[PATCH v6 1/5] bpf: Introduce the bpf_list_del kfunc.

Chengkaitao posted 5 patches 1 month, 1 week ago
There is a newer version of this series
[PATCH v6 1/5] bpf: Introduce the bpf_list_del kfunc.
Posted by Chengkaitao 1 month, 1 week ago
From: Kaitao Cheng <chengkaitao@kylinos.cn>

If a user holds ownership of a node in the middle of a list, they
can directly remove it from the list without strictly adhering to
deletion rules from the head or tail.

We have added an additional parameter bpf_list_head *head to
bpf_list_del, as the verifier requires the head parameter to
check whether the lock is being held.

This is typically paired with bpf_refcount. After calling
bpf_list_del, it is generally necessary to drop the reference to
the list node twice to prevent reference count leaks.

Signed-off-by: Kaitao Cheng <chengkaitao@kylinos.cn>
---
 kernel/bpf/helpers.c  | 27 +++++++++++++++++++++++++++
 kernel/bpf/verifier.c |  6 +++++-
 2 files changed, 32 insertions(+), 1 deletion(-)

diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
index 6eb6c82ed2ee..cc1a096a1f64 100644
--- a/kernel/bpf/helpers.c
+++ b/kernel/bpf/helpers.c
@@ -2459,6 +2459,32 @@ __bpf_kfunc struct bpf_list_node *bpf_list_pop_back(struct bpf_list_head *head)
 	return __bpf_list_del(head, true);
 }
 
+__bpf_kfunc struct bpf_list_node *bpf_list_del(struct bpf_list_head *head,
+					       struct bpf_list_node *node)
+{
+	struct bpf_list_node_kern *knode = (struct bpf_list_node_kern *)node;
+	struct list_head *h = (void *)head;
+
+	/* If list_head was 0-initialized by map, bpf_obj_init_field wasn't
+	 * called on its fields, so init here
+	 */
+	if (unlikely(!h->next)) {
+		INIT_LIST_HEAD(h);
+		return NULL;
+	}
+
+	if (unlikely(!knode))
+		return NULL;
+
+	if (WARN_ON_ONCE(READ_ONCE(knode->owner) != head))
+		return NULL;
+
+	list_del_init(&knode->list_head);
+	WRITE_ONCE(knode->owner, NULL);
+
+	return node;
+}
+
 __bpf_kfunc struct bpf_list_node *bpf_list_front(struct bpf_list_head *head)
 {
 	struct list_head *h = (struct list_head *)head;
@@ -4545,6 +4571,7 @@ BTF_ID_FLAGS(func, bpf_list_push_front_impl)
 BTF_ID_FLAGS(func, bpf_list_push_back_impl)
 BTF_ID_FLAGS(func, bpf_list_pop_front, KF_ACQUIRE | KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_list_pop_back, KF_ACQUIRE | KF_RET_NULL)
+BTF_ID_FLAGS(func, bpf_list_del, KF_ACQUIRE | KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_list_front, KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_list_back, KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_task_acquire, KF_ACQUIRE | KF_RCU | KF_RET_NULL)
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 67c09b43a497..c9557d3fb8dd 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -12461,6 +12461,7 @@ enum special_kfunc_type {
 	KF_bpf_list_push_back_impl,
 	KF_bpf_list_pop_front,
 	KF_bpf_list_pop_back,
+	KF_bpf_list_del,
 	KF_bpf_list_front,
 	KF_bpf_list_back,
 	KF_bpf_cast_to_kern_ctx,
@@ -12521,6 +12522,7 @@ BTF_ID(func, bpf_list_push_front_impl)
 BTF_ID(func, bpf_list_push_back_impl)
 BTF_ID(func, bpf_list_pop_front)
 BTF_ID(func, bpf_list_pop_back)
+BTF_ID(func, bpf_list_del)
 BTF_ID(func, bpf_list_front)
 BTF_ID(func, bpf_list_back)
 BTF_ID(func, bpf_cast_to_kern_ctx)
@@ -12996,6 +12998,7 @@ static bool is_bpf_list_api_kfunc(u32 btf_id)
 	       btf_id == special_kfunc_list[KF_bpf_list_push_back_impl] ||
 	       btf_id == special_kfunc_list[KF_bpf_list_pop_front] ||
 	       btf_id == special_kfunc_list[KF_bpf_list_pop_back] ||
+	       btf_id == special_kfunc_list[KF_bpf_list_del] ||
 	       btf_id == special_kfunc_list[KF_bpf_list_front] ||
 	       btf_id == special_kfunc_list[KF_bpf_list_back];
 }
@@ -13118,7 +13121,8 @@ static bool check_kfunc_is_graph_node_api(struct bpf_verifier_env *env,
 	switch (node_field_type) {
 	case BPF_LIST_NODE:
 		ret = (kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_front_impl] ||
-		       kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_back_impl]);
+		       kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_back_impl] ||
+		       kfunc_btf_id == special_kfunc_list[KF_bpf_list_del]);
 		break;
 	case BPF_RB_NODE:
 		ret = (kfunc_btf_id == special_kfunc_list[KF_bpf_rbtree_remove] ||
-- 
2.50.1 (Apple Git-155)
Re: [PATCH v6 1/5] bpf: Introduce the bpf_list_del kfunc.
Posted by Kumar Kartikeya Dwivedi 1 month, 1 week ago
On Wed, 4 Mar 2026 at 15:50, Chengkaitao <pilgrimtao@gmail.com> wrote:
>
> From: Kaitao Cheng <chengkaitao@kylinos.cn>
>
> If a user holds ownership of a node in the middle of a list, they
> can directly remove it from the list without strictly adhering to
> deletion rules from the head or tail.
>
> We have added an additional parameter bpf_list_head *head to
> bpf_list_del, as the verifier requires the head parameter to
> check whether the lock is being held.
>
> This is typically paired with bpf_refcount. After calling
> bpf_list_del, it is generally necessary to drop the reference to
> the list node twice to prevent reference count leaks.
>
> Signed-off-by: Kaitao Cheng <chengkaitao@kylinos.cn>
> ---
>  kernel/bpf/helpers.c  | 27 +++++++++++++++++++++++++++
>  kernel/bpf/verifier.c |  6 +++++-
>  2 files changed, 32 insertions(+), 1 deletion(-)
>
> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index 6eb6c82ed2ee..cc1a096a1f64 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c
> @@ -2459,6 +2459,32 @@ __bpf_kfunc struct bpf_list_node *bpf_list_pop_back(struct bpf_list_head *head)
>         return __bpf_list_del(head, true);
>  }
>
> +__bpf_kfunc struct bpf_list_node *bpf_list_del(struct bpf_list_head *head,
> +                                              struct bpf_list_node *node)
> +{
> +       struct bpf_list_node_kern *knode = (struct bpf_list_node_kern *)node;
> +       struct list_head *h = (void *)head;
> +
> +       /* If list_head was 0-initialized by map, bpf_obj_init_field wasn't
> +        * called on its fields, so init here
> +        */
> +       if (unlikely(!h->next)) {
> +               INIT_LIST_HEAD(h);
> +               return NULL;
> +       }

nit: I would avoid doing this here, just for symmetry.
If we are part of the list after the owner checks, this should be
initialized correctly anyway.

> +
> +       if (unlikely(!knode))
> +               return NULL;

nit: It seems like node can't (shouldn't) be NULL, so we can lose this check.

> +
> +       if (WARN_ON_ONCE(READ_ONCE(knode->owner) != head))
> +               return NULL;
> +
> +       list_del_init(&knode->list_head);
> +       WRITE_ONCE(knode->owner, NULL);
> +
> +       return node;
> +}
> +
>  __bpf_kfunc struct bpf_list_node *bpf_list_front(struct bpf_list_head *head)
>  {
>         struct list_head *h = (struct list_head *)head;
> @@ -4545,6 +4571,7 @@ BTF_ID_FLAGS(func, bpf_list_push_front_impl)
>  BTF_ID_FLAGS(func, bpf_list_push_back_impl)
>  BTF_ID_FLAGS(func, bpf_list_pop_front, KF_ACQUIRE | KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_list_pop_back, KF_ACQUIRE | KF_RET_NULL)
> +BTF_ID_FLAGS(func, bpf_list_del, KF_ACQUIRE | KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_list_front, KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_list_back, KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_task_acquire, KF_ACQUIRE | KF_RCU | KF_RET_NULL)
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 67c09b43a497..c9557d3fb8dd 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -12461,6 +12461,7 @@ enum special_kfunc_type {
>         KF_bpf_list_push_back_impl,
>         KF_bpf_list_pop_front,
>         KF_bpf_list_pop_back,
> +       KF_bpf_list_del,
>         KF_bpf_list_front,
>         KF_bpf_list_back,
>         KF_bpf_cast_to_kern_ctx,
> @@ -12521,6 +12522,7 @@ BTF_ID(func, bpf_list_push_front_impl)
>  BTF_ID(func, bpf_list_push_back_impl)
>  BTF_ID(func, bpf_list_pop_front)
>  BTF_ID(func, bpf_list_pop_back)
> +BTF_ID(func, bpf_list_del)
>  BTF_ID(func, bpf_list_front)
>  BTF_ID(func, bpf_list_back)
>  BTF_ID(func, bpf_cast_to_kern_ctx)
> @@ -12996,6 +12998,7 @@ static bool is_bpf_list_api_kfunc(u32 btf_id)
>                btf_id == special_kfunc_list[KF_bpf_list_push_back_impl] ||
>                btf_id == special_kfunc_list[KF_bpf_list_pop_front] ||
>                btf_id == special_kfunc_list[KF_bpf_list_pop_back] ||
> +              btf_id == special_kfunc_list[KF_bpf_list_del] ||
>                btf_id == special_kfunc_list[KF_bpf_list_front] ||
>                btf_id == special_kfunc_list[KF_bpf_list_back];
>  }
> @@ -13118,7 +13121,8 @@ static bool check_kfunc_is_graph_node_api(struct bpf_verifier_env *env,
>         switch (node_field_type) {
>         case BPF_LIST_NODE:
>                 ret = (kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_front_impl] ||
> -                      kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_back_impl]);
> +                      kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_back_impl] ||
> +                      kfunc_btf_id == special_kfunc_list[KF_bpf_list_del]);
>                 break;
>         case BPF_RB_NODE:
>                 ret = (kfunc_btf_id == special_kfunc_list[KF_bpf_rbtree_remove] ||
> --
> 2.50.1 (Apple Git-155)
>
>
Re: [PATCH v6 1/5] bpf: Introduce the bpf_list_del kfunc.
Posted by Kumar Kartikeya Dwivedi 1 month, 1 week ago
On Fri, 6 Mar 2026 at 01:14, Kumar Kartikeya Dwivedi <memxor@gmail.com> wrote:
>
> On Wed, 4 Mar 2026 at 15:50, Chengkaitao <pilgrimtao@gmail.com> wrote:
> >
> > From: Kaitao Cheng <chengkaitao@kylinos.cn>
> >
> > If a user holds ownership of a node in the middle of a list, they
> > can directly remove it from the list without strictly adhering to
> > deletion rules from the head or tail.
> >
> > We have added an additional parameter bpf_list_head *head to
> > bpf_list_del, as the verifier requires the head parameter to
> > check whether the lock is being held.
> >
> > This is typically paired with bpf_refcount. After calling
> > bpf_list_del, it is generally necessary to drop the reference to
> > the list node twice to prevent reference count leaks.
> >
> > Signed-off-by: Kaitao Cheng <chengkaitao@kylinos.cn>
> > ---
> >  kernel/bpf/helpers.c  | 27 +++++++++++++++++++++++++++
> >  kernel/bpf/verifier.c |  6 +++++-
> >  2 files changed, 32 insertions(+), 1 deletion(-)
> >
> > diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> > index 6eb6c82ed2ee..cc1a096a1f64 100644
> > --- a/kernel/bpf/helpers.c
> > +++ b/kernel/bpf/helpers.c
> > @@ -2459,6 +2459,32 @@ __bpf_kfunc struct bpf_list_node *bpf_list_pop_back(struct bpf_list_head *head)
> >         return __bpf_list_del(head, true);
> >  }
> >
> > +__bpf_kfunc struct bpf_list_node *bpf_list_del(struct bpf_list_head *head,
> > +                                              struct bpf_list_node *node)
> > +{
> > +       struct bpf_list_node_kern *knode = (struct bpf_list_node_kern *)node;
> > +       struct list_head *h = (void *)head;
> > +
> > +       /* If list_head was 0-initialized by map, bpf_obj_init_field wasn't
> > +        * called on its fields, so init here
> > +        */
> > +       if (unlikely(!h->next)) {
> > +               INIT_LIST_HEAD(h);
> > +               return NULL;
> > +       }
>
> nit: I would avoid doing this here, just for symmetry.
> If we are part of the list after the owner checks, this should be
> initialized correctly anyway.

Hmm, looks like I should take my ack back (just in principle, there's
no bug here).
I see you copied the comment and check from __bpf_list_del.
I was looking at rbtree_remove to see whether it did RB_CLEAR_NODE(n).

I think it's fine to keep this init, but let's try not repeating the
same logic twice.
I would make all 3 (pop_front, pop_back, list_del) use __bpf_list_del.
Just pass h->next, h->prev, or node depending on which one is being called.

