[Draft] Patching in BTF ID + unifying vmlinux and module BTFs acquisition

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

@@ -38,6 +38,7 @@
 #include <linux/avf/virtchnl.h>
 #include <linux/cpu_rmap.h>
 #include <linux/dim.h>
+#include <linux/btf.h>
 #include <net/devlink.h>
 #include <net/ipv6.h>
 #include <net/xdp_sock.h>
@@ -349,6 +350,9 @@ struct ice_vsi {
 	u16 num_xdp_txq;		 /* Used XDP queues */
 	u8 xdp_mapping_mode;		 /* ICE_MAP_MODE_[CONTIG|SCATTER] */
 
+	bool xdp_metadata_support;	 /* true if VSI should support xdp meta */
+	s32 btf_id;
+
 	/* setup back reference, to which aggregator node this VSI
 	 * corresponds to
 	 */

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

@@ -2419,6 +2419,21 @@ static void ice_vsi_assign_bpf_prog(struct ice_vsi *vsi, struct bpf_prog *prog)
 		WRITE_ONCE(vsi->rx_rings[i]->xdp_prog, vsi->xdp_prog);
 }
 
+static void ice_xdp_rings_set_metadata(struct ice_vsi *vsi)
+{
+	int i;
+
+	ice_for_each_rxq(vsi, i) {
+		vsi->rx_rings[i]->xdp_metadata_support = vsi->xdp_metadata_support;
+		vsi->rx_rings[i]->btf_id = vsi->btf_id;
+	}
+
+	for (i = 0; i < vsi->num_xdp_txq; i++) {
+		vsi->xdp_rings[i]->xdp_metadata_support = vsi->xdp_metadata_support;
+		vsi->xdp_rings[i]->btf_id = vsi->btf_id;
+	}
+}
+
 /**
  * ice_prepare_xdp_rings - Allocate, configure and setup Tx rings for XDP
  * @vsi: VSI to bring up Tx rings used by XDP
@@ -2458,6 +2473,8 @@ int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog)
 	if (ice_xdp_alloc_setup_rings(vsi))
 		goto clear_xdp_rings;
 
+	ice_xdp_rings_set_metadata(vsi);
+
 	/* follow the logic from ice_vsi_map_rings_to_vectors */
 	ice_for_each_q_vector(vsi, v_idx) {
 		struct ice_q_vector *q_vector = vsi->q_vectors[v_idx];
@@ -2613,7 +2630,7 @@ static void ice_vsi_rx_napi_schedule(struct ice_vsi *vsi)
  */
 static int
 ice_xdp_setup_prog(struct ice_vsi *vsi, struct bpf_prog *prog,
-		   struct netlink_ext_ack *extack)
+		   struct netlink_ext_ack *extack, u32 flags)
 {
 	int frame_size = vsi->netdev->mtu + ICE_ETH_PKT_HDR_PAD;
 	bool if_running = netif_running(vsi->netdev);
@@ -2633,6 +2650,11 @@ ice_xdp_setup_prog(struct ice_vsi *vsi, struct bpf_prog *prog,
 		}
 	}
 
+	if (flags & XDP_FLAGS_USE_METADATA) {
+		vsi->xdp_metadata_support = true;
+		vsi->btf_id = btf_get_type_id(THIS_MODULE, "xdp_meta_generic", BTF_KIND_STRUCT);
+	}	
+
 	if (!ice_is_xdp_ena_vsi(vsi) && prog) {
 		vsi->num_xdp_txq = vsi->alloc_rxq;
 		xdp_ring_err = ice_prepare_xdp_rings(vsi, prog);
@@ -2686,7 +2708,7 @@ static int ice_xdp(struct net_device *dev, struct netdev_bpf *xdp)
 
 	switch (xdp->command) {
 	case XDP_SETUP_PROG:
-		return ice_xdp_setup_prog(vsi, xdp->prog, xdp->extack);
+		return ice_xdp_setup_prog(vsi, xdp->prog, xdp->extack, xdp->flags);
 	case XDP_SETUP_XSK_POOL:
 		return ice_xsk_pool_setup(vsi, xdp->xsk.pool,
 					  xdp->xsk.queue_id);

