* [PATCH net-next v3 09/13] bonding: use vlan_uses_dev() in __bond_release_one()
From: Veaceslav Falico @ 2013-08-28 16:03 UTC (permalink / raw)
To: netdev; +Cc: Veaceslav Falico, Jay Vosburgh, Andy Gospodarek
In-Reply-To: <1377705842-8276-1-git-send-email-vfalico@redhat.com>
We always hold the rtnl_lock() in __bond_release_one(), so use
vlan_uses_dev() instead of bond_vlan_used().
v1: no change
v2: no change
v3: no change
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
---
drivers/net/bonding/bond_main.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 0fb3122..d61f80a 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -1954,7 +1954,7 @@ static int __bond_release_one(struct net_device *bond_dev,
bond_set_carrier(bond);
eth_hw_addr_random(bond_dev);
- if (bond_vlan_used(bond)) {
+ if (vlan_uses_dev(bond_dev)) {
pr_warning("%s: Warning: clearing HW address of %s while it still has VLANs.\n",
bond_dev->name, bond_dev->name);
pr_warning("%s: When re-adding slaves, make sure the bond's HW address matches its VLANs'.\n",
--
1.7.1
^ permalink raw reply related
* [PATCH net-next v3 08/13] bonding: convert bond_has_this_ip() to use upper devices
From: Veaceslav Falico @ 2013-08-28 16:03 UTC (permalink / raw)
To: netdev; +Cc: Veaceslav Falico, Jay Vosburgh, Andy Gospodarek
In-Reply-To: <1377705842-8276-1-git-send-email-vfalico@redhat.com>
Currently, bond_has_this_ip() is aware only of vlan upper devices, and thus
will return false if the address is associated with the upper bridge or any
other device, and thus will break the arp logic.
Fix this by using the upper device list. For every upper device we verify
if the address associated with it is our address, and if yes - return true.
v1: use netdev_for_each_upper_dev()
v2: use netdev_for_each_upper_dev_rcu()
v3: no change
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
---
drivers/net/bonding/bond_main.c | 25 +++++++++++++------------
1 files changed, 13 insertions(+), 12 deletions(-)
diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 5f38188..0fb3122 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -2392,24 +2392,25 @@ re_arm:
}
}
-static int bond_has_this_ip(struct bonding *bond, __be32 ip)
+static bool bond_has_this_ip(struct bonding *bond, __be32 ip)
{
- struct vlan_entry *vlan;
- struct net_device *vlan_dev;
+ struct net_device *upper;
+ struct list_head *iter;
+ bool ret = false;
if (ip == bond_confirm_addr(bond->dev, 0, ip))
- return 1;
+ return true;
- list_for_each_entry(vlan, &bond->vlan_list, vlan_list) {
- rcu_read_lock();
- vlan_dev = __vlan_find_dev_deep(bond->dev, htons(ETH_P_8021Q),
- vlan->vlan_id);
- rcu_read_unlock();
- if (vlan_dev && ip == bond_confirm_addr(vlan_dev, 0, ip))
- return 1;
+ rcu_read_lock();
+ netdev_for_each_upper_dev_rcu(bond->dev, upper, iter) {
+ if (ip == bond_confirm_addr(upper, 0, ip)) {
+ ret = true;
+ break;
+ }
}
+ rcu_read_unlock();
- return 0;
+ return ret;
}
/*
--
1.7.1
^ permalink raw reply related
* [PATCH net-next v3 06/13] bonding: use netdev_upper list in bond_vlan_used
From: Veaceslav Falico @ 2013-08-28 16:03 UTC (permalink / raw)
To: netdev; +Cc: Veaceslav Falico, Jay Vosburgh, Andy Gospodarek
In-Reply-To: <1377705842-8276-1-git-send-email-vfalico@redhat.com>
Convert bond_vlan_used() to traverse the upper device list to see if we
have any vlans above us. It's protected by rcu, and in case we are holding
rtnl_lock we should call vlan_uses_dev() instead - it's faster.
v1: use netdev_for_each_upper_dev()
v2: use netdev_for_each_upper_dev_rcu()
v3: no change
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
---
drivers/net/bonding/bonding.h | 15 ++++++++++++++-
1 files changed, 14 insertions(+), 1 deletions(-)
diff --git a/drivers/net/bonding/bonding.h b/drivers/net/bonding/bonding.h
index 4bf52d5..230197d 100644
--- a/drivers/net/bonding/bonding.h
+++ b/drivers/net/bonding/bonding.h
@@ -267,9 +267,22 @@ struct bonding {
#endif /* CONFIG_DEBUG_FS */
};
+/* if we hold rtnl_lock() - call vlan_uses_dev() */
static inline bool bond_vlan_used(struct bonding *bond)
{
- return !list_empty(&bond->vlan_list);
+ struct net_device *upper;
+ struct list_head *iter;
+
+ rcu_read_lock();
+ netdev_for_each_upper_dev_rcu(bond->dev, upper, iter) {
+ if (upper->priv_flags & IFF_802_1Q_VLAN) {
+ rcu_read_unlock();
+ return true;
+ }
+ }
+ rcu_read_unlock();
+
+ return false;
}
#define bond_slave_get_rcu(dev) \
--
1.7.1
^ permalink raw reply related
* [PATCH net-next v3 05/13] net: add netdev_for_each_upper_dev_rcu()
From: Veaceslav Falico @ 2013-08-28 16:03 UTC (permalink / raw)
To: netdev
Cc: Veaceslav Falico, David S. Miller, Eric Dumazet, Jiri Pirko,
Alexander Duyck, Cong Wang
In-Reply-To: <1377705842-8276-1-git-send-email-vfalico@redhat.com>
The new macro netdev_for_each_upper_dev_rcu(dev, upper, iter) iterates
through the dev->upper_dev_list starting from the first element, using
the netdev_upper_get_next_dev_rcu(dev, &iter).
Must be called under RCU read lock.
v1: new patch
v2: rename netdev_for_each_upper_dev{,_rcu} and use/export the renamed
netdev_upper_get_next_dev_rcu().
v3: no change
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
---
include/linux/netdevice.h | 10 ++++++++++
1 files changed, 10 insertions(+), 0 deletions(-)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 5ccf5b7..3ad49b8 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -2768,6 +2768,16 @@ extern int bpf_jit_enable;
extern bool netdev_has_upper_dev(struct net_device *dev,
struct net_device *upper_dev);
extern bool netdev_has_any_upper_dev(struct net_device *dev);
+extern struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
+ struct list_head **iter);
+
+/* iterate through upper list, must be called under RCU read lock */
+#define netdev_for_each_upper_dev_rcu(dev, upper, iter) \
+ for (iter = &(dev)->upper_dev_list, \
+ upper = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
+ upper; \
+ upper = netdev_upper_get_next_dev_rcu(dev, &(iter)))
+
extern struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
extern struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
extern int netdev_upper_dev_link(struct net_device *dev,
--
1.7.1
^ permalink raw reply related
* [PATCH net-next v3 04/13] net: add netdev_upper_get_next_dev_rcu(dev, iter)
From: Veaceslav Falico @ 2013-08-28 16:03 UTC (permalink / raw)
To: netdev
Cc: Veaceslav Falico, David S. Miller, Eric Dumazet, Jiri Pirko,
Alexander Duyck, Cong Wang
In-Reply-To: <1377705842-8276-1-git-send-email-vfalico@redhat.com>
This function returns the next dev in the dev->upper_dev_list after the
struct list_head **iter position, and updates *iter accordingly. Returns
NULL if there are no devices left.
Caller must hold RCU read lock.
v1: new patch
v2: rename netdev_upper_get_next_dev{,_rcu} to reflect the RCU, and rename
the netdev_upper to netdev_adjacent
v3: no change
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
---
net/core/dev.c | 25 +++++++++++++++++++++++++
1 files changed, 25 insertions(+), 0 deletions(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index 749925a..6fbb0c9 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4468,6 +4468,31 @@ struct net_device *netdev_master_upper_dev_get(struct net_device *dev)
}
EXPORT_SYMBOL(netdev_master_upper_dev_get);
+/* netdev_upper_get_next_dev_rcu - Get the next dev from upper list
+ * @dev: device
+ * @iter: list_head ** of the current position
+ *
+ * Gets the next device from the dev's upper list, starting from iter
+ * position. The caller must hold RCU read lock.
+ */
+struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
+ struct list_head **iter)
+{
+ struct netdev_adjacent *upper;
+
+ WARN_ON_ONCE(!rcu_read_lock_held());
+
+ upper = list_entry_rcu((*iter)->next, struct netdev_adjacent, list);
+
+ if (&upper->list == &dev->upper_dev_list)
+ return NULL;
+
+ *iter = &upper->list;
+
+ return upper->dev;
+}
+EXPORT_SYMBOL(netdev_upper_get_next_dev_rcu);
+
/**
* netdev_master_upper_dev_get_rcu - Get master upper device
* @dev: device
--
1.7.1
^ permalink raw reply related
* [PATCH net-next v3 03/13] net: remove search_list from netdev_adjacent
From: Veaceslav Falico @ 2013-08-28 16:03 UTC (permalink / raw)
To: netdev
Cc: Veaceslav Falico, David S. Miller, Eric Dumazet, Jiri Pirko,
Alexander Duyck, Cong Wang
In-Reply-To: <1377705842-8276-1-git-send-email-vfalico@redhat.com>
We already don't need it cause we see every upper/lower device in the list
already.
v2: new patch
v3: no change
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
---
net/core/dev.c | 37 +------------------------------------
1 files changed, 1 insertions(+), 36 deletions(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index 2aa914e..749925a 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4381,42 +4381,8 @@ struct netdev_adjacent {
struct list_head list;
struct rcu_head rcu;
- struct list_head search_list;
};
-static void __append_search_uppers(struct list_head *search_list,
- struct net_device *dev)
-{
- struct netdev_adjacent *upper;
-
- list_for_each_entry(upper, &dev->upper_dev_list, list) {
- /* check if this upper is not already in search list */
- if (list_empty(&upper->search_list))
- list_add_tail(&upper->search_list, search_list);
- }
-}
-
-static bool __netdev_search_upper_dev(struct net_device *dev,
- struct net_device *upper_dev)
-{
- LIST_HEAD(search_list);
- struct netdev_adjacent *upper;
- struct netdev_adjacent *tmp;
- bool ret = false;
-
- __append_search_uppers(&search_list, dev);
- list_for_each_entry(upper, &search_list, search_list) {
- if (upper->dev == upper_dev) {
- ret = true;
- break;
- }
- __append_search_uppers(&search_list, upper->dev);
- }
- list_for_each_entry_safe(upper, tmp, &search_list, search_list)
- INIT_LIST_HEAD(&upper->search_list);
- return ret;
-}
-
static struct netdev_adjacent *__netdev_find_adj(struct net_device *dev,
struct net_device *adj_dev,
bool upper)
@@ -4544,7 +4510,6 @@ static int __netdev_adjacent_dev_insert(struct net_device *dev,
adj->master = master;
adj->neighbour = neighbour;
adj->ref_nr = 1;
- INIT_LIST_HEAD(&adj->search_list);
dev_hold(adj_dev);
pr_debug("dev_hold for %s, because of %s link added from %s to %s\n",
@@ -4671,7 +4636,7 @@ static int __netdev_upper_dev_link(struct net_device *dev,
return -EBUSY;
/* To prevent loops, check if dev is not upper device to upper_dev. */
- if (__netdev_search_upper_dev(upper_dev, dev))
+ if (__netdev_find_upper(upper_dev, dev))
return -EBUSY;
if (__netdev_find_upper(dev, upper_dev))
--
1.7.1
^ permalink raw reply related
* [PATCH net-next v3 02/13] net: add lower_dev_list to net_device and make a full mesh
From: Veaceslav Falico @ 2013-08-28 16:03 UTC (permalink / raw)
To: netdev
Cc: Veaceslav Falico, David S. Miller, Eric Dumazet, Jiri Pirko,
Alexander Duyck, Cong Wang
In-Reply-To: <1377705842-8276-1-git-send-email-vfalico@redhat.com>
This patch adds lower_dev_list list_head to net_device, which is the same
as upper_dev_list, only for lower devices, and begins to use it in the same
way as the upper list.
It also changes the way the whole adjacent device lists work - now they
contain *all* of upper/lower devices, not only the first level. The first
level devices are distinguished by the bool neighbour field in
netdev_adjacent, also added by this patch.
There are cases when a device can be added several times to the adjacent
list, the simplest would be:
/---- eth0.10 ---\
eth0- --- bond0
\---- eth0.20 ---/
where both bond0 and eth0 'see' each other in the adjacent lists two times.
To avoid duplication of netdev_adjacent structures ref_nr is being kept as
the number of times the device was added to the list.
The 'full view' is achieved by adding, on link creation, all of the
upper_dev's upper_dev_list devices as upper devices to all of the
lower_dev's lower_dev_list devices (and to the lower_dev itself), and vice
versa. On unlink they are removed using the same logic.
I've tested it with thousands vlans/bonds/bridges, everything works ok and
no observable lags even on a huge number of interfaces.
Memory footprint for 128 devices interconnected with each other via both
upper and lower (which is impossible, but for the comparison) lists would be:
128*128*2*sizeof(netdev_adjacent) = 1.5MB
but in the real world we usualy have at most several devices with slaves
and a lot of vlans, so the footprint will be much lower.
v2: new patch
v3: move the renaming part back to the first patch, where it belongs, and
correct style/comment issues.
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
---
include/linux/netdevice.h | 1 +
net/core/dev.c | 285 ++++++++++++++++++++++++++++++++++++++++-----
2 files changed, 259 insertions(+), 27 deletions(-)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 077363d..5ccf5b7 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1125,6 +1125,7 @@ struct net_device {
struct list_head napi_list;
struct list_head unreg_list;
struct list_head upper_dev_list; /* List of upper devices */
+ struct list_head lower_dev_list;
/* currently active device features */
diff --git a/net/core/dev.c b/net/core/dev.c
index 5072e2c..2aa914e 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4369,7 +4369,16 @@ softnet_break:
struct netdev_adjacent {
struct net_device *dev;
+
+ /* upper master flag, there can only be one master device per list */
bool master;
+
+ /* indicates that this dev is our first-level lower/upper device */
+ bool neighbour;
+
+ /* counter for the number of times this device was added to us */
+ u16 ref_nr;
+
struct list_head list;
struct rcu_head rcu;
struct list_head search_list;
@@ -4408,18 +4417,34 @@ static bool __netdev_search_upper_dev(struct net_device *dev,
return ret;
}
-static struct netdev_adjacent *__netdev_find_upper(struct net_device *dev,
- struct net_device *upper_dev)
+static struct netdev_adjacent *__netdev_find_adj(struct net_device *dev,
+ struct net_device *adj_dev,
+ bool upper)
{
- struct netdev_adjacent *upper;
+ struct netdev_adjacent *adj;
+ struct list_head *dev_list;
- list_for_each_entry(upper, &dev->upper_dev_list, list) {
- if (upper->dev == upper_dev)
- return upper;
+ dev_list = upper ? &dev->upper_dev_list : &dev->lower_dev_list;
+
+ list_for_each_entry(adj, dev_list, list) {
+ if (adj->dev == adj_dev)
+ return adj;
}
return NULL;
}
+static inline struct netdev_adjacent *__netdev_find_upper(struct net_device *dev,
+ struct net_device *udev)
+{
+ return __netdev_find_adj(dev, udev, true);
+}
+
+static inline struct netdev_adjacent *__netdev_find_lower(struct net_device *dev,
+ struct net_device *ldev)
+{
+ return __netdev_find_adj(dev, ldev, false);
+}
+
/**
* netdev_has_upper_dev - Check if device is linked to an upper device
* @dev: device
@@ -4496,10 +4521,149 @@ struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev)
}
EXPORT_SYMBOL(netdev_master_upper_dev_get_rcu);
+static int __netdev_adjacent_dev_insert(struct net_device *dev,
+ struct net_device *adj_dev,
+ bool neighbour, bool master,
+ bool upper)
+{
+ struct netdev_adjacent *adj;
+
+ adj = __netdev_find_adj(dev, adj_dev, upper);
+
+ if (adj) {
+ BUG_ON(neighbour);
+ adj->ref_nr++;
+ return 0;
+ }
+
+ adj = kmalloc(sizeof(*adj), GFP_KERNEL);
+ if (!adj)
+ return -ENOMEM;
+
+ adj->dev = adj_dev;
+ adj->master = master;
+ adj->neighbour = neighbour;
+ adj->ref_nr = 1;
+ INIT_LIST_HEAD(&adj->search_list);
+
+ dev_hold(adj_dev);
+ pr_debug("dev_hold for %s, because of %s link added from %s to %s\n",
+ adj_dev->name, upper ? "upper" : "lower", dev->name,
+ adj_dev->name);
+
+ if (!upper) {
+ list_add_tail_rcu(&adj->list, &dev->lower_dev_list);
+ return 0;
+ }
+
+ /* Ensure that master upper link is always the first item in list. */
+ if (master)
+ list_add_rcu(&adj->list, &dev->upper_dev_list);
+ else
+ list_add_tail_rcu(&adj->list, &dev->upper_dev_list);
+
+ return 0;
+}
+
+static inline int __netdev_upper_dev_insert(struct net_device *dev,
+ struct net_device *udev,
+ bool master, bool neighbour)
+{
+ return __netdev_adjacent_dev_insert(dev, udev, neighbour, master,
+ true);
+}
+
+static inline int __netdev_lower_dev_insert(struct net_device *dev,
+ struct net_device *ldev,
+ bool neighbour)
+{
+ return __netdev_adjacent_dev_insert(dev, ldev, neighbour, false,
+ false);
+}
+
+void __netdev_adjacent_dev_remove(struct net_device *dev,
+ struct net_device *adj_dev, bool upper)
+{
+ struct netdev_adjacent *adj;
+
+ if (upper)
+ adj = __netdev_find_upper(dev, adj_dev);
+ else
+ adj = __netdev_find_lower(dev, adj_dev);
+
+ if (!adj)
+ BUG();
+
+ if (adj->ref_nr > 1) {
+ adj->ref_nr--;
+ return;
+ }
+
+ list_del_rcu(&adj->list);
+ pr_debug("dev_put for %s, because of %s link removed from %s to %s\n",
+ adj_dev->name, upper ? "upper" : "lower", dev->name,
+ adj_dev->name);
+ dev_put(adj_dev);
+ kfree_rcu(adj, rcu);
+}
+
+static inline void __netdev_upper_dev_remove(struct net_device *dev,
+ struct net_device *udev)
+{
+ return __netdev_adjacent_dev_remove(dev, udev, true);
+}
+
+static inline void __netdev_lower_dev_remove(struct net_device *dev,
+ struct net_device *ldev)
+{
+ return __netdev_adjacent_dev_remove(dev, ldev, false);
+}
+
+int __netdev_adjacent_dev_insert_link(struct net_device *dev,
+ struct net_device *upper_dev,
+ bool master, bool neighbour)
+{
+ int ret;
+
+ ret = __netdev_upper_dev_insert(dev, upper_dev, master, neighbour);
+ if (ret)
+ return ret;
+
+ ret = __netdev_lower_dev_insert(upper_dev, dev, neighbour);
+ if (ret) {
+ __netdev_upper_dev_remove(dev, upper_dev);
+ return ret;
+ }
+
+ return 0;
+}
+
+static inline int __netdev_adjacent_dev_link(struct net_device *dev,
+ struct net_device *udev)
+{
+ return __netdev_adjacent_dev_insert_link(dev, udev, false, false);
+}
+
+static inline int __netdev_adjacent_dev_link_neighbour(struct net_device *dev,
+ struct net_device *udev,
+ bool master)
+{
+ return __netdev_adjacent_dev_insert_link(dev, udev, master, true);
+}
+
+void __netdev_adjacent_dev_unlink(struct net_device *dev,
+ struct net_device *upper_dev)
+{
+ __netdev_upper_dev_remove(dev, upper_dev);
+ __netdev_lower_dev_remove(upper_dev, dev);
+}
+
+
static int __netdev_upper_dev_link(struct net_device *dev,
struct net_device *upper_dev, bool master)
{
- struct netdev_adjacent *upper;
+ struct netdev_adjacent *i, *j, *to_i, *to_j;
+ int ret = 0;
ASSERT_RTNL();
@@ -4516,22 +4680,76 @@ static int __netdev_upper_dev_link(struct net_device *dev,
if (master && netdev_master_upper_dev_get(dev))
return -EBUSY;
- upper = kmalloc(sizeof(*upper), GFP_KERNEL);
- if (!upper)
- return -ENOMEM;
+ ret = __netdev_adjacent_dev_link_neighbour(dev, upper_dev, master);
+ if (ret)
+ return ret;
- upper->dev = upper_dev;
- upper->master = master;
- INIT_LIST_HEAD(&upper->search_list);
+ /* Now that we linked these devs, make all the upper_dev's
+ * upper_dev_list visible to every dev's lower_dev_list and vice
+ * versa, and don't forget the devices itself. All of these
+ * links are non-neighbours.
+ */
+ list_for_each_entry(i, &upper_dev->upper_dev_list, list) {
+ list_for_each_entry(j, &dev->lower_dev_list, list) {
+ ret = __netdev_adjacent_dev_link(i->dev, j->dev);
+ if (ret)
+ goto rollback_mesh;
+ }
+ }
+
+ /* add dev to every upper_dev's upper device */
+ list_for_each_entry(i, &upper_dev->upper_dev_list, list) {
+ ret = __netdev_adjacent_dev_link(dev, i->dev);
+ if (ret)
+ goto rollback_upper_mesh;
+ }
+
+ /* add upper_dev to every dev's lower device */
+ list_for_each_entry(i, &dev->lower_dev_list, list) {
+ ret = __netdev_adjacent_dev_link(i->dev, upper_dev);
+ if (ret)
+ goto rollback_lower_mesh;
+ }
- /* Ensure that master upper link is always the first item in list. */
- if (master)
- list_add_rcu(&upper->list, &dev->upper_dev_list);
- else
- list_add_tail_rcu(&upper->list, &dev->upper_dev_list);
- dev_hold(upper_dev);
call_netdevice_notifiers(NETDEV_CHANGEUPPER, dev);
return 0;
+
+rollback_lower_mesh:
+ to_i = i;
+ list_for_each_entry(i, &dev->lower_dev_list, list) {
+ if (i == to_i)
+ break;
+ __netdev_adjacent_dev_unlink(i->dev, upper_dev);
+ }
+
+ i = NULL;
+
+rollback_upper_mesh:
+ to_i = i;
+ list_for_each_entry(i, &upper_dev->upper_dev_list, list) {
+ if (i == to_i)
+ break;
+ __netdev_adjacent_dev_unlink(dev, i->dev);
+ }
+
+ i = j = NULL;
+
+rollback_mesh:
+ to_i = i;
+ to_j = j;
+ list_for_each_entry(i, &dev->lower_dev_list, list) {
+ list_for_each_entry(j, &upper_dev->upper_dev_list, list) {
+ if (i == to_i && j == to_j)
+ break;
+ __netdev_adjacent_dev_unlink(i->dev, j->dev);
+ }
+ if (i == to_i)
+ break;
+ }
+
+ __netdev_adjacent_dev_unlink(dev, upper_dev);
+
+ return ret;
}
/**
@@ -4580,16 +4798,28 @@ EXPORT_SYMBOL(netdev_master_upper_dev_link);
void netdev_upper_dev_unlink(struct net_device *dev,
struct net_device *upper_dev)
{
- struct netdev_adjacent *upper;
-
+ struct netdev_adjacent *i, *j;
ASSERT_RTNL();
- upper = __netdev_find_upper(dev, upper_dev);
- if (!upper)
- return;
- list_del_rcu(&upper->list);
- dev_put(upper_dev);
- kfree_rcu(upper, rcu);
+ __netdev_adjacent_dev_unlink(dev, upper_dev);
+
+ /* Here is the tricky part. We must remove all dev's lower
+ * devices from all upper_dev's upper devices and vice
+ * versa, to maintain the graph relationship.
+ */
+ list_for_each_entry(i, &dev->lower_dev_list, list)
+ list_for_each_entry(j, &upper_dev->upper_dev_list, list)
+ __netdev_adjacent_dev_unlink(i->dev, j->dev);
+
+ /* remove also the devices itself from lower/upper device
+ * list
+ */
+ list_for_each_entry(i, &dev->lower_dev_list, list)
+ __netdev_adjacent_dev_unlink(i->dev, upper_dev);
+
+ list_for_each_entry(i, &upper_dev->upper_dev_list, list)
+ __netdev_adjacent_dev_unlink(dev, i->dev);
+
call_netdevice_notifiers(NETDEV_CHANGEUPPER, dev);
}
EXPORT_SYMBOL(netdev_upper_dev_unlink);
@@ -5850,6 +6080,7 @@ struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
INIT_LIST_HEAD(&dev->unreg_list);
INIT_LIST_HEAD(&dev->link_watch_list);
INIT_LIST_HEAD(&dev->upper_dev_list);
+ INIT_LIST_HEAD(&dev->lower_dev_list);
dev->priv_flags = IFF_XMIT_DST_RELEASE;
setup(dev);
--
1.7.1
^ permalink raw reply related
* [PATCH net-next v3 01/13] net: rename netdev_upper to netdev_adjacent
From: Veaceslav Falico @ 2013-08-28 16:03 UTC (permalink / raw)
To: netdev
Cc: Veaceslav Falico, David S. Miller, Eric Dumazet, Jiri Pirko,
Alexander Duyck, Cong Wang
In-Reply-To: <1377705842-8276-1-git-send-email-vfalico@redhat.com>
Rename the structure to reflect the upcoming addition of lower_dev_list.
v2: new patch
v3: actually rename everything
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
---
net/core/dev.c | 24 ++++++++++++------------
1 files changed, 12 insertions(+), 12 deletions(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index 1ed2b66..5072e2c 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4367,7 +4367,7 @@ softnet_break:
goto out;
}
-struct netdev_upper {
+struct netdev_adjacent {
struct net_device *dev;
bool master;
struct list_head list;
@@ -4378,7 +4378,7 @@ struct netdev_upper {
static void __append_search_uppers(struct list_head *search_list,
struct net_device *dev)
{
- struct netdev_upper *upper;
+ struct netdev_adjacent *upper;
list_for_each_entry(upper, &dev->upper_dev_list, list) {
/* check if this upper is not already in search list */
@@ -4391,8 +4391,8 @@ static bool __netdev_search_upper_dev(struct net_device *dev,
struct net_device *upper_dev)
{
LIST_HEAD(search_list);
- struct netdev_upper *upper;
- struct netdev_upper *tmp;
+ struct netdev_adjacent *upper;
+ struct netdev_adjacent *tmp;
bool ret = false;
__append_search_uppers(&search_list, dev);
@@ -4408,10 +4408,10 @@ static bool __netdev_search_upper_dev(struct net_device *dev,
return ret;
}
-static struct netdev_upper *__netdev_find_upper(struct net_device *dev,
+static struct netdev_adjacent *__netdev_find_upper(struct net_device *dev,
struct net_device *upper_dev)
{
- struct netdev_upper *upper;
+ struct netdev_adjacent *upper;
list_for_each_entry(upper, &dev->upper_dev_list, list) {
if (upper->dev == upper_dev)
@@ -4462,7 +4462,7 @@ EXPORT_SYMBOL(netdev_has_any_upper_dev);
*/
struct net_device *netdev_master_upper_dev_get(struct net_device *dev)
{
- struct netdev_upper *upper;
+ struct netdev_adjacent *upper;
ASSERT_RTNL();
@@ -4470,7 +4470,7 @@ struct net_device *netdev_master_upper_dev_get(struct net_device *dev)
return NULL;
upper = list_first_entry(&dev->upper_dev_list,
- struct netdev_upper, list);
+ struct netdev_adjacent, list);
if (likely(upper->master))
return upper->dev;
return NULL;
@@ -4486,10 +4486,10 @@ EXPORT_SYMBOL(netdev_master_upper_dev_get);
*/
struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev)
{
- struct netdev_upper *upper;
+ struct netdev_adjacent *upper;
upper = list_first_or_null_rcu(&dev->upper_dev_list,
- struct netdev_upper, list);
+ struct netdev_adjacent, list);
if (upper && likely(upper->master))
return upper->dev;
return NULL;
@@ -4499,7 +4499,7 @@ EXPORT_SYMBOL(netdev_master_upper_dev_get_rcu);
static int __netdev_upper_dev_link(struct net_device *dev,
struct net_device *upper_dev, bool master)
{
- struct netdev_upper *upper;
+ struct netdev_adjacent *upper;
ASSERT_RTNL();
@@ -4580,7 +4580,7 @@ EXPORT_SYMBOL(netdev_master_upper_dev_link);
void netdev_upper_dev_unlink(struct net_device *dev,
struct net_device *upper_dev)
{
- struct netdev_upper *upper;
+ struct netdev_adjacent *upper;
ASSERT_RTNL();
--
1.7.1
^ permalink raw reply related
* Re: [nf-next PATCH] netfilter: Extend SYNPROXY with a --continue option
From: Patrick McHardy @ 2013-08-28 16:00 UTC (permalink / raw)
To: Jesper Dangaard Brouer
Cc: Pablo Neira Ayuso, netfilter-devel, netdev, mph, as
In-Reply-To: <20130828152549.8250.97982.stgit@dragon>
On Wed, Aug 28, 2013 at 05:26:55PM +0200, Jesper Dangaard Brouer wrote:
> Packets reaching SYNPROXY are default dropped, as they are most likely
> invalid (given the state matching). In other configurations it might
> be beneficial to let packet not consumed by SYNPROXY (SYN and ACK) to
> continue being processed by the stack.
>
> Introducing a --continue option/flag for the SYNPROXY target to allow
> these unmatched packets to continue being processed. This would also
> help in debugging the module via LOG.
I don't think we should add an option for this. Just DROP packets consumed
by the target and have the other ones continue.
^ permalink raw reply
* Re: [PATCH net-next v10 05/11] ipv6: do not call ndisc_send_rs() with write lock
From: Hannes Frederic Sowa @ 2013-08-28 15:34 UTC (permalink / raw)
To: Cong Wang; +Cc: netdev, David S. Miller
In-Reply-To: <1377667379-2315-6-git-send-email-amwang@redhat.com>
On Wed, Aug 28, 2013 at 01:22:53PM +0800, Cong Wang wrote:
> From: Cong Wang <amwang@redhat.com>
>
> Because vxlan module will call ip6_dst_lookup() in TX path,
> which will hold write lock. So we have to release this write lock
> before calling ndisc_send_rs(), otherwise could deadlock.
>
> Signed-off-by: Cong Wang <amwang@redhat.com>
Change seems fine:
Reviewed-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
> ---
> net/ipv6/addrconf.c | 10 +++++++---
> 1 files changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
> index 8e68466..1bb0861 100644
> --- a/net/ipv6/addrconf.c
> +++ b/net/ipv6/addrconf.c
> @@ -3088,6 +3088,7 @@ static int addrconf_ifdown(struct net_device *dev, int how)
> static void addrconf_rs_timer(unsigned long data)
> {
> struct inet6_dev *idev = (struct inet6_dev *)data;
> + struct net_device *dev = idev->dev;
> struct in6_addr lladdr;
>
> write_lock(&idev->lock);
> @@ -3102,12 +3103,14 @@ static void addrconf_rs_timer(unsigned long data)
> goto out;
>
> if (idev->rs_probes++ < idev->cnf.rtr_solicits) {
> - if (!__ipv6_get_lladdr(idev, &lladdr, IFA_F_TENTATIVE))
> - ndisc_send_rs(idev->dev, &lladdr,
> + write_unlock(&idev->lock);
> + if (!ipv6_get_lladdr(dev, &lladdr, IFA_F_TENTATIVE))
> + ndisc_send_rs(dev, &lladdr,
> &in6addr_linklocal_allrouters);
> else
> - goto out;
> + goto put;
>
> + write_lock(&idev->lock);
> /* The wait after the last probe can be shorter */
> addrconf_mod_rs_timer(idev, (idev->rs_probes ==
> idev->cnf.rtr_solicits) ?
> @@ -3123,6 +3126,7 @@ static void addrconf_rs_timer(unsigned long data)
>
> out:
> write_unlock(&idev->lock);
> +put:
> in6_dev_put(idev);
> }
Maybe it would be cleaner to just update a boolean in the inner block
and call ndisc_send_rs after the write_unlock.
Greetings,
Hannes
^ permalink raw reply
* [nf-next PATCH] netfilter: Extend SYNPROXY with a --continue option
From: Jesper Dangaard Brouer @ 2013-08-28 15:26 UTC (permalink / raw)
To: Pablo Neira Ayuso, Patrick McHardy
Cc: netfilter-devel, netdev, mph, as, Jesper Dangaard Brouer
Packets reaching SYNPROXY are default dropped, as they are most likely
invalid (given the state matching). In other configurations it might
be beneficial to let packet not consumed by SYNPROXY (SYN and ACK) to
continue being processed by the stack.
Introducing a --continue option/flag for the SYNPROXY target to allow
these unmatched packets to continue being processed. This would also
help in debugging the module via LOG.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
---
Comment:
This patch depend on the one send earlier, and I also have the
userspace iptables part ready, if this patch is received positively
include/uapi/linux/netfilter/xt_SYNPROXY.h | 1 +
net/ipv4/netfilter/ipt_SYNPROXY.c | 9 ++++++++-
net/ipv6/netfilter/ip6t_SYNPROXY.c | 9 ++++++++-
3 files changed, 17 insertions(+), 2 deletions(-)
diff --git a/include/uapi/linux/netfilter/xt_SYNPROXY.h b/include/uapi/linux/netfilter/xt_SYNPROXY.h
index 2d59fba..6b788bc 100644
--- a/include/uapi/linux/netfilter/xt_SYNPROXY.h
+++ b/include/uapi/linux/netfilter/xt_SYNPROXY.h
@@ -6,6 +6,7 @@
#define XT_SYNPROXY_OPT_SACK_PERM 0x04
#define XT_SYNPROXY_OPT_TIMESTAMP 0x08
#define XT_SYNPROXY_OPT_ECN 0x10
+#define XT_SYNPROXY_OPT_CONTINUE 0x20
struct xt_synproxy_info {
__u8 options;
diff --git a/net/ipv4/netfilter/ipt_SYNPROXY.c b/net/ipv4/netfilter/ipt_SYNPROXY.c
index 90e489e..3a09759 100644
--- a/net/ipv4/netfilter/ipt_SYNPROXY.c
+++ b/net/ipv4/netfilter/ipt_SYNPROXY.c
@@ -285,9 +285,16 @@ synproxy_tg4(struct sk_buff *skb, const struct xt_action_param *par)
XT_SYNPROXY_OPT_ECN);
synproxy_send_client_synack(skb, th, &opts);
- } else if (th->ack && !(th->fin || th->rst || th->syn))
+ return NF_DROP;
+
+ } else if (th->ack && !(th->fin || th->rst || th->syn)) {
/* ACK from client */
synproxy_recv_client_ack(snet, skb, th, &opts, ntohl(th->seq));
+ return NF_DROP;
+ }
+
+ if (info->options & XT_SYNPROXY_OPT_CONTINUE)
+ return XT_CONTINUE;
return NF_DROP;
}
diff --git a/net/ipv6/netfilter/ip6t_SYNPROXY.c b/net/ipv6/netfilter/ip6t_SYNPROXY.c
index a5af0bf..02f41fa 100644
--- a/net/ipv6/netfilter/ip6t_SYNPROXY.c
+++ b/net/ipv6/netfilter/ip6t_SYNPROXY.c
@@ -300,9 +300,16 @@ synproxy_tg6(struct sk_buff *skb, const struct xt_action_param *par)
XT_SYNPROXY_OPT_ECN);
synproxy_send_client_synack(skb, th, &opts);
- } else if (th->ack && !(th->fin || th->rst || th->syn))
+ return NF_DROP;
+
+ } else if (th->ack && !(th->fin || th->rst || th->syn)) {
/* ACK from client */
synproxy_recv_client_ack(snet, skb, th, &opts, ntohl(th->seq));
+ return NF_DROP;
+ }
+
+ if (info->options & XT_SYNPROXY_OPT_CONTINUE)
