public inbox for netdev@vger.kernel.org
 help / color / mirror / Atom feed
* [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support
@ 2026-03-06 23:00 Chris J Arges
  2026-03-06 23:00 ` [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context Chris J Arges
                   ` (6 more replies)
  0 siblings, 7 replies; 16+ messages in thread
From: Chris J Arges @ 2026-03-06 23:00 UTC (permalink / raw)
  To: michael.chan, pavan.chebbi, joe, kuba, Alexei Starovoitov,
	Daniel Borkmann, David S. Miller, Jesper Dangaard Brouer,
	John Fastabend, Stanislav Fomichev
  Cc: kernel-team, Chris J Arges, netdev, bpf

This series adds XDP RSS hash metadata extraction support for the bnxt_en
driver and includes selftests to validate the functionality. I was able
to test this on a BCM57414 NIC.

Note: I don't have hardware to test v3 completions at the moment.

Patches 1-3 implement the driver support:
  - Patch 1 adds bnxt_xdp_buff to wrap xdp_buff with
    pointers to the hardware RX completion descriptors and the
    completion type. This is similar to other driver implementations.
  - Patch 2 adds the xmo_rx_hash callback for CMP_TYPE_RX_L2_CMP completions.
  - Patch 3 adds handling for CMP_TYPE_RX_L2_V3_CMP completions

Patches 4-5 add selftest coverage:
  - Patch 4 refactors common BPF map helper functions out of the
    existing xdp.py selftest into a shared net/lib/py/bpf.py module.
  - Patch 5 adds an XDP metadata selftest that loads an XDP program
    calling bpf_xdp_metadata_rx_hash. The test skips on devices without
    xdp-rx-metadata-features.

v2: https://lore.kernel.org/netdev/20260303024510.644962-1-carges@cloudflare.com/
  - bnxt xdp: actually fixed RCT issues
  - bnxt xdp: Added patch 3 to handle CMP_TYPE_RX_L2_V3_CMP completions.

v1: https://lore.kernel.org/netdev/cover.1772211065.git.carges@cloudflare.com/
  - selftests: moved tests into net/hw
  - selftests: fix pylinter errors
  - selftests: add check for no error count
  - bnxt_xdp: followed RCT for variable declaration
  - rebased on net-next

Chris J Arges (5):
  bnxt_en: use bnxt_xdp_buff for xdp context
  bnxt_en: Implement XDP RSS hash metadata extraction
  bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP
  selftests: net: move common xdp.py functions into lib
  selftests: drv-net: xdp: Add rss_hash metadata tests

 drivers/net/ethernet/broadcom/bnxt/bnxt.c     |  30 ++--
 drivers/net/ethernet/broadcom/bnxt/bnxt.h     |   2 +
 drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c |  67 +++++++
 drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h |  10 ++
 .../testing/selftests/drivers/net/hw/Makefile |   1 +
 .../drivers/net/hw/lib/py/__init__.py         |   2 +
 .../selftests/drivers/net/hw/xdp_metadata.py  | 146 ++++++++++++++++
 .../selftests/drivers/net/lib/py/__init__.py  |   2 +
 tools/testing/selftests/drivers/net/xdp.py    |  95 +++-------
 .../testing/selftests/net/lib/py/__init__.py  |   2 +
 tools/testing/selftests/net/lib/py/bpf.py     |  68 ++++++++
 .../selftests/net/lib/xdp_metadata.bpf.c      | 163 ++++++++++++++++++
 12 files changed, 506 insertions(+), 82 deletions(-)
 create mode 100644 tools/testing/selftests/drivers/net/hw/xdp_metadata.py
 create mode 100644 tools/testing/selftests/net/lib/py/bpf.py
 create mode 100644 tools/testing/selftests/net/lib/xdp_metadata.bpf.c

--
2.43.0


^ permalink raw reply	[flat|nested] 16+ messages in thread

* [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context
  2026-03-06 23:00 [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Chris J Arges
@ 2026-03-06 23:00 ` Chris J Arges
  2026-03-10  0:52   ` Joe Damato
  2026-03-13  0:14   ` Andy Gospodarek
  2026-03-06 23:00 ` [PATCH net-next v3 2/5] bnxt_en: Implement XDP RSS hash metadata extraction Chris J Arges
                   ` (5 subsequent siblings)
  6 siblings, 2 replies; 16+ messages in thread
From: Chris J Arges @ 2026-03-06 23:00 UTC (permalink / raw)
  To: michael.chan, pavan.chebbi, joe, kuba, Andrew Lunn,
	David S. Miller, Eric Dumazet, Paolo Abeni, Alexei Starovoitov,
	Daniel Borkmann, Jesper Dangaard Brouer, John Fastabend,
	Stanislav Fomichev
  Cc: kernel-team, Chris J Arges, netdev, linux-kernel, bpf

This adds bnxt_xdp_buff which embeds the xdp_buff struct and stores
pointers to hardware RX completion descriptors (rx_cmp and rx_cmp_ext)
along with the completion type.

Signed-off-by: Chris J Arges <carges@cloudflare.com>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c     | 25 +++++++++++--------
 drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h |  7 ++++++
 2 files changed, 22 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index c982aac714d1..ab73aad40593 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -2118,7 +2118,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 	bool xdp_active = false;
 	dma_addr_t dma_addr;
 	struct sk_buff *skb;
-	struct xdp_buff xdp;
+	struct bnxt_xdp_buff bnxt_xdp;
 	u32 flags, misc;
 	u32 cmpl_ts;
 	void *data;
@@ -2231,9 +2231,13 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 	dma_addr = rx_buf->mapping;
 
 	if (bnxt_xdp_attached(bp, rxr)) {
-		bnxt_xdp_buff_init(bp, rxr, cons, data_ptr, len, &xdp);
+		bnxt_xdp.rxcmp = rxcmp;
+		bnxt_xdp.rxcmp1 = rxcmp1;
+		bnxt_xdp.cmp_type = cmp_type;
+
+		bnxt_xdp_buff_init(bp, rxr, cons, data_ptr, len, &bnxt_xdp.xdp);
 		if (agg_bufs) {
-			u32 frag_len = bnxt_rx_agg_netmems_xdp(bp, cpr, &xdp,
+			u32 frag_len = bnxt_rx_agg_netmems_xdp(bp, cpr, &bnxt_xdp.xdp,
 							       cp_cons,
 							       agg_bufs,
 							       false);
@@ -2245,12 +2249,12 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 	}
 
 	if (xdp_active) {
-		if (bnxt_rx_xdp(bp, rxr, cons, &xdp, data, &data_ptr, &len, event)) {
+		if (bnxt_rx_xdp(bp, rxr, cons, &bnxt_xdp.xdp, data, &data_ptr, &len, event)) {
 			rc = 1;
 			goto next_rx;
 		}
-		if (xdp_buff_has_frags(&xdp)) {
-			sinfo = xdp_get_shared_info_from_buff(&xdp);
+		if (xdp_buff_has_frags(&bnxt_xdp.xdp)) {
+			sinfo = xdp_get_shared_info_from_buff(&bnxt_xdp.xdp);
 			agg_bufs = sinfo->nr_frags;
 		} else {
 			agg_bufs = 0;
@@ -2261,7 +2265,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 		if (!xdp_active)
 			skb = bnxt_copy_skb(bnapi, data_ptr, len, dma_addr);
 		else
-			skb = bnxt_copy_xdp(bnapi, &xdp, len, dma_addr);
+			skb = bnxt_copy_xdp(bnapi, &bnxt_xdp.xdp, len, dma_addr);
 		bnxt_reuse_rx_data(rxr, cons, data);
 		if (!skb) {
 			if (agg_bufs) {
@@ -2269,7 +2273,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 					bnxt_reuse_rx_agg_bufs(cpr, cp_cons, 0,
 							       agg_bufs, false);
 				else
-					bnxt_xdp_buff_frags_free(rxr, &xdp);
+					bnxt_xdp_buff_frags_free(rxr, &bnxt_xdp.xdp);
 			}
 			goto oom_next_rx;
 		}
@@ -2293,10 +2297,11 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 			if (!skb)
 				goto oom_next_rx;
 		} else {
-			skb = bnxt_xdp_build_skb(bp, skb, agg_bufs, rxr, &xdp);
+			skb = bnxt_xdp_build_skb(bp, skb, agg_bufs,
+						 rxr, &bnxt_xdp.xdp);
 			if (!skb) {
 				/* we should be able to free the old skb here */
-				bnxt_xdp_buff_frags_free(rxr, &xdp);
+				bnxt_xdp_buff_frags_free(rxr, &bnxt_xdp.xdp);
 				goto oom_next_rx;
 			}
 		}
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
index 8933a0dec09a..8c66698bde11 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
@@ -12,6 +12,13 @@
 
 DECLARE_STATIC_KEY_FALSE(bnxt_xdp_locking_key);
 
+struct bnxt_xdp_buff {
+	struct xdp_buff xdp;
+	struct rx_cmp *rxcmp;
+	struct rx_cmp_ext *rxcmp1;
+	u8 cmp_type;
+};
+
 struct bnxt_sw_tx_bd *bnxt_xmit_bd(struct bnxt *bp,
 				   struct bnxt_tx_ring_info *txr,
 				   dma_addr_t mapping, u32 len,
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* [PATCH net-next v3 2/5] bnxt_en: Implement XDP RSS hash metadata extraction
  2026-03-06 23:00 [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Chris J Arges
  2026-03-06 23:00 ` [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context Chris J Arges
@ 2026-03-06 23:00 ` Chris J Arges
  2026-03-10  0:56   ` Joe Damato
  2026-03-13  0:15   ` Andy Gospodarek
  2026-03-06 23:00 ` [PATCH net-next v3 3/5] bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP Chris J Arges
                   ` (4 subsequent siblings)
  6 siblings, 2 replies; 16+ messages in thread
From: Chris J Arges @ 2026-03-06 23:00 UTC (permalink / raw)
  To: michael.chan, pavan.chebbi, joe, kuba, Andrew Lunn,
	David S. Miller, Eric Dumazet, Paolo Abeni, Alexei Starovoitov,
	Daniel Borkmann, Jesper Dangaard Brouer, John Fastabend,
	Stanislav Fomichev
  Cc: kernel-team, Chris J Arges, netdev, linux-kernel, bpf

Add support for extracting RSS hash values and hash types from hardware
completion descriptors in XDP programs for bnxt_en.

Add IP_TYPE definition for determining if completion is ipv4 or ipv6. In
addition add ITYPE_ICMP flag for identifying ICMP completions.

