[ipxe-devel] result of rom and usb differs (realtek Gbit)

[Oliver Rath]

Hi Robin!

On 27.12.2013 11:11, Robin Smidsrød wrote:
> On 26.12.2013 18:15, Oliver Rath wrote:
>> i have some strange result here: Im starting ipxe-script with dhcp. If i
>> using my ipxe.usb binary, this step seems to work well.
> Could you include the script? Is it embedded or chainloaded?

Embedded. Its simply an

dhcp || sanboot --no-describe --drive 0x80
menu
...
choose ...

So Im waiting one round for dhcp-address and continue with local boot if
the dhcp-server sends no ip-address.

>> But if I flash the same code (built as bin/10ec8168.rom) he doesnt get
>> an ip address. Furthermore, it seems, that the mac-address differs now.
> I would try to perform a packet trace, http://ipxe.org/howto/pcap, in
> both cases and see how the DHCP transaction differs in the two
> situations. I'd also build with DEBUG=realtek to see if the two
> different modes actually end up with different MAC addresses. I'd also
> try both warm boot and cold boot (unplug power) to see if that makes a
> difference. There could be some form of odd reset issue. We've seen it
> before.
Thats a good idea, i will try that.

>
>> In the script, while not getting an ip, i try starting sanboot locally
>> (0x80), but then he gets into an endless ipxe-loop.
> That sounds like a typical chainloading error. See
> http://ipxe.org/howto/chainloading for details on how to get around it.
The looping-problem seems to be solved: I take 0x81 now, it seems
working with local boot. I assume that 0x80 takes the first entry in
boot order which simply is ipxe ;-/, 0x81 seems to take the second entry
in boot order.

>> is this possible or do Im going wrong?
> Maybe, maybe not. Right now it's a bit up in the air. ;)
>
> Happy holidays!
>
Happy holidays to you, too!

Regards,
Oliver