2G is too much for qemu-mips and perhaps also for real mips devices
as we use qemu-usermode during build to run host pieces like mksnapshot
they fail, reducing the allocation range helps
Fixes
| LD_LIBRARY_PATH=/mnt/b/yoe/master/build/tmp/work/mips32r2-yoe-linux/nodejs/14.16.1-r0/node-v14.16.1/out/Release/lib.host:/mnt/b/yoe/master/build/tmp/work/mips32r2-yoe-linux/nodejs/14.16.1-r0/node-v14.16.1
/out/Release/lib.target:$LD_LIBRARY_PATH; export LD_LIBRARY_PATH; cd ../tools/v8_gypfiles; mkdir -p /mnt/b/yoe/master/build/tmp/work/mips32r2-yoe-linux/nodejs/14.16.1-r0/node-v14.16.1/out/Release/obj.target/v
8_snapshot/geni; "/mnt/b/yoe/master/build/tmp/work/mips32r2-yoe-linux/nodejs/14.16.1-r0/node-v14.16.1/out/Release/v8-qemu-wrapper.sh" "/mnt/b/yoe/master/build/tmp/work/mips32r2-yoe-linux/nodejs/14.16.1-r0/nod
e-v14.16.1/out/Release/mksnapshot" --turbo_instruction_scheduling "--target_os=linux" "--target_arch=mips" --startup_src "/mnt/b/yoe/master/build/tmp/work/mips32r2-yoe-linux/nodejs/14.16.1-r0/node-v14.16.1/ou
t/Release/obj.target/v8_snapshot/geni/snapshot.cc" --embedded_variant Default --embedded_src "/mnt/b/yoe/master/build/tmp/work/mips32r2-yoe-linux/nodejs/14.16.1-r0/node-v14.16.1/out/Release/obj.target/v8_snap
shot/geni/embedded.S" --no-native-code-counters
|
| #
| # Fatal process OOM in CodeRange setup: allocate virtual memory
| #
|
| /mnt/b/yoe/master/build/tmp/work/mips32r2-yoe-linux/nodejs/14.16.1-r0/node-v14.16.1/out/Release/v8-qemu-wrapper.sh: line 7: 2292880 Trace/breakpoint trap (core dumped) PSEUDO_UNLOAD=1 qemu-mips -r 3.2.0 -
Signed-off-by: Khem Raj <raj.khem@gmail.com>
So far, we have been trying to build nodejs-native and use the native
host binaries from there, which has worked out ok but always changes
when major upgrade is done, since more binaries or places are required
to be captured. This patch changes this approach to use qemu-user to run
these binaries under during cross build. This lets them run closer to
upstream build process and also removes dependency on nodejs-native
Signed-off-by: Khem Raj <raj.khem@gmail.com>
This is latest maintained LTS release
Forward patches and drop backported patch
License-Update: Update copyright year and drop license of
deps/http_parser as this component is removed in this version
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Commit 836912ab changed the logic in this recipe to stop looking for
"ppc64le". This caused the ppc64le systems used by me to stop working.
There wasn't much in the commit message on why this change occurred but
ppc64le is definitely still needed.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Uprev nodejs in order to fix CVE-2020-8277.
This CVE allows an attacker to trigger a DNS request for a host
of their choice, which could trigger a Denial of Service in
nodejs versions < 12.19.1.
See https://nvd.nist.gov/vuln/detail/CVE-2020-8277 for details.
CVE: CVE-2020-8277
Signed-off-by: Stacy Gaikovaia <Stacy.Gaikovaia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Backport a patch from upstream to take care of build failure e.g.
| ../deps/v8/src/codegen/arm/cpu-arm.cc:38:16: error: write to reserved register 'R7'
| asm volatile("svc 0\n"
| ^
| 1 error generated.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Use system brotli via PACKAGECONFIG by default. So far,
nodejs had been built using its embedded copy of brotli,
which we generally try to avoid, for the known reasons
(independent updates, cve & license checks, etc).
The nodejs patches to enable this have been submitted.
brotli is in meta-oe, so enabling this by default should
not be a problem.
Signed-off-by: André Draszik <andre.draszik@jci.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
During the python3 / nodejs update, the dependencies weren't updated, so
using system-gyp ends up trying to use the python2 version of system-
gyp, which will of course fail.
Fixing this to depend on the python3 version of gyp still doesn't
doesn't make things work, though:
ERROR: nodejs-native-12.14.1-r0 do_configure: Execution of '.../nodejs-native/12.14.1-r0/temp/run.do_configure.26054' failed with exit code 1:
gyp: Error importing pymod_do_mainmodule (ForEachFormat): No module named 'ForEachFormat' while loading dependencies of .../nodejs-native/12.14.1-r0/node-v12.14.1/node.gyp while trying to load .../nodejs-native/12.14.1-r0/node-v12.14.1/node.gyp
Error running GYP
The reason is commit fff922afee6e ("deps,build: compute torque_outputs in v8.gyp")
in NodeJS v12, where they modified their bundled version of gyp to
become incompatible with the upstream version of gyp by adding extra /
unusual search paths to gyp.
Since I'm not sure how to deal with that when using system-gyp, and because
the original intention for using system-gyp was to make the previous nodejs
version compatible with python3 by ultimately switching to the python3 version
of system-gyp which isn't necessary anymore, and given nobody else seems to
be using this PACKAGECONFIG, just drop it.
Signed-off-by: André Draszik <andre.draszik@jci.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
This change makes the parsing go though, we still might have build
issues, which will be reported in world builds seprately
Signed-off-by: Khem Raj <raj.khem@gmail.com>
There are so many useful modules written for node.js I would like to use in
C++/Qt projects.
Run-tested both variants by installing node-red with npm and running it.
Signed-off-by: Andreas Müller <schnitzeltony@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
host usually are using gcc/x86_64/aarch64 and all of them have gcc which
can support atomic intrinsics, but not all of them install libatomic by
default e.g. centos-7, so asking for libatomic unconditionally may not
work always
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Andreas Müller <schnitzeltony@gmail.com>
Acked-by: Tim Orling <timothy.t.orling@linux.intel.com>
Part of the NodeJS build builds V8, which at some stage tries to call
ar with all objects with *absolute* paths (and printf the command line
first).
This will fail if the build path is too long:
make[1]: execvp: printf: Argument list too long
when trying to create Release/obj.target/deps/v8/gypfiles/libv8_base.a
via below gyp-generated out/Makefile rule:
cmd_alink_thin = rm -f $@ && $(AR.$(TOOLSET)) crsT $@ $(filter %.o,$^)
i.e. something like
printf rm -f Release/obj.target/deps/v8/gypfiles/libv8_base.a && arm-poky-linux-musleabi-gcc-ar crsT Release/obj.target/deps/v8/gypfiles/libv8_base.a ...
The above failure happened on a build-directory S with 204 characters
on a Jenkins machine.
While one could probably increase the ulimit on that specific machine,
that would be a pretty specific build machine fix which would need to
be applied everywhere, or switch to non-verbose builds / compilation,
but fortunately we can change all object references to be relative to
the build directory itself by setting the builddir_name make variable
and thus avoid the other two possible work-arounds.
Signed-off-by: André Draszik <git@andred.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
This overrides yocto-provided build flags with its own, e.g we get
arm-poky-linux-musleabi-g++ -mthumb -mfpu=neon -mfloat-abi=hard -mcpu=cortex-a7 \
... \
-march=armv7-a -mfpu=neon -mfloat-abi=hard -marm
Causing the latter to override the former, and compiler warnings:
cc1plus: warning: switch '-mcpu=cortex-a7' conflicts with '-march=armv7-a' switch
Patch this out, so that yocto-provided flags take precedence.
Note that in reality the same should probably be done for all the other
supported architectures, too.
Note that this also switches to Thumb(2) mode (in my case). No obvious
problems have been noted during compilation or runtime.
Upstream-Status: Inappropriate [oe-specific]
Signed-off-by: André Draszik <git@andred.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Now that there is a gyp (python2) recipe in meta-python,
allow its use instead of the bundled one for when
meta-python is enabled.
At the same time, unconditionally inherit pythonnative
as in either case gyp actually uses python, and the
build machine's python and installed modules should
never be used.
Signed-off-by: André Draszik <git@andred.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
We can delete bundled deps where system-provided counterparts
should be used instead.
Amongst others, this ensures they are not used accidentally.
Signed-off-by: André Draszik <git@andred.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
NodeJS comes with an embedded, patched, version of gyp.
Normally, gyp supports compiling for the build machine, e.g.
native tools that need to be compiled to run during the build,
and for the host, using different variables, e.g. CC and
CC.host, etc.
Most of this has been patched out in the NodeJS version of gyp,
and essentially it only supports compiling using one compiler -
${CC}. This modification excludes LDFLAGS for native tools, and
those still evaluate LDFLAGS.host (only).
While this modified behaviour is OK for the OE use-case of building
native and target tools separately, it means that this recipe can
not work as-is with standard gyp, and wrong LDFLAGS are being used
for some of the tools compiled (torque) in either case.
By setting the make variables that gyp-generated makefiles inspect,
we support use of unpatched gyp, and we ensure that all tools
are compiled with correct LDFLAGS in either case.
This now also allows us to drop the patch that had been applied to
work-around this problem.
Signed-off-by: André Draszik <git@andred.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
We can't enable this by default, though, as nghttp2 is
in the meta-networking layer, which might not be enabled.
At least this gives people a simple way to do so if
they want.
Signed-off-by: André Draszik <git@andred.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Use system libuv via PACKAGECONFIG by default. So far,
nodejs had been built using its embedded copy of libuv,
which we generally try to avoid, for the known reasons
(independent updates, cve & license checks, etc).
Signed-off-by: André Draszik <git@andred.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Use system c-ares via PACKAGECONFIG by default. So far,
nodejs had been built using its embedded copy of c-ares,
which we generally try to avoid, for the known reasons
(independent updates, cve & license checks, etc).
Notes:
* otherwise nodejs uses its bundled version of c-ares
* the PACKAGECONFIG variable is 'ares' so as to be in
line with other uses of this (wget & curl recipes in
OE core)
Signed-off-by: André Draszik <git@andred.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
* fixes:
ERROR: QA Issue:
/usr/lib/node_modules/npm/node_modules/node-gyp/gyp/samples/samples
contained in package nodejs-npm requires /usr/bin/python, but no
providers found in RDEPENDS_nodejs-npm? [file-rdeps]
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Update nodejs to active 10.x LTS release 10.16.0.
- make dependencies configurable
License checksum changed due to added bundled dependencies to:
- siphash
- large_pages
- brotli
Signed-off-by: Andrej Valek <andrej.valek@siemens.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
GCC 4.8 compatibility was lost moving from 10.15.1 to 10.15.3. Add
another wrapper for std::make_unique to restore it.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
- do not drop target LDFLAGS
- even if the target LDFLAGS have been specified, tools like torque was
linked with system libraries
Signed-off-by: Andrej Valek <andrej.valek@siemens.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
- replace all occurrences of std::make_unique with wrapper
- sync indent level
Signed-off-by: Andrej Valek <andrej.valek@siemens.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Khem Raj
Andrew Geissler
Clément Péron
Sean Nyekjaer
Stacy Gaikovaia
Jaga
André Draszik
Martin Jansa
Guillaume Burel
Andreas Müller
André Draszik
Adrian Bunk
Andrej Valek
Brad Bishop