Re: [ECOS] N00blet question on networking

Re: [ECOS] N00blet question on networking

[Laurie.Gellatly]

Hi Chuck,
As Redboot and your application share the same NIC they are distinguished by
different IP addresses.
I would guess that the DHCP server, receiving a second request from the same
MAC, will simply reply with the same address again (maybe extending the
lease on the IP).
Have you tried making one of Redboot or the application use a static
address?

...Laurie:{) 

--------- Original Message --------
From: Chuck McManis <ecos@mcmanis.com>
To: ECOS Discussion Group <ecos-discuss@sources.redhat.com>
Subject: [ECOS] N00blet question on networking
Date: 21/02/07 02:21

> Ok, so I have redboot working. Further I can build a libtarget.a with a 
> basic set of options, I tried the simple program to get data from the 
> console and echo it back, and I can boot redboot, 'load -h &lt;my
server&gt; 
> simple' and 'go' to run it. All's cool.
> 
> So I used ecosconfig to add in CYGPKG_NET and FREEBSD_STACK and
ETH_DRIVERS
> (actually there was a point where I had no conflicts but a couple of 
> ethernet driver files were missing, its a CDL bug but I've not assured 
> myself that I know enough about what I'm doing to call it a bug outright 
> yet ;-)
> 
> Now I can build the httpd1.c test and when I run it this is what I get:
> ----------------------------------------------------
> Ethernet eth0: MAC address 00:0a:00:c0:ff:ee
> IP: 192.168.110.124/255.255.255.0, Gateway: 192.168.110.254
> Default server: 0.0.0.0
> 
> RedBoot(tm) bootstrap and debug environment [GRUB]
> Non-certified release, version UNKNOWN - built 21:38:48, Feb 19 2007
> 
> Platform: PC (I386)
> Copyright (C) 2000, 2001, 2002, 2003, 2004 Red Hat, Inc.
> Copyright (C) 2003, 2004, 2005, 2006 eCosCentric Limited
> 
> RAM: 0x00100000-0x040f0000, [0x0018da10-0x01000000] available
> RedBoot&gt; load -h 192.168.110.30 httpd1
> Using default protocol (TFTP)
> Entry point: 0x00208000, address range: 0x00208000-0x00244720
> RedBoot&gt; go
> [cyg_net_init] Init: mbinit(0x00000000)
> [cyg_net_init] Init: cyg_net_init_devs(0x00000000)
> Init device 'vt8235_eth0'
> [cyg_net_init] Init: loopattach(0x00000000)
> [cyg_net_init] Init: ifinit(0x00000000)
> [cyg_net_init] Init: domaininit(0x00000000)
> [cyg_net_init] Init: cyg_net_add_domain(0x00243f80)
> New domain internet at 0x00000000
> [cyg_net_init] Init: cyg_net_add_domain(0x002436c0)
> New domain route at 0x00000000
> [cyg_net_init] Init: call_route_init(0x00000000)
> [cyg_net_init] Done
> [eth_drv_ioctl] Warning: Driver can't set multi-cast mode
> [eth_drv_ioctl] Warning: Driver can't set multi-cast mode
> 
> ----------------------------------------
> 
> Ok, but when I do this, there doesn't seem to be any way to talk to my 
> target board. Eventually it prints :
> 
> BOOTP/DHCP failed on eth0
> [eth_drv_ioctl] Warning: Driver can't set multi-cast mode
> [eth_drv_ioctl] Warning: Driver can't set multi-cast mode
> 
> Which seems unusual given that DHCP worked fine for Redboot.
> 
> So here are some questions:
> 
> What should this test print when it works? Should I be able to ping my 
> target after using init_all_network_interfaces() ? What is different 
> between Redboot's use of my driver and ECOS's use of the driver?
> 
> --Chuck
> 
> 
> -- 
> Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
> and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss
> 
> 

________________________________________________
Message sent using UebiMiau 2.7.9



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

[ECOS] N00blet question on networking

[Chuck McManis]

Ok, so I have redboot working. Further I can build a libtarget.a with a 
basic set of options, I tried the simple program to get data from the 
console and echo it back, and I can boot redboot, 'load -h <my server> 
simple' and 'go' to run it. All's cool.

