CLLF with Hard Hat Lunix "Sending BOOTP requests" problem

CLLF with Hard Hat Lunix "Sending BOOTP requests" problem

[rshaw]

I'm trying to get a CLLF 860T to load a kernel normally.  I'm using Embedded
PowerPC Cross Development Kit Rev. 1.0 Hard Hat Linux.

I'm very new to Linux and networking.  I'm running Red Hat Linux 6.2.  When
I installed Red Hat Linux I choose GNOME Workstation.  I found out later
that I needed to install other packages.  I installed dhcpd, inetd, tftp,
bootparamd and bind packages off the Red Hat install CDROM. I also
uninstalled pump.

I ran

chkconfig dhcpd on

so that the DHCP daemon would be loaded on every boot.

While loading the kernel via tftp I found that in order for the boot process
to work I have to use both the 10 Mbps and 100 Mbps ports.  If I do not do
this it times out at "Sending BOOTP requests". I have had to execute the
following script before loading via tftp during which time the 10 Mbps port
is being used.

#!/bin/bash
arp -s 192.168.0.3 00:10:ec:00:00:00
/etc/rc.d/init.d/dhcpd restart
/etc/rc.d/init.d/nfs start

and it responds with

Shutting down dhcpd:                                       [  OK  ]
Starting dhcpd:                                            [  OK  ]
Starting NFS services:                                     [  OK  ]
Starting NFS quotas:                                       [  OK  ]
Starting NFS mountd:                                       [  OK  ]
Starting NFS daemon:                                       [  OK  ]

then when the line "Linux/PPC load: root=/dev/nfs rw" comes up in minicom I
execute the following script so the 100 Mbps port can be utilized

#!/bin/bash
arp -s 192.168.0.3 00:10:ec:80:00:00
/etc/rc.d/init.d/dhcpd restart
/etc/rc.d/init.d/nfs start

and it responds with

Shutting down dhcpd:                                       [  OK  ]
Starting dhcpd:                                            [  OK  ]
Starting NFS services:                                     [  OK  ]
Starting NFS quotas:
Starting NFS mountd:
Starting NFS daemon: nfssvc: Address already in use
                                                           [FAILED]

The following is a screen dump of the minicom session

DRAM1:wwwwwvvvvvCCCCCVVVVV CaV NVRAM: CaV
--------------------------------------------------
RPX Utility program, Copyright (C) 1998-1999, RPCg LLC, All Rights Reserved.
  Reset(0100): ExHard ExSoft, TESR:0000 Core.Pv/Cp=0020.0050/0004
  BCSR : OK
  RTC  : OK, battery LOW or absent, tics/sec=1, time = 5 sec
  SPIp : OK, I2Cp: OK, STTM: OK, 41.5c, EEcfg: OK, 256b, cs:1D55
  ENET : [zero], 0010EC000000, INET: C0A80003
  FLASH: OK, 1*16 Mb(32x8), id=01:53, cs:F95D es:F95D OK (FFF00000:FFF46200)
  NVRAM: OK, 0 Kb, NVR battery GOOD
  DRAM : OK, 16 Mb, 10 columns
===> RPX Main
     RPX-ClassicLF AY/BW 860T 50Mhz/8Mhz Ic ~Dc RPXU v1.46
  1  test BCSR                    A  Add-on board tests
  2  test dip switches            B  rpxBug command line (ctrl-B)
  3  test LEDs                    C  Configure EEPROM
  4  test SPI I2C STTM EEPR,CFG   L  Load RAM
  5  test FLASH firmware chksum   O  On-board I/O tests
  6  test NVRAM/SRAM              P  Program FLASH
  7  test DRAM (destructive)      R  hard Reset
  8                               S  Scan FLASH for executables
  9  run all tests                T  display Time & Temperature
RPX Main menu: press 123456789 or ABCLOPRST key : <?> : L
===> RPX Load: (ESC - returns to Main menu)
  1  Copy one memory range to another range
  2  Verify one memory range against another range
  3  Fill memory range with generated data
  4
  5  load   using S-records received from serial monitor port
  6  verify using S-records received from serial monitor port
  7  load   using tftp via Ethernet interface
  8  set a start address
  9  transfer control to start address
  T  tftp & start using config info with retries
RPX Load menu: press 123456789T key : <?> : T
 target IP address : C0A80002
 target file name  : z.cllf
 address bias      : 0
Starting tftp download:
inetARP(): OK
.1.2.4.8.16.32.64.128.256.512.1024.2048.4096.8192
load_tftp(): OK
1065943 data bytes, 13494 S-records, 2082 tftp blocks, 0 repeated blocks
start address = 00200000
load_tftp_toMem(): OK
Transferring control to downloaded program
loaded at:     00200000 0020C580
relocated to:  00100000 0010C580
board data at: 001001C4 001001E0
relocated to:  00200100 0020011C
zimage at:     00207000 0026810B
avail ram:     00269000 01000000

