leandrof/linux
1
0
Fork 0
forked from Minki/linux
Code Issues Pull Requests Packages Projects Releases Wiki Activity

tap: Abstract type of virtual interface from tap implementation

Browse Source 
macvlan object is re-structured to hold tap related elements in a separate
entity, tap_dev. Upon NETDEV_REGISTER device_event, tap_dev is registered with
idr and fetched again on tap_open. Few of the tap functions are modified to
accepted tap_dev as argument. tap_dev object includes callbacks to be used by
underlying virtual interface to take care of tx and rx accounting.

Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Sainath Grandhi 2017-02-10 16:03:49 -08:00 committed by David S. Miller
parent ebc05ba7e8
commit 6fe3faf867
4 changed files with 229 additions and 165 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
drivers/net/macvlan.c
View File

@ -1525,7 +1525,6 @@ static const struct nla_policy macvlan_policy[IFLA_MACVLAN_MAX + 1] = {
int macvlan_link_register(struct rtnl_link_ops *ops) int macvlan_link_register(struct rtnl_link_ops *ops)
{ {
/* common fields */ /* common fields */
ops->priv_size = sizeof(struct macvlan_dev);
ops->validate = macvlan_validate; ops->validate = macvlan_validate;
ops->maxtype = IFLA_MACVLAN_MAX; ops->maxtype = IFLA_MACVLAN_MAX;
ops->policy = macvlan_policy; ops->policy = macvlan_policy;
@ -1548,6 +1547,7 @@ static struct rtnl_link_ops macvlan_link_ops = {
.newlink = macvlan_newlink, .newlink = macvlan_newlink,
.dellink = macvlan_dellink, .dellink = macvlan_dellink,
.get_link_net = macvlan_get_link_net, .get_link_net = macvlan_get_link_net,
.priv_size = sizeof(struct macvlan_dev),
}; };
static int macvlan_device_event(struct notifier_block *unused, static int macvlan_device_event(struct notifier_block *unused,

71
drivers/net/macvtap_main.c
View File

@ -24,6 +24,11 @@
#include <linux/virtio_net.h> #include <linux/virtio_net.h>
#include <linux/skb_array.h> #include <linux/skb_array.h>
struct macvtap_dev {
struct macvlan_dev vlan;
struct tap_dev tap;
};
/* /*
* Variables for dealing with macvtaps device numbers. * Variables for dealing with macvtaps device numbers.
*/ */
@ -46,22 +51,55 @@ static struct cdev macvtap_cdev;
#define TUN_OFFLOADS (NETIF_F_HW_CSUM | NETIF_F_TSO_ECN | NETIF_F_TSO | \ #define TUN_OFFLOADS (NETIF_F_HW_CSUM | NETIF_F_TSO_ECN | NETIF_F_TSO | \
NETIF_F_TSO6 | NETIF_F_UFO) NETIF_F_TSO6 | NETIF_F_UFO)
static void macvtap_count_tx_dropped(struct tap_dev *tap)
{
struct macvtap_dev *vlantap = container_of(tap, struct macvtap_dev, tap);
struct macvlan_dev *vlan = &vlantap->vlan;
this_cpu_inc(vlan->pcpu_stats->tx_dropped);
}
static void macvtap_count_rx_dropped(struct tap_dev *tap)
{
struct macvtap_dev *vlantap = container_of(tap, struct macvtap_dev, tap);
struct macvlan_dev *vlan = &vlantap->vlan;
macvlan_count_rx(vlan, 0, 0, 0);
}
static void macvtap_update_features(struct tap_dev *tap,
netdev_features_t features)
{
struct macvtap_dev *vlantap = container_of(tap, struct macvtap_dev, tap);
struct macvlan_dev *vlan = &vlantap->vlan;
vlan->set_features = features;
netdev_update_features(vlan->dev);
}
static int macvtap_newlink(struct net *src_net, static int macvtap_newlink(struct net *src_net,
struct net_device *dev, struct net_device *dev,
struct nlattr *tb[], struct nlattr *tb[],
struct nlattr *data[]) struct nlattr *data[])
{ {
struct macvlan_dev *vlan = netdev_priv(dev); struct macvtap_dev *vlantap = netdev_priv(dev);
int err; int err;
INIT_LIST_HEAD(&vlan->queue_list); INIT_LIST_HEAD(&vlantap->tap.queue_list);
/* Since macvlan supports all offloads by default, make /* Since macvlan supports all offloads by default, make
* tap support all offloads also. * tap support all offloads also.
*/ */
vlan->tap_features = TUN_OFFLOADS; vlantap->tap.tap_features = TUN_OFFLOADS;
err = netdev_rx_handler_register(dev, tap_handle_frame, vlan); /* Register callbacks for rx/tx drops accounting and updating
* net_device features
*/
vlantap->tap.count_tx_dropped = macvtap_count_tx_dropped;
vlantap->tap.count_rx_dropped = macvtap_count_rx_dropped;
vlantap->tap.update_features = macvtap_update_features;
err = netdev_rx_handler_register(dev, tap_handle_frame, &vlantap->tap);
if (err) if (err)
return err; return err;
@ -74,14 +112,18 @@ static int macvtap_newlink(struct net *src_net,
return err; return err;
} }
vlantap->tap.dev = vlantap->vlan.dev;
return 0; return 0;
} }
static void macvtap_dellink(struct net_device *dev, static void macvtap_dellink(struct net_device *dev,
struct list_head *head) struct list_head *head)
{ {
struct macvtap_dev *vlantap = netdev_priv(dev);
netdev_rx_handler_unregister(dev); netdev_rx_handler_unregister(dev);
tap_del_queues(dev); tap_del_queues(&vlantap->tap);
macvlan_dellink(dev, head); macvlan_dellink(dev, head);
} }
@ -96,13 +138,14 @@ static struct rtnl_link_ops macvtap_link_ops __read_mostly = {
.setup = macvtap_setup, .setup = macvtap_setup,
.newlink = macvtap_newlink, .newlink = macvtap_newlink,
.dellink = macvtap_dellink, .dellink = macvtap_dellink,
.priv_size = sizeof(struct macvtap_dev),
}; };
static int macvtap_device_event(struct notifier_block *unused, static int macvtap_device_event(struct notifier_block *unused,
unsigned long event, void *ptr) unsigned long event, void *ptr)
{ {
struct net_device *dev = netdev_notifier_info_to_dev(ptr); struct net_device *dev = netdev_notifier_info_to_dev(ptr);
struct macvlan_dev *vlan; struct macvtap_dev *vlantap;
struct device *classdev; struct device *classdev;
dev_t devt; dev_t devt;
int err; int err;
@ -112,7 +155,7 @@ static int macvtap_device_event(struct notifier_block *unused,
return NOTIFY_DONE; return NOTIFY_DONE;
snprintf(tap_name, IFNAMSIZ, "tap%d", dev->ifindex); snprintf(tap_name, IFNAMSIZ, "tap%d", dev->ifindex);
vlan = netdev_priv(dev); vlantap = netdev_priv(dev);
switch (event) { switch (event) {
case NETDEV_REGISTER: case NETDEV_REGISTER:
@ -120,15 +163,15 @@ static int macvtap_device_event(struct notifier_block *unused,
* been registered but before register_netdevice has * been registered but before register_netdevice has
* finished running. * finished running.
*/ */
err = tap_get_minor(vlan); err = tap_get_minor(&vlantap->tap);
if (err) if (err)
return notifier_from_errno(err); return notifier_from_errno(err);
devt = MKDEV(MAJOR(macvtap_major), vlan->minor); devt = MKDEV(MAJOR(macvtap_major), vlantap->tap.minor);
classdev = device_create(&macvtap_class, &dev->dev, devt, classdev = device_create(&macvtap_class, &dev->dev, devt,
dev, tap_name); dev, tap_name);
if (IS_ERR(classdev)) { if (IS_ERR(classdev)) {
tap_free_minor(vlan); tap_free_minor(&vlantap->tap);
return notifier_from_errno(PTR_ERR(classdev)); return notifier_from_errno(PTR_ERR(classdev));
} }
err = sysfs_create_link(&dev->dev.kobj, &classdev->kobj, err = sysfs_create_link(&dev->dev.kobj, &classdev->kobj,
@ -138,15 +181,15 @@ static int macvtap_device_event(struct notifier_block *unused,
break; break;
case NETDEV_UNREGISTER: case NETDEV_UNREGISTER:
/* vlan->minor == 0 if NETDEV_REGISTER above failed */ /* vlan->minor == 0 if NETDEV_REGISTER above failed */
if (vlan->minor == 0) if (vlantap->tap.minor == 0)
break; break;
sysfs_remove_link(&dev->dev.kobj, tap_name); sysfs_remove_link(&dev->dev.kobj, tap_name);
devt = MKDEV(MAJOR(macvtap_major), vlan->minor); devt = MKDEV(MAJOR(macvtap_major), vlantap->tap.minor);
device_destroy(&macvtap_class, devt); device_destroy(&macvtap_class, devt);
tap_free_minor(vlan); tap_free_minor(&vlantap->tap);
break; break;
case NETDEV_CHANGE_TX_QUEUE_LEN: case NETDEV_CHANGE_TX_QUEUE_LEN:
if (tap_queue_resize(vlan)) if (tap_queue_resize(&vlantap->tap))
return NOTIFY_BAD; return NOTIFY_BAD;
break; break;
} }

264
drivers/net/tap.c
View File

@ -1,5 +1,5 @@
#include <linux/etherdevice.h> #include <linux/etherdevice.h>
#include <linux/if_macvlan.h> #include <linux/if_tap.h>
#include <linux/if_vlan.h> #include <linux/if_vlan.h>
#include <linux/interrupt.h> #include <linux/interrupt.h>
#include <linux/nsproxy.h> #include <linux/nsproxy.h>
@ -23,30 +23,6 @@
#include <linux/virtio_net.h> #include <linux/virtio_net.h>
#include <linux/skb_array.h> #include <linux/skb_array.h>
/*
* A tap queue is the central object of this driver, it connects
* an open character device to a macvlan interface. There can be
* multiple queues on one interface, which map back to queues
* implemented in hardware on the underlying device.
*
* tap_proto is used to allocate queues through the sock allocation
* mechanism.
*
*/
struct tap_queue {
struct sock sk;
struct socket sock;
struct socket_wq wq;
int vnet_hdr_sz;
struct macvlan_dev __rcu *vlan;
struct file *file;
unsigned int flags;
u16 queue_index;
bool enabled;
struct list_head next;
struct skb_array skb_array;
};
#define TAP_IFFEATURES (IFF_VNET_HDR | IFF_MULTI_QUEUE) #define TAP_IFFEATURES (IFF_VNET_HDR | IFF_MULTI_QUEUE)
#define TAP_VNET_LE 0x80000000 #define TAP_VNET_LE 0x80000000
@ -137,7 +113,7 @@ static const struct proto_ops tap_socket_ops;
#define RX_OFFLOADS (NETIF_F_GRO | NETIF_F_LRO) #define RX_OFFLOADS (NETIF_F_GRO | NETIF_F_LRO)
#define TAP_FEATURES (NETIF_F_GSO | NETIF_F_SG | NETIF_F_FRAGLIST) #define TAP_FEATURES (NETIF_F_GSO | NETIF_F_SG | NETIF_F_FRAGLIST)
static struct macvlan_dev *tap_get_vlan_rcu(const struct net_device *dev) static struct tap_dev *tap_dev_get_rcu(const struct net_device *dev)
{ {
return rcu_dereference(dev->rx_handler_data); return rcu_dereference(dev->rx_handler_data);
} }
@ -159,10 +135,9 @@ static struct macvlan_dev *tap_get_vlan_rcu(const struct net_device *dev)
* when both our references and any pending SKBs are gone. * when both our references and any pending SKBs are gone.
*/ */
static int tap_enable_queue(struct net_device *dev, struct file *file, static int tap_enable_queue(struct tap_dev *tap, struct file *file,
struct tap_queue *q) struct tap_queue *q)
{ {
struct macvlan_dev *vlan = netdev_priv(dev);
int err = -EINVAL; int err = -EINVAL;
ASSERT_RTNL(); ASSERT_RTNL();
@ -171,62 +146,60 @@ static int tap_enable_queue(struct net_device *dev, struct file *file,
goto out; goto out;
err = 0; err = 0;
rcu_assign_pointer(vlan->taps[vlan->numvtaps], q); rcu_assign_pointer(tap->taps[tap->numvtaps], q);
q->queue_index = vlan->numvtaps; q->queue_index = tap->numvtaps;
q->enabled = true; q->enabled = true;
vlan->numvtaps++; tap->numvtaps++;
out: out:
return err; return err;
} }
/* Requires RTNL */ /* Requires RTNL */
static int tap_set_queue(struct net_device *dev, struct file *file, static int tap_set_queue(struct tap_dev *tap, struct file *file,
struct tap_queue *q) struct tap_queue *q)
{ {
struct macvlan_dev *vlan = netdev_priv(dev); if (tap->numqueues == MAX_TAP_QUEUES)
if (vlan->numqueues == MAX_TAP_QUEUES)
return -EBUSY; return -EBUSY;
rcu_assign_pointer(q->vlan, vlan); rcu_assign_pointer(q->tap, tap);
rcu_assign_pointer(vlan->taps[vlan->numvtaps], q); rcu_assign_pointer(tap->taps[tap->numvtaps], q);
sock_hold(&q->sk); sock_hold(&q->sk);
q->file = file; q->file = file;
q->queue_index = vlan->numvtaps; q->queue_index = tap->numvtaps;
q->enabled = true; q->enabled = true;
file->private_data = q; file->private_data = q;
list_add_tail(&q->next, &vlan->queue_list); list_add_tail(&q->next, &tap->queue_list);
vlan->numvtaps++; tap->numvtaps++;
vlan->numqueues++; tap->numqueues++;
return 0; return 0;
} }
static int tap_disable_queue(struct tap_queue *q) static int tap_disable_queue(struct tap_queue *q)
{ {
struct macvlan_dev *vlan; struct tap_dev *tap;
struct tap_queue *nq; struct tap_queue *nq;
ASSERT_RTNL(); ASSERT_RTNL();
if (!q->enabled) if (!q->enabled)
return -EINVAL; return -EINVAL;
vlan = rtnl_dereference(q->vlan); tap = rtnl_dereference(q->tap);
if (vlan) { if (tap) {
