I made a fix to ensure that QBox patch doesn’t break the replay. Now I tried it on couple of tests and didn’t see any regression.
Pavel Dovgalyuk From: Artem Pisarenko [mailto:artem.k.pisare...@gmail.com] Sent: Wednesday, October 17, 2018 4:32 PM To: Pavel Dovgalyuk Cc: Paolo Bonzini; Clement Deschamps; qemu-devel@nongnu.org; Pavel Dovgalyuk Subject: Re: [PATCH] Revert "icount: remove obsolete warp call" > I found the source of the bug. As QEMU becomes more multi-threaded and non-> > synchronized, > checkpoints move from thread to thread. > And the event queue that processed at checkpoints should belong to the same > thread > in both record and replay executions. > > Current problem was with the checkpoint for virtual timers. They are > processed from different threads: > from vCPU and from aio_dispatch function. What bug you're talking about ? This patch thread started with intention to fix generic bug with icount, which is being encountered in non-rr mode. As far as I understood your words (and patch content) it will not fix that generic bug. Anyway, it's a good news! -- С уважением, Артем Писаренко
