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

Re: prelink ready binutils



On Sat, 6 Oct 2001, Jack Howarth wrote:

>     If you aren't aware Jakub Jelinek <jakub@redhat.com> is working
> on a prelinker for multiple archs on linux. He has work finished on 
> many archs ready. There are however changes to improve performance of
> the resulting prelinking that have been placed in binutils cvs.
> Would you consider updating binutils to the latest version just
> before the next Debian Woody release is to be finalized and rebuilding
> all of Woody against this newest binutils? The result will be that

I'm packaging HJ Lu's latest (synced with 2001-10-05 CVS) right now.  From
what I can tell, it include prelinking support for ELF/i386 and ELF/PPC,
but I'm not sure about the other archs yet, nor if there's any extra
things that I need to do in order to enable the additional
functionality.  I'm checking into it now.

As for recompiling Woody, that's an awfully ambitious goal, especially
considering how many archs would be affected (it may take a LONG while to
do so).

> we should have a prelinker friendly version of the Woody distro which
> would allow users to only need update their glibc to a prelinker 
> friendly version as well as install prelink to have all of their 
> binaries prelinked. This sounds like something well worth doing.

It probably would be worth it.  Do we have any concrete numbers on how
much of a benefit this provides us?

> I know that it is too late in the Woody development cycle for us
> to add in prelink support however we can prepare many arches to
> be prelink ready this way. Let me know if you are interested and
> I will check with Jakub to discover when he thinks binutils is
> in the best state for prelink to work on the most archs. Thanks
> in advance for considering this request.

I think we should make this a goal for sid for now.  It's a bit late in
the release cycle for Woody proper, but perhaps a point release of woody
can provide this.

I'm cc'ing the rest of the gcc and glibc maintainers in case they have
any thoughts about this.

C




Reply to: