Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
On 01.11.2014, at 13:45, Reimar Döffinger reimar.doeffin...@gmx.de wrote: This is a regression, previous code did not require users to set the profile to use VDPAU. In addition, ASP is a superset of SP, so there is not even a good reason to require it. Are there still objections? As said, to my knowledge every ASP decoder is required to be a fully working SP decoder, so I think this can't break anything with compliant implementations, which is my main justification (besides the regression part). Does anyone disagree? ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
Reimar Döffinger Reimar.Doeffinger at gmx.de writes: Are there still objections? I just wanted to know if we shouldn't always set ASP? Or at least for everything != SP? Carl Eugen ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
On Sun, Nov 02, 2014 at 10:02:50AM +, Carl Eugen Hoyos wrote: Reimar Döffinger Reimar.Doeffinger at gmx.de writes: Are there still objections? I just wanted to know if we shouldn't always set ASP? Or at least for everything != SP? I am unsure. It seemed to me this way would be more reliable and ensure we review it if we were to ever add a new profile. But if anyone has a real opinion on it I'll happily follow it. ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
[FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
This is a regression, previous code did not require users to set the profile to use VDPAU. In addition, ASP is a superset of SP, so there is not even a good reason to require it. --- libavcodec/vdpau_mpeg4.c | 1 + 1 file changed, 1 insertion(+) diff --git a/libavcodec/vdpau_mpeg4.c b/libavcodec/vdpau_mpeg4.c index 2a036b6..02ed4d1 100644 --- a/libavcodec/vdpau_mpeg4.c +++ b/libavcodec/vdpau_mpeg4.c @@ -119,6 +119,7 @@ static int vdpau_mpeg4_init(AVCodecContext *avctx) case FF_PROFILE_MPEG4_SIMPLE: profile = VDP_DECODER_PROFILE_MPEG4_PART2_SP; break; +case FF_PROFILE_UNKNOWN: case FF_PROFILE_MPEG4_ADVANCED_SIMPLE: profile = VDP_DECODER_PROFILE_MPEG4_PART2_ASP; break; -- 2.1.2 ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
On Sat, 1 Nov 2014 13:45:46 +0100 Reimar Döffinger reimar.doeffin...@gmx.de wrote: This is a regression, previous code did not require users to set the profile to use VDPAU. It still doesn't...? There are basically 3 ffmpeg vdpau hwaccel APIs now. In addition, ASP is a superset of SP, so there is not even a good reason to require it. --- libavcodec/vdpau_mpeg4.c | 1 + 1 file changed, 1 insertion(+) diff --git a/libavcodec/vdpau_mpeg4.c b/libavcodec/vdpau_mpeg4.c index 2a036b6..02ed4d1 100644 --- a/libavcodec/vdpau_mpeg4.c +++ b/libavcodec/vdpau_mpeg4.c @@ -119,6 +119,7 @@ static int vdpau_mpeg4_init(AVCodecContext *avctx) case FF_PROFILE_MPEG4_SIMPLE: profile = VDP_DECODER_PROFILE_MPEG4_PART2_SP; break; +case FF_PROFILE_UNKNOWN: case FF_PROFILE_MPEG4_ADVANCED_SIMPLE: profile = VDP_DECODER_PROFILE_MPEG4_PART2_ASP; break; ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
Reimar Döffinger Reimar.Doeffinger at gmx.de writes: +case FF_PROFILE_UNKNOWN: case FF_PROFILE_MPEG4_ADVANCED_SIMPLE: profile = VDP_DECODER_PROFILE_MPEG4_PART2_ASP; Could instead be the default case to set profile to ASP? Or to remove this whole logic? Carl Eugen ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
On 01 Nov, Reimar Döffinger wrote : This is a regression, previous code did not require users to set the profile to use VDPAU. If you don't set the profile, how can you know that the hardware supports it? -- Jean-Baptiste Kempf http://www.jbkempf.com/ - +33 672 704 734 Sent from my Electronic Device ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
Jean-Baptiste Kempf jb at videolan.org writes: On 01 Nov, Reimar Döffinger wrote : This is a regression, previous code did not require users to set the profile to use VDPAU. If you don't set the profile, how can you know that the hardware supports it? Does this have any relevance for ASP? Carl Eugen ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
On Sat, Nov 01, 2014 at 02:18:06PM +0100, Jean-Baptiste Kempf wrote: On 01 Nov, Reimar Döffinger wrote : This is a regression, previous code did not require users to set the profile to use VDPAU. If you don't set the profile, how can you know that the hardware supports it? Firstly: As I said, SP is a subset of ASP, so if the hardware supports ASP it supports SP. Secondly: The code actually interacting with the hardware can handle that. There is no need for the FFmpeg decoder to require that information. Obviously I can just hack around this by setting it to ASP if it's unknown, but I'd like a good reason why that is a better solution than graceful degradation here. Only slightly related: I generally dislike that part of the design as it mixes up the whole VDPAU structure/bitstream generation with the actual hardware decoding, which is a pain if you want to use security feature like process separation of decoding (which is a higher security risk) from actual display. ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
On Sat, Nov 01, 2014 at 01:52:17PM +0100, wm4 wrote: On Sat, 1 Nov 2014 13:45:46 +0100 Reimar Döffinger reimar.doeffin...@gmx.de wrote: This is a regression, previous code did not require users to set the profile to use VDPAU. It still doesn't...? It does if you want the hwaccel API. There are basically 3 ffmpeg vdpau hwaccel APIs now. (possibly 4 depending on how you could, but) Adding the newest one because of this code had the side-effect that you now need to set the profile even for the older ones (not for the oldest one using mpeg4_vdpau though it seems?). Obviously it would be possible to just skip this in the case of the older API, but as said I see no good reason. ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Re: [FFmpeg-devel] [PATCH] vdpau_mpeg4: Do not fail on unknown profile.
On Sat, 1 Nov 2014 15:13:17 +0100 Reimar Döffinger reimar.doeffin...@gmx.de wrote: On Sat, Nov 01, 2014 at 01:52:17PM +0100, wm4 wrote: On Sat, 1 Nov 2014 13:45:46 +0100 Reimar Döffinger reimar.doeffin...@gmx.de wrote: This is a regression, previous code did not require users to set the profile to use VDPAU. It still doesn't...? It does if you want the hwaccel API. There are basically 3 ffmpeg vdpau hwaccel APIs now. (possibly 4 depending on how you could, but) Adding the newest one because of this code had the side-effect that you now need to set the profile even for the older ones (not for the oldest one using mpeg4_vdpau though it seems?). Obviously it would be possible to just skip this in the case of the older API, but as said I see no good reason. That seems indeed like a bug, then. It would mean every API user got broken. Nice job, Libav. ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
