Hi,
On Mon, Jun 5, 2017 at 7:23 AM, James Darnley wrote:
> I forgot to mention in my cover letter that although the dct test
> passes, fate does not. As I mentioned on IRC, changing them causes
> errors elsewhere in fate. I am currently looking into this problem and
> I'm
Hi,
On Sun, Jun 4, 2017 at 1:52 PM, Ilia Valiakhmetov
wrote:
> vp9_diag_downleft_32x32_8bpp_c: 580.2
> vp9_diag_downleft_32x32_8bpp_sse2: 75.6
> vp9_diag_downleft_32x32_8bpp_ssse3: 73.7
> vp9_diag_downleft_32x32_8bpp_avx: 72.7
> vp9_diag_downleft_32x32_10bpp_c: 1101.2
>
Hi,
On Sun, Jun 4, 2017 at 6:32 AM, Nicolas George wrote:
> Le quintidi 15 prairial, an CCXXV, James Almer a écrit :
> > On 6/3/2017 9:25 PM, Michael Niedermayer wrote:
> > I see EINVAL more fit as an error for example if src or dst are NULL,
> > something that would actually
Hi,
On Thu, Jun 1, 2017 at 2:27 AM, raymond zheng
wrote:
> This patch is only for solving the bug introduced by chunksize has been
> initial set as UINT64 MAX.
>
I understand. So what will happen if the server responds that the chunksize
is really actually
Hi,
On May 30, 2017 6:44 AM, "James Darnley" wrote:
On 2017-05-29 16:51, James Darnley wrote:
> Commit message: reindent
Is this acceptable? Should I be more verbose?
Make it x86/idctdsp_init: reindent. Otherwise ok.
Thanks,
Ronald
Hi,
On Mon, May 29, 2017 at 5:26 PM, Michael Niedermayer wrote:
> On Mon, May 29, 2017 at 09:40:49PM +0200, James Darnley wrote:
> > On 2017-05-29 16:51, James Darnley wrote:
> > > ---
> > > Changes:
> > > - Changed type of d4 constant to dwords because it gets
Hi,
On Sun, May 28, 2017 at 5:04 PM, Rostislav Pehlivanov
wrote:
> The native Dirac decoder supports everything the library has and encoding
> support is still provided via the native vc2 (Dirac Pro) encoder.
Isn't the native encoder intra-only, though?
(Not that I want
Hi,
On Fri, May 26, 2017 at 10:36 PM, raymond zheng
wrote:
> Hi:
> I find an issue about http. I don't use chunked, so s->chunksize
> will be set as UINT64_MAX when http open, but because of "if
> (s->chunksize > 0) s->chunksize -= len;" then chunksize will not
Hi,
On Fri, May 26, 2017 at 6:34 PM, Michael Niedermayer <mich...@niedermayer.cc
> wrote:
> On Fri, May 26, 2017 at 06:07:35PM -0400, Ronald S. Bultje wrote:
> > Hi,
> >
> > On Fri, May 26, 2017 at 5:50 PM, Michael Niedermayer
> <mich...@niedermayer.cc
> >
Hi,
On Fri, May 26, 2017 at 5:50 PM, Michael Niedermayer wrote:
> On Fri, May 26, 2017 at 11:18:12PM +0200, Hendrik Leppkes wrote:
> > On Fri, May 26, 2017 at 11:11 PM, Michael Niedermayer
> > wrote:
> > > On Fri, May 26, 2017 at 03:20:14PM
Hi,
On Fri, May 26, 2017 at 6:38 AM, wrote:
> From: Shivraj Patil
>
> Signed-off-by: Shivraj Patil
> ---
> libavcodec/mips/Makefile|2 ++
> libavcodec/mips/blockdsp_init_mips.c|8
Should fix tsan errors in utvideoenc_rgb_left and related tests.
---
libavcodec/frame_thread_encoder.c | 13 +++--
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/libavcodec/frame_thread_encoder.c
b/libavcodec/frame_thread_encoder.c
index 27ae356..6cf1a68 100644
---
Hi,
On May 24, 2017 12:51 PM, "Clément Bœsch" <u...@pkh.me> wrote:
On Wed, May 24, 2017 at 12:15:05PM -0400, Ronald S. Bultje wrote:
> Should fix tsan errors in utvideoenc_rgb_left and related tests.
> ---
> libavcodec/frame_thread_encoder.c | 13 ++---
> 1
Should fix tsan errors in utvideoenc_rgb_left and related tests.
---
libavcodec/frame_thread_encoder.c | 13 ++---
1 file changed, 10 insertions(+), 3 deletions(-)
diff --git a/libavcodec/frame_thread_encoder.c
b/libavcodec/frame_thread_encoder.c
index 27ae356..2746964 100644
---
These variables are shared between frame threads, but they are updated
post-setup_finished() if a EOB/EOS slice type occurs. Moving the EOB/EOS
slices to the next frame thread instance (by parsing them leading into
the next picture instead of trailing behind the last picture) effectively
prevents
Hi,
On Sun, May 21, 2017 at 6:57 AM, Michael Niedermayer <mich...@niedermayer.cc
> wrote:
> On Sun, May 21, 2017 at 12:51:56PM +0200, Michael Niedermayer wrote:
> > On Sat, May 20, 2017 at 10:35:34PM -0400, Ronald S. Bultje wrote:
> > > Hi,
> > >
> > >
Hi,
On Sat, May 20, 2017 at 8:12 PM, Michael Niedermayer wrote:
> Fixes: 1710/clusterfuzz-testcase-minimized-4837032931098624
>
> Found-by: continuous fuzzing process https://github.com/google/oss-
> fuzz/tree/master/projects/ffmpeg
> Signed-off-by: Michael Niedermayer
Hi,
cursory review, I appreciate that this is work in progress so don't read
too much into it.
General points:
- I really like how you're doing a literal translation without trying to
change it, this makes review easier and also reduces the likelihood that
there's performance implications.
- do
Hi,
On Tue, May 16, 2017 at 7:43 PM, Muhammad Faiz <mfc...@gmail.com> wrote:
> On Wed, May 17, 2017 at 6:33 AM, Ronald S. Bultje <rsbul...@gmail.com>
> wrote:
> > Hi,
> >
> > On Tue, May 16, 2017 at 7:23 PM, Muhammad Faiz <mfc...@gmail.com> wrote:
> &
Hi,
On Tue, May 16, 2017 at 7:23 PM, Muhammad Faiz wrote:
> On Sun, May 7, 2017 at 8:18 PM, Nicolas George wrote:
> > L'octidi 18 floréal, an CCXXV, James Almer a écrit :
> >> Will that complete fix be ready and be backport-enabled (AKA, no new
> >> API) in
Hi,
On Mon, May 15, 2017 at 6:53 AM, wm4 wrote:
> On Mon, 15 May 2017 12:40:23 +0200
> Carl Eugen Hoyos wrote:
> > Change-Id: I1cbf000a1bda7b3abf0a84e971e752f176857385
>
> what is this?
Looks like a gerrit (patch review system) patch marker, see e.g.:
Hi,
On Mon, May 15, 2017 at 2:01 AM, Shivraj Patil
wrote:
> Can anyone review the patch please?
>
[..]
> --- a/configure
> +++ b/configure
> @@ -5357,6 +5357,10 @@ elif enabled mips; then
> enabled mipsdsp && check_inline_asm_flags mipsdsp '"addu.qb $t0, $t1,
>
Hi,
On Sun, May 14, 2017 at 6:07 AM, Nicolas George wrote:
> Le duodi 22 floréal, an CCXXV, Michael Niedermayer a écrit :
> > The direct value like 32 feels more natural to me too, but iam fine
> > with either.
> >
> > The avoidance of log() might also favor the other in some
Hi,
On Thu, May 11, 2017 at 10:24 AM, Ronald S. Bultje <rsbul...@gmail.com>
wrote:
> Hi,
>
> On Thu, May 11, 2017 at 9:55 AM, riya khanna <riyakhanna1...@gmail.com>
> wrote:
>
>> Hi,
>>
>> We are a group of PhD students at Georgia Tech carrying
Hi,
On Thu, May 11, 2017 at 9:55 AM, riya khanna
wrote:
> Hi,
>
> We are a group of PhD students at Georgia Tech carrying out a research
> project that explores free software violations by mobile apps. We found
> that over 100 Android apps use GPL version of FFMpeg.
>
Hi,
On Wed, May 10, 2017 at 5:02 PM, Michael Niedermayer wrote:
> On Wed, May 10, 2017 at 10:28:22PM +0200, Nicolas George wrote:
> > Le primidi 21 floréal, an CCXXV, Michael Niedermayer a écrit :
> > > fixed, and now your patchset seems to cause a 1% slowdown of the
>
Hi,
On Wed, May 10, 2017 at 2:11 PM, Ronald S. Bultje <rsbul...@gmail.com>
wrote:
> - return 0, not AVERROR_INVALIDDATA, since alt-ref frames aren't decoding
> errors.
>
I'm also not sure if alt-ref frames in webp (since it's intra-only) make
any sense.
Hi,
On Wed, May 10, 2017 at 12:37 PM, Michael Niedermayer <
mich...@niedermayer.cc> wrote:
> Fixes: 1470/clusterfuzz-testcase-minimized-5404421666111488
> Fixes: 1472/clusterfuzz-testcase-minimized-5677426430443520
>
> Signed-off-by: Michael Niedermayer
> ---
>
Hi,
On Wed, May 10, 2017 at 12:37 PM, Michael Niedermayer <
mich...@niedermayer.cc> wrote:
> Fixes: out of array access
> Fixes: 1434/clusterfuzz-testcase-minimized-6314998085189632
> Fixes: 1435/clusterfuzz-testcase-minimized-6483783723253760
>
> Found-by: continuous fuzzing process
Hi,
On Tue, May 9, 2017 at 9:24 PM, Michael Niedermayer <mich...@niedermayer.cc>
wrote:
> On Tue, May 09, 2017 at 09:08:08PM -0400, Ronald S. Bultje wrote:
> > Hi,
> >
> > On Tue, May 9, 2017 at 8:37 PM, Michael Niedermayer
> <mich...@niedermayer.cc>
>
Hi,
On Tue, May 9, 2017 at 9:24 PM, Michael Niedermayer <mich...@niedermayer.cc>
wrote:
> On Tue, May 09, 2017 at 09:08:08PM -0400, Ronald S. Bultje wrote:
> > Hi,
> >
> > On Tue, May 9, 2017 at 8:37 PM, Michael Niedermayer
> <mich...@niedermayer.cc>
>
Hi,
On Tue, May 9, 2017 at 8:37 PM, Michael Niedermayer
wrote:
> Fixes: out of array access
> Fixes: 1434/clusterfuzz-testcase-minimized-6314998085189632
> Fixes: 1435/clusterfuzz-testcase-minimized-6483783723253760
>
> Found-by: continuous fuzzing process
Hi,
On Mon, May 8, 2017 at 3:08 PM, Zubin Mevawalla
wrote:
> If `bufptr` is NULL, and `len` >= 12, then `buf` is initialized to NULL
> and dereferenced on line 796.
>
> diff --git a/libavformat/rtpdec.c b/libavformat/rtpdec.c
> --- a/libavformat/rtpdec.c
> +++
Hi,
On Sun, May 7, 2017 at 2:55 PM, Hendrik Leppkes <h.lepp...@gmail.com> wrote:
> On Sun, May 7, 2017 at 7:17 PM, Ronald S. Bultje <rsbul...@gmail.com>
> wrote:
> > Hi,
> >
> > On Sun, May 7, 2017 at 11:15 AM, Hendrik Leppkes <h.lepp...@gmail.com>
&
Hi,
On Sun, May 7, 2017 at 7:50 AM, Nicolas George <geo...@nsup.org> wrote:
> L'octidi 18 floréal, an CCXXV, Ronald S. Bultje a écrit :
> > My understanding from what Nicolas said is that the goal is to have
> > higher-level code (utils.c-level, e.g. the callers of filter(
Hi,
On Sun, May 7, 2017 at 11:15 AM, Hendrik Leppkes
wrote:
> Ronald suggested on IRC that the only parts suffering from this are
> the ASM DSP components, and we should just make checkasm tests for all
> of them and let checkasm test the alignment they claim to support,
Hi,
On Sun, May 7, 2017 at 7:34 AM, Marton Balint wrote:
>
> On Sat, 6 May 2017, Nicolas George wrote:
>
> Le septidi 17 floréal, an CCXXV, Muhammad Faiz a écrit :
>>
>>> As far as I know. No new features can go to release branch.
>>> Or maybe I'm wrong.
>>>
>>
>> As I said, if
Hi,
On Fri, May 5, 2017 at 12:17 PM, Nicolas George wrote:
> Le sextidi 16 floréal, an CCXXV, Hendrik Leppkes a écrit :
> > Blocking crash fixes on silly arguments doesn't help anyone. Send a
>
> My argument is not silly: the code to align the frame will be needed,
> but NOT
Hi Muhammad,
On Sat, May 6, 2017 at 10:13 AM, Muhammad Faiz wrote:
> On Fri, May 5, 2017 at 11:22 PM, Nicolas George wrote:
> > Le sextidi 16 floréal, an CCXXV, Muhammad Faiz a écrit :
> >> But if
Hi,
On Sat, May 6, 2017 at 5:20 AM, Nicolas George wrote:
> +/**
> + * Minimum alignment of frame data required by the codec.
> + * All frame data pointers must have the alignment lower bits cleared,
> + * i.e. be a multiple of 1< + * - encoding: set by the
Hi,
On Mon, May 1, 2017 at 12:44 PM, Ronald S. Bultje <rsbul...@gmail.com>
wrote:
> Hi,
>
> On Sun, Apr 30, 2017 at 1:16 PM, wm4 <nfx...@googlemail.com> wrote:
>
>> I think it's probably ok to push your current patch, since the original
>> author of the pat
Hi,
On Sun, Apr 30, 2017 at 1:16 PM, wm4 wrote:
> I think it's probably ok to push your current patch, since the original
> author of the patch who broke this probably won't fix it.
I don't like the patch because it essentially duplicates the C
implementation of the DSP
fate-h264-attachment-631 THREADS=4
> This also reverts a755b725ec1d657609c8bd726ce37e7cf193d03f.
>
> Suggested-by: wm4, Ronald S. Bultje, Marton Balint
> Signed-off-by: Muhammad Faiz <mfc...@gmail.com>
> ---
> libavcodec/decode.c| 21 +++--
>
Hi,
On Thu, Apr 27, 2017 at 10:26 AM, Nicolas George <geo...@nsup.org> wrote:
> L'octidi 8 floréal, an CCXXV, Ronald S. Bultje a écrit :
> > Do we? Where is that documented?
> >
> > https://www.google.com/search?q=twos-complement+ffmpeg+site:
> ffm
Hi,
On Thu, Apr 27, 2017 at 9:56 AM, Michael Niedermayer wrote:
> We require Twos-complement style definition for signed integers
Do we? Where is that documented?
https://www.google.com/search?q=twos-complement+ffmpeg+site:ffmpeg.org+-pipermail
Ronald
Hi,
On Wed, Apr 26, 2017 at 2:20 PM, Muhammad Faiz <mfc...@gmail.com> wrote:
> On Wed, Apr 26, 2017 at 10:34 PM, Ronald S. Bultje <rsbul...@gmail.com>
> wrote:
> > Hi,
> >
> > On Wed, Apr 26, 2017 at 1:21 AM, Muhammad Faiz <mfc...@gmail.com> wrote:
>
Hi,
On Wed, Apr 26, 2017 at 1:21 AM, Muhammad Faiz wrote:
> On Wed, Apr 26, 2017 at 4:09 AM, wm4 wrote:
> > On Tue, 25 Apr 2017 23:52:04 +0700
> > Muhammad Faiz wrote:
> >
> >> when frame is received, not from other threads.
> >>
> >>
Hi,
On Mon, Apr 24, 2017 at 5:57 AM, wm4 wrote:
> On Mon, 24 Apr 2017 10:46:38 +0200
> Carl Eugen Hoyos wrote:
> > 2017-04-24 5:50 GMT+02:00 Aaron Levinson :
> > > On 4/23/2017 7:07 PM, Michael Niedermayer wrote:
> > >>
> > >>
Hi,
On Thu, Apr 20, 2017 at 2:42 PM, James Zern
wrote:
> On Thu, Apr 13, 2017 at 6:44 PM, James Zern wrote:
> > this was added in 1.6.0
> >
> > Signed-off-by: James Zern
> > ---
> > doc/encoders.texi | 2 +-
> >
Hi,
On Tue, Apr 18, 2017 at 10:30 AM, Hendrik Leppkes
wrote:
> This brings our generation of the vpcC box up to date to the latest
> draft version of the VP Codec ISO Media File Format Binding.
>
> Specifically, color/transfer properties are now written with values
> based
Hi,
On Tue, Apr 18, 2017 at 6:31 AM, Michael Niedermayer wrote:
> Signed-off-by: Michael Niedermayer
> ---
> libavcodec/avcodec.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/libavcodec/avcodec.h
Hi,
On Sun, Apr 16, 2017 at 12:59 PM, Aaron Levinson
wrote:
> I like using this approach, but I think such files should only be added to
> the build if --disable-optimizations is passed into configure.
>
Or you could detect in configure if the compiler supports DCE?
Hi,
On Fri, Apr 14, 2017 at 8:14 PM, Ivan Kalvachev
wrote:
> I do remember that in 2014 Ronald was pressuring Libav (e.g. Vittorio) [..]
Keep me out of this discussion, I want nothing to do with this.
Ronald
___
ffmpeg-devel
Hi,
On Fri, Apr 14, 2017 at 9:46 PM, James Darnley wrote:
> The labels get stripped leading to (slightly) nicer disassembly from
> objdump.
> ---
> libavcodec/x86/h264_idct.asm | 24
> 1 file changed, 12 insertions(+), 12 deletions(-)
>
> diff --git
Hi,
On Fri, Apr 14, 2017 at 4:24 AM, Michael Niedermayer wrote:
> On Thu, Apr 13, 2017 at 09:00:33PM +0100, Rostislav Pehlivanov wrote:
> > +We strongly recommend users, distributors, and system integrators to
> > +upgrade unless they use current git master.
>
>
Hi Aaron,
On Thu, Apr 13, 2017 at 4:47 PM, Aaron Levinson
wrote:
> #if ARCH_AARCH64
> if (ARCH_AARCH64)
> return ff_get_cpu_flags_aarch64();
> #endif
>
If you add #if, at least replace the if with it. #if + if is redundant.
> Such a script could be reviewed
Hi,
On Tue, Apr 4, 2017 at 8:01 AM, Ronald S. Bultje <rsbul...@gmail.com> wrote:
> On Tue, Apr 4, 2017 at 6:33 AM, Carl Eugen Hoyos <ceho...@ag.or.at> wrote:
>
>> I believe attached patch makes FFmpeg a little more consistent with
>> itself.
>
>
> Looks l
Hi,
On Mon, Apr 10, 2017 at 6:45 AM, Clément Bœsch wrote:
> On Sun, Apr 02, 2017 at 06:58:39PM +0200, Clément Bœsch wrote:
> [...]
> > Here is a list of con arguments for merging:
> >
> > - speedlosses have been observed in various decoders, in particular on
> >32-bits
Hi,
On Mon, Apr 10, 2017 at 12:33 AM, James Almer wrote:
> Experimental VP9 support was added to the muxer recently.
> ---
> libavformat/movenc.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/libavformat/movenc.c b/libavformat/movenc.c
> index
Hi,
On Fri, Apr 7, 2017 at 7:49 AM, Michael Niedermayer
wrote:
> This avoids potential undefined behavior in debug mode while still allowing
> developers which want to check for potential additional overflows to do so
> by manually enabling this.
>
> Signed-off-by:
Hi,
On Fri, Apr 7, 2017 at 2:37 AM, wm4 <nfx...@googlemail.com> wrote:
> On Thu, 6 Apr 2017 13:48:41 -0400
> "Ronald S. Bultje" <rsbul...@gmail.com> wrote:
>
> > The av_log() is done outside the lock, but this way the accesses to the
> > field (reads
Hi,
On Fri, Apr 7, 2017 at 6:47 AM, Michael Niedermayer <mich...@niedermayer.cc>
wrote:
> On Fri, Apr 07, 2017 at 06:26:05AM -0400, Ronald S. Bultje wrote:
> > On Fri, Apr 7, 2017 at 6:22 AM, Michael Niedermayer
> <mich...@niedermayer.cc>
> > > Id like to appl
Hi,
On Fri, Apr 7, 2017 at 6:22 AM, Michael Niedermayer
wrote:
> if you say these are not bugs by definition even though its
> quite obvious they are bugs then whatever you use as definition must
> be flawed. Or you misunderstood what i wrote.
