[Bug rtl-optimization/57960] New: LRA ICE building glibc

[Bug rtl-optimization/57960] New: LRA ICE building glibc

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57960

            Bug ID: 57960
           Summary: LRA ICE building glibc
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: krebbel at gcc dot gnu.org

Testcase extracted from Glibc e_acosl.c:

typedef union
{
  long double value;
  struct
  {
    unsigned int w0, w1, w2, w3;
  }
  parts32;
}
ieee854_long_double_shape_type;
static const long double one = 1.0L;
long double
__ieee754_acosl (long double x)
{
  long double z, w;
  int ix;
  ieee854_long_double_shape_type u;

      z = (one - u.value) * 0.5;
      u.parts32.w2 = 0;
      u.parts32.w3 = 0;
      w = z - u.value * u.value;
      return 2.0 * w;

}

builds fine with: cc1 -fpreprocessed  -quiet  -O1 -mno-lra

but with: cc1 -fpreprocessed  -quiet  -O1

t.c: In function ‘__ieee754_acosl’:
t.c:25:1: error: unable to generate reloads for:
 }
 ^
(insn 14 13 16 2 (set (strict_low_part (subreg:SI (reg/v:TI 51 [ u ]) 12))
        (const_int 0 [0])) t.c:21 77 {movstrictsi}
     (nil))
t.c:25:1: internal compiler error: in curr_insn_transform, at
lra-constraints.c:2954
0x8051ecb5 _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
        /home/andreas/patched/gcc-head/gcc/rtl-error.c:109
0x8046895b curr_insn_transform
        /home/andreas/patched/gcc-head/gcc/lra-constraints.c:2954
0x804692e1 lra_constraints(bool)
        /home/andreas/patched/gcc-head/gcc/lra-constraints.c:3810
0x804577a9 lra(_IO_FILE*)
        /home/andreas/patched/gcc-head/gcc/lra.c:2319
0x8040c22b do_reload
        /home/andreas/patched/gcc-head/gcc/ira.c:4689
0x8040c22b rest_of_handle_reload
        /home/andreas/patched/gcc-head/gcc/ira.c:4801
>From gcc-bugs-return-426562-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Tue Jul 23 09:06:09 2013
Return-Path: <gcc-bugs-return-426562-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 1133 invoked by alias); 23 Jul 2013 09:06:09 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 30882 invoked by uid 48); 23 Jul 2013 09:04:05 -0000
From: "burnus at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug fortran/57959] ICE with structure constructor with scalar allocatable components
Date: Tue, 23 Jul 2013 09:06:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: fortran
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords: ice-on-valid-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: burnus at gcc dot gnu.org
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-57959-4-t9K31OEvhi@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-57959-4@http.gcc.gnu.org/bugzilla/>
References: <bug-57959-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2013-07/txt/msg01069.txt.bz2
Content-length: 571

http://gcc.gnu.org/bugzilla/show_bug.cgi?idW959

--- Comment #1 from Tobias Burnus <burnus at gcc dot gnu.org> ---
gfc_trans_subcomponent_assign lacks handling of allocatable scalars. It simply
does:
    type1.0.anum = 5.0
instead of:
    (re)allocate of "type1.0.anum"
    *type1.0.anum = 5.0

For arrays, it is handled via gfc_trans_alloc_subarray_assign, which is called
via gfc_trans_structure_assign -> gfc_trans_subcomponent_assign.

TODO: Also check whether coarrays are properly handled. "= TYPE(CAF=5.0)"
should be valid, but the LHS shan't be (re)allocated.

[Bug rtl-optimization/57960] LRA ICE building glibc

[mpolacek at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57960

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

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

--- Comment #1 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
But this is s390x, right?  (Judging from the movstrictsi.)

[Bug rtl-optimization/57960] S/390: LRA ICE building glibc

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57960

Andreas Krebbel <krebbel at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|                            |s390x-ibm-linux
               Host|                            |s390x-ibm-linux
            Summary|LRA ICE building glibc      |S/390: LRA ICE building
                   |                            |glibc
              Build|                            |s390x-ibm-linux

--- Comment #2 from Andreas Krebbel <krebbel at gcc dot gnu.org> ---
(In reply to Marek Polacek from comment #1)
> But this is s390x, right?  (Judging from the movstrictsi.)

Yes.

[Bug rtl-optimization/57960] S/390: LRA ICE building glibc

[vmakarov at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57960

Vladimir Makarov <vmakarov at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |vmakarov at redhat dot com

--- Comment #3 from Vladimir Makarov <vmakarov at redhat dot com> ---
(In reply to Andreas Krebbel from comment #2)
> (In reply to Marek Polacek from comment #1)
> > But this is s390x, right?  (Judging from the movstrictsi.)
> 
> Yes.

Thanks, Andrew.  I've reproduced it.  I guess a fix will be ready on this week
as the bug is in a sensitive part of LRA and the fix will need a lot of testing
on a few machines.

[Bug rtl-optimization/57960] S/390: LRA ICE building glibc

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57960

Andreas Krebbel <krebbel at gcc dot gnu.org> changed:

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

--- Comment #4 from Andreas Krebbel <krebbel at gcc dot gnu.org> ---
Fixed with r201243

2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>

    PR rtl-optimization/57960
    * lra-constraints.c (process_alt_operands): Use the right mode
    when checking strict_low.

2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>

    PR rtl-optimization/57960
    * gcc.target/s390/pr57960.c: New.