[ewg] Re: Continue of "defer skb_orphan() until irqs enabled"

[Moni Shoua]

akepner at sgi.com wrote:
> On Wed, Sep 24, 2008 at 10:22:11AM -0700, Roland Dreier wrote:
>>  > -	spin_unlock_irqrestore(&priv->tx_lock, flags);
>>  > +	local_irq_restore(flags);
>>  > +	if (orphan)
>>  > +		skb_orphan(skb);
>>  > +
>>  > +	spin_unlock(&priv->tx_lock);
>>
>> What did we conclude about the possible skb_orphan()-after-skb_free()
>> scenario here?
>>
> 
> Eli?
> 
> I thought the conclusion was that the "ipoib: defer skb_orphan()..." 
> patch had indeed created a race condition between the skb_orphan() 
> call and the tx completion handler freeing the skb.
> 
This is also what I remember but I think that at least for _next_OFED_ (and next OFED only) it is better
to use the patch rather than staying with the code that has the bug described by Arthur.

Roland,
Don't you think so?