[PATCH v3 3/4] selftests/arm64: GCS must not leak on a failed clone3

[Mark Brown]

On Wed, Oct 28, 2026 at 12:43:14PM -0300, Iván Ezequiel Rodriguez wrote:
> KTAP coverage for clone3(CLONE_VM|CLONE_PIDFD) with a bad pidfd
> pointer: sixteen failures must not grow the parent's GCS VMA count.

Thanks, this is a lot cleaner and easier to understand than the original
version.  It'd probably be nicer to do the refactoring of the existing
code in a separate commit but that's more nice to have than essential.

> +TEST_GEN_PROGS := basic-gcs libc-gcs gcs-locking gcs-stress gcspushm gcsstr \
> +                 gcs-clone-pidfd

We should update .gitignore as well (or merge the test into libc-gcs but
that's a bunch more work and definitely into bikeshed territory).

I verified that this fails with the current kernel and that the kernel
change causes it to pass so with the above:

Reviewed-by: Mark Brown <broonie at kernel.org>
Tested-by: Mark Brown <broonie at kernel.org>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20261028/b2620fbb/attachment.sig>