Fix the folloing build failure with protobuf (enabled since commit
31c68a449e) and gcc 5.3.0:
[ 53%] Building CXX object modules/dnn/CMakeFiles/opencv_dnn.dir/opencv-caffe.pb.cc.o
In file included from /home/peko/autobuild/instance-1/output-1/per-package/opencv3/host/opt/ext-toolchain/mips64el-buildroot-linux-uclibc/include/c++/5.5.0/atomic:38:0,
from /home/peko/autobuild/instance-1/output-1/per-package/opencv3/host/mips64el-buildroot-linux-uclibc/sysroot/usr/include/google/protobuf/io/coded_stream.h:115,
from /home/peko/autobuild/instance-1/output-1/build/opencv3-3.4.12/buildroot-build/modules/dnn/opencv-caffe.pb.h:23,
from /home/peko/autobuild/instance-1/output-1/build/opencv3-3.4.12/buildroot-build/modules/dnn/opencv-caffe.pb.cc:4:
/home/peko/autobuild/instance-1/output-1/per-package/opencv3/host/opt/ext-toolchain/mips64el-buildroot-linux-uclibc/include/c++/5.5.0/bits/c++0x_warning.h:32:2: error: #error This file requires compiler and library support for the ISO C++ 2011 standard. This support must be enabled with the -std=c++11 or -std=gnu++11 compiler options.
#error This file requires compiler and library support \
^
Fixes:
- http://autobuild.buildroot.org/results/7caf175af039054a032b8f63b458b3940d9ec0f3
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
OpenCV-3's buildsystem will try to detect ccache and use it if
available. This may yield a system-installed ccache.
However, in Buildroot, ccache is entirely hidden away and handled in the
toolchain wrapper.
Forcibly disable detection of ccache.
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The imx-gpu-viv install libOpenCL.so.1.2 library and cl.h header,
so declare it as a libopencl provider.
With this support we can select the clinfo package provided by
Buildroot instead of the one provided by imx-gpu-viv package.
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Cc: Gary Bisson <gary.bisson@boundarydevices.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Add all configure options through DHCP_CONFIG_OPTS and avoid splitting
lines when they are less than 80 characters
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
With BR2_RELRO_PARTIAL or BR2_RELRO_FULL, our toolchain wrapper will
forcibly add -Wl,-z,relro to any call to the actual compiler. This
usually works OK, because gcc will only use those options it needs for
the compile step it has to carry: pre-processing, compiling, assembling,
or linking, and ignore those options it does not need.
Excpt in one case: when -v is passed standalone, with no input file,
then gcc will falsely believe it has to do a link stage;
$ gcc -Wl,-z,relro -v
[...]
/usr/bin/ld: /usr/lib/gcc/x86_64-linux-gnu/9/../../../x86_64-linux-gnu/Scrt1.o: in function `_start':
(.text+0x24): undefined reference to `main'
collect2: error: ld returned 1 exit status
Fixing that in our wrapper will not be easy, because we'd have to detect
there is no input file. Doing so would probably require we support
almost all gcc options to differentiate between the parameter of an
option (e.g. -I /some/path) from an actual inpout file. This would not
be very robust, and would have a high risk od breaking when we introduce
the next gcc version.
Since it seems that only rhash is affected, due to its inventive,
custom, hand-written configure script, we just patch it to be a bit more
robust in the face of a compiler that could not accept -v, and fallback
to --version.
Fixes:
- http://autobuild.buildroot.org/results/8605c16cc28316954ce8b9dcc266974390c5da20
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr:
- retain "$CC -v" as default, fallback to "$CC --version", in the hope
that it stands better chance with upstream
- write a commit log to explain the actual root-cause of the build
failure
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Multi format codec (MFC) is the IP present in Samsung Exynos series SoCs
for video encoding/decoding operations.
Signed-off-by: Stefan Agner <stefan@agner.ch>
[yann.morin.1998@free.fr: add all FW versions]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
In buildroot, stripping for the target is configured and implemented
with the global `BR2_STRIP_strip` option that drive the stripping in
the target-finalize step.
So, we explicitly disable stripping at build time for the target
variants.
For the host variants, however, we don't much care about symbols and
stuff, but smaller executables will hopefully load faster than bigger
ones (disputable, given that sections in ELF files are paged-in
on-demand), so we explictly enable stripping.
Signed-off-by: Gleb Mazovetskiy <glex.spb@gmail.com>
[yann.morin.1998@free.fr:
- add burb about the target-finalize step
- enable stripping for host variants
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
quazip requires cmake 3.15 since version 1.0 and
89e7c201f0818adc8224
The rationale for this requirement is that "default locations for the
install(TARGETS command based on the GNUInstallDirs package were only
added in 3.14" and "3.15 is not that much of a difference from 3.14 and
it introduced a lot of useful UI improvements.":
https://github.com/stachenov/quazip/issues/82
Fixes:
- http://autobuild.buildroot.org/results/5d848a46109aef448ea1d1b857a500d9461dc2d9
Note: we also have some patches to allow some packages to build with
cmake-3.10, and this will not be tenable over the long run.
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr: add the "note"]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
