Merge branch 'bpf: expose information about netdev xdp-metadata kfunc support'

Stanislav Fomichev says:

====================
Extend netdev netlink family to expose the bitmask with the
kfuncs that the device implements. The source of truth is the
device's xdp_metadata_ops. There is some amount of auto-generated
netlink boilerplate; the change itself is super minimal.

v2:
- add netdev->xdp_metadata_ops NULL check when dumping to netlink (Martin)

Cc: netdev@vger.kernel.org
Cc: Willem de Bruijn <willemb@google.com>
====================
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>

Documentation/netlink/specs/netdev.yaml

@@ -42,6 +42,19 @@ definitions:
         doc:
           This feature informs if netdev implements non-linear XDP buffer
           support in ndo_xdp_xmit callback.
+  -
+    type: flags
+    name: xdp-rx-metadata
+    render-max: true
+    entries:
+      -
+        name: timestamp
+        doc:
+          Device is capable of exposing receive HW timestamp via bpf_xdp_metadata_rx_timestamp().
+      -
+        name: hash
+        doc:
+          Device is capable of exposing receive packet hash via bpf_xdp_metadata_rx_hash().
 
 attribute-sets:
   -
@@ -68,6 +81,13 @@ attribute-sets:
         type: u32
         checks:
           min: 1
+      -
+        name: xdp-rx-metadata-features
+        doc: Bitmask of supported XDP receive metadata features.
+             See Documentation/networking/xdp-rx-metadata.rst for more details.
+        type: u64
+        enum: xdp-rx-metadata
+        enum-as-flags: true
 
 operations:
   list:
@@ -84,6 +104,7 @@ operations:
             - ifindex
             - xdp-features
             - xdp-zc-max-segs
+            - xdp-rx-metadata-features
       dump:
         reply: *dev-all
     -

Documentation/networking/xdp-rx-metadata.rst

@@ -105,6 +105,13 @@ bpf_tail_call
 Adding programs that access metadata kfuncs to the ``BPF_MAP_TYPE_PROG_ARRAY``
 is currently not supported.
 
+Supported Devices
+=================
+
+It is possible to query which kfunc the particular netdev implements via
+netlink. See ``xdp-rx-metadata-features`` attribute set in
+``Documentation/netlink/specs/netdev.yaml``.
+
 Example
 =======
 

include/net/xdp.h

@@ -383,14 +383,25 @@ void xdp_attachment_setup(struct xdp_attachment_info *info,
 
 #define DEV_MAP_BULK_SIZE XDP_BULK_QUEUE_SIZE
 
+/* Define the relationship between xdp-rx-metadata kfunc and
+ * various other entities:
+ * - xdp_rx_metadata enum
+ * - netdev netlink enum (Documentation/netlink/specs/netdev.yaml)
+ * - kfunc name
+ * - xdp_metadata_ops field
+ */
 #define XDP_METADATA_KFUNC_xxx	\
 	XDP_METADATA_KFUNC(XDP_METADATA_KFUNC_RX_TIMESTAMP, \
-			   bpf_xdp_metadata_rx_timestamp) \
+			   NETDEV_XDP_RX_METADATA_TIMESTAMP, \
+			   bpf_xdp_metadata_rx_timestamp, \
+			   xmo_rx_timestamp) \
 	XDP_METADATA_KFUNC(XDP_METADATA_KFUNC_RX_HASH, \
-			   bpf_xdp_metadata_rx_hash) \
+			   NETDEV_XDP_RX_METADATA_HASH, \
+			   bpf_xdp_metadata_rx_hash, \
+			   xmo_rx_hash) \
 
-enum {
-#define XDP_METADATA_KFUNC(name, _) name,
+enum xdp_rx_metadata {
+#define XDP_METADATA_KFUNC(name, _, __, ___) name,
 XDP_METADATA_KFUNC_xxx
 #undef XDP_METADATA_KFUNC
 MAX_XDP_METADATA_KFUNC,

include/uapi/linux/netdev.h

@@ -38,11 +38,27 @@ enum netdev_xdp_act {
 	NETDEV_XDP_ACT_MASK = 127,
 };
 
+/**
+ * enum netdev_xdp_rx_metadata
+ * @NETDEV_XDP_RX_METADATA_TIMESTAMP: Device is capable of exposing receive HW
+ *   timestamp via bpf_xdp_metadata_rx_timestamp().
+ * @NETDEV_XDP_RX_METADATA_HASH: Device is capable of exposing receive packet
+ *   hash via bpf_xdp_metadata_rx_hash().
+ */
+enum netdev_xdp_rx_metadata {
+	NETDEV_XDP_RX_METADATA_TIMESTAMP = 1,
+	NETDEV_XDP_RX_METADATA_HASH = 2,
+
+	/* private: */
+	NETDEV_XDP_RX_METADATA_MASK = 3,
+};
+
 enum {
 	NETDEV_A_DEV_IFINDEX = 1,
 	NETDEV_A_DEV_PAD,
 	NETDEV_A_DEV_XDP_FEATURES,
 	NETDEV_A_DEV_XDP_ZC_MAX_SEGS,
+	NETDEV_A_DEV_XDP_RX_METADATA_FEATURES,
 
 	__NETDEV_A_DEV_MAX,
 	NETDEV_A_DEV_MAX = (__NETDEV_A_DEV_MAX - 1)

kernel/bpf/offload.c

@@ -845,10 +845,11 @@ void *bpf_dev_bound_resolve_kfunc(struct bpf_prog *prog, u32 func_id)
 	if (!ops)
 		goto out;
 
-	if (func_id == bpf_xdp_metadata_kfunc_id(XDP_METADATA_KFUNC_RX_TIMESTAMP))
-		p = ops->xmo_rx_timestamp;
-	else if (func_id == bpf_xdp_metadata_kfunc_id(XDP_METADATA_KFUNC_RX_HASH))
-		p = ops->xmo_rx_hash;
+#define XDP_METADATA_KFUNC(name, _, __, xmo) \
+	if (func_id == bpf_xdp_metadata_kfunc_id(name)) p = ops->xmo;
+	XDP_METADATA_KFUNC_xxx
+#undef XDP_METADATA_KFUNC
+
 out:
 	up_read(&bpf_devs_lock);
 

net/core/netdev-genl.c

@@ -5,6 +5,7 @@
 #include <linux/rtnetlink.h>
 #include <net/net_namespace.h>
 #include <net/sock.h>
+#include <net/xdp.h>
 
 #include "netdev-genl-gen.h"
 
@@ -12,15 +13,24 @@ static int
 netdev_nl_dev_fill(struct net_device *netdev, struct sk_buff *rsp,
 		   const struct genl_info *info)
 {
+	u64 xdp_rx_meta = 0;
 	void *hdr;
 
 	hdr = genlmsg_iput(rsp, info);
 	if (!hdr)
 		return -EMSGSIZE;
 
+#define XDP_METADATA_KFUNC(_, flag, __, xmo) \
+	if (netdev->xdp_metadata_ops && netdev->xdp_metadata_ops->xmo) \
+		xdp_rx_meta |= flag;
+XDP_METADATA_KFUNC_xxx
+#undef XDP_METADATA_KFUNC
+
 	if (nla_put_u32(rsp, NETDEV_A_DEV_IFINDEX, netdev->ifindex) ||
 	    nla_put_u64_64bit(rsp, NETDEV_A_DEV_XDP_FEATURES,
-			      netdev->xdp_features, NETDEV_A_DEV_PAD)) {
+			      netdev->xdp_features, NETDEV_A_DEV_PAD) ||
+	    nla_put_u64_64bit(rsp, NETDEV_A_DEV_XDP_RX_METADATA_FEATURES,
+			      xdp_rx_meta, NETDEV_A_DEV_PAD)) {
 		genlmsg_cancel(rsp, hdr);
 		return -EINVAL;
 	}

net/core/xdp.c

@@ -741,7 +741,7 @@ __bpf_kfunc int bpf_xdp_metadata_rx_hash(const struct xdp_md *ctx, u32 *hash,
 __diag_pop();
 
 BTF_SET8_START(xdp_metadata_kfunc_ids)
-#define XDP_METADATA_KFUNC(_, name) BTF_ID_FLAGS(func, name, KF_TRUSTED_ARGS)
+#define XDP_METADATA_KFUNC(_, __, name, ___) BTF_ID_FLAGS(func, name, KF_TRUSTED_ARGS)
 XDP_METADATA_KFUNC_xxx
 #undef XDP_METADATA_KFUNC
 BTF_SET8_END(xdp_metadata_kfunc_ids)
@@ -752,7 +752,7 @@ static const struct btf_kfunc_id_set xdp_metadata_kfunc_set = {
 };
 
 BTF_ID_LIST(xdp_metadata_kfunc_ids_unsorted)
-#define XDP_METADATA_KFUNC(name, str) BTF_ID(func, str)
+#define XDP_METADATA_KFUNC(name, _, str, __) BTF_ID(func, str)
 XDP_METADATA_KFUNC_xxx
 #undef XDP_METADATA_KFUNC
 

tools/include/uapi/linux/netdev.h

@@ -38,11 +38,27 @@ enum netdev_xdp_act {
 	NETDEV_XDP_ACT_MASK = 127,
 };
 
+/**
+ * enum netdev_xdp_rx_metadata
+ * @NETDEV_XDP_RX_METADATA_TIMESTAMP: Device is capable of exposing receive HW
+ *   timestamp via bpf_xdp_metadata_rx_timestamp().
+ * @NETDEV_XDP_RX_METADATA_HASH: Device is capable of exposing receive packet
+ *   hash via bpf_xdp_metadata_rx_hash().
+ */
+enum netdev_xdp_rx_metadata {
+	NETDEV_XDP_RX_METADATA_TIMESTAMP = 1,
+	NETDEV_XDP_RX_METADATA_HASH = 2,
+
+	/* private: */
+	NETDEV_XDP_RX_METADATA_MASK = 3,
+};
+
 enum {
 	NETDEV_A_DEV_IFINDEX = 1,
 	NETDEV_A_DEV_PAD,
 	NETDEV_A_DEV_XDP_FEATURES,
 	NETDEV_A_DEV_XDP_ZC_MAX_SEGS,
+	NETDEV_A_DEV_XDP_RX_METADATA_FEATURES,
 
 	__NETDEV_A_DEV_MAX,
 	NETDEV_A_DEV_MAX = (__NETDEV_A_DEV_MAX - 1)

tools/net/ynl/generated/netdev-user.c

@@ -45,12 +45,26 @@ const char *netdev_xdp_act_str(enum netdev_xdp_act value)
 	return netdev_xdp_act_strmap[value];
 }
 
+static const char * const netdev_xdp_rx_metadata_strmap[] = {
+	[0] = "timestamp",
+	[1] = "hash",
+};
+
+const char *netdev_xdp_rx_metadata_str(enum netdev_xdp_rx_metadata value)
+{
+	value = ffs(value) - 1;
+	if (value < 0 || value >= (int)MNL_ARRAY_SIZE(netdev_xdp_rx_metadata_strmap))
+		return NULL;
+	return netdev_xdp_rx_metadata_strmap[value];
+}
+
 /* Policies */
 struct ynl_policy_attr netdev_dev_policy[NETDEV_A_DEV_MAX + 1] = {
 	[NETDEV_A_DEV_IFINDEX] = { .name = "ifindex", .type = YNL_PT_U32, },
 	[NETDEV_A_DEV_PAD] = { .name = "pad", .type = YNL_PT_IGNORE, },
 	[NETDEV_A_DEV_XDP_FEATURES] = { .name = "xdp-features", .type = YNL_PT_U64, },
 	[NETDEV_A_DEV_XDP_ZC_MAX_SEGS] = { .name = "xdp-zc-max-segs", .type = YNL_PT_U32, },
+	[NETDEV_A_DEV_XDP_RX_METADATA_FEATURES] = { .name = "xdp-rx-metadata-features", .type = YNL_PT_U64, },
 };
 