Signed-off-by: Chris J Arges <carges@cloudflare.com>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c     |  5 ++
 drivers/net/ethernet/broadcom/bnxt/bnxt.h     |  2 +
 drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 47 +++++++++++++++++++
 drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h |  3 ++
 4 files changed, 57 insertions(+)

diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index ab73aad40593..c06033cf9f82 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -15906,6 +15906,10 @@ static const struct net_device_ops bnxt_netdev_ops = {
 	.ndo_hwtstamp_set	= bnxt_hwtstamp_set,
 };
 
+static const struct xdp_metadata_ops bnxt_xdp_metadata_ops = {
+	.xmo_rx_hash		= bnxt_xdp_rx_hash,
+};
+
 static void bnxt_get_queue_stats_rx(struct net_device *dev, int i,
 				    struct netdev_queue_stats_rx *stats)
 {
@@ -16790,6 +16794,7 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 		goto init_err_free;
 
 	dev->netdev_ops = &bnxt_netdev_ops;
+	dev->xdp_metadata_ops = &bnxt_xdp_metadata_ops;
 	dev->stat_ops = &bnxt_stat_ops;
 	dev->watchdog_timeo = BNXT_TX_TIMEOUT;
 	dev->ethtool_ops = &bnxt_ethtool_ops;
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.h b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
index 90fa3e93c8d6..34bb98709f2b 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.h
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
@@ -232,6 +232,7 @@ struct rx_cmp {
 	 #define RX_CMP_FLAGS_ITYPE_UDP				 (3 << 12)
 	 #define RX_CMP_FLAGS_ITYPE_FCOE			 (4 << 12)
 	 #define RX_CMP_FLAGS_ITYPE_ROCE			 (5 << 12)
+	 #define RX_CMP_FLAGS_ITYPE_ICMP			 (7 << 12)
 	 #define RX_CMP_FLAGS_ITYPE_PTP_WO_TS			 (8 << 12)
 	 #define RX_CMP_FLAGS_ITYPE_PTP_W_TS			 (9 << 12)
 	#define RX_CMP_LEN					(0xffff << 16)
@@ -311,6 +312,7 @@ struct rx_cmp_ext {
 	#define RX_CMP_FLAGS2_T_IP_CS_CALC			(0x1 << 2)
 	#define RX_CMP_FLAGS2_T_L4_CS_CALC			(0x1 << 3)
 	#define RX_CMP_FLAGS2_META_FORMAT_VLAN			(0x1 << 4)
+	#define RX_CMP_FLAGS2_IP_TYPE				(0x1 << 8)
 	__le32 rx_cmp_meta_data;
 	#define RX_CMP_FLAGS2_METADATA_TCI_MASK			0xffff
 	#define RX_CMP_FLAGS2_METADATA_VID_MASK			0xfff
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
index 85cbeb35681c..1f920464b426 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
@@ -472,3 +472,50 @@ bnxt_xdp_build_skb(struct bnxt *bp, struct sk_buff *skb, u8 num_frags,
 				  xdp_buff_get_skb_flags(xdp));
 	return skb;
 }
+
+int bnxt_xdp_rx_hash(const struct xdp_md *ctx, u32 *hash,
+		     enum xdp_rss_hash_type *rss_type)
+{
+	const struct bnxt_xdp_buff *xdp = (void *)ctx;
+	const struct rx_cmp_ext *rxcmp1 = xdp->rxcmp1;
+	const struct rx_cmp *rxcmp = xdp->rxcmp;
+	enum xdp_rss_hash_type hash_type = 0;
+	u32 itypes;
+
+	if (!rxcmp || !RX_CMP_HASH_VALID(rxcmp))
+		return -ENODATA;
+
+	*hash = le32_to_cpu(rxcmp->rx_cmp_rss_hash);
+
+	if (!rxcmp1) {
+		*rss_type = XDP_RSS_TYPE_L2;
+		return 0;
+	}
+
+	if (xdp->cmp_type == CMP_TYPE_RX_L2_CMP) {
+		itypes = RX_CMP_ITYPES(rxcmp);
+		if (rxcmp1->rx_cmp_flags2 &
+		    cpu_to_le32(RX_CMP_FLAGS2_IP_TYPE)) {
+			hash_type |= XDP_RSS_TYPE_L3_IPV6;
+		} else {
+			hash_type |= XDP_RSS_TYPE_L3_IPV4;
+		}
+
+		switch (itypes) {
+		case RX_CMP_FLAGS_ITYPE_TCP:
+			hash_type |= XDP_RSS_L4 | XDP_RSS_L4_TCP;
+			break;
+		case RX_CMP_FLAGS_ITYPE_UDP:
+			hash_type |= XDP_RSS_L4 | XDP_RSS_L4_UDP;
+			break;
+		case RX_CMP_FLAGS_ITYPE_ICMP:
+			hash_type |= XDP_RSS_L4 | XDP_RSS_L4_ICMP;
+			break;
+		default:
+			break;
+		}
+	}
+
+	*rss_type = hash_type;
+	return 0;
+}
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
index 8c66698bde11..fb4f9143929f 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
@@ -41,4 +41,7 @@ void bnxt_xdp_buff_frags_free(struct bnxt_rx_ring_info *rxr,
 struct sk_buff *bnxt_xdp_build_skb(struct bnxt *bp, struct sk_buff *skb,
 				   u8 num_frags, struct bnxt_rx_ring_info *rxr,
 				   struct xdp_buff *xdp);
+int bnxt_xdp_rx_hash(const struct xdp_md *ctx, u32 *hash,
+		     enum xdp_rss_hash_type *rss_type);
+
 #endif
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* [PATCH net-next v3 3/5] bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP
  2026-03-06 23:00 [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Chris J Arges
  2026-03-06 23:00 ` [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context Chris J Arges
  2026-03-06 23:00 ` [PATCH net-next v3 2/5] bnxt_en: Implement XDP RSS hash metadata extraction Chris J Arges
@ 2026-03-06 23:00 ` Chris J Arges
  2026-03-10  1:02   ` Joe Damato
  2026-03-13  0:15   ` Andy Gospodarek
  2026-03-06 23:00 ` [PATCH net-next v3 4/5] selftests: net: move common xdp.py functions into lib Chris J Arges
                   ` (3 subsequent siblings)
  6 siblings, 2 replies; 16+ messages in thread
From: Chris J Arges @ 2026-03-06 23:00 UTC (permalink / raw)
  To: michael.chan, pavan.chebbi, joe, kuba, Alexei Starovoitov,
	Daniel Borkmann, David S. Miller, Jesper Dangaard Brouer,
	John Fastabend, Stanislav Fomichev, Andrew Lunn, Eric Dumazet,
	Paolo Abeni
  Cc: kernel-team, Chris J Arges, netdev, bpf, linux-kernel

This adds another conditional for if cmp_type is CMP_TYPE_RX_L2_V3_CMP for
drivers that support this completion format.

It uses a similar approach as found in other parts of the driver.

Signed-off-by: Chris J Arges <carges@cloudflare.com>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 20 +++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
index 1f920464b426..25ecef1eb96e 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
@@ -514,6 +514,26 @@ int bnxt_xdp_rx_hash(const struct xdp_md *ctx, u32 *hash,
 		default:
 			break;
 		}
+	} else if (xdp->cmp_type == CMP_TYPE_RX_L2_V3_CMP) {
+		struct bnxt *bp = netdev_priv(xdp->xdp.rxq->dev);
+		u8 ext_op;
+
+		if (rxcmp1->rx_cmp_flags2 & cpu_to_le32(RX_CMP_FLAGS2_IP_TYPE))
+			hash_type |= XDP_RSS_TYPE_L3_IPV6;
+		else
+			hash_type |= XDP_RSS_TYPE_L3_IPV4;
+
+		ext_op = RX_CMP_V3_HASH_TYPE(bp, rxcmp);
+		switch (ext_op) {
+		case EXT_OP_INNER_4:
+		case EXT_OP_OUTER_4:
+		case EXT_OP_INNFL_3:
+		case EXT_OP_OUTFL_3:
+			hash_type |= XDP_RSS_L4;
+			break;
+		default:
+			break;
+		}
 	}
 
 	*rss_type = hash_type;
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* [PATCH net-next v3 4/5] selftests: net: move common xdp.py functions into lib
  2026-03-06 23:00 [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Chris J Arges
                   ` (2 preceding siblings ...)
  2026-03-06 23:00 ` [PATCH net-next v3 3/5] bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP Chris J Arges
@ 2026-03-06 23:00 ` Chris J Arges
  2026-03-10  0:54   ` Joe Damato
  2026-03-06 23:00 ` [PATCH net-next v3 5/5] selftests: drv-net: xdp: Add rss_hash metadata tests Chris J Arges
                   ` (2 subsequent siblings)
  6 siblings, 1 reply; 16+ messages in thread
From: Chris J Arges @ 2026-03-06 23:00 UTC (permalink / raw)
  To: michael.chan, pavan.chebbi, joe, kuba, Andrew Lunn,
	David S. Miller, Eric Dumazet, Paolo Abeni, Shuah Khan,
	Simon Horman, Alexei Starovoitov, Daniel Borkmann,
	Jesper Dangaard Brouer, John Fastabend, Stanislav Fomichev
  Cc: kernel-team, Chris J Arges, linux-kernel, netdev, linux-kselftest,
	bpf

This moves a few functions which can be useful to other python programs
that manipulate XDP programs. This also refactors xdp.py to use the
refactored functions.

Signed-off-by: Chris J Arges <carges@cloudflare.com>
---
 .../selftests/drivers/net/lib/py/__init__.py  |  2 +
 tools/testing/selftests/drivers/net/xdp.py    | 95 +++++--------------
 .../testing/selftests/net/lib/py/__init__.py  |  2 +
 tools/testing/selftests/net/lib/py/bpf.py     | 68 +++++++++++++
 4 files changed, 95 insertions(+), 72 deletions(-)
 create mode 100644 tools/testing/selftests/net/lib/py/bpf.py

diff --git a/tools/testing/selftests/drivers/net/lib/py/__init__.py b/tools/testing/selftests/drivers/net/lib/py/__init__.py
index 5872d114f142..f64e84b90574 100644
--- a/tools/testing/selftests/drivers/net/lib/py/__init__.py
+++ b/tools/testing/selftests/drivers/net/lib/py/__init__.py
@@ -23,6 +23,7 @@ try:
     from net.lib.py import CmdExitFailure
     from net.lib.py import bkg, cmd, bpftool, bpftrace, defer, ethtool, \
         fd_read_timeout, ip, rand_port, rand_ports, wait_port_listen, wait_file
+    from net.lib.py import bpf_map_set, bpf_map_dump, bpf_prog_map_ids
     from net.lib.py import KsftSkipEx, KsftFailEx, KsftXfailEx
     from net.lib.py import ksft_disruptive, ksft_exit, ksft_pr, ksft_run, \
         ksft_setup, ksft_variants, KsftNamedVariant
@@ -36,6 +37,7 @@ try:
                "bkg", "cmd", "bpftool", "bpftrace", "defer", "ethtool",
                "fd_read_timeout", "ip", "rand_port", "rand_ports",
                "wait_port_listen", "wait_file",
+               "bpf_map_set", "bpf_map_dump", "bpf_prog_map_ids",
                "KsftSkipEx", "KsftFailEx", "KsftXfailEx",
                "ksft_disruptive", "ksft_exit", "ksft_pr", "ksft_run",
                "ksft_setup", "ksft_variants", "KsftNamedVariant",
diff --git a/tools/testing/selftests/drivers/net/xdp.py b/tools/testing/selftests/drivers/net/xdp.py
index e54df158dfe9..10d821156db1 100755
--- a/tools/testing/selftests/drivers/net/xdp.py
+++ b/tools/testing/selftests/drivers/net/xdp.py
@@ -16,7 +16,8 @@ from lib.py import KsftNamedVariant, ksft_variants
 from lib.py import KsftFailEx, NetDrvEpEnv
 from lib.py import EthtoolFamily, NetdevFamily, NlError
 from lib.py import bkg, cmd, rand_port, wait_port_listen
-from lib.py import ip, bpftool, defer
+from lib.py import ip, defer
+from lib.py import bpf_map_set, bpf_map_dump, bpf_prog_map_ids
 
 
 class TestConfig(Enum):
@@ -122,47 +123,11 @@ def _load_xdp_prog(cfg, bpf_info):
     xdp_info = ip(f"-d link show dev {cfg.ifname}", json=True)[0]
     prog_info["id"] = xdp_info["xdp"]["prog"]["id"]
     prog_info["name"] = xdp_info["xdp"]["prog"]["name"]
-    prog_id = prog_info["id"]
-
-    map_ids = bpftool(f"prog show id {prog_id}", json=True)["map_ids"]
-    prog_info["maps"] = {}
-    for map_id in map_ids:
-        name = bpftool(f"map show id {map_id}", json=True)["name"]
-        prog_info["maps"][name] = map_id
+    prog_info["maps"] = bpf_prog_map_ids(prog_info["id"])
 
     return prog_info
 
 
-def format_hex_bytes(value):
-    """
-    Helper function that converts an integer into a formatted hexadecimal byte string.
-
-    Args:
-        value: An integer representing the number to be converted.
-
-    Returns:
-        A string representing hexadecimal equivalent of value, with bytes separated by spaces.
-    """
-    hex_str = value.to_bytes(4, byteorder='little', signed=True)
-    return ' '.join(f'{byte:02x}' for byte in hex_str)
-
-
-def _set_xdp_map(map_name, key, value):
-    """
-    Updates an XDP map with a given key-value pair using bpftool.
-
-    Args:
-        map_name: The name of the XDP map to update.
-        key: The key to update in the map, formatted as a hexadecimal string.
-        value: The value to associate with the key, formatted as a hexadecimal string.
-    """
-    key_formatted = format_hex_bytes(key)
-    value_formatted = format_hex_bytes(value)
-    bpftool(
-        f"map update name {map_name} key hex {key_formatted} value hex {value_formatted}"
-    )
-
-
 def _get_stats(xdp_map_id):
     """
     Retrieves and formats statistics from an XDP map.
@@ -177,25 +142,11 @@ def _get_stats(xdp_map_id):
     Raises:
         KsftFailEx: If the stats retrieval fails.
     """
-    stats_dump = bpftool(f"map dump id {xdp_map_id}", json=True)
-    if not stats_dump:
+    stats = bpf_map_dump(xdp_map_id)
+    if not stats:
         raise KsftFailEx(f"Failed to get stats for map {xdp_map_id}")
 
-    stats_formatted = {}
-    for key in range(0, 5):
-        val = stats_dump[key]["formatted"]["value"]
-        if stats_dump[key]["formatted"]["key"] == XDPStats.RX.value:
-            stats_formatted[XDPStats.RX.value] = val
-        elif stats_dump[key]["formatted"]["key"] == XDPStats.PASS.value:
-            stats_formatted[XDPStats.PASS.value] = val
-        elif stats_dump[key]["formatted"]["key"] == XDPStats.DROP.value:
-            stats_formatted[XDPStats.DROP.value] = val
-        elif stats_dump[key]["formatted"]["key"] == XDPStats.TX.value:
-            stats_formatted[XDPStats.TX.value] = val
-        elif stats_dump[key]["formatted"]["key"] == XDPStats.ABORT.value:
-            stats_formatted[XDPStats.ABORT.value] = val
-
-    return stats_formatted
+    return stats
 
 
 def _test_pass(cfg, bpf_info, msg_sz):
@@ -211,8 +162,8 @@ def _test_pass(cfg, bpf_info, msg_sz):
     prog_info = _load_xdp_prog(cfg, bpf_info)
     port = rand_port()
 
-    _set_xdp_map("map_xdp_setup", TestConfig.MODE.value, XDPAction.PASS.value)
-    _set_xdp_map("map_xdp_setup", TestConfig.PORT.value, port)
+    bpf_map_set("map_xdp_setup", TestConfig.MODE.value, XDPAction.PASS.value)
+    bpf_map_set("map_xdp_setup", TestConfig.PORT.value, port)
 
     ksft_eq(_test_udp(cfg, port, msg_sz), True, "UDP packet exchange failed")
     stats = _get_stats(prog_info["maps"]["map_xdp_stats"])
@@ -258,8 +209,8 @@ def _test_drop(cfg, bpf_info, msg_sz):
     prog_info = _load_xdp_prog(cfg, bpf_info)
     port = rand_port()
 
-    _set_xdp_map("map_xdp_setup", TestConfig.MODE.value, XDPAction.DROP.value)
-    _set_xdp_map("map_xdp_setup", TestConfig.PORT.value, port)
+    bpf_map_set("map_xdp_setup", TestConfig.MODE.value, XDPAction.DROP.value)
+    bpf_map_set("map_xdp_setup", TestConfig.PORT.value, port)
 
     ksft_eq(_test_udp(cfg, port, msg_sz), False, "UDP packet exchange should fail")
     stats = _get_stats(prog_info["maps"]["map_xdp_stats"])
@@ -305,8 +256,8 @@ def _test_xdp_native_tx(cfg, bpf_info, payload_lens):
     prog_info = _load_xdp_prog(cfg, bpf_info)
     port = rand_port()
 
-    _set_xdp_map("map_xdp_setup", TestConfig.MODE.value, XDPAction.TX.value)
-    _set_xdp_map("map_xdp_setup", TestConfig.PORT.value, port)
+    bpf_map_set("map_xdp_setup", TestConfig.MODE.value, XDPAction.TX.value)
+    bpf_map_set("map_xdp_setup", TestConfig.PORT.value, port)
 
     expected_pkts = 0
     for payload_len in payload_lens:
@@ -454,15 +405,15 @@ def _test_xdp_native_tail_adjst(cfg, pkt_sz_lst, offset_lst):
     prog_info = _load_xdp_prog(cfg, bpf_info)
 
     # Configure the XDP map for tail adjustment
-    _set_xdp_map("map_xdp_setup", TestConfig.MODE.value, XDPAction.TAIL_ADJST.value)
-    _set_xdp_map("map_xdp_setup", TestConfig.PORT.value, port)
+    bpf_map_set("map_xdp_setup", TestConfig.MODE.value, XDPAction.TAIL_ADJST.value)
+    bpf_map_set("map_xdp_setup", TestConfig.PORT.value, port)
 
     for offset in offset_lst:
         tag = format(random.randint(65, 90), "02x")
 
-        _set_xdp_map("map_xdp_setup", TestConfig.ADJST_OFFSET.value, offset)
+        bpf_map_set("map_xdp_setup", TestConfig.ADJST_OFFSET.value, offset)
         if offset > 0:
-            _set_xdp_map("map_xdp_setup", TestConfig.ADJST_TAG.value, int(tag, 16))
+            bpf_map_set("map_xdp_setup", TestConfig.ADJST_TAG.value, int(tag, 16))
 
         for pkt_sz in pkt_sz_lst:
             test_str = "".join(random.choice(string.ascii_lowercase) for _ in range(pkt_sz))
@@ -574,8 +525,8 @@ def _test_xdp_native_head_adjst(cfg, prog, pkt_sz_lst, offset_lst):
     prog_info = _load_xdp_prog(cfg, BPFProgInfo(prog, "xdp_native.bpf.o", "xdp.frags", 9000))
     port = rand_port()
 
-    _set_xdp_map("map_xdp_setup", TestConfig.MODE.value, XDPAction.HEAD_ADJST.value)
-    _set_xdp_map("map_xdp_setup", TestConfig.PORT.value, port)
+    bpf_map_set("map_xdp_setup", TestConfig.MODE.value, XDPAction.HEAD_ADJST.value)
+    bpf_map_set("map_xdp_setup", TestConfig.PORT.value, port)
 
     hds_thresh = get_hds_thresh(cfg)
     for offset in offset_lst:
@@ -595,11 +546,11 @@ def _test_xdp_native_head_adjst(cfg, prog, pkt_sz_lst, offset_lst):
             test_str = ''.join(random.choice(string.ascii_lowercase) for _ in range(pkt_sz))
             tag = format(random.randint(65, 90), '02x')
 
-            _set_xdp_map("map_xdp_setup",
+            bpf_map_set("map_xdp_setup",
                      TestConfig.ADJST_OFFSET.value,
                      offset)
-            _set_xdp_map("map_xdp_setup", TestConfig.ADJST_TAG.value, int(tag, 16))
-            _set_xdp_map("map_xdp_setup", TestConfig.ADJST_OFFSET.value, offset)
+            bpf_map_set("map_xdp_setup", TestConfig.ADJST_TAG.value, int(tag, 16))
+            bpf_map_set("map_xdp_setup", TestConfig.ADJST_OFFSET.value, offset)
 
             recvd_str = _exchg_udp(cfg, port, test_str)
 
@@ -691,8 +642,8 @@ def test_xdp_native_qstats(cfg, act):
     prog_info = _load_xdp_prog(cfg, bpf_info)
     port = rand_port()
 
