[Bug rtl-optimization/51856] New: [4.7 Regression] ICE in reload_cse_simplify_operands

[Bug rtl-optimization/51856] New: [4.7 Regression] ICE in reload_cse_simplify_operands

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51856

             Bug #: 51856
           Summary: [4.7 Regression] ICE in reload_cse_simplify_operands
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: jakub@gcc.gnu.org
            Target: s390-linux


struct B { int b1; long long b2, b3; int b4; };
struct C { char c1[40], c2, c3[96]; long long c4[5], c5; char c6[596]; };
void fn1 (long long), fn2 (char *, int), fn4 (void);
int r, fn3 (int, const char *, int, char *, int, int);

void
foo (int t, int u, int v, int w, int x, int y, struct B *z)
{
  char c[512], d[512], e;
  struct C g;
  long long f, h[255];
  struct B j;
  __builtin_bzero (&j, sizeof j);
  if (y > w)
    fn4 ();
  __builtin_bzero (&g, sizeof g);
  g.c5 = h[0];
  fn1 (z ? z->b3 : f);
  g.c2 = y;
  fn2 (d, 256);
  if (fn3 (r, "", e, c, 0, 16))
    fn4 ();
}

ICEs with -O2 -m31 on the trunk with:
rh773565.i: In function 'foo':
rh773565.i:23:1: error: insn does not satisfy its constraints:
(insn 51 136 120 8 (set (mem/s/c:QI (plus:SI (reg/f:SI 15 %r15)
                (const_int 1208 [0x4b8])) [0 g.c2+0 S1 A64])
        (mem/c/i:QI (plus:SI (plus:SI (reg/f:SI 15 %r15)
                    (reg:SI 1 %r1))
                (const_int 3 [0x3])) [5 y+3 S1 A32])) rh773565.i:19 73 {*movqi}
     (nil))
rh773565.i:23:1: internal compiler error: in reload_cse_simplify_operands, at
postreload.c:403
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

[Bug rtl-optimization/51856] [4.7 Regression] ICE in reload_cse_simplify_operands

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51856

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |krebbel at gcc dot gnu.org,
                   |                            |uweigand at gcc dot gnu.org
   Target Milestone|---                         |4.7.0

[Bug rtl-optimization/51856] [4.7 Regression] ICE in reload_cse_simplify_operands

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51856

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P4
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-01-19
     Ever Confirmed|0                           |1

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-01-19 12:54:30 UTC ---
Confirmed with a cross from x86_64-linux.

[Bug rtl-optimization/51856] [4.7 Regression] ICE in reload_cse_simplify_operands

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51856

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-01-19 12:59:44 UTC ---
Why P4?  This is a secondary target, failure to build mysql.

[Bug rtl-optimization/51856] [4.7 Regression] ICE in reload_cse_simplify_operands

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51856

--- Comment #3 from Andreas Krebbel <krebbel at gcc dot gnu.org> 2012-01-19 14:38:46 UTC ---
Created attachment 26379
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=26379
Fix

I've introduced that bug with:
http://gcc.gnu.org/ml/gcc-patches/2011-03/msg01527.html

So it's fair to hit s390 first :(

In find_reloads_subreg_address the -1 value is not properly propagated from
find_reloads_address invocation so find_reloads wrongly assumes that the
address has been fully reloaded into a register.

[Bug rtl-optimization/51856] [4.7 Regression] ICE in reload_cse_simplify_operands

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51856

--- Comment #4 from Andreas Krebbel <krebbel at gcc dot gnu.org> 2012-01-20 16:29:11 UTC ---
Author: krebbel
Date: Fri Jan 20 16:29:01 2012
New Revision: 183341

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=183341
Log:
2012-01-20  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>

    PR rtl-optimization/51856
    * reload.c (find_reloads_subreg_address): Set the address_reloaded
    flag to reloaded.

2012-01-20  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>

    * gcc.c-torture/compile/pr51856.c: New testcase.


Added:
    trunk/gcc/testsuite/gcc.c-torture/compile/pr51856.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/reload.c
    trunk/gcc/testsuite/ChangeLog

[Bug rtl-optimization/51856] [4.7 Regression] ICE in reload_cse_simplify_operands

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51856

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-01-24 07:44:36 UTC ---
Fixed.