summaryrefslogtreecommitdiffstats
path: root/build/Android.common.mk
diff options
context:
space:
mode:
authorElliott Hughes <enh@google.com>2013-02-20 14:57:12 -0800
committerElliott Hughes <enh@google.com>2013-02-20 15:46:06 -0800
commit64f574f474aa77c72778640ab21f8cfa72546812 (patch)
treec90d3ef8f56144e61a4a043138a5118c3c9e7b9a /build/Android.common.mk
parent492899e445c21f09db458c76d773d571cc6cc0ac (diff)
downloadart-64f574f474aa77c72778640ab21f8cfa72546812.zip
art-64f574f474aa77c72778640ab21f8cfa72546812.tar.gz
art-64f574f474aa77c72778640ab21f8cfa72546812.tar.bz2
Implement EnableCollection, DisableCollection, IsCollected, and fix Exit.
Rewrites the object registry to not just be one big leak. We use jweaks by default, which means (a) the GC doesn't need to know anything about us and (b) we don't cause unnecessary heap retention. We promote to regular JNI global references if the debugger asks us to. The problem with VirtualMachine.Exit turned out to be that you're supposed to send a success reply to the command before exiting. This is a bit awkward with our current division of responsibilities, but it lets us pass another test. Also log a summary of our replies when -verbose:jdwp is in effect, not just the requests. Change-Id: Idb33e99fe7d8bee7a79152d81fee42e2af00852b
Diffstat (limited to 'build/Android.common.mk')
-rw-r--r--build/Android.common.mk3
1 files changed, 2 insertions, 1 deletions
diff --git a/build/Android.common.mk b/build/Android.common.mk
index 8b64519..5262ce0 100644
--- a/build/Android.common.mk
+++ b/build/Android.common.mk
@@ -189,12 +189,13 @@ LIBART_COMMON_SRC_FILES := \
src/jdwp/jdwp_handler.cc \
src/jdwp/jdwp_main.cc \
src/jdwp/jdwp_socket.cc \
+ src/jdwp/object_registry.cc \
src/jni_internal.cc \
src/jobject_comparator.cc \
src/locks.cc \
src/mem_map.cc \
src/memory_region.cc \
- src/mirror/abstract_method.cc \
+ src/mirror/abstract_method.cc \
src/mirror/array.cc \
src/mirror/class.cc \
src/mirror/dex_cache.cc \