[Bug c/107937] New: ice in find_var_cmp_const, at gimple-predicate-analysis.cc:257

[Bug c/107937] New: ice in find_var_cmp_const, at gimple-predicate-analysis.cc:257

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

            Bug ID: 107937
           Summary: ice in find_var_cmp_const, at
                    gimple-predicate-analysis.cc:257
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dcb314 at hotmail dot com
  Target Milestone: ---

This reduced C code:

double getk_intersect_line_with_ycylinder_r,
    getk_intersect_line_with_ycylinder_a;
getk_intersect_line_with_ycylinder_i;
getk_intersect_line_with_ycylinder(double *k2) {
  if (getk_intersect_line_with_ycylinder_r < 0.)
    return sqrt(getk_intersect_line_with_ycylinder_r);
  *k2 = 0 / getk_intersect_line_with_ycylinder_a;
  {
    double k2;
    for (; getk_intersect_line_with_ycylinder_i;
         getk_intersect_line_with_ycylinder_i++)
      getk_intersect_line_with_ycylinder(&k2);
    vecscale(k2);
  }
}

when compiled by recent gcc trunk and compiler flags -O3 -Wall, does this:

during GIMPLE pass: uninit
bug865.c:4:1: internal compiler error: in find_var_cmp_const, at
gimple-predicate-analysis.cc:257
    4 | getk_intersect_line_with_ycylinder(double *k2) {
      | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0x1b3ad6c find_var_cmp_const(vec<vec<pred_info, va_heap, vl_ptr>, va_heap,
vl_ptr>, gphi*, gimple**, tree_node**)
        ../../trunk.d1/gcc/gimple-predicate-analysis.cc:257
0x1b3ad6c uninit_analysis::overlap(gphi*, unsigned int, hash_set<gphi*, false,
default_hash_traits<gphi*> >*, predicate const&)
        ../../trunk.d1/gcc/gimple-predicate-analysis.cc:648

I'll have a go at finding a git range for the problem.

[Bug tree-optimization/107937] [13 Regression] ice in find_var_cmp_const, at gimple-predicate-analysis.cc:257

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |tree-optimization
   Target Milestone|---                         |13.0
           Keywords|                            |diagnostic,
                   |                            |ice-on-valid-code
            Summary|ice in find_var_cmp_const,  |[13 Regression] ice in
                   |at                          |find_var_cmp_const, at
                   |gimple-predicate-analysis.c |gimple-predicate-analysis.c
                   |c:257                       |c:257

[Bug tree-optimization/107937] [13 Regression] ice in find_var_cmp_const, at gimple-predicate-analysis.cc:257

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
It is working at r13-4401-gf57ff189572575 ....

[Bug tree-optimization/107937] [13 Regression] ice in find_var_cmp_const, at gimple-predicate-analysis.cc:257

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
It might be r13-4408, r13-4407, or r13-4406 .

[Bug tree-optimization/107937] [13 Regression] ice in find_var_cmp_const, at gimple-predicate-analysis.cc:257

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

--- Comment #3 from David Binderman <dcb314 at hotmail dot com> ---
The bug first seems to appear sometime between git hash 4d08c674b0114622
from yesterday and d0a3d55ae4a2656f, from today.

[Bug tree-optimization/107937] [13 Regression] ice in find_var_cmp_const, at gimple-predicate-analysis.cc:257

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to David Binderman from comment #3)
> The bug first seems to appear sometime between git hash 4d08c674b0114622
> from yesterday and d0a3d55ae4a2656f, from today.

Which is in between r13-4382-g4d08c674b01146 and r13-4412-gd0a3d55ae4a265 .
And I know it was working at r13-4401 so that narrows it down a little bit.
r13-4406, r13-4407 and r13-4408 modified the area where the ICE is coming from
which is why I pointed them out.

[Bug tree-optimization/107937] [13 Regression] ICE in find_var_cmp_const, at gimple-predicate-analysis.cc:257 since r13-4408-gb628cad9e093f7a3

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2022-12-01
                 CC|                            |marxin at gcc dot gnu.org,
                   |                            |rguenth at gcc dot gnu.org
            Summary|[13 Regression] ice in      |[13 Regression] ICE in
                   |find_var_cmp_const, at      |find_var_cmp_const, at
                   |gimple-predicate-analysis.c |gimple-predicate-analysis.c
                   |c:257                       |c:257 since
                   |                            |r13-4408-gb628cad9e093f7a3
     Ever confirmed|0                           |1

--- Comment #5 from Martin Liška <marxin at gcc dot gnu.org> ---
Started with r13-4408-gb628cad9e093f7a3.

