4 files changed, 89 insertions, 75 deletions

diff --git a/drivers/net/xen-netback/common.h b/drivers/net/xen-netback/common.h
index 4dd7c4a1923b..2532ce85d718 100644
--- a/drivers/net/xen-netback/common.h
+++ b/drivers/net/xen-netback/common.h
@@ -222,6 +222,7 @@ struct xenvif {
 
         /* Queues */
         struct xenvif_queue *queues;
+        unsigned int num_queues; /* active queues, resource allocated */
 
         /* Miscellaneous private stuff. */
         struct net_device *dev;
diff --git a/drivers/net/xen-netback/interface.c b/drivers/net/xen-netback/interface.c
index 852da34b8961..9e97c7ca0ddd 100644
--- a/drivers/net/xen-netback/interface.c
+++ b/drivers/net/xen-netback/interface.c
@@ -137,32 +137,11 @@ static void xenvif_wake_queue_callback(unsigned long data)
         }
 }
 
-static u16 xenvif_select_queue(struct net_device *dev, struct sk_buff *skb,
-                               void *accel_priv, select_queue_fallback_t fallback)
-{
-        unsigned int num_queues = dev->real_num_tx_queues;
-        u32 hash;
-        u16 queue_index;
-
-        /* First, check if there is only one queue to optimise the
-         * single-queue or old frontend scenario.
-         */
-        if (num_queues == 1) {
-                queue_index = 0;
-        } else {
-                /* Use skb_get_hash to obtain an L4 hash if available */
-                hash = skb_get_hash(skb);
-                queue_index = hash % num_queues;
-        }
-
-        return queue_index;
-}
-
 static int xenvif_start_xmit(struct sk_buff *skb, struct net_device *dev)
 {
         struct xenvif *vif = netdev_priv(dev);
         struct xenvif_queue *queue = NULL;
-        unsigned int num_queues = dev->real_num_tx_queues;
+        unsigned int num_queues = vif->num_queues;
         u16 index;
         int min_slots_needed;
 
@@ -225,7 +204,7 @@ static struct net_device_stats *xenvif_get_stats(struct net_device *dev)
 {
         struct xenvif *vif = netdev_priv(dev);
         struct xenvif_queue *queue = NULL;
-        unsigned int num_queues = dev->real_num_tx_queues;
+        unsigned int num_queues = vif->num_queues;
         unsigned long rx_bytes = 0;
         unsigned long rx_packets = 0;
         unsigned long tx_bytes = 0;
@@ -256,7 +235,7 @@ out:
 static void xenvif_up(struct xenvif *vif)
 {
         struct xenvif_queue *queue = NULL;
-        unsigned int num_queues = vif->dev->real_num_tx_queues;
+        unsigned int num_queues = vif->num_queues;
         unsigned int queue_index;
 
         for (queue_index = 0; queue_index < num_queues; ++queue_index) {
@@ -272,7 +251,7 @@ static void xenvif_up(struct xenvif *vif)
 static void xenvif_down(struct xenvif *vif)
 {
         struct xenvif_queue *queue = NULL;
-        unsigned int num_queues = vif->dev->real_num_tx_queues;
+        unsigned int num_queues = vif->num_queues;
         unsigned int queue_index;
 
         for (queue_index = 0; queue_index < num_queues; ++queue_index) {
@@ -379,7 +358,7 @@ static void xenvif_get_ethtool_stats(struct net_device *dev,
                                      struct ethtool_stats *stats, u64 * data)
 {
         struct xenvif *vif = netdev_priv(dev);
-        unsigned int num_queues = dev->real_num_tx_queues;
+        unsigned int num_queues = vif->num_queues;
         int i;
         unsigned int queue_index;
         struct xenvif_stats *vif_stats;
@@ -424,7 +403,6 @@ static const struct net_device_ops xenvif_netdev_ops = {
         .ndo_fix_features = xenvif_fix_features,
         .ndo_set_mac_address = eth_mac_addr,
         .ndo_validate_addr   = eth_validate_addr,
-        .ndo_select_queue = xenvif_select_queue,
 };
 
 struct xenvif *xenvif_alloc(struct device *parent, domid_t domid,
@@ -438,7 +416,7 @@ struct xenvif *xenvif_alloc(struct device *parent, domid_t domid,
         snprintf(name, IFNAMSIZ - 1, "vif%u.%u", domid, handle);
         /* Allocate a netdev with the max. supported number of queues.
          * When the guest selects the desired number, it will be updated
-         * via netif_set_real_num_tx_queues().
+         * via netif_set_real_num_*_queues().
          */
         dev = alloc_netdev_mq(sizeof(struct xenvif), name, ether_setup,
                               xenvif_max_queues);
@@ -458,11 +436,9 @@ struct xenvif *xenvif_alloc(struct device *parent, domid_t domid,
         vif->dev = dev;
         vif->disabled = false;
 
-        /* Start out with no queues. The call below does not require
-         * rtnl_lock() as it happens before register_netdev().
-         */
+        /* Start out with no queues. */
         vif->queues = NULL;
-        netif_set_real_num_tx_queues(dev, 0);
+        vif->num_queues = 0;
 
         dev->netdev_ops = &xenvif_netdev_ops;
         dev->hw_features = NETIF_F_SG |
@@ -677,7 +653,7 @@ static void xenvif_wait_unmap_timeout(struct xenvif_queue *queue,
 void xenvif_disconnect(struct xenvif *vif)
 {
         struct xenvif_queue *queue = NULL;
-        unsigned int num_queues = vif->dev->real_num_tx_queues;
+        unsigned int num_queues = vif->num_queues;
         unsigned int queue_index;
 
         if (netif_carrier_ok(vif->dev))
@@ -724,7 +700,7 @@ void xenvif_deinit_queue(struct xenvif_queue *queue)
 void xenvif_free(struct xenvif *vif)
 {
         struct xenvif_queue *queue = NULL;
-        unsigned int num_queues = vif->dev->real_num_tx_queues;
+        unsigned int num_queues = vif->num_queues;
         unsigned int queue_index;
         /* Here we want to avoid timeout messages if an skb can be legitimately
          * stuck somewhere else. Realistically this could be an another vif's
@@ -748,12 +724,9 @@ void xenvif_free(struct xenvif *vif)
                 xenvif_deinit_queue(queue);
         }
 
-        /* Free the array of queues. The call below does not require
-         * rtnl_lock() because it happens after unregister_netdev().
-         */
-        netif_set_real_num_tx_queues(vif->dev, 0);
         vfree(vif->queues);
         vif->queues = NULL;
+        vif->num_queues = 0;
 
         free_netdev(vif->dev);
 
diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c
index 1844a47636b6..c65b636bcab9 100644
--- a/drivers/net/xen-netback/netback.c
+++ b/drivers/net/xen-netback/netback.c
@@ -1030,14 +1030,21 @@ static int xenvif_tx_check_gop(struct xenvif_queue *queue,
 {
         struct gnttab_map_grant_ref *gop_map = *gopp_map;
         u16 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
+        /* This always points to the shinfo of the skb being checked, which
+         * could be either the first or the one on the frag_list
+         */
         struct skb_shared_info *shinfo = skb_shinfo(skb);
+        /* If this is non-NULL, we are currently checking the frag_list skb, and
+         * this points to the shinfo of the first one
+         */
+        struct skb_shared_info *first_shinfo = NULL;
         int nr_frags = shinfo->nr_frags;
+        const bool sharedslot = nr_frags &&
+                                frag_get_pending_idx(&shinfo->frags[0]) == pending_idx;
         int i, err;
-        struct sk_buff *first_skb = NULL;
 
         /* Check status of header. */
         err = (*gopp_copy)->status;
-        (*gopp_copy)++;
         if (unlikely(err)) {
                 if (net_ratelimit())
                         netdev_dbg(queue->vif->dev,
@@ -1045,8 +1052,12 @@ static int xenvif_tx_check_gop(struct xenvif_queue *queue,
                                    (*gopp_copy)->status,
                                    pending_idx,
                                    (*gopp_copy)->source.u.ref);
-                xenvif_idx_release(queue, pending_idx, XEN_NETIF_RSP_ERROR);
+                /* The first frag might still have this slot mapped */
+                if (!sharedslot)
+                        xenvif_idx_release(queue, pending_idx,
+                                           XEN_NETIF_RSP_ERROR);
         }
+        (*gopp_copy)++;
 
 check_frags:
         for (i = 0; i < nr_frags; i++, gop_map++) {
@@ -1062,8 +1073,19 @@ check_frags:
                                                 pending_idx,
                                                 gop_map->handle);
                         /* Had a previous error? Invalidate this fragment. */
-                        if (unlikely(err))
+                        if (unlikely(err)) {
                                 xenvif_idx_unmap(queue, pending_idx);
+                                /* If the mapping of the first frag was OK, but
+                                 * the header's copy failed, and they are
+                                 * sharing a slot, send an error
+                                 */
+                                if (i == 0 && sharedslot)
+                                        xenvif_idx_release(queue, pending_idx,
+                                                           XEN_NETIF_RSP_ERROR);
+                                else
+                                        xenvif_idx_release(queue, pending_idx,
+                                                           XEN_NETIF_RSP_OKAY);
+                        }
                         continue;
                 }
 
@@ -1075,42 +1097,53 @@ check_frags:
                                    gop_map->status,
                                    pending_idx,
                                    gop_map->ref);
+
                 xenvif_idx_release(queue, pending_idx, XEN_NETIF_RSP_ERROR);
 
                 /* Not the first error? Preceding frags already invalidated. */
                 if (err)
                         continue;
-                /* First error: invalidate preceding fragments. */
+
+                /* First error: if the header haven't shared a slot with the
+                 * first frag, release it as well.
+                 */
+                if (!sharedslot)
+                        xenvif_idx_release(queue,
+                                           XENVIF_TX_CB(skb)->pending_idx,
+                                           XEN_NETIF_RSP_OKAY);
+
+                /* Invalidate preceding fragments of this skb. */
                 for (j = 0; j < i; j++) {
                         pending_idx = frag_get_pending_idx(&shinfo->frags[j]);
                         xenvif_idx_unmap(queue, pending_idx);
+                        xenvif_idx_release(queue, pending_idx,
+                                           XEN_NETIF_RSP_OKAY);
+                }
+
+                /* And if we found the error while checking the frag_list, unmap
+                 * the first skb's frags
+                 */
+                if (first_shinfo) {
+                        for (j = 0; j < first_shinfo->nr_frags; j++) {
+                                pending_idx = frag_get_pending_idx(&first_shinfo->frags[j]);
+                                xenvif_idx_unmap(queue, pending_idx);
+                                xenvif_idx_release(queue, pending_idx,
+                                                   XEN_NETIF_RSP_OKAY);
+                        }
                 }
 
                 /* Remember the error: invalidate all subsequent fragments. */
                 err = newerr;
         }
 
-        if (skb_has_frag_list(skb)) {
-                first_skb = skb;
-                skb = shinfo->frag_list;
-                shinfo = skb_shinfo(skb);
+        if (skb_has_frag_list(skb) && !first_shinfo) {
+                first_shinfo = skb_shinfo(skb);
+                shinfo = skb_shinfo(skb_shinfo(skb)->frag_list);
                 nr_frags = shinfo->nr_frags;
 
                 goto check_frags;
         }
 
-        /* There was a mapping error in the frag_list skb. We have to unmap
-         * the first skb's frags
-         */
-        if (first_skb && err) {
-                int j;
-                shinfo = skb_shinfo(first_skb);
-                for (j = 0; j < shinfo->nr_frags; j++) {
-                        pending_idx = frag_get_pending_idx(&shinfo->frags[j]);
-                        xenvif_idx_unmap(queue, pending_idx);
-                }
-        }
-
         *gopp_map = gop_map;
         return err;
 }
@@ -1518,7 +1551,16 @@ static int xenvif_tx_submit(struct xenvif_queue *queue)
 
                 /* Check the remap error code. */
                 if (unlikely(xenvif_tx_check_gop(queue, skb, &gop_map, &gop_copy))) {
+                        /* If there was an error, xenvif_tx_check_gop is
+                         * expected to release all the frags which were mapped,
+                         * so kfree_skb shouldn't do it again
+                         */
                         skb_shinfo(skb)->nr_frags = 0;
+                        if (skb_has_frag_list(skb)) {
+                                struct sk_buff *nskb =
+                                                skb_shinfo(skb)->frag_list;
+                                skb_shinfo(nskb)->nr_frags = 0;
+                        }
                         kfree_skb(skb);
                         continue;
                 }
@@ -1822,8 +1864,6 @@ void xenvif_idx_unmap(struct xenvif_queue *queue, u16 pending_idx)
                            tx_unmap_op.status);
                 BUG();
         }
-
-        xenvif_idx_release(queue, pending_idx, XEN_NETIF_RSP_OKAY);
 }
 
 static inline int rx_work_todo(struct xenvif_queue *queue)
diff --git a/drivers/net/xen-netback/xenbus.c b/drivers/net/xen-netback/xenbus.c
index 96c63dc2509e..3d85acd84bad 100644
--- a/drivers/net/xen-netback/xenbus.c
+++ b/drivers/net/xen-netback/xenbus.c
@@ -527,9 +527,7 @@ static void connect(struct backend_info *be)
         /* Use the number of queues requested by the frontend */
         be->vif->queues = vzalloc(requested_num_queues *
                                   sizeof(struct xenvif_queue));
-        rtnl_lock();
-        netif_set_real_num_tx_queues(be->vif->dev, requested_num_queues);
-        rtnl_unlock();
+        be->vif->num_queues = requested_num_queues;
 
         for (queue_index = 0; queue_index < requested_num_queues; ++queue_index) {
                 queue = &be->vif->queues[queue_index];
@@ -546,9 +544,7 @@ static void connect(struct backend_info *be)
                          * earlier queues can be destroyed using the regular
                          * disconnect logic.
                          */
-                        rtnl_lock();
-                        netif_set_real_num_tx_queues(be->vif->dev, queue_index);
-                        rtnl_unlock();
+                        be->vif->num_queues = queue_index;
                         goto err;
                 }
 
@@ -561,13 +557,19 @@ static void connect(struct backend_info *be)
                          * and also clean up any previously initialised queues.
                          */
                         xenvif_deinit_queue(queue);
-                        rtnl_lock();
-                        netif_set_real_num_tx_queues(be->vif->dev, queue_index);
-                        rtnl_unlock();
+                        be->vif->num_queues = queue_index;
                         goto err;
                 }
         }
 
+        /* Initialisation completed, tell core driver the number of
+         * active queues.
+         */
+        rtnl_lock();
+        netif_set_real_num_tx_queues(be->vif->dev, requested_num_queues);
+        netif_set_real_num_rx_queues(be->vif->dev, requested_num_queues);
+        rtnl_unlock();
+
         xenvif_carrier_on(be->vif);
 
         unregister_hotplug_status_watch(be);
@@ -582,13 +584,11 @@ static void connect(struct backend_info *be)
         return;
 
 err:
-        if (be->vif->dev->real_num_tx_queues > 0)
+        if (be->vif->num_queues > 0)
                 xenvif_disconnect(be->vif); /* Clean up existing queues */
         vfree(be->vif->queues);
         be->vif->queues = NULL;
-        rtnl_lock();
-        netif_set_real_num_tx_queues(be->vif->dev, 0);
-        rtnl_unlock();
+        be->vif->num_queues = 0;
         return;
 }
 
@@ -596,7 +596,7 @@ err:
 static int connect_rings(struct backend_info *be, struct xenvif_queue *queue)
 {
         struct xenbus_device *dev = be->dev;
-        unsigned int num_queues = queue->vif->dev->real_num_tx_queues;
+        unsigned int num_queues = queue->vif->num_queues;
         unsigned long tx_ring_ref, rx_ring_ref;
         unsigned int tx_evtchn, rx_evtchn;
         int err;