> Well, since I know some files that will do it, here's the trace from
> one of them, hope it helps.
Hi Tony,
I looked through the NFS debug trace you posted, but couldn't spot
anything that looked wrong. The write that generates the "bad getargs"
complaint seems to be the last block of the file, and it appeared to be
written out of order. This shouldn't be a problem, but it's possible
that this is what the Unixware server is complaining about.
One other thing seemed a little odd, though not necessarily incorrect:
at one point a write at offset 25k is still pending even though the file
position is then much greater.
You mentioned that the problem seems to happen with certain files, so it
would be helpful if you could see what they have in common. Is it
certain sizes, or maybe a certain size modulo the page size (4096)?
Suppose you take the file that triggered the problem and round its size
up to the next even page multiple -- does it still trigger the error?
Also, if anyone else on the kernel list has seen a "bad getargs" error
from a Unixware NFS server, please let me know.
Regards,
Bill