>
That's hypocritical.
Id
Hi,
On Thu, Apr 6, 2017 at 8:10 PM, Michael Niedermayer <mich...@niedermayer.cc>
wrote:
> On Wed, Mar 22, 2017 at 07:09:23PM +0100, Michael Niedermayer wrote:
> > On Wed, Mar 22, 2017 at 09:06:09AM -0400, Ronald S. Bultje wrote:
> > > Hi,
> > >
> > >
The av_log() is done outside the lock, but this way the accesses to the
field (reads and writes) are always protected by a mutex. The av_log()
is not run inside the lock context because it may involve user callbacks
and doing that in performance-sensitive code is probably not a good idea.
This
I'm hoping that this will address the remaining tsan fate-h264 issues:
WARNING: ThreadSanitizer: data race (pid=24478)
Read of size 8 at 0x7dbc0001c828 by main thread (mutexes: write M3243):
#0 ff_h264_ref_picture src/libavcodec/h264_picture.c:107
(ffmpeg+0x013b78d8)
[..]
Previous
Hi,
On Thu, Apr 6, 2017 at 11:36 AM, wm4 wrote:
> Consider the following sequence of events:
>
> - open a codec without AV_CODEC_CAP_DELAY
> - decode call fails with an error
> - ff_thread_flush() is called
> - drain packet is sent
>
> Then the last step would make
Hi,
On Thu, Apr 6, 2017 at 10:22 AM, James Darnley <jdarn...@obe.tv> wrote:
> On 2017-04-05 13:41, Ronald S. Bultje wrote:
> > Hi,
> >
> > On Tue, Apr 4, 2017 at 9:53 PM, James Darnley <jdarn...@obe.tv> wrote:
> >
> >> The labels get strip
Hi,
On Wed, Apr 5, 2017 at 4:53 PM, Ronald S. Bultje <rsbul...@gmail.com> wrote:
> Hi,
>
> On Wed, Apr 5, 2017 at 8:38 AM, Shivraj Patil <shivraj.pa...@imgtec.com>
> wrote:
>
>> Hi,
>>
>>
>>
>> Updated the patch as per comments.
&
Hi,
On Wed, Apr 5, 2017 at 8:38 AM, Shivraj Patil
wrote:
> Hi,
>
>
>
> Updated the patch as per comments.
>
> As hevcpred_init_mips.c needs definition of HAVE_MSA and av_cold, now I
> have included config.h & libavutil/attributes.h in this file.
>
Patch LGTM. I'll
Fixes tsan warnings like this in fate-vp8-test-vector-007:
WARNING: ThreadSanitizer: data race (pid=65909)
Write of size 4 at 0x7d8ce088 by thread T1:
#0 vp8_decode_mb_row_sliced vp8.c:2519 (ffmpeg:x86_64+0x100995ede)
[..]
Previous write of size 4 at 0x7d8ce088 by thread T2:
Fixes tsan warnings like this in fate-vp8-test-vector-007:
WARNING: ThreadSanitizer: data race (pid=3590)
Write of size 4 at 0x7d8ce07c by thread T2:
#0 decode_mb_row_no_filter src/libavcodec/vp8.c:2330 (ffmpeg+0x00ffb59e)
[..]
Previous write of size 4 at 0x7d8ce07c by thread
Hi,
On Mon, Apr 3, 2017 at 10:24 AM, Ronald S. Bultje <rsbul...@gmail.com>
wrote:
> This tries to handle cases where separate invocations of decode_frame()
> (each running in separate threads) write to respective fields in the
> same AVFrame->data[]. Having per-field owners
Hi,
On Wed, Apr 5, 2017 at 3:29 PM, Michael Niedermayer <mich...@niedermayer.cc>
wrote:
> On Mon, Apr 03, 2017 at 10:28:33PM -0400, Ronald S. Bultje wrote:
> > Fixes the following tsan warning when running fate-vsynth_lena-ffvhuff:
> >
> > WARNING: ThreadSan
---
libavcodec/x86/idctdsp_init.c | 30 +++---
1 file changed, 15 insertions(+), 15 deletions(-)
diff --git a/libavcodec/x86/idctdsp_init.c b/libavcodec/x86/idctdsp_init.c
index 3f078e8..fd5ef3f 100644
--- a/libavcodec/x86/idctdsp_init.c
+++
These use the mmx IDCT, but sse2 put/add_pixels_clamped implementations.