int index = q->queue_index; int index = q->queue_index;
BUG_ON(index >= vlan->numvtaps); BUG_ON(index >= tap->numvtaps);
nq = rtnl_dereference(vlan->taps[vlan->numvtaps - 1]); nq = rtnl_dereference(tap->taps[tap->numvtaps - 1]);
nq->queue_index = index; nq->queue_index = index;
rcu_assign_pointer(vlan->taps[index], nq); rcu_assign_pointer(tap->taps[index], nq);
RCU_INIT_POINTER(vlan->taps[vlan->numvtaps - 1], NULL); RCU_INIT_POINTER(tap->taps[tap->numvtaps - 1], NULL);
q->enabled = false; q->enabled = false;
vlan->numvtaps--; tap->numvtaps--;
} }
return 0; return 0;
@ -242,17 +215,17 @@ static int tap_disable_queue(struct tap_queue *q)
*/ */
static void tap_put_queue(struct tap_queue *q) static void tap_put_queue(struct tap_queue *q)
{ {
struct macvlan_dev *vlan; struct tap_dev *tap;
rtnl_lock(); rtnl_lock();
vlan = rtnl_dereference(q->vlan); tap = rtnl_dereference(q->tap);
if (vlan) { if (tap) {
if (q->enabled) if (q->enabled)
BUG_ON(tap_disable_queue(q)); BUG_ON(tap_disable_queue(q));
vlan->numqueues--; tap->numqueues--;
RCU_INIT_POINTER(q->vlan, NULL); RCU_INIT_POINTER(q->tap, NULL);
sock_put(&q->sk); sock_put(&q->sk);
list_del_init(&q->next); list_del_init(&q->next);
} }
@ -270,17 +243,16 @@ static void tap_put_queue(struct tap_queue *q)
* Cache vlan->numvtaps since it can become zero during the execution * Cache vlan->numvtaps since it can become zero during the execution
* of this function. * of this function.
*/ */
static struct tap_queue *tap_get_queue(struct net_device *dev, static struct tap_queue *tap_get_queue(struct tap_dev *tap,
struct sk_buff *skb) struct sk_buff *skb)
{ {
struct macvlan_dev *vlan = netdev_priv(dev); struct tap_queue *queue = NULL;
struct tap_queue *tap = NULL;
/* Access to taps array is protected by rcu, but access to numvtaps /* Access to taps array is protected by rcu, but access to numvtaps
* isn't. Below we use it to lookup a queue, but treat it as a hint * isn't. Below we use it to lookup a queue, but treat it as a hint
* and validate that the result isn't NULL - in case we are * and validate that the result isn't NULL - in case we are
* racing against queue removal. * racing against queue removal.
*/ */
int numvtaps = ACCESS_ONCE(vlan->numvtaps); int numvtaps = ACCESS_ONCE(tap->numvtaps);
__u32 rxq; __u32 rxq;
if (!numvtaps) if (!numvtaps)
@ -292,7 +264,7 @@ static struct tap_queue *tap_get_queue(struct net_device *dev,
/* Check if we can use flow to select a queue */ /* Check if we can use flow to select a queue */
rxq = skb_get_hash(skb); rxq = skb_get_hash(skb);
if (rxq) { if (rxq) {
tap = rcu_dereference(vlan->taps[rxq % numvtaps]); queue = rcu_dereference(tap->taps[rxq % numvtaps]);
goto out; goto out;
} }
@ -302,14 +274,14 @@ static struct tap_queue *tap_get_queue(struct net_device *dev,
while (unlikely(rxq >= numvtaps)) while (unlikely(rxq >= numvtaps))
rxq -= numvtaps; rxq -= numvtaps;
tap = rcu_dereference(vlan->taps[rxq]); queue = rcu_dereference(tap->taps[rxq]);
goto out; goto out;
} }
single: single:
tap = rcu_dereference(vlan->taps[0]); queue = rcu_dereference(tap->taps[0]);
out: out:
return tap; return queue;
} }
/* /*
@ -317,39 +289,38 @@ out:
* that it holds on all queues and safely set the pointer * that it holds on all queues and safely set the pointer
* from the queues to NULL. * from the queues to NULL.
*/ */
void tap_del_queues(struct net_device *dev) void tap_del_queues(struct tap_dev *tap)
{ {
struct macvlan_dev *vlan = netdev_priv(dev);
struct tap_queue *q, *tmp; struct tap_queue *q, *tmp;
ASSERT_RTNL(); ASSERT_RTNL();
list_for_each_entry_safe(q, tmp, &vlan->queue_list, next) { list_for_each_entry_safe(q, tmp, &tap->queue_list, next) {
list_del_init(&q->next); list_del_init(&q->next);
RCU_INIT_POINTER(q->vlan, NULL); RCU_INIT_POINTER(q->tap, NULL);
if (q->enabled) if (q->enabled)
vlan->numvtaps--; tap->numvtaps--;
vlan->numqueues--; tap->numqueues--;
sock_put(&q->sk); sock_put(&q->sk);
} }
BUG_ON(vlan->numvtaps); BUG_ON(tap->numvtaps);
BUG_ON(vlan->numqueues); BUG_ON(tap->numqueues);
/* guarantee that any future tap_set_queue will fail */ /* guarantee that any future tap_set_queue will fail */
vlan->numvtaps = MAX_TAP_QUEUES; tap->numvtaps = MAX_TAP_QUEUES;
} }
rx_handler_result_t tap_handle_frame(struct sk_buff **pskb) rx_handler_result_t tap_handle_frame(struct sk_buff **pskb)
{ {
struct sk_buff *skb = *pskb; struct sk_buff *skb = *pskb;
struct net_device *dev = skb->dev; struct net_device *dev = skb->dev;
struct macvlan_dev *vlan; struct tap_dev *tap;
struct tap_queue *q; struct tap_queue *q;
netdev_features_t features = TAP_FEATURES; netdev_features_t features = TAP_FEATURES;
vlan = tap_get_vlan_rcu(dev); tap = tap_dev_get_rcu(dev);
if (!vlan) if (!tap)
return RX_HANDLER_PASS; return RX_HANDLER_PASS;
q = tap_get_queue(dev, skb); q = tap_get_queue(tap, skb);
if (!q) if (!q)
return RX_HANDLER_PASS; return RX_HANDLER_PASS;
@ -363,7 +334,7 @@ rx_handler_result_t tap_handle_frame(struct sk_buff **pskb)
* enabled. * enabled.
*/ */
if (q->flags & IFF_VNET_HDR) if (q->flags & IFF_VNET_HDR)
features |= vlan->tap_features; features |= tap->tap_features;
if (netif_needs_gso(skb, features)) { if (netif_needs_gso(skb, features)) {
struct sk_buff *segs = __skb_gso_segment(skb, features, false); struct sk_buff *segs = __skb_gso_segment(skb, features, false);
@ -408,50 +379,51 @@ wake_up:
drop: drop:
/* Count errors/drops only here, thus don't care about args. */ /* Count errors/drops only here, thus don't care about args. */
macvlan_count_rx(vlan, 0, 0, 0); if (tap->count_rx_dropped)
tap->count_rx_dropped(tap);
kfree_skb(skb); kfree_skb(skb);
return RX_HANDLER_CONSUMED; return RX_HANDLER_CONSUMED;
} }
int tap_get_minor(struct macvlan_dev *vlan) int tap_get_minor(struct tap_dev *tap)
{ {
int retval = -ENOMEM; int retval = -ENOMEM;
mutex_lock(&macvtap_major.minor_lock); mutex_lock(&macvtap_major.minor_lock);
retval = idr_alloc(&macvtap_major.minor_idr, vlan, 1, TAP_NUM_DEVS, GFP_KERNEL); retval = idr_alloc(&macvtap_major.minor_idr, tap, 1, TAP_NUM_DEVS, GFP_KERNEL);
if (retval >= 0) { if (retval >= 0) {
vlan->minor = retval; tap->minor = retval;
} else if (retval == -ENOSPC) { } else if (retval == -ENOSPC) {
netdev_err(vlan->dev, "Too many tap devices\n"); netdev_err(tap->dev, "Too many tap devices\n");
retval = -EINVAL; retval = -EINVAL;
} }
mutex_unlock(&macvtap_major.minor_lock); mutex_unlock(&macvtap_major.minor_lock);
return retval < 0 ? retval : 0; return retval < 0 ? retval : 0;
} }
void tap_free_minor(struct macvlan_dev *vlan) void tap_free_minor(struct tap_dev *tap)
{ {
mutex_lock(&macvtap_major.minor_lock); mutex_lock(&macvtap_major.minor_lock);
if (vlan->minor) { if (tap->minor) {
idr_remove(&macvtap_major.minor_idr, vlan->minor); idr_remove(&macvtap_major.minor_idr, tap->minor);
vlan->minor = 0; tap->minor = 0;
} }
mutex_unlock(&macvtap_major.minor_lock); mutex_unlock(&macvtap_major.minor_lock);
} }
static struct net_device *dev_get_by_tap_minor(int minor) static struct tap_dev *dev_get_by_tap_minor(int minor)
{ {
struct net_device *dev = NULL; struct net_device *dev = NULL;
struct macvlan_dev *vlan; struct tap_dev *tap;
mutex_lock(&macvtap_major.minor_lock); mutex_lock(&macvtap_major.minor_lock);
vlan = idr_find(&macvtap_major.minor_idr, minor); tap = idr_find(&macvtap_major.minor_idr, minor);
if (vlan) { if (tap) {
dev = vlan->dev; dev = tap->dev;
dev_hold(dev); dev_hold(dev);
} }
mutex_unlock(&macvtap_major.minor_lock); mutex_unlock(&macvtap_major.minor_lock);
return dev; return tap;
} }
static void tap_sock_write_space(struct sock *sk) static void tap_sock_write_space(struct sock *sk)
@ -477,13 +449,13 @@ static void tap_sock_destruct(struct sock *sk)
static int tap_open(struct inode *inode, struct file *file) static int tap_open(struct inode *inode, struct file *file)
{ {
struct net *net = current->nsproxy->net_ns; struct net *net = current->nsproxy->net_ns;
struct net_device *dev; struct tap_dev *tap;
struct tap_queue *q; struct tap_queue *q;
int err = -ENODEV; int err = -ENODEV;
rtnl_lock(); rtnl_lock();
dev = dev_get_by_tap_minor(iminor(inode)); tap = dev_get_by_tap_minor(iminor(inode));
if (!dev) if (!tap)
goto err; goto err;
err = -ENOMEM; err = -ENOMEM;
@ -511,18 +483,18 @@ static int tap_open(struct inode *inode, struct file *file)
* The macvlan supports zerocopy iff the lower device supports zero * The macvlan supports zerocopy iff the lower device supports zero
* copy so we don't have to look at the lower device directly. * copy so we don't have to look at the lower device directly.
*/ */
if ((dev->features & NETIF_F_HIGHDMA) && (dev->features & NETIF_F_SG)) if ((tap->dev->features & NETIF_F_HIGHDMA) && (tap->dev->features & NETIF_F_SG))
sock_set_flag(&q->sk, SOCK_ZEROCOPY); sock_set_flag(&q->sk, SOCK_ZEROCOPY);
err = -ENOMEM; err = -ENOMEM;
if (skb_array_init(&q->skb_array, dev->tx_queue_len, GFP_KERNEL)) if (skb_array_init(&q->skb_array, tap->dev->tx_queue_len, GFP_KERNEL))
goto err_array; goto err_array;
err = tap_set_queue(dev, file, q); err = tap_set_queue(tap, file, q);
if (err) if (err)
goto err_queue; goto err_queue;
dev_put(dev); dev_put(tap->dev);
rtnl_unlock(); rtnl_unlock();
return err; return err;
@ -532,8 +504,8 @@ err_queue:
err_array: err_array:
sock_put(&q->sk); sock_put(&q->sk);
err: err:
if (dev) if (tap)
dev_put(dev); dev_put(tap->dev);
rtnl_unlock(); rtnl_unlock();
return err; return err;
@ -601,7 +573,7 @@ static ssize_t tap_get_user(struct tap_queue *q, struct msghdr *m,
{ {
int good_linear = SKB_MAX_HEAD(TAP_RESERVE); int good_linear = SKB_MAX_HEAD(TAP_RESERVE);
struct sk_buff *skb; struct sk_buff *skb;
struct macvlan_dev *vlan; struct tap_dev *tap;
unsigned long total_len = iov_iter_count(from); unsigned long total_len = iov_iter_count(from);
unsigned long len = total_len; unsigned long len = total_len;
int err; int err;
@ -698,7 +670,7 @@ static ssize_t tap_get_user(struct tap_queue *q, struct msghdr *m,
skb_set_network_header(skb, depth); skb_set_network_header(skb, depth);
rcu_read_lock(); rcu_read_lock();
vlan = rcu_dereference(q->vlan); tap = rcu_dereference(q->tap);
/* copy skb_ubuf_info for callback when skb has no error */ /* copy skb_ubuf_info for callback when skb has no error */
if (zerocopy) { if (zerocopy) {
skb_shinfo(skb)->destructor_arg = m->msg_control; skb_shinfo(skb)->destructor_arg = m->msg_control;
@ -709,8 +681,8 @@ static ssize_t tap_get_user(struct tap_queue *q, struct msghdr *m,
uarg->callback(uarg, false); uarg->callback(uarg, false);
} }
if (vlan) { if (tap) {
skb->dev = vlan->dev; skb->dev = tap->dev;
dev_queue_xmit(skb); dev_queue_xmit(skb);
} else { } else {
kfree_skb(skb); kfree_skb(skb);
@ -724,9 +696,9 @@ err_kfree:
err: err:
rcu_read_lock(); rcu_read_lock();
vlan = rcu_dereference(q->vlan); tap = rcu_dereference(q->tap);
if (vlan) if (tap && tap->count_tx_dropped)
this_cpu_inc(vlan->pcpu_stats->tx_dropped); tap->count_tx_dropped(tap);
rcu_read_unlock(); rcu_read_unlock();
return err; return err;
@ -853,55 +825,55 @@ static ssize_t tap_read_iter(struct kiocb *iocb, struct iov_iter *to)
return ret; return ret;
} }
static struct macvlan_dev *tap_get_vlan(struct tap_queue *q) static struct tap_dev *tap_get_tap_dev(struct tap_queue *q)
{ {
struct macvlan_dev *vlan; struct tap_dev *tap;
ASSERT_RTNL(); ASSERT_RTNL();
vlan = rtnl_dereference(q->vlan); tap = rtnl_dereference(q->tap);
if (vlan) if (tap)
dev_hold(vlan->dev); dev_hold(tap->dev);
return vlan; return tap;
} }
static void tap_put_vlan(struct macvlan_dev *vlan) static void tap_put_tap_dev(struct tap_dev *tap)
{ {
dev_put(vlan->dev); dev_put(tap->dev);
} }
static int tap_ioctl_set_queue(struct file *file, unsigned int flags) static int tap_ioctl_set_queue(struct file *file, unsigned int flags)
{ {
struct tap_queue *q = file->private_data; struct tap_queue *q = file->private_data;
struct macvlan_dev *vlan; struct tap_dev *tap;
int ret; int ret;
vlan = tap_get_vlan(q); tap = tap_get_tap_dev(q);
if (!vlan) if (!tap)
return -EINVAL; return -EINVAL;