-    _set_xdp_map("map_xdp_setup", TestConfig.MODE.value, act.value)
-    _set_xdp_map("map_xdp_setup", TestConfig.PORT.value, port)
+    bpf_map_set("map_xdp_setup", TestConfig.MODE.value, act.value)
+    bpf_map_set("map_xdp_setup", TestConfig.PORT.value, port)
 
     # Discard the input, but we need a listener to avoid ICMP errors
     rx_udp = f"socat -{cfg.addr_ipver} -T 2 -u UDP-RECV:{port},reuseport " + \
diff --git a/tools/testing/selftests/net/lib/py/__init__.py b/tools/testing/selftests/net/lib/py/__init__.py
index a584e7f806a4..ff9b83890773 100644
--- a/tools/testing/selftests/net/lib/py/__init__.py
+++ b/tools/testing/selftests/net/lib/py/__init__.py
@@ -15,6 +15,7 @@ from .nsim import NetdevSim, NetdevSimDev
 from .utils import CmdExitFailure, fd_read_timeout, cmd, bkg, defer, \
     bpftool, ip, ethtool, bpftrace, rand_port, rand_ports, wait_port_listen, \
     wait_file, tool
+from .bpf import bpf_map_set, bpf_map_dump, bpf_prog_map_ids
 from .ynl import NlError, YnlFamily, EthtoolFamily, NetdevFamily, RtnlFamily, RtnlAddrFamily
 from .ynl import NetshaperFamily, DevlinkFamily, PSPFamily
 
@@ -28,6 +29,7 @@ __all__ = ["KSRC",
            "CmdExitFailure", "fd_read_timeout", "cmd", "bkg", "defer",
            "bpftool", "ip", "ethtool", "bpftrace", "rand_port", "rand_ports",
            "wait_port_listen", "wait_file", "tool",
+           "bpf_map_set", "bpf_map_dump", "bpf_prog_map_ids",
            "NetdevSim", "NetdevSimDev",
            "NetshaperFamily", "DevlinkFamily", "PSPFamily", "NlError",
            "YnlFamily", "EthtoolFamily", "NetdevFamily", "RtnlFamily",
diff --git a/tools/testing/selftests/net/lib/py/bpf.py b/tools/testing/selftests/net/lib/py/bpf.py
new file mode 100644
index 000000000000..96b29d41c34b
--- /dev/null
+++ b/tools/testing/selftests/net/lib/py/bpf.py
@@ -0,0 +1,68 @@
+# SPDX-License-Identifier: GPL-2.0
+
+"""
+BPF helper utilities for kernel selftests.
+
+Provides common operations for interacting with BPF maps and programs
+via bpftool, used by XDP and other BPF-based test files.
+"""
+
+from .utils import bpftool
+
+def format_hex_bytes(value):
+    """
+    Helper function that converts an integer into a formatted hexadecimal byte string.
+
+    Args:
+        value: An integer representing the number to be converted.
+
+    Returns:
+        A string representing hexadecimal equivalent of value, with bytes separated by spaces.
+    """
+    hex_str = value.to_bytes(4, byteorder='little', signed=True)
+    return ' '.join(f'{byte:02x}' for byte in hex_str)
+
+
+def bpf_map_set(map_name, key, value):
+    """
+    Updates an XDP map with a given key-value pair using bpftool.
+
+    Args:
+        map_name: The name of the XDP map to update.
+        key: The key to update in the map, formatted as a hexadecimal string.
+        value: The value to associate with the key, formatted as a hexadecimal string.
+    """
+    key_formatted = format_hex_bytes(key)
+    value_formatted = format_hex_bytes(value)
+    bpftool(
+        f"map update name {map_name} key hex {key_formatted} value hex {value_formatted}"
+    )
+
+def bpf_map_dump(map_id):
+    """Dump all entries of a BPF array map.
+
+    Args:
+        map_id: Numeric map ID (as returned by bpftool prog show).
+
+    Returns:
+        A dict mapping formatted key (int) to formatted value (int).
+    """
+    raw = bpftool(f"map dump id {map_id}", json=True)
+    return {e["formatted"]["key"]: e["formatted"]["value"] for e in raw}
+
+
+def bpf_prog_map_ids(prog_id):
+    """Get the map name-to-ID mapping for a loaded BPF program.
+
+    Args:
+        prog_id: Numeric program ID.
+
+    Returns:
+        A dict mapping map name (str) to map ID (int).
+    """
+    map_ids = bpftool(f"prog show id {prog_id}", json=True)["map_ids"]
+    maps = {}
+    for mid in map_ids:
+        name = bpftool(f"map show id {mid}", json=True)["name"]
+        maps[name] = mid
+    return maps
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* [PATCH net-next v3 5/5] selftests: drv-net: xdp: Add rss_hash metadata tests
  2026-03-06 23:00 [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Chris J Arges
                   ` (3 preceding siblings ...)
  2026-03-06 23:00 ` [PATCH net-next v3 4/5] selftests: net: move common xdp.py functions into lib Chris J Arges
@ 2026-03-06 23:00 ` Chris J Arges
  2026-03-10  0:51 ` [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Joe Damato
  2026-03-10  2:37 ` Jakub Kicinski
  6 siblings, 0 replies; 16+ messages in thread
From: Chris J Arges @ 2026-03-06 23:00 UTC (permalink / raw)
  To: michael.chan, pavan.chebbi, joe, kuba, Andrew Lunn,
	David S. Miller, Eric Dumazet, Paolo Abeni, Shuah Khan,
	Simon Horman, Alexei Starovoitov, Daniel Borkmann,
	Jesper Dangaard Brouer, John Fastabend, Stanislav Fomichev
  Cc: kernel-team, Chris J Arges, linux-kernel, netdev, linux-kselftest,
	bpf

This test loads xdp_metadata.bpf which calls bpf_xdp_metadata_rx_hash() on
incoming packets. The metadata from that packet is then sent to a BPF
map for validation. It borrows structure from xdp.py, reusing common
functions.

The test checks the device's xdp-rx-metadata-features via netlink
before running and skips on devices that do not advertise hash support.
This can be run on veth devices as well as real hardware.

The test is fairly simple and just verifies that a TCP or UDP packet can be
identified as an L4 flow. This minimal test also passes if run on a veth
device.

Signed-off-by: Chris J Arges <carges@cloudflare.com>
---
 .../testing/selftests/drivers/net/hw/Makefile |   1 +
 .../drivers/net/hw/lib/py/__init__.py         |   2 +
 .../selftests/drivers/net/hw/xdp_metadata.py  | 146 ++++++++++++++++
 .../selftests/net/lib/xdp_metadata.bpf.c      | 163 ++++++++++++++++++
 4 files changed, 312 insertions(+)
 create mode 100644 tools/testing/selftests/drivers/net/hw/xdp_metadata.py
 create mode 100644 tools/testing/selftests/net/lib/xdp_metadata.bpf.c

diff --git a/tools/testing/selftests/drivers/net/hw/Makefile b/tools/testing/selftests/drivers/net/hw/Makefile
index a64140333a46..9db5b8a62286 100644
--- a/tools/testing/selftests/drivers/net/hw/Makefile
+++ b/tools/testing/selftests/drivers/net/hw/Makefile
@@ -40,6 +40,7 @@ TEST_PROGS = \
 	rss_input_xfrm.py \
 	toeplitz.py \
 	tso.py \
+	xdp_metadata.py \
 	xsk_reconfig.py \
 	#
 