So I used ecosconfig to add in CYGPKG_NET and FREEBSD_STACK and ETH_DRIVERS
(actually there was a point where I had no conflicts but a couple of 
ethernet driver files were missing, its a CDL bug but I've not assured 
myself that I know enough about what I'm doing to call it a bug outright 
yet ;-)

Now I can build the httpd1.c test and when I run it this is what I get:
----------------------------------------------------
Ethernet eth0: MAC address 00:0a:00:c0:ff:ee
IP: 192.168.110.124/255.255.255.0, Gateway: 192.168.110.254
Default server: 0.0.0.0

RedBoot(tm) bootstrap and debug environment [GRUB]
Non-certified release, version UNKNOWN - built 21:38:48, Feb 19 2007

Platform: PC (I386)
Copyright (C) 2000, 2001, 2002, 2003, 2004 Red Hat, Inc.
Copyright (C) 2003, 2004, 2005, 2006 eCosCentric Limited

RAM: 0x00100000-0x040f0000, [0x0018da10-0x01000000] available
RedBoot> load -h 192.168.110.30 httpd1
Using default protocol (TFTP)
Entry point: 0x00208000, address range: 0x00208000-0x00244720
RedBoot> go
[cyg_net_init] Init: mbinit(0x00000000)
[cyg_net_init] Init: cyg_net_init_devs(0x00000000)
Init device 'vt8235_eth0'
[cyg_net_init] Init: loopattach(0x00000000)
[cyg_net_init] Init: ifinit(0x00000000)
[cyg_net_init] Init: domaininit(0x00000000)
[cyg_net_init] Init: cyg_net_add_domain(0x00243f80)
New domain internet at 0x00000000
[cyg_net_init] Init: cyg_net_add_domain(0x002436c0)
New domain route at 0x00000000
[cyg_net_init] Init: call_route_init(0x00000000)
[cyg_net_init] Done
[eth_drv_ioctl] Warning: Driver can't set multi-cast mode
[eth_drv_ioctl] Warning: Driver can't set multi-cast mode

----------------------------------------

Ok, but when I do this, there doesn't seem to be any way to talk to my 
target board. Eventually it prints :

BOOTP/DHCP failed on eth0
[eth_drv_ioctl] Warning: Driver can't set multi-cast mode
[eth_drv_ioctl] Warning: Driver can't set multi-cast mode

Which seems unusual given that DHCP worked fine for Redboot.

So here are some questions:

What should this test print when it works? Should I be able to ping my 
target after using init_all_network_interfaces() ? What is different 
between Redboot's use of my driver and ECOS's use of the driver?

--Chuck


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

Re: [ECOS] N00blet question on networking

[Gary Thomas]

