[llvm] 25e7da0 - AMDGPU/GlobalISel: Remove manual G_FENCE selection

[Matt Arsenault via llvm-commits]

Author: Matt Arsenault
Date: 2020-01-02T17:16:10-05:00
New Revision: 25e7da0c2495182b7c0a8871eaca0cc459578dc1

URL: https://github.com/llvm/llvm-project/commit/25e7da0c2495182b7c0a8871eaca0cc459578dc1
DIFF: https://github.com/llvm/llvm-project/commit/25e7da0c2495182b7c0a8871eaca0cc459578dc1.diff

LOG: AMDGPU/GlobalISel: Remove manual G_FENCE selection

The tablegen emitter now handles the immediate operand correctly, so
let the generatedd matcher works.

Added: 
    

Modified: 
    llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp b/llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp
index 38136d1a8403..d6b0ec6facbc 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp
@@ -1713,11 +1713,6 @@ bool AMDGPUInstructionSelector::select(MachineInstr &I) {
     return selectG_BRCOND(I);
   case TargetOpcode::G_FRAME_INDEX:
     return selectG_FRAME_INDEX(I);
-  case TargetOpcode::G_FENCE:
-    // FIXME: Tablegen importer doesn't handle the imm operands correctly, and
-    // is checking for G_CONSTANT
-    I.setDesc(TII.get(AMDGPU::ATOMIC_FENCE));
-    return true;
   case TargetOpcode::G_PTR_MASK:
     return selectG_PTR_MASK(I);
   default: