This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug sanitizer/81598] -fsanitize=enum does not detect range violation
- From: "marxin at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Thu, 03 Aug 2017 11:27:03 +0000
- Subject: [Bug sanitizer/81598] -fsanitize=enum does not detect range violation
- Auto-submitted: auto-generated
- References: <bug-81598-4@http.gcc.gnu.org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81598
--- Comment #8 from Martin Liška <marxin at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #7)
> As for the switch, I think there shouldn't be any runtime error, there is no
> UB.
> You can have default: label even if you list all the possible in-range
> cases, you can have enum values in between min/max that aren't in the
> enumeration, and you can have case labels outside of the enum range too,
> that is something for a warning, but not a runtime error.
I've just noticed that in this example:
cat enum2.c
int main(int argc, char **argv)
{
char c = (char)argc;
bool x = argc;
switch (c)
{
case -128 ... 127:
break;
default:
return 2;
}
switch (x)
{
case true:
return 2;
case false:
return 3;
default:
return 4;
}
}
We are able to simplify the second switch in vrp:
[...]
Folding statement: switch (x_4) <default: <L5> [33.33%], case 0: <L4> [33.33%],
case 1: <L8> [33.33%]>
removing unreachable case label
[...]
But we can't do it for the first switch. That's probably missing optimization.
I'm interested in the case where we've got call enum values covered and default
is provided: would it be valid to
simply remove the block (via __builtin_unreachable or similarly)?