[Bug c/93894] -Wimplicit-fallthrough false warning with operator %

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93894

Jakub Jelinek  changed:

   What|Removed |Added

 CC||jakub at gcc dot gnu.org

--- Comment #4 from Jakub Jelinek  ---
Yeah, -Wimplicit-fallthrough is intentionally an early warning, so that it
isn't dependent on optimizations etc., and you are asking that it depends on
VRP or some poor man's VRP.  There will always be ways to obfuscate code so
that the compiler doesn't understand something can't fall through.

[Bug c/93894] -Wimplicit-fallthrough false warning with operator %

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93894

--- Comment #3 from Andrew Pinski  ---
The warning is not sensitive to what is being switched on.  That is the inner
most switch is considered as falling through as the switch is not checked for
the values it will be switched on.

[Bug c/93894] -Wimplicit-fallthrough false warning with operator %

[dimhen at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93894

--- Comment #2 from Dmitry G. Dyachenko  ---
`unsigned' changes nothing

$ cat xxu.i 
int f1(int j, unsigned k)
{
switch (j) {
case 0:
switch (k % 2) {
case 0:
return 0;
case 1:
return 1;
}
// return 3; // uncomment to fix warning
default:
return 2;
}
}

$ gcc -fpreprocessed -Wimplicit-fallthrough=2 -Og -c xxu.i 
xx.i: In function ‘f1’:
xx.i:5:2: warning: this statement may fall through [-Wimplicit-fallthrough=]
5 |  switch (k % 2) {
  |  ^~
xx.i:12:5: note: here
   12 | default:
  | ^~~

[Bug c/93894] -Wimplicit-fallthrough false warning with operator %

[sch...@linux-m68k.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93894

--- Comment #1 from Andreas Schwab  ---
case -1 is missing.