[syslinux] PXE2.2 does not work, PXE2.0 does

H. Peter Anvin hpa at zytor.com
Thu Jan 30 11:58:26 PST 2003


Databyte wrote:
> Thank you for your response.
> 
> The older 3Com nics that do work are using the 4.00 MBA version. This MBA is
> using PXE2.
> The newer 3com nics are using MBA 4.30. During boot, they show PXE version
> 2.2
>

This is odd.  MBA 4.30 should work; in fact, I have personally tested
with that version.

> The other visual difference we can see is that the newer 3com's display a
> GUID number (of all zeros). The older 3com's dont show a GUID number. Maybe
> this has something to do with it.
> 
> With the 4.30, the client shows:
> 
> PXELinux 2.00 2002-10-25
> UNDI data segment at:	0009d660
> UNDI data segment size: 3838
> UNDI code segment at:	00090000
> UNDI code segment size:	25A0
> PXE entry point (we hope) at 9D66:00F6
> My IP address seems to be 0AB00166 10.176.1.102
> IP=10.176.1.102:0.0.0.0:10.176.1.254:255.255.255.0
> TFTP prefix: /tftpboot/
> Trying to load: pxelinux.cfg/0AB00166
> Trying to load: pxelinux.cfg/0AB0016
> 
> and so on....
> The time between the tryings is about 30 seconds.
> The only configuration we have is default
>
> On the TFTP server we see the following:
> Connection received from 10.176.1.102 on port 2070
> Read request for file </tftpboot/pxelinux.0>. mode octect
> OACK: <tsize=10820>
> Peer returns ERROR <tftp aborted> -> aborting transfer
> Connection received from 10.176.1.102 on port 2070
> Read request for file </tftpboot/pxelinux.0>. mode octect
> OACK: <blksize=1456>
> <tftboot/pxelinux.0>: sent 8 blks, 10820 bytes in 0s. 0 blks resent

Looks like sending of UDP packets from PXE is broken.

> We are using NT4 DHCP and Tftpd32 by Ph Jourin version 2.21

Do you happen to have a packet capture?

	-hpa





More information about the Syslinux mailing list