[PATCH] D138216: [AMDGPU] Intrinsic to expose s_wait_event for export ready

[Jay Foad via Phabricator via llvm-commits]

foad added a reviewer: AMDGPU.
foad added inline comments.


================
Comment at: llvm/include/llvm/IR/IntrinsicsAMDGPU.td:2070
 
+def int_amdgcn_wait_event_export_ready :
+  Intrinsic<[], [], [IntrNoMem, IntrHasSideEffects, IntrWillReturn]
----------------
Should this be int_amdgcn_s_wait_event_export_ready? I'm not sure how consistent we are about including the "s" or "v" prefix in intrinsic names.


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:8375
                                       Op->getOperand(2), Chain), 0);
+  case Intrinsic::amdgcn_wait_event_export_ready:
+    return SDValue(DAG.getMachineNode(AMDGPU::S_WAIT_EVENT, DL, MVT::Other,
----------------
Why do you need this? Can't you do it with a tablegen selection pattern?


================
Comment at: llvm/test/CodeGen/AMDGPU/llvm.amdgcn.wait.event.ll:1
+; RUN: llc -march=amdgcn -verify-machineinstrs -mcpu=gfx1100 < %s | FileCheck -check-prefix=GCN %s
+
----------------
Should test globalisel as well.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D138216/new/

https://reviews.llvm.org/D138216