[PATCH] D47580: AMDGPU: Preserve metadata when widening loads

Stanislav Mekhanoshin via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu May 31 10:20:27 PDT 2018


rampitec added inline comments.


================
Comment at: lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp:489
+        // Don't make assumptions about the high bits.
+        ConstantAsMetadata::get(ConstantInt::get(I32Ty, 0xffffffff))
+      };
----------------
Range is [a, b). I.e. upper bound is not included in the range. But range can wrap around so 0 should work.
At the same time you will need to check that lower bound is not zero then because it is required that a != b.


https://reviews.llvm.org/D47580





More information about the llvm-commits mailing list