From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:36239) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Qw9bL-0001SA-3T for qemu-devel@nongnu.org; Wed, 24 Aug 2011 05:18:23 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Qw9bJ-0001S2-GY for qemu-devel@nongnu.org; Wed, 24 Aug 2011 05:18:22 -0400 Received: from mx1.redhat.com ([209.132.183.28]:11891) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Qw9bJ-0001Rp-9E for qemu-devel@nongnu.org; Wed, 24 Aug 2011 05:18:21 -0400 Message-ID: <4E54C1D2.8080102@redhat.com> Date: Wed, 24 Aug 2011 12:18:10 +0300 From: Avi Kivity MIME-Version: 1.0 References: <1314144835-29098-1-git-send-email-rth@twiddle.net> In-Reply-To: <1314144835-29098-1-git-send-email-rth@twiddle.net> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 00/16] isa_register_portio_list, v2 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Richard Henderson Cc: qemu-devel@nongnu.org On 08/24/2011 03:13 AM, Richard Henderson wrote: > The problem that malc saw with sb16 was a major think-o on my part > with the whole interface. We can't re-use the const sub-arrays of > the original MemoryRegionPortio array because they have the wrong > offset for the MemoryRegion to which it is attached -- the lookup > in find_portio fails. We must adjust the offsets of the old_portio > array to be based against the MemoryRegion. > > Which means we can easily eliminate the major complaint that came > with the previous round of comments -- the double PORTIO_END_OF_LIST > and the explicit marking of the ranges. All we require of users > is that the array be sorted by offset. > > The entire patch set is at > > git://repo.or.cz/qemu/rth.git mem-api-isa > > and is of course based on > > git://git.kernel.org/pub/scm/virt/kvm/qemu-kvm.git memory/master > > Avi, the first two patches are fixes for compile errors in your > tree. They probably ought to be squashed when next you rebase. Sloppy of me... The patchset looks good, I'll leave it on the list for a few days before pulling to allow further review. Once we're done with the conversion, we should look at extending the idea to mmio registers - decode the registers in the core instead of a per-region switch statement. -- I have a truly marvellous patch that fixes the bug which this signature is too narrow to contain.