This gets rid of a couple instructions, but the actual performance
is almost identical on Cortex A72/A73. On Cortex A53, it is a
handful of cycles faster.
---
libavcodec/aarch64/hevcdsp_qpel_neon.S | 15 +--
1 file changed, 5 insertions(+), 10 deletions(-)
diff --git a/libavcodec/aarc
For widths of 32 pixels and more, loop first horizontally,
then vertically.
Previously, this function would process a 16 pixel wide slice
of the block, looping vertically. After processing the whole
height, it would backtrack and process the next 16 pixel wide
slice.
When doing 8tap filtering hor
This is a pure reordering of code without changing anything in
the individual functions.
---
libavcodec/aarch64/hevcdsp_epel_neon.S | 971 +
1 file changed, 497 insertions(+), 474 deletions(-)
diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S
b/libavcodec/aarch64/hevcds
AWS Graviton 3:
put_hevc_epel_hv4_8_c: 163.7
put_hevc_epel_hv4_8_neon: 52.5
put_hevc_epel_hv4_8_i8mm: 49.5
put_hevc_epel_hv6_8_c: 292.2
put_hevc_epel_hv6_8_neon: 97.7
put_hevc_epel_hv6_8_i8mm: 101.2
put_hevc_epel_hv8_8_c: 471.0
put_hevc_epel_hv8_8_neon: 106.7
put_hevc_epel_hv8_8_i8mm: 102.5
put_hev
---
libavcodec/aarch64/hevcdsp_qpel_neon.S | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/libavcodec/aarch64/hevcdsp_qpel_neon.S
b/libavcodec/aarch64/hevcdsp_qpel_neon.S
index 0fcded344b..062b7d4d0f 100644
--- a/libavcodec/aarch64/hevcdsp_qpel_neon.S
+++ b/libavcodec/aarch6
AWS Graviton 3:
put_hevc_epel_uni_hv4_8_c: 163.5
put_hevc_epel_uni_hv4_8_neon: 59.7
put_hevc_epel_uni_hv4_8_i8mm: 57.5
put_hevc_epel_uni_hv6_8_c: 344.7
put_hevc_epel_uni_hv6_8_neon: 105.0
put_hevc_epel_uni_hv6_8_i8mm: 102.7
put_hevc_epel_uni_hv8_8_c: 552.2
put_hevc_epel_uni_hv8_8_neon: 111.2
put_he
AWS Graviton 3:
put_hevc_epel_uni_w_hv4_8_c: 191.2
put_hevc_epel_uni_w_hv4_8_neon: 87.7
put_hevc_epel_uni_w_hv4_8_i8mm: 83.2
put_hevc_epel_uni_w_hv6_8_c: 349.5
put_hevc_epel_uni_w_hv6_8_neon: 153.0
put_hevc_epel_uni_w_hv6_8_i8mm: 148.5
put_hevc_epel_uni_w_hv8_8_c: 581.2
put_hevc_epel_uni_w_hv8_8_ne
AWS Graviton 3:
put_hevc_epel_h4_8_c: 64.7
put_hevc_epel_h4_8_neon: 25.0
put_hevc_epel_h4_8_i8mm: 21.2
put_hevc_epel_h6_8_c: 130.0
put_hevc_epel_h6_8_neon: 40.7
put_hevc_epel_h6_8_i8mm: 36.5
put_hevc_epel_h8_8_c: 209.0
put_hevc_epel_h8_8_neon: 45.2
put_hevc_epel_h8_8_i8mm: 41.2
put_hevc_epel_h12_8_
AWS Graviton 3:
put_hevc_epel_uni_w_h4_8_c: 97.2
put_hevc_epel_uni_w_h4_8_neon: 41.2
put_hevc_epel_uni_w_h4_8_i8mm: 35.2
put_hevc_epel_uni_w_h6_8_c: 203.7
put_hevc_epel_uni_w_h6_8_neon: 84.7
put_hevc_epel_uni_w_h6_8_i8mm: 74.7
put_hevc_epel_uni_w_h8_8_c: 345.7
put_hevc_epel_uni_w_h8_8_neon: 94.0
pu
The first horizontal filter can use either i8mm or plain neon
versions, while the second part is a pure neon implementation.
---
libavcodec/aarch64/hevcdsp_epel_neon.S | 100 +
1 file changed, 100 insertions(+)
diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S
b/libavco
AWS Graviton 3:
put_hevc_qpel_uni_w_h4_8_c: 159.0
put_hevc_qpel_uni_w_h4_8_neon: 64.2
put_hevc_qpel_uni_w_h4_8_i8mm: 40.0
put_hevc_qpel_uni_w_h6_8_c: 344.7
put_hevc_qpel_uni_w_h6_8_neon: 114.5
put_hevc_qpel_uni_w_h6_8_i8mm: 82.0
put_hevc_qpel_uni_w_h8_8_c: 596.2
put_hevc_qpel_uni_w_h8_8_neon: 132.2
In addition to just templating, this contains one change to
ff_hevc_put_hevc_epel_bi_hv32_8, by setting the w6 register
which ff_hevc_put_hevc_epel_h32_8_neon requires.
AWS Graviton 3:
put_hevc_epel_bi_hv4_8_c: 176.5
put_hevc_epel_bi_hv4_8_neon: 62.0
put_hevc_epel_bi_hv4_8_i8mm: 58.0
put_hevc_epel
---
libavcodec/aarch64/hevcdsp_qpel_neon.S | 94 +++---
1 file changed, 86 insertions(+), 8 deletions(-)
diff --git a/libavcodec/aarch64/hevcdsp_qpel_neon.S
b/libavcodec/aarch64/hevcdsp_qpel_neon.S
index fba063186c..c04e8dbea8 100644
--- a/libavcodec/aarch64/hevcdsp_qpel_neon
While MSVC proper doesn't have a way to signal potential aliasing
corresponding to __attribute__((may_alias)) (which is what
av_alias expands to, if supported), Clang does support it.
When building with Clang in MSVC mode, __GNUC__ isn't defined but
_MSC_VER is, so Clang uses the MSVC implementati
This fixes the test when running in a cross test setup where the
samples are located at a different path between build host and
temote test target.
---
tests/fate/hlsenc.mak | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/fate/hlsenc.mak b/tests/fate/hlsenc.mak
index 7df7
In WinRT mode, we use CreateThread instead of _beginthreadex.
CreateThread takes a LPTHREAD_START_ROUTINE function pointer,
which has got the signature DWORD WINAPI ThreadProc(LPVOID).
_beginthreadex takes a function with the signature
unsigned __stdcall func(void *).
DWORD is defined as an unsig
On Thu, 27 Jul 2023, Rémi Denis-Courmont wrote:
Hi,
The use of RET vs BR also has microarchitectural side effects. AFAIU, RET
should always be paired with an earlier BL/BLR to avoid interfering with branch
prediction.
So depending on the circumstances, either one of these should be
addresse
Use the GCC specific codepath for Clang in MSVC mode too.
This matches the condition used in a number of other places.
MSVC doesn't have a way to signal potential aliasing, while GCC
(and Clang) can use __attribute__((may_alias)) for this purpose.
When building with Clang in MSVC mode, __GNUC__ i
On Wed, 2 Aug 2023, Martin Storsjö wrote:
This fixes the test when running in a cross test setup where the
samples are located at a different path between build host and
temote test target.
---
tests/fate/hlsenc.mak | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/fate
On Wed, 2 Aug 2023, Martin Storsjö wrote:
In WinRT mode, we use CreateThread instead of _beginthreadex.
CreateThread takes a LPTHREAD_START_ROUTINE function pointer,
which has got the signature DWORD WINAPI ThreadProc(LPVOID).
_beginthreadex takes a function with the signature
unsigned
On Thu, 27 Jul 2023, L. E. Segovia wrote:
Including winsock2.h without WIN32_LEAN_AND_MEAN causes bzlib.h to parse
as nonsense, due to an instance of #define char small in rpcndr.h
(included transitively from windows.h).
See: https://stackoverflow.com/a/27794577
Signed-off-by: L. E. Segovia
--
On Fri, 4 Aug 2023, Martin Storsjö wrote:
On Thu, 27 Jul 2023, L. E. Segovia wrote:
Including winsock2.h without WIN32_LEAN_AND_MEAN causes bzlib.h to parse
as nonsense, due to an instance of #define char small in rpcndr.h
(included transitively from windows.h).
