From: Mans Rullgard
The vertically interpolating variants of these functions read
ahead one line to optimise the loop. On the last line processed,
this might be outside the buffer. Fix these invalid reads by
processing the last line outside the loop.
Signed-off-by: Mans Rullgard
---
libavcod
From: Mans Rullgard
Signed-off-by: Mans Rullgard
---
libavcodec/arm/dsputil_neon.S | 62 ++-
1 file changed, 20 insertions(+), 42 deletions(-)
diff --git a/libavcodec/arm/dsputil_neon.S b/libavcodec/arm/dsputil_neon.S
index 4bdcd95..21b1aba 100644
--- a/
Hi,
backports following three commits to 0.8
cbc7d60afa0c5 arm: dsputil: fix overreads in put/avg_pixels functions
2eba6898c96fe arm: dsputil: prettify some conditional instructions in
e54e6f25cfa8b arm/neon: dsputil: use correct size specifiers on
Only the first is required to fix the reported
On 2013-11-21 11:50:54 +0100, Anton Khirnov wrote:
> mpeg12 decoder passes the vaapi pixel format to get_format for both
> mpeg 1 and 2. If the caller chooses vaapi for mpeg1, the decoder will
> crash after failing to find a hwaccel for it.
> ---
> libavcodec/allcodecs.c |1 +
> libavcodec/v
---
release/11 is just an alias of master for now
src/documentation | 2 ++
1 file changed, 2 insertions(+)
diff --git a/src/documentation b/src/documentation
index 56c2444..02a7b26 100644
--- a/src/documentation
+++ b/src/documentation
@@ -37,6 +37,8 @@ developers will be able to better diagnos
Both gas and llvm's integrated assembler accept it.
---
gas-preprocessor.pl | 2 +-
test.S | 9 +
2 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/gas-preprocessor.pl b/gas-preprocessor.pl
index 3020401..fbf0853 100755
--- a/gas-preprocessor.pl
+++ b/gas-preproce
On 2014-08-10 14:43:10 +0300, Martin Storsjö wrote:
> The -MD option (for enabling a dynamically linked crt) gets interpreted
> as a cpp option for generating dependency information (into a file named
> '-.d', when preprocessing to a pipe). We shouldn't be passing
> any and all C compiler flags to
On 2014-08-10 16:07:48 +0300, Martin Storsjö wrote:
> Older versions don't support all the features we test for now.
> ---
> I chose not to fold this into the other patches since it didn't really
> seem to be related to them.
> ---
> configure | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-
On 2014-08-10 16:02:45 +0300, Martin Storsjö wrote:
> On Sun, 10 Aug 2014, Janne Grunau wrote:
>
> >Most of my fate configs needs to have the explicit --as removed.
>
> At least for armasm, the explicit --as should still work, since that
> one passes all tests, so it neve
On 2014-08-10 14:14:23 +0300, Martin Storsjö wrote:
> On Sun, 10 Aug 2014, Janne Grunau wrote:
>
> >I have no setup to test apple's gcc-4.2 with gas.
>
> Tested, almost works, requires the following fixup:
>
> -[ $target_os = "darwin" ] && gas
On 2014-08-09 22:48:48 +0300, Martin Storsjö wrote:
> Since plain 'cpp' is used for preprocessing, it inherits whatever
> platform specific predefined defines.
>
> Previously this was set on the libav configure line, such as
> --as='gas-preprocessor.pl -as-type armasm -- armasm -nologo -U__ELF__ -
On 2014-08-09 22:48:47 +0300, Martin Storsjö wrote:
> This allows doing a simple no-op to test that gas-preprocessor
> exists and works, since armasm doesn't have a -v option.
> ---
> gas-preprocessor.pl | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/gas-preprocessor.pl
On 2014-08-10 10:18:30 +, Anton Khirnov wrote:
> They were only needed until the bump.
> ---
> libavcodec/utils.c | 12
> 1 file changed, 12 deletions(-)
>
> diff --git a/libavcodec/utils.c b/libavcodec/utils.c
> index b115da1..4253e8c 100644
> --- a/libavcodec/utils.c
> +++ b/li
On 2014-08-10 04:17:09 +0100, Kieran Kunhya wrote:
> The Opus decoder in particular uses optimised float_dsp functions that expect
> 32-byte alignment
> ---
> libavcodec/avcodec.h |2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/libavcodec/avcodec.h b/libavcodec/avcode
On 2014-08-09 22:48:29 +0300, Martin Storsjö wrote:
> ---
> This goes on top of Diego's RFC patch.
> ---
> configure | 9 +
> 1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/configure b/configure
> index 4d2439e..877f240 100755
> --- a/configure
> +++ b/configure
> @@ -37
On 2014-08-09 22:48:28 +0300, Martin Storsjö wrote:
> ---
> configure | 4
> 1 file changed, 4 insertions(+)
>
> diff --git a/configure b/configure
> index 7e3f6ff..4d2439e 100755
> --- a/configure
> +++ b/configure
> @@ -2881,6 +2881,10 @@ probe_cc(){
> _cflags_size="-O2 -Munroll=c
'#' at the start of the line indicates on all archs the start of a
comment or a preprocessor directive. This is after preprocessing so
removing the lines as comment is safe. Line number comments are stripped
by this too since gas-preprocessor.pl ignores line numbers. So it
removes just misleading d
On 2014-08-09 08:06:12 -0700, Diego Biurrun wrote:
> ---
>
> Implemented Janne's suggestion. Untested since I don't have the right system.
>
> configure | 15 ---
> 1 file changed, 12 insertions(+), 3 deletions(-)
>
> diff --git a/configure b/configure
> index 46b8391..7c2fab6 1007
On 2014-08-06 12:24:24 +0200, Diego Biurrun wrote:
> On Tue, Aug 05, 2014 at 01:35:22PM +0200, Janne Grunau wrote:
> > From: Janne Grunau
> > --- a/configure
> > +++ b/configure
> > @@ -3760,22 +3760,32 @@ unsigned int endian = 'B' << 24 | 'I
On 2014-08-08 14:58:39 +0200, Diego Biurrun wrote:
> On Thu, Aug 07, 2014 at 06:39:33PM +0200, Diego Biurrun wrote:
> > On Wed, Aug 06, 2014 at 04:41:24PM +0200, Janne Grunau wrote:
> > > On 2014-08-06 05:57:39 -0700, Diego Biurrun wrote:
> > >
> > > the comm
On 2014-08-07 10:13:26 -0700, Diego Biurrun wrote:
> ---
> configure | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/configure b/configure
> index 924fd46..4d3732b 100755
> --- a/configure
> +++ b/configure
> @@ -2199,8 +2199,6 @@ ranlib="ranlib"
> strip="strip"
>
On 2014-08-07 10:13:27 -0700, Diego Biurrun wrote:
> Standalone GNU assembly is not used on x86.
> ---
> configure | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/configure b/configure
> index 4d3732b..c74d910 100755
> --- a/configure
> +++ b/configure
> @@ -3774,7 +3774,
On 2014-08-07 10:13:25 -0700, Diego Biurrun wrote:
> ---
> configure | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/configure b/configure
> index 26ea230..924fd46 100755
> --- a/configure
> +++ b/configure
> @@ -3778,7 +3778,7 @@ EOF
> }
>
> if enabled asm; then
> -
On 2014-08-06 05:57:38 -0700, Diego Biurrun wrote:
> ---
> configure | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/configure b/configure
> index c8c57b9..45da142 100755
> --- a/configure
> +++ b/configure
> @@ -3779,8 +3779,8 @@ EOF
> }
>
> if enabled asm; then
On 2014-08-06 05:57:40 -0700, Diego Biurrun wrote:
> Standalone GNU assembly is not used on x86.
> ---
> configure | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/configure b/configure
> index 81dfc09..a0d60e9 100755
> --- a/configure
> +++ b/configure
> @@ -3775,7 +3775,
On 2014-08-06 05:57:39 -0700, Diego Biurrun wrote:
the commit message describes the commit only partially. It also uses
gas-preprocessor.pl if it's available and not only on darwin. I think
that's not going to work since the plain 'gas-preprocessor.pl' without
options will produce mach-O specif
On 2014-08-06 05:57:38 -0700, Diego Biurrun wrote:
> ---
> configure | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/configure b/configure
> index c8c57b9..45da142 100755
> --- a/configure
> +++ b/configure
> @@ -3779,8 +3779,8 @@ EOF
> }
>
> if enabled asm; then
From: Janne Grunau
Hi,
On 2014-08-05 13:38:47 +0300, Martin Storsjö wrote:
> On Tue, 5 Aug 2014, Janne Grunau wrote:
>
> >From: Janne Grunau
> >
> >This was the last missing feature in llvm's integrated assembler on darwin.
> >There is no need to use g
On 2014-08-05 11:20:18 +0200, Luca Barbato wrote:
> On 05/08/14 09:44, Janne Grunau wrote:
> > All subtargets which should run the fate-filter-pixdesc% need to
> > generate and include tests/pixfmts.mak. Most noteable missing target was
> > fate itself.
> > ---
> >
From: Janne Grunau
This was the last missing feature in llvm's integrated assembler on darwin.
There is no need to use gas-preprocessor.pl if the assembler supports the
asm fully.
---
configure | 6 ++
1 file changed, 6 insertions(+)
diff --git a/configure b/configure
index b2
All subtargets which should run the fate-filter-pixdesc% need to
generate and include tests/pixfmts.mak. Most noteable missing target was
fate itself.
---
tests/fate/filter-video.mak | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/fate/filter-video.mak b/tests/fate/filter
On 2014-08-04 20:32:15 +0200, Diego Biurrun wrote:
> Module: libav
> Branch: master
> Commit: eee813eec7d3c0b0689f80665d3f796401742935
>
> Author:Diego Biurrun
> Committer: Diego Biurrun
> Date: Sun Aug 3 06:21:15 2014 -0700
>
> fate: Only generate tests/pixfmts.mak if some pixfmts f
On 2014-08-04 11:35:46 +0200, Diego Biurrun wrote:
> On Mon, Aug 04, 2014 at 10:09:34AM +0200, Janne Grunau wrote:
> > --- a/tests/Makefile
> > +++ b/tests/Makefile
> > @@ -1,3 +1,4 @@
> > +THREADS := 1
> > VREF = tests/vsynth1/00.pgm
> > AREF = tests/
On 2014-08-04 03:00:58 -0700, Diego Biurrun wrote:
> ---
>
> Now w/o silly typo and not missing fate-list.
>
> tests/fate/filter-video.mak | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/tests/fate/filter-video.mak b/tests/fate/filter-video.mak
> index bbe77ce..cd905
On 2014-08-03 15:48:21 -0700, Diego Biurrun wrote:
> Otherwise the dependency is noticed only after compiling the first file
> that requires it, resulting in double compilation of that file.
requires what? I can't follow why this is necessary. .version and
version.h should be always (re-)made fir
On 2014-08-03 15:48:20 -0700, Diego Biurrun wrote:
> ---
> tests/fate/filter-video.mak | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/tests/fate/filter-video.mak b/tests/fate/filter-video.mak
> index bbe77ce..d1cb721 100644
> --- a/tests/fate/filter-video.mak
> +++ b/
This makes the default of '1' more explicit than defaulting to '1' in
fate-run.sh and regression-funcs.sh if THREADS is not set.
Fixes the reported thread count in fate-cpu if THREADS is not set.
---
tests/Makefile | 1 +
1 file changed, 1 insertion(+)
diff --git a/tests/Makefile b/tests/Makefile
On 2014-08-03 18:06:04 +0200, Diego Biurrun wrote:
> On Sun, Aug 03, 2014 at 04:37:05PM +0200, Janne Grunau wrote:
> > --- a/libavutil/cpu.c
> > +++ b/libavutil/cpu.c
> > @@ -224,17 +228,50 @@ static const struct {
> >
> > +static void print_cpu_flag
On 2014-08-03 09:02:03 -0700, Diego Biurrun wrote:
> From: Anton Khirnov
>
> Signed-off-by: Diego Biurrun
> ---
>
> This fixes fate-xvid-idctmmx with more than one thread.
>
> libavcodec/mpeg4videodec.c | 5 +
> 1 file changed, 5 insertions(+)
>
> diff --git a/libavcodec/mpeg4videodec.c
---
libavutil/aarch64/asm.S | 4
1 file changed, 4 insertions(+)
diff --git a/libavutil/aarch64/asm.S b/libavutil/aarch64/asm.S
index 850f16b..b766e67 100644
--- a/libavutil/aarch64/asm.S
+++ b/libavutil/aarch64/asm.S
@@ -57,7 +57,11 @@ FUNC.func \name
ELF .size \name, . - \name
Add CPU count and number threads as informative values for fate.
---
libavutil/cpu.c | 14 +-
tests/fate/libavutil.mak | 2 +-
2 files changed, 14 insertions(+), 2 deletions(-)
diff --git a/libavutil/cpu.c b/libavutil/cpu.c
index 20f0fe1..a7e5f7f 100644
--- a/libavutil/cpu.c
libavutil/cpu-test prints raw and effective cpu flags to STDERR. Detected
cpu flags can be useful for debugging fate errors.
No comparison of the result against a expected result since that would
require fate config specific references.
---
libavutil/cpu.c | 49 ++
On 2014-08-03 12:23:35 +0300, Martin Storsjö wrote:
> On Sun, 3 Aug 2014, Martin Storsjö wrote:
>
> >On Thu, 24 Jul 2014, Janne Grunau wrote:
> >
> >>---
> >>libavutil/aarch64/asm.S | 9 -
> >>1 file changed, 8 insertions(+), 1 deletion(-)
&g
On 2014-07-24 15:11:24 +0200, Janne Grunau wrote:
> Hi,
>
> following two patches improve support for llvm 3.5's integrated assembler
> on darwin/ios. A locally patched llvm-3.5 with http://reviews.llvm.org/D3507
> passes mak
On 2014-07-30 09:17:01 -0700, Diego Biurrun wrote:
> This allows running all the tests individually and/or in parallel.
> ---
>
> Fixed cross-compilation setups and added a newline to the end of the snippet.
lgtm
Janne
___
libav-devel mailing list
liba
On 2014-07-29 06:18:19 -0700, Diego Biurrun wrote:
> This allows running all the tests individually and/or in parallel.
> ---
>
> On Diegò's 32-core compile host this reduces the time to run fate in
> parallel from 44.5s to 42.5s.
>
> diff --git a/.gitignore b/.gitignore
> index 584ff94..19bb757
On 2014-07-28 01:14:43 -0700, Diego Biurrun wrote:
> From: Bernhard Übelacker
>
> This avoids a segfault in avconv_opt.c:opt_target when trying to
> determine the norm.
>
> A workaround is to specify the norm as prefix to the target (instead
> of "-target dvd" use "-target pal-dvd").
> ---
>
>
Adding 'branch=release/10' to the fate config file will check the
release/10 branch instead of master. If no branch is specified it will
use 'master' so that existing config are still valid.
The server side changes are already deployed, see
https://fate.libav.org/v10/ for an example. The server su
On 2014-07-23 15:51:10 +0200, Diego Biurrun wrote:
> On Mon, Jul 21, 2014 at 11:58:26PM +0200, Janne Grunau wrote:
> > On 2014-07-21 14:21:08 -0700, Diego Biurrun wrote:
> > > The DCT init code takes care of MMX and other arch details on its own.
> >
> > yes and
llvm's integrated assembler does not accept spaces as macro argument
delimiter when targeting darwin. Using a explicit delimiter is a good
idea in principle since it makes case like 'macro 4 -2' vs 'macro 4 - 2'
clear.
---
libavcodec/aarch64/fft_neon.S | 2 +-
libavcodec/aarch64/mpegaudio
---
libavutil/aarch64/asm.S | 9 -
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/libavutil/aarch64/asm.S b/libavutil/aarch64/asm.S
index 850f16b..f765e59 100644
--- a/libavutil/aarch64/asm.S
+++ b/libavutil/aarch64/asm.S
@@ -26,6 +26,12 @@
# define ELF #
#endif
+#ifde
Hi,
following two patches improve support for llvm 3.5's integrated assembler
on darwin/ios. A locally patched llvm-3.5 with http://reviews.llvm.org/D3507
passes make fate.
