BPF List
 help / color / mirror / Atom feed
From: Hawkins Jiawei <yin31149@gmail.com>
To: dhowells@redhat.com, "David S. Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Andrii Nakryiko <andrii@kernel.org>,
	Martin KaFai Lau <kafai@fb.com>, Song Liu <songliubraving@fb.com>,
	Yonghong Song <yhs@fb.com>,
	John Fastabend <john.fastabend@gmail.com>,
	KP Singh <kpsingh@kernel.org>
Cc: linux-kernel@vger.kernel.org, netdev@vger.kernel.org,
	yin31149@gmail.com,
	linux-kernel-mentees@lists.linuxfoundation.org,
	bpf@vger.kernel.org
Subject: Re: [PATCH] net: Fix suspicious RCU usage in bpf_sk_reuseport_detach()
Date: Tue, 16 Aug 2022 18:34:52 +0800	[thread overview]
Message-ID: <20220816103452.479281-1-yin31149@gmail.com> (raw)
In-Reply-To: <166064248071.3502205.10036394558814861778.stgit@warthog.procyon.org.uk>

On Tue, 16 Aug 2022 at 17:34, David Howells <dhowells@redhat.com> wrote:
>
> Fix this by adding a new helper, __locked_read_sk_user_data_with_flags()
> that checks to see if sk->sk_callback_lock() is held and use that here
> instead.
Hi, I wonder if we make this more geniric, for I think maybe the future
code who use __rcu_dereference_sk_user_data_with_flags() may
also meet this bug.

To be more specific, maybe we can refactor
__rcu_dereference_sk_user_data_with_flags() to
__rcu_dereference_sk_user_data_with_flags_check(), like
rcu_dereference() and rcu_dereference_check(). Maybe:

diff --git a/include/net/sock.h b/include/net/sock.h
index 05a1bbdf5805..cf123954eab9 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -578,18 +578,27 @@ static inline bool sk_user_data_is_nocopy(const struct sock *sk)
 #define __sk_user_data(sk) ((*((void __rcu **)&(sk)->sk_user_data)))
 
 /**
- * __rcu_dereference_sk_user_data_with_flags - return the pointer
- * only if argument flags all has been set in sk_user_data. Otherwise
- * return NULL
+ * __rcu_dereference_sk_user_data_with_flags_check - return the pointer
+ * only if argument flags all has been set in sk_user_data, with debug
+ * checking. Otherwise return NULL
  *
- * @sk: socket
- * @flags: flag bits
+ * Do __rcu_dereference_sk_user_data_with_flags(), but check that the
+ * conditions under which the rcu dereference will take place are correct,
+ * which is a bit like rcu_dereference_check() and rcu_derefence().
+ *
+ * @sk		: socket
+ * @flags	: flag bits
+ * @condition	: the conditions under which the rcu dereference will
+ * take place
  */
 static inline void *
-__rcu_dereference_sk_user_data_with_flags(const struct sock *sk,
-					  uintptr_t flags)
+__rcu_dereference_sk_user_data_with_flags_check(const struct sock *sk,
+						uintptr_t flags, bool condition)
 {
-	uintptr_t sk_user_data = (uintptr_t)rcu_dereference(__sk_user_data(sk));
+	uintptr_t sk_user_data;
+
+	sk_user_data = (uintptr_t)rcu_dereference_check(__sk_user_data(sk),
+							condition);
 
 	WARN_ON_ONCE(flags & SK_USER_DATA_PTRMASK);
 
@@ -598,6 +607,8 @@ __rcu_dereference_sk_user_data_with_flags(const struct sock *sk,
 	return NULL;
 }
 
+#define __rcu_dereference_sk_user_data_with_flags(sk, flags) \
+	__rcu_dereference_sk_user_data_with_flags_check(sk, flags, 0)
 #define rcu_dereference_sk_user_data(sk)				\
 	__rcu_dereference_sk_user_data_with_flags(sk, 0)
 #define __rcu_assign_sk_user_data_with_flags(sk, ptr, flags)		\

> +/**
> + * __locked_read_sk_user_data_with_flags - return the pointer
> + * only if argument flags all has been set in sk_user_data. Otherwise
> + * return NULL
> + *
> +               (uintptr_t)rcu_dereference_check(__sk_user_data(sk),
> +                                                lockdep_is_held(&sk->sk_callback_lock));

> diff --git a/kernel/bpf/reuseport_array.c b/kernel/bpf/reuseport_array.c
> index 85fa9dbfa8bf..82c61612f382 100644
> --- a/kernel/bpf/reuseport_array.c
> +++ b/kernel/bpf/reuseport_array.c
> @@ -24,7 +24,7 @@ void bpf_sk_reuseport_detach(struct sock *sk)
>         struct sock __rcu **socks;
> 
>         write_lock_bh(&sk->sk_callback_lock);
> -       socks = __rcu_dereference_sk_user_data_with_flags(sk, SK_USER_DATA_BPF);
> +       socks = __locked_read_sk_user_data_with_flags(sk, SK_USER_DATA_BPF);
>         if (socks) {
>                 WRITE_ONCE(sk->sk_user_data, NULL);
>                 /*
Then, as you point out, we can pass
condition(lockdep_is_held(&sk->sk_callback_lock)) to
__rcu_dereference_sk_user_data_with_flags_check() in order to
make compiler happy as below:

diff --git a/kernel/bpf/reuseport_array.c b/kernel/bpf/reuseport_array.c
index 85fa9dbfa8bf..a772610987c5 100644
--- a/kernel/bpf/reuseport_array.c
+++ b/kernel/bpf/reuseport_array.c
@@ -24,7 +24,10 @@ void bpf_sk_reuseport_detach(struct sock *sk)
 	struct sock __rcu **socks;
 
 	write_lock_bh(&sk->sk_callback_lock);
-	socks = __rcu_dereference_sk_user_data_with_flags(sk, SK_USER_DATA_BPF);
+	socks = __rcu_dereference_sk_user_data_with_flags_check(
+			sk, SK_USER_DATA_BPF,
+			lockdep_is_held(&sk->sk_callback_lock));
+
 	if (socks) {
 		WRITE_ONCE(sk->sk_user_data, NULL);
 		/*

       reply	other threads:[~2022-08-16 11:07 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <166064248071.3502205.10036394558814861778.stgit@warthog.procyon.org.uk>
2022-08-16 10:34 ` Hawkins Jiawei [this message]
2022-08-16 13:09   ` [PATCH] net: Fix suspicious RCU usage in bpf_sk_reuseport_detach() David Howells
2022-08-16 21:49     ` Martin KaFai Lau
2022-08-16 18:21   ` Jakub Kicinski
2022-08-16 20:01     ` David Howells
2022-08-16 21:16   ` David Howells
2022-08-16 23:44     ` Jakub Kicinski
2022-08-17  0:43       ` Martin KaFai Lau
2022-08-17  1:39         ` Jakub Kicinski
2022-08-17  3:00           ` Hawkins Jiawei
2022-08-17 20:55       ` David Howells
2022-08-17 23:42         ` Jakub Kicinski

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20220816103452.479281-1-yin31149@gmail.com \
    --to=yin31149@gmail.com \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=davem@davemloft.net \
    --cc=dhowells@redhat.com \
    --cc=edumazet@google.com \
    --cc=john.fastabend@gmail.com \
    --cc=kafai@fb.com \
    --cc=kpsingh@kernel.org \
    --cc=kuba@kernel.org \
    --cc=linux-kernel-mentees@lists.linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=songliubraving@fb.com \
    --cc=yhs@fb.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox