harmony-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Aleksey Shipilev (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HARMONY-4729) [drlvm][jni] JNI transition should inline hythread_suspend_enable() and hythread_suspend_disable()
Date Thu, 27 Sep 2007 09:31:51 GMT

     [ https://issues.apache.org/jira/browse/HARMONY-4729?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Aleksey Shipilev updated HARMONY-4729:
--------------------------------------

    Attachment: jni-hythr-2.patch

Patch updated for current state of TM.
VM_thread is now equal to hythread, so no additional stubs needed.

> [drlvm][jni] JNI transition should inline hythread_suspend_enable() and hythread_suspend_disable()
> --------------------------------------------------------------------------------------------------
>
>                 Key: HARMONY-4729
>                 URL: https://issues.apache.org/jira/browse/HARMONY-4729
>             Project: Harmony
>          Issue Type: Improvement
>          Components: DRLVM
>            Reporter: Aleksey Shipilev
>         Attachments: jni-hythr-2.patch, jni-hythr.patch
>
>
> JNI calls a pair of hythread_suspend_enable() and hythread_suspend_disable() every time
the transition occurs. That lead to serious overhead due to calls. 
> The idea is to inline these methods by implementing two mnemonics in LIL - hse() and
hsd(), then LIL codegenerator will generate the code for them. More solid solution should
inline method calls in LIL, but it's too complex to achieve.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message