This question is particularly on NFS implementation on HPUX. I
am writing a kernel module to access some files in all supported file
systems in HPUX , it includes NFS (version 2 & 3 both). So there are
multiple issues I am facing incase of NFS, which were working fine for
local file systems, and I am wondering if the implementation is
something drastically different than what I can see in opensolaris
code. Following are the issues I am facing -

1. I get an NFS vp, and I am trying to get a complete filename out of
it, well I have a routine written for it which fetches the filename
from DNLC and works fine 99% of the cases. But for NFS, this routine
doesnt work, specially when I am creating a file, or mostly when I have
done an mkdir operation . I have hooked up mkdir syscall, and just
after vop_mkdir() call I take the vp and run my routine on it... BUT I
dont get any pathname (I never get a pathname)... it probably means
that this vp is stale for NFS. So is there a way that I can refresh my
file handle for NFS ????

2. I am calling VOP_FID() to get the fid related to a vp. This guy
always returns me return code as 71, which is EREMOTE .. I dont know
why. Well I figured out from the debugger that vn_fid pointer in
vnodeops points to nfs_noop in case NFS... why is so ?? I thought fid
makes maximum sense only in case of cachefs or in case of NFS only ?
Well getting this fid is quite critical for my module, so is there any
other way I can get an fid ??

Hoping to get an answer here....