[U-Boot-Users] TFTP with U-Boot

[Jerry Van Baren <gerald.vanbaren@smiths-aerospace.com>]

Robin Randhawa wrote:
> Hi Midhun.
> 
> The tftp client seems to be timing out.
> 
> Some tips that might shed light on your problem :

If you have not done this already, make sure your server is serving the 
file to itself and another computer before trying to use u-boot to get 
the file.

a. On your linux server, do "tftp localhost" and get the target file.

b. On a separate machine, do "tftp 222.222.101.117" (your server_ip per 
below) and get the target file.

You also have a link status "Jun  6 12:17:50 kernel: e100: eth0: 
e100_watchdog: link up, 10Mbps, half-duplex".  Is 10Mbps half duplex 
correct for your link?  If not, you are likely having an autonegotiation 
problem - 10/half is the "other side is stupid, I gotta guess" 
configuration.  If this is wrong, you probably have a PHY configuration 
problem (likely the MII connection to the PHY - H/W and/or S/W).

gvb


> 1. For testing purposes, ensure that you only have the host and the
> target board on the network, either via a crossed ethernet cable or a
> hub.
> 
> 2. As a quick test, run /usr/sbin/in.tftpd "by hand", ie not via the
> init.d scripts and see if that makes a difference.
> 
> 3. Triple-check your portmapper setup. Invariably, tftp is clamped in
> the hosts.allow/deny framework by default.
> 
> 4. Use wire-shark (formerly ethereal) to snoop on the wire and see
> what's going on. This is probably your best bet.
> 
> Cheers,
> Robin
> 
> On Wed, 2007-06-06 at 12:32 +0530, Midhun Agnihotram wrote:
>> Hi All,
>>
>>   I am still not able to do a tftp. I have checked and rechecked my
>> server settings according to the DULG manual.
>>
>> **  I am using a Fedora Core 6 as my host. My /etc/xinetd.d/tftp is as below:
>>
>> service tftp
>> {
>>         disable = no
>>         socket_type             = dgram
>>         protocol                = udp
>>         wait                    = yes
>>         user                    = root
>>         server                  = /usr/sbin/in.tftpd
>>         server_args             = -s /tftpboot
>>         per_source              = 11
>>         cps                     = 100 2
>>         flags                   = IPv4
>> }
>>
>> ** When I start xinetd :
>>
>> # /etc/init.d/xinetd start
>> Starting xinetd:                                           [  OK  ]
>>
>> ** The netstat gives ;
>> # netstat -a | grep tftp
>> udp        0      0 *:tftp                      *:*
>>
>> ** The /etc/services file also contains the correct the port number too.
>>
>> tftp            69/tcp
>> tftp            69/udp
>>
>> ** cAll seems to be fine. But still the tftp does not work.
>>
>> => printenv
>> bootargs=mem=32M console=ttyS0,115200 root=/dev/nfs rw init=/sbin/init ip=on
>> bootdelay=3
>> baudrate=115200
>> netmask=255.255.0.0
>> bootfile="uImage"
>> ethaddr=00:23:31:35:01:56
>> ipaddr=222.222.101.124
>> serverip=222.222.101.117
>> stdin=serial
>> stdout=serial
>> stderr=serial
>>
>> Environment size: 258/131068 bytes
>> => ping 222.222.101.117
>> host 222.222.101.117 is alive
>> => tftp c0200000 firstimg
>> TFTP from server 222.222.101.117; our IP address is 222.222.101.124
>> Filename 'firstimg'.
>> Load address: 0xc0200000
>> Loading: T T T T T T T T T T
>> Retry count exceeded; starting again
>> TFTP from server 222.222.101.117; our IP address is 222.222.101.124
>> Filename 'firstimg'.
>> Load address: 0xc0200000
>> Loading: T T T T T T T T T T
>> Retry count exceeded; starting again
>>
>> ** When I connect the ethernet cable of the target to the host, the
>> /var/log/messages shows :
>>
>> Jun  6 12:17:42 kernel: e100: eth0: e100_watchdog: link down
>> Jun  6 12:17:50 kernel: e100: eth0: e100_watchdog: link up, 10Mbps, half-duplex
>>
>>
>>
>>         Can anybody please let me know what could be the problem? Has
>> anybody used Fedora as their host for U-boot? Is it the problem with
>> the host or the target?
>>
>>
>> Thanks a lot,
>> Midhun Agnihotram