Re: [PATCH net 0/7] xsk: fix AF_XDP multi-buffer Tx descriptor reclaim

[Maciej Fijalkowski <maciej.fijalkowski@intel.com>]

On Thu, Jun 25, 2026 at 09:05:28AM -0700, Stanislav Fomichev wrote:
> On 06/25, Jason Xing wrote:
> > On Thu, Jun 25, 2026 at 12:37 AM Maciej Fijalkowski
> > <maciej.fijalkowski@intel.com> wrote:
> > >
> > > On Wed, Jun 24, 2026 at 08:38:20AM -0700, Stanislav Fomichev wrote:
> > > > On 06/23, Maciej Fijalkowski wrote:
> > > > > Hi,
> > > > >
> > > > > This series fixes several AF_XDP multi-buffer Tx paths where descriptors
> > > > > consumed from the Tx ring are not consistently returned to userspace
> > > > > through the completion ring when the packet is later dropped as invalid.
> > > > >
> > > > > The affected cases are invalid or oversized multi-buffer Tx packets in
> > > > > both the generic and zero-copy paths. In these cases, the kernel can
> > > > > consume one or more Tx descriptors while building or validating a
> > > > > multi-buffer packet, then drop the packet before it reaches the device.
> > > > > Userspace still owns the UMEM buffers only after the corresponding
> > > > > addresses are returned through the CQ. Missing completions therefore
> > > > > make userspace lose track of those buffers.
> > > > >
> > > > > The generic path fixes cover three related cases:
> > > > > * partially built multi-buffer skbs dropped by xsk_drop_skb();
> > > > >   continuation descriptors left in the Tx ring after xsk_build_skb()
> > > > >   reports overflow;
> > > > > * invalid descriptors encountered in the middle of a multi-buffer
> > > > >   packet, including the offending invalid descriptor itself.
> > > > >
> > > > > The zero-copy path is handled separately. The batched Tx parser now
> > > > > distinguishes descriptors that can be passed to the driver from
> > > > > descriptors that are consumed only because they belong to an invalid
> > > > > multi-buffer packet. Reclaim-only descriptors are written to the CQ
> > > > > address area and published in completion order, after any earlier
> > > > > driver-visible Tx descriptors.
> > > > >
> > > > > The ZC batching path can also retain drain state when userspace has not
> > > > > yet provided the end of an invalid multi-buffer packet. To keep this
> > > > > state local to the singular batched path, the series prevents a second
> > > > > Tx socket from joining the same pool while such drain state exists.
> > > > > During the singular-to-shared transition, Tx batching is gated,
> > > > > pre-existing readers are waited out, and bind fails with -EAGAIN if the
> > > > > existing socket still has pending drain state. This avoids adding
> > > > > multi-buffer drain handling to the shared-UMEM fallback path.
> > > > >
> > > > > The last two patches update xskxceiver so the tests account invalid
> > > > > multi-buffer Tx packets as descriptors that must be reclaimed, while
> > > > > still not expecting those invalid packets on the Rx side.
> > > > >
> > > > > This is a follow-up to Jason's changes [0] which were addressing generic
> > > > > xmit only and this set allows me to pass full xskxceiver test suite run
> > > > > against ice driver.
> > > >
> > > > There is a fair amount of feedback from sashiko already :-( So the meta
> > > > question from me is: is it time to scrap our current approach where
> > > > we parse descriptor by descriptor? (and maintain half-baked skb and
> > > > half-consumed descriptor queues)
> > > >
> > > > Should we:
> > > >
> > > > 1. do desc[MAX_SKB_FRAGS] and xskq_cons_peek_desc until we exhaust
> > > > PKT_CONT (if the last packet has PKT_CONT, return EOVERFLOW to userspace
> > > > and do a full stop here)
> > > > 2. now that we really know the number of valid descriptors -> reserve
> > > > the cq space (if not -> EAGAIN)
> > > > 3. pre-allocate everything here (if at any point we have ENOMEM -> cleanup
> > > > locally, don't ever create semi-initialized skb)
> > > > 4. construct the skb
> > > > 5. xmit
> > >
> > > Yeah generic xmit became utterly horrible, haven't gone through sashiko
> > > reviews yet, but bare in mind this set also aligns zc side to what was
> > > previously being addressed by Jason.
> > >
> > > I believe planned logistics were to get these fixes onto net and then
> > > Jason had an implementation of batching on generic xmit, directed towards
> > > -next and that's where we could address current flow.
> > 
> > Agreed. That's what I'm hoping for. There would be much more
> > discussion on how to do batch xmit in an elegant way, I believe.
> 
> This doesn't have to depend on the batch rewrite, we should be able to rewrite
> this non-zc in net, this is still technically fixes, not feature work..
> 
> There was already a couple of revisions with this drain_cont approach
> and every time I look at it feels like the cure is worse than the
> decease :-( Obviously not gonna stop you from going with the current approach,
> but these fixes feel a bit of a wasted effort to me (since the bugs keep
> coming and we are piling more complexity).

Well this is my fault as I took Jason's patches as-is and did not realize
Sashiko had issues with it. I *think* I got ZC side almost right so I'd
like to have at least one last round with trying to make the generic side
right...