Created attachment 241954 [details] www/qt6-webengine: fix build on armv7 The attached patch improved some existing patches so they also work on armv7 (taken from devel/electron19). Unfortunately the build then fails in a later stage due to a crash in node.
(In reply to Robert Clausecker from comment #0) Hi Robert, thanks for the patch. Regarding the node crash you mentioned: Would it help to get a successful build by expanding the conditional in the Makefile to include armv7 as well, e.g.: > .if ${ARCH} == i386 || ${ARCH} == armv7 > EXTRA_PATCHES+= ${PATCHDIR}/extra-patch-i386-compress_files_js > .endif If this helps then we should rename "extra-patch-i386-compress_files_js" to a more generic name, e.g. "extra-patch-32bit-compress_files_js" and update its boiler plate as well.
This indeed gets us further, but now the build fails here: [ 72% 21685/30031] LINK host/root_store_tool FAILED: host/root_store_tool /usr/local/bin/python3.9 "../../../../../qtwebengine-everywhere-src-6.4.2/src/3rdparty/chromium/build/toolchain/gcc_lin k_wrapper.py" --output="host/root_store_tool" -- /usr/local/libexec/ccache/c++ -Wl,--build-id=sha1 -fPIC -Wl,-z,noexecs tack -Wl,-z,relro -Wl,-z,now -no-canonical-prefixes -Wl,-O2 -Wl,--gc-sections -rdynamic -pie -Wl,--disable-new-dtags -L /usr/local/lib -fstack-protector-strong -L/usr/local/lib -o "host/root_store_tool" -Wl,--start-group @"host/root_stor e_tool.rsp" -Wl,--end-group -lpthread -lkvm -lexecinfo -lutil -levent -lnss3 -lsmime3 -lnssutil3 -lplds4 -lplc4 -lnsp r4 -ldl -lz ld: error: undefined symbol: OPENSSL_cpuid_setup >>> referenced by crypto.c >>> boringssl/crypto.o:(.text+0x0) in archive host/obj/third_party/boringssl/libboringssl.a c++: error: linker command failed with exit code 1 (use -v to see invocation) This looks like something I can patch though. Will investigate and report back when I'm done.
Created attachment 242586 [details] www/qt6-webengine: fix build on armv7 Got it! Patch now fixes the build successfully. Will commit with my next batch.
Comment on attachment 242586 [details] www/qt6-webengine: fix build on armv7 (In reply to Robert Clausecker from comment #3) That's great! The updated patch LGTM (with KDE hat on), only the PORTREVISION needs to be adjusted, as qt6-webengine is already at 6.4.2_5 as this time of writing. NB: Perhaps a brief explanation in the commit message of why the PORTREVISION gets bumped, would make sense as well to avoid unnecessary questions.
(In reply to Kai Knoblich from comment #4) I bumped portrevision due to nontrivial changes in the code which should not but may impact other architectures. Unfortunately once I transfered the patch from my development tree to the current ports tree, it now no longer clears the node.js error from earlier: /usr/local/bin/python3.9 ../../../../../qtwebengine-everywhere-src-6.4.2/src/3rdparty/chromium/third_party/node/node.py ../../../../../qtwebengine-everywhere-src-6.4.2/src/3rdparty/chromium/third_party/devtools-frontend/src/scripts/build/compress_files.js --file_list __third_party_devtools-frontend_src_compress__wrkdirs_usr_ports_www_qt6-webengine_work_.build_src_core_target_toolchain_target__rule..rsp Traceback (most recent call last): File "/wrkdirs/usr/ports/www/qt6-webengine/work/.build/src/core/Release/armv7/../../../../../qtwebengine-everywhere-src-6.4.2/src/3rdparty/chromium/third_party/node/node.py", line 58, in <module> RunNode(sys.argv[1:]) File "/wrkdirs/usr/ports/www/qt6-webengine/work/.build/src/core/Release/armv7/../../../../../qtwebengine-everywhere-src-6.4.2/src/3rdparty/chromium/third_party/node/node.py", line 53, in RunNode raise RuntimeError('Command \'%s\' failed\n%s' % (' '.join(cmd), err)) RuntimeError: Command '/usr/local/bin/node ../../../../../qtwebengine-everywhere-src-6.4.2/src/3rdparty/chromium/third_party/devtools-frontend/src/scripts/build/compress_files.js --file_list __third_party_devtools-frontend_src_compress__wrkdirs_usr_ports_www_qt6-webengine_work_.build_src_core_target_toolchain_target__rule..rsp' failed ... My development tree is based on an older ports tree, so there might have been an update inbetween causing the extra patch from i386 to no longer suffice.
I managed to fix the build by changing the BROTLI_PARAM_QUALITY to 4 in the extra patch. Do you approve of this change?
(In reply to Robert Clausecker from comment #6) If a successful build of qt6-webengine on i386 with this change can be confirmed as well: Ship it!
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=dd2c7b778901a7b310d37f044c0018de1603c37f commit dd2c7b778901a7b310d37f044c0018de1603c37f Author: Robert Clausecker <fuz@FreeBSD.org> AuthorDate: 2023-05-03 14:59:15 +0000 Commit: Robert Clausecker <fuz@FreeBSD.org> CommitDate: 2023-06-05 16:53:00 +0000 www/qt6-webengine: fix build on armv7 Improve two patches with more complete variants from devel/electron19. Patch bundled boringssl with additional patches from the same port. Avoid an OOM condition by using the same extra patch previously only used on i386. In that patch, decreate BROTLI_PARAM_QUALITY further to 4 to avoid flakiness on armv7. Obtained from: devel/electron19 Approved by: kai (with hat: kde, maintainer) PR: 271228 www/qt6-webengine/Makefile | 6 +-- ...iles_js => extra-patch-32bit-compress_files_js} | 2 +- ...mium_third__party_boringssl_BUILD.generated.gni | 9 ++-- ..._boringssl_src_crypto_cpu__arm__freebsd.c (new) | 55 ++++++++++++++++++++++ ...y_chromium_third__party_skia_src_core_SkCpu.cpp | 25 +++++++++- ...romium_v8_src_base_platform_platform-freebsd.cc | 2 +- 6 files changed, 90 insertions(+), 9 deletions(-)