From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:51714) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1blQyS-0000mt-5a for qemu-devel@nongnu.org; Sat, 17 Sep 2016 21:32:53 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1blQyQ-00058H-7f for qemu-devel@nongnu.org; Sat, 17 Sep 2016 21:32:51 -0400 Date: Sun, 18 Sep 2016 09:32:28 +0800 From: Yaowei Bai Message-ID: <20160918013130.GA4148@yaowei-K42JY> References: <1473851019-7005-1-git-send-email-baiyaowei@cmss.chinamobile.com> <1473851019-7005-3-git-send-email-baiyaowei@cmss.chinamobile.com> <20160914164017.GR16438@stefanha-x1.localdomain> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20160914164017.GR16438@stefanha-x1.localdomain> Subject: Re: [Qemu-devel] [PATCH 2/2] aio: improve aio_poll performance by checking epoll only once List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Stefan Hajnoczi Cc: famz@redhat.com, jcody@redhat.com, kwolf@redhat.com, mreitz@redhat.com, qemu-block@nongnu.org, qemu-devel@nongnu.org, yjs-kernel@cmss.chinamobile.com On Wed, Sep 14, 2016 at 05:40:17PM +0100, Stefan Hajnoczi wrote: > On Wed, Sep 14, 2016 at 07:03:39AM -0400, Yaowei Bai wrote: > > As epoll whether enabled or not is a global setting, we can just > > check it only once rather than checking it with every node iteration. > > Through this we can avoid a lot of checks when epoll is not enabled. > > > > Signed-off-by: Yaowei Bai > > Reviewed-by: Xiubo Li > > --- > > aio-posix.c | 12 +++++++----- > > 1 file changed, 7 insertions(+), 5 deletions(-) > > The commit message says "improve aio_poll performance" but no benchmark > results were provided. Therefore I can't take this patch as a > performance optimization. I'm still happy to merge the patch since it > makes the if statement simpler but I'll rename it to "aio-posix: avoid > unnecessary aio_epoll_enabled() calls". Sorry for replying late, i just came back from my vacation. And i'd like to say it's ok for me to rename the commit message. Thank you for doing it. > > I don't think this patch gives a measurable performance improvement. If Yes, i agree with you at this point in consideration of there's no too many fds to poll at most time. > you believe otherwise, please post benchmark results. Please let me > know what you think. > > Stefan