As we're about to remove the rpi-userland package, drop support for it
from package/ffmpeg.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
opencv3's code is not compatible with newer versions of ffmpeg, and
opencv3 is no longer maintained, so we have no choice but to disable
its ffmpeg support.
Fixes:
https://autobuild.buildroot.net/results/9ae3911583cccb6362f33cd82e5eaafb059fdc76/
It's not clear which ffmpeg version bump broken the build exactly, but
this issue is definitely present in 2025.02.x as the following
defconfig fails to build in a similar way on 2025.02.x:
BR2_aarch64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN_AARCH64_GLIBC_STABLE=y
BR2_PACKAGE_FFMPEG_NONFREE=y
# BR2_PACKAGE_FFMPEG_FFMPEG is not set
# BR2_PACKAGE_FFMPEG_INDEVS is not set
# BR2_PACKAGE_FFMPEG_OUTDEVS is not set
BR2_PACKAGE_OPENCV3=y
BR2_PACKAGE_OPENCV3_LIB_VIDEOIO=y
BR2_PACKAGE_OPENCV3_BUILD_PERF_TESTS=y
BR2_PACKAGE_OPENCV3_WITH_FFMPEG=y
BR2_PACKAGE_OPENCV3_INSTALL_DATA=y
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
This reverts commit fecc337d21,
re-adding optional support for pulseuadio in ffmpeg.
This is possible now that the circular dependency:
pulseaudio -> libsoxr -> ffmpeg -> pulseaudio
has been broken between libsoxr and ffmpeg.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This reverts commit e44aff6b1b which added
support for linking ffmpeg against libpulse if it is present to provide
libavdevice support for the PulseAudio protocol. However, it turns out
that this creates a dependency cycle when BR2_PACKAGE_LIBSOXR is
enabled, since the pulseaudio daemon can use libsoxr for resampling and
libsoxr can use the FFT implementation from ffmpeg.
Resolving this would probably mean breaking up the pulseaudio package
into the client libraries and the daemon proper (a standalone libpulse
without the daemon is still useful since it can talk to a remote daemon
via the network or more likely today talk to a local instance of
PipeWire, which also implements the PulseAudio protocol).
For now, break the dependency cycle by reverting to the original state
(no pulseaudio client support in ffmpeg).
Reported-by: Titouan Christophe <titouan.christophe@mind.be>
Signed-off-by: Florian Larysch <fl@n621.de>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Version 6.1.2 of ffmpeg fails to build with GCC 14.x due to the V4L2
ioctl code:
libavdevice/v4l2.c:137:17: error: assignment to ‘int (*)(int, long unsigned int, ...)’ from incompatible pointer type ‘int (*)(int, int, ...)’ [-W
incompatible-pointer-types]
137 | s->ioctl_f = prefix ## ioctl; \
| ^
libavdevice/v4l2.c:151:9: note: in expansion of macro ‘SET_WRAPPERS’
151 | SET_WRAPPERS();
| ^~~~~~~~~~~~
This has been fixed upstream in the release/6.1 branch, which has 27
fixes on top of 6.1.2. The commits necessary to fix our issue are:
f71076c009f84917e7a0f2f1ece86b718de2d8d3 configure: improve check for POSIX ioctl
60593d6c06c9b610359bd6af26a268feff1293eb configure: restore autodetection of v4l2 and fbdev
However, since all other commits are fixes, we believe bumping to the
latest commit in the release/6.1 branch is a better idea.
This allows to drop
0008-libavcodec-arm-mlpdsp_armv5te-fix-label-format-to-wo.patch, which
is upstream as of:
4c688845a50f7dce3af9afebe60f0f7a493c4f07 libavcodec/arm/mlpdsp_armv5te: fix label format to work with binutils 2.43
Note that we set FFMPEG_CPE_ID_VERSION to get proper CVE matching even
with FFMPEG_VERSION being set to n6.1.2-27-ge16ff06adb. One who have
ideally set FFMPEG_VERSION to n$(FFMPEG_CPE_ID_VERSION)-ge16ff06adb,
but that makes check-package unhappy with:
WARNING: package/ffmpeg/ffmpeg.mk:7: expecting package version to be set before CPE_ID_VERSION
Fixes:
https://autobuild.buildroot.net/results/fe1574443acd50ca7e576bb4beb24467be1713e3/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Remaining "nios2" strings come from existing patches that
are not removed with along with BR2_nios2 symbol.
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since the update of ffmpeg to 6.1.1 in commit
fa36a6375d there are
a lot of microblaze build failures.
The compilation failure is triggered by GCC Bug 68485.
We need to disable optimisations, or ffmpeg would add -O3, which would
be present after the -O0 we add, overriding it.
Fixes:
- http://autobuild.buildroot.net/results/524/524cef193aec30471e55eb0eecfa39f0d6011a91
And many more.
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
[yann.morin.1998@free.fr: explain why --disable-optimizations]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
This commit allows the package patches to be applied with fuzz factor 0.
The fuzz factor specifies how many lines of the patch can be inexactly
matched, so the value 0 requires all lines to be exactly matched.
