net: decreasing real_num_tx_queues needs to flush qdisc

Reducing real_num_queues needs to flush the qdisc otherwise
skbs with queue_mappings greater then real_num_tx_queues can
be sent to the underlying driver.

The flow for this is,

dev_queue_xmit()
	dev_pick_tx()
		skb_tx_hash()  => hash using real_num_tx_queues
		skb_set_queue_mapping()
	...
	qdisc_enqueue_root() => enqueue skb on txq from hash
...
dev->real_num_tx_queues -= n
...
sch_direct_xmit()
	dev_hard_start_xmit()
		ndo_start_xmit(skb,dev) => skb queue set with old hash

skbs are enqueued on the qdisc with skb->queue_mapping set
0 < queue_mappings < real_num_tx_queues.  When the driver
decreases real_num_tx_queues skb's may be dequeued from the
qdisc with a queue_mapping greater then real_num_tx_queues.

This fixes a case in ixgbe where this was occurring with DCB
and FCoE. Because the driver is using queue_mapping to map
skbs to tx descriptor rings we can potentially map skbs to
rings that no longer exist.

Signed-off-by: John Fastabend <[email protected]>
Tested-by: Ross Brattain <[email protected]>
Signed-off-by: Jeff Kirsher <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

diff --git a/drivers/net/ixgbe/ixgbe_main.c b/drivers/net/ixgbe/ixgbe_main.c
index a0b33165b989e41d8bcbba45d9e253545816ad04..7b5d9764f317c3c776712009dde301c27d9c3b99 100644 (file)
--- a/drivers/net/ixgbe/ixgbe_main.c
+++ b/drivers/net/ixgbe/ixgbe_main.c
@@ -4001,7 +4001,7 @@ static void ixgbe_set_num_queues(struct ixgbe_adapter *adapter)
 
 done:
        /* Notify the stack of the (possibly) reduced Tx Queue count. */
-       adapter->netdev->real_num_tx_queues = adapter->num_tx_queues;
+       netif_set_real_num_tx_queues(adapter->netdev, adapter->num_tx_queues);
 }
 
 static void ixgbe_acquire_msix_vectors(struct ixgbe_adapter *adapter,

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 40291f3750247dfa12102178a6aed457e7735516..5e6188d9f017aa385717f490f4e1c6b896777d36 100644 (file)
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1656,6 +1656,9 @@ static inline int netif_is_multiqueue(const struct net_device *dev)
        return (dev->num_tx_queues > 1);
 }
 
+extern void netif_set_real_num_tx_queues(struct net_device *dev,
+                                        unsigned int txq);
+
 /* Use this variant when it is known for sure that it
  * is executing from hardware interrupt context or with hardware interrupts
  * disabled.

diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index ba749be1e35463f39cf108a89a61ee43d1260615..433604bb3fe8a70df241dcdc1370a867fd57ed92 100644 (file)
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -313,13 +313,12 @@ extern void qdisc_calculate_pkt_len(struct sk_buff *skb,
 extern void tcf_destroy(struct tcf_proto *tp);
 extern void tcf_destroy_chain(struct tcf_proto **fl);
 
-/* Reset all TX qdiscs of a device.  */
-static inline void qdisc_reset_all_tx(struct net_device *dev)
+/* Reset all TX qdiscs greater then index of a device.  */
+static inline void qdisc_reset_all_tx_gt(struct net_device *dev, unsigned int i)
 {
-       unsigned int i;
        struct Qdisc *qdisc;
 
-       for (i = 0; i < dev->num_tx_queues; i++) {
+       for (; i < dev->num_tx_queues; i++) {
                qdisc = netdev_get_tx_queue(dev, i)->qdisc;
                if (qdisc) {
                        spin_lock_bh(qdisc_lock(qdisc));
@@ -329,6 +328,11 @@ static inline void qdisc_reset_all_tx(struct net_device *dev)
        }
 }
 
+static inline void qdisc_reset_all_tx(struct net_device *dev)
+{
+       qdisc_reset_all_tx_gt(dev, 0);
+}
+
 /* Are all TX queues of the device empty?  */
 static inline bool qdisc_all_tx_empty(const struct net_device *dev)
 {

diff --git a/net/core/dev.c b/net/core/dev.c
index 2b3bf53bc687aab3c9cfb32cce902eaebb3c18da..723a34710ad401ff24afd5b1be1af86505f7da33 100644 (file)
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1553,6 +1553,24 @@ static void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev)
        rcu_read_unlock();
 }
 
+/*
+ * Routine to help set real_num_tx_queues. To avoid skbs mapped to queues
+ * greater then real_num_tx_queues stale skbs on the qdisc must be flushed.
+ */
+void netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq)
+{
+       unsigned int real_num = dev->real_num_tx_queues;
+
+       if (unlikely(txq > dev->num_tx_queues))
+               ;
+       else if (txq > real_num)
+               dev->real_num_tx_queues = txq;
+       else if (txq < real_num) {
+               dev->real_num_tx_queues = txq;
+               qdisc_reset_all_tx_gt(dev, txq);
+       }
+}
+EXPORT_SYMBOL(netif_set_real_num_tx_queues);
 
 static inline void __netif_reschedule(struct Qdisc *q)
 {