NoQ added a comment.
> I feel like the 2. is a better solution. Of course, that change might have a
> performance impact as well.
Yes, i'm all for '2.'. There's no need to make this callback more complicated
than it already is.
As for performance, it's messy and suffers from a deeper problem: the number of
escaped symbols is potentially infinite. The following false positive
illustrates that well:
void invalidate(int **x);
void foo(int **x) {
int *y = *x;
if (*y == 0) {
// **x should be invalidated here!
invalidate(x);
}
// Should not warn about division by zero!
1 / *y;
}
Therefore one does not simply compose a list of escaped symbols. We need
something similar to `SymbolReaper` but for invalidation/escapes. And //then//
we'll talk about performance.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D70470/new/
https://reviews.llvm.org/D70470
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