Chuck McManis wrote:
> Ok, so I have redboot working. Further I can build a libtarget.a with a 
> basic set of options, I tried the simple program to get data from the 
> console and echo it back, and I can boot redboot, 'load -h <my server> 
> simple' and 'go' to run it. All's cool.
> 
> So I used ecosconfig to add in CYGPKG_NET and FREEBSD_STACK and ETH_DRIVERS
> (actually there was a point where I had no conflicts but a couple of 
> ethernet driver files were missing, its a CDL bug but I've not assured 
> myself that I know enough about what I'm doing to call it a bug outright 
> yet ;-)
> 
> Now I can build the httpd1.c test and when I run it this is what I get:
> ----------------------------------------------------
> Ethernet eth0: MAC address 00:0a:00:c0:ff:ee
> IP: 192.168.110.124/255.255.255.0, Gateway: 192.168.110.254
> Default server: 0.0.0.0
> 
> RedBoot(tm) bootstrap and debug environment [GRUB]
> Non-certified release, version UNKNOWN - built 21:38:48, Feb 19 2007
> 
> Platform: PC (I386)
> Copyright (C) 2000, 2001, 2002, 2003, 2004 Red Hat, Inc.
> Copyright (C) 2003, 2004, 2005, 2006 eCosCentric Limited
> 
> RAM: 0x00100000-0x040f0000, [0x0018da10-0x01000000] available
> RedBoot> load -h 192.168.110.30 httpd1
> Using default protocol (TFTP)
> Entry point: 0x00208000, address range: 0x00208000-0x00244720
> RedBoot> go
> [cyg_net_init] Init: mbinit(0x00000000)
> [cyg_net_init] Init: cyg_net_init_devs(0x00000000)
> Init device 'vt8235_eth0'
> [cyg_net_init] Init: loopattach(0x00000000)
> [cyg_net_init] Init: ifinit(0x00000000)
> [cyg_net_init] Init: domaininit(0x00000000)
> [cyg_net_init] Init: cyg_net_add_domain(0x00243f80)
> New domain internet at 0x00000000
> [cyg_net_init] Init: cyg_net_add_domain(0x002436c0)
> New domain route at 0x00000000
> [cyg_net_init] Init: call_route_init(0x00000000)
> [cyg_net_init] Done
> [eth_drv_ioctl] Warning: Driver can't set multi-cast mode
> [eth_drv_ioctl] Warning: Driver can't set multi-cast mode
> 
> ----------------------------------------
> 
> Ok, but when I do this, there doesn't seem to be any way to talk to my 
> target board. Eventually it prints :
> 
> BOOTP/DHCP failed on eth0
> [eth_drv_ioctl] Warning: Driver can't set multi-cast mode
> [eth_drv_ioctl] Warning: Driver can't set multi-cast mode
> 
> Which seems unusual given that DHCP worked fine for Redboot.
> 
> So here are some questions:
> 
> What should this test print when it works? Should I be able to ping my 
> target after using init_all_network_interfaces() ? What is different 
> between Redboot's use of my driver and ECOS's use of the driver?

The most likely cause is not getting interrupts from your
ethernet card.  RedBoot doesn't use them but eCos requires
that they work.

Try putting some 'diag_printf()' calls in the ethernet driver to
see what's happening.

Note: at this point, it's best to access RedBoot via the serial port
and leave network based debugging for later.

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

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

Re: [ECOS] N00blet question on networking

[Chuck McManis]

At 04:04 AM 2/22/2007, Gary Thomas wrote:
The most likely cause is not getting interrupts from your
>ethernet card.  RedBoot doesn't use them but eCos requires
>that they work.
>
>Try putting some 'diag_printf()' calls in the ethernet driver to
>see what's happening.

Aye, that I've been doing and sure enough no Rx interrupts. I see the 
transmit interrupts but not the receive ones ... hmmm.

--Chuck


>Note: at this point, it's best to access RedBoot via the serial port
>and leave network based debugging for later.


Sigh :-) I so wish this weren't the case but I understand why it is.



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

Re: [ECOS] N00blet question on networking

[Chuck McManis]

At 04:04 AM 2/22/2007, Gary Thomas wrote:
The most likely cause is not getting interrupts from your
>ethernet card.  RedBoot doesn't use them but eCos requires
>that they work.


Ok so many diag_printfs later ...

One of the things I discovered is that my CR0 register which has 
control bits for enabling transmission and reception. Rx Enable is 
bit 3 and Tx enable is bit 4, NIC "Start" is bit 1 so a "proper" CR0 
should be 0x1a, but at somepoint it gets 0x12 in it and that disables 
the receiver. Since my transmits were working fine slam 0x1a in to 
periodically and I at least got through the handshake to the DHCP 
server (it assigned my address Etc.) that tells me that interrupts 
are hooked correctly if the CR0 register is correct.

Looking at it further, I was wondering if the Half duplex / full 
duplex state of the PHY might have anything to do with it (its hooked 
to a switch and so should be full duplex). Has anyone had an 
experience where if the MAC thought the Phy was half duplex it would 
disable the receiver when it was transmitting? In such a system would 
I be adding code to my TxDone function to turn the receiver enable 
bit back on? Clearly on a Hub it would be half duplex and some other 
lame networking equipment so I want to do the right thing.

--Chuck


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