[Bug target/48273] New: [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[Bug target/48273] New: [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

           Summary: [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx,
                    at sel-sched-ir.c:5604 with -fsel-sched-pipelining
                    -fselective-scheduling2 -march=core2
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: zsojka@seznam.cz
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu


Created attachment 23769
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23769
reduced testcase

Compiler output:
$ gcc -O -fschedule-insns2 -fsel-sched-pipelining -fselective-scheduling2
-funroll-all-loops -march=core2 testcase.C
testcase.C: In function 'void foo()':
testcase.C:7:1: internal compiler error: in create_copy_of_insn_rtx, at
sel-sched-ir.c:5604
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

Tested revisions:
r171337 - crash
4.6 r170955 - crash
4.5 r170955 - crash

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.6.1

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

Uros Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|x86_64-pc-linux-gnu         |i686-pc-linux-gnu
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.03.24 16:43:05
     Ever Confirmed|0                           |1

--- Comment #1 from Uros Bizjak <ubizjak at gmail dot com> 2011-03-24 16:43:05 UTC ---
-m32 is also needed to trigger this ICE, so i686-pc-linux-gnu.

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

--- Comment #2 from Uros Bizjak <ubizjak at gmail dot com> 2011-03-24 16:46:08 UTC ---
(gdb) bt
#0  fancy_abort (file=0xd77f00 "../../gcc-svn/trunk/gcc/sel-sched-ir.c", 
    line=5604, function=0xd78e40 "create_copy_of_insn_rtx")
    at ../../gcc-svn/trunk/gcc/diagnostic.c:892
#1  0x00000000007bac89 in create_copy_of_insn_rtx (insn_rtx=0x7ffff1f33780)
    at ../../gcc-svn/trunk/gcc/sel-sched-ir.c:5604
#2  0x00000000007cdf30 in emit_bookkeeping_insn (insn=0x7ffff1f33820, 
    lparams=0x7fffffffd870, static_params=Unhandled dwarf expression opcode
0xf3
)
    at ../../gcc-svn/trunk/gcc/sel-sched.c:4767
#3  generate_bookkeeping_insn (insn=0x7ffff1f33820, lparams=0x7fffffffd870, 
    static_params=Unhandled dwarf expression opcode 0xf3
) at ../../gcc-svn/trunk/gcc/sel-sched.c:4805
#4  move_op_at_first_insn (insn=0x7ffff1f33820, lparams=0x7fffffffd870, 
    static_params=Unhandled dwarf expression opcode 0xf3
) at ../../gcc-svn/trunk/gcc/sel-sched.c:6073
#5  0x00000000007c7009 in code_motion_path_driver (insn=0x7ffff1f33820, 
    orig_ops=0x0, path=<value optimized out>, local_params_in=0x7fffffffd870, 
    static_params=0x7fffffffdaa0) at ../../gcc-svn/trunk/gcc/sel-sched.c:6638
#6  0x00000000007c668f in code_motion_process_successors (insn=0x7ffff1f33aa0, 
    orig_ops=0x1414978, path=0x14149f0, static_params=0x7fffffffdaa0)
    at ../../gcc-svn/trunk/gcc/sel-sched.c:6354
#7  0x00000000007c71dc in code_motion_path_driver (insn=0x7ffff1f33aa0, 
    orig_ops=0x1414978, path=0x14149f0, local_params_in=0x7fffffffda70, 
    static_params=0x7fffffffdaa0) at ../../gcc-svn/trunk/gcc/sel-sched.c:6600

(gdb) up
#1  0x00000000007bac89 in create_copy_of_insn_rtx (insn_rtx=0x7ffff1f33780)
    at ../../gcc-svn/trunk/gcc/sel-sched-ir.c:5604
5604      gcc_assert (NONJUMP_INSN_P (insn_rtx));
(gdb) li
5599    
5600      if (DEBUG_INSN_P (insn_rtx))
5601        return create_insn_rtx_from_pattern (copy_rtx (PATTERN (insn_rtx)),
5602                         insn_rtx);
5603    
5604      gcc_assert (NONJUMP_INSN_P (insn_rtx));
5605    
5606      res = create_insn_rtx_from_pattern (copy_rtx (PATTERN (insn_rtx)),
5607                                          NULL_RTX);
5608      return res;

(gdb) p debug_rtx (insn_rtx)
(call_insn 6 0 0 3 (call (mem:QI (symbol_ref:SI ("bar") [flags 0x41]
<function_decl 0x7ffff1f24c00 bar>) [0 bar S1 A8])
        (const_int 0 [0])) pr48273.c:6 618 {*call_0}
     (nil)
    (nil))

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

--- Comment #3 from Zdenek Sojka <zsojka at seznam dot cz> 2011-03-24 16:54:05 UTC ---
(In reply to comment #1)
> -m32 is also needed to trigger this ICE, so i686-pc-linux-gnu.

Yes, sorry, originally I tested the code as C++ code, and in that case, -m32 is
not needed.

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[amonakov at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

Alexander Monakov <amonakov at gcc dot gnu.org> changed:

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

--- Comment #4 from Alexander Monakov <amonakov at gcc dot gnu.org> 2011-03-24 17:32:39 UTC ---
So the assert disagrees with the rest of the scheduler on whether calls may be
cloned.  I'll talk to Andrey about amending one or the other.

