[Bug debug/105559] New: -g and -O3 cause timeout

[Bug debug/105559] New: -g and -O3 cause timeout

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

            Bug ID: 105559
           Summary: -g and -O3 cause timeout
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: debug
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dcb314 at hotmail dot com
  Target Milestone: ---

Created attachment 52952
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52952&action=edit
C source code

For the attached C code, flags -g and -O3 cause a timeout:

$ time /home/dcb/gcc/results/bin/gcc  -w -c -O3  bug809.i

real    0m0.656s
user    0m0.587s
sys     0m0.032s
$ (ulimit -t 300; time /home/dcb/gcc/results/bin/gcc  -w -c -g -O3  bug809.i)
gcc: fatal error: Killed signal terminated program cc1
compilation terminated.

real    5m14.868s
user    4m57.685s
sys     0m2.265s
$ (ulimit -t 300; time /home/dcb/gcc/results/bin/gcc  -w -c -g -O2  bug809.i)

real    0m0.732s
user    0m0.683s
sys     0m0.024s
$ /home/dcb/gcc/results/bin/gcc -v 2>&1 | fgrep exp
gcc version 13.0.0 20220510 (experimental) (60ab3d0f2d9f9504) 
$

[Bug c/105559] -g and -O3 cause timeout

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

David Binderman <dcb314 at hotmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|debug                       |c

--- Comment #1 from David Binderman <dcb314 at hotmail dot com> ---
The problem seems to exist since sometime before git revision 82a4c5c704433249,
dated about a month ago.

[Bug c/105559] [12/13 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2022-05-11
                 CC|                            |marxin at gcc dot gnu.org,
                   |                            |rguenth at gcc dot gnu.org
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
            Summary|-g and -O3 cause timeout    |[12/13 Regression] -g and
                   |                            |-O3 cause timeout since
                   |                            |r12-156-g8d4c374c4419a875

--- Comment #2 from Martin Liška <marxin at gcc dot gnu.org> ---
Started with r12-156-g8d4c374c4419a875.

I see purge_dead_edges in perf top after some time dominating.

[Bug c/105559] [12/13 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
   Target Milestone|---                         |12.2
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
I will investigate but definitely the cited revision isn't the culprit but
maybe the exposer.

[Bug rtl-optimization/105559] [12/13 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |compile-time-hog
          Component|c                           |rtl-optimization

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
stuck in RTL jump pass btw, calling delete_trivially_dead_insns at the start
of its invocation (_not_ during the jump threading iteration).

We have a block with very many DEBUG_INSN at end and delete_insn_and_edges
does

/* Like delete_insn but also purge dead edges from BB.
   Return true if any edges are eliminated.  */

bool
delete_insn_and_edges (rtx_insn *insn)
{
  bool purge = false;

  if (INSN_P (insn) && BLOCK_FOR_INSN (insn))
    {
      basic_block bb = BLOCK_FOR_INSN (insn);
      if (BB_END (bb) == insn)
        purge = true;
      else if (DEBUG_INSN_P (BB_END (bb)))
        for (rtx_insn *dinsn = NEXT_INSN (insn);
             DEBUG_INSN_P (dinsn); dinsn = NEXT_INSN (dinsn))
          if (BB_END (bb) == dinsn)
            {
              purge = true;
              break;
            }
    }
  delete_insn (insn);
  if (purge)
    return purge_dead_edges (BLOCK_FOR_INSN (insn));

that's highly inefficient when insn is a debug insn itself.  Fixing that
alone doesn't seem to be enough to make it "fast" but at least it
finishes in 20s for me then with no obvious peak.

[Bug rtl-optimization/105559] [12/13 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Blocks|                            |94618

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Latent on all active branches btw., caused by r10-7769-gc41884a09206be


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94618
[Bug 94618] [8/9 Regression] '-fcompare-debug' failure (length) with -O2
-fnon-call-exceptions since r8-565-g7581ce9a1ad6df9c

[Bug rtl-optimization/105559] [12/13 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:37a8220fa9188470c677abfef50c1b120c0b6c76

commit r13-323-g37a8220fa9188470c677abfef50c1b120c0b6c76
Author: Richard Biener <rguenther@suse.de>
Date:   Wed May 11 13:34:37 2022 +0200

    rtl-optimization/105559 - avoid quadratic behavior in delete_insn_and_edges

    When the insn to delete is a debug insn there's no point in figuring
    out whether it might be the last real insn and thus we have to purge
    dead edges.

    2022-05-11  Richard Biener  <rguenther@suse.de>

            PR rtl-optimization/105559
            * cfgrtl.cc (delete_insn_and_edges): Only perform search to BB_END
            for non-debug insns.

[Bug rtl-optimization/105559] [9/10/11/12 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |13.0
           Priority|P3                          |P2
            Summary|[12/13 Regression] -g and   |[9/10/11/12 Regression] -g
                   |-O3 cause timeout since     |and -O3 cause timeout since
                   |r12-156-g8d4c374c4419a875   |r12-156-g8d4c374c4419a875
   Target Milestone|12.2                        |9.5

[Bug rtl-optimization/105559] [9/10/11/12 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

--- Comment #7 from David Binderman <dcb314 at hotmail dot com> ---
Thanks for the fast work. Less than 2.5 hours from reported to fixed
for a bug that hadn't been found for a couple of years.

I should also mention that this bug was produced from code produced by csmith, 
a most excellent C source code generator program.

There is a school of thought that says running csmith regularly over the gcc
trunk
will find bugs.

[Bug rtl-optimization/105559] [9/10/11/12 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

--- Comment #8 from rguenther at suse dot de <rguenther at suse dot de> ---
On Wed, 11 May 2022, dcb314 at hotmail dot com wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559
> 
> --- Comment #7 from David Binderman <dcb314 at hotmail dot com> ---
> Thanks for the fast work. Less than 2.5 hours from reported to fixed
> for a bug that hadn't been found for a couple of years.
> 
> I should also mention that this bug was produced from code produced by csmith, 
> a most excellent C source code generator program.
> 
> There is a school of thought that says running csmith regularly over the gcc
> trunk
> will find bugs.

Yep, and that's very much appreciated!

[Bug rtl-optimization/105559] [9/10/11/12 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

--- Comment #9 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Richard Biener
<rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:2a040e982945a5795081e81be35094b604d596d3

commit r12-8399-g2a040e982945a5795081e81be35094b604d596d3
Author: Richard Biener <rguenther@suse.de>
Date:   Wed May 11 13:34:37 2022 +0200

    rtl-optimization/105559 - avoid quadratic behavior in delete_insn_and_edges

    When the insn to delete is a debug insn there's no point in figuring
    out whether it might be the last real insn and thus we have to purge
    dead edges.

    2022-05-11  Richard Biener  <rguenther@suse.de>

            PR rtl-optimization/105559
            * cfgrtl.cc (delete_insn_and_edges): Only perform search to BB_END
            for non-debug insns.

    (cherry picked from commit 37a8220fa9188470c677abfef50c1b120c0b6c76)

[Bug rtl-optimization/105559] [9/10/11 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

--- Comment #10 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Richard Biener
<rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:d293946d5c318549af88f96c4c2175dcbb8f4cf1

commit r11-10021-gd293946d5c318549af88f96c4c2175dcbb8f4cf1
Author: Richard Biener <rguenther@suse.de>
Date:   Wed May 11 13:34:37 2022 +0200

    rtl-optimization/105559 - avoid quadratic behavior in delete_insn_and_edges

    When the insn to delete is a debug insn there's no point in figuring
    out whether it might be the last real insn and thus we have to purge
    dead edges.

    2022-05-11  Richard Biener  <rguenther@suse.de>

            PR rtl-optimization/105559
            * cfgrtl.c (delete_insn_and_edges): Only perform search to BB_END
            for non-debug insns.

    (cherry picked from commit 37a8220fa9188470c677abfef50c1b120c0b6c76)

[Bug rtl-optimization/105559] [10 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|9.5                         |10.4

--- Comment #11 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 9 branch is being closed

[Bug rtl-optimization/105559] [10 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

--- Comment #12 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Richard Biener
<rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:6007449a1c3bd116e431fe53d4dfe2d1c67c1076

commit r10-10763-g6007449a1c3bd116e431fe53d4dfe2d1c67c1076
Author: Richard Biener <rguenther@suse.de>
Date:   Wed May 11 13:34:37 2022 +0200

    rtl-optimization/105559 - avoid quadratic behavior in delete_insn_and_edges

    When the insn to delete is a debug insn there's no point in figuring
    out whether it might be the last real insn and thus we have to purge
    dead edges.

    2022-05-11  Richard Biener  <rguenther@suse.de>

            PR rtl-optimization/105559
            * cfgrtl.c (delete_insn_and_edges): Only perform search to BB_END
            for non-debug insns.

    (cherry picked from commit 37a8220fa9188470c677abfef50c1b120c0b6c76)

[Bug rtl-optimization/105559] [10 Regression] -g and -O3 cause timeout since r12-156-g8d4c374c4419a875

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105559

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |10.3.0
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED
      Known to work|                            |10.3.1

--- Comment #13 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.