Merge tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf

Daniel Borkmann says:

====================
pull-request: bpf 2024-01-25

The following pull-request contains BPF updates for your *net* tree.

We've added 12 non-merge commits during the last 2 day(s) which contain
a total of 13 files changed, 190 insertions(+), 91 deletions(-).

The main changes are:

1) Fix bpf_xdp_adjust_tail() in context of XSK zero-copy drivers which
   support XDP multi-buffer. The former triggered a NULL pointer
   dereference upon shrinking, from Maciej Fijalkowski & Tirthendu Sarkar.

2) Fix a bug in riscv64 BPF JIT which emitted a wrong prologue and
   epilogue for struct_ops programs, from Pu Lehui.

* tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
  i40e: update xdp_rxq_info::frag_size for ZC enabled Rx queue
  i40e: set xdp_rxq_info::frag_size
  xdp: reflect tail increase for MEM_TYPE_XSK_BUFF_POOL
  ice: update xdp_rxq_info::frag_size for ZC enabled Rx queue
  intel: xsk: initialize skb_frag_t::bv_offset in ZC drivers
  ice: remove redundant xdp_rxq_info registration
  i40e: handle multi-buffer packets that are shrunk by xdp prog
  ice: work on pre-XDP prog frag count
  xsk: fix usage of multi-buffer BPF helpers for ZC XDP
  xsk: make xsk_buff_pool responsible for clearing xdp_buff::flags
  xsk: recycle buffer in case Rx queue was full
  riscv, bpf: Fix unpredictable kernel crash about RV64 struct_ops
====================

Link: https://lore.kernel.org/r/20240125084416.10876-1-daniel@iogearbox.netSigned-off-by: Paolo Abeni <pabeni@redhat.com>

arch/riscv/net/bpf_jit_comp64.c

@@ -795,6 +795,7 @@ static int __arch_prepare_bpf_trampoline(struct bpf_tramp_image *im,
 	struct bpf_tramp_links *fentry = &tlinks[BPF_TRAMP_FENTRY];
 	struct bpf_tramp_links *fexit = &tlinks[BPF_TRAMP_FEXIT];
 	struct bpf_tramp_links *fmod_ret = &tlinks[BPF_TRAMP_MODIFY_RETURN];
+	bool is_struct_ops = flags & BPF_TRAMP_F_INDIRECT;
 	void *orig_call = func_addr;
 	bool save_ret;
 	u32 insn;
@@ -878,7 +879,7 @@ static int __arch_prepare_bpf_trampoline(struct bpf_tramp_image *im,
 
 	stack_size = round_up(stack_size, 16);
 
-	if (func_addr) {
+	if (!is_struct_ops) {
 		/* For the trampoline called from function entry,
 		 * the frame of traced function and the frame of
 		 * trampoline need to be considered.
@@ -998,7 +999,7 @@ static int __arch_prepare_bpf_trampoline(struct bpf_tramp_image *im,
 
 	emit_ld(RV_REG_S1, -sreg_off, RV_REG_FP, ctx);
 
-	if (func_addr) {
+	if (!is_struct_ops) {
 		/* trampoline called from function entry */
 		emit_ld(RV_REG_T0, stack_size - 8, RV_REG_SP, ctx);
 		emit_ld(RV_REG_FP, stack_size - 16, RV_REG_SP, ctx);

drivers/net/ethernet/intel/i40e/i40e_main.c

@@ -3588,40 +3588,55 @@ static int i40e_configure_rx_ring(struct i40e_ring *ring)
 	struct i40e_hmc_obj_rxq rx_ctx;
 	int err = 0;
 	bool ok;
-	int ret;
 
 	bitmap_zero(ring->state, __I40E_RING_STATE_NBITS);
 
 	/* clear the context structure first */
 	memset(&rx_ctx, 0, sizeof(rx_ctx));
 
-	if (ring->vsi->type == I40E_VSI_MAIN)
-		xdp_rxq_info_unreg_mem_model(&ring->xdp_rxq);
+	ring->rx_buf_len = vsi->rx_buf_len;
+
+	/* XDP RX-queue info only needed for RX rings exposed to XDP */
+	if (ring->vsi->type != I40E_VSI_MAIN)
+		goto skip;
+
+	if (!xdp_rxq_info_is_reg(&ring->xdp_rxq)) {
+		err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
+					 ring->queue_index,
+					 ring->q_vector->napi.napi_id,
+					 ring->rx_buf_len);
+		if (err)
+			return err;
+	}
 
 	ring->xsk_pool = i40e_xsk_pool(ring);
 	if (ring->xsk_pool) {
-		ring->rx_buf_len =
-		  xsk_pool_get_rx_frame_size(ring->xsk_pool);
-		ret = xdp_rxq_info_reg_mem_model(&ring->xdp_rxq,
+		xdp_rxq_info_unreg(&ring->xdp_rxq);
+		ring->rx_buf_len = xsk_pool_get_rx_frame_size(ring->xsk_pool);
+		err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
+					 ring->queue_index,
+					 ring->q_vector->napi.napi_id,
+					 ring->rx_buf_len);
+		if (err)
+			return err;
+		err = xdp_rxq_info_reg_mem_model(&ring->xdp_rxq,
 						 MEM_TYPE_XSK_BUFF_POOL,
 						 NULL);
-		if (ret)
-			return ret;
+		if (err)
+			return err;
 		dev_info(&vsi->back->pdev->dev,
 			 "Registered XDP mem model MEM_TYPE_XSK_BUFF_POOL on Rx ring %d\n",
 			 ring->queue_index);
 
 	} else {
-		ring->rx_buf_len = vsi->rx_buf_len;
-		if (ring->vsi->type == I40E_VSI_MAIN) {
-			ret = xdp_rxq_info_reg_mem_model(&ring->xdp_rxq,
-							 MEM_TYPE_PAGE_SHARED,
-							 NULL);
-			if (ret)
-				return ret;
-		}
+		err = xdp_rxq_info_reg_mem_model(&ring->xdp_rxq,
+						 MEM_TYPE_PAGE_SHARED,
+						 NULL);
+		if (err)
+			return err;
 	}
 
+skip:
 	xdp_init_buff(&ring->xdp, i40e_rx_pg_size(ring) / 2, &ring->xdp_rxq);
 
 	rx_ctx.dbuff = DIV_ROUND_UP(ring->rx_buf_len,

drivers/net/ethernet/intel/i40e/i40e_txrx.c

@@ -1548,7 +1548,6 @@ void i40e_free_rx_resources(struct i40e_ring *rx_ring)
 int i40e_setup_rx_descriptors(struct i40e_ring *rx_ring)
 {
 	struct device *dev = rx_ring->dev;
-	int err;
 
 	u64_stats_init(&rx_ring->syncp);
 
@@ -1569,14 +1568,6 @@ int i40e_setup_rx_descriptors(struct i40e_ring *rx_ring)
 	rx_ring->next_to_process = 0;
 	rx_ring->next_to_use = 0;
 
-	/* XDP RX-queue info only needed for RX rings exposed to XDP */
-	if (rx_ring->vsi->type == I40E_VSI_MAIN) {
-		err = xdp_rxq_info_reg(&rx_ring->xdp_rxq, rx_ring->netdev,
-				       rx_ring->queue_index, rx_ring->q_vector->napi.napi_id);
-		if (err < 0)
-			return err;
-	}
-
 	rx_ring->xdp_prog = rx_ring->vsi->xdp_prog;
 
 	rx_ring->rx_bi =
@@ -2087,7 +2078,8 @@ static void i40e_put_rx_buffer(struct i40e_ring *rx_ring,
 static void i40e_process_rx_buffs(struct i40e_ring *rx_ring, int xdp_res,
 				  struct xdp_buff *xdp)
 {
-	u32 next = rx_ring->next_to_clean;
+	u32 nr_frags = xdp_get_shared_info_from_buff(xdp)->nr_frags;
+	u32 next = rx_ring->next_to_clean, i = 0;
 	struct i40e_rx_buffer *rx_buffer;
 
 	xdp->flags = 0;
@@ -2100,10 +2092,10 @@ static void i40e_process_rx_buffs(struct i40e_ring *rx_ring, int xdp_res,
 		if (!rx_buffer->page)
 			continue;
 
-		if (xdp_res == I40E_XDP_CONSUMED)
-			rx_buffer->pagecnt_bias++;
-		else
+		if (xdp_res != I40E_XDP_CONSUMED)
 			i40e_rx_buffer_flip(rx_buffer, xdp->frame_sz);
+		else if (i++ <= nr_frags)
+			rx_buffer->pagecnt_bias++;
 
 		/* EOP buffer will be put in i40e_clean_rx_irq() */
 		if (next == rx_ring->next_to_process)
@@ -2117,20 +2109,20 @@ static void i40e_process_rx_buffs(struct i40e_ring *rx_ring, int xdp_res,
  * i40e_construct_skb - Allocate skb and populate it
  * @rx_ring: rx descriptor ring to transact packets on
  * @xdp: xdp_buff pointing to the data
- * @nr_frags: number of buffers for the packet
  *
  * This function allocates an skb.  It then populates it with the page
  * data from the current receive descriptor, taking care to set up the
  * skb correctly.
  */
 static struct sk_buff *i40e_construct_skb(struct i40e_ring *rx_ring,
-					  struct xdp_buff *xdp,
-					  u32 nr_frags)
+					  struct xdp_buff *xdp)
 {
 	unsigned int size = xdp->data_end - xdp->data;
 	struct i40e_rx_buffer *rx_buffer;
+	struct skb_shared_info *sinfo;
 	unsigned int headlen;
 	struct sk_buff *skb;
+	u32 nr_frags = 0;
 
 	/* prefetch first cache line of first page */
 	net_prefetch(xdp->data);
@@ -2168,6 +2160,10 @@ static struct sk_buff *i40e_construct_skb(struct i40e_ring *rx_ring,
 	memcpy(__skb_put(skb, headlen), xdp->data,
 	       ALIGN(headlen, sizeof(long)));
 
+	if (unlikely(xdp_buff_has_frags(xdp))) {
+		sinfo = xdp_get_shared_info_from_buff(xdp);
+		nr_frags = sinfo->nr_frags;
+	}
 	rx_buffer = i40e_rx_bi(rx_ring, rx_ring->next_to_clean);
 	/* update all of the pointers */
 	size -= headlen;
@@ -2187,9 +2183,8 @@ static struct sk_buff *i40e_construct_skb(struct i40e_ring *rx_ring,
 	}
 
 	if (unlikely(xdp_buff_has_frags(xdp))) {
-		struct skb_shared_info *sinfo, *skinfo = skb_shinfo(skb);
+		struct skb_shared_info *skinfo = skb_shinfo(skb);
 
-		sinfo = xdp_get_shared_info_from_buff(xdp);
 		memcpy(&skinfo->frags[skinfo->nr_frags], &sinfo->frags[0],
 		       sizeof(skb_frag_t) * nr_frags);
 
@@ -2212,17 +2207,17 @@ static struct sk_buff *i40e_construct_skb(struct i40e_ring *rx_ring,
  * i40e_build_skb - Build skb around an existing buffer
  * @rx_ring: Rx descriptor ring to transact packets on
  * @xdp: xdp_buff pointing to the data
- * @nr_frags: number of buffers for the packet
  *
  * This function builds an skb around an existing Rx buffer, taking care
  * to set up the skb correctly and avoid any memcpy overhead.
  */
 static struct sk_buff *i40e_build_skb(struct i40e_ring *rx_ring,
-				      struct xdp_buff *xdp,
-				      u32 nr_frags)
+				      struct xdp_buff *xdp)
 {
 	unsigned int metasize = xdp->data - xdp->data_meta;
+	struct skb_shared_info *sinfo;
 	struct sk_buff *skb;
+	u32 nr_frags;
 
 	/* Prefetch first cache line of first page. If xdp->data_meta
 	 * is unused, this points exactly as xdp->data, otherwise we
@@ -2231,6 +2226,11 @@ static struct sk_buff *i40e_build_skb(struct i40e_ring *rx_ring,
 	 */
 	net_prefetch(xdp->data_meta);
 
+	if (unlikely(xdp_buff_has_frags(xdp))) {
+		sinfo = xdp_get_shared_info_from_buff(xdp);
+		nr_frags = sinfo->nr_frags;
+	}
+
 	/* build an skb around the page buffer */
 	skb = napi_build_skb(xdp->data_hard_start, xdp->frame_sz);
 	if (unlikely(!skb))
@@ -2243,9 +2243,6 @@ static struct sk_buff *i40e_build_skb(struct i40e_ring *rx_ring,
 		skb_metadata_set(skb, metasize);
 
 	if (unlikely(xdp_buff_has_frags(xdp))) {
-		struct skb_shared_info *sinfo;
-
-		sinfo = xdp_get_shared_info_from_buff(xdp);
 		xdp_update_skb_shared_info(skb, nr_frags,
 					   sinfo->xdp_frags_size,
 					   nr_frags * xdp->frame_sz,
@@ -2589,9 +2586,9 @@ static int i40e_clean_rx_irq(struct i40e_ring *rx_ring, int budget,
 			total_rx_bytes += size;
 		} else {
 			if (ring_uses_build_skb(rx_ring))
-				skb = i40e_build_skb(rx_ring, xdp, nfrags);
+				skb = i40e_build_skb(rx_ring, xdp);
 			else
-				skb = i40e_construct_skb(rx_ring, xdp, nfrags);
+				skb = i40e_construct_skb(rx_ring, xdp);
 
 			/* drop if we failed to retrieve a buffer */
 			if (!skb) {

drivers/net/ethernet/intel/i40e/i40e_xsk.c

@@ -414,7 +414,8 @@ i40e_add_xsk_frag(struct i40e_ring *rx_ring, struct xdp_buff *first,
 	}
 
 	__skb_fill_page_desc_noacc(sinfo, sinfo->nr_frags++,
-				   virt_to_page(xdp->data_hard_start), 0, size);
+				   virt_to_page(xdp->data_hard_start),
+				   XDP_PACKET_HEADROOM, size);
 	sinfo->xdp_frags_size += size;
 	xsk_buff_add_frag(xdp);
 
@@ -498,7 +499,6 @@ int i40e_clean_rx_irq_zc(struct i40e_ring *rx_ring, int budget)
 		xdp_res = i40e_run_xdp_zc(rx_ring, first, xdp_prog);
 		i40e_handle_xdp_result_zc(rx_ring, first, rx_desc, &rx_packets,
 					  &rx_bytes, xdp_res, &failure);
-		first->flags = 0;
 		next_to_clean = next_to_process;
 		if (failure)
 			break;

drivers/net/ethernet/intel/ice/ice_base.c

@@ -547,19 +547,27 @@ int ice_vsi_cfg_rxq(struct ice_rx_ring *ring)
 	ring->rx_buf_len = ring->vsi->rx_buf_len;
 
 	if (ring->vsi->type == ICE_VSI_PF) {
-		if (!xdp_rxq_info_is_reg(&ring->xdp_rxq))
-			/* coverity[check_return] */
-			__xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
-					   ring->q_index,
-					   ring->q_vector->napi.napi_id,
-					   ring->vsi->rx_buf_len);
+		if (!xdp_rxq_info_is_reg(&ring->xdp_rxq)) {
+			err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
+						 ring->q_index,
+						 ring->q_vector->napi.napi_id,
+						 ring->rx_buf_len);
+			if (err)
+				return err;
+		}
 
 		ring->xsk_pool = ice_xsk_pool(ring);
 		if (ring->xsk_pool) {
-			xdp_rxq_info_unreg_mem_model(&ring->xdp_rxq);
+			xdp_rxq_info_unreg(&ring->xdp_rxq);
 
 			ring->rx_buf_len =
 				xsk_pool_get_rx_frame_size(ring->xsk_pool);
+			err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
+						 ring->q_index,
+						 ring->q_vector->napi.napi_id,
+						 ring->rx_buf_len);
+			if (err)
+				return err;
 			err = xdp_rxq_info_reg_mem_model(&ring->xdp_rxq,
 							 MEM_TYPE_XSK_BUFF_POOL,
 							 NULL);
@@ -571,13 +579,14 @@ int ice_vsi_cfg_rxq(struct ice_rx_ring *ring)
 			dev_info(dev, "Registered XDP mem model MEM_TYPE_XSK_BUFF_POOL on Rx ring %d\n",
 				 ring->q_index);
 		} else {
-			if (!xdp_rxq_info_is_reg(&ring->xdp_rxq))
-				/* coverity[check_return] */
-				__xdp_rxq_info_reg(&ring->xdp_rxq,
-						   ring->netdev,
-						   ring->q_index,
-						   ring->q_vector->napi.napi_id,
-						   ring->vsi->rx_buf_len);
+			if (!xdp_rxq_info_is_reg(&ring->xdp_rxq)) {
+				err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
+							 ring->q_index,
+							 ring->q_vector->napi.napi_id,
+							 ring->rx_buf_len);
+				if (err)
+					return err;
+			}
 
 			err = xdp_rxq_info_reg_mem_model(&ring->xdp_rxq,
 							 MEM_TYPE_PAGE_SHARED,

drivers/net/ethernet/intel/ice/ice_txrx.c

@@ -513,11 +513,6 @@ int ice_setup_rx_ring(struct ice_rx_ring *rx_ring)
 	if (ice_is_xdp_ena_vsi(rx_ring->vsi))
 		WRITE_ONCE(rx_ring->xdp_prog, rx_ring->vsi->xdp_prog);
 
-	if (rx_ring->vsi->type == ICE_VSI_PF &&
-	    !xdp_rxq_info_is_reg(&rx_ring->xdp_rxq))
-		if (xdp_rxq_info_reg(&rx_ring->xdp_rxq, rx_ring->netdev,
-				     rx_ring->q_index, rx_ring->q_vector->napi.napi_id))
-			goto err;
 	return 0;
 
 err:
@@ -603,9 +598,7 @@ ice_run_xdp(struct ice_rx_ring *rx_ring, struct xdp_buff *xdp,
 		ret = ICE_XDP_CONSUMED;
 	}
 exit:
-	rx_buf->act = ret;
-	if (unlikely(xdp_buff_has_frags(xdp)))
-		ice_set_rx_bufs_act(xdp, rx_ring, ret);
+	ice_set_rx_bufs_act(xdp, rx_ring, ret);
 }
 
 /**
@@ -893,14 +886,17 @@ ice_add_xdp_frag(struct ice_rx_ring *rx_ring, struct xdp_buff *xdp,
 	}
 
 	if (unlikely(sinfo->nr_frags == MAX_SKB_FRAGS)) {
-		if (unlikely(xdp_buff_has_frags(xdp)))
-			ice_set_rx_bufs_act(xdp, rx_ring, ICE_XDP_CONSUMED);
+		ice_set_rx_bufs_act(xdp, rx_ring, ICE_XDP_CONSUMED);
 		return -ENOMEM;
 	}
 
 	__skb_fill_page_desc_noacc(sinfo, sinfo->nr_frags++, rx_buf->page,
 				   rx_buf->page_offset, size);
 	sinfo->xdp_frags_size += size;
+	/* remember frag count before XDP prog execution; bpf_xdp_adjust_tail()
+	 * can pop off frags but driver has to handle it on its own
+	 */
+	rx_ring->nr_frags = sinfo->nr_frags;
 
 	if (page_is_pfmemalloc(rx_buf->page))
 		xdp_buff_set_frag_pfmemalloc(xdp);
@@ -1251,6 +1247,7 @@ int ice_clean_rx_irq(struct ice_rx_ring *rx_ring, int budget)
 
 		xdp->data = NULL;
 		rx_ring->first_desc = ntc;
+		rx_ring->nr_frags = 0;
 		continue;
 construct_skb:
 		if (likely(ice_ring_uses_build_skb(rx_ring)))
@@ -1266,10 +1263,12 @@ int ice_clean_rx_irq(struct ice_rx_ring *rx_ring, int budget)
 						    ICE_XDP_CONSUMED);
 			xdp->data = NULL;
 			rx_ring->first_desc = ntc;
+			rx_ring->nr_frags = 0;
 			break;
 		}
 		xdp->data = NULL;
 		rx_ring->first_desc = ntc;
+		rx_ring->nr_frags = 0;
 
 		stat_err_bits = BIT(ICE_RX_FLEX_DESC_STATUS0_RXE_S);
 		if (unlikely(ice_test_staterr(rx_desc->wb.status_error0,

drivers/net/ethernet/intel/ice/ice_txrx.h

@@ -358,6 +358,7 @@ struct ice_rx_ring {
 	struct ice_tx_ring *xdp_ring;
 	struct ice_rx_ring *next;	/* pointer to next ring in q_vector */
 	struct xsk_buff_pool *xsk_pool;
+	u32 nr_frags;
 	dma_addr_t dma;			/* physical address of ring */
 	u16 rx_buf_len;
 	u8 dcb_tc;			/* Traffic class of ring */

