Hi,
By the way: people, I totally forgot to mention that 0.0.1 should fix
the fragment problems we have seen courtesy of a patch of lepton. I was
working on another ('proper') fix, but that turned out to be too complicated
(requiring things such as two memory allocations per packet). I still
have it laying around (unfinished), so if someone is interested, give a
shout.
lepton, I'm not totally happy with this patch, as it adds bridge-nf-
specific knowledge (skb->physindev). Would changing it to read something
like
memcpy(skb2->data - 16, skb->data - 16, 16);
help? I assume the oops comes from dereferencing skb->mac.raw?
(16 should be safe because ip_output.c:ip_finish_output2 contains:
read_lock_bh(&hh->hh_lock);
memcpy(skb->data - 16, hh->hh_data, 16);
read_unlock_bh(&hh->hh_lock);
skb_push(skb, hh->hh_len);
cheers,
Lennert
On Mon, Oct 29, 2001 at 03:34:45PM +0800, lepton wrote:
> A little thing needed to fixed:
>
> in ip_output.c
> memcpy(skb2->data-ETH_HLEN,skb->mac.raw,ETH_HLEN);
>
> perhaps change it to
>
> if(skb->physindev)
> memcpy(skb2->data - ETH_HLEN,
>skb->mac.raw, ETH_HLEN);
> would be better.
>
> otherwise you can try oop the kernel with these steps:
>
> #ifconfig lo 127.0.0.1
> #ifconfig lo mtu 1000
> #ping -s 1100 127.0.0.1
>
_______________________________________________
Bridge mailing list
[EMAIL PROTECTED]
http://www.math.leidenuniv.nl/mailman/listinfo/bridge
