[Bug c/97381] New: ice error: invalid types in nop conversion

[Bug c/97381] New: ice error: invalid types in nop conversion

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

            Bug ID: 97381
           Summary: ice error: invalid types in nop conversion
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dcb314 at hotmail dot com
  Target Milestone: ---

Created attachment 49352
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49352&action=edit
C source code

The attached C code seems to ice on flag -O2.

$ ../results/bin/gcc -c -w -O2 bug654.c
destDir/testFile.289.c: In function ‘func_10’:
destDir/testFile.289.c:11479:1: error: invalid types in nop conversion
unsigned char
<<< error >>>
c_7.5374_1690 = (unsigned char) _11366;
destDir/testFile.289.c:11479:1: error: invalid types in nop conversion
unsigned char
<<< error >>>
c_7.5389_1728 = (unsigned char) _11366;
during GIMPLE pass: evrp
destDir/testFile.289.c:11479:1: internal compiler error: verify_gimple failed
0xcf360f verify_gimple_in_cfg(function*, bool)
        ../../trunk.git/gcc/tree-cfg.c:5482

I'll have my usual go at reducing the code and finding a range
for the error.

[Bug c/97381] ice error: invalid types in nop conversion

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

--- Comment #1 from David Binderman <dcb314 at hotmail dot com> ---
The problem first seems to occur sometime between 20201006 and 20201007.

[Bug c/97381] ice error: invalid types in nop conversion

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

--- Comment #2 from David Binderman <dcb314 at hotmail dot com> ---
Reduced C code is:

int a;
void b() {
  char c = 27;
  for (; c <= 85; c += 1) {
    a /= 148372120 * c;
    if (a)
      for (;;)
        ;
  }
}

[Bug tree-optimization/97381] [11 Regression] ice error: invalid types in nop conversion

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2020-10-12
                 CC|                            |aldyh at gcc dot gnu.org
   Target Milestone|---                         |11.0
            Summary|ice error: invalid types in |[11 Regression] ice error:
                   |nop conversion              |invalid types in nop
                   |                            |conversion
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW
          Component|c                           |tree-optimization

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
EVRP

[Bug tree-optimization/97381] [11 Regression] ice error: invalid types in nop conversion

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

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

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

[Bug tree-optimization/97381] [11 Regression] ice error: invalid types in nop conversion

[aldyh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

--- Comment #4 from Aldy Hernandez <aldyh at gcc dot gnu.org> ---
(In reply to David Binderman from comment #2)
> Reduced C code is:
> 
> int a;
> void b() {
>   char c = 27;
>   for (; c <= 85; c += 1) {
>     a /= 148372120 * c;
>     if (a)
>       for (;;)
>         ;
>   }
> }

This looks similar to PR97378.

There is no definition of _8 but there is still a dangling use left:

  c.2_6 = (unsigned char) _8;

[Bug tree-optimization/97381] [11 Regression] ice error: invalid types in nop conversion

[aldyh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381
Bug 97381 depends on bug 97378, which changed state.

Bug 97378 Summary: [11 Regression] ICE in tree check: expected class ‘type’, have ‘exceptional’ (error_mark) in useless_type_conversion_p, at gimple-expr.c:87 since r11-3685-gfcae5121154d1c33
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97378

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

[Bug tree-optimization/97381] [11 Regression] ice error: invalid types in nop conversion

[aldyh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

--- Comment #5 from Aldy Hernandez <aldyh at gcc dot gnu.org> ---
Created attachment 49354
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49354&action=edit
propsed patch in testing

[Bug tree-optimization/97381] [11 Regression] ice error: invalid types in nop conversion

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

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

https://gcc.gnu.org/g:46f4a397ab0907ce5786e1f0b92030a73e8b89ef

commit r11-3836-g46f4a397ab0907ce5786e1f0b92030a73e8b89ef
Author: Andrew MacLeod <amacleod@redhat.com>
Date:   Mon Oct 12 13:00:12 2020 -0400

    Do not ignore failures from gimple_range_calc_op2.

    We were ignoring the return value if op2 returned false and getting garbage
ranges propagated.

    gcc/ChangeLog:

            PR tree-optimization/97381
            * gimple-range-gori.cc (gori_compute::compute_operand2_range): If a
range cannot be
            calculated through operand 2, return false.

    gcc/testsuite/ChangeLog:

            * gcc.dg/pr97381.c: New test.

[Bug tree-optimization/97381] [11 Regression] ice error: invalid types in nop conversion

[aldyh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

Aldy Hernandez <aldyh at gcc dot gnu.org> changed:

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

--- Comment #7 from Aldy Hernandez <aldyh at gcc dot gnu.org> ---
fixed

[Bug tree-optimization/97381] [11 Regression] ice error: invalid types in nop conversion

[amacleod at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97381

--- Comment #8 from Andrew Macleod <amacleod at redhat dot com> ---
in particular:

  _2 = (int) c_8;
  _3 = _2 * 148372120;
  _4 = a.0_1 / _3;
  if (_4 != 0)


we do not wind back thru divides at the moment (unless they are exact divides)
so when processing the false edge to calculate a range for c_8 (which feeds
into a phi eventually)

so when processing 
  [0, 0] = a.0_1 / _3

and trying to calculate a range for _3 , range-ops was returning false,
indicating no further processing should be done (so we wont know anything else
about c_8)

By ignoring that, and continuing to process
  _3 = _2 * 148372120;
with a LHS that was garbage, we were getting some less than ideal results as we
continued back to the cast and figured a value for c_8.