All of lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Magnus Karlsson" <magnus.karlsson@intel.com>
Cc: intel-wired-lan@lists.osuosl.org
Subject: Re: [PATCH net-next 3/5] xsk: ice: add USO zero-copy support
Date: Fri, 20 Nov 2026 16:54:23 +0000	[thread overview]
Message-ID: <sashiko-outbox-225609@kernel.org> (raw)
In-Reply-To: <20261119165340.1071525-4-magnus.karlsson@intel.com>

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [High] ice: xsk: missing descriptor count check in ice_xsk_bufs_fit()
  and USO parse
- [High] ice: xsk: XDP ring mismatch in ice_tx_xsk_pool() frees active
  USO DMA slots
--

commit 68f04de5ef6583147ab705c136de44a0ff2a4e13
Author: Magnus Karlsson <magnus.karlsson@intel.com>

xsk: ice: add USO zero-copy support

This commit adds AF_XDP zero-copy UDP Segmentation Offload (USO) helpers to
the XSK core and implements zero-copy USO in the ice driver using per-ring
DMA-mapped header slots.

> diff --git a/drivers/net/ethernet/intel/ice/ice.h b/drivers/net/ethernet/intel/ice/ice.h
> index 432da05c48c08..e19929b37e110 100644
> --- a/drivers/net/ethernet/intel/ice/ice.h
> +++ b/drivers/net/ethernet/intel/ice/ice.h
> @@ -833,16 +833,24 @@ static inline void ice_rx_xsk_pool(struct ice_rx_ring *ring)
>   * carries a pointer to one of these XDP rings for its own purposes, such as
>   * handling XDP_TX action, therefore we can piggyback here on the
>   * rx_ring->xdp_ring assignment that was done during XDP rings initialization.
> + * The ring must be stopped, as its USO header slots are matched to the pool.
> + * If they cannot be allocated, the ring keeps the pool without USO.
>   */
>  static inline void ice_tx_xsk_pool(struct ice_vsi *vsi, u16 qid)
>  {
> +	struct xsk_buff_pool *pool;
>  	struct ice_tx_ring *ring;
>  
>  	ring = vsi->rx_rings[qid]->xdp_ring;
>  	if (!ring)
>  		return;
>  
> -	WRITE_ONCE(ring->xsk_pool, ice_get_xp_from_qid(vsi, qid));
> +	pool = ice_get_xp_from_qid(vsi, qid);
> +	if (ice_xsk_cfg_uso_slots(ring, pool))
> +		netdev_warn(vsi->netdev,
> +			    "Failed to allocate USO header buffers, USO requests on queue %u are ignored\n",
> +			    qid);
> +	WRITE_ONCE(ring->xsk_pool, pool);
>  }

[Severity: High]
Can ice_tx_xsk_pool() free ring->xsk_uso_pages while
vsi->rx_rings[qid]->xdp_ring is still running with in-flight USO
descriptors?

When vsi->num_q_vectors < vsi->num_xdp_txq (for example after reducing
combined queues with ethtool -L, or when fewer CPUs are online than
possible), ice_map_xdp_rings() maps multiple XDP rings per vector and
vsi->rx_rings[qid]->xdp_ring is not vsi->xdp_rings[qid].  On pool teardown,
ice_xsk_pool_setup() calls ice_qp_dis() and ice_qp_ena(), but ice_qp_dis()
and ice_qp_clean_rings() stop and clean vsi->xdp_rings[qid]:

drivers/net/ethernet/intel/ice/ice_base.c:ice_qp_clean_rings() {
	ice_clean_tx_ring(vsi->tx_rings[q_idx]);
	if (vsi->xdp_rings)
		ice_clean_tx_ring(vsi->xdp_rings[q_idx]);
	ice_clean_rx_ring(vsi->rx_rings[q_idx]);
}

ice_qp_ena() then calls ice_tx_xsk_pool() -> ice_xsk_cfg_uso_slots(ring,
NULL) -> ice_xsk_free_uso_slots() on vsi->rx_rings[qid]->xdp_ring while
that ring still has uncleaned ICE_TX_BUF_XSK_USO descriptors in flight.
Aside from unmapping the DMA pages while the NIC may still read them,
ice_xsk_free_uso_slots() zeroes ring->xsk_uso_slots_per_page and clears
ring->xsk_uso_pages, so the next completion in ice_clean_xsk_buf() ->
ice_xsk_uso_slot() divides by zero and dereferences NULL.

Also, if a pool is disabled while ICE_VSI_DOWN is set,
ice_xsk_pool_disable() returns without freeing the USO header slots that
ice_xsk_pool_enable() allocated on vsi->rx_rings[qid]->xdp_ring.

