This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: Patch to add support for gij for ppc linux
- To: Tom Tromey <tromey at redhat dot com>,khendricks at ivey dot uwo dot ca
- Subject: Re: Patch to add support for gij for ppc linux
- From: Kevin B. Hendricks <khendricks at ivey dot uwo dot ca>
- Date: Sat, 24 Mar 2001 09:23:55 -0500
- Cc: java at gcc dot gnu dot org
- References: <200102201420.f1KEKA713161@ashley.ivey.uwo.ca> <87bsqrq2fu.fsf@creche.redhat.com>
- Reply-To: khendricks at ivey dot uwo dot ca
Hi Tom,
> I've tested the libffi patch and it works fine. I will be checking it
> in on the trunk and the branch shortly.
I still want to "fix" a few minor things to make possibly make it work on ppc
64bit platforms (although I have no way of testing them and IBM has yet to
donate a honkin big 64bit ppc box for me to play with)!
> Thanks a lot for this work! Now PPC is one of our fully supported
> targets!
Glad to do it.
I even sent an e-mail to the Java Linux guys I know at Sun saying how great I
think this project is and how they should find a way to work with you guys
under a full GPL license to get you whatever resources you need. It seems
silly to have to worry about HotSpot and JITs to get even decent jdk
performance when the bulk of the needed performance improvement can come from
ahead of time compiling which gcj is perfect for. Your model gcj/gij is IMHO
the *right* way to do Java.
Interestingly enough, no one from Sun responded in any way. That is very
unusual. That could mean good news or bad news. Perhaps they have something
in the works (I wish it was GPL'ing java!) Either way they are not talking.
> Just to pick a nit: you forgot to write a ChangeLog entry. I'm
> writing one.
Sorry about that. Typically the commit message sets up our changelogs
entries at Blackdown so I often forget to do it on other projects
> I haven't done anything with the first patch. It would have to be
> examined by the regular gcc hackers. Maybe it has been already?
It is really only needed under debug mode since Geoff's original libffi code
has an assert that checks for stack alignment. Technically, the stack
alignment change should not be needed and I can change Geoff's code to remove
that constraint if need be.
I passed along a bug report to gcc list but really have not heard anything
back.
If you run into any problems with it, please let me know. I am still not
sure I have handled exceptions properly and whether anything special needs to
be done when exception stack walkbacks occur through trampoline code.
Good luck with your project. I wish I could help more but until Sun wakes up
I can't help.
Take care,
Kevin
Kevin