[Bug c/113973] New: Pleas issue a warning when using plain character values in bitwise operations

[Bug c/113973] New: Pleas issue a warning when using plain character values in bitwise operations

[gcc-bugzilla at mkarcher dot dialup.fu-berlin.de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113973

            Bug ID: 113973
           Summary: Pleas issue a warning when using plain character
                    values in bitwise operations
           Product: gcc
           Version: 13.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: gcc-bugzilla at mkarcher dot dialup.fu-berlin.de
  Target Milestone: ---

This example program compiles without any kind of warning in gcc:

static char x = 0xD8;

int main(void)
{
    return 0x1200 | x;
}

The value returned from main is 0xFFFFFFD8 on architectures with 32-bit int and
signed characters by default. After just fixing a bug that was caused by an
unexpected sign expansion when building an int from individual bytes, I'd
rather have a warning if

1) A variable of type char is promoted to int.
2) The int value is used in an bitwise expression
3) More than 8 bits of the results are actually used
4) More than 8 bits may be non-zero

Because of condition 3, this will yiels no warning on "char y = x | 0x40;" (top
bits truncated, so condition 3 fails) and no warning on "int y = x & 0x40;"
(all high bits are guaranteed to be zero, so condition 4 fails).

The real-world bug that motivates this enhancement proposal is
https://github.com/hfst/hfst-ospell/issues/43

[Bug c/113973] Pleas issue a warning when using plain character values in bitwise operations

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113973

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2024-02-19
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed, this looks useful (but for arithmetic as well, not just bitwise
ops).

static char x = 0xD8;

int main(void)
{
    return 0x1200 + x;
}