See: https
On Tue, 8 Aug 2023, Casey Smalley wrote:
This patch changes the return instruction in the
tr_32x4 macro from br to ret.
Using ret properly hints that the branch is a
function return.
On devices that support BTI a landing pad is
required when branching with br, or the instruction
can be replace
On Sat, 5 Aug 2023, L. E. Segovia wrote:
Including winsock2.h or windows.h without WIN32_LEAN_AND_MEAN cause
bzlib.h to parse as nonsense, due to an instance of #define char small
in rpcndr.h.
See:
https://stackoverflow.com/a/27794577
Signed-off-by: L. E. Segovia
---
compat/atomics/win32/std
On Mon, 14 Aug 2023, L. E. Segovia wrote:
Including winsock2.h or windows.h without WIN32_LEAN_AND_MEAN cause
bzlib.h to parse as nonsense, due to an instance of #define char small
in rpcndr.h.
See:
https://stackoverflow.com/a/27794577
Reviewed-by: Martin Storsjö
Signed-off-by: L. E
On Sat, 19 Aug 2023, Gyan Doshi wrote:
ddc1cd5cdd breaks the checks for ole32 and thus prevents detection of DXVA2.
@
https://github.com/FFmpeg/FFmpeg/commit/ddc1cd5cdd2570bf3d6ab807ee0ecfacdf09431d
Thanks for noticing this, and sorry for the breakage. It seems like this
is fixable by adju
commit caused ole32 to no longer get detected, which caused
dxva2 to no longer be enabled. This gets fixed by this patch.
Signed-off-by: Martin Storsjö
---
configure | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/configure b/configure
index ff6a5c3600..04bb9fe9dd 100755
The fate-run.sh shell script exports LC_ALL=C before invoking the
test executables; this is probably done for consistency.
When executing Windows binaries with Wine, it normally handles
UTF-8 command line parameters just fine - but with LC_ALL set to
C, it treats them as plain ASCII.
As the unico
On Sun, 20 Aug 2023, Gyan Doshi wrote:
On 2023-08-20 02:42 am, Martin Storsjö wrote:
ddc1cd5cdd2570bf3d6ab807ee0ecfacdf09431d defined WIN32_LEAN_AND_MEAN
globally, which makes for much fewer transitive includes from
windows.h. With that define, CoTaskMemFree no longer gets
implicitly
On Mon, 21 Aug 2023, Martin Storsjö wrote:
The fate-run.sh shell script exports LC_ALL=C before invoking the
test executables; this is probably done for consistency.
When executing Windows binaries with Wine, it normally handles
UTF-8 command line parameters just fine - but with LC_ALL set to
On Wed, 23 Aug 2023, Jean-Baptiste Kempf wrote:
Hello,
On Mon, 21 Aug 2023, at 03:35, Steven Liu wrote:
Jean-Baptiste Kempf 于2023年8月20日周日 16:29写道：
Hello,
Hi
On Sun, 20 Aug 2023, at 06:10, Steven Liu wrote:
> As the enhanced rtmp Extending NetConnection connect Command section
> said, the
On Fri, 25 Aug 2023, Steven Liu wrote:
Add option named rtmp_enhanced_codec,
it would support hvc1,av01,vp09 now,
the fourcc is using Array of strings.