Linux/PPC load: root=/dev/nfs rw
Uncompressing Linux...done.
Now booting the kernel
Linux version 2.2.13 (root@BOBLINUX) (gcc version 2.95.2 19991024 (release))
#50
Boot arguments: root=/dev/nfs rw
time_init: decrementer frequency = 180000000/60
Calibrating delay loop... 47.82 BogoMIPS
Memory: 15160k available (736k kernel code, 452k data, 36k init)
[c0000000,c100]
DENTRY hash table entries: 262144 (order: 9, 2097152 bytes)
Buffer-cache hash table entries: 32768 (order: 5, 131072 bytes)
Page-cache hash table entries: 4096 (order: 2, 16384 bytes)
POSIX conformance testing by UNIFIX
Linux NET4.0 for Linux 2.2
Based upon Swansea University Computer Society NET3.039
NET4: Unix domain sockets 1.0 for Linux NET4.0.
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP, IGMP
TCP: Hash tables configured (ehash 16384 bhash 16384)
Starting kswapd v 1.5
CPM UART driver version 0.03
ttyS00 at 0x0280 is a SMC
ttyS01 at 0x0380 is a SMC
ttyS02 at 0x0100 is a SCC
ttyS03 at 0x0200 is a SCC
pty: 256 Unix98 ptys configured

loadFPGA successful!

RAM disk driver initialized:  16 RAM disks of 4096K size
loop: registered device at major 7
eth0: FEC ENET Version 0.1, 00:10:ec:80:00:00
PPP: version 2.3.7 (demand dialling)
TCP compression code copyright 1989 Regents of the University of California
PPP line discipline registered.
fec: Phy @ 0x17, type 0x01814401
fec: link down
fec: 100 Mbps, Half-Duplex
fec: link down, 100 Mbps, Half-Duplex
Sending BOOTP requests....fec: link up, auto complete, 10 Mbps, Half-Duplex
. OK
IP-Config: Got BOOTP answer from 192.168.0.2, my address is 192.168.0.3
Looking up port of RPC 100003/2 on 192.168.0.2
Root-NFS: Unable to get nfsd port number from server, using default
Looking up port of RPC 100005/1 on 192.168.0.2
VFS: Mounted root (NFS filesystem) readonly.
Freeing unused kernel memory: 36k init
INIT: version 2.77 bootiÿ
INIT: cannot execute "/etc/rc.sysinit"
INIT: Entering runlevel:.
sh-2.03#

OK, This gets us to a Linux shell prompt even though I had to execute the
script files using both the 10 Mbps and 100 Mbps ports, which I shouldn't
have to do. I'm hoping someone might be able to help me with why I need to
use both ports and execute these arp commands.  Is something wrong with my
DHCP configuration?

Here are some of my configuration files.

Here is /etc/dhcpd.conf file contents

allow bootp;
subnet 192.168.0.0 netmask 255.255.0.0 {
  default-lease-time 1209600;   #two weeks
  max-lease-time 31557600;   #one year
  group {
    host CLLFBOARD {
      hardware ethernet 00:10:EC:00:00:00;
      fixed-address 192.168.0.3;
      option root-path "/opt/hardhat/devkit/ppc/8xx/target";
    }
  }
}

Here is the /etc/hosts file contents

127.0.0.1 localhost.localdomain localhost
192.168.0.1 win2000 w2k
192.168.0.2     RANDYLINUX rlx
192.168.0.3     CLLFBOARD cllf

Here is the /etc/hosts.allow file contents

portmap: 192.168.0.

The /etc/hosts.deny file is empty

Here is the /etc/inetd.conf file contents

tftp dgram udp wait root /usr/sbin/tcpd in.tftpd
/opt/hardhat/devkit/ppc/8xx/target/boot
linuxconf stream tcp wait root /bin/linuxconf linuxconf --http
#swat      stream  tcp     nowait.400      root /usr/sbin/swat swat

