Enable concurrent sweep on MIPS and ARM64.

runtime/vm/stub_code_mips.cc

Index: runtime/vm/stub_code_mips.cc
diff --git a/runtime/vm/stub_code_mips.cc b/runtime/vm/stub_code_mips.cc
index 8d1b252e99fc9d00f26243d707c6fa5544d4c985..9b8e79df2f14c30264a05000876f4bd914af4987 100644
--- a/runtime/vm/stub_code_mips.cc
+++ b/runtime/vm/stub_code_mips.cc
@@ -1066,8 +1066,14 @@ void StubCode::GenerateUpdateStoreBufferStub(Assembler* assembler) {
   __ Ret();
 
   __ Bind(&add_to_buffer);
+  // Atomically set the remembered bit of the object header.
+  Label retry;
+  __ Bind(&retry);
+  __ ll(T2, FieldAddress(T0, Object::tags_offset()));
   __ ori(T2, T2, Immediate(1 << RawObject::kRememberedBit));
-  __ sw(T2, FieldAddress(T0, Object::tags_offset()));
+  __ sc(T2, FieldAddress(T0, Object::tags_offset()));
+  // T2 = 1 on success, 0 on failure.
+  __ beq(T2, ZR, &retry);
 
   // Load the StoreBuffer block out of the thread. Then load top_ out of the
   // StoreBufferBlock and add the address to the pointers_.