[9] RFR (S): 8060147: SIGSEGV in Metadata::mark_on_stack() while marking metadata in ciEnv

Roland Westrelin roland.westrelin at oracle.com
Thu Oct 30 12:48:18 UTC 2014


>> _holder = env->get_instance_klass(h_m()->method_holder());
>> 
>> So if jvmti is enabled, we could safepoint before we set the holder field and the method could be unloaded?
> Doesn't method handle (h_m) keep it alive?

A comment in handles.hpp says:

// Metadata Handles.  Unlike oop Handles these are needed to prevent metadata
// from being reclaimed by RedefineClasses.

So that doesn’t help with class unloading?

Roland.


More information about the hotspot-runtime-dev mailing list