Janne
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.or
On 2014-07-22 04:03:02 -0700, Diego Biurrun wrote:
> ---
>
> Changed to also replace the need_memalign variable in configure, which
> essentially indicates the same thing.
>
> I've followed the git-blame path of STRIDE_ALIGN and an now confident
> this is the correct solution for a number of hack
On 2014-07-22 10:01:35 +0200, Diego Biurrun wrote:
> On Tue, Jul 22, 2014 at 12:04:38AM +0200, Janne Grunau wrote:
> > On 2014-07-21 14:21:07 -0700, Diego Biurrun wrote:
> > > It is doubtful if the hack (still) works and PPC is obsolete for encoding.
> >
> > not so
On 2014-07-21 14:21:07 -0700, Diego Biurrun wrote:
> It is doubtful if the hack (still) works and PPC is obsolete for encoding.
not so sure about the latter but xvid had enough time to fix it. no need
to keep this ugly workaround.
> ---
> libavcodec/libxvid.c | 12
> 1 file changed
On 2014-07-21 14:21:08 -0700, Diego Biurrun wrote:
> The DCT init code takes care of MMX and other arch details on its own.
yes and no
> ---
> libavcodec/mpeg4videodec.c | 6 +-
> 1 file changed, 1 insertion(+), 5 deletions(-)
>
> diff --git a/libavcodec/mpeg4videodec.c b/libavcodec/mpeg4vi
---
Makefile | 1 +
configure | 12
2 files changed, 13 insertions(+)
diff --git a/Makefile b/Makefile
index 6302997..b8d1034 100644
--- a/Makefile
+++ b/Makefile
@@ -130,6 +130,7 @@ $(foreach V,$(SUBDIR_VARS),$(eval $(call RESET,$(V
SUBDIR := $(1)/
include $(SRC_PATH)/$(1)/M
From: James Yu
Intrinsics only used on aarch64 since the existing ARMv7 NEON asm
is slightly faster (Cortex-A9, gcc-4.8, micro-benchmarks and full
decoding time).
Signed-off-by: James Yu
Signed-off-by: Janne Grunau
---
libavcodec/mpegvideo.c | 3 +
libavcodec/mpegvideo.h | 1
On 2014-07-20 14:50:59 -0400, Reinhard Tartler wrote:
> Control: forwarded -1 libav-devel@libav.org
>
> Hi,
>
> In Debian, we've received this bugreport regarding a bugfix in our
> video4linux2 handling.
>
> Is the patch below acceptable for master?
looks ok to me
Janne
___
On 2014-07-19 10:42:55 -0400, Reinhard Tartler wrote:
> Hi,
>
> I notice that we currently have 6 open pull requests on github:
>
> https://github.com/libav/libav/pulls
>
> Interestingly, there are also 8 closed requests:
>
> https://github.com/libav/libav/pulls?direction=desc&page=1&sort=creat
On 2014-07-19 18:26:22 +0200, Diego Biurrun wrote:
> On Sat, Jul 19, 2014 at 12:23:58PM -0400, Reinhard Tartler wrote:
> >
> > For release branches, we currently don't have any FATE Farm. It would
> > be great to have something like that, but as far as I understand, it
> > is a lot of effort to se
On 2014-07-21 01:19:15 +0200, Luca Barbato wrote:
> On 20/07/14 13:57, Luca Barbato wrote:
> > On 20/07/14 13:47, Vittorio Giovara wrote:
> >> From: Carl Eugen Hoyos
> >>
> >> ---
> >> libavcodec/tiff.c | 3 +++
> >> 1 file changed, 3 insertions(+)
> >>
> >> diff --git a/libavcodec/tiff.c b/libav
On 2014-07-16 19:33:59 +0300, Martin Storsjö wrote:
> Hi Ben,
>
> On Wed, 16 Jul 2014, Ben Avison wrote:
>
> >>>+ldr v5, =\costable
> >>
> >>Should this perhaps use the movrelx macro?
