[Bug 297681] usr.bin/xinstall/install_test:stdin test is flaky in CI on riscv64

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=297681

--- Comment #5 from Siva Mahadevan <siva@FreeBSD.org> ---
> The output you show does not correspond to the checked-in code.
Oops, that repro was from when I attempted to `chflags nouchg` locally in a
detached testing branch. Fixed now by hard-resetting to main HEAD. Here are the
new local repros on a clean main:

timeout case:
Standard output:
Executing command [ mkdir dst ]
Executing command [ install - dst ]
Executing command [ test ! -e dst/file ]
Executing command [ install - dst/file ]
Executing command [ cmp -s file dst/file ]
Executing command [ rm dst/file ]

Standard error:
Files left in work directory after failure: check.eWp4ct, dst, file
===> Broken tests
usr.bin/xinstall/install_test:stdin  ->  broken: Test case body timed out 
[300.008s]

permission denied case:

Standard output:
Executing command [ mkdir dst ]
Executing command [ install - dst ]
Executing command [ test ! -e dst/file ]
Executing command [ install - dst/file ]
Executing command [ cmp -s file dst/file ]
Executing command [ rm dst/file ]

Standard error:
Fail: incorrect exit status: 1, expected: 0
stdout:

stderr:
rm: dst/file: Operation not permitted

Files left in work directory after failure: dst, file
===> Failed tests
usr.bin/xinstall/install_test:stdin  ->  failed: atf-check failed; see the
output of the test for details  [1.097s]

These repros were done back to back with the same source and same test command.

> Jenkins clearly shows the test timing out, not failing to delete dst/file.

That is on the most recent run:
https://ci.freebsd.org/job/FreeBSD-main-riscv64-test/16689/testReport/

On the previous run, it passes:
https://ci.freebsd.org/job/FreeBSD-main-riscv64-test/16688/testReport/

On other runs, it fails with the output I've seen:

https://ci.freebsd.org/job/FreeBSD-main-riscv64-test/16687/testReport/
https://ci.freebsd.org/job/FreeBSD-main-riscv64-test/16670/testReport/
https://ci.freebsd.org/job/FreeBSD-main-riscv64-test/16666/testReport/

What I mean is that I am locally able to reproduce all three cases, making it
flaky. My mistake in submitting the repro output with my local experiment
aside, it still should be reproducible by others.

-- 
You are receiving this mail because:
You are on the CC list for the bug.