Signed-off-by: Steven Liu
---
doc/protocols.texi | 6 ++
libavformat/rtmpproto.c | 35 +++
2 files cha
On Fri, 25 Aug 2023, Steven Liu wrote:
> +list_len = (fourcc_str_len + 1) / 5;
> +// write the fourCcList field name
> +ff_amf_write_field_name(&p, "fourCcList");
> +
> +// write the fourcc array length
> +ff_amf_write_array_start(&p, list_len);
> +
> +
On Fri, 25 Aug 2023, Steven Liu wrote:
Martin Storsjö 于2023年8月25日周五 17:38写道：
On Fri, 25 Aug 2023, Steven Liu wrote:
>> > +list_len = (fourcc_str_len + 1) / 5;
>> > +// write the fourCcList field name
>> > +ff_amf_write
On Mon, 28 Aug 2023, Steinar H. Gunderson wrote:
When the marker writing code was merged from libav to FFmpeg
in dc62016c, it failed to take into account that the meaning of
cluster_pos had changed in bda5b662; in particular, the special
value for “I'm not currently working on a cluster” had cha
On Mon, 28 Aug 2023, Martin Storsjö wrote:
On Mon, 28 Aug 2023, Steinar H. Gunderson wrote:
When the marker writing code was merged from libav to FFmpeg
in dc62016c, it failed to take into account that the meaning of
cluster_pos had changed in bda5b662; in particular, the special
value for
On Thu, 31 Aug 2023, Andreas Rheinhardt wrote:
It includes various Windows-specific headers when compiling
for Windows and these sometimes cause issues: E.g. winbase.h
defines IGNORE, which clashes with a macro used in the Matroska
muxer (since 884653ee5be03ed38db957c14fad51b300611c8c) and demux
Hi,
Sorry for not tending to your patches sooner.
Unfortunately, this patchset is impossible to apply - there seems to be
garbled whitespace in the patch which would require me to manually apply
all the changes.
Can you try sending the patches again in a way that doesn't corrupt
whitespace?
On Thu, 14 Sep 2023, Logan.Lyu wrote:
Hi Martin,
You can try the attached patchset. If that doesn't work, My code branch
address is https://github.com/myais2023/FFmpeg/tree/hevc-aarch64
Thanks for the patches. Functionally, they seem to work, and the issues i
saw in the code are relatively
On Tue, 26 Sep 2023, Anton Khirnov wrote:
Quoting Andreas Rheinhardt (2023-09-26 01:54:30)
It is of no value to the user, because every muxer can always
be flushed with a NULL packet. As its documentation shows
("If not set, the muxer will not receive a NULL packet in
the write_packet function"
corrected, please let me
know.
在 2023/9/17 5:46, Martin Storsjö 写道:
On Thu, 14 Sep 2023, Logan.Lyu wrote:
Hi Martin,
You can try the attached patchset. If that doesn't work, My code branch
address is https://github.com/myais2023/FFmpeg/tree/hevc-aarch64
Thanks for the patches. Functi
---
Some patches had inconsistent semicolons when invoking these macros.
By omitting the trailing semicolon in the macro, it requires the
callers to use it consistently.
---
libavcodec/aarch64/hevcdsp_init_aarch64.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/libavcod
On Tue, 26 Sep 2023, Andreas Rheinhardt wrote:
Martin Storsjö:
On Tue, 26 Sep 2023, Anton Khirnov wrote:
Quoting Andreas Rheinhardt (2023-09-26 01:54:30)
It is of no value to the user, because every muxer can always
be flushed with a NULL packet. As its documentation shows
("If not set
On Wed, 27 Sep 2023, Rémi Denis-Courmont wrote:
It does not make much sense to me, but GCC somehow optimises the
inline assembler way even when the output are very obviously used.
This reverts commit 09731fbfc3a914ec4f6ffad60aa9062db6a8f6aa.
---
libavutil/riscv/timer.h | 4 ++--
1 file changed,
When linking the main tools, the object files to link are set up
via the variable OBJS-, but for the tools, we've only
used the target's list of dependencies.
In most cases, this has been fine, but it has caused specifying
the libraries to link in a duplicate fashion; the linking command
has looke
On Thu, 28 Sep 2023, Martin Storsjö wrote:
When linking the main tools, the object files to link are set up
via the variable OBJS-, but for the tools, we've only
used the target's list of dependencies.
In most cases, this has been fine, but it has caused specifying
the libraries to
Hi,
On Tue, 18 Aug 2020, Sebastian Pop wrote:
Unrolling by 4 the outer loop in yuv2planeX reduces the number of cache
accesses by 7.5%.
The values loaded for the filter are used in the 4 unrolled iterations and
avoids reloading 3 times the same values.
The performance was measured on an Arm64 N
On Sun, 20 Sep 2020, Jan Ekström wrote:
This can now be re-utilized in other places.
