RE: [ECOS] server_test vmware tribulation

["Vijai Krishnamurthy" <vijai at protosys dot com>]

Andrew, Iztok and all,

I cudn't go down much. Once I reached eth_drv_send
(io/eth/current/src/net/eth_drv.c), I
was tossed around a lot (may be due to lot of context siwtches, I'm not
sure). I put a
breakpoint at lancepci_send
(/devs/eth/amd/lancepci/current/src/if_lancepci.c) and was
expecting to hit but never did.

Reran the whole thing with network debug enabled (verbosity level 6).
Attached please find
the trace (ecos_dump).

Iztok: The ecos appn. was built with the "use DHCP rather than BOOTP"
option,
otherwise I wudn't have hit do_dhcp.

Thanks and regards,
Vijai


-----Original Message-----
From: Andrew Lunn [mailto:andrew.lunn@ascom.ch]
Sent: Friday, November 01, 2002 7:22 PM
To: Vijai Krishnamurthy
Cc: ecos-discuss@sources.redhat.com
Subject: Re: [ECOS] server_test vmware tribulation


> RedBoot defaults to the static IP address and on the screen I can see
these
> prints -
> Can't get BOOTP info, using default IP address
> Ethernet eth0: MAC address 27:39:08:00:00:8e
> IP: 192.168.2.3/255.255.255.0, Gateway: 192.168.2.2
> Default server: 0.0.0.0, DNS server IP: 192.168.2.2
> .
> .
> .
>
> As usual ping and telnets work fine.
>
> At this point I'm little concerned about Redboot's default server being
> 0.0.0.0. Is this
> OK?

Its OK. If you want to use the network to load an image using load -m
tftp etc, you will have to specify the server with -h.

> I couldn't locate a place in ecos.ecc to set this??!!

Neither can I.

> Now from the gdb via serial I load the server_test built on ecos
configured
> for BOOTP/DHCP.
> I go step by step till I hit this line init_all_network_interfaces ->
> do_dhcp -> bring_half_up.
> I step over to see this in tcpdump
> tcpdump -
> 22:40:24.400944 B arp who-has 0.0.0.0 tell 0.0.0.0
>
> Is this arp query OK??

Yes. The interface has just been given the IP address 0.0.0.0 The
stack is trying to find out if there is anybody else on the network
using this IP address. If there is that device should respond to the
arp request. If it receives a reply the stack has a problem, two
devices cannot have the same IP address. Under these conditions it
will probably put the interface down.

>
> As I continue further I observe that a bootp message is being constructed
> and broadcasted (sendto does not
> fail, recvfrom fails) but I don't see anything in the tcpdump.

You see nothing.... Thats strange. You should see a DHCP DISCOVERY
request broadcasted onto the network. Follow the sendto further down
into the device driver. Is the send failling somehow?

     Andrew

Attachment: ecos_dump
Description: Binary data

-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss