[AMDGPU] Simplify VCCZ bug handling

Summary:
VCCZBugHandledSet was used to make sure we don't apply the same
workaround more than once to a single cbranch instruction, but it's not
necessary because the workaround involves inserting an s_waitcnt
instruction, which is enough for subsequent iterations to detect that no
further workaround is necessary.

Also beef up the test case to check that the workaround was only applied
once. I have also manually verified that the test still passes even if I
hack the big do-while loop in runOnMachineFunction to run a minimum of
five iterations.

Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D69621

diff --git a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
index 6a688b7..14b12f9 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
@@ -372,7 +372,6 @@ private:
   AMDGPU::IsaVersion IV;
 
   DenseSet<MachineInstr *> TrackedWaitcntSet;
-  DenseSet<MachineInstr *> VCCZBugHandledSet;
 
   struct BlockInfo {
     MachineBasicBlock *MBB;
@@ -1388,8 +1387,7 @@ bool SIInsertWaitcnts::insertWaitcntInBlock(MachineFunction &MF,
     }
 
     bool VCCZBugWorkAround = false;
-    if (readsVCCZ(Inst) &&
-        (!VCCZBugHandledSet.count(&Inst))) {
+    if (readsVCCZ(Inst)) {
       if (ScoreBrackets.getScoreLB(LGKM_CNT) <
               ScoreBrackets.getScoreUB(LGKM_CNT) &&
           ScoreBrackets.hasPendingEvent(SMEM_ACCESS)) {
@@ -1431,7 +1429,6 @@ bool SIInsertWaitcnts::insertWaitcntInBlock(MachineFunction &MF,
               TII->get(ST->isWave32() ? AMDGPU::S_MOV_B32 : AMDGPU::S_MOV_B64),
               TRI->getVCC())
           .addReg(TRI->getVCC());
-      VCCZBugHandledSet.insert(&Inst);
       Modified = true;
     }
 
@@ -1471,7 +1468,6 @@ bool SIInsertWaitcnts::runOnMachineFunction(MachineFunction &MF) {
       RegisterEncoding.SGPR0 + HardwareLimits.NumSGPRsMax - 1;
 
   TrackedWaitcntSet.clear();
-  VCCZBugHandledSet.clear();
   RpotIdxMap.clear();
   BlockInfos.clear();
 

diff --git a/llvm/test/CodeGen/AMDGPU/smrd-vccz-bug.ll b/llvm/test/CodeGen/AMDGPU/smrd-vccz-bug.ll
index 823785d..146e631 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/smrd-vccz-bug.ll
+++ b/llvm/test/CodeGen/AMDGPU/smrd-vccz-bug.ll
@@ -1,13 +1,13 @@
 ; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=VCCZ-BUG %s
 ; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=VCCZ-BUG %s
-; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=NOVCCZ-BUG %s
+; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
 
 ; GCN-FUNC: {{^}}vccz_workaround:
 ; GCN: s_load_dword s{{[0-9]+}}, s[{{[0-9]+:[0-9]+}}], 0x0
 ; GCN: v_cmp_neq_f32_e64 {{[^,]*}}, s{{[0-9]+}}, 0{{$}}
 ; VCCZ-BUG: s_waitcnt lgkmcnt(0)
 ; VCCZ-BUG: s_mov_b64 vcc, vcc
-; NOVCCZ-BUG-NOT: s_mov_b64 vcc, vcc
+; GCN-NOT: s_mov_b64 vcc, vcc
 ; GCN: s_cbranch_vccnz [[EXIT:[0-9A-Za-z_]+]]
 ; GCN: buffer_store_dword
 ; GCN: [[EXIT]]:
@@ -28,6 +28,8 @@ endif:
 
 ; GCN-FUNC: {{^}}vccz_noworkaround:
 ; GCN: v_cmp_neq_f32_e32 vcc, 0, v{{[0-9]+}}
+; GCN-NOT: s_waitcnt lgkmcnt(0)
+; GCN-NOT: s_mov_b64 vcc, vcc
 ; GCN: s_cbranch_vccnz [[EXIT:[0-9A-Za-z_]+]]
 ; GCN: buffer_store_dword
 ; GCN: [[EXIT]]: