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 | 30 +++++++++++++++++++++++-------
kernel/bpf/verifier.c | 6 +++++-
2 files changed, 28 insertions(+), 8 deletions(-)
diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
index 6eb6c82ed2ee..01b74c4ac00d 100644
--- a/kernel/bpf/helpers.c
+++ b/kernel/bpf/helpers.c
@@ -2426,20 +2426,23 @@ __bpf_kfunc int bpf_list_push_back_impl(struct bpf_list_head *head,
return __bpf_list_add(n, head, true, meta ? meta->record : NULL, off);
}
-static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head, bool tail)
+static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head,
+ struct list_head *n)
{
- struct list_head *n, *h = (void *)head;
+ struct list_head *h = (void *)head;
struct bpf_list_node_kern *node;
/* 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))
+ if (unlikely(!h->next)) {
INIT_LIST_HEAD(h);
- if (list_empty(h))
+ return NULL;
+ }
+
+ if (n == h)
return NULL;
- n = tail ? h->prev : h->next;
node = container_of(n, struct bpf_list_node_kern, list_head);
if (WARN_ON_ONCE(READ_ONCE(node->owner) != head))
return NULL;
@@ -2451,12 +2454,24 @@ static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head, bool tai
__bpf_kfunc struct bpf_list_node *bpf_list_pop_front(struct bpf_list_head *head)
{
- return __bpf_list_del(head, false);
+ struct list_head *h = (void *)head;
+
+ return __bpf_list_del(head, h->next);
}
__bpf_kfunc struct bpf_list_node *bpf_list_pop_back(struct bpf_list_head *head)
{
- return __bpf_list_del(head, true);
+ struct list_head *h = (void *)head;
+
+ return __bpf_list_del(head, h->prev);
+}
+
+__bpf_kfunc struct bpf_list_node *bpf_list_del(struct bpf_list_head *head,
+ struct bpf_list_node *node)
+{
+ struct bpf_list_node_kern *kn = (void *)node;
+
+ return __bpf_list_del(head, &kn->list_head);
}
__bpf_kfunc struct bpf_list_node *bpf_list_front(struct bpf_list_head *head)
@@ -4545,6 +4560,7 @@ 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)
On 8/3/26 21:46, Chengkaitao 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 | 30 +++++++++++++++++++++++-------
> kernel/bpf/verifier.c | 6 +++++-
> 2 files changed, 28 insertions(+), 8 deletions(-)
>
> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index 6eb6c82ed2ee..01b74c4ac00d 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c
> @@ -2426,20 +2426,23 @@ __bpf_kfunc int bpf_list_push_back_impl(struct bpf_list_head *head,
> return __bpf_list_add(n, head, true, meta ? meta->record : NULL, off);
> }
>
> -static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head, bool tail)
> +static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head,
> + struct list_head *n)
> {
> - struct list_head *n, *h = (void *)head;
> + struct list_head *h = (void *)head;
> struct bpf_list_node_kern *node;
>
> /* 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))
> + if (unlikely(!h->next)) {
> INIT_LIST_HEAD(h);
> - if (list_empty(h))
> + return NULL;
> + }
> +
> + if (n == h)
> return NULL;
>
> - n = tail ? h->prev : h->next;
> node = container_of(n, struct bpf_list_node_kern, list_head);
> if (WARN_ON_ONCE(READ_ONCE(node->owner) != head))
> return NULL;
This refactoring is worth, because the "struct list_head *n" seems
better than "bool tail".
But, such refactoring should be a preparatory patch. Importantly,
refactoring should not introduce functional changes.
Thanks,
Leon
> @@ -2451,12 +2454,24 @@ static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head, bool tai
>
> __bpf_kfunc struct bpf_list_node *bpf_list_pop_front(struct bpf_list_head *head)
> {
> - return __bpf_list_del(head, false);
> + struct list_head *h = (void *)head;
> +
> + return __bpf_list_del(head, h->next);
> }
>
> __bpf_kfunc struct bpf_list_node *bpf_list_pop_back(struct bpf_list_head *head)
> {
> - return __bpf_list_del(head, true);
> + struct list_head *h = (void *)head;
> +
> + return __bpf_list_del(head, h->prev);
> +}
> +
> +__bpf_kfunc struct bpf_list_node *bpf_list_del(struct bpf_list_head *head,
> + struct bpf_list_node *node)
> +{
> + struct bpf_list_node_kern *kn = (void *)node;
> +
> + return __bpf_list_del(head, &kn->list_head);
> }
>
> __bpf_kfunc struct bpf_list_node *bpf_list_front(struct bpf_list_head *head)
> @@ -4545,6 +4560,7 @@ 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] ||
On Mon, 9 Mar 2026 at 07:34, Leon Hwang <leon.hwang@linux.dev> wrote:
>
> On 8/3/26 21:46, Chengkaitao 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 | 30 +++++++++++++++++++++++-------
> > kernel/bpf/verifier.c | 6 +++++-
> > 2 files changed, 28 insertions(+), 8 deletions(-)
> >
> > diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> > index 6eb6c82ed2ee..01b74c4ac00d 100644
> > --- a/kernel/bpf/helpers.c
> > +++ b/kernel/bpf/helpers.c
> > @@ -2426,20 +2426,23 @@ __bpf_kfunc int bpf_list_push_back_impl(struct bpf_list_head *head,
> > return __bpf_list_add(n, head, true, meta ? meta->record : NULL, off);
> > }
> >
> > -static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head, bool tail)
> > +static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head,
> > + struct list_head *n)
> > {
> > - struct list_head *n, *h = (void *)head;
> > + struct list_head *h = (void *)head;
> > struct bpf_list_node_kern *node;
> >
> > /* 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))
> > + if (unlikely(!h->next)) {
> > INIT_LIST_HEAD(h);
> > - if (list_empty(h))
> > + return NULL;
> > + }
> > +
> > + if (n == h)
> > return NULL;
Couldn't you keep the list_empty(h) check after INIT_LIST_HEAD(h) as before?
And we don't need n == h either. You could add a comment that n will
never match h.
The verifier should ensure it, since it distinguishes the head and node types.
Let's say the head is uninit. Then list_empty(h) will catch that case.
Otherwise n might be NULL.
After list_empty(h) says false, we definitely have non-null n.
We just need to check list membership using the owner check, and then
we're good.
It will be a less noisy diff.
> >
> > - n = tail ? h->prev : h->next;
> > node = container_of(n, struct bpf_list_node_kern, list_head);
> > if (WARN_ON_ONCE(READ_ONCE(node->owner) != head))
> > return NULL;
>
> This refactoring is worth, because the "struct list_head *n" seems
> better than "bool tail".
>
> But, such refactoring should be a preparatory patch. Importantly,
> refactoring should not introduce functional changes.
>
I think it's fine. Let's address this and avoid too many respins now.
It isn't a lot of code anyway. You could mention in the commit log
that you chose to refactor __bpf_list_del though.
> Thanks,
> Leon
>
> > @@ -2451,12 +2454,24 @@ static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head, bool tai
> >
> > __bpf_kfunc struct bpf_list_node *bpf_list_pop_front(struct bpf_list_head *head)
> > {
> > - return __bpf_list_del(head, false);
> > + struct list_head *h = (void *)head;
> > +
> > + return __bpf_list_del(head, h->next);
> > }
> >
> > __bpf_kfunc struct bpf_list_node *bpf_list_pop_back(struct bpf_list_head *head)
> > {
> > - return __bpf_list_del(head, true);
> > + struct list_head *h = (void *)head;
> > +
> > + return __bpf_list_del(head, h->prev);
> > +}
> > +
> > +__bpf_kfunc struct bpf_list_node *bpf_list_del(struct bpf_list_head *head,
> > + struct bpf_list_node *node)
> > +{
> > + struct bpf_list_node_kern *kn = (void *)node;
> > +
> > + return __bpf_list_del(head, &kn->list_head);
> > }
> >
> > __bpf_kfunc struct bpf_list_node *bpf_list_front(struct bpf_list_head *head)
> > @@ -4545,6 +4560,7 @@ 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] ||
>
>
On Tue, 10 Mar 2026 at 21:10, Kumar Kartikeya Dwivedi <memxor@gmail.com> wrote:
>
> On Mon, 9 Mar 2026 at 07:34, Leon Hwang <leon.hwang@linux.dev> wrote:
> >
> > On 8/3/26 21:46, Chengkaitao 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 | 30 +++++++++++++++++++++++-------
> > > kernel/bpf/verifier.c | 6 +++++-
> > > 2 files changed, 28 insertions(+), 8 deletions(-)
> > >
> > > diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> > > index 6eb6c82ed2ee..01b74c4ac00d 100644
> > > --- a/kernel/bpf/helpers.c
> > > +++ b/kernel/bpf/helpers.c
> > > @@ -2426,20 +2426,23 @@ __bpf_kfunc int bpf_list_push_back_impl(struct bpf_list_head *head,
> > > return __bpf_list_add(n, head, true, meta ? meta->record : NULL, off);
> > > }
> > >
> > > -static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head, bool tail)
> > > +static struct bpf_list_node *__bpf_list_del(struct bpf_list_head *head,
> > > + struct list_head *n)
> > > {
> > > - struct list_head *n, *h = (void *)head;
> > > + struct list_head *h = (void *)head;
> > > struct bpf_list_node_kern *node;
> > >
> > > /* 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))
> > > + if (unlikely(!h->next)) {
> > > INIT_LIST_HEAD(h);
> > > - if (list_empty(h))
> > > + return NULL;
> > > + }
> > > +
> > > + if (n == h)
> > > return NULL;
>
> Couldn't you keep the list_empty(h) check after INIT_LIST_HEAD(h) as before?
> And we don't need n == h either. You could add a comment that n will
> never match h.
> The verifier should ensure it, since it distinguishes the head and node types.
>
> Let's say the head is uninit. Then list_empty(h) will catch that case.
> Otherwise n might be NULL.
>
> After list_empty(h) says false, we definitely have non-null n.
> We just need to check list membership using the owner check, and then
> we're good.
> It will be a less noisy diff.
>
> > >
> > > - n = tail ? h->prev : h->next;
> > > node = container_of(n, struct bpf_list_node_kern, list_head);
> > > if (WARN_ON_ONCE(READ_ONCE(node->owner) != head))
> > > return NULL;
> >
> > This refactoring is worth, because the "struct list_head *n" seems
> > better than "bool tail".
> >
> > But, such refactoring should be a preparatory patch. Importantly,
> > refactoring should not introduce functional changes.
> >
>
> I think it's fine. Let's address this and avoid too many respins now.
> It isn't a lot of code anyway. You could mention in the commit log
> that you chose to refactor __bpf_list_del though.
>
Just to make it clearer, since I feel the language above might be a
bit confusing:
Let's not add more churn and just fix the issues in the existing set,
and try to move towards landing this.
We are quite close, and it's been 7 iterations already.
The bit about the non-owning refs pointed out by the AI bot, you can
do it as a follow up on top after this is accepted.
> [...]
© 2016 - 2026 Red Hat, Inc.