Re: m68k toolchain status?
> > Anyway, I remeber we investigated the ld failure with a simple test case
> > at that time and could not find out where the problem originated. Can't
> > find the mails right now, but I think Stephen came up with the test case
> > (mid-June).
> This pretty much sums up our current problems.
> void main(void)
> printf("Hello world\n");
> $ gcc -pie hello.c
> | /usr/bin/ld: a.out: hidden symbol `__init_array_end' isn't defined
> | /usr/bin/ld: final link failed: Nonrepresentable section on output
> | collect2: ld returned 1 exit status
Right - though -pie is not even used in xulrunner. Maybe this is just the
easiest way to force a hidden symbol during the link stage in the test
The symbol in the xulrunner case is from nsNSSComponent.o which is neither
the first input object nor the last. Other than the parameter types of
nsHashtable do not seem to match the class definition exactly, I cannot
see what's special about this. The (static) library actually containing
the nsHashtable code does not seem to get linked in FWIW.
I'm out of ideas (and time),