| To: | Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [NETLINK] Do not netlink_unicast shared packet in kernel/audit.c |
| From: | "David S. Miller" <davem@xxxxxxxxxxxxx> |
| Date: | Mon, 17 Jan 2005 13:36:40 -0800 |
| Cc: | tommy.christensen@xxxxxxxxx, simon.roscic@xxxxxxxxx, netdev@xxxxxxxxxxx |
| In-reply-to: | <20050116080254.GA2997@xxxxxxxxxxxxxxxxxxx> |
| References: | <E1Cpo1q-0007lI-00@xxxxxxxxxxxxxxxxxxxxxxxx> <41E942AF.3030202@xxxxxxxxx> <20050115183023.GA31211@xxxxxxxxxxxxxxxxxxx> <41E97B1A.30205@xxxxxxxxx> <20050116080254.GA2997@xxxxxxxxxxxxxxxxxxx> |
| Sender: | netdev-bounce@xxxxxxxxxxx |
On Sun, 16 Jan 2005 19:02:54 +1100
Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> wrote:
> So here is a patch to handle shared skb's by cloning them if we're
> going to do a pskb_expand_head. This patch depends on the previous
> one that moves skb_orphan into netlink_trim.
I think this patch is buggy. At least in the final hunk, we
do this:
info.skb = skb;
...
skb = netlink_trim( ... );
Now info.skb can be pointing to a freed up skb.
I'll apply the first patch though.
|
| Previous by Date: | Re: [patch 4/10] s390: network driver., Tommy Christensen |
|---|---|
| Next by Date: | Re: Fwd: [2.6] ethertap and af_inet.c assertion failures, David S. Miller |
| Previous by Thread: | Re: [NETLINK] Do not netlink_unicast shared packet in kernel/audit.c, Tommy Christensen |
| Next by Thread: | Re: [NETLINK] Do not netlink_unicast shared packet in kernel/audit.c, Herbert Xu |
| Indexes: | [Date] [Thread] [Top] [All Lists] |