Add support for Baker read barriers in SystemArrayCopy intrinsics.
Benchmarks (ARM64) score variations on Nexus 5X with CPU
cores clamped at 960000 Hz (aosp_bullhead-userdebug build):
- Ritzperf - average (lower is better): -3.03% (slightly better)
- CaffeineMark - average (higher is better): +1.26% (slightly better)
- DeltaBlue (lower is better): -10.50% (better)
- Richards - average (lower is better): -3.36% (slightly better)
- SciMark2 - average (higher is better): +0.26% (virtually unchanged)
Details about Ritzperf benchmarks with meaningful variations
(lower is better):
- FormulaEvaluationActions.EvaluateAndApplyChanges: -13.26% (better)
- FormulaEvaluationActions.EvaluateCascadingSums: -10.94% (better)
- FormulaEvaluationActions.EvaluateComplexFormulas: -15.50% (better)
- FormulaEvaluationActions.EvaluateFibonacci: -10.41% (better)
- FormulaEvaluationActions.EvaluateLargeSums: +6.02% (worse)
Boot image code size variation on Nexus 5X
(aosp_bullhead-userdebug build):
- total ARM64 framework Oat files size change:
107047632 bytes -> 107154128 bytes (+0.10%)
- total ARM framework Oat files size change:
90932028 bytes -> 91009852 bytes (+0.09%)
Test: ART host and target (ARM, ARM64) tests + Nexus 5X boot.
Bug: 29516905
Bug: 29506760
Bug: 12687968
Change-Id: I85431368d09965687a0301ae2eb3c991f276ce5d
diff --git a/compiler/optimizing/code_generator_arm64.cc b/compiler/optimizing/code_generator_arm64.cc
index d95e7df..14f5b24 100644
--- a/compiler/optimizing/code_generator_arm64.cc
+++ b/compiler/optimizing/code_generator_arm64.cc
@@ -595,7 +595,8 @@
instruction_->IsLoadString() ||
instruction_->IsInstanceOf() ||
instruction_->IsCheckCast() ||
- (instruction_->IsInvokeVirtual()) && instruction_->GetLocations()->Intrinsified())
+ (instruction_->IsInvokeVirtual() && instruction_->GetLocations()->Intrinsified()) ||
+ (instruction_->IsInvokeStaticOrDirect() && instruction_->GetLocations()->Intrinsified()))
<< "Unexpected instruction in read barrier marking slow path: "
<< instruction_->DebugName();
@@ -607,7 +608,8 @@
DCHECK_NE(obj_.reg(), LR);
DCHECK_NE(obj_.reg(), WSP);
DCHECK_NE(obj_.reg(), WZR);
- // WIP0 is used by the slow path as a temp, it can not be the object register.
+ // IP0 is used internally by the ReadBarrierMarkRegX entry point
+ // as a temporary, it cannot be the entry point's input/output.
DCHECK_NE(obj_.reg(), IP0);
DCHECK(0 <= obj_.reg() && obj_.reg() < kNumberOfWRegisters) << obj_.reg();
// "Compact" slow path, saving two moves.
@@ -5239,7 +5241,8 @@
// Introduce a dependency on the lock_word including rb_state,
// to prevent load-load reordering, and without using
// a memory barrier (which would be more expensive).
- // obj is unchanged by this operation, but its value now depends on temp.
+ // `obj` is unchanged by this operation, but its value now depends
+ // on `temp`.
__ Add(obj.X(), obj.X(), Operand(temp.X(), LSR, 32));
// The actual reference load.