Has anyone out there gotten ftape-3.x or 4.x to work with kernel
2.0.36? 2.0.35? What's the latest kernel they've worked with in
the 2.0.x series?
I ask because I'm seeing something very strange. When I load
ftape.o into memory, and dump the kernel's symbol table with
"ksyms", ALL of the exported ftape symbols are mapped to the same
(highly bogus) address as far as the kernel is concerned.
This does not appear to happen with other modules, only ftape.o.
(In fact, I can load another unrelated module on top of ftape.o,
and its addresses are fine, but ftape's are still bogus. I
can also reverse the process.) Also, when I run "insmod -m"
to get a load map, the addresses in the map are totally legitimate.
Of course, when I go to load zftape.o (or any other stacked
module) on top of ftape.o, it blows up because all of the
ftape calls are at bad addresses. (I've confirmed this by
dumping the addresses of about 20 ftape functions from inside
zftape's initialization routine.)
I'm really beginning to think I'm on the trail of a kernel bug
here.
Brad Kaiser
([EMAIL PROTECTED])
