[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: missing __init_array_start, __init_array_end in __libc_csu_init



> > I cannot find a build log for glibc 2.5 later than -3 - that one was still
> > built using binutils 2.17-3, and does not declare a conflict on binutils
> > pre-cvs. What version of binutils were libc6 2.5-10 and -11 built with?
>
> glibc_2.5-11 was built on akire
>
> Toolchain package versions: libc6-dev_2.5-10 linux-kernel-headers_=*=PROVIDED=*= gcc-4.1_4.1.2-12
> g++-4.1_4.1.2-12 binutils_2.17cvs20070426-8 libstdc++6-4.1-dev_4.1.2-12 libstdc++6_4.2-20070516-1
>
> > Anyway, downgrading both gcc and binutils does not help either. Maybe it's
> > a gcc thing after all.
>
> glibc thing?

Right; s/gcc/glibc/ above...

> Should we try building glibc with the latest gcc/binutils?

Either that, or try building gcc with the latest glibc and binutils. I
just downgraded gcc to 4.1.1-21 (old binutils and libc) and that didn't
work either. Well, one difference: it only complains about missing
references to __init_array_start and __init_array_end. Nothing about
nonrepresentable section on output. Nothing about the hidden symbol
__init_array_end not being defined.

Following up on that a bit: keeping binutils 2.17-3 (though libc6-dev
conflicts with it) and gcc 4.1.1-21 while upgrading glibc to 2.5-11 brings
back the nonrepresentable section error as well as the hidden symbol not
defined error.

So it would seem the latter error originates with glibc. The former I'm
unsure about. What is the more important error here - the undefined
reference or the nonrepresentable section?

	Michael



Reply to: