env: fix regression of `--ignore-signal=PIPE` by Ecordonnier · Pull Request #9618 · uutils/coreutils

Ecordonnier · 2025-12-09T20:33:29Z

revert env: use Command::exec() instead of libc::execvp() #9614 because Command::exec() resets the default signal handler for SIGPIPE, interfering with the option --ignore-signal=PIPE
add regression-test for --ignore-signal=PIPE

github-actions · 2025-12-09T20:52:13Z

GNU testsuite comparison:

Skip an intermittent issue tests/tail/overlay-headers (fails in this run but passes in the 'main' branch)

github-actions · 2025-12-10T12:49:13Z

GNU testsuite comparison:

Skipping an intermittent issue tests/tail/overlay-headers (passes in this run but fails in the 'main' branch)

github-actions · 2025-12-12T10:34:40Z

GNU testsuite comparison:

Skipping an intermittent issue tests/tail/overlay-headers (passes in this run but fails in the 'main' branch)

github-actions · 2025-12-16T21:03:36Z

GNU testsuite comparison:

Skip an intermittent issue tests/tail/inotify-dir-recreate (was skipped on 'main', now failing)

github-actions · 2025-12-17T15:39:16Z

GNU testsuite comparison:

Skipping an intermittent issue tests/tail/overlay-headers (passes in this run but fails in the 'main' branch)
Congrats! The gnu test tests/tail/inotify-dir-recreate is now passing!

sylvestre · 2025-12-30T23:03:49Z

please fix the conflicts

Ecordonnier · 2026-01-05T00:28:48Z

please fix the conflicts

I have fixed the conflicts. However while doing it I have noticed that the test fails if the default "yes" is from uutils coreutils on the machine running the test (it works with "yes" from GNU coreutils). This is because uutils yes unconditionally restores the sigpipe handler to its default value (by calling enable_pipe_errors().

I could fix it in this PR as well, but it is not directly related in principle.

github-actions · 2026-01-05T00:42:29Z

GNU testsuite comparison:

GNU test failed: tests/tail/tail-n0f. tests/tail/tail-n0f is passing on 'main'. Maybe you have to rebase?
Note: The gnu test tests/csplit/csplit-heap is now being skipped but was previously passing.

github-actions · 2026-01-05T01:10:06Z

GNU testsuite comparison:

Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)

codspeed-hq · 2026-02-07T20:43:16Z

CodSpeed Performance Report

Merging this PR will degrade performance by 3.57%

_{Comparing Ecordonnier:eco/env/execvp (11708b5) with main (f828549)}

Summary

❌ 1 regressed benchmark
✅ 283 untouched benchmarks
⏩ 38 skipped benchmarks¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	`BASE`	`HEAD`	Efficiency
❌	Simulation	`cp_large_file[16]`	377.8 µs	391.8 µs	-3.57%

38 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

Ecordonnier · 2026-02-07T23:10:07Z

I have changed the test to use seq instead of yes. For some unknown reason, yes doesn't seem to exit when receiving EPIPE in the openbsd CI job. Thus the openbsd test job was hanging forever.

- revert uutils#9614 because `Command::exec()` resets the default signal handler for SIGPIPE, interfering with the option `--ignore-signal=PIPE` - add regression-test for --ignore-signal=PIPE Signed-off-by: Etienne Cordonnier <ecordonnier@snap.com>

github-actions · 2026-02-07T23:23:42Z

GNU testsuite comparison:

GNU test failed: tests/pr/bounded-memory. tests/pr/bounded-memory is passing on 'main'. Maybe you have to rebase?

github-actions · 2026-02-07T23:39:10Z

GNU testsuite comparison:

Skip an intermittent issue tests/shuf/shuf-reservoir (fails in this run but passes in the 'main' branch)
Skip an intermittent issue tests/sort/sort-stale-thread-mem (fails in this run but passes in the 'main' branch)
Congrats! The gnu test tests/pr/bounded-memory is no longer failing!

Ecordonnier · 2026-02-08T00:48:15Z

@sylvestre @cakebaker the PR is good to go. The CodSpeed failure is in cp and is unrelated to this PR.

