In a Noble LXD container, the test setup doesn't work.
In a Noble LXD VM, three tests fail: tracefs_iterate_snapshot_events, , and uprobes.
The first and second tests fail because they try to sched_setaffinity() across 8 cores which the VM host has but the VM itself has only one core exposed and this API returns EINVAL. I'll have a very quick look to see if the CPU detection logic can be changed. Otherwise we won't ever be able to rely on these tests.
The third also fails on my Mantic machine. I cannot create the directory mentioned by Paul anywhere it seems.
I carried this forward.
In a Noble LXD container, the test setup doesn't work.
In a Noble LXD VM, three tests fail: tracefs_ iterate_ snapshot_ events, , and uprobes.
The first and second tests fail because they try to sched_setaffinity() across 8 cores which the VM host has but the VM itself has only one core exposed and this API returns EINVAL. I'll have a very quick look to see if the CPU detection logic can be changed. Otherwise we won't ever be able to rely on these tests.
The third also fails on my Mantic machine. I cannot create the directory mentioned by Paul anywhere it seems.