Re: [RFT][RFC][PATCH] p54: utilize cryptographic accelerator

[Christian Lamparter <chunkeey@web.de>]

On Thursday 27 November 2008 01:15:57 Larry Finger wrote:
> > diff -Nurp a/drivers/net/wireless/p54/p54common.c b/drivers/net/wireless/p54/p54common.c
> > --- a/drivers/net/wireless/p54/p54common.c	2008-11-24 21:05:42.590287518 +0100
> > +++ b/drivers/net/wireless/p54/p54common.c	2008-11-24 22:06:13.951266225 +0100
> > @@ -843,6 +863,8 @@ static int p54_assign_address(struct iee
> >  
> >  	spin_lock_irqsave(&priv->tx_queue.lock, flags);
> >  	left = skb_queue_len(&priv->tx_queue);
> > +	if (left > 30)
> > +		return -ENOMEM;
> 
> The above returns with tx_queue.lock still set.
Oops, this piece should be in another patch.
So, ok it's gone now.

> --snip--
> > +		}
> > +	}
> > +
> > +	mutex_lock(&priv->conf_mutex);
> > +	skb = p54_alloc_skb(dev, P54_HDR_FLAG_CONTROL_OPSET, sizeof(*rxkey) +
> > +			sizeof(struct p54_hdr),	P54_CONTROL_TYPE_RX_KEYCACHE,
> > +			GFP_ATOMIC);
> > +	if (!skb) {
> > +		mutex_unlock(&priv->conf_mutex);
> > +		return -ENOMEM;
> > +	}
> > +
> > +	rxkey = (struct p54_keycache *)skb_put(skb, sizeof(*rxkey));
> > +	rxkey->entry = cpu_to_le16(key->keyidx);
> > +	rxkey->key_id = cpu_to_le16(key->keyidx);
> 
> Sparse objects to the above two statments because key->keyidx is of type s8, and
> both rxkey->entry and rxkey->key_id are of type u8. A conversion to le16 is not
> needed.
> 
> A patch for both problems is below:
Thanks... I guess the final RFT/RFC will be ready on friday/saturday.
there are a two outstanding issues:
	- do we need to swap CCMP's IV as well?
	- does MIC failure & replay detection work?
		(or did you test that, when you were working with TKIP?)

Regards,
	Chr