Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 25 Aug 2001 08:41:51 +0930
From:      Greg Lewis <glewis@eyesbeyond.com>
To:        Tim Liddelow <tim@ideasandassociates.com>
Cc:        Bill Huey <billh@gnuppy.monkey.org>, freebsd-java@FreeBSD.ORG
Subject:   Re: Contributing...
Message-ID:  <20010825084151.A82497@misty.eyesbeyond.com>
In-Reply-To: <3B86D765.249C2214@ideasandassociates.com>; from tim@ideasandassociates.com on Fri, Aug 24, 2001 at 06:38:29PM -0400
References:  <3B8688AA.6956F1BD@ideasandassociates.com> <20010824131403.A3036@gnuppy> <3B86D765.249C2214@ideasandassociates.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Aug 24, 2001 at 06:38:29PM -0400, Tim Liddelow wrote:
> Bill Huey wrote:
> > Yes, and I dumped on it in favor of a pthread implementation instead
> > in light of the KSE effort and how LinuxThreading was hacked into the
> > thread creation/destruction glue layer (wierd SIGCHLD stuff, etc...).
> >
> > This is what led to the current track for getting native threads working.
> 
> The KSE effort seems promising, but my experience with -current is old now (I
> haven't toyed with it since pre SMPng days..and I haven't been active on the
> lists at all), and Julian seems to only be in the initial stages (compilation
> going but unstable).  I wish to focus on contributing to a rock solid, -stable
> port of the recent JDKs and give FreeBSD the chance to show off its other
> powers (great VM system, etc) in a Java environment.  I can also leverage some
> work I am doing here in the commercial sector without encumbrance because the
> result will be utilised by us in production environments (this is probably
> similar to some others here I hope!).

I agree that -STABLE is definitely what we want to focus on.  Apart from
the fact that its used in production, -CURRENT is too much of a moving
target for the current porting team's resources.

> > > Also, what about HotSpot ?  Has anyone attempted to port this to any of
> > > the BSDs ?   I'm looking at other JITs now.  The latest TYA doesn't
> > > build under the 1.3.1 kit.
> >
> > These are the two biggest items on my list while I'm in between jobs.
> >
> > The order of attack is native threading first and then HotSpot. IMO,
> > both can be worked on simultaneously, but you absolutely have to have
> > the threading solid (green, native) before you can even smoke test
> > HotSpot. So getting HotSpot to compile would likely be the hard limit
> > for progress until the surrounding JVM facilities were to solidify.
> 
> I have been lightly following the pthread updates; I know a lot of fixes were
> committed a few months back - how complete now is our pthread support ?   I
> agree with you that native thread support must be done first ... I was hoping
> however to familiarise myself again with the byte code compiler, then take a
> look at HotSpot and being to dissect it - but only initially from the point of
> view of kernel differences between FreeBSD and Linux.

I think the FreeBSD pthread support is pretty complete.  Not only that, but 
its actively maintained by a number of developers who I've found to be
very approachable about any problems and the like.

> > Both the JVM thread (virtual machine craziness + language runtimes) and
> > HotSpot (C++ based with tricky *everything* that come with JIT compilers
> > cores) systems are non-trivial and it would take at least a number of
> > weeks for an accomplished engineer (IMO) to be comfortable with virtual
> > machine abstractions and do basic work with it.
> 
> Sure.  But we have to start somewhere.  I don't promise to be able to make
> everything happen - but if we begin, maybe we will gather more interest and
> hopefully get some momentum going.  Usually things don't get done in this
> space not because of experienced people (to a point!), but usually because of
> motivational and coordinational factors.  Can we get any help from anyone at
> Sun who did the Linux port of HotSpot ?   Nate ?

There is no reason one couldn't start straight on HotSpot to at least try
and get it compiling.  Be sure to have polished up on both your C++ and
x86 assembler though :).

BTW, no need to promise to get everything happening, there are people
ready to help in the porting team :).

-- 
Greg Lewis                            Email : glewis@eyesbeyond.com
Eyes Beyond                           Phone : (801) 765 1887
Information Technology                Web   : http://www.eyesbeyond.com


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-java" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010825084151.A82497>