Use the "GC is marking" information in compiler read barriers (ARM, ARM64).
In compiler-generated code, when deciding whether to mark
a heap reference or not in a read barrier, check whether
the GC is currently marking, instead of checking the gray
bit in the reference's holder's lock word.
This change is only for ARM and ARM64, as it does not
benefit x86 nor x86-64.
Test: Run ART tests in Baker read barrier configuration.
Test: Boot a device in Baker read barrier configuration.
Bug: 29516974
Change-Id: Ia5d90286bb9f753f3bbcb3a6254eb166523a2ff5
diff --git a/compiler/optimizing/code_generator_arm64.h b/compiler/optimizing/code_generator_arm64.h
index 5faf29a..231fb05 100644
--- a/compiler/optimizing/code_generator_arm64.h
+++ b/compiler/optimizing/code_generator_arm64.h
@@ -616,8 +616,8 @@
Location index,
vixl::aarch64::Register temp,
bool needs_null_check);
- // Factored implementation used by GenerateFieldLoadWithBakerReadBarrier
- // and GenerateArrayLoadWithBakerReadBarrier.
+ // Factored implementation, used by GenerateFieldLoadWithBakerReadBarrier,
+ // GenerateArrayLoadWithBakerReadBarrier and some intrinsics.
//
// Load the object reference located at the address
// `obj + offset + (index << scale_factor)`, held by object `obj`, into
@@ -636,6 +636,16 @@
bool use_load_acquire,
bool always_update_field = false);
+ // Generate a heap reference load (with no read barrier).
+ void GenerateRawReferenceLoad(HInstruction* instruction,
+ Location ref,
+ vixl::aarch64::Register obj,
+ uint32_t offset,
+ Location index,
+ size_t scale_factor,
+ bool needs_null_check,
+ bool use_load_acquire);
+
// Generate a read barrier for a heap reference within `instruction`
// using a slow path.
//