On 2016-09-16 11:09, Dietmar Eggemann wrote:
I caught a WALT related hard lockup on a v4.7 kernel with only patch 1 on top. Fairly easy to reproduce by watching a video in firefox browser on Ubuntu 16.04.
$ addr2line -e vmlinux ffffffff810d835e
Thanks a lot for testing and reporting this Dietmar. This is because we need to take both rq locks during migration, and currently the code path doesn't consider that one of the locks may already be taken in some paths. Earlier bug-free versions of code had the double_lock/unlock around set_task_cpu as needed, but since that seemed expensive (bigger critical sections) we moved them into the walt_fixup_busy_time. We'll figure this one out soon.
Thanks, Vikram