> On Jun 20, 2016, at 13:27, Paul Mackerras <pau...@ozlabs.org> wrote: > > On Mon, May 16, 2016 at 03:02:13PM +0800, Li Zhong wrote: >> It seems that we don't need to take the lock before evaluating irq's >> resend flag. What needed is to make sure when we clear the ics's bit >> in the icp's resend_map, we don't miss the resend flag of the irqs >> that set the bit. >> >> And seems this could be ordered through the barrier in test_and_clear_bit(), >> and an newly added wmb when setting irq's resend flag, and icp's resend_map. > > This looks fine to me. Is there a measurable performance improvement > from this? I understand it could be hard to measure. > > Also, you could make the patch description more definite - just say > that we don't need to take the lock, there's no need for "seems”.
OK :) However, we may need to ignore this one for now. To implement the P/Q stuff, we probably need make sure the resend irqs to be resent only once. It’s easier to make sure that with the lock here, and the resend flag can be cleared inside the lock. Thanks, Zhong > > Paul. > _______________________________________________ > Linuxppc-dev mailing list > Linuxppc-dev@lists.ozlabs.org > https://lists.ozlabs.org/listinfo/linuxppc-dev _______________________________________________ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
