> Well, as far as I can tell, on Intel this is caused by libc using
> uint64_t as a dev_t, and the kernel using short as dev_t. I created
> some fixes in the nfs-0.4.22 stuff, and now it works quite well. Under
> the provision that the client does not try to establish the
> whereabouts from a server by sending a null packet to the
> server. (Ultrix 4.3 does this unfortunately)
Where could I get your patch ? I am extremely interested with that...
(fortunately, we have no Ultrices, only Solaris 2.5.1 clients. In the
interim, I'll happy to let your modifs have a go...)
> cc> I have no convenient access to a i386 PC I could recompile Linux on, to
> cc> check whether this problem is specific or not. Has anyone succesfully used
> cc> the knfsd on recent kernel, on any platform ?
> I think this is not arch related....
Well, someone told me a few hours after my initial post that on his i386
and Alphas, that was working fine. Maybe a slightly different casting
convention from short to uint64, across architectures.
> cc> Software :
> cc> S/Linux 2.1.89-VGER-CVS980313
> cc> linux-nfs 0.4.22 from ftp.mathematik.tu-darmstadt.de/pub/okir
> What's your base distribution? Redhat?
Yes. RH4.2 off ftp.lip6.fr (but pretty everything has been recompiled
since, I just kept python1.4 together with python1.5, short of having
found their RPM support module).
(now using the 2.1.91 kernel)
-- Cyrille
------------------------------
Zog Zog
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu