[PATCH 08/10] net/farady: Support NCSI mode

Gavin Shan gwshan at linux.vnet.ibm.com
Sun Jul 3 15:37:36 AEST 2016


On Fri, Jul 01, 2016 at 09:26:56PM +1000, Benjamin Herrenschmidt wrote:
>On Thu, 2016-06-30 at 23:35 +0930, Joel Stanley wrote:
>> We get this message whenever the device is brought up or down. This
>> includes when the host reboots, filling up the bmc's kernel logs.
>> 
>> However, when we had the timer race it was the only way to know what
>> was going on. Can we detect when it's the host bringing the device
>> down opposed to an "ifconfig up" failing to bring it up?
>
>What would be nice would be if we could mirror that into the "carrier"
>state of the interface (which normally corresponds to the PHY link
>state).
>
>The problem however is that when the carrier is off, the network stack
>will ditch packets, which doen't work for us since we still need to
>talk NSCI ...
>
>I'm thinking we should create some netdev flags so that the generic
>link monitor stops bringing it down in that case. That way we can
>set the carrier and eixsting stuff that knows how to read and interpret
>it will work.
>

Ben and Joel, thank you very much for your comments. I'm sorry this comment
isn't covered in the revision I sent to community just. I will evaluate it
later and improve it accordingly. Note this is the only comment not covered
in the revision sent to community, all other comments should be covered there
if I didn't miss any of them.

Thanks,
Gavin

>Cheers,
>Ben.
>



More information about the openbmc mailing list