>
> > +
> > +       if (unlikely(!knode))
> > +               return NULL;
>
> nit: It seems like node can't (shouldn't) be NULL, so we can lose this check.
>
> > +
> > +       if (WARN_ON_ONCE(READ_ONCE(knode->owner) != head))
> > +               return NULL;
> > +
> > +       list_del_init(&knode->list_head);
> > +       WRITE_ONCE(knode->owner, NULL);
> > +
> > +       return node;
> > +}
> > +
> >  __bpf_kfunc struct bpf_list_node *bpf_list_front(struct bpf_list_head *head)
> >  {
> >         struct list_head *h = (struct list_head *)head;
> > @@ -4545,6 +4571,7 @@ BTF_ID_FLAGS(func, bpf_list_push_front_impl)
> >  BTF_ID_FLAGS(func, bpf_list_push_back_impl)
> >  BTF_ID_FLAGS(func, bpf_list_pop_front, KF_ACQUIRE | KF_RET_NULL)
> >  BTF_ID_FLAGS(func, bpf_list_pop_back, KF_ACQUIRE | KF_RET_NULL)
> > +BTF_ID_FLAGS(func, bpf_list_del, KF_ACQUIRE | KF_RET_NULL)
> >  BTF_ID_FLAGS(func, bpf_list_front, KF_RET_NULL)
> >  BTF_ID_FLAGS(func, bpf_list_back, KF_RET_NULL)
> >  BTF_ID_FLAGS(func, bpf_task_acquire, KF_ACQUIRE | KF_RCU | KF_RET_NULL)
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 67c09b43a497..c9557d3fb8dd 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> > @@ -12461,6 +12461,7 @@ enum special_kfunc_type {
> >         KF_bpf_list_push_back_impl,
> >         KF_bpf_list_pop_front,
> >         KF_bpf_list_pop_back,
> > +       KF_bpf_list_del,
> >         KF_bpf_list_front,
> >         KF_bpf_list_back,
> >         KF_bpf_cast_to_kern_ctx,
> > @@ -12521,6 +12522,7 @@ BTF_ID(func, bpf_list_push_front_impl)
> >  BTF_ID(func, bpf_list_push_back_impl)
> >  BTF_ID(func, bpf_list_pop_front)
> >  BTF_ID(func, bpf_list_pop_back)
> > +BTF_ID(func, bpf_list_del)
> >  BTF_ID(func, bpf_list_front)
> >  BTF_ID(func, bpf_list_back)
> >  BTF_ID(func, bpf_cast_to_kern_ctx)
> > @@ -12996,6 +12998,7 @@ static bool is_bpf_list_api_kfunc(u32 btf_id)
> >                btf_id == special_kfunc_list[KF_bpf_list_push_back_impl] ||
> >                btf_id == special_kfunc_list[KF_bpf_list_pop_front] ||
> >                btf_id == special_kfunc_list[KF_bpf_list_pop_back] ||
> > +              btf_id == special_kfunc_list[KF_bpf_list_del] ||
> >                btf_id == special_kfunc_list[KF_bpf_list_front] ||
> >                btf_id == special_kfunc_list[KF_bpf_list_back];
> >  }
> > @@ -13118,7 +13121,8 @@ static bool check_kfunc_is_graph_node_api(struct bpf_verifier_env *env,
> >         switch (node_field_type) {
> >         case BPF_LIST_NODE:
> >                 ret = (kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_front_impl] ||
> > -                      kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_back_impl]);
> > +                      kfunc_btf_id == special_kfunc_list[KF_bpf_list_push_back_impl] ||
> > +                      kfunc_btf_id == special_kfunc_list[KF_bpf_list_del]);
> >                 break;
> >         case BPF_RB_NODE:
> >                 ret = (kfunc_btf_id == special_kfunc_list[KF_bpf_rbtree_remove] ||
> > --
> > 2.50.1 (Apple Git-155)
> >
> >
Re: [PATCH v6 1/5] bpf: Introduce the bpf_list_del kfunc.
Posted by Kumar Kartikeya Dwivedi 1 month, 1 week ago
On Wed, 4 Mar 2026 at 15:50, Chengkaitao <pilgrimtao@gmail.com> wrote:
>
> From: Kaitao Cheng <chengkaitao@kylinos.cn>
>
> If a user holds ownership of a node in the middle of a list, they
> can directly remove it from the list without strictly adhering to
> deletion rules from the head or tail.
>
> We have added an additional parameter bpf_list_head *head to
> bpf_list_del, as the verifier requires the head parameter to
> check whether the lock is being held.
>
> This is typically paired with bpf_refcount. After calling
> bpf_list_del, it is generally necessary to drop the reference to
> the list node twice to prevent reference count leaks.
>
> Signed-off-by: Kaitao Cheng <chengkaitao@kylinos.cn>
> ---

Acked-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>

> [...]