if (flags & IFF_ATTACH_QUEUE) if (flags & IFF_ATTACH_QUEUE)
ret = tap_enable_queue(vlan->dev, file, q); ret = tap_enable_queue(tap, file, q);
else if (flags & IFF_DETACH_QUEUE) else if (flags & IFF_DETACH_QUEUE)
ret = tap_disable_queue(q); ret = tap_disable_queue(q);
else else
ret = -EINVAL; ret = -EINVAL;
tap_put_vlan(vlan); tap_put_tap_dev(tap);
return ret; return ret;
} }
static int set_offload(struct tap_queue *q, unsigned long arg) static int set_offload(struct tap_queue *q, unsigned long arg)
{ {
struct macvlan_dev *vlan; struct tap_dev *tap;
netdev_features_t features; netdev_features_t features;
netdev_features_t feature_mask = 0; netdev_features_t feature_mask = 0;
vlan = rtnl_dereference(q->vlan); tap = rtnl_dereference(q->tap);
if (!vlan) if (!tap)
return -ENOLINK; return -ENOLINK;
features = vlan->dev->features; features = tap->dev->features;
if (arg & TUN_F_CSUM) { if (arg & TUN_F_CSUM) {
feature_mask = NETIF_F_HW_CSUM; feature_mask = NETIF_F_HW_CSUM;
@ -935,9 +907,9 @@ static int set_offload(struct tap_queue *q, unsigned long arg)
/* tap_features are the same as features on tun/tap and /* tap_features are the same as features on tun/tap and
* reflect user expectations. * reflect user expectations.
*/ */
vlan->tap_features = feature_mask; tap->tap_features = feature_mask;
vlan->set_features = features; if (tap->update_features)
netdev_update_features(vlan->dev); tap->update_features(tap, features);
return 0; return 0;
} }
@ -949,7 +921,7 @@ static long tap_ioctl(struct file *file, unsigned int cmd,
unsigned long arg) unsigned long arg)
{ {
struct tap_queue *q = file->private_data; struct tap_queue *q = file->private_data;
struct macvlan_dev *vlan; struct tap_dev *tap;
void __user *argp = (void __user *)arg; void __user *argp = (void __user *)arg;
struct ifreq __user *ifr = argp; struct ifreq __user *ifr = argp;
unsigned int __user *up = argp; unsigned int __user *up = argp;
@ -975,18 +947,18 @@ static long tap_ioctl(struct file *file, unsigned int cmd,
case TUNGETIFF: case TUNGETIFF:
rtnl_lock(); rtnl_lock();
vlan = tap_get_vlan(q); tap = tap_get_tap_dev(q);
if (!vlan) { if (!tap) {
rtnl_unlock(); rtnl_unlock();
return -ENOLINK; return -ENOLINK;
} }
ret = 0; ret = 0;
u = q->flags; u = q->flags;
if (copy_to_user(&ifr->ifr_name, vlan->dev->name, IFNAMSIZ) || if (copy_to_user(&ifr->ifr_name, tap->dev->name, IFNAMSIZ) ||
put_user(u, &ifr->ifr_flags)) put_user(u, &ifr->ifr_flags))
ret = -EFAULT; ret = -EFAULT;
tap_put_vlan(vlan); tap_put_tap_dev(tap);
rtnl_unlock(); rtnl_unlock();
return ret; return ret;
@ -1059,18 +1031,18 @@ static long tap_ioctl(struct file *file, unsigned int cmd,
case SIOCGIFHWADDR: case SIOCGIFHWADDR:
rtnl_lock(); rtnl_lock();
vlan = tap_get_vlan(q); tap = tap_get_tap_dev(q);
if (!vlan) { if (!tap) {
rtnl_unlock(); rtnl_unlock();
return -ENOLINK; return -ENOLINK;
} }
ret = 0; ret = 0;
u = vlan->dev->type; u = tap->dev->type;
if (copy_to_user(&ifr->ifr_name, vlan->dev->name, IFNAMSIZ) || if (copy_to_user(&ifr->ifr_name, tap->dev->name, IFNAMSIZ) ||
copy_to_user(&ifr->ifr_hwaddr.sa_data, vlan->dev->dev_addr, ETH_ALEN) || copy_to_user(&ifr->ifr_hwaddr.sa_data, tap->dev->dev_addr, ETH_ALEN) ||
put_user(u, &ifr->ifr_hwaddr.sa_family)) put_user(u, &ifr->ifr_hwaddr.sa_family))
ret = -EFAULT; ret = -EFAULT;
tap_put_vlan(vlan); tap_put_tap_dev(tap);
rtnl_unlock(); rtnl_unlock();
return ret; return ret;
@ -1078,13 +1050,13 @@ static long tap_ioctl(struct file *file, unsigned int cmd,
