[Bug rtl-optimization/55153] New: [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[Bug rtl-optimization/55153] New: [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

             Bug #: 55153
           Summary: [4.8 Regression] ICE: in begin_move_insn, at
                    sched-ebb.c:205 with -fsched2-use-superblocks and
                    __builtin_prefetch
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: zsojka@seznam.cz


Created attachment 28583
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28583
reduced testcase

Compiler output:
$ gcc -O -fsched2-use-superblocks -fschedule-insns2 testcase.c         
testcase.c: In function 'foo':
testcase.c:7:1: internal compiler error: in begin_move_insn, at sched-ebb.c:205
 }
 ^
0x110a092 begin_move_insn
        /mnt/svn/gcc-trunk/gcc/sched-ebb.c:205
0x10cc59c commit_schedule
        /mnt/svn/gcc-trunk/gcc/haifa-sched.c:5647
0x10cc59c schedule_block(basic_block_def**, void*)
        /mnt/svn/gcc-trunk/gcc/haifa-sched.c:6429
0x110a4f3 schedule_ebb(rtx_def*, rtx_def*, bool)
        /mnt/svn/gcc-trunk/gcc/sched-ebb.c:547
0x110b1f4 schedule_ebbs()
        /mnt/svn/gcc-trunk/gcc/sched-ebb.c:666
0x9b414c rest_of_handle_sched2
        /mnt/svn/gcc-trunk/gcc/sched-rgn.c:3565
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[mpolacek at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-10-31
                 CC|                            |mpolacek at gcc dot gnu.org
     Ever Confirmed|0                           |1

--- Comment #1 from Marek Polacek <mpolacek at gcc dot gnu.org> 2012-10-31 21:59:38 UTC ---
Started with http://gcc.gnu.org/viewcvs?view=revision&revision=192719

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

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> 2012-11-07 11:28:13 UTC ---
I've tried to "fix" this similarly to PR48721:
--- sched-deps.c.jj2    2012-10-17 17:18:21.000000000 +0200
+++ sched-deps.c    2012-11-07 12:15:05.246124552 +0100
@@ -2907,7 +2907,10 @@ sched_analyze_insn (struct deps_desc *de
     {
       rtx next;
       next = next_nonnote_nondebug_insn (insn);
-      if (next && BARRIER_P (next))
+      /* Don't schedule anything after an unconditional return jump.  */
+      if (returnjump_p (insn) && ! condjump_p (insn))
+    reg_pending_barrier = TRUE_BARRIER;
+      else if (next && BARRIER_P (next))
     reg_pending_barrier = MOVE_BARRIER;
       else
     {
but that didn't help.  The difference between reload and LRA seems to be just
in that old reload emitted a deleted note after the prefetch for some reason,
and after adding simple_return jump and barrier after it that note was still
following the barrier, while with LRA there is no such deleted note at the end.
Still the scheduler seems to attempt to do DCE of the prefetch (why?  Isn't the
prefetch meaningful on its own?).  Vlad, can you please have a look?  Thanks.

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

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

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

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[vmakarov at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

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> 2013-01-14 19:44:36 UTC ---
(In reply to comment #2)
>  Vlad, can you please have a look?  Thanks.
Ok, I started to work on this.

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-01-15 21:51:36 UTC ---
Thanks for working on this.  Unfortunately it seems the
http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=195211
change (commit message didn't make it into this PR because of an extra pr after
/ ) regressed i386/pr45352.c testcase on i?86.
./cc1 -m32 -O3 -march=amdfam10 -fselective-scheduling2 -fsel-sched-pipelining
-funroll-all-loops gcc.target/i386/pr45352.c
ICEs now with:
internal compiler error: in add_insn_mem_dependence, at sched-deps.c:1717

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-01-16 11:31:51 UTC ---
Author: vmakarov
Date: Tue Jan 15 16:47:36 2013
New Revision: 195211

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=195211
Log:
2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>

    PR rtl-optimization/pr55153
    * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.

2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>

    PR rtl-optimization/pr55153
    * gcc.dg/pr55153.c: New.


Added:
    trunk/gcc/testsuite/gcc.dg/pr55153.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/sched-deps.c
    trunk/gcc/testsuite/ChangeLog

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

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

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

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-01-16 15:41:46 UTC ---
Fixed, the regression caused by the fix is tracked now as PR56005.

[Bug rtl-optimization/55153] [4.8 Regression] ICE: in begin_move_insn, at sched-ebb.c:205 with -fsched2-use-superblocks and __builtin_prefetch

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=55153

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 43056 has been marked as a duplicate of this bug. ***