+ return XT_CONTINUE;
return NF_DROP;
}
^ permalink raw reply related
* [PATCH net-next 16/16] sfc: Use extended MC_CMD_SENSOR_INFO and MC_CMD_READ_SENSORS
From: Ben Hutchings @ 2013-08-28 15:23 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377702837.2264.21.camel@bwh-desktop.uk.level5networks.com>
We need to use extended requests to read and get metadata for sensors
numbered > 31.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/mcdi_mon.c | 139 ++++++++++++++++++++++++------------
1 file changed, 95 insertions(+), 44 deletions(-)
diff --git a/drivers/net/ethernet/sfc/mcdi_mon.c b/drivers/net/ethernet/sfc/mcdi_mon.c
index 544889b..d7d4566 100644
--- a/drivers/net/ethernet/sfc/mcdi_mon.c
+++ b/drivers/net/ethernet/sfc/mcdi_mon.c
@@ -28,7 +28,7 @@ static const struct {
const char *label;
enum efx_hwmon_type hwmon_type;
int port;
-} efx_mcdi_sensor_type[MC_CMD_SENSOR_ENTRY_MAXNUM] = {
+} efx_mcdi_sensor_type[] = {
#define SENSOR(name, label, hwmon_type, port) \
[MC_CMD_SENSOR_##name] = { label, hwmon_type, port }
SENSOR(CONTROLLER_TEMP, "Controller temp.", EFX_HWMON_TEMP, -1),
@@ -86,6 +86,7 @@ struct efx_mcdi_mon_attribute {
struct device_attribute dev_attr;
unsigned int index;
unsigned int type;
+ enum efx_hwmon_type hwmon_type;
unsigned int limit_value;
char name[12];
};
@@ -93,11 +94,12 @@ struct efx_mcdi_mon_attribute {
static int efx_mcdi_mon_update(struct efx_nic *efx)
{
struct efx_mcdi_mon *hwmon = efx_mcdi_mon(efx);
- MCDI_DECLARE_BUF(inbuf, MC_CMD_READ_SENSORS_IN_LEN);
+ MCDI_DECLARE_BUF(inbuf, MC_CMD_READ_SENSORS_EXT_IN_LEN);
int rc;
- MCDI_SET_QWORD(inbuf, READ_SENSORS_IN_DMA_ADDR,
+ MCDI_SET_QWORD(inbuf, READ_SENSORS_EXT_IN_DMA_ADDR,
hwmon->dma_buf.dma_addr);
+ MCDI_SET_DWORD(inbuf, READ_SENSORS_EXT_IN_LENGTH, hwmon->dma_buf.len);
rc = efx_mcdi_rpc(efx, MC_CMD_READ_SENSORS,
inbuf, sizeof(inbuf), NULL, 0, NULL);
@@ -159,7 +161,7 @@ static ssize_t efx_mcdi_mon_show_value(struct device *dev,
value = EFX_DWORD_FIELD(entry, MC_CMD_SENSOR_VALUE_ENTRY_TYPEDEF_VALUE);
/* Convert temperature from degrees to milli-degrees Celsius */
- if (efx_mcdi_sensor_type[mon_attr->type].hwmon_type == EFX_HWMON_TEMP)
+ if (mon_attr->hwmon_type == EFX_HWMON_TEMP)
value *= 1000;
return sprintf(buf, "%u\n", value);
@@ -176,7 +178,7 @@ static ssize_t efx_mcdi_mon_show_limit(struct device *dev,
value = mon_attr->limit_value;
/* Convert temperature from degrees to milli-degrees Celsius */
- if (efx_mcdi_sensor_type[mon_attr->type].hwmon_type == EFX_HWMON_TEMP)
+ if (mon_attr->hwmon_type == EFX_HWMON_TEMP)
value *= 1000;
return sprintf(buf, "%u\n", value);
@@ -224,6 +226,10 @@ efx_mcdi_mon_add_attr(struct efx_nic *efx, const char *name,
strlcpy(attr->name, name, sizeof(attr->name));
attr->index = index;
attr->type = type;
+ if (type < ARRAY_SIZE(efx_mcdi_sensor_type))
+ attr->hwmon_type = efx_mcdi_sensor_type[type].hwmon_type;
+ else
+ attr->hwmon_type = EFX_HWMON_UNKNOWN;
attr->limit_value = limit_value;
sysfs_attr_init(&attr->dev_attr.attr);
attr->dev_attr.attr.name = attr->name;
@@ -238,35 +244,42 @@ efx_mcdi_mon_add_attr(struct efx_nic *efx, const char *name,
int efx_mcdi_mon_probe(struct efx_nic *efx)
{
struct efx_mcdi_mon *hwmon = efx_mcdi_mon(efx);
- unsigned int n_attrs, n_temp = 0, n_cool = 0, n_in = 0;
+ unsigned int n_temp = 0, n_cool = 0, n_in = 0;
+ MCDI_DECLARE_BUF(inbuf, MC_CMD_SENSOR_INFO_EXT_IN_LEN);
MCDI_DECLARE_BUF(outbuf, MC_CMD_SENSOR_INFO_OUT_LENMAX);
+ unsigned int n_pages, n_sensors, n_attrs, page;
size_t outlen;
char name[12];
u32 mask;
- int rc, i, type;
+ int rc, i, j, type;
- BUILD_BUG_ON(MC_CMD_SENSOR_INFO_IN_LEN != 0);
+ /* Find out how many sensors are present */
+ n_sensors = 0;
+ page = 0;
+ do {
+ MCDI_SET_DWORD(inbuf, SENSOR_INFO_EXT_IN_PAGE, page);
- rc = efx_mcdi_rpc(efx, MC_CMD_SENSOR_INFO, NULL, 0,
- outbuf, sizeof(outbuf), &outlen);
- if (rc)
- return rc;
- if (outlen < MC_CMD_SENSOR_INFO_OUT_LENMIN)
- return -EIO;
-
- /* Find out which sensors are present. Don't create a device
- * if there are none.
- */
- mask = MCDI_DWORD(outbuf, SENSOR_INFO_OUT_MASK);
- if (mask == 0)
+ rc = efx_mcdi_rpc(efx, MC_CMD_SENSOR_INFO, inbuf, sizeof(inbuf),
+ outbuf, sizeof(outbuf), &outlen);
+ if (rc)
+ return rc;
+ if (outlen < MC_CMD_SENSOR_INFO_OUT_LENMIN)
+ return -EIO;
+
+ mask = MCDI_DWORD(outbuf, SENSOR_INFO_OUT_MASK);
+ n_sensors += hweight32(mask & ~(1 << MC_CMD_SENSOR_PAGE0_NEXT));
+ ++page;
+ } while (mask & (1 << MC_CMD_SENSOR_PAGE0_NEXT));
+ n_pages = page;
+
+ /* Don't create a device if there are none */
+ if (n_sensors == 0)
return 0;
- /* Check again for short response */
- if (outlen < MC_CMD_SENSOR_INFO_OUT_LEN(hweight32(mask)))
- return -EIO;
-
- rc = efx_nic_alloc_buffer(efx, &hwmon->dma_buf,
- 4 * MC_CMD_SENSOR_ENTRY_MAXNUM, GFP_KERNEL);
+ rc = efx_nic_alloc_buffer(
+ efx, &hwmon->dma_buf,
+ n_sensors * MC_CMD_SENSOR_VALUE_ENTRY_TYPEDEF_LEN,
+ GFP_KERNEL);
if (rc)
return rc;
@@ -277,7 +290,7 @@ int efx_mcdi_mon_probe(struct efx_nic *efx)
* attributes for this set of sensors: name of the driver plus
* value, min, max, crit, alarm and label for each sensor.
*/
- n_attrs = 1 + 6 * hweight32(mask);
+ n_attrs = 1 + 6 * n_sensors;
hwmon->attrs = kcalloc(n_attrs, sizeof(*hwmon->attrs), GFP_KERNEL);
if (!hwmon->attrs) {
rc = -ENOMEM;
@@ -294,26 +307,63 @@ int efx_mcdi_mon_probe(struct efx_nic *efx)
if (rc)
goto fail;
- for (i = 0, type = -1; ; i++) {
+ for (i = 0, j = -1, type = -1; ; i++) {
+ enum efx_hwmon_type hwmon_type;
const char *hwmon_prefix;
unsigned hwmon_index;
u16 min1, max1, min2, max2;
/* Find next sensor type or exit if there is none */
- type++;
- while (!(mask & (1 << type))) {
+ do {
type++;
- if (type == 32)
- return 0;
- }
- /* Skip sensors specific to a different port */
- if (efx_mcdi_sensor_type[type].hwmon_type != EFX_HWMON_UNKNOWN &&
- efx_mcdi_sensor_type[type].port >= 0 &&
- efx_mcdi_sensor_type[type].port != efx_port_num(efx))
- continue;
+ if ((type % 32) == 0) {
+ page = type / 32;
+ j = -1;
+ if (page == n_pages)
+ return 0;
+
+ MCDI_SET_DWORD(inbuf, SENSOR_INFO_EXT_IN_PAGE,
+ page);
+ rc = efx_mcdi_rpc(efx, MC_CMD_SENSOR_INFO,
+ inbuf, sizeof(inbuf),
+ outbuf, sizeof(outbuf),
+ &outlen);
+ if (rc)
+ goto fail;
+ if (outlen < MC_CMD_SENSOR_INFO_OUT_LENMIN) {
+ rc = -EIO;
+ goto fail;
+ }
+
+ mask = (MCDI_DWORD(outbuf,
+ SENSOR_INFO_OUT_MASK) &
+ ~(1 << MC_CMD_SENSOR_PAGE0_NEXT));
+
+ /* Check again for short response */
+ if (outlen <
+ MC_CMD_SENSOR_INFO_OUT_LEN(hweight32(mask))) {
+ rc = -EIO;
+ goto fail;
+ }
+ }
+ } while (!(mask & (1 << type % 32)));
+ j++;
+
+ if (type < ARRAY_SIZE(efx_mcdi_sensor_type)) {
+ hwmon_type = efx_mcdi_sensor_type[type].hwmon_type;
+
+ /* Skip sensors specific to a different port */
+ if (hwmon_type != EFX_HWMON_UNKNOWN &&
+ efx_mcdi_sensor_type[type].port >= 0 &&
+ efx_mcdi_sensor_type[type].port !=
+ efx_port_num(efx))
+ continue;
+ } else {
+ hwmon_type = EFX_HWMON_UNKNOWN;
+ }
- switch (efx_mcdi_sensor_type[type].hwmon_type) {
+ switch (hwmon_type) {
case EFX_HWMON_TEMP:
hwmon_prefix = "temp";
hwmon_index = ++n_temp; /* 1-based */
@@ -333,13 +383,13 @@ int efx_mcdi_mon_probe(struct efx_nic *efx)
}
min1 = MCDI_ARRAY_FIELD(outbuf, SENSOR_ENTRY,
- SENSOR_INFO_ENTRY, i, MIN1);
+ SENSOR_INFO_ENTRY, j, MIN1);
max1 = MCDI_ARRAY_FIELD(outbuf, SENSOR_ENTRY,
- SENSOR_INFO_ENTRY, i, MAX1);
+ SENSOR_INFO_ENTRY, j, MAX1);
min2 = MCDI_ARRAY_FIELD(outbuf, SENSOR_ENTRY,
- SENSOR_INFO_ENTRY, i, MIN2);
+ SENSOR_INFO_ENTRY, j, MIN2);
max2 = MCDI_ARRAY_FIELD(outbuf, SENSOR_ENTRY,
- SENSOR_INFO_ENTRY, i, MAX2);
+ SENSOR_INFO_ENTRY, j, MAX2);
if (min1 != max1) {
snprintf(name, sizeof(name), "%s%u_input",
@@ -386,7 +436,8 @@ int efx_mcdi_mon_probe(struct efx_nic *efx)
if (rc)
goto fail;
- if (efx_mcdi_sensor_type[type].label) {
+ if (type < ARRAY_SIZE(efx_mcdi_sensor_type) &&
+ efx_mcdi_sensor_type[type].label) {
snprintf(name, sizeof(name), "%s%u_label",
hwmon_prefix, hwmon_index);
rc = efx_mcdi_mon_add_attr(
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 15/16] sfc: Return an error code when a sensor is busy.
From: Ben Hutchings @ 2013-08-28 15:22 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377702837.2264.21.camel@bwh-desktop.uk.level5networks.com>
From: Alexandre Rames <arames@solarflare.com>
[bwh: Also name this new state, though we don't expect to see it in an event]
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/mcdi_mon.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/sfc/mcdi_mon.c b/drivers/net/ethernet/sfc/mcdi_mon.c
index 958c73f..544889b 100644
--- a/drivers/net/ethernet/sfc/mcdi_mon.c
+++ b/drivers/net/ethernet/sfc/mcdi_mon.c
@@ -54,6 +54,7 @@ static const char *const sensor_status_names[] = {
[MC_CMD_SENSOR_STATE_WARNING] = "Warning",
[MC_CMD_SENSOR_STATE_FATAL] = "Fatal",
[MC_CMD_SENSOR_STATE_BROKEN] = "Device failure",
+ [MC_CMD_SENSOR_STATE_NO_READING] = "No reading",
};
void efx_mcdi_sensor_event(struct efx_nic *efx, efx_qword_t *ev)
@@ -144,13 +145,17 @@ static ssize_t efx_mcdi_mon_show_value(struct device *dev,
struct efx_mcdi_mon_attribute *mon_attr =
container_of(attr, struct efx_mcdi_mon_attribute, dev_attr);
efx_dword_t entry;
- unsigned int value;
+ unsigned int value, state;
int rc;
rc = efx_mcdi_mon_get_entry(dev, mon_attr->index, &entry);
if (rc)
return rc;
+ state = EFX_DWORD_FIELD(entry, MC_CMD_SENSOR_VALUE_ENTRY_TYPEDEF_STATE);
+ if (state == MC_CMD_SENSOR_STATE_NO_READING)
+ return -EBUSY;
+
value = EFX_DWORD_FIELD(entry, MC_CMD_SENSOR_VALUE_ENTRY_TYPEDEF_VALUE);
/* Convert temperature from degrees to milli-degrees Celsius */
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 14/16] sfc: Add support for reading packet length from prefix
From: Ben Hutchings @ 2013-08-28 15:22 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377702837.2264.21.camel@bwh-desktop.uk.level5networks.com>
Define a flag for struct efx_rx_buffer and efx_rx_packet() that
indicates packet length must be read from the prefix. If this
is set, read the length in __efx_rx_packet() (when the prefix
should have arrived in cache).
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/net_driver.h | 4 ++++
drivers/net/ethernet/sfc/rx.c | 12 ++++++++++--
2 files changed, 14 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index f52b38c..e85bed0 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -263,6 +263,7 @@ struct efx_rx_buffer {
#define EFX_RX_PKT_CSUMMED 0x0002
#define EFX_RX_PKT_DISCARD 0x0004
#define EFX_RX_PKT_TCP 0x0040
+#define EFX_RX_PKT_PREFIX_LEN 0x0080 /* length is in prefix only */
/**
* struct efx_rx_page_state - Page-based rx buffer state
@@ -668,6 +669,8 @@ struct vfdi_status;
* @rx_prefix_size: Size of RX prefix before packet data
* @rx_packet_hash_offset: Offset of RX flow hash from start of packet data
* (valid only if @rx_prefix_size != 0; always negative)
+ * @rx_packet_len_offset: Offset of RX packet length from start of packet data
+ * (valid only for NICs that set %EFX_RX_PKT_PREFIX_LEN; always negative)
* @rx_hash_key: Toeplitz hash key for RSS
* @rx_indir_table: Indirection table for RSS
* @rx_scatter: Scatter mode enabled for receives
@@ -800,6 +803,7 @@ struct efx_nic {
unsigned int rx_pages_per_batch;
unsigned int rx_prefix_size;
int rx_packet_hash_offset;
+ int rx_packet_len_offset;
u8 rx_hash_key[40];
u32 rx_indir_table[128];
bool rx_scatter;
diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c
index 7646bb3..864b6ff 100644
--- a/drivers/net/ethernet/sfc/rx.c
+++ b/drivers/net/ethernet/sfc/rx.c
@@ -526,7 +526,8 @@ void efx_rx_packet(struct efx_rx_queue *rx_queue, unsigned int index,
/* Validate the number of fragments and completed length */
if (n_frags == 1) {
- efx_rx_packet__check_len(rx_queue, rx_buf, len);
+ if (!(flags & EFX_RX_PKT_PREFIX_LEN))
+ efx_rx_packet__check_len(rx_queue, rx_buf, len);
} else if (unlikely(n_frags > EFX_RX_MAX_FRAGS) ||
unlikely(len <= (n_frags - 1) * EFX_RX_USR_BUF_SIZE) ||
unlikely(len > n_frags * EFX_RX_USR_BUF_SIZE) ||
@@ -554,7 +555,7 @@ void efx_rx_packet(struct efx_rx_queue *rx_queue, unsigned int index,
return;
}
- if (n_frags == 1)
+ if (n_frags == 1 && !(flags & EFX_RX_PKT_PREFIX_LEN))
rx_buf->len = len;
/* Release and/or sync the DMA mapping - assumes all RX buffers
@@ -633,6 +634,13 @@ void __efx_rx_packet(struct efx_channel *channel)
efx_rx_buffer(&channel->rx_queue, channel->rx_pkt_index);
u8 *eh = efx_rx_buf_va(rx_buf);
+ /* Read length from the prefix if necessary. This already
+ * excludes the length of the prefix itself.
+ */
+ if (rx_buf->flags & EFX_RX_PKT_PREFIX_LEN)
+ rx_buf->len = le16_to_cpup((__le16 *)
+ (eh + efx->rx_packet_len_offset));
+
/* If we're in loopback test, then pass the packet directly to the
* loopback layer, and free the rx_buf here
*/
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 13/16] sfc: Add TX merged completion counter
From: Ben Hutchings @ 2013-08-28 15:21 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377702837.2264.21.camel@bwh-desktop.uk.level5networks.com>
Add a counter for TX merged completion events.
This is implemented in the common TX path, because the NIC event
handlers only know how many descriptors were completed, not how many
packets.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/ethtool.c | 1 +
drivers/net/ethernet/sfc/net_driver.h | 2 ++
drivers/net/ethernet/sfc/tx.c | 3 +++
3 files changed, 6 insertions(+)
diff --git a/drivers/net/ethernet/sfc/ethtool.c b/drivers/net/ethernet/sfc/ethtool.c
index 0b8ffdf..6354693 100644
--- a/drivers/net/ethernet/sfc/ethtool.c
+++ b/drivers/net/ethernet/sfc/ethtool.c
@@ -65,6 +65,7 @@ static u64 efx_get_atomic_stat(void *field)
unsigned int, efx_get_uint_stat)
static const struct efx_sw_stat_desc efx_sw_stat_desc[] = {
+ EFX_ETHTOOL_UINT_TXQ_STAT(merge_events),
EFX_ETHTOOL_UINT_TXQ_STAT(tso_bursts),
EFX_ETHTOOL_UINT_TXQ_STAT(tso_long_headers),
EFX_ETHTOOL_UINT_TXQ_STAT(tso_packets),
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index 481e412..f52b38c 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -186,6 +186,7 @@ struct efx_tx_buffer {
* variable indicates that the queue is empty. This is to
* avoid cache-line ping-pong between the xmit path and the
* completion path.
+ * @merge_events: Number of TX merged completion events
* @insert_count: Current insert pointer
* This is the number of buffers that have been added to the
* software ring.
@@ -222,6 +223,7 @@ struct efx_tx_queue {
/* Members used mainly on the completion path */
unsigned int read_count ____cacheline_aligned_in_smp;
unsigned int old_write_count;
+ unsigned int merge_events;
/* Members used only on the xmit path */
unsigned int insert_count ____cacheline_aligned_in_smp;
diff --git a/drivers/net/ethernet/sfc/tx.c b/drivers/net/ethernet/sfc/tx.c
index 4903c4f..85ee647 100644
--- a/drivers/net/ethernet/sfc/tx.c
+++ b/drivers/net/ethernet/sfc/tx.c
@@ -437,6 +437,9 @@ void efx_xmit_done(struct efx_tx_queue *tx_queue, unsigned int index)
efx_dequeue_buffers(tx_queue, index, &pkts_compl, &bytes_compl);
netdev_tx_completed_queue(tx_queue->core_txq, pkts_compl, bytes_compl);
+ if (pkts_compl > 1)
+ ++tx_queue->merge_events;
+
/* See if we need to restart the netif queue. This memory
* barrier ensures that we write read_count (inside
* efx_dequeue_buffers()) before reading the queue status.
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 12/16] sfc: Generalise packet hash lookup to support EF10 RX prefix
From: Ben Hutchings @ 2013-08-28 15:21 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377702837.2264.21.camel@bwh-desktop.uk.level5networks.com>
From: Jon Cooper <jcooper@solarflare.com>
EF10 uses an entirely different RX prefix format from Falcon-arch.
Extend struct efx_nic_type to describe this.
[bwh: Also replace the magic numbers used for the Falcon-arch RX prefix]
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/efx.c | 5 ++++-
drivers/net/ethernet/sfc/falcon.c | 3 ++-
drivers/net/ethernet/sfc/farch_regs.h | 4 ++++
drivers/net/ethernet/sfc/net_driver.h | 11 +++++++++--
drivers/net/ethernet/sfc/rx.c | 15 +++++++--------
drivers/net/ethernet/sfc/siena.c | 3 ++-
6 files changed, 28 insertions(+), 13 deletions(-)
diff --git a/drivers/net/ethernet/sfc/efx.c b/drivers/net/ethernet/sfc/efx.c
index 3af28a8..db2f119 100644
--- a/drivers/net/ethernet/sfc/efx.c
+++ b/drivers/net/ethernet/sfc/efx.c
@@ -573,7 +573,7 @@ static void efx_start_datapath(struct efx_nic *efx)
* support the current MTU, including padding for header
* alignment and overruns.
*/
- efx->rx_dma_len = (efx->type->rx_buffer_hash_size +
+ efx->rx_dma_len = (efx->rx_prefix_size +
EFX_MAX_FRAME_LEN(efx->net_dev->mtu) +
efx->type->rx_buffer_padding);
rx_buf_len = (sizeof(struct efx_rx_page_state) +
@@ -2456,6 +2456,9 @@ static int efx_init_struct(struct efx_nic *efx,
strlcpy(efx->name, pci_name(pci_dev), sizeof(efx->name));
efx->net_dev = net_dev;
+ efx->rx_prefix_size = efx->type->rx_prefix_size;
+ efx->rx_packet_hash_offset =
+ efx->type->rx_hash_offset - efx->type->rx_prefix_size;
spin_lock_init(&efx->stats_lock);
mutex_init(&efx->mac_lock);
efx->phy_op = &efx_dummy_phy_operations;
diff --git a/drivers/net/ethernet/sfc/falcon.c b/drivers/net/ethernet/sfc/falcon.c
index 1140b83..ec77611 100644
--- a/drivers/net/ethernet/sfc/falcon.c
+++ b/drivers/net/ethernet/sfc/falcon.c
@@ -2827,7 +2827,8 @@ const struct efx_nic_type falcon_b0_nic_type = {
.evq_ptr_tbl_base = FR_BZ_EVQ_PTR_TBL,
.evq_rptr_tbl_base = FR_BZ_EVQ_RPTR,
.max_dma_mask = DMA_BIT_MASK(FSF_AZ_TX_KER_BUF_ADDR_WIDTH),
- .rx_buffer_hash_size = 0x10,
+ .rx_prefix_size = FS_BZ_RX_PREFIX_SIZE,
+ .rx_hash_offset = FS_BZ_RX_PREFIX_HASH_OFST,
.rx_buffer_padding = 0,
.can_rx_scatter = true,
.max_interrupt_mode = EFX_INT_MODE_MSIX,
diff --git a/drivers/net/ethernet/sfc/farch_regs.h b/drivers/net/ethernet/sfc/farch_regs.h
index 00ef17a..491b103 100644
--- a/drivers/net/ethernet/sfc/farch_regs.h
+++ b/drivers/net/ethernet/sfc/farch_regs.h
@@ -2925,4 +2925,8 @@
#define FSF_AZ_DRV_GEN_EV_MAGIC_LBN 0
#define FSF_AZ_DRV_GEN_EV_MAGIC_WIDTH 32
+/* RX packet prefix */
+#define FS_BZ_RX_PREFIX_HASH_OFST 12
+#define FS_BZ_RX_PREFIX_SIZE 16
+
#endif /* EFX_FARCH_REGS_H */
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index 21b1c2a..481e412 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -663,6 +663,9 @@ struct vfdi_status;
* @rx_buffer_order: Order (log2) of number of pages for each RX buffer
* @rx_buffer_truesize: Amortised allocation size of an RX buffer,
* for use in sk_buff::truesize
+ * @rx_prefix_size: Size of RX prefix before packet data
+ * @rx_packet_hash_offset: Offset of RX flow hash from start of packet data
+ * (valid only if @rx_prefix_size != 0; always negative)
* @rx_hash_key: Toeplitz hash key for RSS
* @rx_indir_table: Indirection table for RSS
* @rx_scatter: Scatter mode enabled for receives
@@ -793,6 +796,8 @@ struct efx_nic {
unsigned int rx_page_buf_step;
unsigned int rx_bufs_per_page;
unsigned int rx_pages_per_batch;
+ unsigned int rx_prefix_size;
+ int rx_packet_hash_offset;
u8 rx_hash_key[40];
u32 rx_indir_table[128];
bool rx_scatter;
@@ -1009,7 +1014,8 @@ struct efx_mtd_partition {
* @evq_ptr_tbl_base: Event queue pointer table base address
* @evq_rptr_tbl_base: Event queue read-pointer table base address
* @max_dma_mask: Maximum possible DMA mask
- * @rx_buffer_hash_size: Size of hash at start of RX packet
+ * @rx_prefix_size: Size of RX prefix before packet data
+ * @rx_hash_offset: Offset of RX flow hash within prefix
* @rx_buffer_padding: Size of padding at end of RX packet
* @can_rx_scatter: NIC is able to scatter packet to multiple buffers
* @max_interrupt_mode: Highest capability interrupt mode supported
@@ -1126,7 +1132,8 @@ struct efx_nic_type {
unsigned int evq_ptr_tbl_base;
unsigned int evq_rptr_tbl_base;
u64 max_dma_mask;
- unsigned int rx_buffer_hash_size;
+ unsigned int rx_prefix_size;
+ unsigned int rx_hash_offset;
unsigned int rx_buffer_padding;
bool can_rx_scatter;
unsigned int max_interrupt_mode;
diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c
index 1299092..7646bb3 100644
--- a/drivers/net/ethernet/sfc/rx.c
+++ b/drivers/net/ethernet/sfc/rx.c
@@ -61,13 +61,12 @@ static inline u8 *efx_rx_buf_va(struct efx_rx_buffer *buf)
return page_address(buf->page) + buf->page_offset;
}
-static inline u32 efx_rx_buf_hash(const u8 *eh)
+static inline u32 efx_rx_buf_hash(struct efx_nic *efx, const u8 *eh)
{
- /* The ethernet header is always directly after any hash. */
-#if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) || NET_IP_ALIGN % 4 == 0
- return __le32_to_cpup((const __le32 *)(eh - 4));
+#if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS)
+ return __le32_to_cpup((const __le32 *)(eh + efx->rx_packet_hash_offset));
#else
- const u8 *data = eh - 4;
+ const u8 *data = eh + efx->rx_packet_hash_offset;
return (u32)data[0] |
(u32)data[1] << 8 |
(u32)data[2] << 16 |
@@ -439,7 +438,7 @@ efx_rx_packet_gro(struct efx_channel *channel, struct efx_rx_buffer *rx_buf,
}
if (efx->net_dev->features & NETIF_F_RXHASH)
- skb->rxhash = efx_rx_buf_hash(eh);
+ skb->rxhash = efx_rx_buf_hash(efx, eh);
skb->ip_summed = ((rx_buf->flags & EFX_RX_PKT_CSUMMED) ?
CHECKSUM_UNNECESSARY : CHECKSUM_NONE);
@@ -568,8 +567,8 @@ void efx_rx_packet(struct efx_rx_queue *rx_queue, unsigned int index,
*/
prefetch(efx_rx_buf_va(rx_buf));
- rx_buf->page_offset += efx->type->rx_buffer_hash_size;
- rx_buf->len -= efx->type->rx_buffer_hash_size;
+ rx_buf->page_offset += efx->rx_prefix_size;
+ rx_buf->len -= efx->rx_prefix_size;
if (n_frags > 1) {
/* Release/sync DMA mapping for additional fragments.
diff --git a/drivers/net/ethernet/sfc/siena.c b/drivers/net/ethernet/sfc/siena.c
index 3c0a544..89180d4 100644
--- a/drivers/net/ethernet/sfc/siena.c
+++ b/drivers/net/ethernet/sfc/siena.c
@@ -965,7 +965,8 @@ const struct efx_nic_type siena_a0_nic_type = {
.evq_ptr_tbl_base = FR_BZ_EVQ_PTR_TBL,
.evq_rptr_tbl_base = FR_BZ_EVQ_RPTR,
.max_dma_mask = DMA_BIT_MASK(FSF_AZ_TX_KER_BUF_ADDR_WIDTH),
- .rx_buffer_hash_size = 0x10,
+ .rx_prefix_size = FS_BZ_RX_PREFIX_SIZE,
+ .rx_hash_offset = FS_BZ_RX_PREFIX_HASH_OFST,
.rx_buffer_padding = 0,
.can_rx_scatter = true,
.max_interrupt_mode = EFX_INT_MODE_MSIX,
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [v3.11][Regression] skge: add dma_mapping check
From: Joseph Salisbury @ 2013-08-28 15:20 UTC (permalink / raw)
To: stephen; +Cc: davem@davemloft.net, mlindner, netdev@vger.kernel.org,
linux-kernel
Hi Stephen,
A bug was opened against the Ubuntu kernel[0]. A bug has also been
opened in bugzilla[1]. After a kernel bisect, it was found that
reverting the following commit resolved this bug:
commit 136d8f377e1575463b47840bc5f1b22d94bf8f63
Author: stephen hemminger <stephen@networkplumber.org>
Date: Sun Aug 4 17:22:34 2013 -0700
skge: add dma_mapping check
The regression was introduced as of v3.11-rc6.
I see that you are the author of this patch, so I wanted to run this by
you. I was thinking of requesting a revert for v3.11, but I wanted to
get your feedback first.
Thanks,
Joe
[0] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1216745
[1] https://bugzilla.kernel.org/show_bug.cgi?id=60784
^ permalink raw reply
* [PATCH net-next 11/16] sfc: Rename EFX_PAGE_BLOCK_SIZE to EFX_VI_PAGE_SIZE and adjust comments
From: Ben Hutchings @ 2013-08-28 15:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377702837.2264.21.camel@bwh-desktop.uk.level5networks.com>
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/io.h | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/sfc/io.h b/drivers/net/ethernet/sfc/io.h
index 39f6098..19e8b95 100644
--- a/drivers/net/ethernet/sfc/io.h
+++ b/drivers/net/ethernet/sfc/io.h
@@ -204,12 +204,12 @@ static inline void efx_reado_table(struct efx_nic *efx, efx_oword_t *value,
efx_reado(efx, value, reg + index * sizeof(efx_oword_t));
}
-/* Page-mapped register block size */
-#define EFX_PAGE_BLOCK_SIZE 0x2000
+/* Page size used as step between per-VI registers */
+#define EFX_VI_PAGE_SIZE 0x2000
-/* Calculate offset to page-mapped register block */
+/* Calculate offset to page-mapped register */
#define EFX_PAGED_REG(page, reg) \
- ((page) * EFX_PAGE_BLOCK_SIZE + (reg))
+ ((page) * EFX_VI_PAGE_SIZE + (reg))
/* Write the whole of RX_DESC_UPD or TX_DESC_UPD */
static inline void _efx_writeo_page(struct efx_nic *efx, efx_oword_t *value,
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [RFC PATCH net-next 1/4] iptunnels: remove net arg from iptunnel_xmit()
From: Nicolas Dichtel @ 2013-08-28 15:20 UTC (permalink / raw)
To: netdev; +Cc: davem, jesse, dev, Nicolas Dichtel
In-Reply-To: <1377703210-7021-1-git-send-email-nicolas.dichtel@6wind.com>
This argument is not used, let's remove it.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
drivers/net/vxlan.c | 3 +--
include/net/ip_tunnels.h | 3 +--
net/ipv4/ip_tunnel.c | 3 +--
net/ipv4/ip_tunnel_core.c | 3 +--
net/ipv6/sit.c | 4 ++--
net/openvswitch/vport-gre.c | 2 +-
6 files changed, 7 insertions(+), 11 deletions(-)
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 3b21aca0c0c2..3bc27c2ca569 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -1171,8 +1171,7 @@ int vxlan_xmit_skb(struct net *net, struct vxlan_sock *vs,
if (err)
return err;
- return iptunnel_xmit(net, rt, skb, src, dst,
- IPPROTO_UDP, tos, ttl, df);
+ return iptunnel_xmit(rt, skb, src, dst, IPPROTO_UDP, tos, ttl, df);
}
EXPORT_SYMBOL_GPL(vxlan_xmit_skb);
diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
index 0ce316bb3c65..94fe8fd07bc9 100644
--- a/include/net/ip_tunnels.h
+++ b/include/net/ip_tunnels.h
@@ -146,8 +146,7 @@ static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph,
}
int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto);
-int iptunnel_xmit(struct net *net, struct rtable *rt,
- struct sk_buff *skb,
+int iptunnel_xmit(struct rtable *rt, struct sk_buff *skb,
__be32 src, __be32 dst, __u8 proto,
__u8 tos, __u8 ttl, __be16 df);
diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index 830de3f4e293..0a6cf0e69478 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c
@@ -654,8 +654,7 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
}
}
- err = iptunnel_xmit(tunnel->net, rt, skb,
- fl4.saddr, fl4.daddr, protocol,
+ err = iptunnel_xmit(rt, skb, fl4.saddr, fl4.daddr, protocol,
ip_tunnel_ecn_encap(tos, inner_iph, skb), ttl, df);
iptunnel_xmit_stats(err, &dev->stats, dev->tstats);
diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c
index 850525b34899..e82045842351 100644
--- a/net/ipv4/ip_tunnel_core.c
+++ b/net/ipv4/ip_tunnel_core.c
@@ -46,8 +46,7 @@
#include <net/netns/generic.h>
#include <net/rtnetlink.h>
-int iptunnel_xmit(struct net *net, struct rtable *rt,
- struct sk_buff *skb,
+int iptunnel_xmit(struct rtable *rt, struct sk_buff *skb,
__be32 src, __be32 dst, __u8 proto,
__u8 tos, __u8 ttl, __be16 df)
{
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index f18f842ac893..1d1458a3b7c4 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -888,8 +888,8 @@ static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb,
ttl = iph6->hop_limit;
tos = INET_ECN_encapsulate(tos, ipv6_get_dsfield(iph6));
- err = iptunnel_xmit(dev_net(dev), rt, skb, fl4.saddr, fl4.daddr,
- IPPROTO_IPV6, tos, ttl, df);
+ err = iptunnel_xmit(rt, skb, fl4.saddr, fl4.daddr, IPPROTO_IPV6, tos,
+ ttl, df);
iptunnel_xmit_stats(err, &dev->stats, dev->tstats);
return NETDEV_TX_OK;
diff --git a/net/openvswitch/vport-gre.c b/net/openvswitch/vport-gre.c
index 21d5073e148e..9b3713ef831e 100644
--- a/net/openvswitch/vport-gre.c
+++ b/net/openvswitch/vport-gre.c
@@ -176,7 +176,7 @@ static int gre_tnl_send(struct vport *vport, struct sk_buff *skb)
skb->local_df = 1;
- return iptunnel_xmit(net, rt, skb, fl.saddr,
+ return iptunnel_xmit(rt, skb, fl.saddr,
OVS_CB(skb)->tun_key->ipv4_dst, IPPROTO_GRE,
OVS_CB(skb)->tun_key->ipv4_tos,
OVS_CB(skb)->tun_key->ipv4_ttl, df);
--
1.8.2.1
^ permalink raw reply related
* [RFC PATCH net-next 3/4] tunnels: harmonize cleanup done on skb on xmit path
From: Nicolas Dichtel @ 2013-08-28 15:20 UTC (permalink / raw)
To: netdev; +Cc: davem, jesse, dev, Nicolas Dichtel
In-Reply-To: <1377703210-7021-1-git-send-email-nicolas.dichtel@6wind.com>
The goal of this patch is to harmonize cleanup done on a skbuff on xmit path.
Before this patch, behaviors were different depending of the tunnel type.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
drivers/net/vxlan.c | 3 ++-
include/net/ip6_tunnel.h | 1 -
include/net/ip_tunnels.h | 2 +-
net/ipv4/ip_tunnel.c | 6 ++----
net/ipv4/ip_tunnel_core.c | 7 +++----
net/ipv6/ip6_gre.c | 4 ++--
net/ipv6/ip6_tunnel.c | 4 +---
net/ipv6/sit.c | 5 +----
net/openvswitch/vport-gre.c | 2 +-
9 files changed, 13 insertions(+), 21 deletions(-)
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 3bc27c2ca569..aaf9a8f5ee55 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -1171,7 +1171,8 @@ int vxlan_xmit_skb(struct net *net, struct vxlan_sock *vs,
if (err)
return err;
- return iptunnel_xmit(rt, skb, src, dst, IPPROTO_UDP, tos, ttl, df);
+ return iptunnel_xmit(rt, skb, src, dst, IPPROTO_UDP, tos, ttl, df,
+ false);
}
EXPORT_SYMBOL_GPL(vxlan_xmit_skb);
diff --git a/include/net/ip6_tunnel.h b/include/net/ip6_tunnel.h
index 2265b0bf97e5..6d1549c4893c 100644
--- a/include/net/ip6_tunnel.h
+++ b/include/net/ip6_tunnel.h
@@ -75,7 +75,6 @@ static inline void ip6tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
struct net_device_stats *stats = &dev->stats;
int pkt_len, err;
- nf_reset(skb);
pkt_len = skb->len;
err = ip6_local_out(skb);
diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
index 94fe8fd07bc9..a0a4a100f5c9 100644
--- a/include/net/ip_tunnels.h
+++ b/include/net/ip_tunnels.h
@@ -148,7 +148,7 @@ static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph,
int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto);
int iptunnel_xmit(struct rtable *rt, struct sk_buff *skb,
__be32 src, __be32 dst, __u8 proto,
- __u8 tos, __u8 ttl, __be16 df);
+ __u8 tos, __u8 ttl, __be16 df, bool xnet);
static inline void iptunnel_xmit_stats(int err,
struct net_device_stats *err_stats,
diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index b0e74e17088f..88d7d7d1eccb 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c
@@ -614,9 +614,6 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
goto tx_error;
}
- if (!net_eq(tunnel->net, dev_net(dev)))
- skb_scrub_packet(skb, true);
-
if (tunnel->err_count > 0) {
if (time_before(jiffies,
tunnel->err_time + IPTUNNEL_ERR_TIMEO)) {
@@ -655,7 +652,8 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
}
err = iptunnel_xmit(rt, skb, fl4.saddr, fl4.daddr, protocol,
- ip_tunnel_ecn_encap(tos, inner_iph, skb), ttl, df);
+ ip_tunnel_ecn_encap(tos, inner_iph, skb), ttl, df,
+ !net_eq(tunnel->net, dev_net(dev)));
iptunnel_xmit_stats(err, &dev->stats, dev->tstats);
return;
diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c
index e82045842351..d6c856b17fd4 100644
--- a/net/ipv4/ip_tunnel_core.c
+++ b/net/ipv4/ip_tunnel_core.c
@@ -48,16 +48,15 @@
int iptunnel_xmit(struct rtable *rt, struct sk_buff *skb,
__be32 src, __be32 dst, __u8 proto,
- __u8 tos, __u8 ttl, __be16 df)
+ __u8 tos, __u8 ttl, __be16 df, bool xnet)
{
int pkt_len = skb->len;
struct iphdr *iph;
int err;
- nf_reset(skb);
- secpath_reset(skb);
+ skb_scrub_packet(skb, xnet);
+
skb->rxhash = 0;
- skb_dst_drop(skb);
skb_dst_set(skb, &rt->dst);
memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c
index f2d0a42f8057..f179ff1f56ec 100644
--- a/net/ipv6/ip6_gre.c
+++ b/net/ipv6/ip6_gre.c
@@ -694,6 +694,8 @@ static netdev_tx_t ip6gre_xmit2(struct sk_buff *skb,
tunnel->err_count = 0;
}
+ skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(dev)));
+
max_headroom = LL_RESERVED_SPACE(tdev) + gre_hlen + dst->header_len;
if (skb_headroom(skb) < max_headroom || skb_shared(skb) ||
@@ -710,8 +712,6 @@ static netdev_tx_t ip6gre_xmit2(struct sk_buff *skb,
skb = new_skb;
}
- skb_dst_drop(skb);
-
if (fl6->flowi6_mark) {
skb_dst_set(skb, dst);
ndst = NULL;
diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index 72372ac90159..ecbcdbd4bc4f 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -1001,8 +1001,7 @@ static int ip6_tnl_xmit2(struct sk_buff *skb,
goto tx_err_dst_release;
}
- if (!net_eq(t->net, dev_net(dev)))
- skb_scrub_packet(skb, true);
+ skb_scrub_packet(skb, !net_eq(t->net, dev_net(dev)));
/*
* Okay, now see if we can stuff it in the buffer as-is.
