Re: Revision 144098 (d.d. Wed Feb 11 08:56:41 2009 UTC (4 weeks ago)) is not a regression bug fix.

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Thu, Mar 12, 2009 at 8:00 AM, Richard Guenther
<richard.guenther@gmail.com> wrote:
> On Thu, Mar 12, 2009 at 3:56 PM, Steven Bosscher <stevenb.gcc@gmail.com> wrote:
>> - Show quoted text -
>> On Thu, Mar 12, 2009 at 3:34 PM, Paolo Bonzini <bonzini@gnu.org> wrote:
>>>
>>>> Will "REGNO (operands[0]) == REGNO (operands[1])" work here?
>>>
>>> Yes. ?I wanted to be conservative in case one day subregs or who knows
>>> what are allowed. ?I'll defer to maintainers or other people (Steven?),
>>> either way is fine by me.
>>
>> If REGNO works now, I'd use it. ?If "one day" subregs are allowed,
>> REGNO will trigger an rtl-checking failure anyway.
>
> Indeed. ?The patch with REGNO is pre-approved if it bootstraps/regtests.
>

I tried this

Index: i386.md
===================================================================
--- i386.md	(revision 144796)
+++ i386.md	(working copy)
@@ -20820,7 +20820,8 @@
    (parallel [(set (match_dup 0)
                    (match_op_dup 3 [(match_dup 0) (match_dup 1)]))
               (clobber (reg:CC FLAGS_REG))])]
-  "operands[4] = replace_rtx (operands[2], operands[0], operands[1]);")
+  "if (REGNO (operands[0]) == REGNO (operands[1])) abort ();
+  operands[4] = replace_rtx (operands[2], operands[0], operands[1]);")

 (define_peephole2
   [(set (match_operand 0 "register_operand" "")
@@ -20835,7 +20836,7 @@
   [(set (match_dup 0) (match_dup 2))
    (set (match_dup 0)
         (match_op_dup 3 [(match_dup 0) (match_dup 1)]))]
-  "")
+  "if (REGNO (operands[0]) == REGNO (operands[1])) abort ();")

 ; Don't do logical operations with memory outputs

on Linux/Intel64.  There are no regressions:

http://gcc.gnu.org/ml/gcc-testresults/2009-03/msg01199.html

But I did get

[hjl@gnu-33 foo]$ cat /tmp/x.f90
subroutine statin(x,y,pstratr,pconvecr,zhxy,zhxhy,ztmp)
integer :: x,y
real pstratr(x,y),pconvecr(x,y),zhxy(x,y)
real ztmp(4)
do j = 1,y
 do i = 1,x-2
  zttotrainr = zttotrainr + (pstratr(i,j) + pconvecr(i,j))*zhxy(i,j)
  ztstratr   = ztstratr   + pstratr(i,j)
  ztconvecr  = ztconvecr  + pconvecr(i,j)
  ztsenf     = ztsenf     + zhxy(i,j)
  ztlatf     = ztlatf     + zhxy(i,j)
  ztcldtop   = ztcldtop   + zhxy(i,j)
 enddo
enddo
ztmp(1)=zttotrainr
ztmp(2)=ztstratr
ztmp(3)=ztconvecr
ztmp(4)=ztsenf*ztlatf*ztcldtop
end
[hjl@gnu-33 foo]$ ../xgcc -B../ -S -O3 -ffast-math    /tmp/x.f90
/tmp/x.f90: In function ‘statin’:
/tmp/x.f90:18: internal compiler error: in gen_peephole2_2750, at
config/i386/i386.md:20839
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

I am checking:

Index: i386.md
===================================================================
--- i386.md	(revision 144796)
+++ i386.md	(working copy)
@@ -20813,7 +20813,7 @@
                      [(match_dup 0)
                       (match_operand:SI 2 "memory_operand" "")]))
               (clobber (reg:CC FLAGS_REG))])]
-  "operands[0] != operands[1]
+  "REGNO (operands[0]) != REGNO (operands[1])
    && GENERAL_REGNO_P (REGNO (operands[0]))
    && GENERAL_REGNO_P (REGNO (operands[1]))"
   [(set (match_dup 0) (match_dup 4))
@@ -20829,7 +20829,7 @@
                    (match_operator 3 "commutative_operator"
                      [(match_dup 0)
                       (match_operand 2 "memory_operand" "")]))]
-  "operands[0] != operands[1]
+  "REGNO (operands[0]) != REGNO (operands[1])
    && ((MMX_REG_P (operands[0]) && MMX_REG_P (operands[1]))
        || (SSE_REG_P (operands[0]) && SSE_REG_P (operands[1])))"
   [(set (match_dup 0) (match_dup 2))

now.

Thanks.

-- 
H.J.