On Sat, 2007-12-05 at 19:02 +0200, Patrick McHardy wrote:
> jamal wrote:
> > diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> > #define NETDEV_TX_LOCKED -1/* driver tx lock was already taken */
> > +#define NETDEV_TX_DROP -2 /* request caller to drop packet */
>
> Th
jamal wrote:
> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> index f671cd2..718d6fd 100644
> --- a/include/linux/netdevice.h
> +++ b/include/linux/netdevice.h
> @@ -83,6 +83,9 @@ struct wireless_dev;
> #define NETDEV_TX_OK 0 /* driver took care of packet */
>
Ok, heres the next iteration based on feedback from Thomas and Peter.
It seems this is good form in matching what the original code does.
I have a few doubts in labelled as XXX in there. Thomas, if you have
time, do you mind looking at those?
The usual testing happened i.e "if you can see this em
On Sat, 2007-12-05 at 11:46 +0200, Thomas Graf wrote:
> > + struct sk_buff *skb = NULL;
> > +
> > + if (skb = dev->gso_skb)
>
> This must have caused a warning
Possibly (dont recall) - i will move the assignment to the first line.
> and you probably want to inline this function.
will do.
* jamal <[EMAIL PROTECTED]> 2007-05-11 15:29
> diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
> + netif_schedule(dev);
> + return 0;
> +}
> +
> +static struct sk_buff *
> +try_get_tx_pkt(struct net_device *dev, struct Qdisc *q)
> +{
> + struct sk_buff *skb = NULL;
> +
>
On Fri, 2007-11-05 at 15:01 -0700, Waskiewicz Jr, Peter P wrote:
> The current requeue calls in qdisc_restart() returns a value of 1 to the
> upper layer.
Which was not meaningful ;->
i.e qdisc run would break as a result.
Note, this is still in conformance with net-2.6 based on a change in the
> Ok, booting fine to me;
> if you receive this email, it must have worked. OTOH, if you
> didnt receive this it probably failed ;-> (as the presenter
> said "if you cant hear me in the far corner please raise your hand").
>
> clip away ...
+static inline int
+handle_dev_requeue(struct sk_buff *sk
On Fri, 2007-11-05 at 11:46 -0700, Waskiewicz Jr, Peter P wrote:
> Clippers are standing by. :)
Nothing as nice as time off to code away;->
Ok, booting fine to me;
if you receive this email, it must have worked. OTOH, if you didnt
receive this it probably failed ;->
(as the presenter said "if
> For lockless drivers, I think it could be moved up; only
> thing is nothing is stopping it from changing again at transmit time.
> For drivers that are not lockless, i am wondering if its even
> useful to have that check given we have already grabbed the tx lock.
Good point. I live in an LLTX
On Fri, 2007-11-05 at 11:13 -0700, Waskiewicz Jr, Peter P wrote:
> After thinking about this a bit more: even if the queue is stopped,
> you'd end up requeueing anyways. Plus, you'd need to re-acquire
> dev->queue_lock (which is what happens today). I think the best way
> overall would be to che
On Fri, 2007-11-05 at 19:01 +0200, Thomas Graf wrote:
> * jamal <[EMAIL PROTECTED]> 2007-05-10 20:13
> > * Compute the worst case header length according to the protocols
> > diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
> > index f28bb2d..b821040 100644
> > --- a/net/sched/sch_
> On Fri, 2007-11-05 at 08:56 -0700, Waskiewicz Jr, Peter P wrote:
>
> > In qdisc_restart(), you removed any check for if
> > (!netif_queue_stopped(dev)) before calling
> dev_hard_start_xmit(). If
> > the underlying queue is stopped and you send the skb,
> you'll generate
> > a requeue. Is t
On Fri, 2007-11-05 at 08:56 -0700, Waskiewicz Jr, Peter P wrote:
> In qdisc_restart(), you removed any check for if
> (!netif_queue_stopped(dev)) before calling dev_hard_start_xmit(). If
> the underlying queue is stopped and you send the skb, you'll generate a
> requeue. Is there a reason it was
* jamal <[EMAIL PROTECTED]> 2007-05-10 20:13
> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> index f671cd2..718d6fd 100644
> --- a/include/linux/netdevice.h
> +++ b/include/linux/netdevice.h
> @@ -83,6 +83,9 @@ struct wireless_dev;
> #define NETDEV_TX_OK 0 /* d
> This compiles and passes some basic tests - no serious testing.
> Against net-2.6.
>
> The patch is ugly looking, so i have at the end the
> re-written qdisc; you can easily tell the rest from the patch.
>
> Please flush out any fluff - I would like to submit this
> (almost lost it, thanks to
This compiles and passes some basic tests - no serious testing.
Against net-2.6.
The patch is ugly looking, so i have at the end the re-written qdisc;
you can easily tell the rest from the patch.
Please flush out any fluff - I would like to submit this (almost lost
it, thanks to an early posting
16 matches
Mail list logo
