Ensure HandleChunk actually moves into native state
Due to an unnamed ScopedThreadStateChange we were not running the Ddms
code in the correct native mode.
Test: ./test.py --host
Change-Id: Ic8320bf7675aebd57f40af4b63d5ff6e18f26c7c
diff --git a/openjdkjvmti/ti_ddms.cc b/openjdkjvmti/ti_ddms.cc
index 0b4906d..bf063fa 100644
--- a/openjdkjvmti/ti_ddms.cc
+++ b/openjdkjvmti/ti_ddms.cc
@@ -60,7 +60,7 @@
*data_out = nullptr;
art::Thread* self = art::Thread::Current();
- art::ScopedThreadStateChange(self, art::ThreadState::kNative);
+ art::ScopedThreadStateChange stcs(self, art::ThreadState::kNative);
art::ArrayRef<const jbyte> data_arr(data_in, length_in);
std::vector<uint8_t> out_data;