Attempt to make FreeBSD testsuite-green

This includes various bits from my attempt to make FreeBSD pass the testsuite:

• !13618 (closed)
• !13617 (closed)
• Various testing infrastructure changes

The runner configuration is documented in https://gitlab.haskell.org/bgamari/ghc-utils/-/blob/master/rel-eng/setting-up-freebsd.mkd?ref_type=heads.

added FreeBSD label

assigned to @bgamari

added 1 commit

• 3b79cd0b - gitlab-ci: Use system libffi

Compare with previous version

added 9 commits

• 918d8382 - testsuite: Unmark T14028 as broken on FreeBSD
• d0cca7a2 - rts/linker: Fix out-of-bounds mapping logic
• d6c2bdef - rts/linker: Clarify debug output
• d87a3a2e - gitlab-ci: Migrate FreeBSD runner tag to FreeBSD 14
• 54d87b9f - gitlab-ci: Reintroduce FreeBSD 14 job
• aca7c175 - gitlab-ci: Allow use of newer cabal-install bindists
• 57d5f89a - gitlab-ci: Enable documentation build on FreeBSD 14
• b1ad635d - configure: Implement ld override whitelist
• 20268354 - gitlab-ci: Use system libffi

Compare with previous version

added 1 commit

• 9a07d830 - testsuite: Mark linker_unload as broken on FreeeBSD

Compare with previous version

changed the description

mentioned in merge request chreekat/ghc!2

While the testsuite passes for me locally with this branch something appears to be amiss with Python in CI; the testsuite driver appears to fail while emitting no output. Unfortunately I am unable to reproduce this locally and I am out of time for this task. I will leave this for @Kleidukos to investigate when they bring up their runner.

added 2 commits

• fd54336a - gitlab-ci: Prefer system toolchain on FreeBSD
• 0584a454 - testsuite: Mark T21969 as broken on FreeBSD

Compare with previous version

Looking at dmesg, it appears that Python segfaulted:

pid 42703 (python3.11), jid 0, uid 1001: exited on signal 11 (core dumped)

It is possible that this is due to memory pressure.

I have raised the VM's memory limit to 16GB and now find that the job completes. T16180 has failed in the external interpreter way with a timeout in both jobs that have completed. We should investigate this.

@bgamari Interesting to know that 16GB is a hard floor for RAM, thanks!

I believe that the T16180 (as well as linker_unload) timeout is due to #25492 (closed).

added 64 commits

• 0584a454...f7cbffe2 - 54 commits from branch master
• 9b930eb5 - testsuite: Unmark T14028 as broken on FreeBSD
• d71ff9ab - gitlab-ci: Migrate FreeBSD runner tag to FreeBSD 14
• d91c2f70 - gitlab-ci: Reintroduce FreeBSD 14 job
• 2be3c4f6 - gitlab-ci: Allow use of newer cabal-install bindists
• 53ea9d14 - gitlab-ci: Enable documentation build on FreeBSD 14
• 30fd3991 - configure: Implement ld override whitelist
• e9fcf27f - gitlab-ci: Use system libffi on FreeBSD
• 672fc378 - testsuite: Mark linker_unload as broken on FreeeBSD
• 573ac92c - gitlab-ci: Prefer system toolchain on FreeBSD
• 1b0f99a1 - testsuite: Mark T21969 as broken on FreeBSD

Compare with previous version

added 2 commits

• 69671a94 - testsuite: Mark RestartEventLogging as broken on FreeBSD
• 664ec778 - testsuite: Mark T16180 as "broken" on FreeBSD

Compare with previous version

added 9 commits

• 0dcbc1b1 - gitlab-ci: Enable documentation build on FreeBSD 14
• 70fa3b2a - configure: Implement ld override whitelist
• 95a23b96 - gitlab-ci: Use system libffi on FreeBSD
• 3bd8be9f - testsuite: Mark linker_unload as broken on FreeeBSD
• e586afd8 - gitlab-ci: Prefer system toolchain on FreeBSD
• 0bed44aa - testsuite: Mark T21969 as broken on FreeBSD
• 7331ce3e - testsuite: Mark RestartEventLogging as broken on FreeBSD
• 3c6ea281 - testsuite: Mark T16180 as "broken" on FreeBSD
• 3120c80a - ci: Disable T16992 on FreeBSD due to memory requirement