> >
> >Um, maybe. I must admit to not really understanding what the movrel and
> >movrelx macros are
On 2014-07-13 11:42:57 +0300, Martin Storsjö wrote:
> ---
> gas-preprocessor.pl | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/gas-preprocessor.pl b/gas-preprocessor.pl
> index 86293d1..431270a 100755
> --- a/gas-preprocessor.pl
> +++ b/gas-preprocessor.pl
> @@ -877,7 +8
On 2014-07-08 10:58:10 +0300, Martin Storsjö wrote:
> This instruction is deprecated on ARMv8, and it is serializing on
> some ARMv7 cores as well [1].
>
> http://article.gmane.org/gmane.linux.ports.arm.kernel/339293
> ---
> libavcodec/arm/h264dsp_init_arm.c | 7 ++-
> 1 file changed, 6 inser
On 2014-07-06 17:48:14 +0300, Martin Storsjö wrote:
> On Sun, 6 Jul 2014, Janne Grunau wrote:
>
> >On 2014-07-04 18:27:54 +0300, Martin Storsjö wrote:
> >>
> >>[1]
> >>http://lists.infradead.org/pipermail/linux-arm-kernel/2014-July/268504.html
> >>
On 2014-07-04 18:27:54 +0300, Martin Storsjö wrote:
> See [1] for discussion on the issue with using 'setend' on modern
> arm versions.
>
> [1]
> http://lists.infradead.org/pipermail/linux-arm-kernel/2014-July/268504.html
> ---
> libavcodec/arm/h264dsp_init_arm.c | 6 +-
> 1 file changed, 5
On 2014-06-28 00:10:59 +0300, Martin Storsjö wrote:
> When running on a 64 bit kernel, /proc/cpuinfo lists different
> optional features than on 32 bit kernels (because some of them
> are mandatory in the 64 bit implemenations).
>
> The kernel does list the old features properly if they are querie
On 2014-06-23 21:10:14 +0300, Martin Storsjö wrote:
> When running on a 64 bit kernel, /proc/cpuinfo does not list
> the cpu features that aren't optional on such CPUs.
>
> A 32 bit binary which runs on such a kernel, that tries to parse
> /proc/cpuinfo to find the relevant cpu features at runtime
On 2014-06-19 12:26:15 +0200, Janne Grunau wrote:
> Adapt commit 982b596ea6640bfe218a31f6c3fc542d9fe61c31 for the arm and
> aarch64 NEON asm. 5-10% faster on Cortex-A9.
> ---
> libavcodec/aarch64/h264cmc_neon.S | 59 +++---
> libavcodec/arm/h264cmc_
On 2014-06-19 18:46:15 +0200, Diego Biurrun wrote:
> On Thu, Jun 19, 2014 at 12:29:04PM +0200, Janne Grunau wrote:
> > --- a/libavcodec/mpegvideo.c
> > +++ b/libavcodec/mpegvideo.c
> > @@ -391,6 +391,9 @@ av_cold int ff_dct_common_init(MpegEncContext *s)
> > s-
On 2014-06-19 18:55:45 +0200, Diego Biurrun wrote:
> On Thu, Jun 19, 2014 at 12:29:03PM +0200, Janne Grunau wrote:
> > --- a/configure
> > +++ b/configure
> > @@ -4403,6 +4411,17 @@ enabled_all dxva2 CoTaskMemFree &&
> >
> > +for intrin in $INTRINSICS_L
---
Makefile | 1 +
configure | 20
2 files changed, 21 insertions(+)
diff --git a/Makefile b/Makefile
index 6302997..b8d1034 100644
--- a/Makefile
+++ b/Makefile
@@ -130,6 +130,7 @@ $(foreach V,$(SUBDIR_VARS),$(eval $(call RESET,$(V
SUBDIR := $(1)/
include $(SRC_PATH
From: James Yu
Intrinsics only used on aarch64 since the existing ARMv7 NEON asm
is slightly faster (Cortex-A9, gcc-4.8, micro-benchmarks and full
decoding time).
Signed-off-by: James Yu
---
libavcodec/mpegvideo.c | 3 +
libavcodec/mpegvideo.h | 1 +
libavcodec/neon/Makefile|
Adapt commit 982b596ea6640bfe218a31f6c3fc542d9fe61c31 for the arm and
aarch64 NEON asm. 5-10% faster on Cortex-A9.
