libmpv-static and libmpv-shared are disabled by default resulting in the
following build failure when building with gl but without rpi, wayland
or x11:
Checking for OpenGL without platform-specific code (e.g. for libmpv) : libmpv-shared not found
Checking for OpenGL context support : gl-cocoa not found
You manually enabled the feature 'gl', but the autodetection check failed.
Here is an extract of wscript:
} , {
'name': '--plain-gl',
'desc': 'OpenGL without platform-specific code (e.g. for libmpv)',
'deps': 'libmpv-shared || libmpv-static',
'func': check_true,
}, {
'name': '--gl',
'desc': 'OpenGL context support',
'deps': 'gl-cocoa || gl-x11 || egl-x11 || egl-drm || '
+ 'gl-win32 || gl-wayland || rpi || '
+ 'plain-gl',
'func': check_true,
'req': True,
'fmsg': "No OpenGL video output found or enabled. " +
"Aborting. If you really mean to compile without OpenGL " +
"video outputs use --disable-gl.",
}, {
Enabling both the shared and static libraries is not allowed by mpv, so
we consider the BR2_STATIC_LIBS to be static, and otherwise (i.e.
BR2_SHARED_LIBS and BR2_SHARED_STATIC_LIBS) to be shared.
Fixes:
- http://autobuild.buildroot.org/results/590d2a8b6746ef071dfb439e42b636f81dbdc35d
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr:
- expand config log about shared/static icompatibility
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Fixes https://gitlab.com/buildroot.org/buildroot/-/jobs/1297337965
Commit 15a2f9b819 (package/{mesa3d, mesa3d-headers}: bump
version to 21.0.2) marked BR2_PACKAGE_MESA3D_DRI_DRIVER_SWRAST as legacy,
but forgot to update the defconfig. The SW rasterizer isn't really needed
with the Intel GPU, so just drop it.
In addition, X11 now needs some help with loading the modules in the correct
order, similar to how it was done for the test in commit 4a3639bad0
(support/testing: test_glxinfo load X11 modules in the right order).
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Update commit ID to include recent upstream fixes:
- Fix I and D cache synchronization issue (2e2f6faaf105)
- Add carriage return to correct menu formatting (2f6ea51dbb51)
- Add copyright info (7d3413d2ffd9)
- Expand the limit on the size of uboot when update it (623888127a0e)
Signed-off-by: Drew Fustini <drew@beagleboard.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Update the commit id to include upstream fixes:
- Fix print format in load_and_run_ddr(e976d186e69a)
- Update copyright info (f2b049b7fff2)
- Avoid chiplink address exception (86664be28e5d)
Signed-off-by: Drew Fustini <drew@beagleboard.org>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Fix the following build failure with gcc 11:
/data/buildroot-autobuilder/instance-0/output-1/build/qt5base-5.15.2/include/QtCore/../../src/corelib/global/qfloat16.h:300:7: error: 'numeric_limits' is not a class template
300 | class numeric_limits<QT_PREPEND_NAMESPACE(qfloat16)> : public numeric_limits<float>
| ^~~~~~~~~~~~~~
Fixes:
- http://autobuild.buildroot.org/results/9a7a987af40b8408ccdfcae4890008c7090b41a1
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Reviewed-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Disable -Werror to avoid the following build failure:
<command-line>: error: "_FORTIFY_SOURCE" redefined [-Werror]
MEDIA_BUILD_FATAL_WARNINGS option is available since version 18.2.0 and
6932fc0ffb
Fixes:
- http://autobuild.buildroot.org/results/52638d95312e464626d1c4047b3b26d4f57a1cd2
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Add -std=c++11 to fix the following build failure with gcc 11:
/data/buildroot-autobuilder/instance-0/output-1/host/include/cutl/shared-ptr/base.hxx:34:41: error: ISO C++17 does not allow dynamic exception specifications
34 | operator new (std::size_t, cutl::share) throw (std::bad_alloc);
| ^~~~~
Fixes:
- http://autobuild.buildroot.org/results/9cbb8be7a1d8ac5913fbc5e2a78c4c45b5daf8e2
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Building efibootmgr with a musl toolchain is possible.
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Disable dc3dd on riscv32 because of the size of time_t (riscv32 has
never had a 32-bit time, and has always been 64-bit from the onset):
In file included from getdate.y:40:
verify.h:132:30: error: negative width in bit-field 'verify_error_if_negative_size__'
132 | (struct { unsigned int verify_error_if_negative_size__: (R) ? 1 : -1; }))
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
verify.h:138:61: note: in expansion of macro 'verify_true'
138 | # define verify(R) extern int (* verify_function__ (void)) [verify_true (R)]
| ^~~~~~~~~~~
getdate.y:116:1: note: in expansion of macro 'verify'
116 | verify (LONG_MIN <= TYPE_MINIMUM (time_t) && TYPE_MAXIMUM (time_t) <= LONG_MAX);
| ^~~~~~
Fixes:
- http://autobuild.buildroot.org/results/267151dec9d2328a5f8c61ddf224219a4f617e5c
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Parallel build is broken since bump to version 2.03.12 in commit
80997acd35. Commits 4526078d1b, 8a313b019c, and a7186cd1ea tried to fix
that by only installing systemd units when appropriate.
It turns out that there are more cases where parallel build still fails:
http://autobuild.buildroot.org/results/995/995f46ee0033e34261ba7b24b61c41e7a088602b/
>>> lvm2 2.03.12 Installing to staging directory
[...]
[INSTALL] ioctl/libdevmapper.so
[CC] dmsetup.c
/usr/bin/make -C lib install_device-mapper
[CC] dmsetup.c
[...]
[CC] dmsetup
[CC] dmsetup
/nvme/rc-buildroot-test/scripts/instance-0/output-1/host/lib/gcc/arm-buildroot-linux-gnueabihf/10.3.0/../../../../arm-buildroot-linux-gnueabihf/bin/ld: /nvme/rc-buildroot-test/scripts/instance-0/output-1/host/arm-buildroot-linux-gnueabihf/sysroot/usr/lib/Scrt1.o: in function `_start':
(.text+0x54): undefined reference to `main'
collect2: error: ld returned 1 exit status
Makefile:60: recipe for target 'dmsetup' failed
Or:
http://autobuild.buildroot.org/results/a4e/a4ea87da502272dc2e677123b6fbcb0c23106f0b/
>>> lvm2 2.03.12 Installing to staging directory
[...]
[CC] dmsetup
/home/giuliobenetti/autobuild/run/instance-3/output-1/host/lib/gcc/arm-buildroot-linux-musleabihf/9.3.0/../../../../arm-buildroot-linux-musleabihf/bin/ld: /home/giuliobenetti/autobuild/run/instance-3/output-1/host/arm-buildroot-linux-musleabihf/sysroot/lib/Scrt1.o: in function `_start_c':
Scrt1.c:(.text._start_c+0x5c): undefined reference to `main'
collect2: error: ld returned 1 exit status
make[3]: *** [Makefile:61: dmsetup] Error 1
make[3]: Leaving directory '/home/giuliobenetti/autobuild/run/instance-3/output-1/build/lvm2-2.03.12/libdm/dm-tools'
make[2]: *** [../libdm/make.tmpl:315: dm-tools.device-mapper] Error 2
make[2]: *** Waiting for unfinished jobs....
[CC] dmsetup
[INSTALL] dmsetup
[...]
Similar traces in either case: it tries to build dmsetup twice, at
intall time instead of build time.
Fixes:
- http://autobuild.buildroot.org/results/995/995f46ee0033e34261ba7b24b61c41e7a088602b/
- http://autobuild.buildroot.org/results/a4e/a4ea87da502272dc2e677123b6fbcb0c23106f0b/
Note that this is just a workaround for a broken buildsystem anyway:
indeed, the build of dmsetup is done at install time, instead of build
time. More fixes should be worked on with upstream to properly fix the
issue...
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr:
- update after the three partial fix-commits
- extend commit log accordingly
- add new upstream failures
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Import a small patch from the upstream Bugzilla which is needed to allow
building WPE WebKit against uClibc.
Signed-off-by: Adrian Perez de Castro <aperez@igalia.com>
[yann.morin.1998@free.fr: add upstream commit refs in backported patch]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
WebKitGTK 2.32.1 includes fixes for building with the Musl libc, which
also makes it possible to use uClibc as well, therefore arrange
dependencies to allow selecting the package any of the C libraries is in
use. This is done by making the dependencies be more granular, basically
following what the wpewebkit package does.
In order to make make it build against uClibc a small patch that has
been submitted to the upstream's Bugzilla is needed.
Signed-off-by: Adrian Perez de Castro <aperez@igalia.com>
[yann.morin.1998@free.fr: add upstream commit refs in backported patch]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Commit [1] breaks Buildroot Makefile since a Makefile endif has been left
with no sense giving following error:
package/lvm2/lvm2.mk:61: *** extraneous 'endif'. Stop.
So let's remove that forgotten endif.
[1]: https://git.buildroot.net/buildroot/commit/?id=8a313b019c7d7e898186a8b08f9c25ae0194fa16
Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Commit ff0f55e381 (lvm2: replace !BR2_PACKAGE_LVM2_DMSETUP_ONLY by
BR2_PACKAGE_LVM2_STANDARD_INSTALL) changed a negative-logic option to a
positive-logic option.
However, it kept the ordering of the conditional block, which became a
negatice-logic condition.
This is confusing; let's fix that.
Reported-by: Pascal de Bruijn <p.debruijn@unilogic.nl>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Bump version to 2.92 by:
- removing local patches that have been upstreamed
- adding BR2_TOOLCHAIN_HAS_SYNC_4 dependency
- adding libblockdev dependencies
- adding libmount(contained in util-linux package) dependency
- moving download site to github since it's the new upstream address
- disabling all useless udisks's build option to shrink its size and t
use it as a daemon
Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Add Libs.private to libraw.pc to fix the following static build failure
with imagemagick which is raised since commit
2f47cfade4:
/home/giuliobenetti/autobuild/run/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/giuliobenetti/autobuild/run/instance-1/output-1/host/bin/../arm-buildroot-linux-uclibcgnueabi/sysroot/usr/lib/libjasper.a(jpg_enc.c.o): in function `jpg_encode':
jpg_enc.c:(.text+0x1f4): undefined reference to `jpeg_stdio_dest'
Fixes:
- http://autobuild.buildroot.org/results/88e43a1ea2059a684e50b0f5f2af407e8c6df2e1
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Fix CVE-2021-33516: An issue was discovered in GUPnP before 1.0.7 and
1.1.x and 1.2.x before 1.2.5. It allows DNS rebinding. A remote web
server can exploit this vulnerability to trick a victim's browser into
triggering actions against local UPnP services implemented using this
library. Depending on the affected service, this could be used for data
exfiltration, data tempering, etc.
Replace patch by upstream commit as current patch doesn't apply cleanly
https://discourse.gnome.org/t/security-relevant-releases-for-gupnp-issue-cve-2021-33516/6536https://gitlab.gnome.org/GNOME/gupnp/-/blob/gupnp-1.2.6/NEWS
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Disabling tests will fix the following build failures on riscv32:
select_poll_epoll.c:408:16: note: each undeclared identifier is reported only once for each function it appears in
select_poll_epoll.c: In function 'ppoll_fds_ulong_max':
select_poll_epoll.c:440:16: error: 'SYS_ppoll' undeclared (first use in this function); did you mean 'SYS_tkill'?
440 | ret = syscall(SYS_ppoll, ufds, ULONG_MAX, NULL, NULL);
| ^~~~~~~~~
| SYS_tkill
select_poll_epoll.c: In function 'pselect_invalid_fd':
select_poll_epoll.c:488:16: error: 'SYS_pselect6' undeclared (first use in this function); did you mean 'SYS_semctl'?
488 | ret = syscall(SYS_pselect6, fd + 1, &rfds, NULL, NULL, NULL, NULL);
| ^~~~~~~~~~~~
| SYS_semctl
Fixes:
- http://autobuild.buildroot.org/results/dd39188a7191efa512a51f18f4c34d9ee711a6a7
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
vlc unconditionally uses __NR_futex resulting in the following build
failure on riscv32:
linux/thread.c:53:20: error: '__NR_futex' undeclared (first use in this function)
53 | return syscall(__NR_futex, addr, op, val, to, addr2, val3);
| ^~~~~~~~~~
A patch that fixes this was submitted upstream, but rejected [1].
Fixes:
- http://autobuild.buildroot.org/results/be0f2243919d7a052b33118d8a901c922f8d88e7
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
[1] https://patches.videolan.org/patch/30581/
... similar to BR2_ENABLE_DEBUG.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
BR2_ENABLE_DEBUG should just steer the availability of debug symbols and
should have no negative effect on performance.
Introduction of 'assert' statements, 'debug'-type builds with additional
logging, etc. should be steered by BR2_ENABLE_RUNTIME_DEBUG instead.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
BR2_ENABLE_DEBUG should just steer the availability of debug symbols and
should have no negative effect on performance.
Introduction of 'assert' statements, 'debug'-type builds with additional
logging, etc. should be steered by BR2_ENABLE_RUNTIME_DEBUG instead.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Below are the flags set by zmqpp depending on the specified CONFIG variable:
CONFIG_FLAGS =
ifeq ($(CONFIG),debug)
CONFIG_FLAGS = -g -fno-inline -ftemplate-depth-1000
endif
ifeq ($(CONFIG),valgrind)
CONFIG_FLAGS = -g -O1 -DNO_DEBUG_LOG -DNO_TRACE_LOG
endif
ifeq ($(CONFIG),max)
CONFIG_FLAGS = -O3 -funroll-loops -ffast-math -finline-functions \
-fomit-frame-pointer -DNDEBUG
endif
ifneq (,$(findstring $(CONFIG),release loadtest))
CONFIG_FLAGS = -O3 -funroll-loops -ffast-math -finline-functions \
-fomit-frame-pointer -DNO_DEBUG_LOG -DNO_TRACE_LOG -DNDEBUG
endif
For the flags added with CONFIG=debug, '-g' is to be steered by the core
infrastructure (could be '-g1', '-g2' etc.)
The flag '-ftemplate-depth' is only a protection against incorrect code and
not really needed in Buildroot context.
Finally, the flag '-fno-inline' may be useful when really stepping through
zmqpp code, but is a very specific use case.
With the above in mind, not passing CONFIG=debug may actually be better.
Use 'CONFIG=buildroot' instead.
Note that we don't pass an empty 'CONFIG' to avoid confusion, as this
variable is also passed through to the variable BUILD_ENV, even though it is
currently unused.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
BR2_ENABLE_DEBUG should just steer the availability of debug symbols and
should have no negative effect on performance.
Introduction of 'assert' statements, 'debug'-type builds with additional
logging, etc. should be steered by BR2_ENABLE_RUNTIME_DEBUG instead.
The sofia-sip package was setting '--enable-ndebug' conditionally based on
BR2_ENABLE_DEBUG, and this would have to be updated to be based on
BR2_ENABLE_RUNTIME_DEBUG.
