gschafer at zip.com.au
Sat Sep 14 19:06:57 PDT 2002
On Sat, Sep 14, 2002 at 10:31:17PM +0100, Mark Hymers wrote:
> Actually, as a follow up to this one, there is one thing (which may be
> contentious) that I'd like to clarify.
> At the moment in the fileutils instructions we say in one place that
> it's glibc-2.2.3 and above with AMD and in another that it's glibc-2.2.3
> only and AMD.
> AIUI, we've only ever had evidence of it being a glibc-2.2.3 specific
> bug so I've attached a patch to change the other reference to make it
FWIW, I believe the whole warning here is completely bogus. I also believe
that the warning in the glibc section where it says:-
"Also, don't pass the --enable-kernel option to the configure script. It's
known to cause segmentation faults when other packages like fileutils, make
and tar are linked against it."
is directly related to the same issue and thus is also bogus.
Sure, there were definite problems at the time when these warnings went in.
I recently touched on the issue here:-
What we didn't realise at the time was that both glibc-2.2.3 and 2.2.4 had
a buggy configure script which only affected us LFS'ers due to the design
of our build method. We then had the situation where bleeding edge LFS'ers
were building buggy glibc's, then using this bugged LFS as the host to do
further builds and getting bitten with the bugs showing up in fileutils,
make and tar. To complicate matters, there was also the "atexit" issue
whereby buggy gcc's were building buggy glibc's which I suspect is the
reason for the current Ch 5 fileutils patch. I still see folks on the
support list applying this patch even though there is warning in the text
of when and when not to use it.
Folks reading all this waffle may now begin to understand why I spend so
much time working on gcc/glibc testing issues :)
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message
More information about the lfs-dev