Fix lock order issue with large-object space lock
We have a code-path where GC thread acquires large-object space lock (a
higher level lock) after acquire JNI global lock (a lower level lock).
Moving AllocSpaceLock to be lower than JniGlobalsLock to fix this.
Bug: 120782890
Test: art/test/testrunner/testrunner.py --host
Change-Id: I4f169c779cb17f6c14f6e5278857845f9904e62c
diff --git a/runtime/base/locks.h b/runtime/base/locks.h
index 57719f1..b7d8e31 100644
--- a/runtime/base/locks.h
+++ b/runtime/base/locks.h
@@ -71,6 +71,7 @@
kRosAllocGlobalLock,
kRosAllocBracketLock,
kRosAllocBulkFreeLock,
+ kAllocSpaceLock,
kTaggingLockLevel,
kTransactionLogLock,
kCustomTlsLock,
@@ -84,7 +85,6 @@
kReferenceQueueClearedReferencesLock,
kReferenceProcessorLock,
kJitDebugInterfaceLock,
- kAllocSpaceLock,
kBumpPointerSpaceBlockLock,
kArenaPoolLock,
kInternTableLock,