Skip to contentSkip to conversation

net/sched: two fixes for cls_u32

PATCH Applied Eric Dumazet 3 participants 6 messages Apr 13, 2022 – Apr 15, 2022

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover net/sched: two fixes for cls_u32

From: Eric Dumazet <[email protected]>

One syzbot report brought my attention to cls_u32.

This series addresses the syzbot report, and an additional issue discovered in code review.

Eric Dumazet (2):
net/sched: cls_u32: fix netns refcount changes in u32_change() net/sched: cls_u32: fix possible leak in u32_init_knode()

 net/sched/cls_u32.c | 24 ++++++++++++++----------
 1 file changed, 14 insertions(+), 10 deletions(-)
Signature · 1 line
2.35.1.1178.g4f1659d476-goog
#1 lore raw

PATCH 1/2 net/sched: cls_u32: fix netns refcount changes in u32_change()

From: Eric Dumazet <[email protected]>

We are now able to detect extra put_net() at the moment they happen, instead of much later in correct code paths.

u32_init_knode() / tcf_exts_init() populates the ->exts.net pointer, but as mentioned in tcf_exts_init(),
the refcount on netns has not been elevated yet.

The refcount is taken only once tcf_exts_get_net() is called.

So the two u32_destroy_key() calls from u32_change() are attempting to release an invalid reference on the netns.

syzbot report:

refcount_t: decrement hit 0; leaking memory.

WARNING: CPU: 0 PID: 21708 at lib/refcount.c:31 refcount_warn_saturate+0xbf/0x1e0 lib/refcount.c:31
Modules linked in:
CPU: 0 PID: 21708 Comm: syz-executor.5 Not tainted 5.18.0-rc2-next-20220412-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
RIP: 0010:refcount_warn_saturate+0xbf/0x1e0 lib/refcount.c:31
Code: 1d 14 b6 b2 09 31 ff 89 de e8 6d e9 89 fd 84 db 75 e0 e8 84 e5 89 fd 48 c7 c7 40 aa 26 8a c6 05 f4 b5 b2 09 01 e8 e5 81 2e 05 <0f> 0b eb c4 e8 68 e5 89 fd 0f b6 1d e3 b5 b2 09 31 ff 89 de e8 38
RSP: 0018:ffffc900051af1b0 EFLAGS: 00010286
RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000
RDX: 0000000000040000 RSI: ffffffff8160a0c8 RDI: fffff52000a35e28
RBP: 0000000000000004 R08: 0000000000000000 R09: 0000000000000000
R10: ffffffff81604a9e R11: 0000000000000000 R12: 1ffff92000a35e3b
R13: 00000000ffffffef R14: ffff8880211a0194 R15: ffff8880577d0a00
FS:  00007f25d183e700(0000) GS:ffff8880b9c00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f19c859c028 CR3: 0000000051009000 CR4: 00000000003506f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <TASK>
 __refcount_dec include/linux/refcount.h:344 [inline]
 refcount_dec include/linux/refcount.h:359 [inline]
 ref_tracker_free+0x535/0x6b0 lib/ref_tracker.c:118
 netns_tracker_free include/net/net_namespace.h:327 [inline]
 put_net_track include/net/net_namespace.h:341 [inline]
 tcf_exts_put_net include/net/pkt_cls.h:255 [inline]
 u32_destroy_key.isra.0+0xa7/0x2b0 net/sched/cls_u32.c:394
 u32_change+0xe01/0x3140 net/sched/cls_u32.c:909
 tc_new_tfilter+0x98d/0x2200 net/sched/cls_api.c:2148
 rtnetlink_rcv_msg+0x80d/0xb80 net/core/rtnetlink.c:6016
 netlink_rcv_skb+0x153/0x420 net/netlink/af_netlink.c:2495
 netlink_unicast_kernel net/netlink/af_netlink.c:1319 [inline]
 netlink_unicast+0x543/0x7f0 net/netlink/af_netlink.c:1345
 netlink_sendmsg+0x904/0xe00 net/netlink/af_netlink.c:1921
 sock_sendmsg_nosec net/socket.c:705 [inline]
 sock_sendmsg+0xcf/0x120 net/socket.c:725
 ____sys_sendmsg+0x6e2/0x800 net/socket.c:2413
 ___sys_sendmsg+0xf3/0x170 net/socket.c:2467
 __sys_sendmsg+0xe5/0x1b0 net/socket.c:2496
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7f25d0689049
Code: ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f25d183e168 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
RAX: ffffffffffffffda RBX: 00007f25d079c030 RCX: 00007f25d0689049
RDX: 0000000000000000 RSI: 0000000020000340 RDI: 0000000000000005
RBP: 00007f25d06e308d R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 00007ffd0b752e3f R14: 00007f25d183e300 R15: 0000000000022000
 </TASK>
Fixes 35c55fc156d8 ("cls_u32: use tcf_exts_get_net() before call_rcu()")
Cc Jamal Hadi Salim <[email protected]>
Cc Cong Wang <[email protected]>
Cc Jiri Pirko <[email protected]>
---
 net/sched/cls_u32.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
net/sched/cls_u32.c +10 -6
index cf5649292ee00941e5c4a4d5b11b1c3dc98cce3f..fcba6c43ba509a069c593d525daf2943b4079538 100644
@@ -386,14 +386,19 @@ static int u32_init(struct tcf_proto *tp)
 	return 0;
 }
 
