Re: [patch] syncppp.c and skb->priority

Gergely Madarasz (gorgo@sztaki.hu)
Mon, 12 Jul 1999 17:54:35 +0200 (MET DST)


On Mon, 12 Jul 1999, Jan Kasprzak wrote:

> Gergely Madarasz wrote:
> : > : shouldn't this read skb->priority=TC_PRIO_CONTROL ?
> : > :
> : > I don't know - I can see only one occurence of TC_PRIO_CONTROL
> : > in the 2.2.10 and 2.3.9 trees. What is TC_PRIO_CONTROL expected to do?
> :
> : I guess in this case it would mean link control protocol... higher
> : priority than normal packets. Anyway Alan told me to use it, so I guess it
> : does the right thing (and I tested it, it really does the right thing :))
> :
> I have tested it, it works (but didn't find why - or why setting
> it to 0 or 1 does not work). This must be hard-coded somewhere,
> the TC_PRIO_CONTROL is not used anywhere.

I don't think the packet scheduler would check for exact values when
scheduling packets, just compare them, so there is no need to use this
constant in the packet scheduler code.

>
> Gergely, can you send Alan the patches to syncppp.c and dev.c?
> I think it should be incorporated in the kernel sources.

Alexey just pointed out that the IFF_RUNNING setting is dangerous in
syncppp.c, so I'll have to find another way of reporting the status of the
link protocol. Btw, Alan, what you said about IFF_RUNNING earlier, does
this mean that it can't be done ?

> BTW, I have found another problem with syncppp.c. In Sync PPP mode
> it does not work when I manually ifconfig one of the ends down and then up.
> It seems the LCP does not end. It helps to unplug the line physically
> for MAXALIVECNT keepalive packets - both ends then will agree the line
> is down and restart the LCP from the beginning.

Hmm... sorry, can't help you with this one. It should be related with your
board not reporting modem line status though. The driver should be
notified of modem line status changes and call at least sppp_reopen() when
the line changes to up. In my drivers I even call sppp_close() when the
line goes down and sppp_open() when it comes back.

Greg

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/