[PATCH bpf-next v3 0/7] Add attach_type in bpf_link

[PATCH bpf-next v3 0/7] Add attach_type in bpf_link

[Tao Chen]

Andrii suggested moving the attach_type into bpf_link, the previous discussion
is as follows:
https://lore.kernel.org/bpf/CAEf4BzY7TZRjxpCJM-+LYgEqe23YFj5Uv3isb7gat2-HU4OSng@mail.gmail.com

patch1 add attach_type in bpf_link, and pass it to bpf_link_init, which
will init the attach_type field.

patch2-7 remove the attach_type in struct bpf_xx_link, update the info
with bpf_link attach_type.

There are some functions finally call bpf_link_init but do not have bpf_attr
from user or do not need to init attach_type from user like bpf_raw_tracepoint_open,
now use prog->expected_attach_type to init attach_type.

bpf_struct_ops_map_update_elem
bpf_raw_tracepoint_open
bpf_struct_ops_test_run

Feedback of any kind is welcome, thanks.

Tao Chen (7):
  bpf: Add attach_type in bpf_link
  bpf: Remove attach_type in bpf_cgroup_link
  bpf: Remove attach_type in sockmap_link
  bpf: Remove location field in tcx_link
  bpf: Remove attach_type in bpf_netns_link
  bpf: Remove attach_type in bpf_tracing_link
  netkit: Remove location field in netkit_link

 drivers/net/netkit.c           | 10 ++++-----
 include/linux/bpf-cgroup.h     |  1 -
 include/linux/bpf.h            | 29 ++++++++++++++----------
 include/net/tcx.h              |  1 -
 kernel/bpf/bpf_iter.c          |  3 ++-
 kernel/bpf/bpf_struct_ops.c    |  5 +++--
 kernel/bpf/cgroup.c            | 17 +++++++--------
 kernel/bpf/net_namespace.c     |  8 +++----
 kernel/bpf/syscall.c           | 40 ++++++++++++++++++++--------------
 kernel/bpf/tcx.c               | 16 +++++++-------
 kernel/bpf/trampoline.c        | 10 +++++----
 kernel/trace/bpf_trace.c       |  4 ++--
 net/bpf/bpf_dummy_struct_ops.c |  3 ++-
 net/core/dev.c                 |  3 ++-
 net/core/sock_map.c            | 13 +++++------
 net/netfilter/nf_bpf_link.c    |  3 ++-
 16 files changed, 89 insertions(+), 77 deletions(-)

Change list:
 v2 -> v3:
  - move sleepable field to the end in bpf_link to fill the byte hole.(Jiri)
  - Acked from Jiri
 v2: https://lore.kernel.org/bpf/20250708082228.824766-1-chen.dylane@linux.dev

 v1 -> v2:
  - fix build error.(Jiri)
 v1: https://lore.kernel.org/bpf/20250707153916.802802-1-chen.dylane@linux.dev
-- 
2.48.1

[PATCH bpf-next v3 1/7] bpf: Add attach_type in bpf_link

[Tao Chen]

Attach_type will be set when link created from user, it is better
to record attach_type in bpf_link directly suggested by Andrii. So
add the attach_type field in bpf_link and move the sleepable field to
the end just to fill the byte hole.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Tao Chen <chen.dylane@linux.dev>
---
 drivers/net/netkit.c           |  2 +-
 include/linux/bpf.h            | 28 ++++++++++++++++-----------
 kernel/bpf/bpf_iter.c          |  3 ++-
 kernel/bpf/bpf_struct_ops.c    |  5 +++--
 kernel/bpf/cgroup.c            |  4 ++--
 kernel/bpf/net_namespace.c     |  2 +-
 kernel/bpf/syscall.c           | 35 +++++++++++++++++++++-------------
 kernel/bpf/tcx.c               |  3 ++-
 kernel/bpf/trampoline.c        | 10 ++++++----
 kernel/trace/bpf_trace.c       |  4 ++--
 net/bpf/bpf_dummy_struct_ops.c |  3 ++-
 net/core/dev.c                 |  3 ++-
 net/core/sock_map.c            |  3 ++-
 net/netfilter/nf_bpf_link.c    |  3 ++-
 14 files changed, 66 insertions(+), 42 deletions(-)

diff --git a/drivers/net/netkit.c b/drivers/net/netkit.c
index d072a7968f5..5928c99eac7 100644
--- a/drivers/net/netkit.c
+++ b/drivers/net/netkit.c
@@ -775,7 +775,7 @@ static int netkit_link_init(struct netkit_link *nkl,
 			    struct bpf_prog *prog)
 {
 	bpf_link_init(&nkl->link, BPF_LINK_TYPE_NETKIT,
-		      &netkit_link_lops, prog);
+		      &netkit_link_lops, prog, attr->link_create.attach_type);
 	nkl->location = attr->link_create.attach_type;
 	nkl->dev = dev;
 	return bpf_link_prime(&nkl->link, link_primer);
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 34dd90ec7fa..dd5070039de 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1729,12 +1729,10 @@ struct bpf_link {
 	enum bpf_link_type type;
 	const struct bpf_link_ops *ops;
 	struct bpf_prog *prog;
-	/* whether BPF link itself has "sleepable" semantics, which can differ
-	 * from underlying BPF program having a "sleepable" semantics, as BPF
-	 * link's semantics is determined by target attach hook
-	 */
-	bool sleepable;
+
 	u32 flags;
+	enum bpf_attach_type attach_type;
+
 	/* rcu is used before freeing, work can be used to schedule that
 	 * RCU-based freeing before that, so they never overlap
 	 */
@@ -1742,6 +1740,11 @@ struct bpf_link {
 		struct rcu_head rcu;
 		struct work_struct work;
 	};
+	/* whether BPF link itself has "sleepable" semantics, which can differ
+	 * from underlying BPF program having a "sleepable" semantics, as BPF
+	 * link's semantics is determined by target attach hook
+	 */
+	bool sleepable;
 };
 
 struct bpf_link_ops {
@@ -2034,11 +2037,13 @@ int bpf_prog_ctx_arg_info_init(struct bpf_prog *prog,
 
 #if defined(CONFIG_CGROUP_BPF) && defined(CONFIG_BPF_LSM)
 int bpf_trampoline_link_cgroup_shim(struct bpf_prog *prog,
-				    int cgroup_atype);
+				    int cgroup_atype,
+				    enum bpf_attach_type attach_type);
 void bpf_trampoline_unlink_cgroup_shim(struct bpf_prog *prog);
 #else
 static inline int bpf_trampoline_link_cgroup_shim(struct bpf_prog *prog,
-						  int cgroup_atype)
+						  int cgroup_atype,
+						  enum bpf_attach_type attach_type)
 {
 	return -EOPNOTSUPP;
 }
@@ -2528,10 +2533,11 @@ int bpf_map_new_fd(struct bpf_map *map, int flags);
 int bpf_prog_new_fd(struct bpf_prog *prog);
 
 void bpf_link_init(struct bpf_link *link, enum bpf_link_type type,
-		   const struct bpf_link_ops *ops, struct bpf_prog *prog);
+		   const struct bpf_link_ops *ops, struct bpf_prog *prog,
+		   enum bpf_attach_type attach_type);
 void bpf_link_init_sleepable(struct bpf_link *link, enum bpf_link_type type,
 			     const struct bpf_link_ops *ops, struct bpf_prog *prog,
-			     bool sleepable);
+			     bool sleepable, enum bpf_attach_type attach_type);
 int bpf_link_prime(struct bpf_link *link, struct bpf_link_primer *primer);
 int bpf_link_settle(struct bpf_link_primer *primer);
 void bpf_link_cleanup(struct bpf_link_primer *primer);
@@ -2883,13 +2889,13 @@ bpf_prog_inc_not_zero(struct bpf_prog *prog)
 
 static inline void bpf_link_init(struct bpf_link *link, enum bpf_link_type type,
 				 const struct bpf_link_ops *ops,
-				 struct bpf_prog *prog)
+				 struct bpf_prog *prog, enum bpf_attach_type attach_type)
 {
 }
 
 static inline void bpf_link_init_sleepable(struct bpf_link *link, enum bpf_link_type type,
 					   const struct bpf_link_ops *ops, struct bpf_prog *prog,
-					   bool sleepable)
+					   bool sleepable, enum bpf_attach_type attach_type)
 {
 }
 
diff --git a/kernel/bpf/bpf_iter.c b/kernel/bpf/bpf_iter.c
index 303ab1f42d3..0cbcae72707 100644
--- a/kernel/bpf/bpf_iter.c
+++ b/kernel/bpf/bpf_iter.c
@@ -552,7 +552,8 @@ int bpf_iter_link_attach(const union bpf_attr *attr, bpfptr_t uattr,
 	if (!link)
 		return -ENOMEM;
 
-	bpf_link_init(&link->link, BPF_LINK_TYPE_ITER, &bpf_iter_link_lops, prog);
+	bpf_link_init(&link->link, BPF_LINK_TYPE_ITER, &bpf_iter_link_lops, prog,
+		      attr->link_create.attach_type);
 	link->tinfo = tinfo;
 
 	err = bpf_link_prime(&link->link, &link_primer);
diff --git a/kernel/bpf/bpf_struct_ops.c b/kernel/bpf/bpf_struct_ops.c
index 96113633e39..687a3e9c76f 100644
--- a/kernel/bpf/bpf_struct_ops.c
+++ b/kernel/bpf/bpf_struct_ops.c
@@ -808,7 +808,7 @@ static long bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 			goto reset_unlock;
 		}
 		bpf_link_init(&link->link, BPF_LINK_TYPE_STRUCT_OPS,
-			      &bpf_struct_ops_link_lops, prog);
+			      &bpf_struct_ops_link_lops, prog, prog->expected_attach_type);
 		*plink++ = &link->link;
 
 		ksym = kzalloc(sizeof(*ksym), GFP_USER);
@@ -1351,7 +1351,8 @@ int bpf_struct_ops_link_create(union bpf_attr *attr)
 		err = -ENOMEM;
 		goto err_out;
 	}
-	bpf_link_init(&link->link, BPF_LINK_TYPE_STRUCT_OPS, &bpf_struct_ops_map_lops, NULL);
+	bpf_link_init(&link->link, BPF_LINK_TYPE_STRUCT_OPS, &bpf_struct_ops_map_lops, NULL,
+		      attr->link_create.attach_type);
 
 	err = bpf_link_prime(&link->link, &link_primer);
 	if (err)
diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
index cd220e861d6..bacdd0ca741 100644
--- a/kernel/bpf/cgroup.c
+++ b/kernel/bpf/cgroup.c
@@ -867,7 +867,7 @@ static int __cgroup_bpf_attach(struct cgroup *cgrp,
 	cgrp->bpf.flags[atype] = saved_flags;
 
 	if (type == BPF_LSM_CGROUP) {
-		err = bpf_trampoline_link_cgroup_shim(new_prog, atype);
+		err = bpf_trampoline_link_cgroup_shim(new_prog, atype, type);
 		if (err)
 			goto cleanup;
 	}
@@ -1495,7 +1495,7 @@ int cgroup_bpf_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
 		goto out_put_cgroup;
 	}
 	bpf_link_init(&link->link, BPF_LINK_TYPE_CGROUP, &bpf_cgroup_link_lops,
-		      prog);
+		      prog, attr->link_create.attach_type);
 	link->cgroup = cgrp;
 	link->type = attr->link_create.attach_type;
 
diff --git a/kernel/bpf/net_namespace.c b/kernel/bpf/net_namespace.c
index 868cc2c4389..63702c86275 100644
--- a/kernel/bpf/net_namespace.c
+++ b/kernel/bpf/net_namespace.c
@@ -501,7 +501,7 @@ int netns_bpf_link_create(const union bpf_attr *attr, struct bpf_prog *prog)
 		goto out_put_net;
 	}
 	bpf_link_init(&net_link->link, BPF_LINK_TYPE_NETNS,
-		      &bpf_netns_link_ops, prog);
+		      &bpf_netns_link_ops, prog, type);
 	net_link->net = net;
 	net_link->type = type;
 	net_link->netns_type = netns_type;
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 7db7182a305..14883b3040a 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3069,7 +3069,7 @@ static int bpf_obj_get(const union bpf_attr *attr)
  */
 void bpf_link_init_sleepable(struct bpf_link *link, enum bpf_link_type type,
 			     const struct bpf_link_ops *ops, struct bpf_prog *prog,
-			     bool sleepable)
+			     bool sleepable, enum bpf_attach_type attach_type)
 {
 	WARN_ON(ops->dealloc && ops->dealloc_deferred);
 	atomic64_set(&link->refcnt, 1);
@@ -3078,12 +3078,14 @@ void bpf_link_init_sleepable(struct bpf_link *link, enum bpf_link_type type,
 	link->id = 0;
 	link->ops = ops;
 	link->prog = prog;
+	link->attach_type = attach_type;
 }
 
 void bpf_link_init(struct bpf_link *link, enum bpf_link_type type,
-		   const struct bpf_link_ops *ops, struct bpf_prog *prog)
+		   const struct bpf_link_ops *ops, struct bpf_prog *prog,
+		   enum bpf_attach_type attach_type)
 {
-	bpf_link_init_sleepable(link, type, ops, prog, false);
+	bpf_link_init_sleepable(link, type, ops, prog, false, attach_type);
 }
 
 static void bpf_link_free_id(int id)
@@ -3443,7 +3445,8 @@ static const struct bpf_link_ops bpf_tracing_link_lops = {
 static int bpf_tracing_prog_attach(struct bpf_prog *prog,
 				   int tgt_prog_fd,
 				   u32 btf_id,
-				   u64 bpf_cookie)
+				   u64 bpf_cookie,
+				   enum bpf_attach_type attach_type)
 {
 	struct bpf_link_primer link_primer;
 	struct bpf_prog *tgt_prog = NULL;
@@ -3511,7 +3514,8 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
 		goto out_put_prog;
 	}
 	bpf_link_init(&link->link.link, BPF_LINK_TYPE_TRACING,
-		      &bpf_tracing_link_lops, prog);
+		      &bpf_tracing_link_lops, prog, attach_type);
+
 	link->attach_type = prog->expected_attach_type;
 	link->link.cookie = bpf_cookie;
 
@@ -4049,7 +4053,8 @@ static int bpf_perf_link_attach(const union bpf_attr *attr, struct bpf_prog *pro
 		err = -ENOMEM;
 		goto out_put_file;
 	}
-	bpf_link_init(&link->link, BPF_LINK_TYPE_PERF_EVENT, &bpf_perf_link_lops, prog);
+	bpf_link_init(&link->link, BPF_LINK_TYPE_PERF_EVENT, &bpf_perf_link_lops, prog,
+		      attr->link_create.attach_type);
 	link->perf_file = perf_file;
 
 	err = bpf_link_prime(&link->link, &link_primer);
@@ -4081,7 +4086,8 @@ static int bpf_perf_link_attach(const union bpf_attr *attr, struct bpf_prog *pro
 #endif /* CONFIG_PERF_EVENTS */
 
 static int bpf_raw_tp_link_attach(struct bpf_prog *prog,
-				  const char __user *user_tp_name, u64 cookie)
+				  const char __user *user_tp_name, u64 cookie,
+				  enum bpf_attach_type attach_type)
 {
 	struct bpf_link_primer link_primer;
 	struct bpf_raw_tp_link *link;
@@ -4104,7 +4110,7 @@ static int bpf_raw_tp_link_attach(struct bpf_prog *prog,
 			tp_name = prog->aux->attach_func_name;
 			break;
 		}
-		return bpf_tracing_prog_attach(prog, 0, 0, 0);
+		return bpf_tracing_prog_attach(prog, 0, 0, 0, attach_type);
 	case BPF_PROG_TYPE_RAW_TRACEPOINT:
 	case BPF_PROG_TYPE_RAW_TRACEPOINT_WRITABLE:
 		if (strncpy_from_user(buf, user_tp_name, sizeof(buf) - 1) < 0)
@@ -4127,7 +4133,7 @@ static int bpf_raw_tp_link_attach(struct bpf_prog *prog,
 	}
 	bpf_link_init_sleepable(&link->link, BPF_LINK_TYPE_RAW_TRACEPOINT,
 				&bpf_raw_tp_link_lops, prog,
-				tracepoint_is_faultable(btp->tp));
+				tracepoint_is_faultable(btp->tp), attach_type);
 	link->btp = btp;
 	link->cookie = cookie;
 
@@ -4168,7 +4174,7 @@ static int bpf_raw_tracepoint_open(const union bpf_attr *attr)
 
 	tp_name = u64_to_user_ptr(attr->raw_tracepoint.name);
 	cookie = attr->raw_tracepoint.cookie;
-	fd = bpf_raw_tp_link_attach(prog, tp_name, cookie);
+	fd = bpf_raw_tp_link_attach(prog, tp_name, cookie, prog->expected_attach_type);
 	if (fd < 0)
 		bpf_prog_put(prog);
 	return fd;
@@ -5536,7 +5542,8 @@ static int link_create(union bpf_attr *attr, bpfptr_t uattr)
 		ret = bpf_tracing_prog_attach(prog,
 					      attr->link_create.target_fd,
 					      attr->link_create.target_btf_id,
-					      attr->link_create.tracing.cookie);
+					      attr->link_create.tracing.cookie,
+					      attr->link_create.attach_type);
 		break;
 	case BPF_PROG_TYPE_LSM:
 	case BPF_PROG_TYPE_TRACING:
@@ -5545,7 +5552,8 @@ static int link_create(union bpf_attr *attr, bpfptr_t uattr)
 			goto out;
 		}
 		if (prog->expected_attach_type == BPF_TRACE_RAW_TP)
-			ret = bpf_raw_tp_link_attach(prog, NULL, attr->link_create.tracing.cookie);
+			ret = bpf_raw_tp_link_attach(prog, NULL, attr->link_create.tracing.cookie,
+						     attr->link_create.attach_type);
 		else if (prog->expected_attach_type == BPF_TRACE_ITER)
 			ret = bpf_iter_link_attach(attr, uattr, prog);
 		else if (prog->expected_attach_type == BPF_LSM_CGROUP)
@@ -5554,7 +5562,8 @@ static int link_create(union bpf_attr *attr, bpfptr_t uattr)
 			ret = bpf_tracing_prog_attach(prog,
 						      attr->link_create.target_fd,
 						      attr->link_create.target_btf_id,
-						      attr->link_create.tracing.cookie);
+						      attr->link_create.tracing.cookie,
+						      attr->link_create.attach_type);
 		break;
 	case BPF_PROG_TYPE_FLOW_DISSECTOR:
 	case BPF_PROG_TYPE_SK_LOOKUP:
diff --git a/kernel/bpf/tcx.c b/kernel/bpf/tcx.c
index 2e4885e7781..e6a14f408d9 100644
--- a/kernel/bpf/tcx.c
+++ b/kernel/bpf/tcx.c
@@ -301,7 +301,8 @@ static int tcx_link_init(struct tcx_link *tcx,
 			 struct net_device *dev,
 			 struct bpf_prog *prog)
 {
-	bpf_link_init(&tcx->link, BPF_LINK_TYPE_TCX, &tcx_link_lops, prog);
+	bpf_link_init(&tcx->link, BPF_LINK_TYPE_TCX, &tcx_link_lops, prog,
+		      attr->link_create.attach_type);
 	tcx->location = attr->link_create.attach_type;
 	tcx->dev = dev;
 	return bpf_link_prime(&tcx->link, link_primer);
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index b1e358c16ee..0e364614c3a 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -674,7 +674,8 @@ static const struct bpf_link_ops bpf_shim_tramp_link_lops = {
 
 static struct bpf_shim_tramp_link *cgroup_shim_alloc(const struct bpf_prog *prog,
 						     bpf_func_t bpf_func,
-						     int cgroup_atype)
+						     int cgroup_atype,
+						     enum bpf_attach_type attach_type)
 {
 	struct bpf_shim_tramp_link *shim_link = NULL;
 	struct bpf_prog *p;
@@ -701,7 +702,7 @@ static struct bpf_shim_tramp_link *cgroup_shim_alloc(const struct bpf_prog *prog
 	p->expected_attach_type = BPF_LSM_MAC;
 	bpf_prog_inc(p);
 	bpf_link_init(&shim_link->link.link, BPF_LINK_TYPE_UNSPEC,
-		      &bpf_shim_tramp_link_lops, p);
+		      &bpf_shim_tramp_link_lops, p, attach_type);
 	bpf_cgroup_atype_get(p->aux->attach_btf_id, cgroup_atype);
 
 	return shim_link;
@@ -726,7 +727,8 @@ static struct bpf_shim_tramp_link *cgroup_shim_find(struct bpf_trampoline *tr,
 }
 
 int bpf_trampoline_link_cgroup_shim(struct bpf_prog *prog,
-				    int cgroup_atype)
+				    int cgroup_atype,
+				    enum bpf_attach_type attach_type)
 {
 	struct bpf_shim_tramp_link *shim_link = NULL;
 	struct bpf_attach_target_info tgt_info = {};
@@ -763,7 +765,7 @@ int bpf_trampoline_link_cgroup_shim(struct bpf_prog *prog,
 
 	/* Allocate and install new shim. */
 
-	shim_link = cgroup_shim_alloc(prog, bpf_func, cgroup_atype);
+	shim_link = cgroup_shim_alloc(prog, bpf_func, cgroup_atype, attach_type);
 	if (!shim_link) {
 		err = -ENOMEM;
 		goto err;
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index e7f97a9a8bb..ffdde840abb 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -2986,7 +2986,7 @@ int bpf_kprobe_multi_link_attach(const union bpf_attr *attr, struct bpf_prog *pr
 	}
 
 	bpf_link_init(&link->link, BPF_LINK_TYPE_KPROBE_MULTI,
-		      &bpf_kprobe_multi_link_lops, prog);
+		      &bpf_kprobe_multi_link_lops, prog, attr->link_create.attach_type);
 
 	err = bpf_link_prime(&link->link, &link_primer);
 	if (err)
@@ -3441,7 +3441,7 @@ int bpf_uprobe_multi_link_attach(const union bpf_attr *attr, struct bpf_prog *pr
 	link->link.flags = flags;
 
 	bpf_link_init(&link->link, BPF_LINK_TYPE_UPROBE_MULTI,
-		      &bpf_uprobe_multi_link_lops, prog);
+		      &bpf_uprobe_multi_link_lops, prog, attr->link_create.attach_type);
 
 	for (i = 0; i < cnt; i++) {
 		uprobes[i].uprobe = uprobe_register(d_real_inode(link->path.dentry),
diff --git a/net/bpf/bpf_dummy_struct_ops.c b/net/bpf/bpf_dummy_struct_ops.c
index f71f67c6896..812457819b5 100644
--- a/net/bpf/bpf_dummy_struct_ops.c
+++ b/net/bpf/bpf_dummy_struct_ops.c
@@ -171,7 +171,8 @@ int bpf_struct_ops_test_run(struct bpf_prog *prog, const union bpf_attr *kattr,
 	}
 	/* prog doesn't take the ownership of the reference from caller */
 	bpf_prog_inc(prog);
-	bpf_link_init(&link->link, BPF_LINK_TYPE_STRUCT_OPS, &bpf_struct_ops_link_lops, prog);
+	bpf_link_init(&link->link, BPF_LINK_TYPE_STRUCT_OPS, &bpf_struct_ops_link_lops, prog,
+		      prog->expected_attach_type);
 
 	op_idx = prog->expected_attach_type;
 	err = bpf_struct_ops_prepare_trampoline(tlinks, link,
diff --git a/net/core/dev.c b/net/core/dev.c
index be97c440ecd..7969fddc94e 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -10364,7 +10364,8 @@ int bpf_xdp_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
 		goto unlock;
 	}
 
-	bpf_link_init(&link->link, BPF_LINK_TYPE_XDP, &bpf_xdp_link_lops, prog);
+	bpf_link_init(&link->link, BPF_LINK_TYPE_XDP, &bpf_xdp_link_lops, prog,
+		      attr->link_create.attach_type);
 	link->dev = dev;
 	link->flags = attr->link_create.flags;
 
diff --git a/net/core/sock_map.c b/net/core/sock_map.c
index 82a14f131d0..fbe9a33ddf1 100644
--- a/net/core/sock_map.c
+++ b/net/core/sock_map.c
@@ -1866,7 +1866,8 @@ int sock_map_link_create(const union bpf_attr *attr, struct bpf_prog *prog)
 	}
 
 	attach_type = attr->link_create.attach_type;
-	bpf_link_init(&sockmap_link->link, BPF_LINK_TYPE_SOCKMAP, &sock_map_link_ops, prog);
+	bpf_link_init(&sockmap_link->link, BPF_LINK_TYPE_SOCKMAP, &sock_map_link_ops, prog,
+		      attach_type);
 	sockmap_link->map = map;
 	sockmap_link->attach_type = attach_type;
 
diff --git a/net/netfilter/nf_bpf_link.c b/net/netfilter/nf_bpf_link.c
index 06b08484470..a054d3b216d 100644
--- a/net/netfilter/nf_bpf_link.c
+++ b/net/netfilter/nf_bpf_link.c
@@ -225,7 +225,8 @@ int bpf_nf_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
 	if (!link)
 		return -ENOMEM;
 
-	bpf_link_init(&link->link, BPF_LINK_TYPE_NETFILTER, &bpf_nf_link_lops, prog);
+	bpf_link_init(&link->link, BPF_LINK_TYPE_NETFILTER, &bpf_nf_link_lops, prog,
+		      attr->link_create.attach_type);
 
 	link->hook_ops.hook = nf_hook_run_bpf;
 	link->hook_ops.hook_ops_type = NF_HOOK_OP_BPF;
-- 
2.48.1

Re: [PATCH bpf-next v3 1/7] bpf: Add attach_type in bpf_link

[Daniel Borkmann]

On 7/9/25 5:07 AM, Tao Chen wrote:
> Attach_type will be set when link created from user, it is better
> to record attach_type in bpf_link directly suggested by Andrii. So
> add the attach_type field in bpf_link and move the sleepable field to
> the end just to fill the byte hole.
> 
> Acked-by: Jiri Olsa <jolsa@kernel.org>
> Signed-off-by: Tao Chen <chen.dylane@linux.dev>
> ---
>   drivers/net/netkit.c           |  2 +-
>   include/linux/bpf.h            | 28 ++++++++++++++++-----------
>   kernel/bpf/bpf_iter.c          |  3 ++-
>   kernel/bpf/bpf_struct_ops.c    |  5 +++--
>   kernel/bpf/cgroup.c            |  4 ++--
>   kernel/bpf/net_namespace.c     |  2 +-
>   kernel/bpf/syscall.c           | 35 +++++++++++++++++++++-------------
>   kernel/bpf/tcx.c               |  3 ++-
>   kernel/bpf/trampoline.c        | 10 ++++++----
>   kernel/trace/bpf_trace.c       |  4 ++--
>   net/bpf/bpf_dummy_struct_ops.c |  3 ++-
>   net/core/dev.c                 |  3 ++-
>   net/core/sock_map.c            |  3 ++-
>   net/netfilter/nf_bpf_link.c    |  3 ++-
>   14 files changed, 66 insertions(+), 42 deletions(-)
[...]

> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index 34dd90ec7fa..dd5070039de 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -1729,12 +1729,10 @@ struct bpf_link {
>   	enum bpf_link_type type;
>   	const struct bpf_link_ops *ops;
>   	struct bpf_prog *prog;
> -	/* whether BPF link itself has "sleepable" semantics, which can differ
> -	 * from underlying BPF program having a "sleepable" semantics, as BPF
> -	 * link's semantics is determined by target attach hook
> -	 */
> -	bool sleepable;
> +
>   	u32 flags;
> +	enum bpf_attach_type attach_type;
> +
>   	/* rcu is used before freeing, work can be used to schedule that
>   	 * RCU-based freeing before that, so they never overlap
>   	 */
> @@ -1742,6 +1740,11 @@ struct bpf_link {
>   		struct rcu_head rcu;
>   		struct work_struct work;
>   	};
> +	/* whether BPF link itself has "sleepable" semantics, which can differ
> +	 * from underlying BPF program having a "sleepable" semantics, as BPF
> +	 * link's semantics is determined by target attach hook
> +	 */
> +	bool sleepable;
>   };

lgtm, it looks a bit weird at the end here after rcu/work but in future if other
attributes get added to the link obj and a new hole frees up we can move it up
again. Definitely good the rationale for the move is documented in the commit msg.

Acked-by: Daniel Borkmann <daniel@iogearbox.net>

[PATCH bpf-next v3 2/7] bpf: Remove attach_type in bpf_cgroup_link

[Tao Chen]

Use attach_type in bpf_link, and remove it in bpf_cgroup_link.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Tao Chen <chen.dylane@linux.dev>
---
 include/linux/bpf-cgroup.h |  1 -
 kernel/bpf/cgroup.c        | 13 ++++++-------
 2 files changed, 6 insertions(+), 8 deletions(-)

diff --git a/include/linux/bpf-cgroup.h b/include/linux/bpf-cgroup.h
index 70c8b94e797..082ccd8ad96 100644
--- a/include/linux/bpf-cgroup.h
+++ b/include/linux/bpf-cgroup.h
@@ -103,7 +103,6 @@ struct bpf_cgroup_storage {
 struct bpf_cgroup_link {
 	struct bpf_link link;
 	struct cgroup *cgroup;
-	enum bpf_attach_type type;
 };
 
 struct bpf_prog_list {
diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
index bacdd0ca741..72c8b50dca0 100644
--- a/kernel/bpf/cgroup.c
+++ b/kernel/bpf/cgroup.c
@@ -984,7 +984,7 @@ static int __cgroup_bpf_replace(struct cgroup *cgrp,
 	struct hlist_head *progs;
 	bool found = false;
 
-	atype = bpf_cgroup_atype_find(link->type, new_prog->aux->attach_btf_id);
+	atype = bpf_cgroup_atype_find(link->link.attach_type, new_prog->aux->attach_btf_id);
 	if (atype < 0)
 		return -EINVAL;
 
@@ -1396,8 +1396,8 @@ static void bpf_cgroup_link_release(struct bpf_link *link)
 	}
 
 	WARN_ON(__cgroup_bpf_detach(cg_link->cgroup, NULL, cg_link,
-				    cg_link->type, 0));
-	if (cg_link->type == BPF_LSM_CGROUP)
+				    link->attach_type, 0));
+	if (link->attach_type == BPF_LSM_CGROUP)
 		bpf_trampoline_unlink_cgroup_shim(cg_link->link.prog);
 
 	cg = cg_link->cgroup;
@@ -1439,7 +1439,7 @@ static void bpf_cgroup_link_show_fdinfo(const struct bpf_link *link,
 		   "cgroup_id:\t%llu\n"
 		   "attach_type:\t%d\n",
 		   cg_id,
-		   cg_link->type);
+		   link->attach_type);
 }
 
 static int bpf_cgroup_link_fill_link_info(const struct bpf_link *link,
@@ -1455,7 +1455,7 @@ static int bpf_cgroup_link_fill_link_info(const struct bpf_link *link,
 	cgroup_unlock();
 
 	info->cgroup.cgroup_id = cg_id;
-	info->cgroup.attach_type = cg_link->type;
+	info->cgroup.attach_type = link->attach_type;
 	return 0;
 }
 
@@ -1497,7 +1497,6 @@ int cgroup_bpf_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
 	bpf_link_init(&link->link, BPF_LINK_TYPE_CGROUP, &bpf_cgroup_link_lops,
 		      prog, attr->link_create.attach_type);
 	link->cgroup = cgrp;
-	link->type = attr->link_create.attach_type;
 
 	err = bpf_link_prime(&link->link, &link_primer);
 	if (err) {
@@ -1506,7 +1505,7 @@ int cgroup_bpf_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
 	}
 
 	err = cgroup_bpf_attach(cgrp, NULL, NULL, link,
-				link->type, BPF_F_ALLOW_MULTI | attr->link_create.flags,
+				link->link.attach_type, BPF_F_ALLOW_MULTI | attr->link_create.flags,
 				attr->link_create.cgroup.relative_fd,
 				attr->link_create.cgroup.expected_revision);
 	if (err) {
-- 
2.48.1

Re: [PATCH bpf-next v3 2/7] bpf: Remove attach_type in bpf_cgroup_link

[Daniel Borkmann]

On 7/9/25 5:07 AM, Tao Chen wrote:
> Use attach_type in bpf_link, and remove it in bpf_cgroup_link.
> 
> Acked-by: Jiri Olsa <jolsa@kernel.org>
> Signed-off-by: Tao Chen <chen.dylane@linux.dev>
> ---
>   include/linux/bpf-cgroup.h |  1 -
>   kernel/bpf/cgroup.c        | 13 ++++++-------
>   2 files changed, 6 insertions(+), 8 deletions(-)

Acked-by: Daniel Borkmann <daniel@iogearbox.net>

[PATCH bpf-next v3 3/7] bpf: Remove attach_type in sockmap_link

[Tao Chen]

Use attach_type in bpf_link, and remove it in sockmap_link.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Tao Chen <chen.dylane@linux.dev>
---
 net/core/sock_map.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/net/core/sock_map.c b/net/core/sock_map.c
index fbe9a33ddf1..5947b38e4f8 100644
--- a/net/core/sock_map.c
+++ b/net/core/sock_map.c
@@ -1709,7 +1709,6 @@ EXPORT_SYMBOL_GPL(sock_map_close);
 struct sockmap_link {
 	struct bpf_link link;
 	struct bpf_map *map;
-	enum bpf_attach_type attach_type;
 };
 
 static void sock_map_link_release(struct bpf_link *link)
@@ -1721,7 +1720,7 @@ static void sock_map_link_release(struct bpf_link *link)
 		goto out;
 