-static int u32_destroy_key(struct tc_u_knode *n, bool free_pf)
+static void __u32_destroy_key(struct tc_u_knode *n)
 {
 	struct tc_u_hnode *ht = rtnl_dereference(n->ht_down);
 
 	tcf_exts_destroy(&n->exts);
-	tcf_exts_put_net(&n->exts);
 	if (ht && --ht->refcnt == 0)
 		kfree(ht);
+	kfree(n);
+}
+
+static void u32_destroy_key(struct tc_u_knode *n, bool free_pf)
+{
+	tcf_exts_put_net(&n->exts);
 #ifdef CONFIG_CLS_U32_PERF
 	if (free_pf)
 		free_percpu(n->pf);
@@ -402,8 +407,7 @@ static int u32_destroy_key(struct tc_u_knode *n, bool free_pf)
 	if (free_pf)
 		free_percpu(n->pcpu_success);
 #endif
-	kfree(n);
-	return 0;
+	__u32_destroy_key(n);
 }
 
 /* u32_delete_key_rcu should be called when free'ing a copied
@@ -900,13 +904,13 @@ static int u32_change(struct net *net, struct sk_buff *in_skb,
 				    extack);
 
 		if (err) {
-			u32_destroy_key(new, false);
+			__u32_destroy_key(new);
 			return err;
 		}
 
 		err = u32_replace_hw_knode(tp, new, flags, extack);
 		if (err) {
-			u32_destroy_key(new, false);
+			__u32_destroy_key(new);
 			return err;
 		}
 
Signature · 1 line
2.35.1.1178.g4f1659d476-goog
maintainer Acked-by on v1 1/2
#2 lore raw

Re: [PATCH net 1/2] net/sched: cls_u32: fix netns refcount changes in u32_change()

Quoted 9 lines · Eric Dumazet

From: Eric Dumazet <[email protected]>

We are now able to detect extra put_net() at the moment they happen, instead of much later in correct code paths.

u32_init_knode() / tcf_exts_init() populates the ->exts.net pointer, but as mentioned in tcf_exts_init(),
the refcount on netns has not been elevated yet.

Quoted 6 lines

The refcount is taken only once tcf_exts_get_net() is called.

So the two u32_destroy_key() calls from u32_change() are attempting to release an invalid reference on the netns.

Looks good to me.

cheers,
jamal

#3 lore raw

PATCH 2/2 net/sched: cls_u32: fix possible leak in u32_init_knode()

In reply to Eric Dumazet

From: Eric Dumazet <[email protected]>

While investigating a related syzbot report,
I found that whenever call to tcf_exts_init()
from u32_init_knode() is failing, we end up
with an elevated refcount on ht->refcnt

To avoid that, only increase the refcount after
all possible errors have been evaluated.

Fixes b9a24bb76bf6 ("net_sched: properly handle failure case of tcf_exts_init()")
Cc Jamal Hadi Salim <[email protected]>
Cc Cong Wang <[email protected]>
Cc Jiri Pirko <[email protected]>
---
 net/sched/cls_u32.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
net/sched/cls_u32.c +4 -4
index fcba6c43ba509a069c593d525daf2943b4079538..4d27300c287c46d11bf9d44f8c66eded9e734581 100644
@@ -815,10 +815,6 @@ static struct tc_u_knode *u32_init_knode(struct net *net, struct tcf_proto *tp,
 	new->flags = n->flags;
 	RCU_INIT_POINTER(new->ht_down, ht);
 
-	/* bump reference count as long as we hold pointer to structure */
-	if (ht)
-		ht->refcnt++;
-
 #ifdef CONFIG_CLS_U32_PERF
 	/* Statistics may be incremented by readers during update
 	 * so we must keep them in tact. When the node is later destroyed
@@ -840,6 +836,10 @@ static struct tc_u_knode *u32_init_knode(struct net *net, struct tcf_proto *tp,
 		return NULL;
 	}
 
+	/* bump reference count as long as we hold pointer to structure */
+	if (ht)
+		ht->refcnt++;
+
 	return new;
 }
 
Signature · 1 line
2.35.1.1178.g4f1659d476-goog
maintainer Acked-by on v1 2/2
#4 lore raw

Re: [PATCH net 2/2] net/sched: cls_u32: fix possible leak in u32_init_knode()

Quoted 7 lines · Eric Dumazet

From: Eric Dumazet<[email protected]>

While investigating a related syzbot report, I found that whenever call to tcf_exts_init() from u32_init_knode() is failing, we end up with an elevated refcount on ht->refcnt

Quoted 8 lines

To avoid that, only increase the refcount after all possible errors have been evaluated.

Fixes b9a24bb76bf6 ("net_sched: properly handle failure case of tcf_exts_init()")
Cc Jamal Hadi Salim<[email protected]>
Cc Cong Wang<[email protected]>
Cc Jiri Pirko<[email protected]>

cheers,
jamal

#5 lore raw
In reply to Eric Dumazet
This series was applied to netdev/net.git (master) by Jakub Kicinski <[email protected]>:

Hello:

This series was applied to netdev/net.git (master) by Jakub Kicinski <[email protected]>:

Quoted 8 lines · On Wed, 13 Apr 2022 10:35:40 -0700 you wrote:

From: Eric Dumazet <[email protected]>

One syzbot report brought my attention to cls_u32.

This series addresses the syzbot report, and an additional issue discovered in code review.

[...]

Here is the summary with links:
- [net,1/2] net/sched: cls_u32: fix netns refcount changes in u32_change()

    https://git.kernel.org/netdev/net/c/3db09e762dc7
  - [net,2/2] net/sched: cls_u32: fix possible leak in u32_init_knode()
    https://git.kernel.org/netdev/net/c/ec5b0f605b10

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html