Re: [PATCH 4.19 016/133] cgroup: fix cgroup_sk_alloc() for sk_clone_lock()

From: Yang Yingliang
Date: Thu Aug 13 2020 - 08:07:50 EST

On 2020/8/13 19:41, Greg Kroah-Hartman wrote:
On Thu, Aug 13, 2020 at 07:30:55PM +0800, Yang Yingliang wrote:

On 2020/7/20 23:36, Greg Kroah-Hartman wrote:
From: Cong Wang <xiyou.wangcong@xxxxxxxxx>

[ Upstream commit ad0f75e5f57ccbceec13274e1e242f2b5a6397ed ]

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.

The global variable 'cgroup_sk_alloc_disabled' is used to determine
whether to take these reference counts. It is impossible to make
the reference counting correct unless we save this bit of information
in skcd->val. So, add a new bit there to record whether the socket
has already taken the reference counts. This obviously relies on
kmalloc() to align cgroup pointers to at least 4 bytes,
ARCH_KMALLOC_MINALIGN is certainly larger than that.

This bug seems to be introduced since the beginning, commit
d979a39d7242 ("cgroup: duplicate cgroup reference when cloning sockets")
tried to fix it but not compeletely. It seems not easy to trigger until
the recent commit 090e28b229af
("netprio_cgroup: Fix unlimited memory leak of v2 cgroups") was merged.

Fixes: bd1060a1d671 ("sock, cgroup: add sock->sk_cgroup")
Reported-by: Cameron Berkenpas <cam@xxxxxxxxxxx>
Reported-by: Peter Geis <pgwipeout@xxxxxxxxx>
Reported-by: Lu Fengqi <lufq.fnst@xxxxxxxxxxxxxx>
Reported-by: Daniël Sonck <dsonck92@xxxxxxxxx>
Reported-by: Zhang Qiang <qiang.zhang@xxxxxxxxxxxxx>
Tested-by: Cameron Berkenpas <cam@xxxxxxxxxxx>
Tested-by: Peter Geis <pgwipeout@xxxxxxxxx>
Tested-by: Thomas Lamprecht <t.lamprecht@xxxxxxxxxxx>
Cc: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
Cc: Zefan Li <lizefan@xxxxxxxxxx>
Cc: Tejun Heo <tj@xxxxxxxxxx>
Cc: Roman Gushchin <guro@xxxxxx>
Signed-off-by: Cong Wang <xiyou.wangcong@xxxxxxxxx>
Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
+void cgroup_sk_clone(struct sock_cgroup_data *skcd)
+ /* Socket clone path */
+ if (skcd->val) {
Compare to mainline patch, it's missing *if (skcd->no_refcnt)* check here.

Is it a mistake here ?
Possibly, it is in the cgroup_sk_free() call. Can you send a patch to
fix this up?

OK, I checked other stable branches, it also need be fixed in stable-4.9 and stable-4.14.

I will send the patches to these branches.


greg k-h