diff --git a/tools/testing/selftests/drivers/net/hw/lib/py/__init__.py b/tools/testing/selftests/drivers/net/hw/lib/py/__init__.py
index 1971577d47e9..ac4a0a9cf0f2 100644
--- a/tools/testing/selftests/drivers/net/hw/lib/py/__init__.py
+++ b/tools/testing/selftests/drivers/net/hw/lib/py/__init__.py
@@ -24,6 +24,7 @@ try:
     from net.lib.py import bkg, cmd, bpftool, bpftrace, defer, ethtool, \
         fd_read_timeout, ip, rand_port, rand_ports, wait_port_listen, \
         wait_file, tool
+    from net.lib.py import bpf_map_set, bpf_map_dump, bpf_prog_map_ids
     from net.lib.py import KsftSkipEx, KsftFailEx, KsftXfailEx
     from net.lib.py import ksft_disruptive, ksft_exit, ksft_pr, ksft_run, \
         ksft_setup, ksft_variants, KsftNamedVariant
@@ -39,6 +40,7 @@ try:
                "bkg", "cmd", "bpftool", "bpftrace", "defer", "ethtool",
                "fd_read_timeout", "ip", "rand_port", "rand_ports",
                "wait_port_listen", "wait_file", "tool",
+               "bpf_map_set", "bpf_map_dump", "bpf_prog_map_ids",
                "KsftSkipEx", "KsftFailEx", "KsftXfailEx",
                "ksft_disruptive", "ksft_exit", "ksft_pr", "ksft_run",
                "ksft_setup", "ksft_variants", "KsftNamedVariant",
diff --git a/tools/testing/selftests/drivers/net/hw/xdp_metadata.py b/tools/testing/selftests/drivers/net/hw/xdp_metadata.py
new file mode 100644
index 000000000000..33a1985356d9
--- /dev/null
+++ b/tools/testing/selftests/drivers/net/hw/xdp_metadata.py
@@ -0,0 +1,146 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: GPL-2.0
+
+"""
+Tests for XDP metadata kfuncs (e.g. bpf_xdp_metadata_rx_hash).
+
+These tests load device-bound XDP programs from xdp_metadata.bpf.o
+that call metadata kfuncs, send traffic, and verify the extracted
+metadata via BPF maps.
+"""
+from lib.py import ksft_run, ksft_eq, ksft_exit, ksft_ge, ksft_ne, ksft_pr
+from lib.py import KsftNamedVariant, ksft_variants
+from lib.py import CmdExitFailure, KsftSkipEx, NetDrvEpEnv
+from lib.py import NetdevFamily
+from lib.py import bkg, cmd, rand_port, wait_port_listen
+from lib.py import ip, bpftool, defer
+from lib.py import bpf_map_set, bpf_map_dump, bpf_prog_map_ids
+
+
+def _load_xdp_metadata_prog(cfg, prog_name, bpf_file="xdp_metadata.bpf.o"):
+    """Load a device-bound XDP metadata program and return prog/map info.
+
+    Returns:
+        dict with 'id', 'name', and 'maps' (name -> map_id).
+    """
+    abs_path = cfg.net_lib_dir / bpf_file
+    pin_dir = "/sys/fs/bpf/xdp_metadata_test"
+
+    cmd(f"rm -rf {pin_dir}", shell=True, fail=False)
+    cmd(f"mkdir -p {pin_dir}", shell=True)
+
+    try:
+        bpftool(f"prog loadall {abs_path} {pin_dir} type xdp "
+                f"xdpmeta_dev {cfg.ifname}")
+    except CmdExitFailure as e:
+        cmd(f"rm -rf {pin_dir}", shell=True, fail=False)
+        raise KsftSkipEx(
+            f"Failed to load device-bound XDP program '{prog_name}'"
+        ) from e
+    defer(cmd, f"rm -rf {pin_dir}", shell=True, fail=False)
+
+    pin_path = f"{pin_dir}/{prog_name}"
+    ip(f"link set dev {cfg.ifname} xdpdrv pinned {pin_path}")
+    defer(ip, f"link set dev {cfg.ifname} xdpdrv off")
+
+    xdp_info = ip(f"-d link show dev {cfg.ifname}", json=True)[0]
+    prog_id = xdp_info["xdp"]["prog"]["id"]
+
+    return {"id": prog_id,
+            "name": xdp_info["xdp"]["prog"]["name"],
+            "maps": bpf_prog_map_ids(prog_id)}
+
+
+def _send_probe(cfg, port, proto="tcp"):
+    """Send a single payload from the remote end using socat.
+
+    Args:
+        cfg: Configuration object containing network settings.
+        port: Port number for the exchange.
+        proto: Protocol to use, either "tcp" or "udp".
+    """
+    cfg.require_cmd("socat", remote=True)
+
+    if proto == "tcp":
+        rx_cmd = f"socat -{cfg.addr_ipver} -T 2 TCP-LISTEN:{port},reuseport STDOUT"
+        tx_cmd = f"echo -n rss_hash_test | socat -t 2 -u STDIN TCP:{cfg.baddr}:{port}"
+    else:
+        rx_cmd = f"socat -{cfg.addr_ipver} -T 2 -u UDP-RECV:{port},reuseport STDOUT"
+        tx_cmd = f"echo -n rss_hash_test | socat -t 2 -u STDIN UDP:{cfg.baddr}:{port}"
+
+    with bkg(rx_cmd, exit_wait=True):
+        wait_port_listen(port, proto=proto)
+        cmd(tx_cmd, host=cfg.remote, shell=True)
+
+
+# BPF map keys matching the enums in xdp_metadata.bpf.c
+_SETUP_KEY_PORT = 1
+
+_RSS_KEY_HASH = 0
+_RSS_KEY_TYPE = 1
+_RSS_KEY_PKT_CNT = 2
+_RSS_KEY_ERR_CNT = 3
+
+XDP_RSS_L4 = 0x8  # BIT(3) from enum xdp_rss_hash_type
+
+
+@ksft_variants([
+    KsftNamedVariant("tcp", "tcp"),
+    KsftNamedVariant("udp", "udp"),
+])
+def test_xdp_rss_hash(cfg, proto):
+    """Test RSS hash metadata extraction via bpf_xdp_metadata_rx_hash().
+
+    This test will only run on devices that support xdp-rx-metadata-features.
+
+    Loads the xdp_rss_hash program from xdp_metadata, sends a packet using
+    the specified protocol, and verifies that the program extracted a non-zero
+    hash with an L4 hash type.
+    """
+    dev_info = cfg.netnl.dev_get({"ifindex": cfg.ifindex})
+    rx_meta = dev_info.get("xdp-rx-metadata-features", [])
+    if "hash" not in rx_meta:
+        raise KsftSkipEx("device does not support XDP rx hash metadata")
+
+    prog_info = _load_xdp_metadata_prog(cfg, "xdp_rss_hash")
+
+    port = rand_port()
+    bpf_map_set("map_xdp_setup", _SETUP_KEY_PORT, port)
+
+    rss_map_id = prog_info["maps"]["map_rss"]
+
+    _send_probe(cfg, port, proto=proto)
+
+    rss = bpf_map_dump(rss_map_id)
+
+    pkt_cnt = rss.get(_RSS_KEY_PKT_CNT, 0)
+    err_cnt = rss.get(_RSS_KEY_ERR_CNT, 0)
+    hash_val = rss.get(_RSS_KEY_HASH, 0)
+    hash_type = rss.get(_RSS_KEY_TYPE, 0)
+
+    ksft_ge(pkt_cnt, 1, comment="should have received at least one packet")
+    ksft_eq(err_cnt, 0, comment=f"RSS hash error count: {err_cnt}")
+
+    ksft_ne(hash_val, 0,
+            f"RSS hash should be non-zero for {proto.upper()} traffic")
+    ksft_pr(f"  RSS hash: {hash_val:#010x}")
+
+    ksft_pr(f"  RSS hash type: {hash_type:#06x}")
+    ksft_ne(hash_type & XDP_RSS_L4, 0,
+            f"RSS hash type should include L4 for {proto.upper()} traffic")
+
+
+def main():
+    """Run XDP metadata kfunc tests against a real device."""
+    with NetDrvEpEnv(__file__) as cfg:
+        cfg.netnl = NetdevFamily()
+        ksft_run(
+            [
+                test_xdp_rss_hash,
+            ],
+            args=(cfg,))
+    ksft_exit()
+
+
+if __name__ == "__main__":
+    main()
diff --git a/tools/testing/selftests/net/lib/xdp_metadata.bpf.c b/tools/testing/selftests/net/lib/xdp_metadata.bpf.c
new file mode 100644
index 000000000000..f71f59215239
--- /dev/null
+++ b/tools/testing/selftests/net/lib/xdp_metadata.bpf.c
@@ -0,0 +1,163 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <stddef.h>
+#include <linux/bpf.h>
+#include <linux/in.h>
+#include <linux/if_ether.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/udp.h>
+#include <linux/tcp.h>
+#include <bpf/bpf_endian.h>
+#include <bpf/bpf_helpers.h>
+
+enum {
+	XDP_PORT = 1,
+	XDP_PROTO = 4,
+} xdp_map_setup_keys;
+
+struct {
+	__uint(type, BPF_MAP_TYPE_ARRAY);
+	__uint(max_entries, 5);
+	__type(key, __u32);
+	__type(value, __s32);
+} map_xdp_setup SEC(".maps");
+
+/* RSS hash results: key 0 = hash, key 1 = hash type,
+ * key 2 = packet count, key 3 = error count.
+ */
+enum {
+	RSS_KEY_HASH = 0,
+	RSS_KEY_TYPE = 1,
+	RSS_KEY_PKT_CNT = 2,
+	RSS_KEY_ERR_CNT = 3,
+};
+
+struct {
+	__uint(type, BPF_MAP_TYPE_ARRAY);
+	__type(key, __u32);
+	__type(value, __u32);
+	__uint(max_entries, 4);
+} map_rss SEC(".maps");
+
+/* Mirror of enum xdp_rss_hash_type from include/net/xdp.h.
+ * Needed because the enum is not part of UAPI headers.
+ */
+enum xdp_rss_hash_type {
+	XDP_RSS_L3_IPV4 = 1U << 0,
+	XDP_RSS_L3_IPV6 = 1U << 1,
+	XDP_RSS_L3_DYNHDR = 1U << 2,
+	XDP_RSS_L4 = 1U << 3,
+	XDP_RSS_L4_TCP = 1U << 4,
+	XDP_RSS_L4_UDP = 1U << 5,
+	XDP_RSS_L4_SCTP = 1U << 6,
+	XDP_RSS_L4_IPSEC = 1U << 7,
+	XDP_RSS_L4_ICMP = 1U << 8,
+};
+
+extern int bpf_xdp_metadata_rx_hash(const struct xdp_md *ctx, __u32 *hash,
+				    enum xdp_rss_hash_type *rss_type) __ksym;
+
+static __always_inline __u16 get_dest_port(void *l4, void *data_end,
+					   __u8 protocol)
+{
+	if (protocol == IPPROTO_UDP) {
+		struct udphdr *udp = l4;
+
+		if ((void *)(udp + 1) > data_end)
+			return 0;
+		return udp->dest;
+	} else if (protocol == IPPROTO_TCP) {
+		struct tcphdr *tcp = l4;
+
+		if ((void *)(tcp + 1) > data_end)
+			return 0;
+		return tcp->dest;
+	}
+
+	return 0;
+}
+
+SEC("xdp")
+int xdp_rss_hash(struct xdp_md *ctx)
+{
+	void *data_end = (void *)(long)ctx->data_end;
+	void *data = (void *)(long)ctx->data;
+	enum xdp_rss_hash_type rss_type = 0;
+	struct ethhdr *eth = data;
+	__u8 l4_proto = 0;
+	__u32 hash = 0;
+	__u32 key, val;
+	void *l4 = NULL;
+	__u32 *cnt;
+	int ret;
+
+	if ((void *)(eth + 1) > data_end)
+		return XDP_PASS;
+
+	if (eth->h_proto == bpf_htons(ETH_P_IP)) {
+		struct iphdr *iph = (void *)(eth + 1);
+
+		if ((void *)(iph + 1) > data_end)
+			return XDP_PASS;
+		l4_proto = iph->protocol;
+		l4 = (void *)(iph + 1);
+	} else if (eth->h_proto == bpf_htons(ETH_P_IPV6)) {
+		struct ipv6hdr *ip6h = (void *)(eth + 1);
+
+		if ((void *)(ip6h + 1) > data_end)
+			return XDP_PASS;
+		l4_proto = ip6h->nexthdr;
+		l4 = (void *)(ip6h + 1);
+	}
+
+	if (!l4)
+		return XDP_PASS;
+
+	/* Filter on the configured protocol (map_xdp_setup key XDP_PROTO).
+	 * When set, only process packets matching the requested L4 protocol.
+	 */
+	key = XDP_PROTO;
+	__s32 *proto_cfg = bpf_map_lookup_elem(&map_xdp_setup, &key);
+
+	if (proto_cfg && *proto_cfg != 0 && l4_proto != (__u8)*proto_cfg)
+		return XDP_PASS;
+
+	/* Filter on the configured port (map_xdp_setup key XDP_PORT).
+	 * Only applies to protocols with ports (UDP, TCP).
+	 */
+	key = XDP_PORT;
+	__s32 *port_cfg = bpf_map_lookup_elem(&map_xdp_setup, &key);
+
+	if (port_cfg && *port_cfg != 0) {
+		__u16 dest = get_dest_port(l4, data_end, l4_proto);
+
+		if (!dest || bpf_ntohs(dest) != (__u16)*port_cfg)
+			return XDP_PASS;
+	}
+
+	ret = bpf_xdp_metadata_rx_hash(ctx, &hash, &rss_type);
+	if (ret < 0) {
+		key = RSS_KEY_ERR_CNT;
+		cnt = bpf_map_lookup_elem(&map_rss, &key);
+		if (cnt)
+			__sync_fetch_and_add(cnt, 1);
+		return XDP_PASS;
+	}
+
+	key = RSS_KEY_HASH;
+	bpf_map_update_elem(&map_rss, &key, &hash, BPF_ANY);
+
+	key = RSS_KEY_TYPE;
+	val = (__u32)rss_type;
+	bpf_map_update_elem(&map_rss, &key, &val, BPF_ANY);
+
+	key = RSS_KEY_PKT_CNT;
+	cnt = bpf_map_lookup_elem(&map_rss, &key);
+	if (cnt)
+		__sync_fetch_and_add(cnt, 1);
+
+	return XDP_PASS;
+}
+
+char _license[] SEC("license") = "GPL";
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support
  2026-03-06 23:00 [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Chris J Arges
                   ` (4 preceding siblings ...)
  2026-03-06 23:00 ` [PATCH net-next v3 5/5] selftests: drv-net: xdp: Add rss_hash metadata tests Chris J Arges
@ 2026-03-10  0:51 ` Joe Damato
  2026-03-10  2:06   ` Chris Arges
  2026-03-10  2:37 ` Jakub Kicinski
  6 siblings, 1 reply; 16+ messages in thread
From: Joe Damato @ 2026-03-10  0:51 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, kuba, Alexei Starovoitov,
	Daniel Borkmann, David S. Miller, Jesper Dangaard Brouer,
	John Fastabend, Stanislav Fomichev, kernel-team, netdev, bpf

