[PATCH] D73558: AMDGPU: Correct memory size for image intrinsics

[Nicolai Hähnle via Phabricator via llvm-commits]

nhaehnle added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:916-922
+      // TODO: Account for dmask reducing loaded size.
+      if (auto *ST = dyn_cast<StructType>(CI.getType())) {
         // Some intrinsics return an aggregate type - special case to work out
         // the correct memVT
-        Info.memVT = memVTFromAggregate(CI.getType());
-      }
+        Info.memVT = memVTFromImageReturn(*ST);
+      } else
+        Info.memVT = MVT::getVT(CI.getType(), true);
----------------
I think I'd prefer to have all of this logic inside of memVTFromImageReturn. That would make the function name fit better, since with this change it's really more "memVTFromImageReturnIfItsAStruct"...


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D73558/new/

https://reviews.llvm.org/D73558