[PATCH] D59608: [AMDGPU] Fix clamp bit DAG operand

[Matt Arsenault via Phabricator via llvm-commits]

arsenm requested changes to this revision.
arsenm added inline comments.
This revision now requires changes to proceed.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp:1124-1125
             SubOp = AMDGPU::V_SUB_U32_e64;
-            Opnds.push_back(Zero); // clamp bit
+            Opnds.push_back(
+                CurDAG->getTargetConstant(0, {}, MVT::i1)); // clamp bit
           }
----------------
This is missing a test case 


================
Comment at: llvm/test/CodeGen/AMDGPU/uaddo.ll:176
+; GCN: s_endpgm
+define amdgpu_kernel void @s_uaddo_clamp_bit(i32 addrspace(1)* %out, i1 addrspace(1)* %carryout, i32 %a, i32 %b) #0 {
+entry:
----------------
This should really use VGPR inputs


================
Comment at: llvm/test/CodeGen/AMDGPU/uaddo.ll:178
+entry:
+  %uadd = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
+  %val = extractvalue { i32, i1 } %uadd, 0
----------------
Should  also cover sub version


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D59608/new/

https://reviews.llvm.org/D59608