[Bug rtl-optimization/48133] New: [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[Bug rtl-optimization/48133] New: [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

           Summary: [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at
                    cfgloop.c:831 with -O -funroll-loops -fthread-jumps
                    -fno-tree-ch
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: zsojka@seznam.cz
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu


Created attachment 23662
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23662
reduced testcase (from gcc.dg/pr47899.c)

+++ This bug was initially created as a clone of Bug #47899 +++

Compiler output:
$ gcc -O -funroll-loops -fthread-jumps -fno-tree-ch testcase.c
testcase.c: In function 'foo':
testcase.c:11:1: internal compiler error: in get_loop_body, at cfgloop.c:831
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

(gdb) bt
#0  fancy_abort (file=0x10f5258 "/mnt/svn/gcc-4_6/gcc/cfgloop.c", line=831,
function=0x10f5792 "get_loop_body")
    at /mnt/svn/gcc-4_6/gcc/diagnostic.c:892
#1  0x00000000005f1511 in get_loop_body (loop=0x7ffff7ed2d48) at
/mnt/svn/gcc-4_6/gcc/cfgloop.c:831
#2  0x00000000005f2fe8 in verify_loop_structure () at
/mnt/svn/gcc-4_6/gcc/cfgloop.c:1343
#3  0x00000000007bffbc in peel_loops_completely (flags=2) at
/mnt/svn/gcc-4_6/gcc/loop-unroll.c:259
#4  unroll_and_peel_loops (flags=2) at /mnt/svn/gcc-4_6/gcc/loop-unroll.c:165
#5  0x00000000007afc48 in rtl_unroll_and_peel_loops () at
/mnt/svn/gcc-4_6/gcc/loop-init.c:332
#6  0x00000000007f6b46 in execute_one_pass (pass=0x1639e20) at
/mnt/svn/gcc-4_6/gcc/passes.c:1556
#7  0x00000000007f6e45 in execute_pass_list (pass=0x1639e20) at
/mnt/svn/gcc-4_6/gcc/passes.c:1611
#8  0x00000000007f6e57 in execute_pass_list (pass=0x163a000) at
/mnt/svn/gcc-4_6/gcc/passes.c:1612
#9  0x00000000007f6e57 in execute_pass_list (pass=0x163a220) at
/mnt/svn/gcc-4_6/gcc/passes.c:1612
#10 0x0000000000939d76 in tree_rest_of_compilation (fndecl=0x7ffff5bcef00) at
/mnt/svn/gcc-4_6/gcc/tree-optimize.c:422
#11 0x0000000000b01c12 in cgraph_expand_function (node=0x7ffff5bec000) at
/mnt/svn/gcc-4_6/gcc/cgraphunit.c:1576
#12 0x0000000000b0435a in cgraph_expand_all_functions () at
/mnt/svn/gcc-4_6/gcc/cgraphunit.c:1635
#13 cgraph_optimize () at /mnt/svn/gcc-4_6/gcc/cgraphunit.c:1899
#14 0x0000000000b048da in cgraph_finalize_compilation_unit () at
/mnt/svn/gcc-4_6/gcc/cgraphunit.c:1096
#15 0x00000000005097fc in c_write_global_declarations () at
/mnt/svn/gcc-4_6/gcc/c-decl.c:9872
#16 0x00000000008e2d08 in compile_file (argc=16, argv=0x7fffffffdc28) at
/mnt/svn/gcc-4_6/gcc/toplev.c:591
#17 do_compile (argc=16, argv=0x7fffffffdc28) at
/mnt/svn/gcc-4_6/gcc/toplev.c:1900
#18 toplev_main (argc=16, argv=0x7fffffffdc28) at
/mnt/svn/gcc-4_6/gcc/toplev.c:1963
#19 0x00007ffff648cb6d in __libc_start_main () from /lib/libc.so.6
#20 0x00000000004f03b1 in _start ()

Tested revisions:
r170964 - crash
4.6 r170955 - crash
4.5 r170955 - crash
4.4 r170955 - OK

[Bug rtl-optimization/48133] [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.03.15 11:44:24
   Target Milestone|---                         |4.5.3
     Ever Confirmed|0                           |1

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-03-15 11:44:24 UTC ---
Confirmed.

[Bug rtl-optimization/48133] [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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> 2011-03-15 14:13:22 UTC ---
Started failing with http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=149206

[Bug rtl-optimization/48133] [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

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

[Bug rtl-optimization/48133] [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

--- Comment #3 from Zdenek Sojka <zsojka at seznam dot cz> 2011-04-14 19:01:43 UTC ---
Created attachment 23988
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23988
C testcase (reduced from libgomp.c/autopar-1.c)

$ gcc -O -ftree-parallelize-loops=4 -fno-tree-ch -fno-tree-dominator-opts
testcase.c
$ ./a.out 
Aborted

[Bug rtl-optimization/48133] [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

Zdenek Sojka <zsojka at seznam dot cz> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #23988|0                           |1
        is obsolete|                            |

--- Comment #4 from Zdenek Sojka <zsojka at seznam dot cz> 2011-04-14 19:03:16 UTC ---
Comment on attachment 23988
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23988
C testcase (reduced from libgomp.c/autopar-1.c)

Sorry, it belongs to PR46886, I didn't realise bugzilla moves me to another bug
in the bug list when I make a change.

[Bug rtl-optimization/48133] [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.5.3                       |4.5.4

--- Comment #5 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-04-28 14:51:19 UTC ---
GCC 4.5.3 is being released, adjusting target milestone.

[Bug rtl-optimization/48133] [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-01-27 14:07:35 UTC ---
Honza, ping?

[Bug rtl-optimization/48133] [4.5/4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

--- Comment #7 from Uros Bizjak <ubizjak at gmail dot com> 2012-02-27 20:50:46 UTC ---
I have done a bit of debugging here.

The problem starts in peel_loop_completely, loop-unroll.c. Before remove_path
(ein), we have:

loop_0 (header = 0, latch = 1, niter = )
{
  bb_2 (preds = {bb_0 }, succs = {bb_3 })
  bb_9 (preds = {bb_8 }, succs = {bb_1 })
  loop_1 (header = 3, latch = 10, niter = )
  {
    bb_3 (preds = {bb_2 bb_10 }, succs = {bb_4 })
    bb_10 (preds = {bb_5 }, succs = {bb_3 })
    loop_2 (header = 4, latch = 11, niter = )
    {
      bb_4 (preds = {bb_11 bb_3 }, succs = {bb_7 bb_5 })
      bb_5 (preds = {bb_4 }, succs = {bb_6 bb_10 })
      bb_6 (preds = {bb_7 bb_5 }, succs = {bb_7 })
      bb_7 (preds = {bb_4 bb_6 }, succs = {bb_6 bb_8 })
      bb_8 (preds = {bb_7 }, succs = {bb_11 bb_9 })
      bb_11 (preds = {bb_8 }, succs = {bb_4 })
    }
  }
}

We cancel loop2 completely and path ( 8 -> 11), and we get:

loop_0 (header = 0, latch = 1, niter = )
{
  bb_2 (preds = {bb_0 }, succs = {bb_3 })
  bb_8 (preds = {bb_7 }, succs = {bb_9 })
  bb_9 (preds = {bb_8 }, succs = {bb_1 })
  loop_1 (header = 3, latch = 10, niter = )
  {
    bb_3 (preds = {bb_2 bb_10 }, succs = {bb_4 })
    bb_4 (preds = {bb_3 }, succs = {bb_7 bb_5 })
    bb_5 (preds = {bb_4 }, succs = {bb_6 bb_10 })
    bb_10 (preds = {bb_5 }, succs = {bb_3 })
    bb_6 (preds = {bb_7 bb_5 }, succs = {bb_7 })
    bb_7 (preds = {bb_4 bb_6 }, succs = {bb_6 bb_8 })
  }
}

Analysing loop_1 through get_loop_body -> get_loop_body_with_size ->
dfs_enumerate_from (..., 1 /* reverse */, ...), we visit only BBs 3, 10, 5 and
4, leaving out BBs 6 and 7.

This crashes compilation later in get_loop_body:

  gcc_assert (tv == loop->num_nodes);

since tv == 4 and loop->num_nodes == 6.

[Bug rtl-optimization/48133] [4.5/4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[dcb314 at hotmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

dcb <dcb314 at hotmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dcb314 at hotmail dot com

--- Comment #8 from dcb <dcb314 at hotmail dot com> 2012-03-28 19:36:17 UTC ---

I am also seeing this problem when compiling the new Linux kernel version 3.3

[Bug rtl-optimization/48133] [4.5/4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[mario-baumann at web dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

--- Comment #9 from Mario Baumann <mario-baumann at web dot de> 2012-03-29 13:16:59 UTC ---
Created attachment 27032
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=27032
fortran90 source file

[Bug rtl-optimization/48133] [4.5/4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[mario-baumann at web dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

Mario Baumann <mario-baumann at web dot de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mario-baumann at web dot de

--- Comment #10 from Mario Baumann <mario-baumann at web dot de> 2012-03-29 13:17:20 UTC ---
added a fortran testcase `goo.f90` which fails with the same error message!

> gfortran -O2 -c goo.f90
goo.f90: In function 'goo':
goo.f90:16:0: internal compiler error: in get_loop_body, at cfgloop.c:831
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

> gfortran -v
Using built-in specs.
COLLECT_GCC=/app2/gcc/4.8.0-20120329-svn185950/i686/bin/gfortran
COLLECT_LTO_WRAPPER=/app2/gcc/4.8.0-20120329-svn185950/i686/libexec/gcc/i686-unknown-linux-gnu/4.8.0/lto-wrapper
Target: i686-unknown-linux-gnu
Configured with: ./configure --prefix=/app2/gcc/4.8.0-20120329-svn185950/i686
--enable-languages=c,c++,fortran --disable-nls
--with-gmp=/app2/gcc/4.8.0-20120329-svn185950/i686/aux
--with-mpfr=/app2/gcc/4.8.0-20120329-svn185950/i686/aux
--with-mpc=/app2/gcc/4.8.0-20120329-svn185950/i686/aux
--with-ppl=/app2/gcc/4.8.0-20120329-svn185950/i686/aux
--with-cloog=/app2/gcc/4.8.0-20120329-svn185950/i686/aux --disable-multilib
--build=i686-unknown-linux-gnu
Thread model: posix
gcc version 4.8.0 20120329 (experimental) (GCC)