Here is a minicom screen dump showing the CLLF EEPROM configuration

RPX Configure menu: press 123456789 key : <?> : 7
Board          : F CCP CCE                : <F>            :
Processor      : 821 855 860              : <860>          :
Variant        : SR EN T MH DT E SE P DP . : <T>            :
Freq cpu/xtal  : 40/8 50/8 66/8           : <50/8>         :
Enable Caches  : N I                      : <I>            :
DRAM columns   : 8 9 10 11 12 13          : <10>           :
DRAM1 size(Mb) : 0 4 16 32 64 128         : <16>           :
NVRAM size(Kb) : 0 32 128 512             : <0>            :
Ethernet addr  : [format:hex]             : <0010EC000000> :
IP address     : [format:hex]             : <C0A80003>     :
Mon Baud/100   : 3 12 24 96 384 576 1152  : <96>           :
STTM bias(C*10): [format:dec]             : <131>          :
AutoBoot name  : [format:asc]             : < >            :
TFTP IP addr   : [format:hex]             : <C0A80002>     :
TFTP filename  : [format:asc]             : <z.cllf>       :
low except vec : N Y                      : <[missing]>    :
no twirl prompt: N Y                      : <[missing]>    :
inv monitor SMC: N Y                      : <[missing]>    :
Lock KAPWR regs: N Y                      : <[missing]>    :
RPX Configure menu: press 123456789 key : <?> : ESC

Thanks for any help!

Randy Shaw <rshaw@graftononline.net>


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: CLLF with Hard Hat Lunix "Sending BOOTP requests" problem

[Matthew Locke]

rshaw@graftononline.net wrote:
>
> I'm trying to get a CLLF 860T to load a kernel normally.  I'm using Embedded
> PowerPC Cross Development Kit Rev. 1.0 Hard Hat Linux.

you might want to upgrade to 1.2. it is on our ftp site.  It is not
nescessary, but it is always good to use the latest stable stuff.

> While loading the kernel via tftp I found that in order for the boot process
> to work I have to use both the 10 Mbps and 100 Mbps ports.

the EP boot loader uses the 10Mbps port for tftp loading.  You don't use
the 10/100 for tftp.

>If I do not do
> this it times out at "Sending BOOTP requests". I have had to execute the
> following script before loading via tftp during which time the 10 Mbps port
> is being used.

you only need to use the 10Mbps port, you don't need the 100Mbps port
connected  to the network.

>
> #!/bin/bash
> arp -s 192.168.0.3 00:10:ec:00:00:00

that ethernet address is wrong.  This may be causing some of your
problems.  Contact Embedded Planet support for help to reprogram the
correct enet address for your board


> OK, This gets us to a Linux shell prompt even though I had to execute the
> script files using both the 10 Mbps and 100 Mbps ports, which I shouldn't
> have to do. I'm hoping someone might be able to help me with why I need to
> use both ports and execute these arp commands.  Is something wrong with my
> DHCP configuration?

look into changing the enet address to the right one.   the
configuration looks fine.  I hope the w2k host is not really running
win2k.  If it is, all bets are off.


> Here is the /etc/hosts file contents
>
> 127.0.0.1 localhost.localdomain localhost
> 192.168.0.1 win2000 w2k
> 192.168.0.2     RANDYLINUX rlx
> 192.168.0.3     CLLFBOARD cllf
>

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: CLLF with Hard Hat Lunix "Sending BOOTP requests" problem

[Graham Stoney]

rshaw@graftononline.net writes:
> While loading the kernel via tftp I found that in order for the boot process
> to work I have to use both the 10 Mbps and 100 Mbps ports.

Which version of the ROM monitor are you using?  The old one couldn't download
over the 100 Mbps port, so we had to connect both.  We used the 10 Mbps port
to download the kernel, and the 100 Mbps port once the kernel comes up.  Of
course, you need a 10/100 switch to do this.  I believe that Embedded
Planet are working on a new ROM monitor, which hopefully will correct this
problem; I don't know if they're shipping it with production boards yet
though.

Note that there is a hack in the fec driver which munges the ethernet MAC
address for the FEC port by adding 80:00:00 to the number that the monitor
reports on the CPM port.  You'll need to configure your dhcp & nfs server to
recognise the board via the munged address.  Your tftp server on the other
hand needs to recognise the unmunged address.

> OK, This gets us to a Linux shell prompt even though I had to execute the
> script files using both the 10 Mbps and 100 Mbps ports, which I shouldn't
> have to do. I'm hoping someone might be able to help me with why I need to
> use both ports and execute these arp commands.  Is something wrong with my
> DHCP configuration?
>
> Here are some of my configuration files.
>
> Here is /etc/dhcpd.conf file contents
>
> allow bootp;
> subnet 192.168.0.0 netmask 255.255.0.0 {
>   default-lease-time 1209600;   #two weeks
>   max-lease-time 31557600;   #one year
>   group {
>     host CLLFBOARD {
>       hardware ethernet 00:10:EC:00:00:00;
>       fixed-address 192.168.0.3;
>       option root-path "/opt/hardhat/devkit/ppc/8xx/target";
>     }
>   }
> }

You're using the unmunged address here, where it should be 00:10:EC:80:00:00.
This is probably why your bootp server doesn't respond.  By the way, they
didn't _really_ sell you a board with a MAC address ending in all-zero, did
they?