Compare with previous version

added 3 commits

• f973e48e - testsuite: Mark RestartEventLogging as broken on FreeBSD
• a76d398d - testsuite: Mark T16180 as "broken" on FreeBSD
• 8b86203c - ci: Disable T16992 on FreeBSD due to memory requirement

Compare with previous version

added 2 commits

• 838bca2c - testsuite: Skip T16992 unless in slow speed
• 0bc18c68 - gitlab/ci: Don't clobber RUNTEST_ARGS

Compare with previous version

approved this merge request

Horray, CI green! Let's merge.

unassigned @bgamari

assigned to @marge-bot

I will attempt to batch this MR (!13698 (closed))...

unassigned @marge-bot

Breaks multiple lint jobs in https://gitlab.haskell.org/ghc/ghc/-/pipelines/104235

Batch MR !13698 (closed) failed: CI failed! I will retry later...

added 22 commits

• 0bc18c68...dfd1db48 - 9 commits from branch master
• dfd1db48...02491d4b - 3 earlier commits
• 5bfa3b75 - gitlab-ci: Allow use of newer cabal-install bindists
• bec6c369 - gitlab-ci: Enable documentation build on FreeBSD 14
• 49f3f92a - configure: Implement ld override whitelist
• e2a855e8 - gitlab-ci: Use system libffi on FreeBSD
• fc638b56 - testsuite: Mark linker_unload as broken on FreeeBSD
• 6a67a2ca - gitlab-ci: Prefer system toolchain on FreeBSD
• 4e0285c3 - testsuite: Mark T21969 as broken on FreeBSD
• 12973e73 - testsuite: Mark RestartEventLogging as broken on FreeBSD
• f0c3b998 - testsuite: Mark T16180 as "broken" on FreeBSD
• 91a26d7f - testsuite: Skip T16992 unless in slow speed

Compare with previous version

I have moved the errant commit into !13702 (closed) to allow this to go in.

assigned to @marge-bot

I will attempt to batch this MR (!13706 (closed))...

Batch MR !13706 (closed) failed: CI failed! I will retry later...

I will attempt to batch this MR (!13707 (closed))...

added full-ci label

unassigned @marge-bot

Breaks aarch64/riscv64 cross job configure checks in https://gitlab.haskell.org/ghc/ghc/-/pipelines/104295

Batch MR !13707 (closed) failed: CI failed! I will retry later...

Thanks @TerrorJack !

added 7 commits

• d8aeea2f - gitlab-ci: Use system libffi on FreeBSD
• 5ac3f203 - testsuite: Mark linker_unload as broken on FreeeBSD
• ebc03d0b - gitlab-ci: Prefer system toolchain on FreeBSD
• bfebcc29 - testsuite: Mark T21969 as broken on FreeBSD
• 9963ddf1 - testsuite: Mark RestartEventLogging as broken on FreeBSD
• c939ca01 - testsuite: Mark T16180 as "broken" on FreeBSD
• 4ab378d0 - testsuite: Skip T16992 unless in slow speed

Compare with previous version

I have dropped !13617 (closed) from this MR, avoiding this failure.

assigned to @marge-bot

added 29 commits

• 4ab378d0...99225996 - 17 commits from branch master
• 99225996...d7c0eb5a - 2 earlier commits
• 7246dacc - gitlab-ci: Reintroduce FreeBSD 14 job
• 4af936da - gitlab-ci: Allow use of newer cabal-install bindists
• cbf38c1b - gitlab-ci: Enable documentation build on FreeBSD 14
• d68107fb - gitlab-ci: Use system libffi on FreeBSD
• fea3b590 - testsuite: Mark linker_unload as broken on FreeeBSD
• ccf171ee - gitlab-ci: Prefer system toolchain on FreeBSD
• cfb34738 - testsuite: Mark T21969 as broken on FreeBSD
• 0b64e37c - testsuite: Mark RestartEventLogging as broken on FreeBSD
• 3b412019 - testsuite: Mark T16180 as "broken" on FreeBSD
• 57e3cab5 - testsuite: Skip T16992 unless in slow speed

Compare with previous version

I will attempt to batch this MR (!13715 (closed))...

We believe in you, Marge.

I will attempt to batch this MR (!13716 (closed))...

Merged in 57e3cab5

closed

mentioned in issue #25512 (closed)

mentioned in issue #14012 (closed)

mentioned in epic &5