[Bug target/49206] New: RA failure in spill_failure, at reload1.c:2113

[Bug target/49206] New: RA failure in spill_failure, at reload1.c:2113

[arthur.j.odwyer at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

           Summary: RA failure in spill_failure, at reload1.c:2113
           Product: gcc
           Version: 4.5.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: arthur.j.odwyer@gmail.com


Created attachment 24383
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=24383
Output of "ajo-gcc -O2 -w -c test.c -v"

This register-allocator failure reproduces for me with svn revision 173843
(2011-05-17), and also with gcc 4.5.1. It doesn't reproduce with gcc 4.4.5. I'm
on Ubuntu 10.10, x86-64.

cat >test.c <<EOF
struct S1 { short f0; };
extern volatile struct S1 g_5;
extern int g_120, i;
extern short g_339;

int func_72(int x, int y, struct S1 z) {
    for (z.f0 = -3; z.f0 > 16; z.f0 += 1)
      foo();
    return g_120;
}
int main() {
    while (g_339 % 0x200003) continue;
    if (i)
      func_72(0, 0, g_5);
    foo(&g_339);
    return 0;
}
EOF
gcc -O2 -w -c test.c

test.c: In function ‘main’:
test.c:17:1: error: unable to find a register to spill in class ‘DREG’
test.c:17:1: error: this is the insn:
(insn 7 6 8 2 (parallel [
            (set (reg:SI 2 cx [64])
                (truncate:SI (lshiftrt:DI (mult:DI (sign_extend:DI (reg:SI 4 si
[orig:61 pretmp.11 ] [61]))
                            (sign_extend:DI (reg:SI 2 cx [66])))
                        (const_int 32 [0x20]))))
            (clobber (scratch:SI))
            (clobber (reg:CC 17 flags))
        ]) test.c:12 343 {*smulsi3_highpart_1}
     (expr_list:REG_DEAD (reg:SI 2 cx [66])
        (expr_list:REG_UNUSED (reg:CC 17 flags)
            (expr_list:REG_EQUAL (truncate:SI (lshiftrt:DI (mult:DI
(sign_extend:DI (reg:SI 4 si [orig:61 pretmp.11 ] [61]))
                            (const_int 2147480577 [0x7ffff401]))
                        (const_int 32 [0x20])))
                (nil)))))
test.c:17:1: internal compiler error: in spill_failure, at reload1.c:2113


This test case is reduced from the output of Csmith 2.1.0 (git hash 01aa8b04,
https://github.com/Quuxplusone/csmith/), using the following command line:
csmith --paranoid --longlong --no-pointers --no-arrays --jumps --consts
--volatiles --checksum --no-divs --muls --no-bitfields --no-packed-struct -s
855007444

[Bug target/49206] [4.5/4.6/4.7 Regression] RA failure in spill_failure, at reload1.c:2113

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.07.24 20:54:39
   Target Milestone|---                         |4.5.4
            Summary|RA failure in               |[4.5/4.6/4.7 Regression] RA
                   |spill_failure, at           |failure in spill_failure,
                   |reload1.c:2113              |at reload1.c:2113
     Ever Confirmed|0                           |1

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-07-24 20:54:39 UTC ---
Confirmed.

[Bug target/49206] [4.5/4.6/4.7 Regression] RA failure in spill_failure, at reload1.c:2113

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

[Bug target/49206] [4.5/4.6/4.7 Regression] RA failure in spill_failure, at reload1.c:2113

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-08-22 07:29:16 UTC ---
Can only reproduce this with
http://gcc.gnu.org/viewcvs?view=revision&revision=172023
and later.

[Bug target/49206] [4.5/4.6/4.7 Regression] RA failure in spill_failure, at reload1.c:2113

[sergos.gnu at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

--- Comment #3 from Sergey Ostanevich <sergos.gnu at gmail dot com> 2011-08-22 16:37:54 UTC ---
is it right that while() is an infinite loop? at least some phases can rely on
this?

