fix for seg faults in static fileutils-4.1

[markh@…]

mca came up with this fix which solves the segfault problem that's been seen when anything like ls is used in chroot having been compiled in chapter 5. So we should add this to the install instructions for fileutils-4.1 in static only unless anyone has any objections/problems.

sed 's/\(.*\)\(fopen-safer\.c \)
/\1\2atexit.c
/' lib/Makefile.in > tmp && sed 's/\(.*\)\(idcache\$U\.\$.*\)
/\1\2atexit$U.$(OBJEXT)
/' tmp > lib/Makefile.in

[markh@…]

I've put this in cvs.

[gerard@…]

The command explanation section hasn't been updated yet.

We're suspecting this only affects AMD systems, I will build on Intel tomorrow (Wednesday 8th of august) and see how it affects me.

[markh@…]

I'm not convinced that even if we *think* it only affects AMD systems, we shouldn't just leave it in anyways if it doesn't cause problems on Intel etc... It's only for that static build anyways after all.

[gerard@…]

I got an email a minute ago from somebody who just upgraded to an Athlon and starting getting the seg faults. That's 7 or 8 people I know of who have this problem and they all are using some kind of AMD. Not a coincidence anymore.

Just leave the sed in the book and add in the explanation that this fixes a problem if you run an AMD system.

Btw why doesn't this happen in chroot? I understood this is a Glibc-2.2.3 related problem so should happen inside chroot too? Or does it happen inside chroot too if you were to link statically (so that makes it a strictly static link problem)

[markh@…]

Hmm. I think it's a static linking problem. It certainly looks like it. Can I just clarify - you want me to leave the sed in but put in the explanation that it is to fix a bug which occurs on AMD systems?

[gerard@…]

Yes that's what I meant

[markh@…]

I've put the explanation (including AMD comment) in the book.