[clang] [Clang] Add __builtin_counted_by_ref builtin (PR #114495)

[via cfe-commits]

================
@@ -13,7 +13,10 @@ void init(void * __attribute__((pass_dynamic_object_size(0))));
 // CHECK-LABEL: define dso_local void @_ZN3foo3barC1Ev(
 // CHECK-SAME: ptr noundef nonnull align 4 dereferenceable(1) [[THIS:%.*]]) unnamed_addr #[[ATTR0:[0-9]+]] align 2 {
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    tail call void @_Z4initPvU25pass_dynamic_object_size0(ptr noundef nonnull align 4 dereferenceable(1) [[THIS]], i64 noundef -1) #[[ATTR2:[0-9]+]]
----------------
Sirraide wrote:

Essentially, what I meant was that there is no reason to ever write a codegen test w/ optimisation enabled—unless the code that you’re introducing has different behaviour depending on optimisation levels (which iirc isn’t the case for `__builtin_counted_by_ref`, right?).

*Other* tests might still require the `-O2`; and if that is the case, then we generally shouldn’t add tests for something that is invariant under optimisation (wrt how the frontend behaves; the IR would be different of course, but then has nothing to do w/ what Clang emits) to a file that contains tests that *do* care about optimisation levels. I thought at first that this is what you did here, but looking at it again, that doesn’t seem to be the case because there *is* in fact a separate file w/ tests for this pr.

So yeah, my bad if I wasn’t too clear about this ealier.

https://github.com/llvm/llvm-project/pull/114495