 	WARN_ON_ONCE(sock_map_prog_update(sockmap_link->map, NULL, link->prog, link,
-					  sockmap_link->attach_type));
+					  link->attach_type));
 
 	bpf_map_put_with_uref(sockmap_link->map);
 	sockmap_link->map = NULL;
@@ -1772,7 +1771,7 @@ static int sock_map_link_update_prog(struct bpf_link *link,
 	}
 
 	ret = sock_map_prog_link_lookup(sockmap_link->map, &pprog, &plink,
-					sockmap_link->attach_type);
+					link->attach_type);
 	if (ret)
 		goto out;
 
@@ -1817,7 +1816,7 @@ static int sock_map_link_fill_info(const struct bpf_link *link,
 	u32 map_id = sock_map_link_get_map_id(sockmap_link);
 
 	info->sockmap.map_id = map_id;
-	info->sockmap.attach_type = sockmap_link->attach_type;
+	info->sockmap.attach_type = link->attach_type;
 	return 0;
 }
 
@@ -1828,7 +1827,7 @@ static void sock_map_link_show_fdinfo(const struct bpf_link *link,
 	u32 map_id = sock_map_link_get_map_id(sockmap_link);
 
 	seq_printf(seq, "map_id:\t%u\n", map_id);
-	seq_printf(seq, "attach_type:\t%u\n", sockmap_link->attach_type);
+	seq_printf(seq, "attach_type:\t%u\n", link->attach_type);
 }
 
 static const struct bpf_link_ops sock_map_link_ops = {
@@ -1869,7 +1868,6 @@ int sock_map_link_create(const union bpf_attr *attr, struct bpf_prog *prog)
 	bpf_link_init(&sockmap_link->link, BPF_LINK_TYPE_SOCKMAP, &sock_map_link_ops, prog,
 		      attach_type);
 	sockmap_link->map = map;
-	sockmap_link->attach_type = attach_type;
 
 	ret = bpf_link_prime(&sockmap_link->link, &link_primer);
 	if (ret) {
-- 
2.48.1

Re: [PATCH bpf-next v3 3/7] bpf: Remove attach_type in sockmap_link

[Daniel Borkmann]

On 7/9/25 5:07 AM, Tao Chen wrote:
> Use attach_type in bpf_link, and remove it in sockmap_link.
> 
> Acked-by: Jiri Olsa <jolsa@kernel.org>
> Signed-off-by: Tao Chen <chen.dylane@linux.dev>

Acked-by: Daniel Borkmann <daniel@iogearbox.net>

[PATCH bpf-next v3 4/7] bpf: Remove location field in tcx_link

[Tao Chen]

Use attach_type in bpf_link to replace the location filed, and
remove location field in tcx_link.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Tao Chen <chen.dylane@linux.dev>
---
 include/net/tcx.h |  1 -
 kernel/bpf/tcx.c  | 13 ++++++-------
 2 files changed, 6 insertions(+), 8 deletions(-)

diff --git a/include/net/tcx.h b/include/net/tcx.h
index 5ce0ce9e0c0..23a61af1354 100644
--- a/include/net/tcx.h
+++ b/include/net/tcx.h
@@ -20,7 +20,6 @@ struct tcx_entry {
 struct tcx_link {
 	struct bpf_link link;
 	struct net_device *dev;
-	u32 location;
 };
 
 static inline void tcx_set_ingress(struct sk_buff *skb, bool ingress)
diff --git a/kernel/bpf/tcx.c b/kernel/bpf/tcx.c
index e6a14f408d9..efd987ea687 100644
--- a/kernel/bpf/tcx.c
+++ b/kernel/bpf/tcx.c
@@ -142,7 +142,7 @@ static int tcx_link_prog_attach(struct bpf_link *link, u32 flags, u32 id_or_fd,
 				u64 revision)
 {
 	struct tcx_link *tcx = tcx_link(link);
-	bool created, ingress = tcx->location == BPF_TCX_INGRESS;
+	bool created, ingress = link->attach_type == BPF_TCX_INGRESS;
 	struct bpf_mprog_entry *entry, *entry_new;
 	struct net_device *dev = tcx->dev;
 	int ret;
@@ -169,7 +169,7 @@ static int tcx_link_prog_attach(struct bpf_link *link, u32 flags, u32 id_or_fd,
 static void tcx_link_release(struct bpf_link *link)
 {
 	struct tcx_link *tcx = tcx_link(link);
-	bool ingress = tcx->location == BPF_TCX_INGRESS;
+	bool ingress = link->attach_type == BPF_TCX_INGRESS;
 	struct bpf_mprog_entry *entry, *entry_new;
 	struct net_device *dev;
 	int ret = 0;
@@ -204,7 +204,7 @@ static int tcx_link_update(struct bpf_link *link, struct bpf_prog *nprog,
 			   struct bpf_prog *oprog)
 {
 	struct tcx_link *tcx = tcx_link(link);
-	bool ingress = tcx->location == BPF_TCX_INGRESS;
+	bool ingress = link->attach_type == BPF_TCX_INGRESS;
 	struct bpf_mprog_entry *entry, *entry_new;
 	struct net_device *dev;
 	int ret = 0;
@@ -260,8 +260,8 @@ static void tcx_link_fdinfo(const struct bpf_link *link, struct seq_file *seq)
 
 	seq_printf(seq, "ifindex:\t%u\n", ifindex);
 	seq_printf(seq, "attach_type:\t%u (%s)\n",
-		   tcx->location,
-		   tcx->location == BPF_TCX_INGRESS ? "ingress" : "egress");
+		   link->attach_type,
+		   link->attach_type == BPF_TCX_INGRESS ? "ingress" : "egress");
 }
 
 static int tcx_link_fill_info(const struct bpf_link *link,
@@ -276,7 +276,7 @@ static int tcx_link_fill_info(const struct bpf_link *link,
 	rtnl_unlock();
 
 	info->tcx.ifindex = ifindex;
-	info->tcx.attach_type = tcx->location;
+	info->tcx.attach_type = link->attach_type;
 	return 0;
 }
 
@@ -303,7 +303,6 @@ static int tcx_link_init(struct tcx_link *tcx,
 {
 	bpf_link_init(&tcx->link, BPF_LINK_TYPE_TCX, &tcx_link_lops, prog,
 		      attr->link_create.attach_type);
-	tcx->location = attr->link_create.attach_type;
 	tcx->dev = dev;
 	return bpf_link_prime(&tcx->link, link_primer);
 }
-- 
2.48.1

Re: [PATCH bpf-next v3 4/7] bpf: Remove location field in tcx_link

[Daniel Borkmann]

On 7/9/25 5:07 AM, Tao Chen wrote:
> Use attach_type in bpf_link to replace the location filed, and
> remove location field in tcx_link.
> 
> Acked-by: Jiri Olsa <jolsa@kernel.org>
> Signed-off-by: Tao Chen <chen.dylane@linux.dev>

Acked-by: Daniel Borkmann <daniel@iogearbox.net>

[PATCH bpf-next v3 5/7] bpf: Remove attach_type in bpf_netns_link

[Tao Chen]

