This time I ran under X, with a console window and saw
NFS: error 55, which is ENOBUFS.  So does NFS just hang
if it runs out of buffers???

>>>>> On Tue, 04 Oct 1994 10:04:22 -0700, I said:

    DG> I am running NetBSD-current on a PC attached to a SUN via
    DG> 19200 baud serial cable.  Everything is wonderful except
    DG> NFS, which crashes after transferring 30-40K of stuff, and
    DG> the PC freezes.

    DG> I have a cheapo serial card with no fifo, and I see error messages
    DG> "3 silo overflows", where 3 is variable.  I conclude that PPP
    DG> is not error-resistant??? though I see ack/nak and error checking
    DG> code.

    DG> I'm running PPP-2.1.2 on the Sun side, and the distributed PPP on
    DG> the PC side.

    DG> Kermit transfers work fine (that's how the whole 200 megs of Unix
    DG> and X got onto the PC), though they're about twice as fast under
    DG> DOS as under UNIX.

    DG> So:

    DG> 1) is there a recommended fix?

    DG> and 

    DG> 2) if not, is there a recommended approach for debugging?