On Fri, Mar 06, 2026 at 05:00:15PM -0600, Chris J Arges wrote:
> This series adds XDP RSS hash metadata extraction support for the bnxt_en
> driver and includes selftests to validate the functionality. I was able
> to test this on a BCM57414 NIC.
> 
> Note: I don't have hardware to test v3 completions at the moment.
> 
> Patches 1-3 implement the driver support:
>   - Patch 1 adds bnxt_xdp_buff to wrap xdp_buff with
>     pointers to the hardware RX completion descriptors and the
>     completion type. This is similar to other driver implementations.
>   - Patch 2 adds the xmo_rx_hash callback for CMP_TYPE_RX_L2_CMP completions.
>   - Patch 3 adds handling for CMP_TYPE_RX_L2_V3_CMP completions
> 
> Patches 4-5 add selftest coverage:
>   - Patch 4 refactors common BPF map helper functions out of the
>     existing xdp.py selftest into a shared net/lib/py/bpf.py module.
>   - Patch 5 adds an XDP metadata selftest that loads an XDP program
>     calling bpf_xdp_metadata_rx_hash. The test skips on devices without
>     xdp-rx-metadata-features.
> 
> v2: https://lore.kernel.org/netdev/20260303024510.644962-1-carges@cloudflare.com/
>   - bnxt xdp: actually fixed RCT issues
>   - bnxt xdp: Added patch 3 to handle CMP_TYPE_RX_L2_V3_CMP completions.
> 
> v1: https://lore.kernel.org/netdev/cover.1772211065.git.carges@cloudflare.com/
>   - selftests: moved tests into net/hw
>   - selftests: fix pylinter errors
>   - selftests: add check for no error count
>   - bnxt_xdp: followed RCT for variable declaration
>   - rebased on net-next

I think I commented this on the v2 [1], but it would be super helpful to put
the summary of changes in each patch itself to make it easier for reviewers.

I think you dropped my Reviewed-by's, so I'm not sure which patches changed
and which didn't :(

[1]: https://lore.kernel.org/netdev/aacsQvvOoL29Rvyw@devvm20253.cco0.facebook.com/

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context
  2026-03-06 23:00 ` [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context Chris J Arges
@ 2026-03-10  0:52   ` Joe Damato
  2026-03-13  0:14   ` Andy Gospodarek
  1 sibling, 0 replies; 16+ messages in thread
From: Joe Damato @ 2026-03-10  0:52 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, kuba, Andrew Lunn, David S. Miller,
	Eric Dumazet, Paolo Abeni, Alexei Starovoitov, Daniel Borkmann,
	Jesper Dangaard Brouer, John Fastabend, Stanislav Fomichev,
	kernel-team, netdev, linux-kernel, bpf

On Fri, Mar 06, 2026 at 05:00:16PM -0600, Chris J Arges wrote:
> This adds bnxt_xdp_buff which embeds the xdp_buff struct and stores
> pointers to hardware RX completion descriptors (rx_cmp and rx_cmp_ext)
> along with the completion type.
> 
> Signed-off-by: Chris J Arges <carges@cloudflare.com>
> ---
>  drivers/net/ethernet/broadcom/bnxt/bnxt.c     | 25 +++++++++++--------
>  drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h |  7 ++++++
>  2 files changed, 22 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> index c982aac714d1..ab73aad40593 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> @@ -2118,7 +2118,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
>  	bool xdp_active = false;
>  	dma_addr_t dma_addr;
>  	struct sk_buff *skb;
> -	struct xdp_buff xdp;
> +	struct bnxt_xdp_buff bnxt_xdp;

I think I mentioned in the v2 [1], not sure RCT matters here.

But in either case:

Reviewed-by: Joe Damato <joe@dama.to>

[1]: https://lore.kernel.org/netdev/aacrSocsBAV9wZsT@devvm20253.cco0.facebook.com/

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 4/5] selftests: net: move common xdp.py functions into lib
  2026-03-06 23:00 ` [PATCH net-next v3 4/5] selftests: net: move common xdp.py functions into lib Chris J Arges
@ 2026-03-10  0:54   ` Joe Damato
  0 siblings, 0 replies; 16+ messages in thread
From: Joe Damato @ 2026-03-10  0:54 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, kuba, Andrew Lunn, David S. Miller,
	Eric Dumazet, Paolo Abeni, Shuah Khan, Simon Horman,
	Alexei Starovoitov, Daniel Borkmann, Jesper Dangaard Brouer,
	John Fastabend, Stanislav Fomichev, kernel-team, linux-kernel,
	netdev, linux-kselftest, bpf

On Fri, Mar 06, 2026 at 05:00:19PM -0600, Chris J Arges wrote:
> This moves a few functions which can be useful to other python programs
> that manipulate XDP programs. This also refactors xdp.py to use the
> refactored functions.
> 
> Signed-off-by: Chris J Arges <carges@cloudflare.com>
> ---
>  .../selftests/drivers/net/lib/py/__init__.py  |  2 +
>  tools/testing/selftests/drivers/net/xdp.py    | 95 +++++--------------
>  .../testing/selftests/net/lib/py/__init__.py  |  2 +
>  tools/testing/selftests/net/lib/py/bpf.py     | 68 +++++++++++++
>  4 files changed, 95 insertions(+), 72 deletions(-)
>  create mode 100644 tools/testing/selftests/net/lib/py/bpf.py

Looks unchanged from the v2?

Reviewed-by: Joe Damato <joe@dama.to>

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 2/5] bnxt_en: Implement XDP RSS hash metadata extraction
  2026-03-06 23:00 ` [PATCH net-next v3 2/5] bnxt_en: Implement XDP RSS hash metadata extraction Chris J Arges
@ 2026-03-10  0:56   ` Joe Damato
  2026-03-13  0:15   ` Andy Gospodarek
  1 sibling, 0 replies; 16+ messages in thread
From: Joe Damato @ 2026-03-10  0:56 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, kuba, Andrew Lunn, David S. Miller,
	Eric Dumazet, Paolo Abeni, Alexei Starovoitov, Daniel Borkmann,
	Jesper Dangaard Brouer, John Fastabend, Stanislav Fomichev,
	kernel-team, netdev, linux-kernel, bpf

On Fri, Mar 06, 2026 at 05:00:17PM -0600, Chris J Arges wrote:
> Add support for extracting RSS hash values and hash types from hardware
> completion descriptors in XDP programs for bnxt_en.
> 
> Add IP_TYPE definition for determining if completion is ipv4 or ipv6. In
> addition add ITYPE_ICMP flag for identifying ICMP completions.
> 
> Signed-off-by: Chris J Arges <carges@cloudflare.com>
> ---
>  drivers/net/ethernet/broadcom/bnxt/bnxt.c     |  5 ++
>  drivers/net/ethernet/broadcom/bnxt/bnxt.h     |  2 +
>  drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 47 +++++++++++++++++++
>  drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h |  3 ++
>  4 files changed, 57 insertions(+)

Looks unchanged from the v2, so:

Reviewed-by: Joe Damato <joe@dama.to>

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 3/5] bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP
  2026-03-06 23:00 ` [PATCH net-next v3 3/5] bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP Chris J Arges
@ 2026-03-10  1:02   ` Joe Damato
  2026-03-13  0:15   ` Andy Gospodarek
  1 sibling, 0 replies; 16+ messages in thread
From: Joe Damato @ 2026-03-10  1:02 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, kuba, Alexei Starovoitov,
	Daniel Borkmann, David S. Miller, Jesper Dangaard Brouer,
	John Fastabend, Stanislav Fomichev, Andrew Lunn, Eric Dumazet,
	Paolo Abeni, kernel-team, netdev, bpf, linux-kernel

On Fri, Mar 06, 2026 at 05:00:18PM -0600, Chris J Arges wrote:
> This adds another conditional for if cmp_type is CMP_TYPE_RX_L2_V3_CMP for
> drivers that support this completion format.
> 
> It uses a similar approach as found in other parts of the driver.
> 
> Signed-off-by: Chris J Arges <carges@cloudflare.com>
> ---
>  drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 20 +++++++++++++++++++
>  1 file changed, 20 insertions(+)
> 
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> index 1f920464b426..25ecef1eb96e 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> @@ -514,6 +514,26 @@ int bnxt_xdp_rx_hash(const struct xdp_md *ctx, u32 *hash,
>  		default:
>  			break;
>  		}
> +	} else if (xdp->cmp_type == CMP_TYPE_RX_L2_V3_CMP) {
> +		struct bnxt *bp = netdev_priv(xdp->xdp.rxq->dev);
> +		u8 ext_op;
> +
> +		if (rxcmp1->rx_cmp_flags2 & cpu_to_le32(RX_CMP_FLAGS2_IP_TYPE))
> +			hash_type |= XDP_RSS_TYPE_L3_IPV6;
> +		else
> +			hash_type |= XDP_RSS_TYPE_L3_IPV4;
> +
> +		ext_op = RX_CMP_V3_HASH_TYPE(bp, rxcmp);
> +		switch (ext_op) {
> +		case EXT_OP_INNER_4:
> +		case EXT_OP_OUTER_4:
> +		case EXT_OP_INNFL_3:
> +		case EXT_OP_OUTFL_3:
> +			hash_type |= XDP_RSS_L4;
> +			break;
> +		default:
> +			break;
> +		}

Looks like the above code is duplicated from bnxt_rss_ext_op?

Maybe that could be refactored into a helper in bnxt.h and called from here to
reduce code duplciation?

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support
  2026-03-10  0:51 ` [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Joe Damato
@ 2026-03-10  2:06   ` Chris Arges
  0 siblings, 0 replies; 16+ messages in thread
From: Chris Arges @ 2026-03-10  2:06 UTC (permalink / raw)
  To: Joe Damato, michael.chan, pavan.chebbi, kuba, Alexei Starovoitov,
	Daniel Borkmann, David S. Miller, Jesper Dangaard Brouer,
	John Fastabend, Stanislav Fomichev, kernel-team, netdev, bpf

On 2026-03-09 17:51:38, Joe Damato wrote:
> On Fri, Mar 06, 2026 at 05:00:15PM -0600, Chris J Arges wrote:
> > This series adds XDP RSS hash metadata extraction support for the bnxt_en
> > driver and includes selftests to validate the functionality. I was able
> > to test this on a BCM57414 NIC.
> > 
> > Note: I don't have hardware to test v3 completions at the moment.
> > 
> > Patches 1-3 implement the driver support:
> >   - Patch 1 adds bnxt_xdp_buff to wrap xdp_buff with
> >     pointers to the hardware RX completion descriptors and the
> >     completion type. This is similar to other driver implementations.
> >   - Patch 2 adds the xmo_rx_hash callback for CMP_TYPE_RX_L2_CMP completions.
> >   - Patch 3 adds handling for CMP_TYPE_RX_L2_V3_CMP completions
> > 
> > Patches 4-5 add selftest coverage:
> >   - Patch 4 refactors common BPF map helper functions out of the
> >     existing xdp.py selftest into a shared net/lib/py/bpf.py module.
> >   - Patch 5 adds an XDP metadata selftest that loads an XDP program
> >     calling bpf_xdp_metadata_rx_hash. The test skips on devices without
> >     xdp-rx-metadata-features.
> > 
> > v2: https://lore.kernel.org/netdev/20260303024510.644962-1-carges@cloudflare.com/
> >   - bnxt xdp: actually fixed RCT issues
> >   - bnxt xdp: Added patch 3 to handle CMP_TYPE_RX_L2_V3_CMP completions.
> > 
> > v1: https://lore.kernel.org/netdev/cover.1772211065.git.carges@cloudflare.com/
> >   - selftests: moved tests into net/hw
> >   - selftests: fix pylinter errors
> >   - selftests: add check for no error count
> >   - bnxt_xdp: followed RCT for variable declaration
> >   - rebased on net-next
> 
> I think I commented this on the v2 [1], but it would be super helpful to put
> the summary of changes in each patch itself to make it easier for reviewers.
> 
> I think you dropped my Reviewed-by's, so I'm not sure which patches changed
> and which didn't :(
> 
> [1]: https://lore.kernel.org/netdev/aacsQvvOoL29Rvyw@devvm20253.cco0.facebook.com/

Thanks for the reviews and feedback, I'm learning. So when re-submitting, I'll
be sure to keep your reviewed by if nothing has changed. In addition I'll make
sure I put the changelog in each patch rather than just the cover letter.

--chris

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support
  2026-03-06 23:00 [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Chris J Arges
                   ` (5 preceding siblings ...)
  2026-03-10  0:51 ` [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Joe Damato
@ 2026-03-10  2:37 ` Jakub Kicinski
  6 siblings, 0 replies; 16+ messages in thread
From: Jakub Kicinski @ 2026-03-10  2:37 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, joe, Alexei Starovoitov,
	Daniel Borkmann, David S. Miller, Jesper Dangaard Brouer,
	John Fastabend, Stanislav Fomichev, kernel-team, netdev, bpf

On Fri,  6 Mar 2026 17:00:15 -0600 Chris J Arges wrote:
> This series adds XDP RSS hash metadata extraction support for the bnxt_en
> driver and includes selftests to validate the functionality. I was able
> to test this on a BCM57414 NIC.

Looks like we hit a conflict, again, you'll have to respin.
-- 
pw-bot: cr

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context
  2026-03-06 23:00 ` [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context Chris J Arges
  2026-03-10  0:52   ` Joe Damato
@ 2026-03-13  0:14   ` Andy Gospodarek
  1 sibling, 0 replies; 16+ messages in thread
From: Andy Gospodarek @ 2026-03-13  0:14 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, joe, kuba, Andrew Lunn,
	David S. Miller, Eric Dumazet, Paolo Abeni, Alexei Starovoitov,
	Daniel Borkmann, Jesper Dangaard Brouer, John Fastabend,
	Stanislav Fomichev, kernel-team, netdev, linux-kernel, bpf

On Fri, Mar 06, 2026 at 05:00:16PM -0600, Chris J Arges wrote:
> This adds bnxt_xdp_buff which embeds the xdp_buff struct and stores
> pointers to hardware RX completion descriptors (rx_cmp and rx_cmp_ext)
> along with the completion type.
> 
> Signed-off-by: Chris J Arges <carges@cloudflare.com>
> Reviewed-by: Joe Damato <joe@dama.to>
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>

> ---
>  drivers/net/ethernet/broadcom/bnxt/bnxt.c     | 25 +++++++++++--------
>  drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h |  7 ++++++
>  2 files changed, 22 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> index c982aac714d1..ab73aad40593 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> @@ -2118,7 +2118,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
>  	bool xdp_active = false;
>  	dma_addr_t dma_addr;
>  	struct sk_buff *skb;
> -	struct xdp_buff xdp;
> +	struct bnxt_xdp_buff bnxt_xdp;
>  	u32 flags, misc;
>  	u32 cmpl_ts;
>  	void *data;
> @@ -2231,9 +2231,13 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
>  	dma_addr = rx_buf->mapping;
>  
>  	if (bnxt_xdp_attached(bp, rxr)) {
> -		bnxt_xdp_buff_init(bp, rxr, cons, data_ptr, len, &xdp);
> +		bnxt_xdp.rxcmp = rxcmp;
> +		bnxt_xdp.rxcmp1 = rxcmp1;
> +		bnxt_xdp.cmp_type = cmp_type;
> +
> +		bnxt_xdp_buff_init(bp, rxr, cons, data_ptr, len, &bnxt_xdp.xdp);
>  		if (agg_bufs) {
> -			u32 frag_len = bnxt_rx_agg_netmems_xdp(bp, cpr, &xdp,
> +			u32 frag_len = bnxt_rx_agg_netmems_xdp(bp, cpr, &bnxt_xdp.xdp,
>  							       cp_cons,
>  							       agg_bufs,
>  							       false);
> @@ -2245,12 +2249,12 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
>  	}
>  
>  	if (xdp_active) {
> -		if (bnxt_rx_xdp(bp, rxr, cons, &xdp, data, &data_ptr, &len, event)) {
> +		if (bnxt_rx_xdp(bp, rxr, cons, &bnxt_xdp.xdp, data, &data_ptr, &len, event)) {
>  			rc = 1;
>  			goto next_rx;
>  		}
> -		if (xdp_buff_has_frags(&xdp)) {
> -			sinfo = xdp_get_shared_info_from_buff(&xdp);
> +		if (xdp_buff_has_frags(&bnxt_xdp.xdp)) {
> +			sinfo = xdp_get_shared_info_from_buff(&bnxt_xdp.xdp);
>  			agg_bufs = sinfo->nr_frags;
>  		} else {
>  			agg_bufs = 0;
> @@ -2261,7 +2265,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
>  		if (!xdp_active)
>  			skb = bnxt_copy_skb(bnapi, data_ptr, len, dma_addr);
>  		else
> -			skb = bnxt_copy_xdp(bnapi, &xdp, len, dma_addr);
> +			skb = bnxt_copy_xdp(bnapi, &bnxt_xdp.xdp, len, dma_addr);
>  		bnxt_reuse_rx_data(rxr, cons, data);
>  		if (!skb) {
>  			if (agg_bufs) {
> @@ -2269,7 +2273,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
>  					bnxt_reuse_rx_agg_bufs(cpr, cp_cons, 0,
>  							       agg_bufs, false);
>  				else
> -					bnxt_xdp_buff_frags_free(rxr, &xdp);
> +					bnxt_xdp_buff_frags_free(rxr, &bnxt_xdp.xdp);
>  			}
>  			goto oom_next_rx;
>  		}
> @@ -2293,10 +2297,11 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
>  			if (!skb)
>  				goto oom_next_rx;
>  		} else {
> -			skb = bnxt_xdp_build_skb(bp, skb, agg_bufs, rxr, &xdp);
> +			skb = bnxt_xdp_build_skb(bp, skb, agg_bufs,
> +						 rxr, &bnxt_xdp.xdp);
>  			if (!skb) {
>  				/* we should be able to free the old skb here */
> -				bnxt_xdp_buff_frags_free(rxr, &xdp);
> +				bnxt_xdp_buff_frags_free(rxr, &bnxt_xdp.xdp);
>  				goto oom_next_rx;
>  			}
>  		}
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
> index 8933a0dec09a..8c66698bde11 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
> @@ -12,6 +12,13 @@
>  
>  DECLARE_STATIC_KEY_FALSE(bnxt_xdp_locking_key);
>  
> +struct bnxt_xdp_buff {
> +	struct xdp_buff xdp;
> +	struct rx_cmp *rxcmp;
> +	struct rx_cmp_ext *rxcmp1;
> +	u8 cmp_type;
> +};
> +
>  struct bnxt_sw_tx_bd *bnxt_xmit_bd(struct bnxt *bp,
>  				   struct bnxt_tx_ring_info *txr,
>  				   dma_addr_t mapping, u32 len,

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 2/5] bnxt_en: Implement XDP RSS hash metadata extraction
  2026-03-06 23:00 ` [PATCH net-next v3 2/5] bnxt_en: Implement XDP RSS hash metadata extraction Chris J Arges
  2026-03-10  0:56   ` Joe Damato
@ 2026-03-13  0:15   ` Andy Gospodarek
  1 sibling, 0 replies; 16+ messages in thread
From: Andy Gospodarek @ 2026-03-13  0:15 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, joe, kuba, Andrew Lunn,
	David S. Miller, Eric Dumazet, Paolo Abeni, Alexei Starovoitov,
	Daniel Borkmann, Jesper Dangaard Brouer, John Fastabend,
	Stanislav Fomichev, kernel-team, netdev, linux-kernel, bpf