if (copy_from_user(&sa, &ifr->ifr_hwaddr, sizeof(sa))) if (copy_from_user(&sa, &ifr->ifr_hwaddr, sizeof(sa)))
return -EFAULT; return -EFAULT;
rtnl_lock(); rtnl_lock();
vlan = tap_get_vlan(q); tap = tap_get_tap_dev(q);
if (!vlan) { if (!tap) {
rtnl_unlock(); rtnl_unlock();
return -ENOLINK; return -ENOLINK;
} }
ret = dev_set_mac_address(vlan->dev, &sa); ret = dev_set_mac_address(tap->dev, &sa);
tap_put_vlan(vlan); tap_put_tap_dev(tap);
rtnl_unlock(); rtnl_unlock();
return ret; return ret;
@ -1167,19 +1139,19 @@ struct socket *tap_get_socket(struct file *file)
} }
EXPORT_SYMBOL_GPL(tap_get_socket); EXPORT_SYMBOL_GPL(tap_get_socket);
int tap_queue_resize(struct macvlan_dev *vlan) int tap_queue_resize(struct tap_dev *tap)
{ {
struct net_device *dev = vlan->dev; struct net_device *dev = tap->dev;
struct tap_queue *q; struct tap_queue *q;
struct skb_array **arrays; struct skb_array **arrays;
int n = vlan->numqueues; int n = tap->numqueues;
int ret, i = 0; int ret, i = 0;
arrays = kmalloc(sizeof *arrays * n, GFP_KERNEL); arrays = kmalloc(sizeof *arrays * n, GFP_KERNEL);
if (!arrays) if (!arrays)
return -ENOMEM; return -ENOMEM;
list_for_each_entry(q, &vlan->queue_list, next) list_for_each_entry(q, &tap->queue_list, next)
arrays[i++] = &q->skb_array; arrays[i++] = &q->skb_array;
ret = skb_array_resize_multiple(arrays, n, ret = skb_array_resize_multiple(arrays, n,

57
include/linux/if_tap.h
View File

@ -14,11 +14,60 @@ static inline struct socket *tap_get_socket(struct file *f)
} }
#endif /* CONFIG_MACVTAP */ #endif /* CONFIG_MACVTAP */
#include <net/sock.h>
#include <linux/skb_array.h>
#define MAX_TAP_QUEUES 256
struct tap_queue;
struct tap_dev {
struct net_device *dev;
u16 flags;
/* This array tracks active taps. */
struct tap_queue __rcu *taps[MAX_TAP_QUEUES];
/* This list tracks all taps (both enabled and disabled) */
struct list_head queue_list;
int numvtaps;
int numqueues;
netdev_features_t tap_features;
int minor;
void (*update_features)(struct tap_dev *tap, netdev_features_t features);
void (*count_tx_dropped)(struct tap_dev *tap);
void (*count_rx_dropped)(struct tap_dev *tap);
};
/*
* A tap queue is the central object of tap module, it connects
* an open character device to virtual interface. There can be
* multiple queues on one interface, which map back to queues
* implemented in hardware on the underlying device.
*
* tap_proto is used to allocate queues through the sock allocation
* mechanism.
*
*/
struct tap_queue {
struct sock sk;
struct socket sock;
struct socket_wq wq;
int vnet_hdr_sz;
struct tap_dev __rcu *tap;
struct file *file;
unsigned int flags;
u16 queue_index;
bool enabled;
struct list_head next;
struct skb_array skb_array;
};
rx_handler_result_t tap_handle_frame(struct sk_buff **pskb); rx_handler_result_t tap_handle_frame(struct sk_buff **pskb);
void tap_del_queues(struct net_device *dev); void tap_del_queues(struct tap_dev *tap);
int tap_get_minor(struct macvlan_dev *vlan); int tap_get_minor(struct tap_dev *tap);
void tap_free_minor(struct macvlan_dev *vlan); void tap_free_minor(struct tap_dev *tap);
int tap_queue_resize(struct macvlan_dev *vlan); int tap_queue_resize(struct tap_dev *tap);
int tap_create_cdev(struct cdev *tap_cdev, int tap_create_cdev(struct cdev *tap_cdev,
dev_t *tap_major, const char *device_name); dev_t *tap_major, const char *device_name);
void tap_destroy_cdev(dev_t major, struct cdev *tap_cdev); void tap_destroy_cdev(dev_t major, struct cdev *tap_cdev);