From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:37938) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1b7dhZ-0007Yl-8c for qemu-devel@nongnu.org; Tue, 31 May 2016 03:02:58 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1b7dhT-00057j-8I for qemu-devel@nongnu.org; Tue, 31 May 2016 03:02:56 -0400 References: <1464338922-13102-1-git-send-email-mark.cave-ayland@ilande.co.uk> <20160530211911.GA27823@aurel32.net> From: Mark Cave-Ayland Message-ID: <574D36EE.5030200@ilande.co.uk> Date: Tue, 31 May 2016 08:02:06 +0100 MIME-Version: 1.0 In-Reply-To: <20160530211911.GA27823@aurel32.net> Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 0/2] macio: switch over to new byte-aligned DMA helpers List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Aurelien Jarno Cc: pbonzini@redhat.com, qemu-ppc@nongnu.org, qemu-devel@nongnu.org On 30/05/16 22:19, Aurelien Jarno wrote: > On 2016-05-27 09:48, Mark Cave-Ayland wrote: >> Here is a tidied up version of my patch to convert the macio controller over to >> using the new byte-aligned DMA helpers. >> >> The first patch is just a hack and temporarily disables unaligned iovec >> truncation in the DMA helper (as discussed in the recent thread) until Paolo or >> someone else can devise a proper solution. Without this, the subsequent switch >> over to the DMA helpers will appear to work during a Darwin PPC install but the >> resulting image is corrupt and will fail to boot. >> >> The second patch is the real one and switches the macio controller over to use >> the new byte-aligned DMA helpers. Here I see a speed-up of around 2.5x-3x for >> a typical Darwin PPC installation compared to the previous code. >> >> Aurelien, I'd be grateful if you could test the TRIM path as I know this is >> something you've had issues with before and I couldn't quite figure out how to >> reproduce your TRIM tests from before. > > I have just tested the TRIM path, all works fine with your 2 patches > applied. Thanks a lot for testing this, Aurelien. In that case I'll resubmit patch 2 officially once Paolo has figured out the correct solution for the iovec truncation. ATB, Mark.