---
libavcodec/aarch64/h264cmc_neon.S | 59 +++---
libavcodec/arm/h264cmc_neon.S | 60 ---
2 files changed, 111
On 2014-06-17 13:59:17 +0200, Diego Biurrun wrote:
> On Tue, Jun 10, 2014 at 10:43:33PM +0200, Janne Grunau wrote:
> > --- a/configure
> > +++ b/configure
> > @@ -1361,6 +1361,7 @@ HAVE_LIST_PUB="
> > HEADERS_LIST="
> > alsa_asoundl
On 2014-06-10 22:43:33 +0200, Janne Grunau wrote:
> ---
> configure | 8
> 1 file changed, 8 insertions(+)
>
> diff --git a/configure b/configure
> index 8359c68..5c21a33 100755
> --- a/configure
> +++ b/configure
> @@ -1361,6 +1361,7 @@ HAVE
On 2014-06-11 19:51:35 +0200, Janne Grunau wrote:
> Since error resilience uses AVFrame pointers instead of references it
> has to copy NULL pointers too. After a codec flush the last/next frame
> pointers in MpegEncContext are NULL and the old pointers remaining in
> ERContext are in
Since error resilience uses AVFrame pointers instead of references it
has to copy NULL pointers too. After a codec flush the last/next frame
pointers in MpegEncContext are NULL and the old pointers remaining in
ERContext are invalid. Fixes a crash in vlc for android thumbnailer.
Reported and debugg
From: James Yu
Intrinsics only used on aarch64 since the existing ARMv7 NEON asm
is slightly faster (Cortex-A9, gcc-4.8, micro-benchmarks and full
decoding time).
Signed-off-by: James Yu
---
libavcodec/aarch64/Makefile | 4 ++
libavcodec/arm/Makefile | 4 ++
libavcodec/arm/neon
---
configure | 8
1 file changed, 8 insertions(+)
diff --git a/configure b/configure
index 8359c68..5c21a33 100755
--- a/configure
+++ b/configure
@@ -1361,6 +1361,7 @@ HAVE_LIST_PUB="
HEADERS_LIST="
alsa_asoundlib_h
altivec_h
+arm_neon_h
arpa_inet_h
cdio_parano
Hi, sorry for the late reply, I missed this updated patch
On 2014-05-17 12:21:10 +0800, james...@linaro.org wrote:
> From: James Yu
>
> Using NEON intrinsics to support ARMv7 and ARMv8 simultaneously.
> Rewrite ARMv7 assembly to intinsics version.
>
> Toolchain:
> armv7: gcc-linaro-arm-linux-gnue
Prevents a triggered assert during fate-vsynth{1,2}-rv20 in
dct_unquantize_h263_intra_c().
---
libavcodec/ituh263enc.c | 8 +++-
1 file changed, 3 insertions(+), 5 deletions(-)
diff --git a/libavcodec/ituh263enc.c b/libavcodec/ituh263enc.c
index cd4ed84..5f15b2f 100644
--- a/libavcodec/ituh26
On 2014-06-08 01:55:47 +0300, Martin Storsjö wrote:
> We know that the called function (ff_chroma_inter_body_mmxext)
> doesn't touch the redzone, and thus will be kept intact - thus,
> this doesn't fix any bug per se.
>
> However, valgrind's memcheck tool intentionally assumes that the
> redzone i
av_dict_set leaks it key/value arguments if AV_DICT_DONT_OVERWRITE is
combined with AV_DICT_DONT_STRDUP_{KEY,VAL} and the key exists.
---
libavutil/dict.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/libavutil/dict.c b/libavutil/dict.c
index 3b95aba..112e082 100644
--- a
On 2014-06-03 13:40:12 +0200, Marc-Antoine ARNAUD wrote:
> I found my error, I don't pass sample directory, I think it's that.
> It looks okay now.
>
> But yes the FATE for Mpeg2 will change. What's the best way ? Resubmit a
> complet patch (including FATE changes),
yes, please resubmit a patch i
On 2014-06-03 11:18:31 +0200, Marc-Antoine ARNAUD wrote:
> I don't find mpeg12 fate using : make fate-list | grep mpeg
> What I need to update for fix my commit ?
