[etherlab-users] RTL8168

[Aaron Edsinger]

We're also very interested in getting a RTL8168  driver working but 
don't have the skill set in-house. We've looked around to contract the 
work out but with no luck yet (we have only modest resources). I'm 
wondering if anyone on the list:

a) is willing to pool resources to fund the work
b) would be willing to do the work given a fee

Of course we would contribute the results to the Etherlab package. 
Florian, do you have an estimate for how long it would take a skilled 
device coder to do this work?

Best,
Aaron

Erwin Burgstaller wrote:
> * Thu, Dec 11, 2008 - 13:41:33 +0800, li.peng - li.peng at advantech.com.cn:
>   
>> Sorry.I test it with slave.It can send data, but can not receive data.
>> Could anyone help me to fix it, Thanks in advance.
>>     
>
> Sorry, I can't help. I didn't got your driver to run, so I decided to
> patch one by myself, but didn't get very far.
>
> I took the source code for gigabit RealTek cards from kernel 2.6.25: 
>
>   
>> r8169.c: RealTek 8169/8168/8101 ethernet driver.
>>     
>
> As this good looking driver would not only support the 8168 and is made
> by people from RealTek, this source is in my opinion worth much more for
> patching than the one you've used.
>
> The only thing I had to change to get it running in 2.6.22 as regular
> Ethernet NIC had been this:
>
>  static void mdio_write(void __iomem *ioaddr, int reg_addr, int value)
>  {
>         int i;
> @@ -1089,7 +1105,11 @@
>         .get_wol                = rtl8169_get_wol,
>         .set_wol                = rtl8169_set_wol,
>         .get_strings            = rtl8169_get_strings,
> -       .get_sset_count         = rtl8169_get_sset_count,
> +#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,22)
> +       .get_sset_count        = rtl8169_get_sset_count,
> +#else
> +       .get_stats_count        = rtl8169_get_sset_count,
> +#endif
>         .get_ethtool_stats      = rtl8169_get_ethtool_stats,
>  };
>  
> So it would cover basically all kernel versions from 2.6.22 to 2.6.25 so
> far.
>
> I would have much more to get involved in the basics of that source and
> in EtherCAT, to patch it right. Just referring to the patches in the
> 8139too lead me to dead end street, and at the moment I've a lot more
> important things to do, so we switched simply the hardware for the
> moment (Advantech uno-2172 -> uno-3282).
>
> But at the long run, I think, it would be important for the etherlab
> Master to support more than just one currently available NIC, the e1000
> based ones.
>  
> Greets,
> 	Erwin
>
>   


-- 
Aaron Edsinger, Ph.D.
San Francisco, CA
415.994.0727
www.mekabot.com
www.hee3.com
www.csail.mit.edu/~edsinger