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Arnout Vandecappelle <arnout@mind.be>
Fix the following build failure:
/tmp/ccw5V2Ti.s: Assembler messages:
/tmp/ccw5V2Ti.s:2013: Error: selected processor does not support `itt gt' in Thumb mode
/tmp/ccw5V2Ti.s:2014: Error: Thumb does not support conditional execution
/tmp/ccw5V2Ti.s:2015: Error: Thumb does not support conditional execution
/tmp/ccw5V2Ti.s:2017: Error: selected processor does not support `it le' in Thumb mode
/tmp/ccw5V2Ti.s:2018: Error: Thumb does not support conditional execution
/tmp/ccw5V2Ti.s:2020: Error: selected processor does not support `it gt' in Thumb mode
/tmp/ccw5V2Ti.s:2021: Error: Thumb does not support conditional execution
Fixes:
http://autobuild.buildroot.net/results/3fc/3fc68f756cf1661673941c411c4a2477e901e361/
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Fix the following build failure:
In file included from libavcodec/ppc/audiodsp.c:31:
libavcodec/ppc/audiodsp.c: In function 'scalarproduct_int16_altivec':
./libavutil/ppc/util_altivec.h:123:5: error: implicit declaration of function 'vec_vsx_ld'; did you mean 'vec_vslh'? [-Werror=implicit-function-declaration]
123 | vec_vsx_ld(offset, b)
| ^~~~~~~~~~
Fixes:
- http://autobuild.buildroot.org/results/b772d285f978ff9bc3b07872d009633c943f20b1
VSX is indeed an extension to AltiVec, so havinf VSX implies having
AltiVec [0], so we can condition he altivec support on LE ,on VSX being
available.
To be noted, however, is that ffmpeg has a configre switch dedicated to
VSX: --enable-vsx. We do not use it add support for that here, as we are
just fixing the AltiVec support. Adding VSX configure flag is left as an
excercise for a future feature addition.
[0] https://en.wikipedia.org/wiki/AltiVec#VSX_(Vector_Scalar_Extension)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr:
- add comment in .mk
- exend commit log to explain VSX implies AltiVec
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
mips_32 is not supported by ffmpeg and it tries to build with loongson3
SIMD support that leads to build failure due to:
/tmp/ccFO2LRa.s: Assembler messages:
/tmp/ccFO2LRa.s:15314: Error: opcode not supported on this processor: mips32 (mips32) `dmult $2,$6'
/tmp/ccFO2LRa.s:15316: Error: opcode not supported on this processor: mips32 (mips32) `dsrl $2,$2,32'
So let's --disable-asm to prevent using those unsupported opcodes for every
mips architecture according to Arnout.
Fixes:
http://autobuild.buildroot.net/results/f01/f01d9cedec8e1b371308d0f7af561a75883fa27c/
Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
-latomic was added to extralibs to fix static build of ffmpeg in commit
fc8798197b. However, extralibs is not
added to libavformat.pc resulting in the following static build failure
of motion:
/home/buildroot/autobuild/instance-1/output-1/host/opt/ext-toolchain/bin/../lib/gcc/arm-buildroot-linux-uclibcgnueabi/9.3.0/../../../../arm-buildroot-linux-uclibcgnueabi/bin/ld: /home/buildroot/autobuild/instance-1/output-1/host/bin/../arm-buildroot-linux-uclibcgnueabi/sysroot/usr/lib/libavformat.a(fifo.o): in function `fifo_init':
/home/buildroot/autobuild/instance-1/output-1/build/ffmpeg-4.4/libavformat/fifo.c:519: undefined reference to `__atomic_store_8'
So add a patch to add extralibs (and so -latomic) to all pkg-config
files
Fixes:
- http://autobuild.buildroot.org/results/62ec618e40081a250b8129ec6f5a178eb06fba1d
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Adding these patches brings our ffmpeg package to the same level as the
kodi-provided ffmpeg package allowing us to switch the kodi build to
use the system-provided ffmpeg instead of its internal build when kodi
is bumped to version 19.0-Matrix.
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
opencv 2.x has seen no release since July 2018 and all buildroot
packages are already compatible with opencv3 so drop it to ease the
addition of opencv4 which is not fully backward compatible with opencv3
as "a lot of C API from OpenCV 1.x has been removed.":
https://opencv.org/opencv-4-0.
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Add a patch to fix build without ssse3.
Update the license hash due to various changes since August 2019 [0]:
- some files were relicensed from GPL-2.0+ to LGPL-2.1+; no impact on
the licensing information;
- clarification about the licensing when linked with external
libraries, which now requires passing --enable-version3; Buildroot
is not concerned, because it passes --disable-version3;
- various clarification about applicable licensing terms.
[0] http://git.videolan.org/?p=ffmpeg.git;a=history;f=LICENSE.md;h=613070e1b631a919a74c862f2a75a0f249a631ff;hb=8e12af29d1a3f95c9e952d78354e3c8b1c0431a8
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
[yann.morin.1998@free.fr:
- fix the SoB order in the patch
- clarify the impact of the licensing changes
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
cbs_jpeg_split_fragment in libavcodec/cbs_jpeg.c in FFmpeg 4.2.2 has a
heap-based buffer overflow during JPEG_MARKER_SOS handling because of a
missing length check.
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Bernd Kuhls