Re: Is that a kernel bug?
H.J. Lu (hjl@lucon.org)
Sat, 22 Nov 1997 14:16:06 -0800 (PST)
>
> On Fri, Nov 21, 1997 at 09:24:18PM -0800, H.J. Lu wrote:
> > When new kernel, at least from 2.1.62 and above, I am not longer
> > able to debug the dynamic linker.
>
> I changed the way an ET_DYN program is run. It is now shifted
> up to a known empty spot so that it is not in the way of any
> program it might try to map. This was needed in particular for
> Sparc32, which normally starts the application at 64k -- not
> enough room for ld.so to live at 0.
>
> There are several ways to continue debugging ld.so. One that
> works really the best is to actually link ld.so as an ET_EXEC.
>
How does it work? If the kernel maps an ELF file by its e_type value,
where will the ld.so be mapped to if we make it ET_EXEC?
--
H.J. Lu (hjl@gnu.org)