[syslinux] tftp problems with VIA EPIA-M board

[Hugo Mills]

On Tue, Jun 03, 2003 at 02:42:16PM -0700, H. Peter Anvin wrote:
> Hugo Mills wrote:
> >    I'm trying to boot linux on a VIA EPIA-M board (a Nehemiah-based
> > M10000), and I'm having some trouble with it. I'm watching the network
> > transactions with ethereal, and I'm seeing a failure at what appears
> > to be the tftp stage.
> > 
> >    The M10000 picks up its IP address from the dhcp server (ISC
> > DHCP3), and reports it correctly on the screen. Then the tftp session
> > starts, and I get the following packets in ethereal:
> > 
> > Client udp/2070 to Server udp/69: Read request /tftpboot/pxelinux.0
> > Server udp/32872 to Client udp/2070: Option acknowledgement, tsize=10832
> > Client udp/2070 to Server udp/32872: Error code, not defined (0), TFTP aborted
> > 
> >    This is with tftpd-hpa. I've tried with atftpd, and I get much the
> > same conversation -- an error code immediately after an option ack.
> 
> That's actually perfectly normal; it's standard behaviour for most PXE
> stacks.

   To abort on receipt of an option acknowledgement? How does it
expect to continue? Should it just carry on regardless after sending
the abort packet back?

> >    The hardware is a VIA-Rhine II (VT6102), claiming "Intel UNDI,
> > PXE-2.0 (build 082)" when the PEX boot sequence starts. Is this a duff
> > PXE implementation, or something stupid I've done?
> 
> No idea; you don't give us enoguh information.

   What additional information do you need?

   Hugo.

-- 
=== Hugo Mills: hugo at ... carfax.org.uk | darksatanic.net | lug.org.uk ===
  PGP key: 1C335860 from wwwkeys.eu.pgp.net or http://www.carfax.org.uk
    --- We demand rigidly defined areas of doubt and uncertainty! ---