[llvm] [Clang][inlineasm] Add special support for "rm" output constraints (PR #92040)

[Bill Wendling via llvm-commits]

================
@@ -496,6 +496,8 @@ class TargetPassConfig : public ImmutablePass {
 void registerCodeGenCallback(PassInstrumentationCallbacks &PIC,
                              LLVMTargetMachine &);
 
+bool usesGreedyOrDefaultRegisterAllocator();
----------------
bwendling wrote:

> I thought the result of the issue thread was fast RA would just start spilling everything?

And how do we know that we're in the fast register allocator before we reach the fast register allocator? Also, how does this not work in general?

https://github.com/llvm/llvm-project/pull/92040