>>> On 09.11.17 at 11:36, <raist...@linux.it> wrote: > Well, I'm afraid I only see two solutions: > 1) we get rid of lazy context switch; > 2) whatever it is that is happening at point c above, it needs to be > aware that we use lazy context switch, and make sure to sync the > context before playing with or altering it;
3) Better centralize the updating of v->processor, so that it becomes reasonable to sync state there. Igor's idea of flushing state once it is known (or at least pretty certain) that the vCPU won't run on the prior pCPU next time it gets scheduled is certainly a reasonable one. It just doesn't fit well with how the individual schedulers currently behave. Jan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org https://lists.xen.org/xen-devel