However, the sofia-sip option '--enable-ndebug' only sets the 'NDEBUG'
preprocessor macro, and the core package infrastructure already sets this
macro correctly based on BR2_ENABLE_RUNTIME_DEBUG.
This means that the explicit '--enable-ndebug' flag can be removed.
Suggested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
The preprocessor option NDEBUG, triggered by the configure option
'--enable-ndebug', should be read as 'no-debug'. When NDEBUG is set, asserts
are _disabled_.
The sofia-sip package had inverted logic, and set '--enable-ndebug' when
BR2_ENABLE_DEBUG was enabled, while it should be the other way around.
Reported-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
BR2_ENABLE_DEBUG should just steer the availability of debug symbols and
should have no negative effect on performance.
Introduction of 'assert' statements, 'debug'-type builds with additional
logging, etc. should be steered by BR2_ENABLE_RUNTIME_DEBUG instead.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
BR2_ENABLE_DEBUG should just steer the availability of debug symbols and
should have no negative effect on performance.
Introduction of 'assert' statements, 'debug'-type builds with additional
logging, etc. should be steered by BR2_ENABLE_RUNTIME_DEBUG instead.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
BR2_ENABLE_DEBUG should just steer the availability of debug symbols and
should have no negative effect on performance.
Introduction of 'assert' statements, 'debug'-type builds with additional
logging, etc. should be steered by BR2_ENABLE_RUNTIME_DEBUG instead.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
BR2_ENABLE_DEBUG should just steer the availability of debug symbols and
should have no negative effect on performance.
Introduction of 'assert' statements, 'debug'-type builds with additional
logging, etc. should be steered by BR2_ENABLE_RUNTIME_DEBUG instead.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
sysrepo explicitly sets CMAKE_BUILD_TYPE=Release, ignoring any possible
value of BR2_ENABLE_DEBUG (previously) or BR2_ENABLE_RUNTIME_DEBUG (now).
With the introduction of BR2_ENABLE_RUNTIME_DEBUG, this change should no
longer be necessary. Users that do not wish to have additional runtime
debugging just keep BR2_ENABLE_RUNTIME_DEBUG disabled (default value).
As the 'Debug' build type enables tests, disable them explicitly.
As the 'Debug' build type uses a custom REPO_PATH which does not exist on
target, force /etc/sysrepo like in the 'Release' build type.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Use the default build type for host-sysrepo.
In the current version of sysrepo, this happens to be 'Debug'.
As 'Debug' also enables tests, explicitly disable them.
Suggested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
flare-engine enables profiling when CMAKE_BUILD_TYPE is 'Debug'. The
Buildroot package explicitly avoided that by forcing CMAKE_BUILD_TYPE to
'RelWithDebInfo' when pkg-cmake.mk would normally set it to 'Debug'. Until
recently, this was the case when BR2_ENABLE_DEBUG was enabled.
A previous commit changed the condition under which CMAKE_BUILD_TYPE=Debug
was set, from BR2_ENABLE_DEBUG=y to BR2_ENABLE_RUNTIME_DEBUG=y, so logically
the flare-engine package would have to be updated accordingly.
However, apart from the profiling flag, the flare-engine package only uses
CMAKE_BUILD_TYPE to determine flags that Buildroot wants to control itself,
like optimization and debugging flags.
This means we can fake CMAKE_BUILD_TYPE to a value that has no meaning for
flare-engine itself, without needing to check BR2_ENABLE_DEBUG nor
BR2_ENABLE_RUNTIME_DEBUG.
Incidentally, this trick was already done in case
BR2_TOOLCHAIN_HAS_GCC_BUG_85180 was true, so move that line out of this
condition.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
The CMAKE_BUILD_TYPE is currently set as 'Debug' in case BR2_ENABLE_DEBUG is
set, and as 'Release' in other cases. However, while the description of
BR2_ENABLE_DEBUG is to enable debug symbols (no runtime impact), the 'Debug'
build type in CMake can actually have runtime impact. For one, because it
does not set -DNDEBUG like is done for 'Release', but also because packages
may do custom things based on it.
The question of which CMAKE_BUILD_TYPE Buildroot should set, be it 'Debug',
'Release', 'RelWithDebInfo' or others, has come up several times in the
past. See some references below:
- July 2016: switch from Debug to RelWithDebInfo:
https://git.buildroot.org/buildroot/commit/?id=4b0120183404913f7f7788ef4f0f6b51498ef363
- October 2016: switch from RelWithDebInfo back to Debug:
https://git.buildroot.org/buildroot/commit/?id=104bb29e0490bfb487e2e665448dd3ca07fcc2b5
and changes to make sure Buildroot's flags are respected:
https://git.buildroot.org/buildroot/commit/?id=12494ef48f893684d0800e7f6fe39a2ceaed0451
- August 2017: bug #10246 - "BR2_ENABLE_DEBUG does not have the expected
effect for cmake packages"
https://bugs.busybox.net/show_bug.cgi?id=10246
- August 2017: mail thread following bug #10246:
http://lists.busybox.net/pipermail/buildroot/2017-August/200778.html
In the last mail thread, Samuel Martin confirmed that the 'Release' build
type could be used in all cases, because Buildroot is actually making sure
that the optimization flags are those determined by Buildroot, not the
defaults of cmake, thanks to commit 12494ef48f.
But Arnout Vandecappelle objected to using always 'Release', stating that
users may actually want the extra assertions.
With the introduction of BR2_ENABLE_RUNTIME_DEBUG, Buildroot can now cater
for all cases:
- use CMAKE_BUILD_TYPE=Release by default. This makes sure that there is no
unexpected performance degradation triggered by enabling BR2_ENABLE_DEBUG.
- users can optionally enable BR2_ENABLE_RUNTIME_DEBUG if they want runtime
debug info like assertions, at the risk of introducing performance
degradation. In this case, we switch to CMAKE_BUILD_TYPE=Debug.
- orthogonally to the above, BR2_ENABLE_DEBUG still determines passing the
'-g' flag to enable debug symbols, and BR2_OPTIMIZE_X still determines the
used optimization flags.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
The 'assert' statement in glibc honors the 'NDEBUG' preprocessor macro: if
it is set, then the assert statement is compiled away.
Define this 'NDEBUG' macro when BR2_ENABLE_RUNTIME_DEBUG is disabled (the
default case).
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Reviewed-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Some packages have optional runtime assertions, extra traces, or other
elements that can help in debugging problems. However, such runtime elements
can negatively influence performance.
In a test program performing 100K gRPC calls from a client to a local server
and receiving the returned response, we see following execution time:
- runtime debug enabled: 1065 seconds
- runtime debug disabled: 48 seconds
This is more than a factor 20 (!) difference. Analysis shows that the
problem mostly stems from libabseil-cpp (a dependency of gRPC) which enables
mutex deadlock analysis when the preprocessor flag 'NDEBUG' is not set,
which adds a 'backtrace()' call on every lock/unlock. Potentially worse,
when libunwind is enabled and linked with the test program, 'backtrace()' is
not provided by glibc but by libunwind itself.
For production systems, users expect good performance out-of-the-box. In the
example above, the difference is huge and unless explicitly tested and
analyzed, users may not realize that the performance could be much better.
Address this problem by introducing a new option BR2_ENABLE_RUNTIME_DEBUG,
which can be used by packages or package infrastructures to set the
necessary flags.
Note that BR2_ENABLE_RUNTIME_DEBUG is orthogonal to BR2_ENABLE_DEBUG: the
former changes runtime behavior, while the latter is only expected to add
debug symbols to the build. Today, the cmake build system does introduce a
runtime impact when BR2_ENABLE_DEBUG is set, but that will be rectified in a
subsequent commit.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
