RFR: 8146416: java.lang.OutOfMemoryError triggers: assert(current_bci == 0) failed: bci isn't zero for do_not_unlock_if_synchronized
rwestrel at redhat.com
Thu May 26 09:29:29 UTC 2016
> my worry is this, we do all computation (vframe array) using one
> exception or no exception at all.. and later change it to realloc
> exception.. wouldn't this cause and issue at unpack.
JVMCI does something similar so we can suppose it's safe.
> isn' t checking exception at interpreter return and pop all frames is
> better thing to do, here?
In general, I think we want to avoid platform dependent changes as they
are painful to implement and maintain. Also I find the exception
propagation code quite hard to follow so not adding another special case
seems like a good thing.
More information about the hotspot-compiler-dev