---
libavformat/movenc.c | 39 +--
1 file changed, 29 insertions(+), 10 deletions(-)
diff --git a/libavformat/movenc.c b/libavformat/movenc.c
index 12471c7d68..1962f2 1006
On Sun, 20 Sep 2020, Jan Ekström wrote:
This way we have a single location in movenc which utilizes the
CPB properties.
---
libavformat/movenc.c | 9 -
1 file changed, 4 insertions(+), 5 deletions(-)
Looks ok
// Martin
___
ffmpeg-devel mailin
On Sun, 20 Sep 2020, Jan Ekström wrote:
This is utilized by various media ingests to figure out the bit
rate of the content you are pushing towards it, so write it by
default for video, audio and subtitle tracks. It is only mentioned
for timed metadata sample descriptions in QTFF, so limit it on
On Sun, 20 Sep 2020, Jan Ekström wrote:
If the average bit rate cannot be calculated, such as in the case
of streamed fragmented mp4, utilize various available parameters
in priority order.
Tests are updated where the esds or btrt or ISML manifest boxes'
output changes.
---
libavformat/movenc.c
On Sun, 20 Sep 2020, Jan Ekström wrote:
The newly calculated average bit rate value is pretty much what is
being done here.
---
libavformat/movenc.c | 13 +++--
1 file changed, 3 insertions(+), 10 deletions(-)
Ok
// Martin
___
ffmpeg-devel ma
On Mon, 21 Sep 2020, Jan Ekström wrote:
This is utilized by various media ingests to figure out the bit
rate of the content you are pushing towards it, so write it by
default for video, audio and subtitle tracks. It is only mentioned
for timed metadata sample descriptions in QTFF, so limit it on
---
gas-preprocessor.pl | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/gas-preprocessor.pl b/gas-preprocessor.pl
index e51849b..bc36300 100755
--- a/gas-preprocessor.pl
+++ b/gas-preprocessor.pl
@@ -1204,7 +1204,7 @@ sub handle_serialized_line {
$line =~ s/\.arm/ARM/x
If preprocessing with cl.exe, the preprocessor doesn't take care of
concatenating continued lines.
---
gas-preprocessor.pl | 19 ---
1 file changed, 16 insertions(+), 3 deletions(-)
diff --git a/gas-preprocessor.pl b/gas-preprocessor.pl
index 126ee50..e9baeba 100755
--- a/gas-prep
When invoking gas-preprocessor to preprocess only to get the dependency
list, we still need to use cpp, as cl doesn't support the -M options.
This allows using gas-preprocessor in an environment with only MSVC
available, no GCC/Clang providing a cpp.exe, allowing building e.g.
dav1d with MSVC targ
On Wed, 23 Sep 2020, Martin Storsjö wrote:
When invoking gas-preprocessor to preprocess only to get the dependency
list, we still need to use cpp, as cl doesn't support the -M options.
This allows using gas-preprocessor in an environment with only MSVC
available, no GCC/Clang provid
Hi,
On Sun, 27 Sep 2020, Zhao Zhili wrote:
Fix #8840.
Steps to reproduce:
1. sender:
./ffmpeg -i test.mp4 -c copy -f rtsp -rtsp_transport udp
rtsp://localhost:12345/live.sdp
2. receiver:
./ffmpeg_g -y -rtsp_flags listen -timeout 100 -i
rtsp://localhost:12345/live.sdp -c copy test.mp4
---
v3
In listen mode with UDP transport, once the sender has sent
the TEARDOWN and closed the connection, poll will indicate that
one can read from the connection (indicating that the socket has
reached EOF and should be closed by the receiver as well). In this
case, parse_rtsp_message won't try to parse
On Sun, 27 Sep 2020, Zhao Zhili wrote:
1. Remove the assumption that the message method is TEARDOWN.
2. Don't ignore the error code of ff_rtsp_parse_streaming_commands.
---
libavformat/rtsp.c | 6 +-
1 file changed, 1 insertion(+), 5 deletions(-)
diff --git a/libavformat/rtsp.c b/libavformat
On Wed, 30 Sep 2020, Zhao Zhili wrote:
Hi Martin,
On 9/30/20 5:41 PM, Martin Storsjö wrote:
In listen mode with UDP transport, once the sender has sent
the TEARDOWN and closed the connection, poll will indicate that
one can read from the connection (indicating that the socket has
reached EOF
Hi,
On Thu, 1 Oct 2020, Andriy Gelman wrote:
On Wed, 30. Sep 12:41, Martin Storsjö wrote:
Hi,
On Sun, 27 Sep 2020, Zhao Zhili wrote:
> Fix #8840.
