Re: New optimization for reload_combine

[IainS <idsandoe@googlemail.com>]

On 17 Jul 2010, at 18:17, Bernd Schmidt wrote:

> On 07/17/2010 06:14 PM, H.J. Lu wrote:
>> On Sat, Jul 17, 2010 at 8:59 AM, Bernd Schmidt <bernds@codesourcery.com 
>> > wrote:
>>> On 07/17/2010 05:03 PM, H.J. Lu wrote:
>>>> It is more than that. It failed to boostrap on Linux/ia32 when
>>>> configured with
>>>>
>>>> --enable-clocale=gnu --with-system-zlib --enable-shared
>>>> --with-demangler-in-ld  --with-fpmath=sse
>>>
>>> I can't seem to reproduce this.  Is that the full command line?
>>>
>>>
>>
>> I used:
>>
>> ../src-trunk/configure \
>> 		 --enable-clocale=gnu --with-system-zlib --enable-shared --with-
>> demangler-in-ld -with-plugin-ld=ld.gold --enable-gold --with- 
>> fpmath=sse
>>
>> on Fedora 12/ia32.
>
> I'm on Gentoo, without gold - not sure whether that made a difference,
> but I'm not seeing these failures.  I don't have access to SPEC2k6
> either.  Can you isolate any testcases?

I don't have gold either - on darwin.

on i686-apple-darwin:
recog.o  and
reg-stack.o fail stage2/3 compare

here is recog.o (stage2 vs stage3)
stripped binaries => otool -tv ( basically turn the text section into  
disassembled).

It doesn't look like a debug-related diff to me - but more like what  
you posted earlier.
cheers
Iain


--- r2-code.s   2010-07-17 18:25:11.000000000 +0100
+++ r3-code.s   2010-07-17 18:25:19.000000000 +0100
@@ -1,4 +1,4 @@
-r2.o.stripped:
+r3.o.stripped:
  (__TEXT,__text) section
  _memory_address_addr_space_p:
  00000000       pushl   %ebx
@@ -6400,8 +6400,8 @@
  0000596f       leal    0x40(%esp),%edi
  00005973       nopw    0x00(%eax,%eax)
  00005979       nopl    0x00000000(%eax)
-00005980       leal    0xfc(%esi),%eax
-00005983       addl    $0x01,%esi
+00005980       addl    $0x01,%esi
+00005983       leal    0xfb(%esi),%eax
  00005986       cmpl    $0x05,%esi
  00005989       cmovgel %eax,%esi
  0000598c       movl    0x00(%ebp,%esi,8),%eax