[PATCH] D94429: AMDGPU: Move handling of allocation of fixed ABI inputs

[Sebastian Neubauer via Phabricator via llvm-commits]

Flakebi added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:1883-1885
+  // If this has a fixed position, we still should allocate the register in the
+  // CCInfo state. Technically we could get away with this for values passed
+  // outside of the normal argument range.
----------------
This comment does make more sense in the function above, where the register is allocated in the CCInfo.


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:1887-1888
+  if (Arg) {
+    allocateFixedSGPRInputImpl(CCInfo, &AMDGPU::SGPR_32RegClass,
+                               Arg.getRegister());
+  } else
----------------
This should assign to Arg probably


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:1895-1896
+  if (Arg) {
+    allocateFixedSGPRInputImpl(CCInfo, &AMDGPU::SGPR_64RegClass,
+                               Arg.getRegister());
+  } else
----------------
This should assign to Arg probably


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D94429/new/

https://reviews.llvm.org/D94429