MaskRay added a comment. In D148827#4422498 <https://reviews.llvm.org/D148827#4422498>, @glandium wrote:
> In D148827#4379764 <https://reviews.llvm.org/D148827#4379764>, @MaskRay wrote: > >> These cases are UB and should be caught. It's not an excuse that they use C. > > Are they really though? They are. > struct A { > int a; > }; > > int foo(struct A *a) { return 42; } > > int bar(void *a) { return foo(a); } > > int main(void) { > struct A a; > bar(&a); > int (*qux)(void *) = (int (*)(void *))foo; > qux(&a); // If this is UB, why isn't the call to foo from bar? > return 0; > } > > Likewise for `int foo()` and `int bar(void)`/`int (*qux)(void)`. > Likewise for `struct A* foo(void)` and `void *bar(void)`/`void *(*qux)(void)` > (surprisingly, clang doesn't emit this error for `struct A* foo()` and `void* > bar()`/`void*(*qux)()`) C11 6.5.2.2p9 (as mentioned in the patch summary) > If the function is defined with a type that is not compatible with the type > (of the expression) pointed to by the expression that denotes the called > function, the behavior is undefined. `void *` and `struct A*` are not compatible per 6.7..6.1p2 > For two pointer types to be compatible, both shall be identically qualified > and both shall be pointers to compatible types. For old-style (parameter-less) definitions (warned by `-Wstrict-prototypes` unless `-std=c2x`), we have the following behaviors: int foo() { return 42; } // old-style (parameter-less) definition, warned by -Wstrict-prototypes unless -std=c2x int bar(int a) { return 42; } int main(void) { //int (*qux)(int) = (int (*)(int))foo; //qux(42); // rejected by -fsanitize=function int (*qux)() = (int (*)())foo; qux(); // not rejected by -fsanitize=function return 0; } If you enable `-fsanitize=undefined` for some C proijects with the UB, you can consider suppressing them with `ubsan_ignorelist.txt` or enabling `-fno-sanitize=function` for the C files. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D148827/new/ https://reviews.llvm.org/D148827 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
