From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43)
	id 1KQT1s-0000Vg-QB
	for qemu-devel@nongnu.org; Tue, 05 Aug 2008 16:21:12 -0400
Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43)
	id 1KQT1r-0000TD-4e
	for qemu-devel@nongnu.org; Tue, 05 Aug 2008 16:21:12 -0400
Received: from [199.232.76.173] (port=51399 helo=monty-python.gnu.org)
	by lists.gnu.org with esmtp (Exim 4.43) id 1KQT1r-0000Sv-0h
	for qemu-devel@nongnu.org; Tue, 05 Aug 2008 16:21:11 -0400
Received: from csl.cornell.edu ([128.84.224.10]:3795 helo=vlsi.csl.cornell.edu)
	by monty-python.gnu.org with esmtp (Exim 4.60)
	(envelope-from <vince@stanley.csl.cornell.edu>) id 1KQT1q-0008NT-JK
	for qemu-devel@nongnu.org; Tue, 05 Aug 2008 16:21:10 -0400
Date: Tue, 5 Aug 2008 16:20:58 -0400 (EDT)
From: Vince Weaver <vince@csl.cornell.edu>
In-Reply-To: <f43fc5580807291018n7f47d010l97276f704b5afb52@mail.gmail.com>
Message-ID: <20080805161615.S57426@stanley.csl.cornell.edu>
References: <20080728225136.C26546@stanley.csl.cornell.edu>
	<f43fc5580807291018n7f47d010l97276f704b5afb52@mail.gmail.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
Subject: [Qemu-devel] another SPARC issue
Reply-To: qemu-devel@nongnu.org
List-Id: qemu-devel.nongnu.org
List-Unsubscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.gnu.org/pipermail/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: qemu-devel@nongnu.org
Cc: Blue Swirl <blauwirbel@gmail.com>

Hello

so I'm still trying to get bzip to work under sparc32plus-linux-user

This time the problem is with the "ldda" instruction, which is loading off
of the block-read %asi (I am learning way more than I want to about 
obscure parts of the SPARC architecture).

The code involved is doing an optimized memory copy, so it's using the 
floating point unit to accelerate this.

The instruction is
    ldda  [ %o1 + 0x40 ] %asi, %f16

but under Qemu for some reason 0x80 is added to %o1 (instead of 0x40).

Indeed if you look at the generated TCG (see below), the 0x40 is being 
added in twice.  I was digging around in translate.c but can't seem to see 
why this is happning.

Any help would be appreciated...

Thanks

Vince


IN: __memcpy_large
0x000000000002b73c:  ldda  [ %o1 + 0x40 ] %asi, %f16

OP after la:
  ld_i32 tmp11,regwptr,$0x8
  ld_i32 tmp12,regwptr,$0xc
  movi_i32 tmp15,$0x40
  movi_i32 tmp16,$0x0
  add2_i32 loc9,loc10,tmp11,tmp12,tmp15,tmp16
  movi_i32 pc_0,$0x2b73c
  movi_i32 pc_1,$0x0
  movi_i32 npc_0,$0x2b740
  movi_i32 npc_1,$0x0
  movi_i32 tmp15,$0x40
  movi_i32 tmp16,$0x0
  add2_i32 loc9,loc10,loc9,loc10,tmp15,tmp16
  ld_i32 tmp17,env,$0xd9c8
  movi_i32 tmp18,$0x8
  movi_i32 tmp19,$0x10
  movi_i32 tmp20,$helper_ldf_asi
  call tmp20,$0x0,$0,loc9,loc10,tmp17,tmp18,tmp19
  movi_i32 pc_0,$0x2b740
  movi_i32 pc_1,$0x0
  movi_i32 npc_0,$0x2b744
  movi_i32 npc_1,$0x0
  movi_i32 tmp17,$helper_debug
  call tmp17,$0x0,$0
  exit_tb $0x0
  end