[Bug target/49206] [4.5/4.6/4.7 Regression] RA failure in spill_failure, at reload1.c:2113

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-08-22 16:43:28 UTC ---
It is a conditional infinite loop (identical to if (g_339 % 0x200003) for (;;);
).

[Bug target/49206] [4.5/4.6/4.7 Regression] RA failure in spill_failure, at reload1.c:2113

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

--- Comment #5 from Uros Bizjak <ubizjak at gmail dot com> 2012-03-01 19:57:27 UTC ---
The testcase works OK for 4.7, 64bit and 32bit targets.

GCC: (GNU) 4.7.0 20120229 (experimental) [trunk revision 184671]

I propose to add the testcase to the 4.7 torture testsuite and close the bug.

[Bug target/49206] [4.5/4.6/4.7 Regression] RA failure in spill_failure, at reload1.c:2113

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-03-01 21:31:33 UTC ---
This went away with http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=181075
but was it really a fix for this, rather than just something that made the bug
latent?

[Bug target/49206] [4.5/4.6/4.7 Regression] RA failure in spill_failure, at reload1.c:2113

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

--- Comment #7 from Uros Bizjak <ubizjak at gmail dot com> 2012-03-01 23:04:12 UTC ---
(In reply to comment #6)
> This went away with http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=181075
> but was it really a fix for this, rather than just something that made the bug
> latent?

I am afraid that the change is not relevant for this bug, so the later...

[Bug target/49206] [4.5/4.6/4.7/4.8 Regression] RA failure in spill_failure, at reload1.c:2113

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.5.4                       |4.6.4

--- Comment #8 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-07-02 11:40:05 UTC ---
The 4.5 branch is being closed, adjusting target milestone.

[Bug target/49206] [4.6/4.7/4.8 Regression] RA failure in spill_failure, at reload1.c:2113

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

--- Comment #9 from Uros Bizjak <ubizjak at gmail dot com> 2012-09-02 08:35:45 UTC ---
(In reply to comment #7)

> > This went away with http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=181075
> > but was it really a fix for this, rather than just something that made the bug
> > latent?
> 
> I am afraid that the change is not relevant for this bug, so the later...

I have backed up r181075 on the mainline, and the test still passes.  I will
put the test to compile torture testsuite, as this is the maximum we can do
now.

[Bug target/49206] [4.6/4.7/4.8 Regression] RA failure in spill_failure, at reload1.c:2113

[uros at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

--- Comment #10 from uros at gcc dot gnu.org 2012-09-02 08:49:37 UTC ---
Author: uros
Date: Sun Sep  2 08:49:31 2012
New Revision: 190857

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=190857
Log:
    PR target/49206
    * gcc.c-torture/compile/pr49206.c: New test.


Added:
    trunk/gcc/testsuite/gcc.c-torture/compile/pr49206.c
Modified:
    trunk/gcc/testsuite/ChangeLog

[Bug target/49206] [4.6/4.7/4.8 Regression] RA failure in spill_failure, at reload1.c:2113

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

Uros Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.7.1
   Target Milestone|4.6.4                       |4.8.0
      Known to fail|                            |4.6.3

--- Comment #11 from Uros Bizjak <ubizjak at gmail dot com> 2012-09-02 08:58:52 UTC ---
Closed as fixed in 4.8.0.

[Bug target/49206] [4.6 Regression] RA failure in spill_failure, at reload1.c:2113

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

Uros Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.8.0                       |4.6.4
            Summary|[4.6/4.7/4.8 Regression] RA |[4.6 Regression] RA failure
                   |failure in spill_failure,   |in spill_failure, at
                   |at reload1.c:2113           |reload1.c:2113

--- Comment #12 from Uros Bizjak <ubizjak at gmail dot com> 2012-09-02 09:10:15 UTC ---
Let's keep this PR open as a 4.6 regression.

[Bug target/49206] [4.6 Regression] RA failure in spill_failure, at reload1.c:2113

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49206

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|4.6.4                       |4.7.0

--- Comment #13 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-04-12 16:17:48 UTC ---
The 4.6 branch has been closed, fixed in GCC 4.7.0.