[Bug tree-optimization/107937] [13 Regression] ICE in find_var_cmp_const, at gimple-predicate-analysis.cc:257 since r13-4408-gb628cad9e093f7a3

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

--- Comment #6 from Martin Liška <marxin at gcc dot gnu.org> ---
Another test-case:

$ cat warn.i
int regs, vm_debug_engine_vp_0, vm_debug_engine_vp_2;

void
vm_dispatch_hook();


void
vm_debug_engine() {
  int fp;
  void *jump_table = &&l_nop;
l_nop:
  if (__builtin_expect(vm_debug_engine_vp_2, 0))
    vm_dispatch_hook();
  if (_setjmp(regs)) {
    fp = fp;
    vm_dispatch_hook();
    goto *jump_table;
  }
  vm_debug_engine_vp_0 = fp;
}

$ gcc -c warn.i -O1 -Wall
warn.i: In function ‘vm_debug_engine’:
warn.i:14:7: warning: implicit declaration of function ‘_setjmp’
[-Wimplicit-function-declaration]
   14 |   if (_setjmp(regs)) {
      |       ^~~~~~~
during GIMPLE pass: uninit
warn.i:8:1: internal compiler error: in find_var_cmp_const, at
gimple-predicate-analysis.cc:257
    8 | vm_debug_engine() {
      | ^~~~~~~~~~~~~~~
0x84abb6 find_var_cmp_const
        /home/marxin/Programming/gcc/gcc/gimple-predicate-analysis.cc:257
0x1c83ff2 uninit_analysis::overlap(gphi*, unsigned int, hash_set<gphi*, false,
default_hash_traits<gphi*> >*, predicate const&)
        /home/marxin/Programming/gcc/gcc/gimple-predicate-analysis.cc:648
0x1c83b45 uninit_analysis::is_use_guarded(gimple*, basic_block_def*, gphi*,
unsigned int, hash_set<gphi*, false, default_hash_traits<gphi*> >*)
        /home/marxin/Programming/gcc/gcc/gimple-predicate-analysis.cc:2215
0x1c840ea uninit_analysis::is_use_guarded(gimple*, basic_block_def*, gphi*,
unsigned int)
        /home/marxin/Programming/gcc/gcc/gimple-predicate-analysis.cc:2249
0x11188e7 find_uninit_use
        /home/marxin/Programming/gcc/gcc/tree-ssa-uninit.cc:1234
0x111923e warn_uninitialized_phi
        /home/marxin/Programming/gcc/gcc/tree-ssa-uninit.cc:1304
0x111923e execute_late_warn_uninitialized
        /home/marxin/Programming/gcc/gcc/tree-ssa-uninit.cc:1425
0x111923e execute
        /home/marxin/Programming/gcc/gcc/tree-ssa-uninit.cc:1442
Please submit a full bug report, with preprocessed source (by using
-freport-bug).
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.

[Bug tree-optimization/107937] [13 Regression] ICE in find_var_cmp_const, at gimple-predicate-analysis.cc:257 since r13-4408-gb628cad9e093f7a3

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

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

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

--- Comment #7 from Richard Biener <rguenth at gcc dot gnu.org> ---
Mine.

[Bug tree-optimization/107937] [13 Regression] ICE in find_var_cmp_const, at gimple-predicate-analysis.cc:257 since r13-4408-gb628cad9e093f7a3

[zhendong.su at inf dot ethz.ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

Zhendong Su <zhendong.su at inf dot ethz.ch> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |zhendong.su at inf dot ethz.ch

--- Comment #8 from Zhendong Su <zhendong.su at inf dot ethz.ch> ---
Another test:

[546] % gcctk -O3 -Wall -w small.c
during GIMPLE pass: uninit
small.c: In function ‘main’:
small.c:3:5: internal compiler error: in find_var_cmp_const, at
gimple-predicate-analysis.cc:257
    3 | int main() {
      |     ^~~~
0x818725 find_var_cmp_const
        ../../gcc-trunk/gcc/gimple-predicate-analysis.cc:257
0x1c9bfbf uninit_analysis::overlap(gphi*, unsigned int, hash_set<gphi*, false,
default_hash_traits<gphi*> >*, predicate const&)
        ../../gcc-trunk/gcc/gimple-predicate-analysis.cc:648
0x1c9bafa uninit_analysis::is_use_guarded(gimple*, basic_block_def*, gphi*,
unsigned int, hash_set<gphi*, false, default_hash_traits<gphi*> >*)
        ../../gcc-trunk/gcc/gimple-predicate-analysis.cc:2215
0x1c9c0ec uninit_analysis::is_use_guarded(gimple*, basic_block_def*, gphi*,
unsigned int)
        ../../gcc-trunk/gcc/gimple-predicate-analysis.cc:2249
0x110d500 find_uninit_use
        ../../gcc-trunk/gcc/tree-ssa-uninit.cc:1234
0x110de96 warn_uninitialized_phi
        ../../gcc-trunk/gcc/tree-ssa-uninit.cc:1304
0x110de96 execute_late_warn_uninitialized
        ../../gcc-trunk/gcc/tree-ssa-uninit.cc:1425
0x110de96 execute
        ../../gcc-trunk/gcc/tree-ssa-uninit.cc:1442
Please submit a full bug report, with preprocessed source (by using
-freport-bug).
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.
[547] % cat small.c
int printf(const char *, ...);
int a, b;
int main() {
  int c, d;
  while (b) {
    if (a) {
      &d;
      c = ~c;
    }
    printf("0");
    while (b && d && c)
      ;
  }
  return 0;
}

[Bug tree-optimization/107937] [13 Regression] ICE in find_var_cmp_const, at gimple-predicate-analysis.cc:257 since r13-4408-gb628cad9e093f7a3

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

--- Comment #9 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:abf05583dbc86a6667b63f5bda6ba57fe55a1b25

commit r13-4437-gabf05583dbc86a6667b63f5bda6ba57fe55a1b25
Author: Richard Biener <rguenther@suse.de>
Date:   Thu Dec 1 10:12:28 2022 +0100

    tree-optimization/107937 - uninit predicate simplification fixup

    The following changes the predicate representation to record the
    value of a predicate with an empty set of AND predicates.  That's
    necessary to properly represent the conservative fallback for the
    def vs use predicates.  Since simplification now can result in
    such an empty set this distinction becomes important and we need
    to check for this as we otherwise ICE.

            PR tree-optimization/107937
            * gimple-predicate-analysis.h (predicate::is_true): New.
            (predicate::is_false): Likewise.
            (predicate::empty_val): Likewise.
            (uninit_analysis::uninit_analysis): Properly initialize
            def_preds.
            * gimple-predicate-analysis.cc (simplify_1b): Indicate
            whether the chain became empty.
            (predicate::simplify): Release emptied chain before removing it.
            (predicate::normalize): Replace temporary object with assertion.
            (uninit_analysis::is_use_guarded): Deal with predicates
            that simplify to true/false.

            * gcc.dg/pr107937.c: New testcase.

[Bug tree-optimization/107937] [13 Regression] ICE in find_var_cmp_const, at gimple-predicate-analysis.cc:257 since r13-4408-gb628cad9e093f7a3

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107937

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

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

--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> ---
So we have

        ((getk_intersect_line_with_ycylinder_r.0_19 < 0.0) AND
(getk_intersect_line_with_ycylinder_r.0_19 u>= 0.0) AND (NOT (_180 != 0)))

from

<bb 7> [local count: 711980952]:
# k2_184 = PHI <k2_20(74), k2_220(D)(81)>
# prephitmp_96 = PHI <_182(74), getk_intersect_line_with_ycylinder_i.4_30(81)>
getk_intersect_line_with_ycylinder_r.0_19 =
getk_intersect_line_with_ycylinder_r;
if (getk_intersect_line_with_ycylinder_r.0_19 < 0.0)
  goto <bb 8>; [14.99%]
else
  goto <bb 11>; [85.01%]

<bb 8> [local count: 106727384]:
if (getk_intersect_line_with_ycylinder_r.0_19 u>= 0.0)
  goto <bb 9>; [99.95%]
else
  goto <bb 10>; [0.05%]

and we simplify that to true which is bogus.  It's correct to remove
this false sub-predicate from the (a || b || c ...) chain but when
it's the last the result is false, not true.

Fixed.