From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:56581) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1XuPNv-0006hm-4H for qemu-devel@nongnu.org; Fri, 28 Nov 2014 12:31:18 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1XuPNn-0007vN-II for qemu-devel@nongnu.org; Fri, 28 Nov 2014 12:31:10 -0500 Received: from plane.gmane.org ([80.91.229.3]:38041) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1XuPNn-0007v9-B8 for qemu-devel@nongnu.org; Fri, 28 Nov 2014 12:31:03 -0500 Received: from list by plane.gmane.org with local (Exim 4.69) (envelope-from ) id 1XuPNm-0000vq-4o for qemu-devel@nongnu.org; Fri, 28 Nov 2014 18:31:02 +0100 Received: from net-93-146-133-240.cust.vodafonedsl.it ([93.146.133.240]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Fri, 28 Nov 2014 18:31:02 +0100 Received: from pbonzini by net-93-146-133-240.cust.vodafonedsl.it with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Fri, 28 Nov 2014 18:31:02 +0100 From: Paolo Bonzini Date: Fri, 28 Nov 2014 18:30:50 +0100 Message-ID: <5478B14A.1060002@redhat.com> References: <1417183941-26329-1-git-send-email-pbonzini@redhat.com> <1417183941-26329-6-git-send-email-pbonzini@redhat.com> <20141128164023.GH4035@noname.redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit In-Reply-To: <20141128164023.GH4035@noname.redhat.com> Subject: Re: [Qemu-devel] [PATCH 5/7] coroutine: rewrite pool to avoid mutex List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: ming.lei@canonical.com, pl@kamp.de, stefanha@redhat.com On 28/11/2014 17:40, Kevin Wolf wrote: >> > I still believe we will end with some kind of coroutine bypass scheme >> > (even coroutines _do_ allocate an AIOCB, so calling bdrv_aio_readv >> > directly can help), but hey it cannot hurt to optimize hot code. > > Not sure if speculations about the future belong into commit messages, > but while it may turn out that a bypass is required in the end (I hope > it doesn't), the part about AIOCBs is wrong if you really consistently > use coroutines all the way down from the device to the block driver. This is much harder from virtio-scsi than from virtio-blk, though. Paolo