Aleksey Shipilev said the following on 10.04.2008 13:08:
[Charging... Clear... *ZAP*]
Colleagues,
I had updated JNI transition optimization umbrella issue [1] with
distribution of clean/dirty patches. There are 2 issues ready for
commit, can some VM guru review them and commit?
I am going to take care of HARMONY-4705.
[1] https://issues.apache.org/jira/browse/HARMONY-4704
On Sun, Jan 27, 2008 at 12:34 AM, Aleksey Shipilev
<[EMAIL PROTECTED]> wrote:
[reviving thread]
Hi, all!
During profiling of some native-intensive workload I had noticed the
whole bunch of JNI-related methods are hot on VM side.
These include:
jthread_get_vm_thread_ptr_stub
exn_raised
m2n_pop_local_handles
free_local_object_handles2
oh_allocate_handle
m2n_get_last_frame
oh_allocate_local_handle_internal
I had remembered I had filed a bunch of experimental JNI improvements
for Harmony [1]. Sure, some of them are dirty and not intended for
inclusion, but at least one [2] is worth commiting, it will solve the
problem with calling "get TLS" method during JNI transition.
Can VM gurus have an input here? Any objections?
Thanks,
Aleksey.
[1] https://issues.apache.org/jira/browse/HARMONY-4704
[2] https://issues.apache.org/jira/browse/HARMONY-4705
--
Gregory
