From: Cong Wang <xiyou.wangcong@gmail.com>
To: Cameron Berkenpas <cam@neo-zeon.de>
Cc: "Roman Gushchin" <guro@fb.com>, "Zefan Li" <lizefan@huawei.com>,
"Linux Kernel Network Developers" <netdev@vger.kernel.org>,
"Peter Geis" <pgwipeout@gmail.com>,
"Lu Fengqi" <lufq.fnst@cn.fujitsu.com>,
"Daniël Sonck" <dsonck92@gmail.com>,
"Daniel Borkmann" <daniel@iogearbox.net>,
"Tejun Heo" <tj@kernel.org>
Subject: Re: [Patch net] cgroup: fix cgroup_sk_alloc() for sk_clone_lock()
Date: Fri, 26 Jun 2020 10:58:14 -0700 [thread overview]
Message-ID: <CAM_iQpU1ji2x9Pgb6Xs7Kqoh3mmFRN3R9GKf5QoVUv82mZb8hg@mail.gmail.com> (raw)
In-Reply-To: <b3a5298d-3c4e-ba51-7045-9643c3986054@neo-zeon.de>
[-- Attachment #1: Type: text/plain, Size: 748 bytes --]
On Thu, Jun 25, 2020 at 10:23 PM Cameron Berkenpas <cam@neo-zeon.de> wrote:
>
> Hello,
>
> Somewhere along the way I got the impression that it generally takes
> those affected hours before their systems lock up. I'm (generally) able
> to reproduce this issue much faster than that. Regardless, I can help test.
>
> Are there any patches that need testing or is this all still pending
> discussion around the best way to resolve the issue?
Yes. I come up with a (hopefully) much better patch in the attachment.
Can you help to test it? You need to unapply the previous patch before
applying this one.
(Just in case of any confusion: I still believe we should check NULL on
top of this refcnt fix. But it should be a separate patch.)
Thank you!
[-- Attachment #2: cgroup-bpf-ref.patch --]
[-- Type: text/x-patch, Size: 4774 bytes --]
commit 259150604c0b77c717fdaab057da5722e2dfd922
Author: Cong Wang <xiyou.wangcong@gmail.com>
Date: Sat Jun 13 12:34:40 2020 -0700
cgroup: fix cgroup_sk_alloc() for sk_clone_lock()
When we clone a socket in sk_clone_lock(), its sk_cgrp_data is
copied, so the cgroup refcnt must be taken too. And, unlike the
sk_alloc() path, sock_update_netprioidx() is not called here.
Therefore, it is safe and necessary to grab the cgroup refcnt
even when cgroup_sk_alloc is disabled.
sk_clone_lock() is in BH context anyway, the in_interrupt()
would terminate this function if called there. And for sk_alloc()
skcd->val is always zero. So it's safe to factor out the code
to make it more readable.
Fixes: 4bfc0bb2c60e ("bpf: decouple the lifetime of cgroup_bpf from cgroup itself")
Reported-by: Cameron Berkenpas <cam@neo-zeon.de>
Reported-by: Peter Geis <pgwipeout@gmail.com>
Reported-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Reported-by: Daniël Sonck <dsonck92@gmail.com>
Tested-by: Cameron Berkenpas <cam@neo-zeon.de>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Zefan Li <lizefan@huawei.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
diff --git a/include/linux/cgroup-defs.h b/include/linux/cgroup-defs.h
index 52661155f85f..4f1cd0edc57d 100644
--- a/include/linux/cgroup-defs.h
+++ b/include/linux/cgroup-defs.h
@@ -790,7 +790,8 @@ struct sock_cgroup_data {
union {
#ifdef __LITTLE_ENDIAN
struct {
- u8 is_data;
+ u8 is_data : 1;
+ u8 no_refcnt : 1;
u8 padding;
u16 prioidx;
u32 classid;
@@ -800,7 +801,8 @@ struct sock_cgroup_data {
u32 classid;
u16 prioidx;
u8 padding;
- u8 is_data;
+ u8 no_refcnt : 1;
+ u8 is_data : 1;
} __packed;
#endif
u64 val;
diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index 4598e4da6b1b..618838c48313 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -822,6 +822,7 @@ extern spinlock_t cgroup_sk_update_lock;
void cgroup_sk_alloc_disable(void);
void cgroup_sk_alloc(struct sock_cgroup_data *skcd);
+void cgroup_sk_clone(struct sock_cgroup_data *skcd);
void cgroup_sk_free(struct sock_cgroup_data *skcd);
static inline struct cgroup *sock_cgroup_ptr(struct sock_cgroup_data *skcd)
@@ -835,7 +836,7 @@ static inline struct cgroup *sock_cgroup_ptr(struct sock_cgroup_data *skcd)
*/
v = READ_ONCE(skcd->val);
- if (v & 1)
+ if (v & 3)
return &cgrp_dfl_root.cgrp;
return (struct cgroup *)(unsigned long)v ?: &cgrp_dfl_root.cgrp;
@@ -847,6 +848,7 @@ static inline struct cgroup *sock_cgroup_ptr(struct sock_cgroup_data *skcd)
#else /* CONFIG_CGROUP_DATA */
static inline void cgroup_sk_alloc(struct sock_cgroup_data *skcd) {}
+static inline void cgroup_sk_clone(struct sock_cgroup_data *skcd) {}
static inline void cgroup_sk_free(struct sock_cgroup_data *skcd) {}
#endif /* CONFIG_CGROUP_DATA */
diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index 1ea181a58465..dd247747ec14 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -6439,18 +6439,8 @@ void cgroup_sk_alloc_disable(void)
void cgroup_sk_alloc(struct sock_cgroup_data *skcd)
{
- if (cgroup_sk_alloc_disabled)
- return;
-
- /* Socket clone path */
- if (skcd->val) {
- /*
- * We might be cloning a socket which is left in an empty
- * cgroup and the cgroup might have already been rmdir'd.
