Re: gcj 4.3.0 on mpc860 kernel 2.4.24

[Andrew Haley <aph at redhat dot com>]

Ben Gardiner wrote:
> Hello all, this is my first posting to the java@gcc list. First off: I'd
> like to thank the gcj team for making java 5 a reality; we're very
> excited to start using generics and more in our embedded application.
> And then...
> 
> I have been trying to get a working cross toolchain (i686 -> mpc860)
> working. So far I have been able to build a working chroot on the device
> (busybox, ssh, rsync, ...) using a 4.3.0 toolchain (binutils 2.18, glibc
> 2.5, kernel 2.4.24). So I think this is a gcj problem...

OK.

> However, when trying to execute a java hello world, I get a segfault. In
> a debugger, when i 'stepi' the program execution never gets past
> natString.cc:60
> 58      _Jv_StringFindSlot (jchar* data, jint len, jint hash)
> 59      {
> 60        JvSynchronize sync (&java::lang::String::class$);
> 
> It's not the first time into this sync, and it always occurs the second
> time that data has been created from the string "". I have read a little
> about the changes in the threading libs and I'm getting scared. Perhaps
> linuxthreads-2.5 on 2.4.24 isn't 'posix' enough for gcj 4.3.0 ?

Linuxthreads?  I don't like the sound of that.  gcj doesn't have any
explicit dependencies on NPTL, but it doesn't get tested on Linuxthreads.

> I was hoping some nice person out there could either:
> 1) tell me it is not possible to use gcj 4.3.0 on my glibc/kernel
> versions; or
> 2) please help me to fix my build of gcj 4.3.0 on 2.4.24: I have stack
> traces, the configure commands, flags etc. (I didn't want to spam the
> list if 1 is the case).

You're using a configuration that, to the best of my knowledge, no-one
else has ever tried.  gcj works well on PowerPC Apple systems, so there
is no reason to suspect it can't work.  There is probably some porting
to be done, however.  Whether this is going to work probably depends on
your expertise with this platform.

Where does it segfault?  I mean, where *exactly*?

Andrew.