| Summary: | memcheck/tests/linux/stack_changes fails on qemuarm64 builds | ||
|---|---|---|---|
| Product: | [Developer tools] valgrind | Reporter: | Tony Tascioglu <tony.tascioglu> |
| Component: | memcheck | Assignee: | Julian Seward <jseward> |
| Status: | CONFIRMED --- | ||
| Severity: | normal | CC: | mark, pjfloyd, rwmacleod, tony.tascioglu |
| Priority: | NOR | ||
| Version First Reported In: | 3.15 SVN | ||
| Target Milestone: | --- | ||
| Platform: | Other | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed/Implemented In: | ||
| Sentry Crash Report: | |||
| Attachments: | expected vs actual test result from stderr | ||
|
Description
Tony Tascioglu
2021-06-23 16:40:08 UTC
My apologies, that log is from an older testing attempt that was run without the --valgrind and --valgrind-lib parameters. When run correctly as follows, the test fails with the incorrect diff seen during testing. root@qemuarm64:/usr/lib/valgrind/ptest# ./tests/vg_regtest --valgrind=/usr/bin/valgrind --valgrind-lib=/usr/libexec/valgrind memcheck/tests/linux/stack_changes stack_changes: valgrind -q ./stack_changes *** stack_changes failed (stderr) *** == 1 test, 1 stderr failure, 0 stdout failures, 0 stderrB failures, 0 stdoutB failures, 0 post failures == memcheck/tests/linux/stack_changes (stderr) I have attached the stderr diff between the expected output and the result from the test that causes the failure. Created attachment 139623 [details]
expected vs actual test result from stderr
This works fine on x86_64, and it works when compiled on a Raspberry Pi, but when cross-compiled with Yocto and run on qemuarm64 (on an x86_64 build host), it fails close to 100% of the time. Is there something we can do to help gather more information to help debug this or make progress on the bug? Thanks, Tony Do you still have this problem? If so could you attach the unfiltered log file? Run perl tests/vg_regtest --keep-unfiltered memcheck/tests/linux/stack_changes and then attach stack_changes.stderr.out.unfiltered.out I'll try it on FreeBSD - reading the commit comments, it's only Darwin that doesn't work with this test. We are suddenly seeing this on fedora 40, 41 and eln: https://kojipkgs.fedoraproject.org//packages/valgrind/3.23.0/4.fc40/data/logs/aarch64/build.log https://kojipkgs.fedoraproject.org//packages/valgrind/3.23.0/4.fc41/data/logs/aarch64/build.log https://kojipkgs.fedoraproject.org//packages/valgrind/3.23.0/4.eln141/data/logs/aarch64/build.log It doesn't seem to consistently fail though. |