Once you've got it set up, your arp scripts will be unnecessary.

Regards,
Graham
--
Graham Stoney
Principal Hardware/Software Engineer
Canon Information Systems Research Australia
Ph: +61 2 9805 2909  Fax: +61 2 9805 2929

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: CLLF with Hard Hat Lunix "Sending BOOTP requests" problem

[Dan Malek]

Graham Stoney wrote:

> Which version of the ROM monitor are you using?  The old one couldn't download
> over the 100 Mbps port, so we had to connect both.  We used the 10 Mbps port
> to download the kernel, and the 100 Mbps port once the kernel comes up.

The "old" one could download over the 10/100 port.  It was an option
through the test menu.  The new one won't download over the 10/100 port,
this was in response to customers asking for the smallest possible boot
rom image.

> Note that there is a hack in the fec driver which munges the ethernet MAC
> address for the FEC port by adding 80:00:00 to the number

This will probably remain as the method for finding a MAC address
for the second port on the CLLF.  Embedded Planet has said they will
sequentially update the single MAC address in the on-board EEPROM, and
we can safely do this.  However, I think it is always wise for anyone that
creates Ethernet devices to request their own MAC addresses from the
IEEE and use those.  It allows flexibility and control during production.

> .........  By the way, they
> didn't _really_ sell you a board with a MAC address ending in all-zero, did
> they?

I have seen prototype and semi-custom boards from Embedded Planet
that have been tested in their engineering department and rushed to
customers (i.e. bypassed the manufacturing floor) arrive with all
zeros in the MAC address (at least the last three bytes).  I forget
the magic rom command to set these.  Call their support staff and
they will straighten it out.


	-- Dan

---

	I like MMUs because I don't have a real life.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: CLLF with Hard Hat Lunix "Sending BOOTP requests" problem

[Graham Stoney]

Hi Dan,

Dan Malek writes:
> The "old" one could download over the 10/100 port.  It was an option
> through the test menu.

Sure, the option is there, but I couldn't ever get it to actually work.
Tried real hard too; even with help from EP's tech support I couldn't get it
to go.  Eventually gave up and connected both ports.

> The new one won't download over the 10/100 port, this was in response to
> customers asking for the smallest possible boot rom image.

Sounds like my protestations about 100 Mbps download being broken didn't get
through their tech support firewall at all.  I may have stopped hassling
their tech support about it constantly once our workaround was going, but I
certainly did ask them to fix it.

> > Note that there is a hack in the fec driver which munges the ethernet MAC
> > address for the FEC port by adding 80:00:00 to the number
>
> This will probably remain as the method for finding a MAC address
> for the second port on the CLLF.  Embedded Planet has said they will
> sequentially update the single MAC address in the on-board EEPROM, and
> we can safely do this.

This works fine, so long as the user realises that the MAC address that the
ROM monitor reports isn't exactly the one that the FEC port will use, and
configures their DHCP server correctly.

Regards,
Graham
--
Graham Stoney
Principal Hardware/Software Engineer
Canon Information Systems Research Australia
Ph: +61 2 9805 2909  Fax: +61 2 9805 2929

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: CLLF with Hard Hat Lunix "Sending BOOTP requests" problem

[Dan Malek]

Graham Stoney wrote:

> Sure, the option is there, but I couldn't ever get it to actually work.

Hmm...I guess I was just lucky.

> ... Eventually gave up and connected both ports.

I did too, because the 10 Mbit had the "autostart" option.  Just
one menu key and you are off and running.

> Sounds like my protestations about 100 Mbps download being broken didn't get
> through their tech support firewall at all.

Well, to be a little fair, that was kind of an engineering hack that
was never intended to be "supported".  I am not trying to justify
their responses they may have given to you, but considering it was
never documented as a feature......

The hardware worked, there was a documented method for downloading
over the Ethernet (although it was the 10 Mbit port), and connecting
two cables solved the "problem" for Linux.

> This works fine, so long as the user realises that the MAC address that the
> ROM monitor reports isn't exactly the one that the FEC port will use, and
> configures their DHCP server correctly.

Again, this is something unique to the Linux driver, so it is probably
up to us (me :-) to ensure it is documented somewhere more clearly.
At least it is printed during boot up, which is usually the way I
find the MAC addresses on most systems anyway.


	-- Dan

--

	I like MMUs because I don't have a real life.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/