- * Don't use cgroup_get_live().
- */
- cgroup_get(sock_cgroup_ptr(skcd));
- cgroup_bpf_get(sock_cgroup_ptr(skcd));
+ if (cgroup_sk_alloc_disabled) {
+ skcd->no_refcnt = 1;
return;
}
@@ -6475,10 +6465,27 @@ void cgroup_sk_alloc(struct sock_cgroup_data *skcd)
rcu_read_unlock();
}
+void cgroup_sk_clone(struct sock_cgroup_data *skcd)
+{
+ if (skcd->val) {
+ if (skcd->no_refcnt)
+ return;
+ /*
+ * We might be cloning a socket which is left in an empty
+ * cgroup and the cgroup might have already been rmdir'd.
+ * Don't use cgroup_get_live().
+ */
+ cgroup_get(sock_cgroup_ptr(skcd));
+ cgroup_bpf_get(sock_cgroup_ptr(skcd));
+ }
+}
+
void cgroup_sk_free(struct sock_cgroup_data *skcd)
{
struct cgroup *cgrp = sock_cgroup_ptr(skcd);
+ if (skcd->no_refcnt)
+ return;
cgroup_bpf_put(cgrp);
cgroup_put(cgrp);
}
diff --git a/net/core/sock.c b/net/core/sock.c
index d832c650287c..2e5b7870e5d3 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -1926,7 +1926,7 @@ struct sock *sk_clone_lock(const struct sock *sk, const gfp_t priority)
/* sk->sk_memcg will be populated at accept() time */
newsk->sk_memcg = NULL;
- cgroup_sk_alloc(&newsk->sk_cgrp_data);
+ cgroup_sk_clone(&newsk->sk_cgrp_data);
rcu_read_lock();
filter = rcu_dereference(sk->sk_filter);
next prev parent reply other threads:[~2020-06-26 17:58 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-16 18:03 [Patch net] cgroup: fix cgroup_sk_alloc() for sk_clone_lock() Cong Wang
2020-06-18 1:44 ` Zefan Li
2020-06-18 19:19 ` Cong Wang
2020-06-18 19:36 ` Roman Gushchin
2020-06-18 21:09 ` Cong Wang
2020-06-18 21:26 ` Roman Gushchin
2020-06-18 22:45 ` Peter Geis
2020-06-19 6:40 ` Zefan Li
2020-06-19 19:51 ` Cong Wang
2020-06-20 0:45 ` Zefan Li
2020-06-20 0:51 ` Zefan Li
2020-06-20 3:31 ` Cong Wang
2020-06-20 7:52 ` Zefan Li
2020-06-20 16:04 ` Roman Gushchin
2020-06-23 22:21 ` Roman Gushchin
2020-06-26 5:23 ` Cameron Berkenpas
2020-06-26 17:58 ` Cong Wang [this message]
2020-06-26 22:03 ` Cameron Berkenpas
2020-06-27 22:59 ` Cameron Berkenpas
2020-06-30 22:16 ` Cong Wang
2020-06-27 23:41 ` Roman Gushchin
2020-06-30 22:22 ` Cong Wang
2020-06-30 22:48 ` Roman Gushchin
2020-07-01 1:18 ` Zefan Li
2020-07-02 4:48 ` Cong Wang
2020-07-02 8:12 ` Thomas Lamprecht
2020-07-02 16:02 ` Roman Gushchin
2020-07-02 16:24 ` Peter Geis
2020-07-03 1:17 ` Zefan Li
2020-06-20 0:51 ` Roman Gushchin
2020-06-20 1:00 ` Zefan Li
2020-06-20 1:14 ` Roman Gushchin
2020-06-20 2:48 ` Zefan Li
2020-06-20 3:00 ` Cong Wang
2020-06-20 15:57 ` Roman Gushchin
2020-06-22 18:14 ` Cong Wang
2020-06-22 20:39 ` Roman Gushchin
2020-06-23 8:45 ` Zhang,Qiang
2020-06-23 17:56 ` Cong Wang
2020-06-23 8:54 ` Zhang,Qiang
2020-06-23 9:01 ` Zhang,Qiang
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=CAM_iQpU1ji2x9Pgb6Xs7Kqoh3mmFRN3R9GKf5QoVUv82mZb8hg@mail.gmail.com \
--to=xiyou.wangcong@gmail.com \
--cc=cam@neo-zeon.de \
--cc=daniel@iogearbox.net \
--cc=dsonck92@gmail.com \
--cc=guro@fb.com \
--cc=lizefan@huawei.com \
--cc=lufq.fnst@cn.fujitsu.com \
--cc=netdev@vger.kernel.org \
--cc=pgwipeout@gmail.com \
--cc=tj@kernel.org \
/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;
as well as URLs for NNTP newsgroup(s).