ChrisDryden · 2026-02-09T20:37:16Z

I was keeping an eye on this PR for when the 9.10 upgrade happened since the env-signal gnu test began failing because it added a test for SIGPIPE, when running the tests locally I was able to confirm that this fixes all of the tests related to env and sigpipes. The only reason the GNU test is still failing is because we're missing RTMIN and RTMAX in our signal library

ChrisDryden · 2026-02-09T23:33:15Z

Was fuzzing to see the other impacts of this change by using strace and by trying to explore what exec does under the hood and I haven't come across any other behaviors that this would regress and it fixes multiple test cases the env signal handler gnu test. We should also do a similar change in timeout eventually and it will allow us to remove a bunch of unsafe blocks.

- revert uutils#9614 because `Command::exec()` resets the default signal handler for SIGPIPE, interfering with the option `--ignore-signal=PIPE` - add regression-test for --ignore-signal=PIPE Signed-off-by: Etienne Cordonnier <ecordonnier@snap.com>

Ecordonnier mentioned this pull request Dec 9, 2025

env: use Command::exec() instead of libc::execvp() #9614

Merged

Ecordonnier force-pushed the eco/env/execvp branch from c362cfb to 62e4f7b Compare December 10, 2025 12:15

Ecordonnier force-pushed the eco/env/execvp branch from 62e4f7b to 67a4741 Compare December 12, 2025 10:18

Ecordonnier force-pushed the eco/env/execvp branch from 67a4741 to 8a264a0 Compare December 16, 2025 20:47

Ecordonnier force-pushed the eco/env/execvp branch from 8a264a0 to 73a0844 Compare December 17, 2025 15:20

Ecordonnier force-pushed the eco/env/execvp branch from 73a0844 to 98e348d Compare January 5, 2026 00:26

Ecordonnier force-pushed the eco/env/execvp branch from 98e348d to cffd508 Compare January 5, 2026 00:47

Ecordonnier force-pushed the eco/env/execvp branch 2 times, most recently from 6130ebb to 0cd6f10 Compare February 7, 2026 20:16

Ecordonnier force-pushed the eco/env/execvp branch from 0cd6f10 to ec859ea Compare February 7, 2026 23:08

env: fix "--ignore-signal=PIPE"

11708b5

- revert uutils#9614 because `Command::exec()` resets the default signal handler for SIGPIPE, interfering with the option `--ignore-signal=PIPE` - add regression-test for --ignore-signal=PIPE Signed-off-by: Etienne Cordonnier <ecordonnier@snap.com>

Ecordonnier force-pushed the eco/env/execvp branch from ec859ea to 11708b5 Compare February 7, 2026 23:24

ChrisDryden merged commit e4e95a7 into uutils:main Feb 9, 2026
196 of 198 checks passed

Uh oh!

Conversation

Ecordonnier commented Dec 9, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Dec 9, 2025

Uh oh!

github-actions bot commented Dec 10, 2025

Uh oh!

github-actions bot commented Dec 12, 2025

Uh oh!

github-actions bot commented Dec 16, 2025

Uh oh!

github-actions bot commented Dec 17, 2025

Uh oh!

sylvestre commented Dec 30, 2025

Uh oh!

Ecordonnier commented Jan 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Jan 5, 2026

Uh oh!

github-actions bot commented Jan 5, 2026

Uh oh!

codspeed-hq bot commented Feb 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CodSpeed Performance Report

Merging this PR will degrade performance by 3.57%

Summary

Performance Changes

Footnotes

Uh oh!

Ecordonnier commented Feb 7, 2026

Uh oh!

github-actions bot commented Feb 7, 2026

Uh oh!

github-actions bot commented Feb 7, 2026

Uh oh!

Ecordonnier commented Feb 8, 2026

Uh oh!

ChrisDryden commented Feb 9, 2026

Uh oh!

ChrisDryden commented Feb 9, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Ecordonnier commented Dec 9, 2025 •

edited

Loading

Ecordonnier commented Jan 5, 2026 •

edited

Loading

codspeed-hq bot commented Feb 7, 2026 •

edited

Loading