[PATCH] D62537: [AMDGPU] Correct the handling of inlineasm output registers.

Michael Liao via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue May 28 12:34:34 PDT 2019


This revision was automatically updated to reflect the committed changes.
Closed by commit rL361868: [AMDGPU] Correct the handling of inlineasm output registers. (authored by hliao, committed by ).

Changed prior to commit:
  https://reviews.llvm.org/D62537?vs=201730&id=201742#toc

Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D62537/new/

https://reviews.llvm.org/D62537

Files:
  llvm/trunk/lib/Target/AMDGPU/SIISelLowering.cpp
  llvm/trunk/test/CodeGen/AMDGPU/inline-asm.ll


Index: llvm/trunk/test/CodeGen/AMDGPU/inline-asm.ll
===================================================================
--- llvm/trunk/test/CodeGen/AMDGPU/inline-asm.ll
+++ llvm/trunk/test/CodeGen/AMDGPU/inline-asm.ll
@@ -277,3 +277,23 @@
   tail call void asm sideeffect "; sgpr96 $0", "s"(<3 x i32> <i32 10, i32 11, i32 12>) #1
   ret void
 }
+
+; Check aggregate types are handled properly.
+; CHECK-LABEL: mad_u64
+; CHECK: v_mad_u64_u32
+define void @mad_u64(i32 %x) {
+entry:
+  br i1 undef, label %exit, label %false
+
+false:
+  %s0 = tail call { i64, i64 } asm sideeffect "v_mad_u64_u32 $0, $1, $2, $3, $4", "=v,=s,v,v,v"(i32 -766435501, i32 %x, i64 0)
+  br label %exit
+
+exit:
+  %s1 = phi { i64, i64} [ undef, %entry ], [ %s0, %false]
+  %v0 = extractvalue { i64, i64 } %s1, 0
+  %v1 = extractvalue { i64, i64 } %s1, 1
+  tail call void asm sideeffect "; use $0", "v"(i64 %v0)
+  tail call void asm sideeffect "; use $0", "v"(i64 %v1)
+  ret void
+}
Index: llvm/trunk/lib/Target/AMDGPU/SIISelLowering.cpp
===================================================================
--- llvm/trunk/lib/Target/AMDGPU/SIISelLowering.cpp
+++ llvm/trunk/lib/Target/AMDGPU/SIISelLowering.cpp
@@ -10244,8 +10244,7 @@
           unsigned AssignedReg;
           const TargetRegisterClass *RC;
           std::tie(AssignedReg, RC) = getRegForInlineAsmConstraint(
-              SIRI, TC.ConstraintCode,
-              getSimpleValueType(MF.getDataLayout(), CS.getType()));
+              SIRI, TC.ConstraintCode, TC.ConstraintVT);
           if (RC) {
             MachineRegisterInfo &MRI = MF.getRegInfo();
             if (AssignedReg != 0 && SIRI->isSGPRReg(MRI, AssignedReg))


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D62537.201742.patch
Type: text/x-patch
Size: 1676 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190528/66a9e883/attachment-0001.bin>


More information about the llvm-commits mailing list