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

@@ -1135,6 +1135,10 @@ int ice_clean_rx_irq(struct ice_ring *rx_ring, int budget)
 		hard_start = page_address(rx_buf->page) + rx_buf->page_offset -
 			     offset;
 		xdp_prepare_buff(&xdp, hard_start, offset, size, true);
+
+		if (likely(rx_ring->xdp_metadata_support))
+			ice_xdp_set_meta(&xdp, rx_desc, rx_ring->btf_id);
+
 #if (PAGE_SIZE > 4096)
 		/* At larger PAGE_SIZE, frame_sz depend on len size */
 		xdp.frame_sz = ice_rx_frame_truesize(rx_ring, size);

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

@@ -276,6 +276,7 @@ struct ice_ring {
 	u16 q_handle;			/* Queue handle per TC */
 
 	u8 ring_active:1;		/* is ring online or not */
+	u8 xdp_metadata_support:1;	/* is xdp metadata support */
 
 	u16 count;			/* Number of descriptors */
 	u16 reg_idx;			/* HW register index of the ring */
@@ -301,6 +302,9 @@ struct ice_ring {
 	/* CL3 - 3rd cacheline starts here */
 	struct xdp_rxq_info xdp_rxq;
 	struct sk_buff *skb;
+	s32 btf_id;
+
+
 	/* CLX - the below items are only accessed infrequently and should be
 	 * in their own cache line if possible
 	 */

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

@@ -46,6 +46,26 @@ static inline void ice_xdp_ring_update_tail(struct ice_ring *xdp_ring)
 	writel_relaxed(xdp_ring->next_to_use, xdp_ring->tail);
 }
 
+static inline void ice_xdp_set_meta(struct xdp_buff *xdp, union ice_32b_rx_flex_desc *desc, s32 btf_id)
+{
+	struct ice_32b_rx_flex_desc_nic *flex = (struct ice_32b_rx_flex_desc_nic *)desc;
+	struct xdp_meta_generic *md = xdp->data - sizeof(struct xdp_meta_generic);
+
+	md->rxcvid = le16_to_cpu(flex->flex_ts.flex.vlan_id);
+	md->hash = le32_to_cpu(flex->rss_hash);
+
+	md->flags = 0;
+	md->free_slot = 0;
+	md->csum_off = 0;
+	md->txcvid = 0;
+	md->csum = 0;
+	md->tstamp = 0;
+
+	md->btf_id = btf_id;
+
+	xdp->data_meta = md;
+}
+
 void ice_finalize_xdp_rx(struct ice_ring *rx_ring, unsigned int xdp_res);
 int ice_xmit_xdp_buff(struct xdp_buff *xdp, struct ice_ring *xdp_ring);
 int ice_xmit_xdp_ring(void *data, u16 size, struct ice_ring *xdp_ring);

include/linux/btf.h

@@ -94,6 +94,7 @@ u32 btf_obj_id(const struct btf *btf);
 bool btf_is_kernel(const struct btf *btf);
 bool btf_is_module(const struct btf *btf);
 struct module *btf_try_get_module(const struct btf *btf);
+s32 btf_get_type_id(const struct module *mod, char *name, u32 kind);
 u32 btf_nr_types(const struct btf *btf);
 bool btf_member_is_reg_int(const struct btf *btf, const struct btf_type *s,
 			   const struct btf_member *m,

include/net/xdp.h

@@ -76,6 +76,24 @@ struct xdp_buff {
 	u32 frame_sz; /* frame size to deduce data_hard_end/reserved tailroom*/
 };
 
+struct xdp_meta_generic {
+	// Tx part
+	u32        flags;
+	u16        free_slot;
+	u16        csum_off;
+	u16        txcvid;
+
+	// Rx part
+	u16        rxcvid;
+	u32        csum;
+	u32        hash;
+	u64        tstamp;
+
+	// BTF ID
+	u32        btf_id;
+} __packed __aligned(8);
+static_assert(sizeof(struct xdp_meta_generic) == 32);
+
 static __always_inline void
 xdp_init_buff(struct xdp_buff *xdp, u32 frame_sz, struct xdp_rxq_info *rxq)
 {

include/uapi/linux/bpf.h

@@ -873,6 +873,7 @@ enum bpf_cmd {
 	BPF_ITER_CREATE,
 	BPF_LINK_DETACH,
 	BPF_PROG_BIND_MAP,
+	BPF_BTF_VMLINUX_INFO,
 };
 
 enum bpf_map_type {
@@ -1396,6 +1397,11 @@ union bpf_attr {
 		__aligned_u64	info;
 	} info;
 