>
> Steps to reproduce:
> 1. sender:
> ./ffmpeg -i test.mp4 -c copy -f rtsp -rtsp_transport udp
rtsp://localhost:12345/live.sd
On Wed, 30 Sep 2020, Andreas Rheinhardt wrote:
There are two possible kinds of timecode tracks (with tag "tmcd") in the
mov muxer: Tracks created internally by the muxer and timecode tracks
sent by the user. If any of the latter exists, the former are
deactivated. The former all belong to anothe
On Thu, 1 Oct 2020, Andriy Gelman wrote:
On Thu, 01. Oct 22:00, Zhao Zhili wrote:
On 10/1/20 4:15 AM, Martin Storsjö wrote:
> On Wed, 30 Sep 2020, Zhao Zhili wrote:
>
> > Hi Martin,
> >
> > On 9/30/20 5:41 PM, Martin Storsjö wrote:
> > > In listen mode wi
On Mon, 28 Sep 2020, Jan Ekström wrote:
The generated text streams for chapters lack an AVStream since they
are but an internal concept within movenc.
Fixes #8190
---
libavformat/movenc.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/libavformat/movenc.c b/libavformat/
On Wed, 14 Oct 2020, lance.lmw...@gmail.com wrote:
From: Limin Wang
If the rtp input is discontinued, we'll drop all the new received packets now,
It's better to update the previous seq with the new seq so that the following
packets will not be dropped always.
Signed-off-by: Limin Wang
---
l
This makes sure that small seeks forward on https don't end up
doing new requests.
---
libavformat/tls_gnutls.c | 7 +++
libavformat/tls_libtls.c | 7 +++
libavformat/tls_mbedtls.c | 7 +++
libavformat/tls_openssl.c | 7 +++
libavformat/tls_schanne
On Thu, 29 Oct 2020, Andreas Rheinhardt wrote:
Martin Storsjö:
This makes sure that small seeks forward on https don't end up
doing new requests.
---
libavformat/tls_gnutls.c | 7 +++
libavformat/tls_libtls.c | 7 +++
libavformat/tls_mbedtls.c
This makes sure that small seeks forward on https don't end up
doing new requests.
---
libavformat/tls.h | 7 +++
1 file changed, 7 insertions(+)
diff --git a/libavformat/tls.h b/libavformat/tls.h
index 4f3315427d..d33af80454 100644
--- a/libavformat/tls.h
+++ b/libavformat/tls.h
@@ -56,10 +5
---
libavformat/tls.h | 13 +
libavformat/tls_gnutls.c | 11 ++-
libavformat/tls_libtls.c | 11 ++-
libavformat/tls_mbedtls.c | 11 ++-
libavformat/tls_openssl.c | 11 ++-
libavformat/tls_schannel.c
The previous threshold, 4 KB, maybe was reasonable when it was set
(in 2010), but in today's settings and with typical network speeds
and data sizes, it's pretty small. 32 KB probably is a more reasonable
default now, regardless of input.
---
libavformat/aviobuf.c | 2 +-
1 file changed, 1 inserti
On Fri, 30 Oct 2020, Martin Storsjö wrote:
The previous threshold, 4 KB, maybe was reasonable when it was set
(in 2010), but in today's settings and with typical network speeds
and data sizes, it's pretty small. 32 KB probably is a more reasonable
default now, regardles
On Thu, 29 Oct 2020, Andreas Rheinhardt wrote:
Martin Storsjö:
This makes sure that small seeks forward on https don't end up
doing new requests.