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[amonakov at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

--- Comment #5 from Alexander Monakov <amonakov at gcc dot gnu.org> 2011-04-06 14:40:28 UTC ---
There's little value in allowing bookkeeping for calls, and the (untested yet)
patch below forces calls to be unique.  The test case also shows that our
remove_insns_that_need_bookkeeping function works incorrectly for infinite
loops (the single basic block dominates itself, and SPEC of the pipelined
instruction is not increased because there's no conditional branch). It's
possible to fix up that function by also looking at INSN_SCHED_TIMES, but the
simpler way is to forbid pipelining for infinite loops.

diff --git a/gcc/sel-sched-ir.c b/gcc/sel-sched-ir.c
index 67484dd..95c1431 100644
--- a/gcc/sel-sched-ir.c
+++ b/gcc/sel-sched-ir.c
@@ -2905,6 +2905,7 @@ init_global_and_expr_for_insn (insn_t insn)
       if (CANT_MOVE (insn)
           || INSN_ASM_P (insn)
           || SCHED_GROUP_P (insn)
+      || CALL_P (insn)
           /* Exception handling insns are always unique.  */
           || (cfun->can_throw_non_call_exceptions && can_throw_internal
(insn))
           /* TRAP_IF though have an INSN code is control_flow_insn_p ().  */
diff --git a/gcc/sel-sched.c b/gcc/sel-sched.c
index 0e8173b..415eaa4 100644
--- a/gcc/sel-sched.c
+++ b/gcc/sel-sched.c
@@ -6781,7 +6781,9 @@ sel_setup_region_sched_flags (void)
   bookkeeping_p = 1;
   pipelining_p = (bookkeeping_p
                   && (flag_sel_sched_pipelining != 0)
-          && current_loop_nest != NULL);
+          && current_loop_nest != NULL
+          /* Check that there is at least one exit.  */
+          && current_loop_nest->exits->next->e);
   max_insns_to_rename = PARAM_VALUE (PARAM_SELSCHED_INSNS_TO_RENAME);
   max_ws = MAX_WS;
 }

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[abel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

Andrey Belevantsev <abel at gcc dot gnu.org> changed:

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

--- Comment #6 from Andrey Belevantsev <abel at gcc dot gnu.org> 2011-04-07 07:25:19 UTC ---
(In reply to comment #5)
> +          && current_loop_nest != NULL
> +          /* Check that there is at least one exit.  */
> +          && current_loop_nest->exits->next->e);
This exposes loop optimizer internals, so please add a helper to cfgloop.c
instead.

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[abel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

--- Comment #7 from Andrey Belevantsev <abel at gcc dot gnu.org> 2011-04-07 08:00:28 UTC ---
And some of our sel-sched-ir.h helpers can be moved to cfgloop.c as well (e.g.
get_loop_exit_edges_unique_dests).

[Bug target/48273] [4.6/4.7 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[amonakov at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

--- Comment #8 from Alexander Monakov <amonakov at gcc dot gnu.org> 2011-04-08 11:49:47 UTC ---
Author: amonakov
Date: Fri Apr  8 11:49:43 2011
New Revision: 172175

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=172175
Log:
    PR target/48273
    * cfgloop.h (loop_has_exit_edges): New helper.
    * sel-sched-ir.c (init_global_and_expr_for_insn): Make CALLs
    non-clonable.
    * sel-sched.c (sel_setup_region_sched_flags): Don't pipeline loops
    that have no exit edges.

testsuite:
    * g++.dg/opt/pr48273.C: New.


Added:
    trunk/gcc/testsuite/g++.dg/opt/pr48273.C
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/cfgloop.h
    trunk/gcc/sel-sched-ir.c
    trunk/gcc/sel-sched.c
    trunk/gcc/testsuite/ChangeLog

[Bug target/48273] [4.6 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

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

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

[Bug target/48273] [4.6 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.6.1                       |4.6.2

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-06-27 12:33:17 UTC ---
GCC 4.6.1 is being released.

[Bug target/48273] [4.6 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[amonakov at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

--- Comment #10 from Alexander Monakov <amonakov at gcc dot gnu.org> 2011-06-28 12:19:23 UTC ---
Author: amonakov
Date: Tue Jun 28 12:19:18 2011
New Revision: 175581

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=175581
Log:
    Backport from mainline
    2011-04-08  Alexander Monakov  <amonakov@ispras.ru>

    PR target/48273
    * cfgloop.h (loop_has_exit_edges): New helper.
    * sel-sched-ir.c (init_global_and_expr_for_insn): Make CALLs
    non-clonable.
    * sel-sched.c (sel_setup_region_sched_flags): Don't pipeline loops
    that have no exit edges.

    * g++.dg/opt/pr48273.C: New.


Added:
    branches/gcc-4_6-branch/gcc/testsuite/g++.dg/opt/pr48273.C
Modified:
    branches/gcc-4_6-branch/gcc/ChangeLog
    branches/gcc-4_6-branch/gcc/cfgloop.h
    branches/gcc-4_6-branch/gcc/sel-sched-ir.c
    branches/gcc-4_6-branch/gcc/sel-sched.c
    branches/gcc-4_6-branch/gcc/testsuite/ChangeLog

[Bug target/48273] [4.6 Regression] ICE: in create_copy_of_insn_rtx, at sel-sched-ir.c:5604 with -fsel-sched-pipelining -fselective-scheduling2 -march=core2

[amonakov at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48273

Alexander Monakov <amonakov at gcc dot gnu.org> changed:

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

--- Comment #11 from Alexander Monakov <amonakov at gcc dot gnu.org> 2011-06-28 12:21:27 UTC ---
Fixed.