On Fri, Mar 06, 2026 at 05:00:17PM -0600, Chris J Arges wrote:
> Add support for extracting RSS hash values and hash types from hardware
> completion descriptors in XDP programs for bnxt_en.
> 
> Add IP_TYPE definition for determining if completion is ipv4 or ipv6. In
> addition add ITYPE_ICMP flag for identifying ICMP completions.
> 
> Signed-off-by: Chris J Arges <carges@cloudflare.com>
> Reviewed-by: Joe Damato <joe@dama.to>
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>

> ---
>  drivers/net/ethernet/broadcom/bnxt/bnxt.c     |  5 ++
>  drivers/net/ethernet/broadcom/bnxt/bnxt.h     |  2 +
>  drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 47 +++++++++++++++++++
>  drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h |  3 ++
>  4 files changed, 57 insertions(+)
> 
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> index ab73aad40593..c06033cf9f82 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> @@ -15906,6 +15906,10 @@ static const struct net_device_ops bnxt_netdev_ops = {
>  	.ndo_hwtstamp_set	= bnxt_hwtstamp_set,
>  };
>  
> +static const struct xdp_metadata_ops bnxt_xdp_metadata_ops = {
> +	.xmo_rx_hash		= bnxt_xdp_rx_hash,
> +};
> +
>  static void bnxt_get_queue_stats_rx(struct net_device *dev, int i,
>  				    struct netdev_queue_stats_rx *stats)
>  {
> @@ -16790,6 +16794,7 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
>  		goto init_err_free;
>  
>  	dev->netdev_ops = &bnxt_netdev_ops;
> +	dev->xdp_metadata_ops = &bnxt_xdp_metadata_ops;
>  	dev->stat_ops = &bnxt_stat_ops;
>  	dev->watchdog_timeo = BNXT_TX_TIMEOUT;
>  	dev->ethtool_ops = &bnxt_ethtool_ops;
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.h b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
> index 90fa3e93c8d6..34bb98709f2b 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt.h
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
> @@ -232,6 +232,7 @@ struct rx_cmp {
>  	 #define RX_CMP_FLAGS_ITYPE_UDP				 (3 << 12)
>  	 #define RX_CMP_FLAGS_ITYPE_FCOE			 (4 << 12)
>  	 #define RX_CMP_FLAGS_ITYPE_ROCE			 (5 << 12)
> +	 #define RX_CMP_FLAGS_ITYPE_ICMP			 (7 << 12)
>  	 #define RX_CMP_FLAGS_ITYPE_PTP_WO_TS			 (8 << 12)
>  	 #define RX_CMP_FLAGS_ITYPE_PTP_W_TS			 (9 << 12)
>  	#define RX_CMP_LEN					(0xffff << 16)
> @@ -311,6 +312,7 @@ struct rx_cmp_ext {
>  	#define RX_CMP_FLAGS2_T_IP_CS_CALC			(0x1 << 2)
>  	#define RX_CMP_FLAGS2_T_L4_CS_CALC			(0x1 << 3)
>  	#define RX_CMP_FLAGS2_META_FORMAT_VLAN			(0x1 << 4)
> +	#define RX_CMP_FLAGS2_IP_TYPE				(0x1 << 8)
>  	__le32 rx_cmp_meta_data;
>  	#define RX_CMP_FLAGS2_METADATA_TCI_MASK			0xffff
>  	#define RX_CMP_FLAGS2_METADATA_VID_MASK			0xfff
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> index 85cbeb35681c..1f920464b426 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> @@ -472,3 +472,50 @@ bnxt_xdp_build_skb(struct bnxt *bp, struct sk_buff *skb, u8 num_frags,
>  				  xdp_buff_get_skb_flags(xdp));
>  	return skb;
>  }
> +
> +int bnxt_xdp_rx_hash(const struct xdp_md *ctx, u32 *hash,
> +		     enum xdp_rss_hash_type *rss_type)
> +{
> +	const struct bnxt_xdp_buff *xdp = (void *)ctx;
> +	const struct rx_cmp_ext *rxcmp1 = xdp->rxcmp1;
> +	const struct rx_cmp *rxcmp = xdp->rxcmp;
> +	enum xdp_rss_hash_type hash_type = 0;
> +	u32 itypes;
> +
> +	if (!rxcmp || !RX_CMP_HASH_VALID(rxcmp))
> +		return -ENODATA;
> +
> +	*hash = le32_to_cpu(rxcmp->rx_cmp_rss_hash);
> +
> +	if (!rxcmp1) {
> +		*rss_type = XDP_RSS_TYPE_L2;
> +		return 0;
> +	}
> +
> +	if (xdp->cmp_type == CMP_TYPE_RX_L2_CMP) {
> +		itypes = RX_CMP_ITYPES(rxcmp);
> +		if (rxcmp1->rx_cmp_flags2 &
> +		    cpu_to_le32(RX_CMP_FLAGS2_IP_TYPE)) {
> +			hash_type |= XDP_RSS_TYPE_L3_IPV6;
> +		} else {
> +			hash_type |= XDP_RSS_TYPE_L3_IPV4;
> +		}
> +
> +		switch (itypes) {
> +		case RX_CMP_FLAGS_ITYPE_TCP:
> +			hash_type |= XDP_RSS_L4 | XDP_RSS_L4_TCP;
> +			break;
> +		case RX_CMP_FLAGS_ITYPE_UDP:
> +			hash_type |= XDP_RSS_L4 | XDP_RSS_L4_UDP;
> +			break;
> +		case RX_CMP_FLAGS_ITYPE_ICMP:
> +			hash_type |= XDP_RSS_L4 | XDP_RSS_L4_ICMP;
> +			break;
> +		default:
> +			break;
> +		}
> +	}
> +
> +	*rss_type = hash_type;
> +	return 0;
> +}
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
> index 8c66698bde11..fb4f9143929f 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.h
> @@ -41,4 +41,7 @@ void bnxt_xdp_buff_frags_free(struct bnxt_rx_ring_info *rxr,
>  struct sk_buff *bnxt_xdp_build_skb(struct bnxt *bp, struct sk_buff *skb,
>  				   u8 num_frags, struct bnxt_rx_ring_info *rxr,
>  				   struct xdp_buff *xdp);
> +int bnxt_xdp_rx_hash(const struct xdp_md *ctx, u32 *hash,
> +		     enum xdp_rss_hash_type *rss_type);
> +
>  #endif

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH net-next v3 3/5] bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP
  2026-03-06 23:00 ` [PATCH net-next v3 3/5] bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP Chris J Arges
  2026-03-10  1:02   ` Joe Damato
@ 2026-03-13  0:15   ` Andy Gospodarek
  1 sibling, 0 replies; 16+ messages in thread
From: Andy Gospodarek @ 2026-03-13  0:15 UTC (permalink / raw)
  To: Chris J Arges
  Cc: michael.chan, pavan.chebbi, joe, kuba, Alexei Starovoitov,
	Daniel Borkmann, David S. Miller, Jesper Dangaard Brouer,
	John Fastabend, Stanislav Fomichev, Andrew Lunn, Eric Dumazet,
	Paolo Abeni, kernel-team, netdev, bpf, linux-kernel

On Fri, Mar 06, 2026 at 05:00:18PM -0600, Chris J Arges wrote:
> This adds another conditional for if cmp_type is CMP_TYPE_RX_L2_V3_CMP for
> drivers that support this completion format.
> 
> It uses a similar approach as found in other parts of the driver.
> 
> Signed-off-by: Chris J Arges <carges@cloudflare.com>
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>

> ---
>  drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c | 20 +++++++++++++++++++
>  1 file changed, 20 insertions(+)
> 
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> index 1f920464b426..25ecef1eb96e 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_xdp.c
> @@ -514,6 +514,26 @@ int bnxt_xdp_rx_hash(const struct xdp_md *ctx, u32 *hash,
>  		default:
>  			break;
>  		}
> +	} else if (xdp->cmp_type == CMP_TYPE_RX_L2_V3_CMP) {
> +		struct bnxt *bp = netdev_priv(xdp->xdp.rxq->dev);
> +		u8 ext_op;
> +
> +		if (rxcmp1->rx_cmp_flags2 & cpu_to_le32(RX_CMP_FLAGS2_IP_TYPE))
> +			hash_type |= XDP_RSS_TYPE_L3_IPV6;
> +		else
> +			hash_type |= XDP_RSS_TYPE_L3_IPV4;
> +
> +		ext_op = RX_CMP_V3_HASH_TYPE(bp, rxcmp);
> +		switch (ext_op) {
> +		case EXT_OP_INNER_4:
> +		case EXT_OP_OUTER_4:
> +		case EXT_OP_INNFL_3:
> +		case EXT_OP_OUTFL_3:
> +			hash_type |= XDP_RSS_L4;
> +			break;
> +		default:
> +			break;
> +		}
>  	}
>  
>  	*rss_type = hash_type;

^ permalink raw reply	[flat|nested] 16+ messages in thread

end of thread, other threads:[~2026-03-13  0:15 UTC | newest]

Thread overview: 16+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-03-06 23:00 [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Chris J Arges
2026-03-06 23:00 ` [PATCH net-next v3 1/5] bnxt_en: use bnxt_xdp_buff for xdp context Chris J Arges
2026-03-10  0:52   ` Joe Damato
2026-03-13  0:14   ` Andy Gospodarek
2026-03-06 23:00 ` [PATCH net-next v3 2/5] bnxt_en: Implement XDP RSS hash metadata extraction Chris J Arges
2026-03-10  0:56   ` Joe Damato
2026-03-13  0:15   ` Andy Gospodarek
2026-03-06 23:00 ` [PATCH net-next v3 3/5] bnxt_en: Implement XDP RSS hash metadata extraction for V3_CMP Chris J Arges
2026-03-10  1:02   ` Joe Damato
2026-03-13  0:15   ` Andy Gospodarek
2026-03-06 23:00 ` [PATCH net-next v3 4/5] selftests: net: move common xdp.py functions into lib Chris J Arges
2026-03-10  0:54   ` Joe Damato
2026-03-06 23:00 ` [PATCH net-next v3 5/5] selftests: drv-net: xdp: Add rss_hash metadata tests Chris J Arges
2026-03-10  0:51 ` [PATCH net-next v3 0/5] bnxt_en: Add XDP RSS hash metadata support Joe Damato
2026-03-10  2:06   ` Chris Arges
2026-03-10  2:37 ` Jakub Kicinski

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox