shujit

shujit

Post by Dag-Erling Smorgr » Tue, 17 Jun 2003 16:53:35



--=-=-=

[redirected from private mail to -java]



> > You've marked java/shujit BROKEN for OSVERSION >= 500113, yet I've had
> > no trouble at all building and running it on 5.1-RELEASE and recent
> > -CURRENT.  What's up?
> Probably you have some stale headers on your system that are allowing
> the build to spuriously succeed.

You're right.  I've narrowed it down to bogus code in one of the
shujit headers which uses struct osigcontext instead of struct
sigcontext for FreeBSD >= 4.x, and apparently struct osigcontext
recently disappeared from the system headers.  I don't think using
struct osigcontext is correct for *any* FreeBSD version, but I haven't
had a chance to test my (attached) patch on 4.x yet.

Any objections to my picking up the maintainership on this port?  I
use it quite a bit since - unlike sobomax :) - I'm not entirely
comfortable with JDK 1.4.1 yet, and shujit makes a big difference on
1.3.1's performance.

DES
--

--=-=-=
Content-Type: text/x-patch
Content-Disposition: attachment; filename=shujit.diff

Index: Makefile
===================================================================
RCS file: /home/pcvs/ports/java/shujit/Makefile,v
retrieving revision 1.39
diff -u -r1.39 Makefile
--- Makefile    4 Jun 2003 22:43:29 -0000       1.39

 .include <bsd.port.pre.mk>

-.if ${OSVERSION} >= 500113
-BROKEN=         "Does not compile"
-.endif
-
 post-configure:

Index: files/patch-compiler.h
===================================================================
RCS file: /home/pcvs/ports/java/shujit/files/patch-compiler.h,v
retrieving revision 1.4
diff -u -r1.4 patch-compiler.h
--- files/patch-compiler.h      17 Feb 2003 15:20:08 -0000      1.4

  #  ifndef _UINT16_T
  #    define _UINT16_T

+ #    include <linux/version.h>  // for kernel version
+ #  elif defined(__FreeBSD__) || defined(__NetBSD__)
+ #    ifdef __FreeBSD__
+-#      if __FreeBSD__ >= 4 && JDK_VER < 12
++#      if __FreeBSD__ == 4 && JDK_VER < 12
+       // In this case, the 3rd arg. of sig. handler is not (sigcontext *).
+ #      define SIGCONTEXT_T struct osigcontext
+ #      else

--=-=-=
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________

http://lists.freebsd.org/mailman/listinfo/freebsd-java

--=-=-=--

 
 
 

shujit

Post by Greg Lew » Wed, 18 Jun 2003 09:23:37




> > Any objections to my picking up the maintainership on this port?  I
> > use it quite a bit since - unlike sobomax :) - I'm not entirely
> > comfortable with JDK 1.4.1 yet, and shujit makes a big difference on
> > 1.3.1's performance.

> > DES
> Do you mean it's even better than 1.3.1 w/ HotSpot?

Define better :).  Do you mean faster, stabler or what?  I suspect that
its a bit slower than HotSpot (in general) but a bit stabler, but thats
just a guess :).

Note that Shudo-san usually keeps up with the mailing list so he is
probably interested in the patches :).

--

Eyes Beyond                         Web     : http://www.eyesbeyond.com

_______________________________________________

http://lists.freebsd.org/mailman/listinfo/freebsd-java


 
 
 

1. offtopic question [Re: Shujit doesn't work with Forte and JDK1.3.1

This is very interesting. Is this a common practice to eliminate tail
recursion in Java JIT compilers or is it only shujit that does that?
In general is it a common optimization in procedural languages like Java,
C etc?
Thanks!


with "unsubscribe freebsd-java" in the body of the message

2. AIXServ

3. shujit with Resin

4. FAQ questions

5. Shujit doesn't work with Forte and JDK1.3.1 on FreeBSD

6. simple question about ports

7. ShuJit compile problem

8. changing user ftp directory

9. shuJIT 0.7.4

10. offtopic question [Re: Shujit doesn't work with Forte and

11. Jdk 1.2.2-beta and shujit problem

12. shuJIT-Problems since Upgrade FreeBSD 4.3 -> 4.4

13. shuJIT doesn't work with patchlevel 6 of the FreeBSD JDK1.3.1