You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I don't have full proof, but looking at spikes of SYZFATAL errors I think the following can happen.
We create cgroups of the for "/syzcgroup/*/syz{{PROCID}}" and then use this cgroup for all tests for PROCID, we don't create per-test cgroups. So if a test messes with cgroup settings, e.g. restricts memory to a very low value, all subsequent tests for the same PROCID will fail. Even if executor is restarted.
We either should create per-test cgroups, or at least per loop process cgropus, so that restarting executor fixes things.
The text was updated successfully, but these errors were encountered:
I don't have full proof, but looking at spikes of SYZFATAL errors I think the following can happen.
We create cgroups of the for "/syzcgroup/*/syz{{PROCID}}" and then use this cgroup for all tests for PROCID, we don't create per-test cgroups. So if a test messes with cgroup settings, e.g. restricts memory to a very low value, all subsequent tests for the same PROCID will fail. Even if executor is restarted.
We either should create per-test cgroups, or at least per loop process cgropus, so that restarting executor fixes things.
The text was updated successfully, but these errors were encountered: