[Click] 2.6 update - Bug + Fix
Eddie Kohler
kohler at cs.ucla.edu
Wed Jun 28 13:39:46 EDT 2006
Great! And no problem, I extremely appreciate the patch!!!
E
Beyers Cronje wrote:
> Hi Eddie,
>
> Because of C precedence rules this gets parsed as
>
> !( (atomic_read(&skb_shinfo(skb)->dataref) - skb->nohdr)
> ? (1 << SKB_DATAREF_SHIFT) + 1
> : 1 )
>
> which is always true. So skbs never get recycled in your code.
>
>
> Bad bad code from me.
>
> There is no need for a following atomic_sub, since dataref is set to
> 1 below
> anyway.
>
>
> 2 Bonehead mistakes in a two liner patch, more bad. In my defense the
> bug and solution found me at 05:00am right before I went to bed after an
> all-nighter and I just had to quickly check if it worked :)
>
> Does this work for you?
>
>
> Works 100%, thanks!
>
> Beyers
>
More information about the click
mailing list