Re: [PATCH net] net/sched: fix pedit partial COW leading to page cache corruption

[Jakub Kicinski <kuba@kernel.org>]

On Sat, 23 May 2026 08:13:21 -0400 Jamal Hadi Salim wrote:
> > > @@ -474,8 +473,6 @@ TC_INDIRECT_SCOPE int tcf_pedit_act(struct sk_buff *skb,
> > >                 }
> > >
> > >                 *ptr = ((*ptr & tkey->mask) ^ val);
> > > -               if (ptr == &hdata)
> > > -                       skb_store_bits(skb, hoffset + offset, ptr, 4);
> > >         }
> > >
> > >         goto done;  
> >
> > I see you are trying to get rid of the skb_header_pointer() /
> > skb_store_bits() piece. Sure looks cleaner if we must linearize.  
> 
> The other thing (i may be over thinking) with pskb_may_pull is: if the
> data is already linear (in a clone), wouldn't we corrupt the shared
> linear data of the clone?

I said

  for the portion of the problem that's "we are writing to frags"

IOW not replacing the rest of the patch (assuming we care).