libavcodec/pthread_frame.c
@@ -905,6 +905,14 @@ static av_cold int init_thread(PerThreadContext *p, int *threads_to_free,
if (err < 0)
return err;
}
+ } else {
+ for (int i = 0; i < avctx->nb_decoded_side_data; i++) {
+ err = av_frame_side_data_clone(©->decoded_side_data,
+ ©->nb_decoded_side_data,
+ avctx->decoded_side_data[i], 0);
+ if (err < 0)
+ return err;
+ }
}
atomic_init(&p->debug_threads, (copy->debug & FF_DEBUG_THREADS) != 0);
--
2.49.0
_______________________________________________
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".
^ permalink raw reply [flat|nested] 4+ messages in thread
* [FFmpeg-devel] [PATCH 2/2] avcodec/decode: sync side_data_pref_mask across worker threads
2025-04-26 23:41 [FFmpeg-devel] [PATCH] avcodec/pthread_frame: sync decoded_side_data arrays with all worker threads James Almer
@ 2025-04-27 1:41 ` James Almer
2025-04-27 18:52 ` Andreas Rheinhardt
0 siblings, 1 reply; 4+ messages in thread
From: James Almer @ 2025-04-27 1:41 UTC (permalink / raw)
To: ffmpeg-devel
side_data_pref_mask is set during ff_decode_preinit() and then never copied to
worker threads. As such, any decoder calling one of the side data helpers that
rely on side_data_pref() in frame threaded scenarios will ignore the user set
list of side data types.
Signed-off-by: James Almer <jamrial@gmail.com>
---
libavcodec/decode.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavcodec/decode.c b/libavcodec/decode.c
index c2b2dd6e3b..20e99b3ffb 100644
--- a/libavcodec/decode.c
+++ b/libavcodec/decode.c
@@ -2209,6 +2209,7 @@ void ff_decode_internal_sync(AVCodecContext *dst, const AVCodecContext *src)
const DecodeContext *src_dc = decode_ctx(src->internal);
DecodeContext *dst_dc = decode_ctx(dst->internal);
+ dst_dc->side_data_pref_mask = src_dc->side_data_pref_mask;
av_refstruct_replace(&dst_dc->lcevc, src_dc->lcevc);
}
--
2.49.0
_______________________________________________
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".
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [FFmpeg-devel] [PATCH 2/2] avcodec/decode: sync side_data_pref_mask across worker threads
2025-04-27 1:41 ` [FFmpeg-devel] [PATCH 2/2] avcodec/decode: sync side_data_pref_mask across " James Almer
@ 2025-04-27 18:52 ` Andreas Rheinhardt
2025-04-27 19:19 ` James Almer
0 siblings, 1 reply; 4+ messages in thread
From: Andreas Rheinhardt @ 2025-04-27 18:52 UTC (permalink / raw)
To: ffmpeg-devel
James Almer:
> side_data_pref_mask is set during ff_decode_preinit() and then never copied to
> worker threads. As such, any decoder calling one of the side data helpers that
> rely on side_data_pref() in frame threaded scenarios will ignore the user set
> list of side data types.
>
> Signed-off-by: James Almer <jamrial@gmail.com>
> ---
> libavcodec/decode.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/libavcodec/decode.c b/libavcodec/decode.c
> index c2b2dd6e3b..20e99b3ffb 100644
> --- a/libavcodec/decode.c
> +++ b/libavcodec/decode.c
> @@ -2209,6 +2209,7 @@ void ff_decode_internal_sync(AVCodecContext *dst, const AVCodecContext *src)
> const DecodeContext *src_dc = decode_ctx(src->internal);
> DecodeContext *dst_dc = decode_ctx(dst->internal);
>
> + dst_dc->side_data_pref_mask = src_dc->side_data_pref_mask;
> av_refstruct_replace(&dst_dc->lcevc, src_dc->lcevc);
> }
>
Why is this done for every frame, even though it does not change after
preinit?
- 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".
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [FFmpeg-devel] [PATCH 2/2] avcodec/decode: sync side_data_pref_mask across worker threads
2025-04-27 18:52 ` Andreas Rheinhardt
@ 2025-04-27 19:19 ` James Almer
0 siblings, 0 replies; 4+ messages in thread
From: James Almer @ 2025-04-27 19:19 UTC (permalink / raw)
To: ffmpeg-devel
[-- Attachment #1.1.1: Type: text/plain, Size: 1232 bytes --]
On 4/27/2025 3:52 PM, Andreas Rheinhardt wrote:
> James Almer:
>> side_data_pref_mask is set during ff_decode_preinit() and then never copied to
>> worker threads. As such, any decoder calling one of the side data helpers that
>> rely on side_data_pref() in frame threaded scenarios will ignore the user set
>> list of side data types.
>>
>> Signed-off-by: James Almer <jamrial@gmail.com>
>> ---
>> libavcodec/decode.c | 1 +
>> 1 file changed, 1 insertion(+)
>>
>> diff --git a/libavcodec/decode.c b/libavcodec/decode.c
>> index c2b2dd6e3b..20e99b3ffb 100644
>> --- a/libavcodec/decode.c
>> +++ b/libavcodec/decode.c
>> @@ -2209,6 +2209,7 @@ void ff_decode_internal_sync(AVCodecContext *dst, const AVCodecContext *src)
>> const DecodeContext *src_dc = decode_ctx(src->internal);
>> DecodeContext *dst_dc = decode_ctx(dst->internal);
>>
>> + dst_dc->side_data_pref_mask = src_dc->side_data_pref_mask;
>> av_refstruct_replace(&dst_dc->lcevc, src_dc->lcevc);
>> }
>>
>
> Why is this done for every frame, even though it does not change after
> preinit?
I think the ff_decode_internal_sync() call in
update_context_from_thread() from pthread_frame.c could be removed?
[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 495 bytes --]
[-- Attachment #2: Type: text/plain, Size: 251 bytes --]
_______________________________________________
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".
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2025-04-27 19:19 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-04-26 23:41 [FFmpeg-devel] [PATCH] avcodec/pthread_frame: sync decoded_side_data arrays with all worker threads James Almer
2025-04-27 1:41 ` [FFmpeg-devel] [PATCH 2/2] avcodec/decode: sync side_data_pref_mask across " James Almer
2025-04-27 18:52 ` Andreas Rheinhardt
2025-04-27 19:19 ` James Almer
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