[Bug c/36594] New: [4.4 Regression] multiple regressions on powerpc at rev.136976

[Bug c/36594] New: [4.4 Regression] multiple regressions on powerpc at rev.136976

[dominiq at lps dot ens dot fr]

Multiple regressions have appeared at revision 136976 (last known working
136953), see:

http://gcc.gnu.org/ml/gcc-testresults/2008-06/msg01644.html
http://gcc.gnu.org/ml/gcc-testresults/2008-06/msg01650.html

A typical failure is:

[karma] f90/bug% /opt/gcc/gcc4.4w/bin/gcc
/opt/gcc/_gcc_clean/gcc/testsuite/gcc.c-torture/execute/nestfunc-5.c
/opt/gcc/_gcc_clean/gcc/testsuite/gcc.c-torture/execute/nestfunc-5.c: In
function 'do_goto':
/opt/gcc/_gcc_clean/gcc/testsuite/gcc.c-torture/execute/nestfunc-5.c:11:
internal compiler error: Bus error


-- 
           Summary: [4.4 Regression] multiple regressions on powerpc at
                    rev.136976
           Product: gcc
           Version: 4.4.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: dominiq at lps dot ens dot fr
GCC target triplet: powerpc-*


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36594

[Bug c/36594] [4.4 Regression] multiple regressions on powerpc at rev.136976

[dominiq at lps dot ens dot fr]

------- Comment #1 from dominiq at lps dot ens dot fr  2008-06-22 11:18 -------
If I revert revision 136959, the test in comment#0 passes.


-- 

dominiq at lps dot ens dot fr changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ubizjak at gmail dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36594

[Bug c/36594] [4.4 Regression] multiple regressions on powerpc at rev.136976

[ubizjak at gmail dot com]

------- Comment #2 from ubizjak at gmail dot com  2008-06-22 12:00 -------
Author: hubicka
Date: Thu Jun 19 18:00:12 2008
New Revision: 136959

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=136959
Log:
        * builtins.c (expand_builtin_nonlocal_goto): Stabilize r_sp before
        clobbering framepointer.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/builtins.c


-- 

ubizjak at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|ubizjak at gmail dot com    |hubicka at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36594

[Bug middle-end/36594] [4.4 Regression] multiple regressions on powerpc at rev.136976

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |middle-end
           Keywords|                            |ice-on-valid-code
   Target Milestone|---                         |4.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36594

[Bug middle-end/36594] [4.4 Regression] multiple regressions on powerpc at rev.136976

[dje at gcc dot gnu dot org]

------- Comment #3 from dje at gcc dot gnu dot org  2008-06-22 23:42 -------
Yes, similarly multiple regressions on AIX.  Reverting Honza's patch fixes it.

The failure on AIX is:

gcc/testsuite/gcc.c-torture/compile/20050122-2.c:11: internal 
compiler error: in copy_rtx, at rtl.c:314

rs6000.md restore_stack_nonlocal now receives a REG as operand[1] due to the
copy_to_reg change.  The pattern calls adjust_address_nv on the operand,
adjust_address_1 blindly dereferences the operand it thinks is a memref to
extract the address and applies copy_rtx() to the address.  copy_rtx receives
RTL with code UNKNOWN.  Fun and hilarity ensues.


-- 

dje at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dje at gcc dot gnu dot org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
           Priority|P3                          |P2
   Last reconfirmed|0000-00-00 00:00:00         |2008-06-22 23:42:26
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36594

[Bug middle-end/36594] [4.4 Regression] multiple regressions on powerpc at rev.136976

[pinskia at gcc dot gnu dot org]

------- Comment #4 from pinskia at gcc dot gnu dot org  2008-06-23 22:12 -------
This is the patch which I am testing (which fixes the original issue Honza ran
into and this one too):
Index: builtins.c
===================================================================
--- builtins.c  (revision 137049)
+++ builtins.c  (working copy)
@@ -873,6 +873,9 @@ expand_builtin_nonlocal_goto (tree exp)
   r_label = convert_memory_address (Pmode, r_label);
   r_save_area = expand_normal (t_save_area);
   r_save_area = convert_memory_address (Pmode, r_save_area);
+  /* Copy the address of the save location to a register just in case it was
based
+    on the frame pointer.   */
+  r_save_area = copy_to_reg (r_save_area);
   r_fp = gen_rtx_MEM (Pmode, r_save_area);
   r_sp = gen_rtx_MEM (STACK_SAVEAREA_MODE (SAVE_NONLOCAL),
                      plus_constant (r_save_area, GET_MODE_SIZE (Pmode)));
@@ -887,7 +890,6 @@ expand_builtin_nonlocal_goto (tree exp)
 #endif
     {
       r_label = copy_to_reg (r_label);
-      r_sp = copy_to_reg (r_sp);

       emit_clobber (gen_rtx_MEM (BLKmode, gen_rtx_SCRATCH (VOIDmode)));
       emit_clobber (gen_rtx_MEM (BLKmode, hard_frame_pointer_rtx));


-- 

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|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36594

[Bug middle-end/36594] [4.4 Regression] multiple regressions on powerpc at rev.136976

[pinskia at gcc dot gnu dot org]

------- Comment #5 from pinskia at gcc dot gnu dot org  2008-06-24 22:42 -------
Fixed.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36594

[Bug middle-end/36594] [4.4 Regression] multiple regressions on powerpc at rev.136976

[pinskia at gcc dot gnu dot org]

------- Comment #6 from pinskia at gcc dot gnu dot org  2008-06-24 22:42 -------
Subject: Bug 36594

Author: pinskia
Date: Tue Jun 24 22:42:10 2008
New Revision: 137089

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=137089
Log:
2008-06-24  Andrew Pinski  <andrew_pinski@playstation.sony.com>

        PR middle-end/36594
        * builtins.c (expand_builtin_nonlocal_goto): Stabilize the address of
        the memory instead of the memory itself for the save area.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/builtins.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36594

[Bug middle-end/36594] [4.4 Regression] multiple regressions on powerpc at rev.136976

[pinskia at gcc dot gnu dot org]

------- Comment #7 from pinskia at gcc dot gnu dot org  2008-06-24 22:43 -------
Fixed I said.


-- 

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=36594