Am 09.07.2014 01:43, schrieb Francois Romieu:
> Richard Weinberger <rich...@nod.at> :
>> Am 09.07.2014 00:47, schrieb Francois Romieu:
> [...]
>>> What are you taking about ? netconsole does not need to receive.
>>
>> Isn't netconsole is only one user of netpoll ?
> 
> Out of tree users are irrelevant. See netpoll related comments in
> cd6362befe4cc7bf589a5236d2a780af2d47bcc9

Thanks lot for pointing this out!

>> Of course netconsole needs only to transmit SKBs.
>> But if you look at other ->ndo_poll_controller implementations
>> you'll notice that they care also about receiving.
> 
> It's just the long, illuminating history of netpoll :o)
> 
> Some limited Rx netpoll support may be done but it needs more work
> than was originally advertised.
> 
>>> hyperv start_xmit handler almost does its own Tx completion as you have
>>> noticed. The situation is imho close to a virtual device one as was veth
>>> in bb446c19fefd7b4435adb12a9dd7666adc5b553a.
>>
>> Bad commit reference: bb446c19fefd7b4435adb12a9dd7666adc5b553a
> 
> Sorry, it currently belongs to davem's net-next.

Found it already. :)

Thanks,
//richard
_______________________________________________
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

Reply via email to