[clang] [AMDGPU] Mark int64_t and intmax_t as LP64 (PR #214865)

[Matt Arsenault via cfe-commits]

================
@@ -10,7 +10,7 @@ void test_atomic_inc32_av(volatile unsigned *p, unsigned val) {
   [[clang::amdgpu_av("none")]] __builtin_amdgcn_atomic_inc32(p, val, __ATOMIC_SEQ_CST, "agent"); // expected-warning {{only applies to atomic expressions or Clang atomic builtins}}
 }
 
-void test_atomic_dec64_av(volatile unsigned long long *p, unsigned long long val) {
----------------
arsenm wrote:

is long long i128 now?

https://github.com/llvm/llvm-project/pull/214865