> diff --git a/drivers/net/ethernet/intel/ice/ice_xsk.c b/drivers/net/ethernet/intel/ice/ice_xsk.c
> index 0c1ce3e86a45b..7fa1127480b66 100644
> --- a/drivers/net/ethernet/intel/ice/ice_xsk.c
> +++ b/drivers/net/ethernet/intel/ice/ice_xsk.c

[ ... ]

> @@ -791,6 +946,272 @@ static void ice_fill_tx_hw_ring(struct ice_tx_ring *xdp_ring,
>  		ice_xmit_pkt(xdp_ring, xsk_pool, &descs[i], total_bytes);
>  }
>  

[ ... ]

> +/* A buffer longer than one data descriptor is split, so a packet without USO
> + * can then need more data descriptors than the hardware allows.
> + */
> +static bool ice_xsk_bufs_fit(const struct xsk_buff_pool *pool,
> +			     const struct xdp_desc *descs, u32 nr)
> +{
> +	u32 buffers = 0, i;
> +
> +	if (likely(pool->chunk_size <= ICE_MAX_DATA_PER_TXD_ALIGNED))
> +		return true;
> +	for (i = 0; i < nr; i++)
> +		buffers += DIV_ROUND_UP(descs[i].len,
> +					ICE_MAX_DATA_PER_TXD_ALIGNED);
> +	return buffers <= ICE_MAX_BUF_TXD;
> +}

[Severity: High]
Does ice_xsk_bufs_fit() allow a non-USO packet with more than
ICE_MAX_BUF_TXD descriptors through when pool->chunk_size <=
ICE_MAX_DATA_PER_TXD_ALIGNED?

When multiple sockets share a UMEM without XDP_UMEM_SG_FLAG,
xsk_tx_peek_release_desc_batch() uses xsk_tx_peek_release_fallback(), which
peeks descriptors via xsk_tx_peek_desc() without enforcing
pool->xdp_zc_max_segs.  In ice_xmit_zc_metadata(), a multi-descriptor chain
with XDP_PKT_CONTD set can therefore accumulate nr > ICE_MAX_BUF_TXD
descriptors into one packet.  With standard 2K or 4K chunks,
ice_xsk_bufs_fit() returns true without checking nr <= ICE_MAX_BUF_TXD, and
ice_xsk_xmit_descs() emits more than 8 hardware descriptors for a single
unsegmented packet, triggering an MDD event on the Tx queue.

For USO packets on that same fallback path, neither xsk_tx_uso_parse() nor
ice_xsk_uso_fits() checks nr <= pool->xdp_zc_max_segs either (since
ice_xsk_uso_fits() resets buffers to 1 at each segment boundary).

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261119165340.1071525-1-magnus.karlsson@intel.com?part=3

  parent reply	other threads:[~2026-11-20 16:54 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-19 16:53 [PATCH net-next 0/5] xsk: ice: USO support for AF_XDP Magnus Karlsson
2026-11-19 16:53 ` [PATCH net-next 1/5] xsk: add USO support for skb path Magnus Karlsson
2026-11-19 18:47   ` Maciej Fijalkowski
2026-11-20  8:55     ` Magnus Karlsson
2026-11-20 11:43       ` Maciej Fijalkowski
2026-11-20 17:22   ` Stanislav Fomichev
2026-11-19 16:53 ` [PATCH net-next 2/5] ice: add Tx descriptor builders and header length check Magnus Karlsson
2026-11-20 11:59   ` Maciej Fijalkowski
2026-11-19 16:53 ` [PATCH net-next 3/5] xsk: ice: add USO zero-copy support Magnus Karlsson
2026-11-20 14:22   ` Maciej Fijalkowski
2026-11-20 16:54   ` sashiko-bot [this message]
2026-11-19 16:53 ` [PATCH net-next 4/5] xsk: add KUnit tests for AF_XDP USO header helpers Magnus Karlsson
2026-11-19 16:53 ` [PATCH net-next 5/5] selftests: net: add AF_XDP USO test Magnus Karlsson
2026-11-19 19:05   ` Maciej Fijalkowski
2026-11-20  8:35     ` Magnus Karlsson
  -- strict thread matches above, loose matches on Subject: below --
2026-11-19 14:10 [PATCH net-next 0/5] xsk: ice: USO support for AF_XDP Magnus Karlsson
2026-11-19 14:10 ` [PATCH net-next 3/5] xsk: ice: add USO zero-copy support Magnus Karlsson
2026-11-19 14:59   ` sashiko-bot

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=sashiko-outbox-225609@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=intel-wired-lan@lists.osuosl.org \
    --cc=magnus.karlsson@intel.com \
    --cc=sashiko-reviews@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.