Hello Tejun, Thanks for the review. All five points are addressed in v4: https://lore.kernel.org/r/[email protected]
The test is now test_pids_events_notify() in test_pids.c, runs on both mount types, drops the skip and the duplicate counter check, and cites 1765a153d985. The v1 to v2 note moved to the changelog. I tested it in QEMU on the cgroup for-7.4 branch. It passes on a default mount and on a pids_localevents mount, and it fails on a pids_localevents mount with 1765a153d985 reverted. Separately, the existing test_pids_events fails on a pids_localevents mount, with and without this patch. I have not looked into it further.
