[PATCH v1 bpf-next 0/2] bpf: Misc changes around AF_UNIX.

[PATCH v1 bpf-next 0/2] bpf: Misc changes around AF_UNIX.

[Kuniyuki Iwashima]

Patch 1 adapts sk_is_XXX() helpers in __cgroup_bpf_run_filter_sock_addr().
Patch 2 removes an unnecessary sk_fullsock() in bpf_skc_to_unix_sock().


Kuniyuki Iwashima (2):
  bpf: Use sk_is_inet() and sk_is_unix() in
    __cgroup_bpf_run_filter_sock_addr().
  bpf: Don't check sk_fullsock() in bpf_skc_to_unix_sock().

 kernel/bpf/cgroup.c | 6 +-----
 net/core/filter.c   | 2 +-
 2 files changed, 2 insertions(+), 6 deletions(-)

-- 
2.53.0.rc2.204.g2597b5adb4-goog

[PATCH v1 bpf-next 1/2] bpf: Use sk_is_inet() and sk_is_unix() in __cgroup_bpf_run_filter_sock_addr().

[Kuniyuki Iwashima]

sk->sk_family should be read with READ_ONCE() in
__cgroup_bpf_run_filter_sock_addr() due to IPV6_ADDRFORM.

Also, the comment there is a bit stale since commit 859051dd165e
("bpf: Implement cgroup sockaddr hooks for unix sockets"), and the
kdoc has the same comment.

Let's use sk_is_inet() and sk_is_unix() and remove the comment.

Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
---
 kernel/bpf/cgroup.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
index 69988af44b37..b029f0369ecf 100644
--- a/kernel/bpf/cgroup.c
+++ b/kernel/bpf/cgroup.c
@@ -1680,11 +1680,7 @@ int __cgroup_bpf_run_filter_sock_addr(struct sock *sk,
 	struct cgroup *cgrp;
 	int ret;
 
-	/* Check socket family since not all sockets represent network
-	 * endpoint (e.g. AF_UNIX).
-	 */
-	if (sk->sk_family != AF_INET && sk->sk_family != AF_INET6 &&
-	    sk->sk_family != AF_UNIX)
+	if (!sk_is_inet(sk) && !sk_is_unix(sk))
 		return 0;
 
 	if (!ctx.uaddr) {
-- 
2.53.0.rc2.204.g2597b5adb4-goog

[PATCH v1 bpf-next 2/2] bpf: Don't check sk_fullsock() in bpf_skc_to_unix_sock().

[Kuniyuki Iwashima]

AF_UNIX does not use TCP_NEW_SYN_RECV nor TCP_TIME_WAIT and
checking sk->sk_family is sufficient.

Let's remove sk_fullsock() and use sk_is_unix() in
bpf_skc_to_unix_sock().

Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
---
 net/core/filter.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/core/filter.c b/net/core/filter.c
index f04982d79d72..90bfe5a6c1db 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -12021,7 +12021,7 @@ BPF_CALL_1(bpf_skc_to_unix_sock, struct sock *, sk)
 	 * trigger an explicit type generation here.
 	 */
 	BTF_TYPE_EMIT(struct unix_sock);
-	if (sk && sk_fullsock(sk) && sk->sk_family == AF_UNIX)
+	if (sk && sk_is_unix(sk))
 		return (unsigned long)sk;
 
 	return (unsigned long)NULL;
-- 
2.53.0.rc2.204.g2597b5adb4-goog

Re: [PATCH v1 bpf-next 0/2] bpf: Misc changes around AF_UNIX.

[Stanislav Fomichev]

On 02/03, Kuniyuki Iwashima wrote:
> Patch 1 adapts sk_is_XXX() helpers in __cgroup_bpf_run_filter_sock_addr().
> Patch 2 removes an unnecessary sk_fullsock() in bpf_skc_to_unix_sock().
> 
> 
> Kuniyuki Iwashima (2):
>   bpf: Use sk_is_inet() and sk_is_unix() in
>     __cgroup_bpf_run_filter_sock_addr().
>   bpf: Don't check sk_fullsock() in bpf_skc_to_unix_sock().

Acked-by: Stanislav Fomichev <sdf@fomichev.me>

Re: [PATCH v1 bpf-next 0/2] bpf: Misc changes around AF_UNIX.

[Martin KaFai Lau]

On 2/3/26 1:30 PM, Kuniyuki Iwashima wrote:
> Patch 1 adapts sk_is_XXX() helpers in __cgroup_bpf_run_filter_sock_addr().
> Patch 2 removes an unnecessary sk_fullsock() in bpf_skc_to_unix_sock().
> 
> 
> Kuniyuki Iwashima (2):
>    bpf: Use sk_is_inet() and sk_is_unix() in
>      __cgroup_bpf_run_filter_sock_addr().
>    bpf: Don't check sk_fullsock() in bpf_skc_to_unix_sock().
> 
>   kernel/bpf/cgroup.c | 6 +-----
>   net/core/filter.c   | 2 +-

The bot is slow. The change in filter.c is small and the cycle is 
closing, so I have landed them to the bpf_next/master.

I just noticed the netdev list is not in cc, so added it here also.