From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1LAvsQ-0004Ot-Cl for qemu-devel@nongnu.org; Thu, 11 Dec 2008 19:27:30 -0500 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1LAvsP-0004Oh-Q7 for qemu-devel@nongnu.org; Thu, 11 Dec 2008 19:27:30 -0500 Received: from [199.232.76.173] (port=41905 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1LAvsP-0004Oe-LH for qemu-devel@nongnu.org; Thu, 11 Dec 2008 19:27:29 -0500 Received: from mx2.redhat.com ([66.187.237.31]:43852) by monty-python.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1LAvsP-0001V8-6n for qemu-devel@nongnu.org; Thu, 11 Dec 2008 19:27:29 -0500 Date: Fri, 12 Dec 2008 01:27:21 +0100 From: Andrea Arcangeli Subject: Re: [Qemu-devel] [RFC] Replace posix-aio with custom thread pool Message-ID: <20081212002721.GH6809@random.random> References: <493FFAB6.2000106@codemonkey.ws> <493FFC8E.9080802@redhat.com> <49400F69.8080707@codemonkey.ws> <20081210190810.GG18814@random.random> <20081211131222.GA14908@random.random> <494130B5.2080800@redhat.com> <20081211155335.GE14908@random.random> <49413B9C.3030703@redhat.com> <20081211164947.GD6809@random.random> <20081211213213.GB2399@lst.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20081211213213.GB2399@lst.de> Reply-To: qemu-devel@nongnu.org List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Christoph Hellwig Cc: qemu-devel@nongnu.org, kvm-devel , Gerd Hoffmann On Thu, Dec 11, 2008 at 10:32:13PM +0100, Christoph Hellwig wrote: > Mainline kernel aio only implements O_DIRECT. Some RHEL version had > support for buffered kernel AIO. Not just RHEL, I'm sure SLES supported this too and it was quite cleanly implemented, not sure why this wasn't picked up by mainline. At the light of this preadv/pwritev or having the file opened by each thread (if proven safe with nfs) remains the only option. Zerocopy dma must be supported without O_DIRECT too even if it's probably more important when O_DIRECT is on.