This way we don't need to use the ff_put/add_pixels_clamped function
pointers.
---
libavcodec/x86/idctdsp_init.c | 38 ++
libavcodec/x86/simple_idct.c | 15 +--
This makes using the function pointer ff_add_pixels_clamped() unnecessary,
since we always know what the best implementation is at compile-time.
---
libavcodec/x86/cavsdsp.c| 15 +-
libavcodec/x86/cavsidct.asm | 48 -
2 files changed, 61
Hi,
On Tue, Apr 4, 2017 at 9:53 PM, James Darnley wrote:
> The labels get stripped leading to (slightly) nicer disassembly from
> objdump.
>
[..]
> -jz .cycle%1end
> +jz %%skip
Can you preserve the leading dot? I don't mind the %%skip, but please make
it .%%skip.
Hi,
On Tue, Apr 4, 2017 at 9:54 PM, Michael Niedermayer
wrote:
> if the idct depends on external asm then it would need to be under
> HAVE_EXTERNAL_*, you are correct. Also that is ugly
It's temporary, I think James/Kieran were talking about converting the IDCT
to yasm
Hi,
On Tue, Apr 4, 2017 at 5:52 PM, Michael Niedermayer <mich...@niedermayer.cc>
wrote:
> On Tue, Apr 04, 2017 at 12:48:17PM -0400, Ronald S. Bultje wrote:
> > These use the mmx IDCT, but sse2 put/add_pixels_clamped implementations.
> > This way we don't need to use the ff_p
Hi,
On Tue, Apr 4, 2017 at 5:18 PM, Michael Niedermayer <mich...@niedermayer.cc>
wrote:
> On Tue, Apr 04, 2017 at 12:48:13PM -0400, Ronald S. Bultje wrote:
> > Instead, hardcode the use of the _arm implementation of this function.
> > Since there's separate implementatio
Instead, hardcode the use of the _arm implementation of add_pixels,
and use the C version for put_pixels (as no arm-optimized version
exists). Since there's separate implementations of idct{,_put,_add}
for neon, this has no practical impact on performance.
---
libavcodec/arm/idctdsp_init_arm.c |
Hi,
On Tue, Apr 4, 2017 at 3:07 PM, James Almer <jamr...@gmail.com> wrote:
> On 4/4/2017 1:48 PM, Ronald S. Bultje wrote:
> > ---
> > libavcodec/x86/Makefile | 2 +-
> > libavcodec/x86/cavsdsp.c| 165 ++
> --
&g
These use the mmx IDCT, but sse2 put/add_pixels_clamped implementations.
This way we don't need to use the ff_put/add_pixels_clamped function
pointers.
---
libavcodec/x86/idctdsp_init.c | 10 ++
libavcodec/x86/simple_idct.c | 15 +--
libavcodec/x86/simple_idct.h | 3 +++
3
quot;
-);
-}
-
+ff_cavs_idct8_mmx(b2, block);
ff_add_pixels_clamped(b2, dst, stride);
}
-#endif /* HAVE_MMX_INLINE */
+#endif /* HAVE_MMX_EXTERNAL */
#if (HAVE_MMXEXT_INLINE || HAVE_AMD3DNOW_INLINE)
@@ -529,12 +376,10 @@ static av_cold void cavsdsp_init_mmx(CAVSDSPContext
This removes the last use of the ff_put/add_pixels_clamped global
function pointers, and as such they are removed.
This patch has a negative effect on performance on MIPS, since there's
a SIMD-optimized put/add_pixels_clamped, but no xvid or jrev. From a
code maintenance point of view, that is
This makes using the function pointer ff_add_pixels_clamped() unnecessary,
since we always know what the best implementation is at compile-time.
---
libavcodec/x86/cavsdsp.c| 15 +-
libavcodec/x86/cavsidct.asm | 48 -
2 files changed, 61
Since there's separate SSE2 implementations of xvid_idct_put/add, this
patch has no practical impact on performance.
---
libavcodec/x86/xvididct_init.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/libavcodec/x86/xvididct_init.c b/libavcodec/x86/xvididct_init.c
index
Instead, hardcode the use of the _arm implementation of this function.
Since there's separate implementations of idct{,_put,_add} for neon,
this has no practical effect on performance.
---
libavcodec/arm/idctdsp_init_arm.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git
Hi,
On Tue, Apr 4, 2017 at 10:07 AM, Michael Niedermayer <mich...@niedermayer.cc
> wrote:
> On Tue, Apr 04, 2017 at 09:09:36AM -0400, Ronald S. Bultje wrote:
> > This fixes the following tsan warnings when running fate-dnxhr-parse:
> >
> > WARNING: ThreadSani
Hi,
On Tue, Apr 4, 2017 at 9:09 AM, Ronald S. Bultje <rsbul...@gmail.com> wrote:
> This fixes the following tsan warning:
>
> WARNING: ThreadSanitizer: data race (pid=95987)
> Write of size 4 at 0x7d504278 by main thread:
> #0 get_frame_defaults frame.c:117 (ffmp
This fixes the following tsan warning:
WARNING: ThreadSanitizer: data race (pid=95987)
Write of size 4 at 0x7d504278 by main thread:
#0 get_frame_defaults frame.c:117 (ffmpeg:x86_64+0x1016c5b18)
[..]
Previous write of size 4 at 0x7d504278 by thread T2:
[failed to restore the
This fixes the following tsan warnings when running fate-dnxhr-parse:
WARNING: ThreadSanitizer: data race (pid=29917)
Write of size 8 at 0x025b12d8 by thread T2 (mutexes: write M1543):
#0 ff_idctdsp_init src/libavcodec/idctdsp.c:313 (ffmpeg+0x0044b68e)
[..]
Previous write of size
On Tue, Apr 4, 2017 at 8:05 AM, Shivraj Patil
wrote:
> Hi,
>
> > diff --git a/libavcodec/hevcpred.h b/libavcodec/hevcpred.h
> > [..]
> > @@ -26,6 +26,8 @@
> > #include
> > #include
> >
> > +#include "get_bits.h"
> > +
>
hevcpred does not depend on get_bits.h and
Hi,
On Tue, Apr 4, 2017 at 6:33 AM, Carl Eugen Hoyos wrote:
> Hi!
>
> I believe attached patch makes FFmpeg a little more consistent with itself.
Looks like it, yes.
This patch does expose a bug we've discussed before though. Since the
return value of handle_jpeg() is
Fixes the following tsan warning when running fate-vsynth_lena-ffvhuff:
WARNING: ThreadSanitizer: data race (pid=6484)
Write of size 8 at 0x7d64000154b8 by main thread (mutexes: write M1331):
#0 update_context_from_user src/libavcodec/pthread_frame.c:331
(ffmpeg+0x00dca887)
[..]
Hi Betty,
On Mon, Apr 3, 2017 at 11:00 AM, Betty Wu wrote:
> +typedef double number_t;
>
Why?
+static int ansnr_filter(const uint8_t* src_image, number_t* dst_image, int
> img_row, int img_col, const double *filter, int stride)
>
[..]
> +number_t **imme_image;
> +
Fixes tsan warnings in fate-apng:
WARNING: ThreadSanitizer: data race (pid=51230)
Read of size 4 at 0x7d5042fc by main thread (mutexes: write M1000):
#0 frame_copy_props frame.c:302 (ffmpeg:x86_64+0x1019a35d6)
[..]
Previous write of size 4 at 0x7d5042fc by thread T1 (mutexes:
This tries to handle cases where separate invocations of decode_frame()
(each running in separate threads) write to respective fields in the
same AVFrame->data[]. Having per-field owners makes interaction between
readers (the referencing thread) and writers (the decoding thread)
slightly more
Fixes a reported (but false) race condition in tsan for fate-apng:
WARNING: ThreadSanitizer: data race (pid=6274)
Read of size 4 at 0x7d680001ec78 by main thread (mutexes: write M1338):
#0 update_thread_context src/libavcodec/pngdec.c:1456
(ffmpeg+0x00dacf0c)
[..]
Previous write of
501 - 600 of 1885 matches
Mail list logo