 struct ynl_policy_nest netdev_dev_nest = {
@@ -97,6 +111,11 @@ int netdev_dev_get_rsp_parse(const struct nlmsghdr *nlh, void *data)
 				return MNL_CB_ERROR;
 			dst->_present.xdp_zc_max_segs = 1;
 			dst->xdp_zc_max_segs = mnl_attr_get_u32(attr);
+		} else if (type == NETDEV_A_DEV_XDP_RX_METADATA_FEATURES) {
+			if (ynl_attr_validate(yarg, attr))
+				return MNL_CB_ERROR;
+			dst->_present.xdp_rx_metadata_features = 1;
+			dst->xdp_rx_metadata_features = mnl_attr_get_u64(attr);
 		}
 	}
 

tools/net/ynl/generated/netdev-user.h

@@ -18,6 +18,7 @@ extern const struct ynl_family ynl_netdev_family;
 /* Enums */
 const char *netdev_op_str(int op);
 const char *netdev_xdp_act_str(enum netdev_xdp_act value);
+const char *netdev_xdp_rx_metadata_str(enum netdev_xdp_rx_metadata value);
 
 /* Common nested types */
 /* ============== NETDEV_CMD_DEV_GET ============== */
@@ -48,11 +49,13 @@ struct netdev_dev_get_rsp {
 		__u32 ifindex:1;
 		__u32 xdp_features:1;
 		__u32 xdp_zc_max_segs:1;
+		__u32 xdp_rx_metadata_features:1;
 	} _present;
 
 	__u32 ifindex;
 	__u64 xdp_features;
 	__u32 xdp_zc_max_segs;
+	__u64 xdp_rx_metadata_features;
 };
 
 void netdev_dev_get_rsp_free(struct netdev_dev_get_rsp *rsp);

tools/net/ynl/samples/Makefile

@@ -4,7 +4,7 @@ include ../Makefile.deps
 
 CC=gcc
 CFLAGS=-std=gnu11 -O2 -W -Wall -Wextra -Wno-unused-parameter -Wshadow \
-	-I../lib/ -I../generated/ -idirafter $(UAPI_PATH)
+	-I../../../include/uapi -I../lib/ -I../generated/ -idirafter $(UAPI_PATH)
 ifeq ("$(DEBUG)","1")
   CFLAGS += -g -fsanitize=address -fsanitize=leak -static-libasan
 endif

tools/net/ynl/samples/netdev.c

@@ -32,12 +32,18 @@ static void netdev_print_device(struct netdev_dev_get_rsp *d, unsigned int op)
 	if (!d->_present.xdp_features)
 		return;
 
-	printf("%llx:", d->xdp_features);
+	printf("xdp-features (%llx):", d->xdp_features);
 	for (int i = 0; d->xdp_features > 1U << i; i++) {
 		if (d->xdp_features & (1U << i))
 			printf(" %s", netdev_xdp_act_str(1 << i));
 	}
 
+	printf(" xdp-rx-metadata-features (%llx):", d->xdp_rx_metadata_features);
+	for (int i = 0; d->xdp_rx_metadata_features > 1U << i; i++) {
+		if (d->xdp_rx_metadata_features & (1U << i))
+			printf(" %s", netdev_xdp_rx_metadata_str(1 << i));
+	}
+
 	printf(" xdp-zc-max-segs=%u", d->xdp_zc_max_segs);
 
 	name = netdev_op_str(op);