[prev in list] [next in list] [prev in thread] [next in thread]
List: openjdk-serviceability-dev
Subject: Re: JEP 158 support for JIT
From: Yasumasa Suenaga <yasuenag () gmail ! com>
Date: 2017-01-05 12:49:53
Message-ID: 376b811d-2daa-cf1c-e2f0-0e325388953a () gmail ! com
[Download RAW message or body]
Hi Vladimir,
> Yes, please, file "Enchantment" in JBS for hotspot/compiler category and set "Fix \
> version" 10. We can review it now - no need to wait when jdk 10 is opened (should \
> be soon).
Thanks!
I filed them to JBS.
> > Compilation logging: \
> > http://cr.openjdk.java.net/~ysuenaga/ul-jit/compilation/
>
> In all places you have unified logging followed by the same normal out may be you \
> should factor out into separate method which does both outputs instead of having \
> almost duplicated code all over places.
I've sent review request:
http://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/2017-January/025275.html
According to JEP, I guess we should remove existed logging code.
So I just add new function for UL to CompileTask class.
> > CodeCache flushing logging: \
> > http://cr.openjdk.java.net/~ysuenaga/ul-jit/codecache/
>
> codeCache.cpp - may be use const char* locals for messages instead of duplicating \
> them.
I've sent review request:
http://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/2017-January/025276.html
I fixed them in this webrev.
Thanks,
Yasumasa
On 2017/01/05 3:47, Vladimir Kozlov wrote:
> Thank you, Yasumasa
>
> Yes, please, file "Enchantment" in JBS for hotspot/compiler category and set "Fix \
> version" 10. We can review it now - no need to wait when jdk 10 is opened (should \
> be soon).
> On 1/4/17 4:21 AM, Yasumasa Suenaga wrote:
> > Hi,
> >
> > > Contributions are always welcome! :)
> >
> > I uploaded patches for using UL.
> > I want to get JIT logs as below:
> >
> > Compilation logging: \
> > http://cr.openjdk.java.net/~ysuenaga/ul-jit/compilation/
>
> In all places you have unified logging followed by the same normal out may be you \
> should factor out into separate method which does both outputs instead of having \
> almost duplicated code all over places.
> > CodeCache flushing logging: \
> > http://cr.openjdk.java.net/~ysuenaga/ul-jit/codecache/
>
> codeCache.cpp - may be use const char* locals for messages instead of duplicating \
> them.
> Thanks,
> Vladimir
>
> >
> >
> > > But for JDK 9 we already passed Features freeze dates.
> >
> > Should I file them to JBS and send webrev after jdk10 repos is opened?
> >
> >
> > Thanks,
> >
> > Yasumasa
> >
> >
> > On 2017/01/04 8:06, Vladimir Kozlov wrote:
> > > Contributions are always welcome! :)
> > >
> > > But for JDK 9 we already passed Features freeze dates.
> > >
> > > Thanks,
> > > Vladimir
> > >
> > > On 1/3/17 3:01 PM, Christian Thalinger wrote:
> > > >
> > > > > On Jan 2, 2017, at 6:08 PM, Yasumasa Suenaga <yasuenag@gmail.com> wrote:
> > > > >
> > > > > Thanks Vladimir,
> > > > >
> > > > > > Definitely not in JDK 9. And I can't say when it could be done or done at \
> > > > > > all.
> > > > >
> > > > > I hope this feature will be implemented ASAP.
> > > >
> > > > You can always contribute an implementation. Shouldn't be too difficult.
> > > >
> > > > >
> > > > >
> > > > > Yasumasa
> > > > >
> > > > >
> > > > > On 2017/01/03 12:00, Vladimir Kozlov wrote:
> > > > > > On 1/2/17 6:33 PM, Yasumasa Suenaga wrote:
> > > > > > > Hi all,
> > > > > > >
> > > > > > > Java 9 has JEP 158: Unified JVM Logging.
> > > > > > > This JEP describes that existing 'tty->print...' logging should use \
> > > > > > > unified logging as output. However, C2 compiler does
> > > > > > > not seem to use it.
> > > > > > >
> > > > > > > Do you have any plan to use JEP 158 in JIT codes?
> > > > > >
> > > > > > Definitely not in JDK 9. And I can't say when it could be done or done at \
> > > > > > all.
> > > > > > Regards,
> > > > > > Vladimir
> > > > > >
> > > > > > >
> > > > > > > I uploaded Unified JVM logging viewer to GitHub [1].
> > > > > > > I want to draw chart(s) or list all JIT'ed methods on it if possible.
> > > > > > > (Especially I want to get log from PrintCompilation and PrintIntrinsics \
> > > > > > > through Unified JVM logging)
> > > > > > >
> > > > > > > Thanks,
> > > > > > >
> > > > > > > Yasumasa
> > > > > > >
> > > > > > >
> > > > > > > [1] https://github.com/YaSuenag/ulviewer
> > > > > > >
> > > >
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic