[Bug rtl-optimization/55055] New: [4.8 regression] RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1123

[Bug rtl-optimization/55055] New: [4.8 regression] RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1123

[dimhen at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55055

             Bug #: 55055
           Summary: [4.8 regression] RTL check: expected code 'reg', have
                    'subreg' in rhs_regno, at rtl.h:1123
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: dimhen@gmail.com


$ gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/local/gcc_current/libexec/gcc/x86_64-unknown-linux-gnu/4.8.0/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: /home/dimhen/src/gcc-current/configure
--prefix=/usr/local/gcc_current --enable-__cxa_atexit --enable-shared
--enable-checking=yes,df,fold,rtl,tree --enable-gnu-unique-object
--enable-linker-build-id --enable-languages=c,c++,lto --enable-plugin
--enable-version-specific-runtime-libs --with-tune=generic
Thread model: posix
gcc version 4.8.0 20121024 (experimental) [trunk revision 192761] (GCC) 


PASS -02
FAIL -O2 -ftree-vectorize

$ gcc -fpreprocessed -Wall -O2 -ftree-vectorize -c 1.i
1.i: In function 'fn1':
1.i:12:1: internal compiler error: RTL check: expected code 'reg', have
'subreg' in rhs_regno, at rtl.h:1123
 }
 ^
0x8e0847 rtl_check_failed_code1(rtx_def const*, rtx_code, char const*, int,
char const*)
    /home/dimhen/src/gcc-current/gcc/rtl.c:772
0xfa555f rhs_regno
    /home/dimhen/src/gcc-current/gcc/rtl.h:1123
0xd4f490 rhs_regno
    /home/dimhen/src/gcc-current/gcc/rtl.h:1122
0xf823be peephole2_insns(rtx_def*, rtx_def*, int*)
    /home/dimhen/src/gcc-current/gcc/config/i386/i386.md:17380
0x890b20 peephole2_optimize
    /home/dimhen/src/gcc-current/gcc/recog.c:3551
0x890b20 rest_of_handle_peephole2
    /home/dimhen/src/gcc-current/gcc/recog.c:3735
Please submit a full bug report,

reduced test
$ cat 1.i
long a;
int *b, *c;
void fn1 ()
{
    int l, d;
    d = a;
    for (; d; d--, c++)
    {
        l = *b++;
        *c = l;
    }
}

[Bug rtl-optimization/55055] [4.8 regression] RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1123

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55055

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.8.0

[Bug rtl-optimization/55055] [4.8 regression] RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1123

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55055

Uros Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ra
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-10-24
                 CC|                            |vmakarov at gcc dot gnu.org
     Ever Confirmed|0                           |1

--- Comment #1 from Uros Bizjak <ubizjak at gmail dot com> 2012-10-24 15:17:26 UTC ---
There should be no subregs in peephole2 pass.

In .203r.ira, we have:

(insn 153 156 169 9 (set (reg:SI 154 [ D.1790 ])
        (plus:SI (subreg:SI (reg:DI 87 [ a.0 ]) 0)
            (const_int -1 [0xffffffffffffffff]))) 240 {*leasi}
     (expr_list:REG_DEAD (reg:DI 87 [ a.0 ])
        (nil)))

and in .204r.reload, we still have:

(insn 153 156 169 9 (set (reg:SI 37 r8 [orig:154 D.1790 ] [154])
        (plus:SI (subreg:SI (reg:DI 37 r8 [orig:87 a.0 ] [87]) 0)
            (const_int -1 [0xffffffffffffffff]))) 240 {*leasi}
     (expr_list:REG_DEAD (reg:DI 37 r8 [orig:87 a.0 ] [87])
        (nil)))

I will take an uneducated guess that this is due to new LRA.

Adding CC.

[Bug rtl-optimization/55055] [4.8 regression] RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1123

[vmakarov at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55055

--- Comment #2 from Vladimir Makarov <vmakarov at gcc dot gnu.org> 2012-10-24 20:02:12 UTC ---
Author: vmakarov
Date: Wed Oct 24 20:02:08 2012
New Revision: 192779

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=192779
Log:
2012-10-24  Vladimir Makarov  <vmakarov@redhat.com>

    PR rtl-optimization/55055
    * lra-spills.c (alter_subregs): New function.
    (lra_hard_reg_substitution): Use it.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/lra-spills.c

[Bug rtl-optimization/55055] [4.8 regression] RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1123

[dimhen at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55055

Dmitry G. Dyachenko <dimhen at gmail dot com> changed:

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

--- Comment #3 from Dmitry G. Dyachenko <dimhen at gmail dot com> 2012-10-27 16:50:39 UTC ---
fixed in 192779