From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1MqX85-0005Nj-GS for qemu-devel@nongnu.org; Wed, 23 Sep 2009 15:03:53 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1MqX80-0005Lz-Tv for qemu-devel@nongnu.org; Wed, 23 Sep 2009 15:03:53 -0400 Received: from [199.232.76.173] (port=50881 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1MqX80-0005Lw-OE for qemu-devel@nongnu.org; Wed, 23 Sep 2009 15:03:48 -0400 Received: from mail-bw0-f211.google.com ([209.85.218.211]:33095) by monty-python.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1MqX80-0008Ns-85 for qemu-devel@nongnu.org; Wed, 23 Sep 2009 15:03:48 -0400 Received: by bwz7 with SMTP id 7so869627bwz.34 for ; Wed, 23 Sep 2009 12:03:46 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <20090923184041.GA18947@redhat.com> References: <20090923125818.GA13875@redhat.com> <4ABA62D0.2060205@codemonkey.ws> <20090923184041.GA18947@redhat.com> From: Blue Swirl Date: Wed, 23 Sep 2009 22:03:26 +0300 Message-ID: Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Subject: [Qemu-devel] Re: comments on: get page size in device init List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Michael S. Tsirkin" Cc: qemu-devel@nongnu.org On Wed, Sep 23, 2009 at 9:40 PM, Michael S. Tsirkin wrote: > On Wed, Sep 23, 2009 at 01:02:56PM -0500, Anthony Liguori wrote: >> Blue Swirl wrote: >>> On Wed, Sep 23, 2009 at 3:58 PM, Michael S. Tsirkin wr= ote: >>> >>>>> =C2=A0 =C2=A0 Compile msix only once >>>>> >>>>> =C2=A0 =C2=A0 Get page size in device init. >>>>> >>>>> =C2=A0 =C2=A0 Signed-off-by: Blue Swirl >>>>> >>>> What was the motivation for the page size change? >>>> >>> >>> "Compile msix only once" >>> >>> >>>> It seems the only user passes TARGET_PAGE_SIZE anyway, >>>> using a constant seems clearer and probably generates >>>> less code. No? >>>> >>> >>> Yes, but then the code would depend on TARGET_PAGE_SIZE, making it >>> impossible to compile the code only once. >>> >> >> We could probably get away with doing >> >> #define TARGET_PAGE_SIZE target_get_page_size() >> >> And take care of a big chunk of this without passing page size >> parameters around. > > > Sounds good. That would work and target_get_page_size() together with get_ram_size() would also handle the virtio case nicely, except for the if (!kvm_enabled() || kvm_has_sync_mmu()) part.