[PATCH net] net/packet: remove data races in fanout operations

[PATCH net] net/packet: remove data races in fanout operations

[Eric Dumazet]

From: Eric Dumazet <edumazet@google.com>

af_packet fanout uses RCU rules to ensure f->arr elements
are not dismantled before RCU grace period.

However, it lacks rcu accessors to make sure KCSAN and other tools
wont detect data races. Stupid compilers could also play games.

Fixes: dc99f600698d ("packet: Add fanout support.")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: "Gong, Sishuai" <sishuai@purdue.edu>
Cc: Willem de Bruijn <willemb@google.com>
---
This probably can target net-next instead of net, this is not a problem in practice.

 net/packet/af_packet.c | 15 +++++++++------
 net/packet/internal.h  |  2 +-
 2 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index e24b2841c643a1c2c471416ceb78e82a85f9a740..9611e41c7b8be7c7fa8961d9e6e91f023dd7023f 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -1359,7 +1359,7 @@ static unsigned int fanout_demux_rollover(struct packet_fanout *f,
 	struct packet_sock *po, *po_next, *po_skip = NULL;
 	unsigned int i, j, room = ROOM_NONE;
 
-	po = pkt_sk(f->arr[idx]);
+	po = pkt_sk(rcu_dereference(f->arr[idx]));
 
 	if (try_self) {
 		room = packet_rcv_has_room(po, skb);
@@ -1371,7 +1371,7 @@ static unsigned int fanout_demux_rollover(struct packet_fanout *f,
 
 	i = j = min_t(int, po->rollover->sock, num - 1);
 	do {
-		po_next = pkt_sk(f->arr[i]);
+		po_next = pkt_sk(rcu_dereference(f->arr[i]));
 		if (po_next != po_skip && !READ_ONCE(po_next->pressure) &&
 		    packet_rcv_has_room(po_next, skb) == ROOM_NORMAL) {
 			if (i != j)
@@ -1466,7 +1466,7 @@ static int packet_rcv_fanout(struct sk_buff *skb, struct net_device *dev,
 	if (fanout_has_flag(f, PACKET_FANOUT_FLAG_ROLLOVER))
 		idx = fanout_demux_rollover(f, skb, idx, true, num);
 
-	po = pkt_sk(f->arr[idx]);
+	po = pkt_sk(rcu_dereference(f->arr[idx]));
 	return po->prot_hook.func(skb, dev, &po->prot_hook, orig_dev);
 }
 
@@ -1480,7 +1480,7 @@ static void __fanout_link(struct sock *sk, struct packet_sock *po)
 	struct packet_fanout *f = po->fanout;
 
 	spin_lock(&f->lock);
-	f->arr[f->num_members] = sk;
+	rcu_assign_pointer(f->arr[f->num_members], sk);
 	smp_wmb();
 	f->num_members++;
 	if (f->num_members == 1)
@@ -1495,11 +1495,14 @@ static void __fanout_unlink(struct sock *sk, struct packet_sock *po)
 
 	spin_lock(&f->lock);
 	for (i = 0; i < f->num_members; i++) {
-		if (f->arr[i] == sk)
+		if (rcu_dereference_protected(f->arr[i],
+					      lockdep_is_held(&f->lock)) == sk)
 			break;
 	}
 	BUG_ON(i >= f->num_members);
-	f->arr[i] = f->arr[f->num_members - 1];
+	rcu_assign_pointer(f->arr[i],
+			   rcu_dereference_protected(f->arr[f->num_members - 1],
+						     lockdep_is_held(&f->lock)));
 	f->num_members--;
 	if (f->num_members == 0)
 		__dev_remove_pack(&f->prot_hook);
diff --git a/net/packet/internal.h b/net/packet/internal.h
index 5f61e59ebbffaa25a8fdfe31f79211fe6a755c51..48af35b1aed2565267c0288e013e23ff51f2fcac 100644
--- a/net/packet/internal.h
+++ b/net/packet/internal.h
@@ -94,7 +94,7 @@ struct packet_fanout {
 	spinlock_t		lock;
 	refcount_t		sk_ref;
 	struct packet_type	prot_hook ____cacheline_aligned_in_smp;
-	struct sock		*arr[];
+	struct sock	__rcu	*arr[];
 };
 
 struct packet_rollover {
-- 
2.31.1.295.g9ea45b61b8-goog

Re: [PATCH net] net/packet: remove data races in fanout operations

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net-next.git (refs/heads/master):

On Wed, 14 Apr 2021 12:36:44 -0700 you wrote:
> From: Eric Dumazet <edumazet@google.com>
> 
> af_packet fanout uses RCU rules to ensure f->arr elements
> are not dismantled before RCU grace period.
> 
> However, it lacks rcu accessors to make sure KCSAN and other tools
> wont detect data races. Stupid compilers could also play games.
> 
> [...]

Here is the summary with links:
  - [net] net/packet: remove data races in fanout operations
    https://git.kernel.org/netdev/net-next/c/94f633ea8ade

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html