summaryrefslogtreecommitdiff
path: root/openjdkjvmti/deopt_manager.cc
diff options
context:
space:
mode:
Diffstat (limited to 'openjdkjvmti/deopt_manager.cc')
-rw-r--r--openjdkjvmti/deopt_manager.cc15
1 files changed, 6 insertions, 9 deletions
diff --git a/openjdkjvmti/deopt_manager.cc b/openjdkjvmti/deopt_manager.cc
index 312a797b9d..03042eccfd 100644
--- a/openjdkjvmti/deopt_manager.cc
+++ b/openjdkjvmti/deopt_manager.cc
@@ -61,16 +61,13 @@
namespace openjdkjvmti {
-// TODO We should make this much more selective in the future so we only return true when we
+// We could make this much more selective in the future so we only return true when we
// actually care about the method at this time (ie active frames had locals changed). For now we
-// just assume that if anything has changed any frame's locals we care about all methods. If nothing
-// has we only care about methods with active breakpoints on them. In the future we should probably
-// rewrite all of this to instead do this at the ShadowFrame or thread granularity.
-bool JvmtiMethodInspectionCallback::IsMethodBeingInspected(art::ArtMethod* method) {
- // In non-java-debuggable runtimes the breakpoint check would miss if we have breakpoints on
- // methods that are inlined. Since these features are best effort in non-java-debuggable
- // runtimes it is OK to be less precise. For debuggable runtimes, inlining is disabled.
- return manager_->HaveLocalsChanged() || manager_->MethodHasBreakpoints(method);
+// just assume that if anything has changed any frame's locals we care about all methods. This only
+// impacts whether we are able to OSR or not so maybe not really important to maintain frame
+// specific information.
+bool JvmtiMethodInspectionCallback::HaveLocalsChanged() {
+ return manager_->HaveLocalsChanged();
}
DeoptManager::DeoptManager()