From aba845e835d749155c431c559a7fb19d80cb6b25 Mon Sep 17 00:00:00 2001
From: Chad Rosier <mcrosier@codeaurora.org>
Date: Wed, 2 Jul 2014 16:46:08 +0000
Subject: [PATCH] Revert "Revert "MachineScheduler: better book-keeping for
 asserts.""

This reverts commit r212109, which reverted r212088.

However, disable the assert as it's not necessary for correctness.  There are
several corner cases that the assert needed to handle better for in-order
scheduling, but none of them are incorrect scheduler behavior. The assert is
mainly there to collect good unit tests like this and ensure that the
target-independent scheduler is working as expected with the various machine
models.

llvm-svn: 212187
---
 llvm/lib/CodeGen/MachineScheduler.cpp              | 18 ++++++++------
 .../CodeGen/AArch64/arm64-misched-basic-A53.ll     | 29 ++++++++++++++++++++++
 2 files changed, 40 insertions(+), 7 deletions(-)

diff --git a/llvm/lib/CodeGen/MachineScheduler.cpp b/llvm/lib/CodeGen/MachineScheduler.cpp
index 0075242..44191f7 100644
--- a/llvm/lib/CodeGen/MachineScheduler.cpp
+++ b/llvm/lib/CodeGen/MachineScheduler.cpp
@@ -1689,7 +1689,7 @@ bool SchedBoundary::checkHazard(SUnit *SU) {
       unsigned NRCycle = getNextResourceCycle(PI->ProcResourceIdx, PI->Cycles);
       if (NRCycle > CurrCycle) {
 #ifndef NDEBUG
-        MaxObservedStall = std::max(NRCycle - CurrCycle, MaxObservedStall);
+        MaxObservedStall = std::max(PI->Cycles, MaxObservedStall);
 #endif
         DEBUG(dbgs() << "  SU(" << SU->NodeNum << ") "
               << SchedModel->getResourceName(PI->ProcResourceIdx)
@@ -1953,10 +1953,12 @@ void SchedBoundary::bumpNode(SUnit *SU) {
              PE = SchedModel->getWriteProcResEnd(SC); PI != PE; ++PI) {
         unsigned PIdx = PI->ProcResourceIdx;
         if (SchedModel->getProcResource(PIdx)->BufferSize == 0) {
-          ReservedCycles[PIdx] = isTop() ? NextCycle + PI->Cycles : NextCycle;
-#ifndef NDEBUG
-          MaxObservedStall = std::max(PI->Cycles, MaxObservedStall);
-#endif
+          if (isTop()) {
+            ReservedCycles[PIdx] =
+              std::max(getNextResourceCycle(PIdx, 0), NextCycle + PI->Cycles);
+          }
+          else
+            ReservedCycles[PIdx] = NextCycle;
         }
       }
     }
@@ -2059,8 +2061,10 @@ SUnit *SchedBoundary::pickOnlyChoice() {
     }
   }
   for (unsigned i = 0; Available.empty(); ++i) {
-    assert(i <= (HazardRec->getMaxLookAhead() + MaxObservedStall) &&
-           "permanent hazard"); (void)i;
+//  FIXME: Re-enable assert once PR20057 is resolved.
+//    assert(i <= (HazardRec->getMaxLookAhead() + MaxObservedStall) &&
+//           "permanent hazard");
+    (void)i;
     bumpCycle(CurrCycle + 1);
     releasePending();
   }
diff --git a/llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll b/llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll
index a03d36b..bc7ed7f 100644
--- a/llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll
@@ -172,3 +172,32 @@ entry:
 }
 
 declare void @llvm.trap()
+
+; Regression test for PR20057: "permanent hazard"'
+; Resource contention on LDST.
+; CHECK: ********** MI Scheduling **********
+; CHECK: testLdStConflict
+; CHECK: *** Final schedule for BB#1 ***
+; CHECK: LD4Fourv2d
+; CHECK: STRQui
+; CHECK: ********** INTERVALS **********
+define void @testLdStConflict() {
+entry:
+  br label %loop
+
+loop:
+  %0 = call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.aarch64.neon.ld4.v2i64.p0i8(i8* null)
+  %ptr = bitcast i8* undef to <2 x i64>*
+  store <2 x i64> zeroinitializer, <2 x i64>* %ptr, align 4
+  %ptr1 = bitcast i8* undef to <2 x i64>*
+  store <2 x i64> zeroinitializer, <2 x i64>* %ptr1, align 4
+  %ptr2 = bitcast i8* undef to <2 x i64>*
+  store <2 x i64> zeroinitializer, <2 x i64>* %ptr2, align 4
+  %ptr3 = bitcast i8* undef to <2 x i64>*
+  store <2 x i64> zeroinitializer, <2 x i64>* %ptr3, align 4
+  %ptr4 = bitcast i8* undef to <2 x i64>*
+  store <2 x i64> zeroinitializer, <2 x i64>* %ptr4, align 4
+  br label %loop
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.aarch64.neon.ld4.v2i64.p0i8(i8*)
-- 
2.7.4