netdev - [patch net-next v2] sched: sch_api: add missing rcu read lock to silence the warning

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <20200720081041.8711-1-jiri@resnulli.us>
Date:   Mon, 20 Jul 2020 10:10:41 +0200
From:   Jiri Pirko <jiri@...nulli.us>
To:     netdev@...r.kernel.org
Cc:     davem@...emloft.net, kuba@...nel.org, jhs@...atatu.com,
        xiyou.wangcong@...il.com, idosch@...lanox.com, mlxsw@...lanox.com
Subject: [patch net-next v2] sched: sch_api: add missing rcu read lock to silence the warning

From: Jiri Pirko <jiri@...lanox.com>

In case the qdisc_match_from_root function() is called from non-rcu path
with rtnl mutex held, a suspiciout rcu usage warning appears:

[  241.504354] =============================
[  241.504358] WARNING: suspicious RCU usage
[  241.504366] 5.8.0-rc4-custom-01521-g72a7c7d549c3 #32 Not tainted
[  241.504370] -----------------------------
[  241.504378] net/sched/sch_api.c:270 RCU-list traversed in non-reader section!!
[  241.504382]
               other info that might help us debug this:
[  241.504388]
               rcu_scheduler_active = 2, debug_locks = 1
[  241.504394] 1 lock held by tc/1391:
[  241.504398]  #0: ffffffff85a27850 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x49a/0xbd0
[  241.504431]
               stack backtrace:
[  241.504440] CPU: 0 PID: 1391 Comm: tc Not tainted 5.8.0-rc4-custom-01521-g72a7c7d549c3 #32
[  241.504446] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-2.fc32 04/01/2014
[  241.504453] Call Trace:
[  241.504465]  dump_stack+0x100/0x184
[  241.504482]  lockdep_rcu_suspicious+0x153/0x15d
[  241.504499]  qdisc_match_from_root+0x293/0x350

Fix this by passing the rtnl held lockdep condition down to
hlist_for_each_entry_rcu()

Reported-by: Ido Schimmel <idosch@...lanox.com>
Signed-off-by: Jiri Pirko <jiri@...lanox.com>
---
v1->v2:
- reworked to use the lockdep condition passing suggested by Ido.
---
 include/linux/hashtable.h | 4 ++--
 net/sched/sch_api.c       | 3 ++-
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/include/linux/hashtable.h b/include/linux/hashtable.h
index 78b6ea5fa8ba..f6c666730b8c 100644
--- a/include/linux/hashtable.h
+++ b/include/linux/hashtable.h
@@ -173,9 +173,9 @@ static inline void hash_del_rcu(struct hlist_node *node)
  * @member: the name of the hlist_node within the struct
  * @key: the key of the objects to iterate over
  */
-#define hash_for_each_possible_rcu(name, obj, member, key)		\
+#define hash_for_each_possible_rcu(name, obj, member, key, cond...)	\
 	hlist_for_each_entry_rcu(obj, &name[hash_min(key, HASH_BITS(name))],\
-		member)
+		member, ## cond)
 
 /**
  * hash_for_each_possible_rcu_notrace - iterate over all possible objects hashing
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 11ebba60da3b..2a76a2f5ed88 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -267,7 +267,8 @@ static struct Qdisc *qdisc_match_from_root(struct Qdisc *root, u32 handle)
 	    root->handle == handle)
 		return root;
 
-	hash_for_each_possible_rcu(qdisc_dev(root)->qdisc_hash, q, hash, handle) {
+	hash_for_each_possible_rcu(qdisc_dev(root)->qdisc_hash, q, hash, handle,
+				   lockdep_rtnl_is_held()) {
 		if (q->handle == handle)
 			return q;
 	}
-- 
2.21.3