From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:53933)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <tommusta@gmail.com>) id 1WkKx7-0006Sr-No
	for qemu-devel@nongnu.org; Tue, 13 May 2014 18:13:43 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <tommusta@gmail.com>) id 1WkKx1-0007h1-CY
	for qemu-devel@nongnu.org; Tue, 13 May 2014 18:13:37 -0400
Received: from mail-yh0-x236.google.com ([2607:f8b0:4002:c01::236]:36168)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <tommusta@gmail.com>) id 1WkKx1-0007gj-8L
	for qemu-devel@nongnu.org; Tue, 13 May 2014 18:13:31 -0400
Received: by mail-yh0-f54.google.com with SMTP id i57so964900yha.27
	for <qemu-devel@nongnu.org>; Tue, 13 May 2014 15:13:30 -0700 (PDT)
Message-ID: <53729904.8000601@gmail.com>
Date: Tue, 13 May 2014 17:13:24 -0500
From: Tom Musta <tommusta@gmail.com>
MIME-Version: 1.0
References: <1400018359-6121-1-git-send-email-rth@twiddle.net>
In-Reply-To: <1400018359-6121-1-git-send-email-rth@twiddle.net>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [PATCH] tcg: Fix tcg_reg_alloc_mov vs no-op
	truncation
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Richard Henderson <rth@twiddle.net>, qemu-devel@nongnu.org
Cc: peter.maydell@linaro.org, thuth@linux.vnet.ibm.com

On 5/13/2014 4:59 PM, Richard Henderson wrote:
> Commit af3cbfbe8018ccc16fb3a0048e928f66f0d05e87 hoisted some "common"
> loads of the temporary type, forgetting that the types could differ
> during truncating moves.  This affects the correctness of the memory
> offset on big-endian hosts.
> 
> Signed-off-by: Richard Henderson <rth@twiddle.net>

Tested-by: Tom Musta <tommusta@gmail.com>

Works functionally and also the low level traces look better:

---

IN:
0x00000fff79d62ae4:  rlwimi  r4,r4,8,16,23

OP:
 ld_i32 tmp0,env,$0xfffffffffffffffc
 movi_i32 tmp1,$0x0
 brcond_i32 tmp0,tmp1,ne,$0x0
 ---- 0xfff79d62ae4
 mov_i32 tmp0,r4
 movi_i32 tmp1,$0x8
 rotl_i32 tmp0,tmp0,tmp1
 movi_i64 tmp3,$0xffffffff
 and_i64 tmp2,tmp0,tmp3
 movi_i64 tmp4,$0xff00
 and_i64 tmp2,tmp2,tmp4
 movi_i64 tmp4,$0xffffffffffff00ff
 and_i64 tmp3,r4,tmp4
 or_i64 r4,tmp2,tmp3
 goto_tb $0x0
 movi_i64 nip,$0xfff79d62ae8
 exit_tb $0xfff79e7dc30
 set_label $0x0
 exit_tb $0xfff79e7dc33

OUT: [size=136]
0x60363920:  lwz     r14,-4(r27)
0x60363924:  cmpwi   cr7,r14,0
0x60363928:  bne-    cr7,0x60363994
0x6036392c:  ld      r14,32(r27)  # TEM -- looks better
0x60363930:  mr      r15,r14
0x60363934:  rotlwi  r15,r15,8
0x60363938:  andi.   r15,r15,65280
0x6036393c:  lis     r0,-1
0x60363940:  ori     r0,r0,255
0x60363944:  and     r14,r14,r0
0x60363948:  or      r14,r15,r14
0x6036394c:  std     r14,32(r27)
0x60363950:  .long 0x0
0x60363954:  .long 0x0
0x60363958:  .long 0x0
...
NIP 00000fff79d62ae8   LR 00000fff79d55af0 CTR 0000000000000007 XER 0000000000000000
MSR 8000000002006000 HID0 0000000000000000  HF 0000000002006000 idx 0
TB 00000000 00000000
GPR00 00000000100002d9 0000004000a0e040 00000fff79d88ae8 00000000100002d9
GPR04 0000000000002424 0000000000000000 0000000000000000 0000000000000000        # TEM so does GPR04
GPR08 0000000000000000 0000000010010978 000000007fffffff 0000000000000001