>
> During make fate I have:
> TESTfft-4
> make: *** [fate-fft-4] Erreur 127
That should not happen. Is there anything special abo
Clang's integrated assembler (after 3.4) does not yet support the
'.altmacro' directive which is only used in arm asm. Support is planned:
http://llvm.org/bugs/show_bug.cgi?id=18918
---
configure | 4
1 file changed, 4 insertions(+)
diff --git a/configure b/configure
index 59f7273..8359c68 1
On 2014-06-02 14:49:39 +0200, Diego Biurrun wrote:
> On Fri, May 30, 2014 at 08:56:30PM +0200, Janne Grunau wrote:
> > Clang's integrated assembler (after 3.4) does not yet support the
> > '.altmacro' directive which is only used in arm asm. Support is pla
Not supported by Clang's integrated assembler. Since it just adds
debug information it can safely omitted.
---
configure | 6 ++
libavutil/aarch64/asm.S | 12 +---
libavutil/arm/asm.S | 12 +---
3 files changed, 24 insertions(+), 6 deletions(-)
diff --git a/
Move the GNU as check before the arch specific asm checks since the .dn
check requires gas compatible assembler.
Disable the VC-1 motion compensation NEON asm which is the only part
using that directive. The integrated assembler in the upcoming clang 3.5
does not support .dn/.qn without plans to c
Both gnu as and clang treat lines starting with '#' as comments if they
aren't consumed by the C-style preprocessor.
Using '//' does not work with clang since comments are removed before
macro expansion.
---
reformulated the commit message
libavutil/aarch64/asm.S | 2 +-
1 file changed, 1 inserti
I've decided to handle the darwin styl relocations explicitly with #if
in the hope that a future version of apple clang supports our asm fully.
Janne
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
---
libavutil/aarch64/asm.S | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/libavutil/aarch64/asm.S b/libavutil/aarch64/asm.S
index e9cc995..4896bf4 100644
--- a/libavutil/aarch64/asm.S
+++ b/libavutil/aarch64/asm.S
@@ -63,7 +63,10 @@ ELF .size \name, . - \name
.endm
Clang's integrated assembler does not support it.
---
libavutil/aarch64/asm.S | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libavutil/aarch64/asm.S b/libavutil/aarch64/asm.S
index 4896bf4..850f16b 100644
--- a/libavutil/aarch64/asm.S
+++ b/libavutil/aarch64/asm.S
@@ -67,7 +67
On 2014-05-30 22:14:03 +0300, Martin Storsjö wrote:
> On Fri, 30 May 2014, Janne Grunau wrote:
>
> >Clang's integrated assembler does not support it and it is optional
> >for GNU as.
> >---
> >libavutil/aarch64/asm.S | 2 +-
> >1 file changed, 1 insert
On 2014-05-30 22:20:14 +0300, Martin Storsjö wrote:
> On Fri, 30 May 2014, Janne Grunau wrote:
>
> >Disable the VC-1 motion compensation NEON asm which is the only part
> >using that directive. The integrated assembler in the upcoming clang 3.5
> >does not support .dn/.qn
On 2014-05-30 22:14:03 +0300, Martin Storsjö wrote:
> On Fri, 30 May 2014, Janne Grunau wrote:
>
> >Clang's integrated assembler does not support it and it is optional
> >for GNU as.
> >---
> >libavutil/aarch64/asm.S | 2 +-
> >1 file changed, 1 insert
On 2014-05-30 22:17:42 +0300, Martin Storsjö wrote:
> On Fri, 30 May 2014, Janne Grunau wrote:
>
> >Defining symbols conditionally to '//' does not work with clang's
> >integrated assembler. Comments seems to be removed before the symbols
> >are substituted
Defining symbols conditionally to '//' does not work with clang's
integrated assembler. Comments seems to be removed before the symbols
are substituted.
---
libavutil/aarch64/asm.S | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libavutil/aarch64/asm.S b/libavutil/aarch64/asm.S
501 - 600 of 2711 matches
Mail list logo