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

[Matt Arsenault via cfe-commits]

================
@@ -0,0 +1,60 @@
+// Standalone amdgcn is LP64, so 'long' is used for the 64-bit and maximal
+// integer types.
+//
+// RUN: %clang_cc1 -E -dM -triple=amdgcn-amd-amdhsa < /dev/null \
+// RUN:   | FileCheck -match-full-lines -check-prefix=LP64 %s
+// RUN: %clang_cc1 -x c++ -E -dM -triple=amdgcn-amd-amdhsa < /dev/null \
+// RUN:   | FileCheck -match-full-lines -check-prefix=LP64 %s
+// RUN: %clang_cc1 -E -dM -triple=amdgcn-amd-amdhsa -target-cpu gfx90a < /dev/null \
+// RUN:   | FileCheck -match-full-lines -check-prefix=LP64 %s
----------------
arsenm wrote:

Subarch triples 

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