PR#17: Add riscv64 support - rpms/abseil-cpp

rpms / abseil-cpp

#17 Add riscv64 support

Opened 3 months ago by u2fsdgvkx1. Modified 3 months ago

rpms/ u2fsdgvkx1/abseil-cpp rawhide into rawhide

Add riscv64 support

Songsong Zhang • 3 months ago

b114b50

abseil-cpp.spec

file modified

+9 -2

		`@@ -3,7 +3,7 @@`

		`Name: abseil-cpp`
		`Version: 20240116.0`
		`- Release: 1%{?dist}`
		`+ Release: 2%{?dist}`
		`Summary: C++ Common Libraries`

		`# The entire source is Apache-2.0, except:`
		`@@ -110,7 +110,11 @@`
		`%cmake_install`

		`%check`
		`- %ctest`
		`+ CTEST_OPT_ARGS=""`
		`+ %ifarch riscv64`
		`+ CTEST_OPT_ARGS+="-E absl_log_format_test\|absl_failure_signal_handler_test"`
		`+ %endif`
		`+ %ctest $CTEST_OPT_ARGS`

		`%files`
		`%license LICENSE`
		`@@ -234,6 +238,9 @@`
		`%{_libdir}/pkgconfig/absl_*.pc`

		`%changelog`
		`+ * Wed Feb 21 2024 Songsong Zhang <U2FsdGVkX1@gmail.com> - 20240116.0-2`
		`+ - Add riscv64 support`
		`+`
		`* Wed Jan 24 2024 Benjamin A. Beasley <code@musicinmybrain.net> - 20240116.0-1`
		`- Update to 20240116.0`

u2fsdgvkx1 commented 3 months ago

risc-v based on rawhide branch

cc @rjones @davidlt @tekkamanninja

music commented 3 months ago

Is there an upstream bug report or something that we can link to? I’m curious if the skipped tests represent something that actually needs to be fixed.

music commented 3 months ago

Otherwise, the method of skipping these tests looks OK. I confirmed that it ends up as

+ /usr/bin/ctest --test-dir redhat-linux-build --output-on-failure --force-new-ctest-process -j16 -E 'absl_log_format_test|absl_failure_signal_handler_test'

and the | doesn’t end up unquoted and interpreted as a shell pipeline.

A minor simplification would be to omit the line

CTEST_OPT_ARGS=""

and write

%ctest ${CTEST_OPT_ARGS-}

u2fsdgvkx1 commented 3 months ago

Is there an upstream bug report or something that we can link to? I’m curious if the skipped tests represent something that actually needs to be fixed.

https://openkoji.iscas.ac.cn/kojifiles/work/tasks/5819/5075819/build.log
^build log^

The test unit doesn't pass in risc-v, I don't know what the reason is yet, I think it's a bug in risc-v

zuul commented 3 months ago

Build succeeded.
https://fedora.softwarefactory-project.io/zuul/buildset/449f580379b14f60ba1072ae60e873d7

check-for-arches : SUCCESS in 1m 37s
check-for-eln : SUCCESS in 17s
eln-rpm-scratch-build : SUCCESS in 14m 52s
rpm-scratch-build : SUCCESS in 16m 12s
rpm-scratch-build-s390x : SUCCESS in 2h 03m 59s (non-voting)
rpm-scratch-build-ppc64le : SUCCESS in 28m 28s (non-voting)
rpm-scratch-build-i686 : SUCCESS in 16m 36s (non-voting)
rpm-scratch-build-aarch64 : SUCCESS in 18m 27s (non-voting)
rpm-linter : FAILURE in 5m 45s (non-voting)
rpm-rpminspect : SUCCESS in 9m 48s (non-voting)
check-for-sti-tests : SUCCESS in 14s
check-for-fmf-tests : SUCCESS in 14s
rpm-install-test : SUCCESS in 1m 13s
Skipped 2 jobs

music commented 3 months ago

I don’t like that the skipped tests represent bugs, but nobody knows what the bugs really are, and they haven’t been reported anywhere, so they can’t be tracked and are unlikely to be fixed.

On the other hand, I don’t have access to riscv64 hardware or a straightforward way to test in emulation, so there isn’t really anything I can do to make this PR better.

I guess this could be merged, under the circumstances.

music commented 3 months ago

@rmattes , any thoughts?