[Bug rtl-optimization/26272] [4.2 Regresssion] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[Bug rtl-optimization/26272] [4.2 Regresssion] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.2 Regresssion] gcc.c-    |[4.2 Regresssion] gcc.c-
                   |torture/execute/built-in-   |torture/execute/built-in-
                   |setjmp.c fails at -O2 and   |setjmp.c fails at -O2 and
                   |above                       |above on powerpc-darwin
   Target Milestone|---                         |4.2.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272

[Bug rtl-optimization/26272] New: [4.2 Regresssion] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above

[pinskia at gcc dot gnu dot org]

[dandelion:~/src/gcc/local] pinskia% /Volumes/temp/gcc/local/bin/gcc -O2
gcc/gcc/testsuite/gcc.c-torture/execute/built-in-setjmp.c
[dandelion:~/src/gcc/local] pinskia% !./
./a.out
Bus error
[dandelion:~/src/gcc/local] pinskia% /Volumes/temp/gcc/local/bin/gcc -O2
gcc/gcc/testsuite/gcc.c-torture/execute/built-in-setjmp.c
-fno-move-loop-invariants
[dandelion:~/src/gcc/local] pinskia% !./
./a.out
[dandelion:~/src/gcc/local] pinskia% 

It is moving the reloading of the PIC register from inside the loop to outside:
(insn 45 144 41 2 (set (reg:SI 168)
        (unspec:SI [
                (symbol_ref:SI ("*LSJR4"))
            ] 15)) 16 {load_macho_picbase_si} (nil)
    (nil))
;; End of basic block 2, registers live:
 (nil)

Is that really valid (or should there be an use on pc)?


-- 
           Summary: [4.2 Regresssion] gcc.c-torture/execute/built-in-
                    setjmp.c fails at -O2 and above
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pinskia at gcc dot gnu dot org
GCC target triplet: powerpc-darwin7.9.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272

[Bug rtl-optimization/26272] [4.2 Regression] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2006-02-14 00:37 -------
I am going to test the following patch (but I would like to know if this is the
correct approach):
Index: config/rs6000/darwin.md
===================================================================
--- config/rs6000/darwin.md     (revision 110948)
+++ config/rs6000/darwin.md     (working copy)
@@ -254,8 +254,8 @@ Boston, MA 02110-1301, USA.  */

 (define_insn "load_macho_picbase_si"
   [(set (match_operand:SI 0 "register_operand" "=l")
-       (unspec:SI [(match_operand:SI 1 "immediate_operand" "s")]
-                  UNSPEC_LD_MPIC))]
+       (unspec:SI [(match_operand:SI 1 "immediate_operand" "s")
+                   (pc)] UNSPEC_LD_MPIC))]
   "(DEFAULT_ABI == ABI_DARWIN) && flag_pic"
   "bcl 20,31,%1\\n%1:"
   [(set_attr "type" "branch")
@@ -263,7 +263,8 @@ Boston, MA 02110-1301, USA.  */

 (define_insn "load_macho_picbase_di"
   [(set (match_operand:DI 0 "register_operand" "=l")
-       (unspec:DI [(match_operand:DI 1 "immediate_operand" "s")]
UNSPEC_LD_MPIC))]
+       (unspec:DI [(match_operand:DI 1 "immediate_operand" "s")
+                   (pc)] UNSPEC_LD_MPIC))]
   "(DEFAULT_ABI == ABI_DARWIN) && flag_pic && TARGET_64BIT"
   "bcl 20,31,%1\\n%1:"
   [(set_attr "type" "branch")


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272

[Bug rtl-optimization/26272] [4.2 Regression] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[rakdver at gcc dot gnu dot org]

------- Comment #2 from rakdver at gcc dot gnu dot org  2006-02-14 00:51 -------
Without further information (use of pc) loop-invariant has no way how to know
that this insn cannot be moved; so I think your patch should be correct.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272

[Bug rtl-optimization/26272] [4.2 Regression] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[pinskia at gcc dot gnu dot org]

------- Comment #3 from pinskia at gcc dot gnu dot org  2006-02-14 03:31 -------
Confirmed based on:
http://gcc.gnu.org/ml/gcc-testresults/2006-02/msg00688.html

Plus I think I have a fix so mine.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |pinskia at gcc dot gnu dot
                   |dot org                     |org
             Status|UNCONFIRMED                 |ASSIGNED
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2006-02-14 03:31:50
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272

[Bug rtl-optimization/26272] [4.2 Regression] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[pinskia at gcc dot gnu dot org]

------- Comment #4 from pinskia at gcc dot gnu dot org  2006-02-16 04:25 -------
Patch posted:
http://gcc.gnu.org/ml/gcc-patches/2006-02/msg01310.html


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2006-
                   |                            |02/msg01310.html
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272

[Bug rtl-optimization/26272] [4.2 Regression] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[patchapp at dberlin dot org]

------- Comment #5 from patchapp at dberlin dot org  2006-02-16 04:26 -------
Subject: Bug number PR target/26272

A patch for this bug has been added to the patch tracker.
The mailing list url for the patch is
http://gcc.gnu.org/ml/gcc-patches/2006-02/msg01310.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272

[Bug rtl-optimization/26272] [4.2 Regression] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[pinskia at gcc dot gnu dot org]

------- Comment #7 from pinskia at gcc dot gnu dot org  2006-02-17 19:52 -------
Fixed.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272

[Bug rtl-optimization/26272] [4.2 Regression] gcc.c-torture/execute/built-in-setjmp.c fails at -O2 and above on powerpc-darwin

[pinskia at gcc dot gnu dot org]

------- Comment #6 from pinskia at gcc dot gnu dot org  2006-02-17 19:52 -------
Subject: Bug 26272

Author: pinskia
Date: Fri Feb 17 19:52:28 2006
New Revision: 111206

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=111206
Log:
2006-02-17  Andrew Pinski  <pinskia@physics.uc.edu>

        PR target/26272
        * config/rs6000/darwin.md (load_macho_picbase_si): Add
        pc to the pattern.
        (load_macho_picbase_di): Likewise.



Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/rs6000/darwin.md


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26272