How to attach higher ethernet stacks to the MAC driver

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

How to attach higher ethernet stacks to the MAC driver

Dennis S
Hi all,
 
I have a question regarding the connection between a mac-/phy-driver and a third
party tcp / ip stack. In the mac layer I have functions like
y_eth_isr, y_eth_init, y_eth_send, y_eth_recv, ... and many more.
 
In y_eth_init I attached an ISR (y_eth_isr) and the default DSR (eth_drv_dsr).
 
As far as I understood, the data flow is as follows:
1) The MAC (Hardware) receives an ethernet frame and creates an interrupt
2) y_eth_isr is called ( ret = (CYG_ISR_HANDLED | CYG_ISR_CALL_DSR); )
3) The default eth_drv_dsr is called
4) y_eth_deliver is called
5) y_eth_poll is called
6) The default eth_drv_recv is called
7) y_eth_recv in the mac driver is called
 
This is not what I really could proove but what I investigated by reading the
documentation and source code. Now.. my consideration was that I simply have to
add the call to the receive function of the tcp/ip stack in y_eth_recv. Is this
the right approach?
 
Another point: how can I reliable test the function of my hardware? I plugged in
my device into an ethernet switch which says that there is a link. Using ipconfig
under Windows just says -not connected-. Using ping is not a solution of course,
because I have no ip address...
 
And a third question: in the send function of the tcp/ip stack I have to forward
the send buffer to y_eth_send in the mac layer. I simply do this by calling
y_eth0_sc.funs->send(...). Is this the suitable way?
 
Kind Regards,
Dennis
 

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

Reply | Threaded
Open this post in threaded view
|

Re: How to attach higher ethernet stacks to the MAC driver

Leschke Serafin (lesc)

Hi

On 21.05.2015 18:39, [hidden email] wrote:

> Hi all,
>  
> I have a question regarding the connection between a mac-/phy-driver and a third
> party tcp / ip stack. In the mac layer I have functions like
> y_eth_isr, y_eth_init, y_eth_send, y_eth_recv, ... and many more.
>  
> In y_eth_init I attached an ISR (y_eth_isr) and the default DSR (eth_drv_dsr).
>  
> As far as I understood, the data flow is as follows:
> 1) The MAC (Hardware) receives an ethernet frame and creates an interrupt
> 2) y_eth_isr is called ( ret = (CYG_ISR_HANDLED | CYG_ISR_CALL_DSR); )
> 3) The default eth_drv_dsr is called
> 4) y_eth_deliver is called
> 5) y_eth_poll is called
> 6) The default eth_drv_recv is called
> 7) y_eth_recv in the mac driver is called
This seems correct so far.
>  
> This is not what I really could proove but what I investigated by reading the
> documentation and source code. Now.. my consideration was that I simply have to
> add the call to the receive function of the tcp/ip stack in y_eth_recv. Is this
> the right approach?
Well I thik it's depends on the Stack. The FreeBSD-Stack does some more
things on the eth_drv Level.
ether_input Net/bsd_tcpip/current/sry/sys/net/if_ethersubr.c
ether_demux
        Net/bsd_tcpip/current/sry/sys/net/if_ethersubr.c


after this two routines the packet is handed over to ip_input wich runs
on a lower priority. I would presume you would also need to do some
things before calling the tcp stack to be 802.1 compliant. Maybe these
files give you an input.
>  
> Another point: how can I reliable test the function of my hardware? I plugged in
> my device into an ethernet switch which says that there is a link. Using ipconfig
> under Windows just says -not connected-. Using ping is not a solution of course,
> because I have no ip address...
This is a hard one. Testing on this level is not easy. If you just want
to see if the stack is running. You could also use arping and see with
wireshark if you detect an answer. Testing of the whole system is much
more difficult, you would need some sort of Mock /virtalisation of the
different layers, which is really hard because of the timings etc.
>  
> And a third question: in the send function of the tcp/ip stack I have to forward
> the send buffer to y_eth_send in the mac layer. I simply do this by calling
> y_eth0_sc.funs->send(...). Is this the suitable way?
>  
This sounds also fine. Of course someone must handle the memory
management of the sendbuffer.
> Kind Regards,
> Dennis
>  

I hope this helps
Serafin

--
Serafin Leschke, BSc in Computer Science UAS Zurich
Research Assistant




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