[PATCH] D120265: AMDGPU: Use the implicit kernargs for code object version 5

[Changpeng Fang via Phabricator via llvm-commits]

cfang marked 4 inline comments as done.
cfang added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:1839
+    // Pointer address
+    B.materializePtrAdd(LoadAddr, KernargPtrReg, LLT::scalar(64), Offset);
+    // Load address
----------------
arsenm wrote:
> Use buildPtrAdd (not sure why we still have materializePtrAdd)
Use buildPtrAdd ! Thanks.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:4850
+  MachineFunction &MF = B.getMF();
+  const GCNSubtarget &ST = MF.getSubtarget<GCNSubtarget>();
+  const LLT S64 = LLT::scalar(64);
----------------
arsenm wrote:
> ST is available in the class
Remove it. Thanks.


================
Comment at: llvm/lib/Target/AMDGPU/SIDefines.h:784
+// Implicit kernel argument offset for code object version 5.
+enum ImplicitKernargOffset : unsigned {
+  HOSTCALL_PTR_OFFSET = 80,
----------------
arsenm wrote:
> Add a COV5 suffix? Probably should also wrap in a namespace
Add COV5 suffix, and wrap in a namespace of ImplicitArg. Rename the type to
Offset. So it is of AMDGPU::ImplicitArg::Offset type.  


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D120265/new/

https://reviews.llvm.org/D120265