Use attach_type in bpf_link, and remove it in bpf_netns_link.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Tao Chen <chen.dylane@linux.dev>
---
 kernel/bpf/net_namespace.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/kernel/bpf/net_namespace.c b/kernel/bpf/net_namespace.c
index 63702c86275..6d27bd97c95 100644
--- a/kernel/bpf/net_namespace.c
+++ b/kernel/bpf/net_namespace.c
@@ -11,7 +11,6 @@
 
 struct bpf_netns_link {
 	struct bpf_link	link;
-	enum bpf_attach_type type;
 	enum netns_bpf_attach_type netns_type;
 
 	/* We don't hold a ref to net in order to auto-detach the link
@@ -216,7 +215,7 @@ static int bpf_netns_link_fill_info(const struct bpf_link *link,
 	mutex_unlock(&netns_bpf_mutex);
 
 	info->netns.netns_ino = inum;
-	info->netns.attach_type = net_link->type;
+	info->netns.attach_type = link->attach_type;
 	return 0;
 }
 
@@ -230,7 +229,7 @@ static void bpf_netns_link_show_fdinfo(const struct bpf_link *link,
 		   "netns_ino:\t%u\n"
 		   "attach_type:\t%u\n",
 		   info.netns.netns_ino,
-		   info.netns.attach_type);
+		   link->attach_type);
 }
 
 static const struct bpf_link_ops bpf_netns_link_ops = {
@@ -503,7 +502,6 @@ int netns_bpf_link_create(const union bpf_attr *attr, struct bpf_prog *prog)
 	bpf_link_init(&net_link->link, BPF_LINK_TYPE_NETNS,
 		      &bpf_netns_link_ops, prog, type);
 	net_link->net = net;
-	net_link->type = type;
 	net_link->netns_type = netns_type;
 
 	err = bpf_link_prime(&net_link->link, &link_primer);
-- 
2.48.1

Re: [PATCH bpf-next v3 5/7] bpf: Remove attach_type in bpf_netns_link

[Jakub Sitnicki]

On Wed, Jul 09, 2025 at 11:08 AM +08, Tao Chen wrote:
> Use attach_type in bpf_link, and remove it in bpf_netns_link.
>
> Acked-by: Jiri Olsa <jolsa@kernel.org>
> Signed-off-by: Tao Chen <chen.dylane@linux.dev>
> ---
>  kernel/bpf/net_namespace.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
>
> diff --git a/kernel/bpf/net_namespace.c b/kernel/bpf/net_namespace.c
> index 63702c86275..6d27bd97c95 100644
> --- a/kernel/bpf/net_namespace.c
> +++ b/kernel/bpf/net_namespace.c
> @@ -11,7 +11,6 @@
>  
>  struct bpf_netns_link {
>  	struct bpf_link	link;
> -	enum bpf_attach_type type;
>  	enum netns_bpf_attach_type netns_type;
>  
>  	/* We don't hold a ref to net in order to auto-detach the link

Nit: Doesn't that create a hole? Maybe move netns_type to the end.

[...]

Re: [PATCH bpf-next v3 5/7] bpf: Remove attach_type in bpf_netns_link

[Tao Chen]

在 2025/7/9 16:21, Jakub Sitnicki 写道:
> On Wed, Jul 09, 2025 at 11:08 AM +08, Tao Chen wrote:
>> Use attach_type in bpf_link, and remove it in bpf_netns_link.
>>
>> Acked-by: Jiri Olsa <jolsa@kernel.org>
>> Signed-off-by: Tao Chen <chen.dylane@linux.dev>
>> ---
>>   kernel/bpf/net_namespace.c | 6 ++----
>>   1 file changed, 2 insertions(+), 4 deletions(-)
>>
>> diff --git a/kernel/bpf/net_namespace.c b/kernel/bpf/net_namespace.c
>> index 63702c86275..6d27bd97c95 100644
>> --- a/kernel/bpf/net_namespace.c
>> +++ b/kernel/bpf/net_namespace.c
>> @@ -11,7 +11,6 @@
>>   
>>   struct bpf_netns_link {
>>   	struct bpf_link	link;
>> -	enum bpf_attach_type type;
>>   	enum netns_bpf_attach_type netns_type;
>>   
>>   	/* We don't hold a ref to net in order to auto-detach the link
> 
> Nit: Doesn't that create a hole? Maybe move netns_type to the end.
> 

Hi Jakub,
You are right, i will change it in v4, thanks.

pahole -C bpf_netns_link vmlinux
struct bpf_netns_link {
	struct bpf_link            link __attribute__((__aligned__(8))); /* 
0    80 */

	/* XXX last struct has 7 bytes of padding */

	/* --- cacheline 1 boundary (64 bytes) was 16 bytes ago --- */
	enum netns_bpf_attach_type netns_type;           /*    80     4 */

	/* XXX 4 bytes hole, try to pack */

	struct net *               net;                  /*    88     8 */
	struct list_head           node;                 /*    96    16 */

	/* size: 112, cachelines: 2, members: 4 */
	/* sum members: 108, holes: 1, sum holes: 4 */
	/* paddings: 1, sum paddings: 7 */
	/* forced alignments: 1 */
	/* last cacheline: 48 bytes */
} __attribute__((__aligned__(8)));

> [...]


-- 
Best Regards
Tao Chen

[PATCH bpf-next v3 6/7] bpf: Remove attach_type in bpf_tracing_link

[Tao Chen]

Use attach_type in bpf_link, and remove it in bpf_tracing_link.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Tao Chen <chen.dylane@linux.dev>
---
 include/linux/bpf.h  | 1 -
 kernel/bpf/syscall.c | 5 ++---
 2 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index dd5070039de..976ae571522 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1784,7 +1784,6 @@ struct bpf_shim_tramp_link {
 
 struct bpf_tracing_link {
 	struct bpf_tramp_link link;
-	enum bpf_attach_type attach_type;
 	struct bpf_trampoline *trampoline;
 	struct bpf_prog *tgt_prog;
 };
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 14883b3040a..bed523bf92c 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3414,7 +3414,7 @@ static void bpf_tracing_link_show_fdinfo(const struct bpf_link *link,
 		   "target_obj_id:\t%u\n"
 		   "target_btf_id:\t%u\n"
 		   "cookie:\t%llu\n",
-		   tr_link->attach_type,
+		   link->attach_type,
 		   target_obj_id,
 		   target_btf_id,
 		   tr_link->link.cookie);
@@ -3426,7 +3426,7 @@ static int bpf_tracing_link_fill_link_info(const struct bpf_link *link,
 	struct bpf_tracing_link *tr_link =
 		container_of(link, struct bpf_tracing_link, link.link);
 