drivers/net/ethernet/intel/ice/ice_txrx_lib.h

@@ -12,26 +12,39 @@
  * act: action to store onto Rx buffers related to XDP buffer parts
  *
  * Set action that should be taken before putting Rx buffer from first frag
- * to one before last. Last one is handled by caller of this function as it
- * is the EOP frag that is currently being processed. This function is
- * supposed to be called only when XDP buffer contains frags.
+ * to the last.
  */
 static inline void
 ice_set_rx_bufs_act(struct xdp_buff *xdp, const struct ice_rx_ring *rx_ring,
 		    const unsigned int act)
 {
-	const struct skb_shared_info *sinfo = xdp_get_shared_info_from_buff(xdp);
-	u32 first = rx_ring->first_desc;
-	u32 nr_frags = sinfo->nr_frags;
+	u32 sinfo_frags = xdp_get_shared_info_from_buff(xdp)->nr_frags;
+	u32 nr_frags = rx_ring->nr_frags + 1;
+	u32 idx = rx_ring->first_desc;
 	u32 cnt = rx_ring->count;
 	struct ice_rx_buf *buf;
 
 	for (int i = 0; i < nr_frags; i++) {
-		buf = &rx_ring->rx_buf[first];
+		buf = &rx_ring->rx_buf[idx];
 		buf->act = act;
 
-		if (++first == cnt)
-			first = 0;
+		if (++idx == cnt)
+			idx = 0;
+	}
+
+	/* adjust pagecnt_bias on frags freed by XDP prog */
+	if (sinfo_frags < rx_ring->nr_frags && act == ICE_XDP_CONSUMED) {
+		u32 delta = rx_ring->nr_frags - sinfo_frags;
+
+		while (delta) {
+			if (idx == 0)
+				idx = cnt - 1;
+			else
+				idx--;
+			buf = &rx_ring->rx_buf[idx];
+			buf->pagecnt_bias--;
+			delta--;
+		}
 	}
 }
 

drivers/net/ethernet/intel/ice/ice_xsk.c

@@ -825,7 +825,8 @@ ice_add_xsk_frag(struct ice_rx_ring *rx_ring, struct xdp_buff *first,
 	}
 
 	__skb_fill_page_desc_noacc(sinfo, sinfo->nr_frags++,
-				   virt_to_page(xdp->data_hard_start), 0, size);
+				   virt_to_page(xdp->data_hard_start),
+				   XDP_PACKET_HEADROOM, size);
 	sinfo->xdp_frags_size += size;
 	xsk_buff_add_frag(xdp);
 
@@ -895,7 +896,6 @@ int ice_clean_rx_irq_zc(struct ice_rx_ring *rx_ring, int budget)
 
 		if (!first) {
 			first = xdp;
-			xdp_buff_clear_frags_flag(first);
 		} else if (ice_add_xsk_frag(rx_ring, first, xdp, size)) {
 			break;
 		}

include/net/xdp_sock_drv.h

@@ -159,11 +159,29 @@ static inline struct xdp_buff *xsk_buff_get_frag(struct xdp_buff *first)
 	return ret;
 }
 
+static inline void xsk_buff_del_tail(struct xdp_buff *tail)
+{
+	struct xdp_buff_xsk *xskb = container_of(tail, struct xdp_buff_xsk, xdp);
+
+	list_del(&xskb->xskb_list_node);
+}
+
+static inline struct xdp_buff *xsk_buff_get_tail(struct xdp_buff *first)
+{
+	struct xdp_buff_xsk *xskb = container_of(first, struct xdp_buff_xsk, xdp);
+	struct xdp_buff_xsk *frag;
+
+	frag = list_last_entry(&xskb->pool->xskb_list, struct xdp_buff_xsk,
+			       xskb_list_node);
+	return &frag->xdp;
+}
+
 static inline void xsk_buff_set_size(struct xdp_buff *xdp, u32 size)
 {
 	xdp->data = xdp->data_hard_start + XDP_PACKET_HEADROOM;
 	xdp->data_meta = xdp->data;
 	xdp->data_end = xdp->data + size;
+	xdp->flags = 0;
 }
 
 static inline dma_addr_t xsk_buff_raw_get_dma(struct xsk_buff_pool *pool,
@@ -350,6 +368,15 @@ static inline struct xdp_buff *xsk_buff_get_frag(struct xdp_buff *first)
 	return NULL;
 }
 
+static inline void xsk_buff_del_tail(struct xdp_buff *tail)
+{
+}
+
+static inline struct xdp_buff *xsk_buff_get_tail(struct xdp_buff *first)
+{
+	return NULL;
+}
+
 static inline void xsk_buff_set_size(struct xdp_buff *xdp, u32 size)
 {
 }

net/core/filter.c

@@ -83,6 +83,7 @@
 #include <net/netfilter/nf_conntrack_bpf.h>
 #include <net/netkit.h>
 #include <linux/un.h>
+#include <net/xdp_sock_drv.h>
 
 #include "dev.h"
 
@@ -4092,10 +4093,46 @@ static int bpf_xdp_frags_increase_tail(struct xdp_buff *xdp, int offset)
 	memset(skb_frag_address(frag) + skb_frag_size(frag), 0, offset);
 	skb_frag_size_add(frag, offset);
 	sinfo->xdp_frags_size += offset;
+	if (rxq->mem.type == MEM_TYPE_XSK_BUFF_POOL)
+		xsk_buff_get_tail(xdp)->data_end += offset;
 
 	return 0;
 }
 
+static void bpf_xdp_shrink_data_zc(struct xdp_buff *xdp, int shrink,
+				   struct xdp_mem_info *mem_info, bool release)
+{
+	struct xdp_buff *zc_frag = xsk_buff_get_tail(xdp);
+
+	if (release) {
+		xsk_buff_del_tail(zc_frag);
+		__xdp_return(NULL, mem_info, false, zc_frag);
+	} else {
+		zc_frag->data_end -= shrink;
+	}
+}
+
+static bool bpf_xdp_shrink_data(struct xdp_buff *xdp, skb_frag_t *frag,
+				int shrink)
+{
+	struct xdp_mem_info *mem_info = &xdp->rxq->mem;
+	bool release = skb_frag_size(frag) == shrink;
+
+	if (mem_info->type == MEM_TYPE_XSK_BUFF_POOL) {
+		bpf_xdp_shrink_data_zc(xdp, shrink, mem_info, release);
+		goto out;
+	}
+
+	if (release) {
+		struct page *page = skb_frag_page(frag);
+
+		__xdp_return(page_address(page), mem_info, false, NULL);
+	}
+
+out:
+	return release;
+}
+
 static int bpf_xdp_frags_shrink_tail(struct xdp_buff *xdp, int offset)
 {
 	struct skb_shared_info *sinfo = xdp_get_shared_info_from_buff(xdp);
@@ -4110,12 +4147,7 @@ static int bpf_xdp_frags_shrink_tail(struct xdp_buff *xdp, int offset)
 
 		len_free += shrink;
 		offset -= shrink;
-
-		if (skb_frag_size(frag) == shrink) {
-			struct page *page = skb_frag_page(frag);
-
-			__xdp_return(page_address(page), &xdp->rxq->mem,
-				     false, NULL);
+		if (bpf_xdp_shrink_data(xdp, frag, shrink)) {
 			n_frags_free++;
 		} else {
 			skb_frag_size_sub(frag, shrink);

net/xdp/xsk.c

@@ -167,8 +167,10 @@ static int xsk_rcv_zc(struct xdp_sock *xs, struct xdp_buff *xdp, u32 len)
 		contd = XDP_PKT_CONTD;
 
 	err = __xsk_rcv_zc(xs, xskb, len, contd);
-	if (err || likely(!frags))
-		goto out;
+	if (err)
+		goto err;
+	if (likely(!frags))
+		return 0;
 
 	xskb_list = &xskb->pool->xskb_list;
 	list_for_each_entry_safe(pos, tmp, xskb_list, xskb_list_node) {
@@ -177,11 +179,13 @@ static int xsk_rcv_zc(struct xdp_sock *xs, struct xdp_buff *xdp, u32 len)
 		len = pos->xdp.data_end - pos->xdp.data;
 		err = __xsk_rcv_zc(xs, pos, len, contd);
 		if (err)
-			return err;
+			goto err;
 		list_del(&pos->xskb_list_node);
 	}
 
-out:
+	return 0;
+err:
+	xsk_buff_free(xdp);
 	return err;
 }
 

net/xdp/xsk_buff_pool.c

@@ -555,6 +555,7 @@ struct xdp_buff *xp_alloc(struct xsk_buff_pool *pool)
 
 	xskb->xdp.data = xskb->xdp.data_hard_start + XDP_PACKET_HEADROOM;
 	xskb->xdp.data_meta = xskb->xdp.data;
+	xskb->xdp.flags = 0;
 
 	if (pool->dma_need_sync) {
 		dma_sync_single_range_for_device(pool->dev, xskb->dma, 0,