From 79d401905fcf519da732b33ee9dffd18f2da3b10 Mon Sep 17 00:00:00 2001 From: Michael Liao Date: Wed, 25 Dec 2019 00:47:18 -0500 Subject: [PATCH] [amdgpu] Fix scoreboard updating on `s_waitcnt_vscnt`. Summary: - Other counters are accidentally cleared. Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71866 --- llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp | 2 +- llvm/test/CodeGen/AMDGPU/waitcnt-vscnt.mir | 17 +++++++++++++++++ 2 files changed, 18 insertions(+), 1 deletion(-) create mode 100644 llvm/test/CodeGen/AMDGPU/waitcnt-vscnt.mir diff --git a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp index bfeaebd..927826c 100644 --- a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp +++ b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp @@ -1081,7 +1081,7 @@ bool SIInsertWaitcnts::generateWaitcntInstBefore( assert(II->getOpcode() == AMDGPU::S_WAITCNT_VSCNT); assert(II->getOperand(0).getReg() == AMDGPU::SGPR_NULL); ScoreBrackets.applyWaitcnt( - AMDGPU::Waitcnt(0, 0, 0, II->getOperand(1).getImm())); + AMDGPU::Waitcnt(~0u, ~0u, ~0u, II->getOperand(1).getImm())); } } } diff --git a/llvm/test/CodeGen/AMDGPU/waitcnt-vscnt.mir b/llvm/test/CodeGen/AMDGPU/waitcnt-vscnt.mir new file mode 100644 index 0000000..e9d8ebe --- /dev/null +++ b/llvm/test/CodeGen/AMDGPU/waitcnt-vscnt.mir @@ -0,0 +1,17 @@ +# RUN: llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs -run-pass si-insert-waitcnts -o - %s | FileCheck -check-prefix=GFX10 %s + +# GFX10-LABEL: waitcnt-vscnt +# GFX10: GLOBAL_ATOMIC_ADD_RTN +# GFX10-NEXT: S_WAITCNT 49279 +--- +name: waitcnt-vscnt +machineFunctionInfo: + isEntryFunction: true +body: | + bb.0: + liveins: $sgpr0_sgpr1 + $sgpr4 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 4, 0, 0 :: (dereferenceable invariant load 4 from `i32 addrspace(4)* undef`) + S_WAITCNT_VSCNT undef $sgpr_null, 0 + $vgpr0 = GLOBAL_ATOMIC_ADD_RTN $vgpr0_vgpr1, $vgpr2, 0, 0, implicit $exec :: (load store syncscope("agent") seq_cst 4, addrspace 1) + S_CMP_LG_U32 killed $sgpr4, 0, implicit-def $scc +... -- 2.7.4