@@ -1021,7 +1020,6 @@ static int ip6_tnl_xmit2(struct sk_buff *skb,
consume_skb(skb);
skb = new_skb;
}
- skb_dst_drop(skb);
if (fl6->flowi6_mark) {
skb_dst_set(skb, dst);
ndst = NULL;
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index b2e44f478e14..82b425b9b7d2 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -860,9 +860,6 @@ static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb,
tunnel->err_count = 0;
}
- if (!net_eq(tunnel->net, dev_net(dev)))
- skb_scrub_packet(skb, true);
-
/*
* Okay, now see if we can stuff it in the buffer as-is.
*/
@@ -889,7 +886,7 @@ static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb,
tos = INET_ECN_encapsulate(tos, ipv6_get_dsfield(iph6));
err = iptunnel_xmit(rt, skb, fl4.saddr, fl4.daddr, IPPROTO_IPV6, tos,
- ttl, df);
+ ttl, df, !net_eq(tunnel->net, dev_net(dev)));
iptunnel_xmit_stats(err, &dev->stats, dev->tstats);
return NETDEV_TX_OK;
diff --git a/net/openvswitch/vport-gre.c b/net/openvswitch/vport-gre.c
index 9b3713ef831e..c99dea543d64 100644
--- a/net/openvswitch/vport-gre.c
+++ b/net/openvswitch/vport-gre.c
@@ -179,7 +179,7 @@ static int gre_tnl_send(struct vport *vport, struct sk_buff *skb)
return iptunnel_xmit(rt, skb, fl.saddr,
OVS_CB(skb)->tun_key->ipv4_dst, IPPROTO_GRE,
OVS_CB(skb)->tun_key->ipv4_tos,
- OVS_CB(skb)->tun_key->ipv4_ttl, df);
+ OVS_CB(skb)->tun_key->ipv4_ttl, df, false);
err_free_rt:
ip_rt_put(rt);
error:
--
1.8.2.1
^ permalink raw reply related
* [PATCH net-next 10/16] sfc: Remove early call to efx_nic_type::reconfigure_mac in efx_reset_up()
From: Ben Hutchings @ 2013-08-28 15:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377702837.2264.21.camel@bwh-desktop.uk.level5networks.com>
efx_reset_up() calls efx_nic_type::reconfigure_mac once directly,
then again through efx_start_all() -> efx_start_port() ->
efx->type->reconfigure_mac().
This first call is also made too early to work properly on EF10.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/efx.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/drivers/net/ethernet/sfc/efx.c b/drivers/net/ethernet/sfc/efx.c
index efad5f7..3af28a8 100644
--- a/drivers/net/ethernet/sfc/efx.c
+++ b/drivers/net/ethernet/sfc/efx.c
@@ -2202,8 +2202,6 @@ int efx_reset_up(struct efx_nic *efx, enum reset_type method, bool ok)
"could not restore PHY settings\n");
}
- efx->type->reconfigure_mac(efx);
-
efx_enable_interrupts(efx);
efx_restore_filters(efx);
efx_sriov_reset(efx);
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [RFC PATCH net-next 4/4] tunnels: harmonize cleanup done on skb on rx path
From: Nicolas Dichtel @ 2013-08-28 15:20 UTC (permalink / raw)
To: netdev-u79uwXL29TY76Z2rM5mHXA
Cc: dev-yBygre7rU0TnMu66kgdUjQ, Nicolas Dichtel,
davem-fT/PcQaiUtIeIZ0/mPfg9Q
In-Reply-To: <1377703210-7021-1-git-send-email-nicolas.dichtel-pdR9zngts4EAvxtiuMwx3w@public.gmane.org>
The goal of this patch is to harmonize cleanup done on a skbuff on rx path.
Before this patch, behaviors were different depending of the tunnel type.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel-pdR9zngts4EAvxtiuMwx3w@public.gmane.org>
---
include/net/dst.h | 12 +++++++-----
net/ipv4/ip_tunnel.c | 3 +--
net/ipv4/ipmr.c | 3 +--
net/ipv6/ip6_gre.c | 5 +----
net/ipv6/ip6_tunnel.c | 7 +------
net/ipv6/ip6mr.c | 3 +--
net/ipv6/sit.c | 6 +-----
7 files changed, 13 insertions(+), 26 deletions(-)
diff --git a/include/net/dst.h b/include/net/dst.h
index 1f8fd109e225..3bc4865f8267 100644
--- a/include/net/dst.h
+++ b/include/net/dst.h
@@ -311,11 +311,13 @@ static inline void skb_dst_force(struct sk_buff *skb)
* __skb_tunnel_rx - prepare skb for rx reinsert
* @skb: buffer
* @dev: tunnel device
+ * @net: netns for packet i/o
*
* After decapsulation, packet is going to re-enter (netif_rx()) our stack,
* so make some cleanups. (no accounting done)
*/
-static inline void __skb_tunnel_rx(struct sk_buff *skb, struct net_device *dev)
+static inline void __skb_tunnel_rx(struct sk_buff *skb, struct net_device *dev,
+ struct net *net)
{
skb->dev = dev;
@@ -327,8 +329,7 @@ static inline void __skb_tunnel_rx(struct sk_buff *skb, struct net_device *dev)
if (!skb->l4_rxhash)
skb->rxhash = 0;
skb_set_queue_mapping(skb, 0);
- skb_dst_drop(skb);
- nf_reset(skb);
+ skb_scrub_packet(skb, !net_eq(net, dev_net(dev)));
}
/**
@@ -340,12 +341,13 @@ static inline void __skb_tunnel_rx(struct sk_buff *skb, struct net_device *dev)
* so make some cleanups, and perform accounting.
* Note: this accounting is not SMP safe.
*/
-static inline void skb_tunnel_rx(struct sk_buff *skb, struct net_device *dev)
+static inline void skb_tunnel_rx(struct sk_buff *skb, struct net_device *dev,
+ struct net *net)
{
/* TODO : stats should be SMP safe */
dev->stats.rx_packets++;
dev->stats.rx_bytes += skb->len;
- __skb_tunnel_rx(skb, dev);
+ __skb_tunnel_rx(skb, dev, net);
}
/* Children define the path of the packet through the
diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index 88d7d7d1eccb..ac9fabe0300f 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c
@@ -461,8 +461,7 @@ int ip_tunnel_rcv(struct ip_tunnel *tunnel, struct sk_buff *skb,
skb->dev = tunnel->dev;
}
- if (!net_eq(tunnel->net, dev_net(tunnel->dev)))
- skb_scrub_packet(skb, true);
+ skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(tunnel->dev)));
gro_cells_receive(&tunnel->gro_cells, skb);
return 0;
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index bacc0bcf48ce..9ae54b09254f 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -2067,9 +2067,8 @@ static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb,
skb_reset_network_header(skb);
skb->protocol = htons(ETH_P_IP);
skb->ip_summed = CHECKSUM_NONE;
- skb->pkt_type = PACKET_HOST;
- skb_tunnel_rx(skb, reg_dev);
+ skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev));
netif_rx(skb);
diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c
index f179ff1f56ec..db992a373011 100644
--- a/net/ipv6/ip6_gre.c
+++ b/net/ipv6/ip6_gre.c
@@ -509,8 +509,6 @@ static int ip6gre_rcv(struct sk_buff *skb)
goto drop;
}
- secpath_reset(skb);
-
skb->protocol = gre_proto;
/* WCCP version 1 and 2 protocol decoding.
* - Change protocol to IP
@@ -525,7 +523,6 @@ static int ip6gre_rcv(struct sk_buff *skb)
skb->mac_header = skb->network_header;
__pskb_pull(skb, offset);
skb_postpull_rcsum(skb, skb_transport_header(skb), offset);
- skb->pkt_type = PACKET_HOST;
if (((flags&GRE_CSUM) && csum) ||
(!(flags&GRE_CSUM) && tunnel->parms.i_flags&GRE_CSUM)) {
@@ -557,7 +554,7 @@ static int ip6gre_rcv(struct sk_buff *skb)
skb_postpull_rcsum(skb, eth_hdr(skb), ETH_HLEN);
}
- __skb_tunnel_rx(skb, tunnel->dev);
+ __skb_tunnel_rx(skb, tunnel->dev, tunnel->net);
skb_reset_network_header(skb);
diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index ecbcdbd4bc4f..55999d923f26 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -802,14 +802,12 @@ static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
rcu_read_unlock();
goto discard;
}
- secpath_reset(skb);
skb->mac_header = skb->network_header;
skb_reset_network_header(skb);
skb->protocol = htons(protocol);
- skb->pkt_type = PACKET_HOST;
memset(skb->cb, 0, sizeof(struct inet6_skb_parm));
- __skb_tunnel_rx(skb, t->dev);
+ __skb_tunnel_rx(skb, t->dev, t->net);
err = dscp_ecn_decapsulate(t, ipv6h, skb);
if (unlikely(err)) {
@@ -829,9 +827,6 @@ static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
tstats->rx_packets++;
tstats->rx_bytes += skb->len;
- if (!net_eq(t->net, dev_net(t->dev)))
- skb_scrub_packet(skb, true);
-
netif_rx(skb);
rcu_read_unlock();
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index a60a84ef04f7..f365310bfcca 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -672,9 +672,8 @@ static int pim6_rcv(struct sk_buff *skb)
skb_reset_network_header(skb);
skb->protocol = htons(ETH_P_IPV6);
skb->ip_summed = CHECKSUM_NONE;
- skb->pkt_type = PACKET_HOST;
- skb_tunnel_rx(skb, reg_dev);
+ skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev));
netif_rx(skb);
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index 82b425b9b7d2..19abcc9d6a1a 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -581,12 +581,10 @@ static int ipip6_rcv(struct sk_buff *skb)
tunnel->parms.iph.protocol != 0)
goto out;
- secpath_reset(skb);
skb->mac_header = skb->network_header;
skb_reset_network_header(skb);
IPCB(skb)->flags = 0;
skb->protocol = htons(ETH_P_IPV6);
- skb->pkt_type = PACKET_HOST;
if (tunnel->dev->priv_flags & IFF_ISATAP) {
if (!isatap_chksrc(skb, iph, tunnel)) {
@@ -603,7 +601,7 @@ static int ipip6_rcv(struct sk_buff *skb)
}
}
- __skb_tunnel_rx(skb, tunnel->dev);
+ __skb_tunnel_rx(skb, tunnel->dev, tunnel->net);
err = IP_ECN_decapsulate(iph, skb);
if (unlikely(err)) {
@@ -621,8 +619,6 @@ static int ipip6_rcv(struct sk_buff *skb)
tstats->rx_packets++;
tstats->rx_bytes += skb->len;
- if (!net_eq(tunnel->net, dev_net(tunnel->dev)))
- skb_scrub_packet(skb, true);
netif_rx(skb);
return 0;
--
1.8.2.1
^ permalink raw reply related
* [RFC PATCH net-next 2/4] skb: allow skb_scrub_packet() to be used by tunnels
From: Nicolas Dichtel @ 2013-08-28 15:20 UTC (permalink / raw)
To: netdev-u79uwXL29TY76Z2rM5mHXA
Cc: dev-yBygre7rU0TnMu66kgdUjQ, Nicolas Dichtel,
davem-fT/PcQaiUtIeIZ0/mPfg9Q
In-Reply-To: <1377703210-7021-1-git-send-email-nicolas.dichtel-pdR9zngts4EAvxtiuMwx3w@public.gmane.org>
This function was only used when a packet was sent to another netns. Now, it can
also be used after tunnel encapsulation or decapsulation.
Only skb_orphan() should not be done when a packet is not crossing netns.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel-pdR9zngts4EAvxtiuMwx3w@public.gmane.org>
---
include/linux/skbuff.h | 2 +-
net/core/dev.c | 2 +-
net/core/skbuff.c | 19 ++++++++++++-------
net/ipv4/ip_tunnel.c | 4 ++--
net/ipv6/ip6_tunnel.c | 4 ++--
net/ipv6/sit.c | 4 ++--
6 files changed, 20 insertions(+), 15 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 5ac96f31d546..a3eeabe7c04b 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -2392,7 +2392,7 @@ extern void skb_split(struct sk_buff *skb,
struct sk_buff *skb1, const u32 len);
extern int skb_shift(struct sk_buff *tgt, struct sk_buff *skb,
int shiftlen);
-extern void skb_scrub_packet(struct sk_buff *skb);
+extern void skb_scrub_packet(struct sk_buff *skb, bool xnet);
extern struct sk_buff *skb_segment(struct sk_buff *skb,
netdev_features_t features);
diff --git a/net/core/dev.c b/net/core/dev.c
index 1ed2b66a10a6..d55493e72af6 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1697,7 +1697,7 @@ int dev_forward_skb(struct net_device *dev, struct sk_buff *skb)
* call skb_scrub_packet() after it to clear pkt_type _after_ calling
* eth_type_trans().
*/
- skb_scrub_packet(skb);
+ skb_scrub_packet(skb, true);
return netif_rx(skb);
}
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 2c3d0f53d198..d81cff119f73 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -3500,17 +3500,22 @@ bool skb_try_coalesce(struct sk_buff *to, struct sk_buff *from,
EXPORT_SYMBOL(skb_try_coalesce);
/**
- * skb_scrub_packet - scrub an skb before sending it to another netns
+ * skb_scrub_packet - scrub an skb
*
* @skb: buffer to clean
- *
- * skb_scrub_packet can be used to clean an skb before injecting it in
- * another namespace. We have to clear all information in the skb that
- * could impact namespace isolation.
+ * @xnet: packet is crossing netns
+ *
+ * skb_scrub_packet can be used after encapsulating or decapsulting a packet
+ * into/from a tunnel. Some information have to be cleared during these
+ * operations.
+ * skb_scrub_packet can also be used to clean a skb before injecting it in
+ * another namespace (@xnet == true). We have to clear all information in the
+ * skb that could impact namespace isolation.
*/
-void skb_scrub_packet(struct sk_buff *skb)
+void skb_scrub_packet(struct sk_buff *skb, bool xnet)
{
- skb_orphan(skb);
+ if (xnet)
+ skb_orphan(skb);
skb->tstamp.tv64 = 0;
skb->pkt_type = PACKET_HOST;
skb->skb_iif = 0;
diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index 0a6cf0e69478..b0e74e17088f 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c
@@ -462,7 +462,7 @@ int ip_tunnel_rcv(struct ip_tunnel *tunnel, struct sk_buff *skb,
}
if (!net_eq(tunnel->net, dev_net(tunnel->dev)))
- skb_scrub_packet(skb);
+ skb_scrub_packet(skb, true);
gro_cells_receive(&tunnel->gro_cells, skb);
return 0;
@@ -615,7 +615,7 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
}
if (!net_eq(tunnel->net, dev_net(dev)))
- skb_scrub_packet(skb);
+ skb_scrub_packet(skb, true);
if (tunnel->err_count > 0) {
if (time_before(jiffies,
diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index d6e00a39274c..72372ac90159 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -830,7 +830,7 @@ static int ip6_tnl_rcv(struct sk_buff *skb, __u16 protocol,
tstats->rx_bytes += skb->len;
if (!net_eq(t->net, dev_net(t->dev)))
- skb_scrub_packet(skb);
+ skb_scrub_packet(skb, true);
netif_rx(skb);
@@ -1002,7 +1002,7 @@ static int ip6_tnl_xmit2(struct sk_buff *skb,
}
if (!net_eq(t->net, dev_net(dev)))
- skb_scrub_packet(skb);
+ skb_scrub_packet(skb, true);
/*
* Okay, now see if we can stuff it in the buffer as-is.
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index 1d1458a3b7c4..b2e44f478e14 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -622,7 +622,7 @@ static int ipip6_rcv(struct sk_buff *skb)
tstats->rx_bytes += skb->len;
if (!net_eq(tunnel->net, dev_net(tunnel->dev)))
- skb_scrub_packet(skb);
+ skb_scrub_packet(skb, true);
netif_rx(skb);
return 0;
@@ -861,7 +861,7 @@ static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb,
}
if (!net_eq(tunnel->net, dev_net(dev)))
- skb_scrub_packet(skb);
+ skb_scrub_packet(skb, true);
/*
* Okay, now see if we can stuff it in the buffer as-is.
--
1.8.2.1
^ permalink raw reply related
* [RFC PATCH net-next 0/4] tunnels: harmonize skb scrubbing during encapsulation/decapsulation
From: Nicolas Dichtel @ 2013-08-28 15:20 UTC (permalink / raw)
To: netdev-u79uwXL29TY76Z2rM5mHXA
Cc: dev-yBygre7rU0TnMu66kgdUjQ, davem-fT/PcQaiUtIeIZ0/mPfg9Q
We talk recently about harmonizing tunnels so they behave consistently wrt. SKB
orphaning, cleaning netfilter state, etc.
The goal of this serie is to achieve this.
Note that I only made some basic tests with ipip, sit and ip6_tunnels modules.
drivers/net/vxlan.c | 4 ++--
include/linux/skbuff.h | 2 +-
include/net/dst.h | 12 +++++++-----
include/net/ip6_tunnel.h | 1 -
include/net/ip_tunnels.h | 5 ++---
net/core/dev.c | 2 +-
net/core/skbuff.c | 19 ++++++++++++-------
net/ipv4/ip_tunnel.c | 12 ++++--------
net/ipv4/ip_tunnel_core.c | 10 ++++------
net/ipv4/ipmr.c | 3 +--
net/ipv6/ip6_gre.c | 9 +++------
net/ipv6/ip6_tunnel.c | 11 ++---------
net/ipv6/ip6mr.c | 3 +--
net/ipv6/sit.c | 13 +++----------
net/openvswitch/vport-gre.c | 4 ++--
15 files changed, 45 insertions(+), 65 deletions(-)
Comments are welcome.
Regards,
Nicolas
^ permalink raw reply
* [PATCH net-next 09/16] sfc: use MCDI epoch flag to improve MC reboot detection in the driver
From: Ben Hutchings @ 2013-08-28 15:19 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377702837.2264.21.camel@bwh-desktop.uk.level5networks.com>
From: Daniel Pieczko <dpieczko@solarflare.com>
The Huntington MC will reject all MCDI requests after an MC reboot until it sees
one with the NOT_EPOCH flag clear. This flag is set by default for all requests,
and then cleared on the first request after we detect that an MC reboot has
occurred.
The old MCDI_STATUS_DELAY_COUNT gave a timeout of 10ms, which was not long enough
for the driver to detect that a reboot had occurred based on the warm boot count
while calling efx_mcdi_poll_reboot() from the loop in efx_mcdi_ev_death().
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/mcdi.c | 19 +++++++++++++------
drivers/net/ethernet/sfc/mcdi.h | 2 ++
2 files changed, 15 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/sfc/mcdi.c b/drivers/net/ethernet/sfc/mcdi.c
index 3810634..63121ad 100644
--- a/drivers/net/ethernet/sfc/mcdi.c
+++ b/drivers/net/ethernet/sfc/mcdi.c
@@ -26,9 +26,10 @@
/* A reboot/assertion causes the MCDI status word to be set after the
* command word is set or a REBOOT event is sent. If we notice a reboot
- * via these mechanisms then wait 10ms for the status word to be set. */
+ * via these mechanisms then wait 20ms for the status word to be set.
+ */
#define MCDI_STATUS_DELAY_US 100
-#define MCDI_STATUS_DELAY_COUNT 100
+#define MCDI_STATUS_DELAY_COUNT 200
#define MCDI_STATUS_SLEEP_MS \
(MCDI_STATUS_DELAY_US * MCDI_STATUS_DELAY_COUNT / 1000)
@@ -56,6 +57,7 @@ int efx_mcdi_init(struct efx_nic *efx)
mcdi->mode = MCDI_MODE_POLL;
(void) efx_mcdi_poll_reboot(efx);
+ mcdi->new_epoch = true;
/* Recover from a failed assertion before probing */
return efx_mcdi_handle_assertion(efx);
@@ -85,24 +87,26 @@ static void efx_mcdi_copyin(struct efx_nic *efx, unsigned cmd,
if (efx->type->mcdi_max_ver == 1) {
/* MCDI v1 */
- EFX_POPULATE_DWORD_6(hdr[0],
+ EFX_POPULATE_DWORD_7(hdr[0],
MCDI_HEADER_RESPONSE, 0,
MCDI_HEADER_RESYNC, 1,
MCDI_HEADER_CODE, cmd,
MCDI_HEADER_DATALEN, inlen,
MCDI_HEADER_SEQ, seqno,
- MCDI_HEADER_XFLAGS, xflags);
+ MCDI_HEADER_XFLAGS, xflags,
+ MCDI_HEADER_NOT_EPOCH, !mcdi->new_epoch);
hdr_len = 4;
} else {
/* MCDI v2 */
BUG_ON(inlen > MCDI_CTL_SDU_LEN_MAX_V2);
- EFX_POPULATE_DWORD_6(hdr[0],
+ EFX_POPULATE_DWORD_7(hdr[0],
MCDI_HEADER_RESPONSE, 0,
MCDI_HEADER_RESYNC, 1,
MCDI_HEADER_CODE, MC_CMD_V2_EXTN,
MCDI_HEADER_DATALEN, 0,
MCDI_HEADER_SEQ, seqno,
- MCDI_HEADER_XFLAGS, xflags);
+ MCDI_HEADER_XFLAGS, xflags,
+ MCDI_HEADER_NOT_EPOCH, !mcdi->new_epoch);
EFX_POPULATE_DWORD_2(hdr[1],
MC_CMD_V2_EXTN_IN_EXTENDED_CMD, cmd,
MC_CMD_V2_EXTN_IN_ACTUAL_LEN, inlen);
@@ -373,6 +377,7 @@ int efx_mcdi_rpc_start(struct efx_nic *efx, unsigned cmd,
spin_unlock_bh(&mcdi->iface_lock);
efx_mcdi_copyin(efx, cmd, inbuf, inlen);
+ mcdi->new_epoch = false;
return 0;
}
@@ -435,6 +440,7 @@ int efx_mcdi_rpc_finish(struct efx_nic *efx, unsigned cmd, size_t inlen,
if (rc == -EIO || rc == -EINTR) {
msleep(MCDI_STATUS_SLEEP_MS);
efx_mcdi_poll_reboot(efx);
+ mcdi->new_epoch = true;
}
}
@@ -530,6 +536,7 @@ static void efx_mcdi_ev_death(struct efx_nic *efx, int rc)
break;
udelay(MCDI_STATUS_DELAY_US);
}
+ mcdi->new_epoch = true;
}
spin_unlock(&mcdi->iface_lock);
diff --git a/drivers/net/ethernet/sfc/mcdi.h b/drivers/net/ethernet/sfc/mcdi.h
index 5f67ac3..303d9e8 100644
--- a/drivers/net/ethernet/sfc/mcdi.h
+++ b/drivers/net/ethernet/sfc/mcdi.h
@@ -36,6 +36,7 @@ enum efx_mcdi_mode {
* @state: Request handling state. Waited for by @wq.
* @mode: Poll for mcdi completion, or wait for an mcdi_event.
* @wq: Wait queue for threads waiting for @state != %MCDI_STATE_RUNNING
+ * @new_epoch: Indicates start of day or start of MC reboot recovery
* @iface_lock: Serialises access to all the following fields
* @seqno: The next sequence number to use for mcdi requests.
* @credits: Number of spurious MCDI completion events allowed before we
@@ -49,6 +50,7 @@ struct efx_mcdi_iface {
enum efx_mcdi_mode mode;
wait_queue_head_t wq;
spinlock_t iface_lock;
+ bool new_epoch;
unsigned int credits;
unsigned int seqno;
int resprc;
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox