[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1447029684.8727.38.camel@kernel.crashing.org>
Date: Mon, 09 Nov 2015 11:41:24 +1100
From: Benjamin Herrenschmidt <benh@...nel.crashing.org>
To: Gavin Shan <gwshan@...ux.vnet.ibm.com>, netdev@...r.kernel.org
Cc: davem@...emloft.net, sergei.shtylyov@...entembedded.com
Subject: Re: [PATCH RFC 2/6] net/ncsi: Packet handler
On Mon, 2015-11-09 at 11:10 +1100, Gavin Shan wrote:
> +
> + /* Send NCSI packet */
> + skb_get(nr->nr_cmd);
> + ret = dev_queue_xmit_sk(NULL, nr->nr_cmd);
> + if (ret)
> + goto out;
> +
Probably just dev_queue_xmit() in recent kernels
This means the queue needs to be running, which means the carrier needs
to be "on".
Dave, this is a problem actually. The NC-SI connection can be to
several NICs that may or may not have a link up. The "stack" will chose
one based on policies that we can discuss separately but basically, I
think we should reflect whether we found a NIC and it has a link up in
the "carrier" setting of our driver.
The problem with doing that is that if the carrier is "off", then
there's some code nowadays that will put a noop qdisc in and we won't
be able to send the low level NC-SI packets to communicate with the NIC
and thus find a new one or get a new carrier indication.
So at the moment, we must have the carrier of the driver appear as
"always on" when using NC-SI.
Cheers,
Ben.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists