netdev - [PATCH net-next V1] net: flush sd->defer_list on unregister

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: <168198515529.808959.12962138073127060724.stgit@firesoul>
Date:   Thu, 20 Apr 2023 12:05:55 +0200
From:   Jesper Dangaard Brouer <brouer@...hat.com>
To:     netdev@...r.kernel.org, edumazet@...gle.com
Cc:     Jesper Dangaard Brouer <brouer@...hat.com>, pabeni@...hat.com,
        kuba@...nel.org, hawk@...nel.org, davem@...emloft.net,
        lorenzo@...nel.org
Subject: [PATCH net-next V1] net: flush sd->defer_list on unregister_netdevice

When removing a net_device (that use NAPI), the sd->defer_list
system can still hold on to SKBs that have a dst_entry which can
have a netdev_hold reference.

Choose simple solution of flushing the softnet_data defer_list
system as part of unregister_netdevice flush_all_backlogs().

Signed-off-by: Jesper Dangaard Brouer <brouer@...hat.com>
---
 net/core/dev.c |   26 +++++++++++++++++++++-----
 1 file changed, 21 insertions(+), 5 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 3fc4dba71f9d..b63edd2c21e4 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -5780,6 +5780,8 @@ EXPORT_SYMBOL(netif_receive_skb_list);
 
 static DEFINE_PER_CPU(struct work_struct, flush_works);
 
+static void skb_defer_free_flush(struct softnet_data *sd, bool napi_safe);
+
 /* Network device is going away, flush any packets still pending */
 static void flush_backlog(struct work_struct *work)
 {
@@ -5806,6 +5808,8 @@ static void flush_backlog(struct work_struct *work)
 			input_queue_head_incr(sd);
 		}
 	}
+
+	skb_defer_free_flush(sd, false);
 	local_bh_enable();
 }
 
@@ -5824,6 +5828,9 @@ static bool flush_required(int cpu)
 		   !skb_queue_empty_lockless(&sd->process_queue);
 	rps_unlock_irq_enable(sd);
 
+	if (READ_ONCE(sd->defer_list))
+		do_flush = true;
+
 	return do_flush;
 #endif
 	/* without RPS we can't safely check input_pkt_queue: during a
@@ -6623,23 +6630,32 @@ static int napi_threaded_poll(void *data)
 	return 0;
 }
 
-static void skb_defer_free_flush(struct softnet_data *sd)
+static void skb_defer_free_flush(struct softnet_data *sd, bool napi_safe)
 {
 	struct sk_buff *skb, *next;
+	unsigned long flags;
 
 	/* Paired with WRITE_ONCE() in skb_attempt_defer_free() */
 	if (!READ_ONCE(sd->defer_list))
 		return;
 
-	spin_lock_irq(&sd->defer_lock);
+	if (napi_safe)
+		spin_lock_irq(&sd->defer_lock);
+	else
+		spin_lock_irqsave(&sd->defer_lock, flags);
+
 	skb = sd->defer_list;
 	sd->defer_list = NULL;
 	sd->defer_count = 0;
-	spin_unlock_irq(&sd->defer_lock);
+
+	if (napi_safe)
+		spin_unlock_irq(&sd->defer_lock);
+	else
+		spin_unlock_irqrestore(&sd->defer_lock, flags);
 
 	while (skb != NULL) {
 		next = skb->next;
-		napi_consume_skb(skb, 1);
+		napi_consume_skb(skb, napi_safe);
 		skb = next;
 	}
 }
@@ -6662,7 +6678,7 @@ static __latent_entropy void net_rx_action(struct softirq_action *h)
 	for (;;) {
 		struct napi_struct *n;
 
-		skb_defer_free_flush(sd);
+		skb_defer_free_flush(sd, true);
 
 		if (list_empty(&list)) {
 			if (list_empty(&repoll)) {