[llvm] [AMDGPU] Handle memcpy()-like ops in LowerBufferFatPointers (PR #126621)

[Krzysztof Drewniak via llvm-commits]

================
@@ -597,6 +622,47 @@ bool StoreFatPtrsAsIntsVisitor::visitStoreInst(StoreInst &SI) {
   return true;
 }
 
+bool StoreFatPtrsAsIntsAndExpandMemcpyVisitor::visitMemCpyInst(
+    MemCpyInst &MCI) {
+  // TODO: Allow memcpy.p7.p3 as a synonym for the direct-to-LDS copy, which'll
+  // need loop expansion here.
+  if (MCI.getSourceAddressSpace() != AMDGPUAS::BUFFER_FAT_POINTER &&
+      MCI.getDestAddressSpace() != AMDGPUAS::BUFFER_FAT_POINTER)
+    return false;
+  llvm::expandMemCpyAsLoop(&MCI,
+                           TM->getTargetTransformInfo(*MCI.getFunction()));
----------------
krzysz00 wrote:

I mean, as mentioned in the PR, there's a special case for buffer -> LDS that's future work, but, other than that ... because ptr addrspace(8) can't carry an offset and because ptr addrspace(7) has to disappear before MIR formation (because SelectionDAG, especially because SelectionDAG doesn't have ptradd), this has to be rewritten away here

https://github.com/llvm/llvm-project/pull/126621