From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1MsgXq-0007Hy-60 for qemu-devel@nongnu.org; Tue, 29 Sep 2009 13:31:22 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1MsgXl-0007BR-5G for qemu-devel@nongnu.org; Tue, 29 Sep 2009 13:31:21 -0400 Received: from [199.232.76.173] (port=55359 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1MsgXk-0007BE-V0 for qemu-devel@nongnu.org; Tue, 29 Sep 2009 13:31:16 -0400 Received: from mail-qy0-f173.google.com ([209.85.221.173]:40406) by monty-python.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1MsgXk-0000dZ-8T for qemu-devel@nongnu.org; Tue, 29 Sep 2009 13:31:16 -0400 Received: by qyk3 with SMTP id 3so4122364qyk.4 for ; Tue, 29 Sep 2009 10:31:15 -0700 (PDT) Message-ID: <4AC24460.1040305@codemonkey.ws> Date: Tue, 29 Sep 2009 12:31:12 -0500 From: Anthony Liguori MIME-Version: 1.0 References: <4AB7A01A.3000206@redhat.com> <20090929085450.GE25389@redhat.com> <4AC211C8.1060202@codemonkey.ws> <200909300258.17297.rusty@rustcorp.com.au> In-Reply-To: <200909300258.17297.rusty@rustcorp.com.au> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Subject: [Qemu-devel] Re: [PATCH 0/2] fix virtio_blk serial pci config breakage List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Rusty Russell Cc: john cooper , "Michael S. Tsirkin" , john cooper , qemu-devel@nongnu.org, Avi Kivity , jens.axboe@oracle.com, Vadim Rozenfeld Rusty Russell wrote: > On Tue, 29 Sep 2009 11:25:20 pm Anthony Liguori wrote: > >> 2) Passing an ATA identity page is goofy. We should just pass the >> serial number and let Linux generate the identity page. Just because >> Linux requires this as it's user space interface, that doesn't mean that >> other guests will (like Windows). Instead of exposing an opaque blob, >> we should expose the information we need in a structured way. >> > > I think John did this on my prompting, because it already existed as a > defined ABI. If someone is going to make stuff up, isn't it better that > kvm does it (some of those fields might have useful values?) than linux? > We should expose Linux ABI quirks in the virtio ABI. Down the road, someone may decide that there's a better way to expose things like S/N down to userspace and then in our drivers we would have to parse the ATA identity page in order to figure out the S/N to expose that properly through the new interface. Moreover, there's probably an OS out there that we already have to do that for. That's a pretty ugly thing to have to do. > (I don't care: my patches stick with the current scheme, but changing is > fairly easy, but needs to be decided before commit). > Everything is tremendously simplified if we just expose the S/N in the ABI since we can avoid solving the config size limitation. I'm a big fan of deferring difficult problems when there's an immediate easy solution :-) > Thanks, > Rusty, >