---
libavformat/tls_gnutls.c | 7 +++
libavformat/tls_libtls.c | 7 +++
libavformat/tls_mbedtls.c
On Wed, 4 Nov 2020, lance.lmw...@gmail.com wrote:
From: Limin Wang
Signed-off-by: Limin Wang
---
libavformat/rtspdec.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libavformat/rtspdec.c b/libavformat/rtspdec.c
index ef084a8..3abf34b 100644
--- a/libavformat/rtspdec.c
+++
The previous threshold, 4 KB, maybe was reasonable when it was set
(in 2010), but in today's settings and with typical network speeds
and data sizes, it's pretty small. 32 KB probably is a more reasonable
default now, regardless of input.
This changes the test references for two seek tests.
When
On Thu, 5 Nov 2020, Martin Storsjö wrote:
The previous threshold, 4 KB, maybe was reasonable when it was set
(in 2010), but in today's settings and with typical network speeds
and data sizes, it's pretty small. 32 KB probably is a more reasonable
default now, regardless of input.
Th
This fixes tests if built for x86 with x87 FPU.
---
This requires someone to upload a reference file.
---
tests/fate/mpc.mak | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/tests/fate/mpc.mak b/tests/fate/mpc.mak
index bb1c03d250..cde6e55177 100644
--- a/tests/fate/mpc.mak
+
On Wed, 11 Nov 2020, David Michael wrote:
There are two issues:
The unused global integer does not make it into the compiled object
file, so declare it static to correct this.
Umm, what? That sounds entirely counterintuitive to me.
If the global integer is non-static, it could be referred to
On Thu, 12 Nov 2020, Andreas Rheinhardt wrote:
Martin Storsjö:
This fixes tests if built for x86 with x87 FPU.
---
This requires someone to upload a reference file.
---
tests/fate/mpc.mak | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/tests/fate/mpc.mak b/tests/fate
When the check was added (in 3668701f9600, in 2015), some IO
functions returned 0 on EOF (in particular, the TCP protocol
did, but the TLS protocol returned AVERROR_EOF). Since
0e1f771d2200d in 2017, the TCP protocol also returns AVERROR_EOF
instead of 0, making the check for premature end never ha
IO functions are expected to return AVERROR_EOF instead of 0
nowadays. This is also expected by other higher level layers
within the http protocol itself (e.g. the reconnect mechanism).
---
libavformat/http.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/libavformat/http.
On Fri, 13 Nov 2020, Martin Storsjö wrote:
When the check was added (in 3668701f9600, in 2015), some IO
functions returned 0 on EOF (in particular, the TCP protocol
did, but the TLS protocol returned AVERROR_EOF). Since
0e1f771d2200d in 2017, the TCP protocol also returns AVERROR_EOF
instead of
On Thu, 12 Nov 2020, Martin Storsjö wrote:
On Thu, 12 Nov 2020, Andreas Rheinhardt wrote:
Martin Storsjö:
This fixes tests if built for x86 with x87 FPU.
---
This requires someone to upload a reference file.
---
tests/fate/mpc.mak | 4 +++-
1 file changed, 3 insertions(+), 1 deletion
On Fri, 20 Nov 2020, Jan Ekström wrote:
On Fri, Nov 13, 2020 at 12:17 AM Martin Storsjö wrote:
When the check was added (in 3668701f9600, in 2015), some IO
functions returned 0 on EOF (in particular, the TCP protocol
did, but the TLS protocol returned AVERROR_EOF). Since
0e1f771d2200d in
On Fri, 20 Nov 2020, Jan Ekström wrote:
On Fri, Nov 13, 2020 at 12:32 AM Martin Storsjö wrote:
IO functions are expected to return AVERROR_EOF instead of 0
nowadays. This is also expected by other higher level layers
within the http protocol itself (e.g. the reconnect mechanism
On Mon, 12 Oct 2020, Andriy Gelman wrote:
From: Andriy Gelman
As per the docs network initialization is required before ff_url_join().
Furthermore, because the ff_network_init() was skipped, this makes
one additional call to ff_network_close() if the stream exits without
errors.
