https://bugzilla.kernel.org/show_bug.cgi?id=201825 --- Comment #5 from Larkin Lowrey (llowrey@xxxxxxxxxxxxxxxxx) --- Still occurs in 4.19.5. As always, there is a call to try_async_pf. That is clearly part of the problem. Interestingly, I have not experienced the issue when running zswap. Even then there is still quite a bit of on-disk swap usage so that io path is still being used. Still, it seems that when zswap in enabled the async page fault always completes but without zswap the path sometimes does not complete. -- You are receiving this mail because: You are watching the assignee of the bug.
