gij runs HelloWorld on PPC Linux

Andrew Haley
Mon Feb 12 10:19:00 GMT 2001

Tom Tromey writes:
 > >>>>> "Jeff" == Jeff Sturm <> writes:
 > Jeff> I agree it's a special case.  There was an analogous fix for the compiler:
 > Jeff>
 > Hmm, I seem to remember that :-)
 > I checked in a variant of your patch.
 > Jeff> Odd thing is, the compiler doesn't seem to have anything like
 > Jeff> NULLCHECK.  Maybe it isn't needed on any interesting platforms.
 > Oh, it is.  Andrew has a patch for this but I don't know how heavily
 > it has been tested yet though.  We never got around to implementing it
 > because we've generally been using systems where you can catch the
 > segv.

I intend to post the patch, which is effectively a generalization of
the one above, in the next day or so.  It will allow us to run on all
targets without needing sjlj exceptions or aynchronous exceptions, so
reducing initial porting effort.  Indeed, on embedded OSes that don't
allow catching SEGV this will make running gcj possible.

I think this should be the default on targets that we haven't done any
porting work for.

However, this patch doesn't help with null checks in native methods.
I'm going to have a look at ways of doing that.


More information about the Java mailing list