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

[Jakub Kicinski <kuba@kernel.org>]

On Thu, 21 May 2026 11:16:35 -0400 Jamal Hadi Salim wrote:
> On Thu, May 21, 2026 at 10:35 AM Jakub Kicinski <kuba@kernel.org> wrote:
> > > To be precise, something like attached (untested, uncompiled)  
> >
> > Can we not pull the headers? Do you know of anyone modifying payloads
> > with pedit?
> 
> not sure - but it is not unreasonable if someone used it for such a case.

Ack, we'd still support modifying any offset in the packet.
But not optimize for modifying frags until we know user exists.

> > The concept of "shared frags" is silly IMHO, if I'm checking right only
> > rxrpc and xfrm think that it's a thing. I'm hoping to delete that and
> > reclaim the flag id in net-next...  
> 
> All these issues stem from the shared frags point, but the patchlet i
> showed is unrelated to SKBFL_SHARED_FRAG, it is more related to
> cloned+frags.

Right, right. Just pointing it out cause Rajat's patch adds another
skb_has_shared_frag(). Which we'll then have to delete in net-next :\

> If you remove it in net-next - does that necessitate removing all
> "fixes" that tried to address that issue? It seems like that would be
> sensible, but it makes me wonder why the shared frags concept exists
> in the first place.

IDK. We have in the tree plenty of "good ideas that went nowhere"..
As I said elsewhere IMO having the maintain this flag correctly
100% of the time will only create more CVEs, and there's no practical
reason to optimize for frags not being shared.