[PATCH] D91356: AMDGPU: Factor out large flat offset splitting

[Jay Foad via Phabricator via llvm-commits]

foad added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIInstrInfo.cpp:6988
+    int64_t D = 1LL << (NumBits - 1);
+    RemainderOffset = (static_cast<int64_t>(COffsetVal) / D) * D;
+    ImmField = COffsetVal - RemainderOffset;
----------------
Don't need this cast.


================
Comment at: llvm/lib/Target/AMDGPU/SIInstrInfo.cpp:6990
+    ImmField = COffsetVal - RemainderOffset;
+  } else if (static_cast<int64_t>(COffsetVal) >= 0) {
+    ImmField = COffsetVal & maskTrailingOnes<uint64_t>(NumBits);
----------------
Or this one.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D91356/new/

https://reviews.llvm.org/D91356