On 2026-04-28 01:48, Mike Rapoport wrote:
On Mon, Apr 27, 2026 at 01:19:07PM -0400, Luiz Capitulino wrote:
On 2026-04-27 12:51, Mike Rapoport wrote:
On Mon, Apr 27, 2026 at 08:56:27AM -0400, Luiz Capitulino wrote:
On 2026-04-26 06:58, Mike Rapoport wrote:
And there's another issue: when running the migration test individually,
private_anon_htlb gets skipped. I guess it's because the previous test
is restoring the HugeTLB state:
It could be.
I pushed the updated version that has the fixes for both zombie and
signal issues:
https://git.kernel.org/pub/scm/linux/kernel/git/rppt/linux.git/log/?h=mm-selftest/v3
Would be great if you can test it in your setup.
This works! Both issues are fixed: the migration test completes w/o
hanging and when I run it individually all tests run.
Thanks for confirming!
One last question: should we munmap() the mappings in this test? Though
I understand this goes beyond your series goal.
We could, but the tests _exit() anyway as a part of kselftest_harness, so
it does not really matter.
OK. Just to clarify that I meant exercising the munmap() path, not that
we're not freeing resources. But if we know that _exit() is sufficient
for this, then it's fine.
exit tears down all the mappings, but the code path is slightly different
than calling munmap().
We can add munmap() calls there, but let's do it separately from this
series.
Agreed.