[Bug rtl-optimization/48133] [4.5/4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

Jan Hubicka <hubicka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
         AssignedTo|unassigned at gcc dot       |hubicka at gcc dot gnu.org
                   |gnu.org                     |

--- Comment #11 from Jan Hubicka <hubicka at gcc dot gnu.org> 2012-03-29 14:29:01 UTC ---
I tought I solved this problem already by making path removal code to also
cancel loops, but this testcase seems different.
I will look into it.  The patch cited above just triggers previously latent
problem.

Honza

[Bug rtl-optimization/48133] [4.5/4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

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

--- Comment #12 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-07-02 11:10:41 UTC ---
The 4.5 branch is being closed, adjusting target milestone.

[Bug rtl-optimization/48133] [4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

Jan Hubicka <hubicka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hjl.tools at gmail dot com

--- Comment #13 from Jan Hubicka <hubicka at gcc dot gnu.org> 2012-08-10 05:38:10 UTC ---
This seems to work for me now
jh@gcc10:~/trunk/build/gcc$ ./gfortran -O2 -c t.f90 -B ./
jh@gcc10:~/trunk/build/gcc$ cat t.f90
SUBROUTINE goo()
IMPLICIT NONE
CHARACTER(len=9),SAVE :: s
INTEGER,SAVE :: i,j,k
i=0
j=0
DO WHILE (i==0)
   CALL goo1(s)
   IF (INDEX(s,'$')/=1 .AND. INDEX(s,'!')/=1) THEN
      CALL goo2(k)
      IF (k>0) THEN
         i=1
      END IF
   END IF
END DO
END SUBROUTINE
! gfortran -O2 -c goo.f90

(similarly with the C testcase).  It would be nice to know what fixed it. H.J,
do you think you can track it?

[Bug rtl-optimization/48133] [4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

--- Comment #14 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-08-10 06:17:44 UTC ---
(In reply to comment #13)
 seems to work for me now
> jh@gcc10:~/trunk/build/gcc$ ./gfortran -O2 -c t.f90 -B ./
> jh@gcc10:~/trunk/build/gcc$ cat t.f90
> SUBROUTINE goo()
> IMPLICIT NONE
> CHARACTER(len=9),SAVE :: s
> INTEGER,SAVE :: i,j,k
> i=0
> j=0
> DO WHILE (i==0)
>    CALL goo1(s)
>    IF (INDEX(s,'$')/=1 .AND. INDEX(s,'!')/=1) THEN
>       CALL goo2(k)
>       IF (k>0) THEN
>          i=1
>       END IF
>    END IF
> END DO
> END SUBROUTINE
> ! gfortran -O2 -c goo.f90
> 
> (similarly with the C testcase).  It would be nice to know what fixed it. H.J,
> do you think you can track it?

The #c0 testcase stopped ICEing with
http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=185913
and the #c13 testcase just ICEd a few revisions around r185950.

[Bug rtl-optimization/48133] [4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[mpolacek at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

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

--- Comment #15 from Marek Polacek <mpolacek at gcc dot gnu.org> 2013-02-08 17:37:04 UTC ---
The testcases no longer fail for me; can this be closed now?

[Bug rtl-optimization/48133] [4.6/4.7/4.8 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[mpolacek at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

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

--- Comment #16 from Marek Polacek <mpolacek at gcc dot gnu.org> 2013-02-13 17:31:29 UTC ---
Seems like it, thus -> fixed.

[Bug rtl-optimization/48133] [4.6/4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
      Known to work|                            |4.8.0
         Resolution|FIXED                       |
            Summary|[4.6/4.7/4.8 Regression]    |[4.6/4.7 Regression] ICE:
                   |ICE: in get_loop_body, at   |in get_loop_body, at
                   |cfgloop.c:831 with -O       |cfgloop.c:831 with -O
                   |-funroll-loops              |-funroll-loops
                   |-fthread-jumps -fno-tree-ch |-fthread-jumps -fno-tree-ch

--- Comment #17 from Richard Biener <rguenth at gcc dot gnu.org> 2013-02-14 10:34:01 UTC ---
Doesn't it still fail with 4.6 and 4.7? ... yes, at least with 4.7.

[Bug rtl-optimization/48133] [4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.6.4                       |4.7.4

--- Comment #18 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-04-12 15:16:44 UTC ---
GCC 4.6.4 has been released and the branch has been closed.

[Bug rtl-optimization/48133] [4.7 Regression] ICE: in get_loop_body, at cfgloop.c:831 with -O -funroll-loops -fthread-jumps -fno-tree-ch

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=48133

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|---                         |FIXED
   Target Milestone|4.7.4                       |4.8.0
      Known to fail|                            |4.7.4

--- Comment #19 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed for 4.8.0.