-	info->tracing.attach_type = tr_link->attach_type;
+	info->tracing.attach_type = link->attach_type;
 	info->tracing.cookie = tr_link->link.cookie;
 	bpf_trampoline_unpack_key(tr_link->trampoline->key,
 				  &info->tracing.target_obj_id,
@@ -3516,7 +3516,6 @@ static int bpf_tracing_prog_attach(struct bpf_prog *prog,
 	bpf_link_init(&link->link.link, BPF_LINK_TYPE_TRACING,
 		      &bpf_tracing_link_lops, prog, attach_type);
 
-	link->attach_type = prog->expected_attach_type;
 	link->link.cookie = bpf_cookie;
 
 	mutex_lock(&prog->aux->dst_mutex);
-- 
2.48.1

[PATCH bpf-next v3 7/7] netkit: Remove location field in netkit_link

[Tao Chen]

Use attach_type in bpf_link to replace the location field, and
remove location field in netkit_link.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Tao Chen <chen.dylane@linux.dev>
---
 drivers/net/netkit.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/drivers/net/netkit.c b/drivers/net/netkit.c
index 5928c99eac7..492be60f2e7 100644
--- a/drivers/net/netkit.c
+++ b/drivers/net/netkit.c
@@ -32,7 +32,6 @@ struct netkit {
 struct netkit_link {
 	struct bpf_link link;
 	struct net_device *dev;
-	u32 location;
 };
 
 static __always_inline int
@@ -733,8 +732,8 @@ static void netkit_link_fdinfo(const struct bpf_link *link, struct seq_file *seq
 
 	seq_printf(seq, "ifindex:\t%u\n", ifindex);
 	seq_printf(seq, "attach_type:\t%u (%s)\n",
-		   nkl->location,
-		   nkl->location == BPF_NETKIT_PRIMARY ? "primary" : "peer");
+		   link->attach_type,
+		   link->attach_type == BPF_NETKIT_PRIMARY ? "primary" : "peer");
 }
 
 static int netkit_link_fill_info(const struct bpf_link *link,
@@ -749,7 +748,7 @@ static int netkit_link_fill_info(const struct bpf_link *link,
 	rtnl_unlock();
 
 	info->netkit.ifindex = ifindex;
-	info->netkit.attach_type = nkl->location;
+	info->netkit.attach_type = link->attach_type;
 	return 0;
 }
 
@@ -776,7 +775,6 @@ static int netkit_link_init(struct netkit_link *nkl,
 {
 	bpf_link_init(&nkl->link, BPF_LINK_TYPE_NETKIT,
 		      &netkit_link_lops, prog, attr->link_create.attach_type);
-	nkl->location = attr->link_create.attach_type;
 	nkl->dev = dev;
 	return bpf_link_prime(&nkl->link, link_primer);
 }
-- 
2.48.1

Re: [PATCH bpf-next v3 7/7] netkit: Remove location field in netkit_link

[Nikolay Aleksandrov]

On 7/9/25 06:08, Tao Chen wrote:
> Use attach_type in bpf_link to replace the location field, and
> remove location field in netkit_link.
> 
> Acked-by: Jiri Olsa <jolsa@kernel.org>
> Signed-off-by: Tao Chen <chen.dylane@linux.dev>
> ---
>  drivers/net/netkit.c | 8 +++-----
>  1 file changed, 3 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/net/netkit.c b/drivers/net/netkit.c
> index 5928c99eac7..492be60f2e7 100644
> --- a/drivers/net/netkit.c
> +++ b/drivers/net/netkit.c
> @@ -32,7 +32,6 @@ struct netkit {
>  struct netkit_link {
>  	struct bpf_link link;
>  	struct net_device *dev;
> -	u32 location;
>  };
>  
>  static __always_inline int
> @@ -733,8 +732,8 @@ static void netkit_link_fdinfo(const struct bpf_link *link, struct seq_file *seq
>  
>  	seq_printf(seq, "ifindex:\t%u\n", ifindex);
>  	seq_printf(seq, "attach_type:\t%u (%s)\n",
> -		   nkl->location,
> -		   nkl->location == BPF_NETKIT_PRIMARY ? "primary" : "peer");
> +		   link->attach_type,
> +		   link->attach_type == BPF_NETKIT_PRIMARY ? "primary" : "peer");
>  }
>  
>  static int netkit_link_fill_info(const struct bpf_link *link,
> @@ -749,7 +748,7 @@ static int netkit_link_fill_info(const struct bpf_link *link,
>  	rtnl_unlock();
>  
>  	info->netkit.ifindex = ifindex;
> -	info->netkit.attach_type = nkl->location;
> +	info->netkit.attach_type = link->attach_type;
>  	return 0;
>  }
>  
> @@ -776,7 +775,6 @@ static int netkit_link_init(struct netkit_link *nkl,
>  {
>  	bpf_link_init(&nkl->link, BPF_LINK_TYPE_NETKIT,
>  		      &netkit_link_lops, prog, attr->link_create.attach_type);
> -	nkl->location = attr->link_create.attach_type;
>  	nkl->dev = dev;
>  	return bpf_link_prime(&nkl->link, link_primer);
>  }

LGTM for netkit,
Acked-by: Nikolay Aleksandrov <razor@blackwall.org>

Re: [PATCH bpf-next v3 7/7] netkit: Remove location field in netkit_link

[Daniel Borkmann]

On 7/9/25 5:08 AM, Tao Chen wrote:
> Use attach_type in bpf_link to replace the location field, and
> remove location field in netkit_link.
> 
> Acked-by: Jiri Olsa <jolsa@kernel.org>
> Signed-off-by: Tao Chen <chen.dylane@linux.dev>

Acked-by: Daniel Borkmann <daniel@iogearbox.net>

Re: [PATCH bpf-next v3 0/7] Add attach_type in bpf_link

[Jakub Sitnicki]

On Wed, Jul 09, 2025 at 11:07 AM +08, Tao Chen wrote:
> Andrii suggested moving the attach_type into bpf_link, the previous discussion
> is as follows:
> https://lore.kernel.org/bpf/CAEf4BzY7TZRjxpCJM-+LYgEqe23YFj5Uv3isb7gat2-HU4OSng@mail.gmail.com
>
> patch1 add attach_type in bpf_link, and pass it to bpf_link_init, which
> will init the attach_type field.
>
> patch2-7 remove the attach_type in struct bpf_xx_link, update the info
> with bpf_link attach_type.
>
> There are some functions finally call bpf_link_init but do not have bpf_attr
> from user or do not need to init attach_type from user like bpf_raw_tracepoint_open,
> now use prog->expected_attach_type to init attach_type.
>
> bpf_struct_ops_map_update_elem
> bpf_raw_tracepoint_open
> bpf_struct_ops_test_run
>
> Feedback of any kind is welcome, thanks.
>
> Tao Chen (7):
>   bpf: Add attach_type in bpf_link
>   bpf: Remove attach_type in bpf_cgroup_link
>   bpf: Remove attach_type in sockmap_link
>   bpf: Remove location field in tcx_link
>   bpf: Remove attach_type in bpf_netns_link
>   bpf: Remove attach_type in bpf_tracing_link
>   netkit: Remove location field in netkit_link

For the series:

Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>