AMDGPU: Stop emitting an error on illegal addrspacecasts (llvm#127487) (llvm#127751)

These cannot be static compile errors, and should be treated as
poison. Invalid casts may be introduced which are dynamically dead.

For example:

```
  void foo(volatile generic int* x) {
    __builtin_assume(is_shared(x));
    *x = 4;
  }

  void bar() {
    private int y;
    foo(&y); // violation, wrong address space
  }
```

This could produce a compile time backend error or not depending on
the optimization level. Similarly, the new test demonstrates a failure
on a lowered atomicrmw which required inserting runtime address
space checks. The invalid cases are dynamically dead, we should not
error, and the AtomicExpand pass shouldn't have to consider the details
of the incoming pointer to produce valid IR.

This should go to the release branch. This fixes broken -O0 compiles
with 64-bit atomics which would have started failing in
1d03708.

(cherry picked from commit 18ea6c9)

Author: arsenm

llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp

@@ -2426,11 +2426,8 @@ bool AMDGPULegalizerInfo::legalizeAddrSpaceCast(
     return true;
   }
 
-  DiagnosticInfoUnsupported InvalidAddrSpaceCast(
-      MF.getFunction(), "invalid addrspacecast", B.getDebugLoc());
-
-  LLVMContext &Ctx = MF.getFunction().getContext();
-  Ctx.diagnose(InvalidAddrSpaceCast);
+  // Invalid casts are poison.
+  // TODO: Should return poison
   B.buildUndef(Dst);
   MI.eraseFromParent();
   return true;

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

@@ -7340,11 +7340,8 @@ SDValue SITargetLowering::lowerADDRSPACECAST(SDValue Op,
 
   // global <-> flat are no-ops and never emitted.
 
-  const MachineFunction &MF = DAG.getMachineFunction();
-  DiagnosticInfoUnsupported InvalidAddrSpaceCast(
-      MF.getFunction(), "invalid addrspacecast", SL.getDebugLoc());
-  DAG.getContext()->diagnose(InvalidAddrSpaceCast);
-
+  // Invalid casts are poison.
+  // TODO: Should return poison
   return DAG.getUNDEF(Op->getValueType(0));
 }