The was forgo
On Mon, 12 Oct 2020, Andriy Gelman wrote:
From: Andriy Gelman
Signed-off-by: Andriy Gelman
---
libavformat/rtspdec.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/libavformat/rtspdec.c b/libavformat/rtspdec.c
index 1ff84cdd56..5a383b5a19 100644
--- a/libavformat/rtspdec
On Mon, 12 Oct 2020, Andriy Gelman wrote:
From: Andriy Gelman
Fixes #6334
Signed-off-by: Andriy Gelman
---
libavformat/rtspdec.c | 14 +-
1 file changed, 9 insertions(+), 5 deletions(-)
diff --git a/libavformat/rtspdec.c b/libavformat/rtspdec.c
index 5a383b5a19..7c617837ca 100644
On Mon, 12 Oct 2020, Andriy Gelman wrote:
From: Andriy Gelman
Make error check style consistent with rest of function.
---
libavformat/rtspdec.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/libavformat/rtspdec.c b/libavformat/rtspdec.c
index 7c617837ca..b9fbc45190 100
On Mon, 12 Oct 2020, Andriy Gelman wrote:
From: Andriy Gelman
In this error path ret still stores the number of bytes read in
ffurl_read().
Signed-off-by: Andriy Gelman
---
libavformat/rtsp.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavformat/rtsp.c b/libavformat/rtsp.c
index e9f
On Mon, 12 Oct 2020, Andriy Gelman wrote:
From: Andriy Gelman
In sdp_read_header() some ff_network_close() calls were missed.
Also in rtp_read_header() update comment to explain why a single
call to ff_network_close() is enough to cover all cases even if
sdp_read_header() returns an error.
S
These inline assembly functions rely on being inlined into the
caller, so that the parameter "int p" can be a known assembly time
constant, instead of a variable parameter.
__OPTIMIZE__ is a built-in define which is set by both GCC and Clang
(the two main compilers supporting our inline assembly)
This fixes building for arm targets with optimizations disabled.
---
libavutil/arm/intmath.h | 24 ++--
1 file changed, 18 insertions(+), 6 deletions(-)
diff --git a/libavutil/arm/intmath.h b/libavutil/arm/intmath.h
index 5311a7d52b..f19b21e98d 100644
--- a/libavutil/arm/intma
ymbols that are accessed from AArch64 assembly
as hidden, so that they are resolved fully at link time even without
the version script and -Wl,-Bsymbolic.
Signed-off-by: Martin Storsjö
---
v4: Moved the attribute definition to a new, standalone header (which
only depends on libavut
On Sat, 27 Aug 2022, Martin Storsjö wrote:
The AArch64 assembly accesses those symbols directly, without
indirection via e.g. the GOT on ELF. In order for this not to
require text relocations, those symbols need to be resolved fully
at link time, i.e. those symbols can't be interpo
On Fri, 26 Aug 2022, Martin Storsjö wrote:
This fixes building for arm targets with optimizations disabled.
---
libavutil/arm/intmath.h | 24 ++--
1 file changed, 18 insertions(+), 6 deletions(-)
diff --git a/libavutil/arm/intmath.h b/libavutil/arm/intmath.h
index 5311a7d52b
On Tue, 9 Aug 2022, Martin Storsjö wrote:
These were missed when h264_chroma_mc_func was changed in
e4a94d8b36c48d95a7d412c40d7b558422ff659c.
Signed-off-by: Martin Storsjö
---
libavcodec/arm/rv40dsp_init_arm.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
OK'd by Andre
On Mon, 22 Aug 2022, Hubert Mazur wrote:
Add vectorized implementation of nsse16 function.
Performance comparison tests are shown below.
- nsse_0_c: 707.0
- nsse_0_neon: 120.0
Benchmarks and tests run with checkasm tool on AWS Graviton 3.
Signed-off-by: Hubert Mazur
---
libavcodec/aarch64/me
On Mon, 22 Aug 2022, Hubert Mazur wrote:
Provide optimized implementation of vsad16 function for arm64.
Performance comparison tests are shown below.
- vsad_0_c: 285.0
- vsad_0_neon: 42.5
Benchmarks and tests are run with checkasm tool on AWS Graviton 3.
Signed-off-by: Hubert Mazur
---
libav
On Sat, 3 Sep 2022, Andreas Rheinhardt wrote:
It is advantageous for ff_crop_tab, as the base pointer used to
access this table is not the first element of it. But the real
base pointer is still at a constant offset from the code/the GOT
and can therefore be accessed relative to the instruction
601 - 700 of 1517 matches
Mail list logo
