Il 05/04/2013 14:54, Anthony Liguori ha scritto: > I guess this works with migration because we assume that after migration > the main loop will do a complete run?
Yes, migration will terminate in an fd handler, and the next round of the main loop will re-evaluate chr_read. > Is this a safe assumption or does > there need to be a qemu_notify_event() somewhere after migration to make > sure this doesn't cause a hang? There could be a qemu_chr_accept_input() for all character devices after migration. I think that would be a separate patch. Regarding the need or not for Peter's patch: the patch might be needed this kind of busy-wait fix was required often. As far as I recall, this is the first we ever had, and it came after an almost-complete rewrite. It seems rare enough, that it's much better to fix the root causes when they appear---not the symptoms. Paolo
