From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:57932) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YuPvV-0006D6-Jr for qemu-devel@nongnu.org; Mon, 18 May 2015 14:38:10 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YuPvQ-0001bM-HR for qemu-devel@nongnu.org; Mon, 18 May 2015 14:38:09 -0400 Received: from mail-qk0-x243.google.com ([2607:f8b0:400d:c09::243]:35992) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YuPvQ-0001bC-D3 for qemu-devel@nongnu.org; Mon, 18 May 2015 14:38:04 -0400 Received: by qkan20 with SMTP id n20so879848qka.3 for ; Mon, 18 May 2015 11:38:03 -0700 (PDT) Sender: Richard Henderson Message-ID: <555A3186.7000200@twiddle.net> Date: Mon, 18 May 2015 11:37:58 -0700 From: Richard Henderson MIME-Version: 1.0 References: <1431665391-2389-1-git-send-email-crosthwaite.peter@gmail.com> <555613BC.4070408@twiddle.net> <55562A2C.9080005@twiddle.net> In-Reply-To: Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH] microblaze: Remove uses of TCGv and target_ulong List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Peter Crosthwaite Cc: Peter Maydell , Peter Crosthwaite , Peter Crosthwaite , "qemu-devel@nongnu.org Developers" , Edgar Iglesias , =?UTF-8?B?QW5kcmVhcyBGw6RyYmVy?= On 05/15/2015 01:52 PM, Peter Crosthwaite wrote: > 2: Core code is converted to not use "target_ulong" (tcg_core_ulong?) > and target_ulong becomes a target-specific code concept. This is my favorite. I did poke around with new tcg opcodes to break some of the TARGET_LONG_BITS usage in the tcg backend. It wasn't terribly difficult until I reached the memory helpers. Aside from target_ulong addr, there's also TARGET_PAGE_MASK, TARGET_PAGE_SIZE, TARGET_PAGE_BITS, CPU_TLB_SIZE. What's our position on the compatibility of the mmu between the multi-arch platforms? I assume at the moment they must be all the same? Of course, I've mentioned before that ARM might well benefit from having these values be variable and not constant at all. Primarily because of the armv5 1K pages affecting aarch64. r~