+	struct { /* anonymous struct used by BPF_BTF_VMLINUX_INFO */
+		__u32 info_len;
+		__aligned_u64 info;
+	} info_vmlinux;
+
 	struct { /* anonymous struct used by BPF_PROG_QUERY command */
 		__u32		target_fd;	/* container object to query */
 		__u32		attach_type;

include/uapi/linux/if_link.h

@@ -1182,11 +1182,13 @@ enum {
 #define XDP_FLAGS_DRV_MODE		(1U << 2)
 #define XDP_FLAGS_HW_MODE		(1U << 3)
 #define XDP_FLAGS_REPLACE		(1U << 4)
+#define XDP_FLAGS_USE_METADATA		(1U << 5)
 #define XDP_FLAGS_MODES			(XDP_FLAGS_SKB_MODE | \
 					 XDP_FLAGS_DRV_MODE | \
 					 XDP_FLAGS_HW_MODE)
 #define XDP_FLAGS_MASK			(XDP_FLAGS_UPDATE_IF_NOEXIST | \
-					 XDP_FLAGS_MODES | XDP_FLAGS_REPLACE)
+					 XDP_FLAGS_MODES | XDP_FLAGS_REPLACE | \
+					 XDP_FLAGS_USE_METADATA)
 
 /* These are stored into IFLA_XDP_ATTACHED on dump. */
 enum {

kernel/bpf/btf.c

@@ -6152,6 +6152,43 @@ struct module *btf_try_get_module(const struct btf *btf)
 	return res;
 }
 
+struct btf *btf_get_from_module(const struct module *module)
+{
+	struct btf *res = NULL;
+#ifdef CONFIG_DEBUG_INFO_BTF_MODULES
+	struct btf_module *btf_mod, *tmp;
+
+	mutex_lock(&btf_module_mutex);
+	list_for_each_entry_safe(btf_mod, tmp, &btf_modules, list) {
+		if (btf_mod->module != module)
+			continue;
+
+		res = btf_mod->btf;
+
+		break;
+	}
+	mutex_unlock(&btf_module_mutex);
+#endif
+
+	return res;
+}
+
+s32 btf_get_type_id(const struct module *mod, char *name, u32 kind)
+{
+	struct btf *btf;
+
+	if (mod)
+		btf = btf_get_from_module(mod);
+	else
+		btf = bpf_get_btf_vmlinux();
+
+	if (!btf)
+		return 0;
+
+	return btf_find_by_name_kind(btf, name, kind);
+}
+EXPORT_SYMBOL_GPL(btf_get_type_id);
+
 BPF_CALL_4(bpf_btf_find_by_name_kind, char *, name, int, name_sz, u32, kind, int, flags)
 {
 	struct btf *btf;

kernel/bpf/cpumap.c

@@ -215,6 +215,7 @@ static int cpu_map_bpf_prog_run_xdp(struct bpf_cpu_map_entry *rcpu,
 {
 	struct xdp_rxq_info rxq;
 	struct xdp_buff xdp;
+	volatile struct xdp_meta_generic xdp_meta_generic;
 	int i, nframes = 0;
 
 	xdp_set_return_frame_no_direct();

kernel/bpf/syscall.c

@@ -4542,6 +4542,25 @@ static int bpf_prog_bind_map(union bpf_attr *attr)
 	return ret;
 }
 
+#define BPF_BTF_VMLINUX_INFO_LAST_FIELD info.info
+
+static int bpf_btf_get_vmlinux_info(const union bpf_attr *attr,
+				    union bpf_attr __user *uattr)
+{
+	struct bpf_btf_info __user *uinfo = u64_to_user_ptr(attr->info.info);
+	u32 info_len = attr->info.info_len;
+	int err;
+
+	if (CHECK_ATTR(BPF_BTF_VMLINUX_INFO))
+		return -EINVAL;
+
+	err = bpf_check_uarg_tail_zero(USER_BPFPTR(uinfo), sizeof(*uinfo), info_len);
+	if (err)
+		return err;
+
+	return btf_get_info_by_fd(bpf_get_btf_vmlinux(), attr, uattr);
+}
+
 static int __sys_bpf(int cmd, bpfptr_t uattr, unsigned int size)
 {
 	union bpf_attr attr;
@@ -4678,6 +4697,9 @@ static int __sys_bpf(int cmd, bpfptr_t uattr, unsigned int size)
 	case BPF_PROG_BIND_MAP:
 		err = bpf_prog_bind_map(&attr);
 		break;
+	case BPF_BTF_VMLINUX_INFO:
+		err = bpf_btf_get_vmlinux_info(&attr, uattr.user);
+		break;
 	default:
 		err = -EINVAL;
 		break;

samples/bpf/Makefile

@@ -57,6 +57,7 @@ tprogs-y += xdp_redirect_map_multi
 tprogs-y += xdp_redirect_map
 tprogs-y += xdp_redirect
 tprogs-y += xdp_monitor
+tprogs-y += xdp_meta
 
 # Libbpf dependencies
 LIBBPF = $(TOOLS_PATH)/lib/bpf/libbpf.a
@@ -118,6 +119,7 @@ xdp_redirect_cpu-objs := xdp_redirect_cpu_user.o $(XDP_SAMPLE)
 xdp_redirect_map-objs := xdp_redirect_map_user.o $(XDP_SAMPLE)
 xdp_redirect-objs := xdp_redirect_user.o $(XDP_SAMPLE)
 xdp_monitor-objs := xdp_monitor_user.o $(XDP_SAMPLE)
+xdp_meta-objs := xdp_meta_user.o
 
 # Tell kbuild to always build the programs
 always-y := $(tprogs-y)
@@ -314,6 +316,7 @@ $(obj)/xdp_redirect_map_multi_user.o: $(obj)/xdp_redirect_map_multi.skel.h
 $(obj)/xdp_redirect_map_user.o: $(obj)/xdp_redirect_map.skel.h
 $(obj)/xdp_redirect_user.o: $(obj)/xdp_redirect.skel.h
 $(obj)/xdp_monitor_user.o: $(obj)/xdp_monitor.skel.h
+$(obj)/xdp_meta_user.o: $(obj)/xdp_meta.skel.h
 
 $(obj)/tracex5_kern.o: $(obj)/syscall_nrs.h
 $(obj)/hbm_out_kern.o: $(src)/hbm.h $(src)/hbm_kern.h
@@ -371,17 +374,18 @@ $(obj)/%.bpf.o: $(src)/%.bpf.c $(obj)/vmlinux.h $(src)/xdp_sample.bpf.h $(src)/x
 		-c $(filter %.bpf.c,$^) -o $@
 
 LINKED_SKELS := xdp_redirect_cpu.skel.h xdp_redirect_map_multi.skel.h \
-		xdp_redirect_map.skel.h xdp_redirect.skel.h xdp_monitor.skel.h
+		xdp_redirect_map.skel.h xdp_redirect.skel.h xdp_monitor.skel.h \
+		xdp_meta.skel.h
 clean-files += $(LINKED_SKELS)
 
 xdp_redirect_cpu.skel.h-deps := xdp_redirect_cpu.bpf.o xdp_sample.bpf.o
 xdp_redirect_map_multi.skel.h-deps := xdp_redirect_map_multi.bpf.o xdp_sample.bpf.o
 xdp_redirect_map.skel.h-deps := xdp_redirect_map.bpf.o xdp_sample.bpf.o
 xdp_redirect.skel.h-deps := xdp_redirect.bpf.o xdp_sample.bpf.o
 xdp_monitor.skel.h-deps := xdp_monitor.bpf.o xdp_sample.bpf.o
+xdp_meta.skel.h-deps := xdp_meta.bpf.o
 
 LINKED_BPF_SRCS := $(patsubst %.bpf.o,%.bpf.c,$(foreach skel,$(LINKED_SKELS),$($(skel)-deps)))
-
 BPF_SRCS_LINKED := $(notdir $(wildcard $(src)/*.bpf.c))
 BPF_OBJS_LINKED := $(patsubst %.bpf.c,$(obj)/%.bpf.o, $(BPF_SRCS_LINKED))
 BPF_SKELS_LINKED := $(addprefix $(obj)/,$(LINKED_SKELS))