* [PATCH 2/2] batman-adv: fix packet loss for broadcasted DHCP packets to a server
From: Simon Wunderlich @ 2018-03-26 12:47 UTC (permalink / raw)
To: davem
Cc: netdev, b.a.t.m.a.n, Linus Lüssing, Sven Eckelmann,
Simon Wunderlich
In-Reply-To: <20180326124754.5897-1-sw@simonwunderlich.de>
From: Linus Lüssing <linus.luessing@c0d3.blue>
DHCP connectivity issues can currently occur if the following conditions
are met:
1) A DHCP packet from a client to a server
2) This packet has a multicast destination
3) This destination has a matching entry in the translation table
(FF:FF:FF:FF:FF:FF for IPv4, 33:33:00:01:00:02/33:33:00:01:00:03
for IPv6)
4) The orig-node determined by TT for the multicast destination
does not match the orig-node determined by best-gateway-selection
In this case the DHCP packet will be dropped.
The "gateway-out-of-range" check is supposed to only be applied to
unicasted DHCP packets to a specific DHCP server.
In that case dropping the the unicasted frame forces the client to
retry via a broadcasted one, but now directed to the new best
gateway.
A DHCP packet with broadcast/multicast destination is already ensured to
always be delivered to the best gateway. Dropping a multicasted
DHCP packet here will only prevent completing DHCP as there is no
other fallback.
So far, it seems the unicast check was implicitly performed by
expecting the batadv_transtable_search() to return NULL for multicast
destinations. However, a multicast address could have always ended up in
the translation table and in fact is now common.
To fix this potential loss of a DHCP client-to-server packet to a
multicast address this patch adds an explicit multicast destination
check to reliably bail out of the gateway-out-of-range check for such
destinations.
The issue and fix were tested in the following three node setup:
- Line topology, A-B-C
- A: gateway client, DHCP client
- B: gateway server, hop-penalty increased: 30->60, DHCP server
- C: gateway server, code modifications to announce FF:FF:FF:FF:FF:FF
Without this patch, A would never transmit its DHCP Discover packet
due to an always "out-of-range" condition. With this patch,
a full DHCP handshake between A and B was possible again.
Fixes: be7af5cf9cae ("batman-adv: refactoring gateway handling code")
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
---
net/batman-adv/gateway_client.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/net/batman-adv/gateway_client.c b/net/batman-adv/gateway_client.c
index 37fe9a644f22..808d2dd4a839 100644
--- a/net/batman-adv/gateway_client.c
+++ b/net/batman-adv/gateway_client.c
@@ -746,7 +746,7 @@ bool batadv_gw_out_of_range(struct batadv_priv *bat_priv,
{
struct batadv_neigh_node *neigh_curr = NULL;
struct batadv_neigh_node *neigh_old = NULL;
- struct batadv_orig_node *orig_dst_node;
+ struct batadv_orig_node *orig_dst_node = NULL;
struct batadv_gw_node *gw_node = NULL;
struct batadv_gw_node *curr_gw = NULL;
struct batadv_neigh_ifinfo *curr_ifinfo, *old_ifinfo;
@@ -757,6 +757,9 @@ bool batadv_gw_out_of_range(struct batadv_priv *bat_priv,
vid = batadv_get_vid(skb, 0);
+ if (is_multicast_ether_addr(ethhdr->h_dest))
+ goto out;
+
orig_dst_node = batadv_transtable_search(bat_priv, ethhdr->h_source,
ethhdr->h_dest, vid);
if (!orig_dst_node)
--
2.11.0
^ permalink raw reply related
* [PATCH 1/2] batman-adv: fix multicast-via-unicast transmission with AP isolation
From: Simon Wunderlich @ 2018-03-26 12:47 UTC (permalink / raw)
To: davem
Cc: netdev, b.a.t.m.a.n, Linus Lüssing, Sven Eckelmann,
Simon Wunderlich
In-Reply-To: <20180326124754.5897-1-sw@simonwunderlich.de>
From: Linus Lüssing <linus.luessing@c0d3.blue>
For multicast frames AP isolation is only supposed to be checked on
the receiving nodes and never on the originating one.
Furthermore, the isolation or wifi flag bits should only be intepreted
as such for unicast and never multicast TT entries.
By injecting flags to the multicast TT entry claimed by a single
target node it was verified in tests that this multicast address
becomes unreachable, leading to packet loss.
Omitting the "src" parameter to the batadv_transtable_search() call
successfully skipped the AP isolation check and made the target
reachable again.
Fixes: 1d8ab8d3c176 ("batman-adv: Modified forwarding behaviour for multicast packets")
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
---
net/batman-adv/multicast.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/batman-adv/multicast.c b/net/batman-adv/multicast.c
index d70640135e3a..ee56af5c43e0 100644
--- a/net/batman-adv/multicast.c
+++ b/net/batman-adv/multicast.c
@@ -814,8 +814,8 @@ static struct batadv_orig_node *
batadv_mcast_forw_tt_node_get(struct batadv_priv *bat_priv,
struct ethhdr *ethhdr)
{
- return batadv_transtable_search(bat_priv, ethhdr->h_source,
- ethhdr->h_dest, BATADV_NO_FLAGS);
+ return batadv_transtable_search(bat_priv, NULL, ethhdr->h_dest,
+ BATADV_NO_FLAGS);
}
/**
--
2.11.0
^ permalink raw reply related
* [PATCH 0/2] pull request for net: batman-adv 2018-03-26
From: Simon Wunderlich @ 2018-03-26 12:47 UTC (permalink / raw)
To: davem; +Cc: netdev, b.a.t.m.a.n, Simon Wunderlich
Hi David,
here are two late bugfixes for batman-adv which we would like to have in net
if still possible.
Please pull or let me know of any problem!
Thank you,
Simon
The following changes since commit fc04fdb2c8a894283259f5621d31d75610701091:
batman-adv: Fix skbuff rcsum on packet reroute (2018-03-18 13:20:32 +0100)
are available in the git repository at:
git://git.open-mesh.org/linux-merge.git tags/batadv-net-for-davem-20180326
for you to fetch changes up to a752c0a4524889cdc0765925258fd1fd72344100:
batman-adv: fix packet loss for broadcasted DHCP packets to a server (2018-03-24 10:25:49 +0100)
----------------------------------------------------------------
Here are some batman-adv bugfixes:
- fix multicast-via-unicast transmissions for AP isolation and gateway
extension, by Linus Luessing (2 patches)
----------------------------------------------------------------
Linus Lüssing (2):
batman-adv: fix multicast-via-unicast transmission with AP isolation
batman-adv: fix packet loss for broadcasted DHCP packets to a server
net/batman-adv/gateway_client.c | 5 ++++-
net/batman-adv/multicast.c | 4 ++--
2 files changed, 6 insertions(+), 3 deletions(-)
^ permalink raw reply
* [RFC PATCH v3] net: phy: Added dev-addr device tree binding and MDIO enablement of non-standard register addr space
From: Vicentiu Galanopulo @ 2018-03-26 12:18 UTC (permalink / raw)
To: netdev, linux-kernel, robh+dt, mark.rutland, davem, marcel,
devicetree
Cc: madalin.bucur, alexandru.marginean
Enabling the discovery on the MDIO bus
of the INPHI PHY which has a vendor specific
address space for accessing the C45 MDIO registers
A search of the dev-addr property is done in
of_mdiobus_register. If the property is found in
the PHY node, of_mdiobus_register_vend_spec_phy()
is called. This is a wrapper function for
of_mdiobus_register_phy() which finds the device
in package based on dev-addr and fills devices_addrs:
devices_addrs is a new field added to phy_c45_device_ids.
This new field will store the dev-addr property on the same
index where the device in package has been found.
of_mdiobus_register_phy() signature call has been changed
and struct phy_c45_device_ids *c45_ids was added.
If c45_ids is not NULL, get_vend_spec_addr_phy_device() is
called and c45_ids are propagated all the way to
get_phy_c45_ids().
Having dev-addr stored in devices_addrs, in get_phy_c45_ids(),
when probing the identifiers, dev-addr can be extracted from
devices_addrs and probed if devices_addrs[current_identifier]
is not 0.
This should probably be in a separate patch but I
found it in the context of enabling the discovery of
the INPHI PHY: num_ids in get_phy_c45_ids,
has the value 8 (ARRAY_SIZE(c45_ids->device_ids)),
but the u32 *devs can store 32 devices in the bitfield.
If a device is stored in *devs, in bits 32 to 9, it
will not be found. This is the reason for changing
in phy.h, the size of device_ids array
Signed-off-by: Vicentiu Galanopulo <vicentiu.galanopulo@nxp.com>
---
Documentation/devicetree/bindings/net/phy.txt | 6 ++
drivers/net/phy/phy_device.c | 49 ++++++++++-
drivers/of/of_mdio.c | 113 +++++++++++++++++++++++++-
include/linux/phy.h | 16 +++-
4 files changed, 176 insertions(+), 8 deletions(-)
diff --git a/Documentation/devicetree/bindings/net/phy.txt b/Documentation/devicetree/bindings/net/phy.txt
index d2169a5..82692e2 100644
--- a/Documentation/devicetree/bindings/net/phy.txt
+++ b/Documentation/devicetree/bindings/net/phy.txt
@@ -61,6 +61,11 @@ Optional Properties:
- reset-deassert-us: Delay after the reset was deasserted in microseconds.
If this property is missing the delay will be skipped.
+- dev-addr: If set, it indicates the device address of the PHY to be used
+ when accessing the C45 PHY registers over MDIO. It is used for vendor specific
+ register space addresses that do no conform to standard address for the MDIO
+ registers (e.g. MMD30)
+
Example:
ethernet-phy@0 {
@@ -72,4 +77,5 @@ ethernet-phy@0 {
reset-gpios = <&gpio1 4 GPIO_ACTIVE_LOW>;
reset-assert-us = <1000>;
reset-deassert-us = <2000>;
+ dev-addr = <0x1e>;
};
diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
index b285323..3cbae4b 100644
--- a/drivers/net/phy/phy_device.c
+++ b/drivers/net/phy/phy_device.c
@@ -457,7 +457,7 @@ static int get_phy_c45_devs_in_pkg(struct mii_bus *bus, int addr, int dev_addr,
static int get_phy_c45_ids(struct mii_bus *bus, int addr, u32 *phy_id,
struct phy_c45_device_ids *c45_ids) {
int phy_reg;
- int i, reg_addr;
+ int i, reg_addr, dev_addr;
const int num_ids = ARRAY_SIZE(c45_ids->device_ids);
u32 *devs = &c45_ids->devices_in_package;
@@ -493,13 +493,23 @@ static int get_phy_c45_ids(struct mii_bus *bus, int addr, u32 *phy_id,
if (!(c45_ids->devices_in_package & (1 << i)))
continue;
- reg_addr = MII_ADDR_C45 | i << 16 | MII_PHYSID1;
+ /* if c45_ids->devices_addrs for the current id is not 0,
+ * then dev-addr was defined in the PHY device tree node,
+ * and the PHY has been seen as a valid device, and added
+ * in the package. In this case we can use the
+ * dev-addr(c45_ids->devices_addrs[i]) to do the MDIO
+ * reading of the PHY ID.
+ */
+ dev_addr = !!c45_ids->devices_addrs[i] ?
+ c45_ids->devices_addrs[i] : i;
+
+ reg_addr = MII_ADDR_C45 | dev_addr << 16 | MII_PHYSID1;
phy_reg = mdiobus_read(bus, addr, reg_addr);
if (phy_reg < 0)
return -EIO;
c45_ids->device_ids[i] = (phy_reg & 0xffff) << 16;
- reg_addr = MII_ADDR_C45 | i << 16 | MII_PHYSID2;
+ reg_addr = MII_ADDR_C45 | dev_addr << 16 | MII_PHYSID2;
phy_reg = mdiobus_read(bus, addr, reg_addr);
if (phy_reg < 0)
return -EIO;
@@ -551,6 +561,39 @@ static int get_phy_id(struct mii_bus *bus, int addr, u32 *phy_id,
}
/**
+ * get_vend_spec_addr_phy_device - reads the specified PHY device
+ * and returns its @phy_device struct
+ * @bus: the target MII bus
+ * @addr: PHY address on the MII bus
+ * @is_c45: If true the PHY uses the 802.3 clause 45 protocol
+ * @c45_ids: Query the c45_ids to see if a PHY with a vendor specific
+ * register address space was defined in the PHY device tree
+ * node by adding the "dev-addr" property to the node.
+ * Store the c45 ID information about the rest of the PHYs
+ * found PHYs on the MDIO bus during probing.
+ *
+ * Description: Reads the ID registers of the PHY at @addr on the
+ * @bus, then allocates and returns the phy_device to represent it.
+ */
+struct phy_device *get_vend_spec_addr_phy_device(struct mii_bus *bus,
+ int addr, bool is_c45,
+ struct phy_c45_device_ids *c45_ids)
+{
+ u32 phy_id = 0;
+ int r;
+
+ r = get_phy_id(bus, addr, &phy_id, is_c45, c45_ids);
+ if (r)
+ return ERR_PTR(r);
+
+ /* If the phy_id is mostly Fs, there is no device there */
+ if ((phy_id & 0x1fffffff) == 0x1fffffff)
+ return ERR_PTR(-ENODEV);
+
+ return phy_device_create(bus, addr, phy_id, is_c45, c45_ids);
+}
+
+/**
* get_phy_device - reads the specified PHY device and returns its @phy_device
* struct
* @bus: the target MII bus
diff --git a/drivers/of/of_mdio.c b/drivers/of/of_mdio.c
index 8c0c927..52e8bfb 100644
--- a/drivers/of/of_mdio.c
+++ b/drivers/of/of_mdio.c
@@ -45,7 +45,8 @@ static int of_get_phy_id(struct device_node *device, u32 *phy_id)
}
static int of_mdiobus_register_phy(struct mii_bus *mdio,
- struct device_node *child, u32 addr)
+ struct device_node *child, u32 addr,
+ struct phy_c45_device_ids *c45_ids)
{
struct phy_device *phy;
bool is_c45;
@@ -58,7 +59,12 @@ static int of_mdiobus_register_phy(struct mii_bus *mdio,
if (!is_c45 && !of_get_phy_id(child, &phy_id))
phy = phy_device_create(mdio, addr, phy_id, 0, NULL);
else
- phy = get_phy_device(mdio, addr, is_c45);
+ if (c45_ids)
+ phy = get_vend_spec_addr_phy_device(mdio,
+ addr, is_c45,
+ c45_ids);
+ else
+ phy = get_phy_device(mdio, addr, is_c45);
if (IS_ERR(phy))
return PTR_ERR(phy);
@@ -190,6 +196,72 @@ static bool of_mdiobus_child_is_phy(struct device_node *child)
return false;
}
+static void of_fill_c45_devices_addrs(u32 dev_addr,
+ struct phy_c45_device_ids *c45_ids)
+{
+ int i;
+ const int num_ids = ARRAY_SIZE(c45_ids->device_ids);
+
+ /* Search through all Device Identifiers
+ * and set dev_addr in c45_ids->devices_addrs,
+ * if the device bit is set in
+ * c45_ids->devices_in_package
+ */
+ for (i = 1; i < num_ids; i++) {
+ if (!(c45_ids->devices_in_package & (1 << i)))
+ continue;
+
+ c45_ids->devices_addrs[i] = dev_addr;
+ }
+}
+
+static int of_find_devaddr_in_pkg(struct mii_bus *bus, u32 addr, u32 dev_addr,
+ struct phy_c45_device_ids *c45_ids)
+{
+ u32 *devs = &c45_ids->devices_in_package;
+ int phy_reg, reg_addr;
+
+ reg_addr = MII_ADDR_C45 | dev_addr << 16 | MDIO_DEVS2;
+ phy_reg = mdiobus_read(bus, addr, reg_addr);
+ if (phy_reg < 0)
+ return -EIO;
+
+ *devs = (phy_reg & 0xffff) << 16;
+
+ reg_addr = MII_ADDR_C45 | dev_addr << 16 | MDIO_DEVS1;
+ phy_reg = mdiobus_read(bus, addr, reg_addr);
+ if (phy_reg < 0)
+ return -EIO;
+
+ *devs |= (phy_reg & 0xffff);
+
+ return 0;
+}
+
+/*
+ * Finds the device in package and populates the c45_ids
+ * if any device is found at dev_addr address. After this
+ * the PHY is registered
+ */
+static int of_mdiobus_register_vend_spec_phy(struct mii_bus *mdio,
+ struct device_node *child,
+ u32 addr, u32 dev_addr)
+{
+ struct phy_c45_device_ids c45_ids = {0};
+ int dev_err = 0;
+
+ if (!dev_addr)
+ goto register_phy;
+
+ dev_err = of_find_devaddr_in_pkg(mdio, addr, dev_addr, &c45_ids);
+
+ if (!dev_err)
+ of_fill_c45_devices_addrs(dev_addr, &c45_ids);
+
+register_phy:
+ return of_mdiobus_register_phy(mdio, child, addr, &c45_ids);
+}
+
/**
* of_mdiobus_register - Register mii_bus and create PHYs from the device tree
* @mdio: pointer to mii_bus structure
@@ -202,7 +274,10 @@ int of_mdiobus_register(struct mii_bus *mdio, struct device_node *np)
{
struct device_node *child;
bool scanphys = false;
+ bool dev_addr_found = true;
int addr, rc;
+ int dev_addr = 0;
+ int ret;
/* Do not continue if the node is disabled */
if (!of_device_is_available(np))
@@ -226,6 +301,14 @@ int of_mdiobus_register(struct mii_bus *mdio, struct device_node *np)
/* Loop over the child nodes and register a phy_device for each phy */
for_each_available_child_of_node(np, child) {
+ /* Check if dev-addr is set in the PHY node */
+ ret = of_property_read_u32(child, "dev-addr", &dev_addr);
+
+ if (ret < 0) {
+ /* either not set or invalid */
+ dev_addr_found = false;
+ }
+
addr = of_mdio_parse_addr(&mdio->dev, child);
if (addr < 0) {
scanphys = true;
@@ -233,7 +316,14 @@ int of_mdiobus_register(struct mii_bus *mdio, struct device_node *np)
}
if (of_mdiobus_child_is_phy(child))
- rc = of_mdiobus_register_phy(mdio, child, addr);
+ if (dev_addr_found)
+ rc = of_mdiobus_register_vend_spec_phy(mdio,
+ child,
+ addr,
+ dev_addr);
+ else
+ rc = of_mdiobus_register_phy(mdio, child,
+ addr, NULL);
else
rc = of_mdiobus_register_device(mdio, child, addr);
@@ -248,8 +338,16 @@ int of_mdiobus_register(struct mii_bus *mdio, struct device_node *np)
if (!scanphys)
return 0;
+ /* reset device found variable */
+ dev_addr_found = true;
+
/* auto scan for PHYs with empty reg property */
for_each_available_child_of_node(np, child) {
+ /* Check if dev-addr is set in the PHY node,
+ * for PHYs which don't have reg property set
+ */
+ ret = of_property_read_u32(child, "dev-addr", &dev_addr);
+
/* Skip PHYs with reg property set */
if (of_find_property(child, "reg", NULL))
continue;
@@ -264,7 +362,14 @@ int of_mdiobus_register(struct mii_bus *mdio, struct device_node *np)
child->name, addr);
if (of_mdiobus_child_is_phy(child)) {
- rc = of_mdiobus_register_phy(mdio, child, addr);
+ if (dev_addr_found)
+ rc = of_mdiobus_register_vend_spec_phy(mdio,
+ child,
+ addr,
+ dev_addr);
+ else
+ rc = of_mdiobus_register_phy(mdio, child,
+ addr, NULL);
if (rc && rc != -ENODEV)
goto unregister;
}
diff --git a/include/linux/phy.h b/include/linux/phy.h
index 5a9b175..f665945 100644
--- a/include/linux/phy.h
+++ b/include/linux/phy.h
@@ -357,10 +357,13 @@ enum phy_state {
* struct phy_c45_device_ids - 802.3-c45 Device Identifiers
* @devices_in_package: Bit vector of devices present.
* @device_ids: The device identifer for each present device.
+ * @devices_addrs: The devices addresses from the device tree
+ * for each present device.
*/
struct phy_c45_device_ids {
u32 devices_in_package;
- u32 device_ids[8];
+ u32 device_ids[32];
+ u32 devices_addrs[32];
};
/* phy_device: An instance of a PHY
@@ -904,6 +907,9 @@ struct phy_device *phy_device_create(struct mii_bus *bus, int addr, int phy_id,
struct phy_c45_device_ids *c45_ids);
#if IS_ENABLED(CONFIG_PHYLIB)
struct phy_device *get_phy_device(struct mii_bus *bus, int addr, bool is_c45);
+struct phy_device *get_vend_spec_addr_phy_device(struct mii_bus *bus, int addr,
+ bool is_c45,
+ struct phy_c45_device_ids *c45_ids);
int phy_device_register(struct phy_device *phy);
void phy_device_free(struct phy_device *phydev);
#else
@@ -913,6 +919,14 @@ struct phy_device *get_phy_device(struct mii_bus *bus, int addr, bool is_c45)
return NULL;
}
+static inline
+struct phy_device *get_vend_spec_addr_phy_device(struct mii_bus *bus, int addr,
+ bool is_c45,
+ struct phy_c45_device_ids *c45_ids)
+{
+ return NULL;
+}
+
static inline int phy_device_register(struct phy_device *phy)
{
return 0;
--
2.7.4
^ permalink raw reply related
* [PATCH net-next 15/15] mlxsw: spectrum: Add multicast router trap for PIMv6
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Add a new trap for PIMv6 packets. As PIM already has a designated trap
group [ & rate limiter], simply use the same for PIMv6 as well.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 1 +
drivers/net/ethernet/mellanox/mlxsw/trap.h | 1 +
2 files changed, 2 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 7885fc475f7e..4aa84442e357 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -3380,6 +3380,7 @@ static const struct mlxsw_listener mlxsw_sp_listener[] = {
MLXSW_SP_RXL_NO_MARK(ACL0, TRAP_TO_CPU, IP2ME, false),
/* Multicast Router Traps */
MLXSW_SP_RXL_MARK(IPV4_PIM, TRAP_TO_CPU, PIM, false),
+ MLXSW_SP_RXL_MARK(IPV6_PIM, TRAP_TO_CPU, PIM, false),
MLXSW_SP_RXL_MARK(RPF, TRAP_TO_CPU, RPF, false),
MLXSW_SP_RXL_MARK(ACL1, TRAP_TO_CPU, MULTICAST, false),
MLXSW_SP_RXL_MR_MARK(ACL2, TRAP_TO_CPU, MULTICAST, false),
diff --git a/drivers/net/ethernet/mellanox/mlxsw/trap.h b/drivers/net/ethernet/mellanox/mlxsw/trap.h
index ec6cef8267ae..399e9d6993f7 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/trap.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/trap.h
@@ -77,6 +77,7 @@ enum {
MLXSW_TRAP_ID_IPV6_DHCP = 0x69,
MLXSW_TRAP_ID_IPV6_ALL_ROUTERS_LINK = 0x6F,
MLXSW_TRAP_ID_RTR_INGRESS0 = 0x70,
+ MLXSW_TRAP_ID_IPV6_PIM = 0x79,
MLXSW_TRAP_ID_IPV4_BGP = 0x88,
MLXSW_TRAP_ID_IPV6_BGP = 0x89,
MLXSW_TRAP_ID_L3_IPV6_ROUTER_SOLICITATION = 0x8A,
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 14/15] mlxsw: spectrum_router: Process IP6MR fib notification
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Following previous patches driver is ready to handle notifications
arriving from ip6mr - start processing those when they arrive following
the same manner ipmr currently goes through.
This should enable driver to start offloading ipv6 multicast routes.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c | 17 ++++++++++-------
1 file changed, 10 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 14b887b96cbb..a9ccd974c620 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -5396,13 +5396,10 @@ static int __mlxsw_sp_router_set_abort_trap(struct mlxsw_sp *mlxsw_sp,
static struct mlxsw_sp_mr_table *
mlxsw_sp_router_fibmr_family_to_table(struct mlxsw_sp_vr *vr, int family)
{
- switch (family) {
- case RTNL_FAMILY_IPMR:
- return vr->mr_table[MLXSW_SP_L3_PROTO_IPV4];
- default:
- WARN_ON(1);
+ if (family == RTNL_FAMILY_IPMR)
return vr->mr_table[MLXSW_SP_L3_PROTO_IPV4];
- }
+ else
+ return vr->mr_table[MLXSW_SP_L3_PROTO_IPV6];
}
static int mlxsw_sp_router_fibmr_add(struct mlxsw_sp *mlxsw_sp,
@@ -5844,6 +5841,10 @@ static int mlxsw_sp_router_fib_rule_event(unsigned long event,
if (!ipmr_rule_default(rule) && !rule->l3mdev)
err = -1;
break;
+ case RTNL_FAMILY_IP6MR:
+ if (!ip6mr_rule_default(rule) && !rule->l3mdev)
+ err = -1;
+ break;
}
if (err < 0)
@@ -5863,7 +5864,8 @@ static int mlxsw_sp_router_fib_event(struct notifier_block *nb,
if (!net_eq(info->net, &init_net) ||
(info->family != AF_INET && info->family != AF_INET6 &&
- info->family != RTNL_FAMILY_IPMR))
+ info->family != RTNL_FAMILY_IPMR &&
+ info->family != RTNL_FAMILY_IP6MR))
return NOTIFY_DONE;
router = container_of(nb, struct mlxsw_sp_router, fib_nb);
@@ -5893,6 +5895,7 @@ static int mlxsw_sp_router_fib_event(struct notifier_block *nb,
INIT_WORK(&fib_work->work, mlxsw_sp_router_fib6_event_work);
mlxsw_sp_router_fib6_event(fib_work, info);
break;
+ case RTNL_FAMILY_IP6MR:
case RTNL_FAMILY_IPMR:
INIT_WORK(&fib_work->work, mlxsw_sp_router_fibmr_event_work);
mlxsw_sp_router_fibmr_event(fib_work, info);
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 13/15] mlxsw: spectrum_mr: Add ipv6 specific operations
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Populate the various operation structures meant for IPv6 with logic
unique to that protocol suite.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c | 56 +++++++++++++++++++++++
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h | 1 +
2 files changed, 57 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
index ba6bcbe6f75b..a82539609d49 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
@@ -33,6 +33,7 @@
*/
#include <linux/rhashtable.h>
+#include <net/ipv6.h>
#include "spectrum_mr.h"
#include "spectrum_router.h"
@@ -840,11 +841,60 @@ static bool mlxsw_sp_mr_vif4_is_regular(const struct mlxsw_sp_mr_vif *vif)
return !(vif->vif_flags & (VIFF_TUNNEL | VIFF_REGISTER));
}
+static bool
+mlxsw_sp_mr_route6_validate(const struct mlxsw_sp_mr_table *mr_table,
+ const struct mr_mfc *c)
+{
+ struct mfc6_cache *mfc = (struct mfc6_cache *) c;
+
+ /* If the route is a (*,*) route, abort, as these kind of routes are
+ * used for proxy routes.
+ */
+ if (ipv6_addr_any(&mfc->mf6c_origin) &&
+ ipv6_addr_any(&mfc->mf6c_mcastgrp)) {
+ dev_warn(mr_table->mlxsw_sp->bus_info->dev,
+ "Offloading proxy routes is not supported.\n");
+ return false;
+ }
+ return true;
+}
+
+static void mlxsw_sp_mr_route6_key(struct mlxsw_sp_mr_table *mr_table,
+ struct mlxsw_sp_mr_route_key *key,
+ struct mr_mfc *c)
+{
+ const struct mfc6_cache *mfc = (struct mfc6_cache *) c;
+
+ memset(key, 0, sizeof(*key));
+ key->vrid = mr_table->vr_id;
+ key->proto = MLXSW_SP_L3_PROTO_IPV6;
+ key->group.addr6 = mfc->mf6c_mcastgrp;
+ memset(&key->group_mask.addr6, 0xff, sizeof(key->group_mask.addr6));
+ key->source.addr6 = mfc->mf6c_origin;
+ if (!ipv6_addr_any(&mfc->mf6c_origin))
+ memset(&key->source_mask.addr6, 0xff,
+ sizeof(key->source_mask.addr6));
+}
+
+static bool mlxsw_sp_mr_route6_starg(const struct mlxsw_sp_mr_table *mr_table,
+ const struct mlxsw_sp_mr_route *mr_route)
+{
+ return ipv6_addr_any(&mr_route->key.source_mask.addr6);
+}
+
+static bool mlxsw_sp_mr_vif6_is_regular(const struct mlxsw_sp_mr_vif *vif)
+{
+ return !(vif->vif_flags & MIFF_REGISTER);
+}
+
static struct
mlxsw_sp_mr_vif_ops mlxsw_sp_mr_vif_ops_arr[] = {
{
.is_regular = mlxsw_sp_mr_vif4_is_regular,
},
+ {
+ .is_regular = mlxsw_sp_mr_vif6_is_regular,
+ },
};
static struct
@@ -854,6 +904,12 @@ mlxsw_sp_mr_table_ops mlxsw_sp_mr_table_ops_arr[] = {
.key_create = mlxsw_sp_mr_route4_key,
.is_route_starg = mlxsw_sp_mr_route4_starg,
},
+ {
+ .is_route_valid = mlxsw_sp_mr_route6_validate,
+ .key_create = mlxsw_sp_mr_route6_key,
+ .is_route_starg = mlxsw_sp_mr_route6_starg,
+ },
+
};
struct mlxsw_sp_mr_table *mlxsw_sp_mr_table_create(struct mlxsw_sp *mlxsw_sp,
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h
index 6f94fc9eea1b..7c864a86811d 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h
@@ -36,6 +36,7 @@
#define _MLXSW_SPECTRUM_MCROUTER_H
#include <linux/mroute.h>
+#include <linux/mroute6.h>
#include "spectrum_router.h"
#include "spectrum.h"
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 12/15] mlxsw: spectrum_router: Make IPMR-related APIs family agnostic
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
spectrum_router and spectrum_mr have several APIs that are used to
manipulate configurations originating from ipmr fib notifications.
Following previous patches all the protocol-specifics that are necessary
for the configuration are hidden within spectrum_mr. This allows us to
clean the API and make sure that other than choosing the mr_table based
on the fib notification family, spectrum_router wouldn't care about the
source of the notification when passing it onward to spectrum_mr.
This would later allow us to leverage the same code for fib
notifications originating from ip6mr.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c | 52 ++++++++--------------
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h | 8 ++--
.../net/ethernet/mellanox/mlxsw/spectrum_router.c | 33 ++++++++++----
3 files changed, 47 insertions(+), 46 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
index ae8c5b5387db..ba6bcbe6f75b 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
@@ -323,8 +323,8 @@ static void mlxsw_sp_mr_route_erase(struct mlxsw_sp_mr_table *mr_table,
}
static struct mlxsw_sp_mr_route *
-mlxsw_sp_mr_route4_create(struct mlxsw_sp_mr_table *mr_table,
- struct mfc_cache *mfc)
+mlxsw_sp_mr_route_create(struct mlxsw_sp_mr_table *mr_table,
+ struct mr_mfc *mfc)
{
struct mlxsw_sp_mr_route_vif_entry *rve, *tmp;
struct mlxsw_sp_mr_route *mr_route;
@@ -339,13 +339,13 @@ mlxsw_sp_mr_route4_create(struct mlxsw_sp_mr_table *mr_table,
/* Find min_mtu and link iVIF and eVIFs */
mr_route->min_mtu = ETH_MAX_MTU;
- mr_cache_hold(&mfc->_c);
- mr_route->mfc = &mfc->_c;
+ mr_cache_hold(mfc);
+ mr_route->mfc = mfc;
mr_table->ops->key_create(mr_table, &mr_route->key, mr_route->mfc);
mr_route->mr_table = mr_table;
for (i = 0; i < MAXVIFS; i++) {
- if (mfc->_c.mfc_un.res.ttls[i] != 255) {
+ if (mfc->mfc_un.res.ttls[i] != 255) {
err = mlxsw_sp_mr_route_evif_link(mr_route,
&mr_table->vifs[i]);
if (err)
@@ -356,44 +356,30 @@ mlxsw_sp_mr_route4_create(struct mlxsw_sp_mr_table *mr_table,
}
}
mlxsw_sp_mr_route_ivif_link(mr_route,
- &mr_table->vifs[mfc->_c.mfc_parent]);
+ &mr_table->vifs[mfc->mfc_parent]);
mr_route->route_action = mlxsw_sp_mr_route_action(mr_route);
return mr_route;
err:
- mr_cache_put(&mfc->_c);
+ mr_cache_put(mfc);
list_for_each_entry_safe(rve, tmp, &mr_route->evif_list, route_node)
mlxsw_sp_mr_route_evif_unlink(rve);
kfree(mr_route);
return ERR_PTR(err);
}
-static void mlxsw_sp_mr_route4_destroy(struct mlxsw_sp_mr_table *mr_table,
- struct mlxsw_sp_mr_route *mr_route)
+static void mlxsw_sp_mr_route_destroy(struct mlxsw_sp_mr_table *mr_table,
+ struct mlxsw_sp_mr_route *mr_route)
{
struct mlxsw_sp_mr_route_vif_entry *rve, *tmp;
mlxsw_sp_mr_route_ivif_unlink(mr_route);
- mr_cache_put((struct mr_mfc *)mr_route->mfc);
+ mr_cache_put(mr_route->mfc);
list_for_each_entry_safe(rve, tmp, &mr_route->evif_list, route_node)
mlxsw_sp_mr_route_evif_unlink(rve);
kfree(mr_route);
}
-static void mlxsw_sp_mr_route_destroy(struct mlxsw_sp_mr_table *mr_table,
- struct mlxsw_sp_mr_route *mr_route)
-{
- switch (mr_table->proto) {
- case MLXSW_SP_L3_PROTO_IPV4:
- mlxsw_sp_mr_route4_destroy(mr_table, mr_route);
- break;
- case MLXSW_SP_L3_PROTO_IPV6:
- /* fall through */
- default:
- WARN_ON_ONCE(1);
- }
-}
-
static void mlxsw_sp_mr_mfc_offload_set(struct mlxsw_sp_mr_route *mr_route,
bool offload)
{
@@ -422,18 +408,18 @@ static void __mlxsw_sp_mr_route_del(struct mlxsw_sp_mr_table *mr_table,
mlxsw_sp_mr_route_destroy(mr_table, mr_route);
}
-int mlxsw_sp_mr_route4_add(struct mlxsw_sp_mr_table *mr_table,
- struct mfc_cache *mfc, bool replace)
+int mlxsw_sp_mr_route_add(struct mlxsw_sp_mr_table *mr_table,
+ struct mr_mfc *mfc, bool replace)
{
struct mlxsw_sp_mr_route *mr_orig_route = NULL;
struct mlxsw_sp_mr_route *mr_route;
int err;
- if (!mr_table->ops->is_route_valid(mr_table, &mfc->_c))
+ if (!mr_table->ops->is_route_valid(mr_table, mfc))
return -EINVAL;
/* Create a new route */
- mr_route = mlxsw_sp_mr_route4_create(mr_table, mfc);
+ mr_route = mlxsw_sp_mr_route_create(mr_table, mfc);
if (IS_ERR(mr_route))
return PTR_ERR(mr_route);
@@ -478,7 +464,7 @@ int mlxsw_sp_mr_route4_add(struct mlxsw_sp_mr_table *mr_table,
&mr_orig_route->ht_node,
mlxsw_sp_mr_route_ht_params);
list_del(&mr_orig_route->node);
- mlxsw_sp_mr_route4_destroy(mr_table, mr_orig_route);
+ mlxsw_sp_mr_route_destroy(mr_table, mr_orig_route);
}
mlxsw_sp_mr_mfc_offload_update(mr_route);
@@ -491,17 +477,17 @@ int mlxsw_sp_mr_route4_add(struct mlxsw_sp_mr_table *mr_table,
list_del(&mr_route->node);
err_no_orig_route:
err_duplicate_route:
- mlxsw_sp_mr_route4_destroy(mr_table, mr_route);
+ mlxsw_sp_mr_route_destroy(mr_table, mr_route);
return err;
}
-void mlxsw_sp_mr_route4_del(struct mlxsw_sp_mr_table *mr_table,
- struct mfc_cache *mfc)
+void mlxsw_sp_mr_route_del(struct mlxsw_sp_mr_table *mr_table,
+ struct mr_mfc *mfc)
{
struct mlxsw_sp_mr_route *mr_route;
struct mlxsw_sp_mr_route_key key;
- mr_table->ops->key_create(mr_table, &key, &mfc->_c);
+ mr_table->ops->key_create(mr_table, &key, mfc);
mr_route = rhashtable_lookup_fast(&mr_table->route_ht, &key,
mlxsw_sp_mr_route_ht_params);
if (mr_route)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h
index 5d26a122af49..6f94fc9eea1b 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h
@@ -109,10 +109,10 @@ struct mlxsw_sp_mr_table;
int mlxsw_sp_mr_init(struct mlxsw_sp *mlxsw_sp,
const struct mlxsw_sp_mr_ops *mr_ops);
void mlxsw_sp_mr_fini(struct mlxsw_sp *mlxsw_sp);
-int mlxsw_sp_mr_route4_add(struct mlxsw_sp_mr_table *mr_table,
- struct mfc_cache *mfc, bool replace);
-void mlxsw_sp_mr_route4_del(struct mlxsw_sp_mr_table *mr_table,
- struct mfc_cache *mfc);
+int mlxsw_sp_mr_route_add(struct mlxsw_sp_mr_table *mr_table,
+ struct mr_mfc *mfc, bool replace);
+void mlxsw_sp_mr_route_del(struct mlxsw_sp_mr_table *mr_table,
+ struct mr_mfc *mfc);
int mlxsw_sp_mr_vif_add(struct mlxsw_sp_mr_table *mr_table,
struct net_device *dev, vifi_t vif_index,
unsigned long vif_flags,
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index caa17e0a39fa..14b887b96cbb 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -5393,10 +5393,23 @@ static int __mlxsw_sp_router_set_abort_trap(struct mlxsw_sp *mlxsw_sp,
return 0;
}
+static struct mlxsw_sp_mr_table *
+mlxsw_sp_router_fibmr_family_to_table(struct mlxsw_sp_vr *vr, int family)
+{
+ switch (family) {
+ case RTNL_FAMILY_IPMR:
+ return vr->mr_table[MLXSW_SP_L3_PROTO_IPV4];
+ default:
+ WARN_ON(1);
+ return vr->mr_table[MLXSW_SP_L3_PROTO_IPV4];
+ }
+}
+
static int mlxsw_sp_router_fibmr_add(struct mlxsw_sp *mlxsw_sp,
struct mfc_entry_notifier_info *men_info,
bool replace)
{
+ struct mlxsw_sp_mr_table *mrt;
struct mlxsw_sp_vr *vr;
if (mlxsw_sp->router->aborted)
@@ -5406,14 +5419,14 @@ static int mlxsw_sp_router_fibmr_add(struct mlxsw_sp *mlxsw_sp,
if (IS_ERR(vr))
return PTR_ERR(vr);
- return mlxsw_sp_mr_route4_add(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4],
- (struct mfc_cache *) men_info->mfc,
- replace);
+ mrt = mlxsw_sp_router_fibmr_family_to_table(vr, men_info->info.family);
+ return mlxsw_sp_mr_route_add(mrt, men_info->mfc, replace);
}
static void mlxsw_sp_router_fibmr_del(struct mlxsw_sp *mlxsw_sp,
struct mfc_entry_notifier_info *men_info)
{
+ struct mlxsw_sp_mr_table *mrt;
struct mlxsw_sp_vr *vr;
if (mlxsw_sp->router->aborted)
@@ -5423,8 +5436,8 @@ static void mlxsw_sp_router_fibmr_del(struct mlxsw_sp *mlxsw_sp,
if (WARN_ON(!vr))
return;
- mlxsw_sp_mr_route4_del(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4],
- (struct mfc_cache *) men_info->mfc);
+ mrt = mlxsw_sp_router_fibmr_family_to_table(vr, men_info->info.family);
+ mlxsw_sp_mr_route_del(mrt, men_info->mfc);
mlxsw_sp_vr_put(mlxsw_sp, vr);
}
@@ -5432,6 +5445,7 @@ static int
mlxsw_sp_router_fibmr_vif_add(struct mlxsw_sp *mlxsw_sp,
struct vif_entry_notifier_info *ven_info)
{
+ struct mlxsw_sp_mr_table *mrt;
struct mlxsw_sp_rif *rif;
struct mlxsw_sp_vr *vr;
@@ -5442,9 +5456,9 @@ mlxsw_sp_router_fibmr_vif_add(struct mlxsw_sp *mlxsw_sp,
if (IS_ERR(vr))
return PTR_ERR(vr);
+ mrt = mlxsw_sp_router_fibmr_family_to_table(vr, ven_info->info.family);
rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, ven_info->dev);
- return mlxsw_sp_mr_vif_add(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4],
- ven_info->dev,
+ return mlxsw_sp_mr_vif_add(mrt, ven_info->dev,
ven_info->vif_index,
ven_info->vif_flags, rif);
}
@@ -5453,6 +5467,7 @@ static void
mlxsw_sp_router_fibmr_vif_del(struct mlxsw_sp *mlxsw_sp,
struct vif_entry_notifier_info *ven_info)
{
+ struct mlxsw_sp_mr_table *mrt;
struct mlxsw_sp_vr *vr;
if (mlxsw_sp->router->aborted)
@@ -5462,8 +5477,8 @@ mlxsw_sp_router_fibmr_vif_del(struct mlxsw_sp *mlxsw_sp,
if (WARN_ON(!vr))
return;
- mlxsw_sp_mr_vif_del(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4],
- ven_info->vif_index);
+ mrt = mlxsw_sp_router_fibmr_family_to_table(vr, ven_info->info.family);
+ mlxsw_sp_mr_vif_del(mrt, ven_info->vif_index);
mlxsw_sp_vr_put(mlxsw_sp, vr);
}
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 11/15] mlxsw: spectrum_mr: Convert into using mr_mfc
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Current multicast routing logic in driver assumes it's always meant to
deal with IPv4 multicast routes, leaving several placeholders for
later IPv6 support [currently usually WARN()].
This patch changes the driver's internal multicast route struct into
holding a common mr_mfc instead of the IPv4 mfc_cache.
The various placeholders are grouped into 2:
- Functions that require only the common bits; These remain and the
restriction for IPv4-only is lifted.
- Function that require IPv4-specifics - for handling these functions
we add sets of operations that encapsulate the protocol differences
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c | 192 ++++++++++++----------
1 file changed, 108 insertions(+), 84 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
index a8f4927ff150..ae8c5b5387db 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
@@ -47,6 +47,11 @@ struct mlxsw_sp_mr {
/* priv has to be always the last item */
};
+struct mlxsw_sp_mr_vif;
+struct mlxsw_sp_mr_vif_ops {
+ bool (*is_regular)(const struct mlxsw_sp_mr_vif *vif);
+};
+
struct mlxsw_sp_mr_vif {
struct net_device *dev;
const struct mlxsw_sp_rif *rif;
@@ -61,6 +66,9 @@ struct mlxsw_sp_mr_vif {
* instance is used as an ingress VIF
*/
struct list_head route_ivif_list;
+
+ /* Protocol specific operations for a VIF */
+ const struct mlxsw_sp_mr_vif_ops *ops;
};
struct mlxsw_sp_mr_route_vif_entry {
@@ -70,6 +78,17 @@ struct mlxsw_sp_mr_route_vif_entry {
struct mlxsw_sp_mr_route *mr_route;
};
+struct mlxsw_sp_mr_table;
+struct mlxsw_sp_mr_table_ops {
+ bool (*is_route_valid)(const struct mlxsw_sp_mr_table *mr_table,
+ const struct mr_mfc *mfc);
+ void (*key_create)(struct mlxsw_sp_mr_table *mr_table,
+ struct mlxsw_sp_mr_route_key *key,
+ struct mr_mfc *mfc);
+ bool (*is_route_starg)(const struct mlxsw_sp_mr_table *mr_table,
+ const struct mlxsw_sp_mr_route *mr_route);
+};
+
struct mlxsw_sp_mr_table {
struct list_head node;
enum mlxsw_sp_l3proto proto;
@@ -78,6 +97,7 @@ struct mlxsw_sp_mr_table {
struct mlxsw_sp_mr_vif vifs[MAXVIFS];
struct list_head route_list;
struct rhashtable route_ht;
+ const struct mlxsw_sp_mr_table_ops *ops;
char catchall_route_priv[0];
/* catchall_route_priv has to be always the last item */
};
@@ -88,7 +108,7 @@ struct mlxsw_sp_mr_route {
struct mlxsw_sp_mr_route_key key;
enum mlxsw_sp_mr_route_action route_action;
u16 min_mtu;
- struct mfc_cache *mfc4;
+ struct mr_mfc *mfc;
void *route_priv;
const struct mlxsw_sp_mr_table *mr_table;
/* A list of route_vif_entry structs that point to the egress VIFs */
@@ -104,14 +124,9 @@ static const struct rhashtable_params mlxsw_sp_mr_route_ht_params = {
.automatic_shrinking = true,
};
-static bool mlxsw_sp_mr_vif_regular(const struct mlxsw_sp_mr_vif *vif)
-{
- return !(vif->vif_flags & (VIFF_TUNNEL | VIFF_REGISTER));
-}
-
static bool mlxsw_sp_mr_vif_valid(const struct mlxsw_sp_mr_vif *vif)
{
- return mlxsw_sp_mr_vif_regular(vif) && vif->dev && vif->rif;
+ return vif->ops->is_regular(vif) && vif->dev && vif->rif;
}
static bool mlxsw_sp_mr_vif_exists(const struct mlxsw_sp_mr_vif *vif)
@@ -122,18 +137,9 @@ static bool mlxsw_sp_mr_vif_exists(const struct mlxsw_sp_mr_vif *vif)
static bool
mlxsw_sp_mr_route_ivif_in_evifs(const struct mlxsw_sp_mr_route *mr_route)
{
- vifi_t ivif;
+ vifi_t ivif = mr_route->mfc->mfc_parent;
- switch (mr_route->mr_table->proto) {
- case MLXSW_SP_L3_PROTO_IPV4:
- ivif = mr_route->mfc4->_c.mfc_parent;
- return mr_route->mfc4->_c.mfc_un.res.ttls[ivif] != 255;
- case MLXSW_SP_L3_PROTO_IPV6:
- /* fall through */
- default:
- WARN_ON_ONCE(1);
- }
- return false;
+ return mr_route->mfc->mfc_un.res.ttls[ivif] != 255;
}
static int
@@ -149,19 +155,6 @@ mlxsw_sp_mr_route_valid_evifs_num(const struct mlxsw_sp_mr_route *mr_route)
return valid_evifs;
}
-static bool mlxsw_sp_mr_route_starg(const struct mlxsw_sp_mr_route *mr_route)
-{
- switch (mr_route->mr_table->proto) {
- case MLXSW_SP_L3_PROTO_IPV4:
- return mr_route->key.source_mask.addr4 == htonl(INADDR_ANY);
- case MLXSW_SP_L3_PROTO_IPV6:
- /* fall through */
- default:
- WARN_ON_ONCE(1);
- }
- return false;
-}
-
static enum mlxsw_sp_mr_route_action
mlxsw_sp_mr_route_action(const struct mlxsw_sp_mr_route *mr_route)
{
@@ -174,7 +167,8 @@ mlxsw_sp_mr_route_action(const struct mlxsw_sp_mr_route *mr_route)
/* The kernel does not match a (*,G) route that the ingress interface is
* not one of the egress interfaces, so trap these kind of routes.
*/
- if (mlxsw_sp_mr_route_starg(mr_route) &&
+ if (mr_route->mr_table->ops->is_route_starg(mr_route->mr_table,
+ mr_route) &&
!mlxsw_sp_mr_route_ivif_in_evifs(mr_route))
return MLXSW_SP_MR_ROUTE_ACTION_TRAP;
@@ -195,25 +189,11 @@ mlxsw_sp_mr_route_action(const struct mlxsw_sp_mr_route *mr_route)
static enum mlxsw_sp_mr_route_prio
mlxsw_sp_mr_route_prio(const struct mlxsw_sp_mr_route *mr_route)
{
- return mlxsw_sp_mr_route_starg(mr_route) ?
+ return mr_route->mr_table->ops->is_route_starg(mr_route->mr_table,
+ mr_route) ?
MLXSW_SP_MR_ROUTE_PRIO_STARG : MLXSW_SP_MR_ROUTE_PRIO_SG;
}
-static void mlxsw_sp_mr_route4_key(struct mlxsw_sp_mr_table *mr_table,
- struct mlxsw_sp_mr_route_key *key,
- const struct mfc_cache *mfc)
-{
- bool starg = (mfc->mfc_origin == htonl(INADDR_ANY));
-
- memset(key, 0, sizeof(*key));
- key->vrid = mr_table->vr_id;
- key->proto = mr_table->proto;
- key->group.addr4 = mfc->mfc_mcastgrp;
- key->group_mask.addr4 = htonl(0xffffffff);
- key->source.addr4 = mfc->mfc_origin;
- key->source_mask.addr4 = htonl(starg ? 0 : 0xffffffff);
-}
-
static int mlxsw_sp_mr_route_evif_link(struct mlxsw_sp_mr_route *mr_route,
struct mlxsw_sp_mr_vif *mr_vif)
{
@@ -356,12 +336,13 @@ mlxsw_sp_mr_route4_create(struct mlxsw_sp_mr_table *mr_table,
if (!mr_route)
return ERR_PTR(-ENOMEM);
INIT_LIST_HEAD(&mr_route->evif_list);
- mlxsw_sp_mr_route4_key(mr_table, &mr_route->key, mfc);
/* Find min_mtu and link iVIF and eVIFs */
mr_route->min_mtu = ETH_MAX_MTU;
mr_cache_hold(&mfc->_c);
- mr_route->mfc4 = mfc;
+ mr_route->mfc = &mfc->_c;
+ mr_table->ops->key_create(mr_table, &mr_route->key, mr_route->mfc);
+
mr_route->mr_table = mr_table;
for (i = 0; i < MAXVIFS; i++) {
if (mfc->_c.mfc_un.res.ttls[i] != 255) {
@@ -393,7 +374,7 @@ static void mlxsw_sp_mr_route4_destroy(struct mlxsw_sp_mr_table *mr_table,
struct mlxsw_sp_mr_route_vif_entry *rve, *tmp;
mlxsw_sp_mr_route_ivif_unlink(mr_route);
- mr_cache_put((struct mr_mfc *)mr_route->mfc4);
+ mr_cache_put((struct mr_mfc *)mr_route->mfc);
list_for_each_entry_safe(rve, tmp, &mr_route->evif_list, route_node)
mlxsw_sp_mr_route_evif_unlink(rve);
kfree(mr_route);
@@ -416,18 +397,10 @@ static void mlxsw_sp_mr_route_destroy(struct mlxsw_sp_mr_table *mr_table,
static void mlxsw_sp_mr_mfc_offload_set(struct mlxsw_sp_mr_route *mr_route,
bool offload)
{
- switch (mr_route->mr_table->proto) {
- case MLXSW_SP_L3_PROTO_IPV4:
- if (offload)
- mr_route->mfc4->_c.mfc_flags |= MFC_OFFLOAD;
- else
- mr_route->mfc4->_c.mfc_flags &= ~MFC_OFFLOAD;
- break;
- case MLXSW_SP_L3_PROTO_IPV6:
- /* fall through */
- default:
- WARN_ON_ONCE(1);
- }
+ if (offload)
+ mr_route->mfc->mfc_flags |= MFC_OFFLOAD;
+ else
+ mr_route->mfc->mfc_flags &= ~MFC_OFFLOAD;
}
static void mlxsw_sp_mr_mfc_offload_update(struct mlxsw_sp_mr_route *mr_route)
@@ -456,15 +429,8 @@ int mlxsw_sp_mr_route4_add(struct mlxsw_sp_mr_table *mr_table,
struct mlxsw_sp_mr_route *mr_route;
int err;
- /* If the route is a (*,*) route, abort, as these kind of routes are
- * used for proxy routes.
- */
- if (mfc->mfc_origin == htonl(INADDR_ANY) &&
- mfc->mfc_mcastgrp == htonl(INADDR_ANY)) {
- dev_warn(mr_table->mlxsw_sp->bus_info->dev,
- "Offloading proxy routes is not supported.\n");
+ if (!mr_table->ops->is_route_valid(mr_table, &mfc->_c))
return -EINVAL;
- }
/* Create a new route */
mr_route = mlxsw_sp_mr_route4_create(mr_table, mfc);
@@ -535,7 +501,7 @@ void mlxsw_sp_mr_route4_del(struct mlxsw_sp_mr_table *mr_table,
struct mlxsw_sp_mr_route *mr_route;
struct mlxsw_sp_mr_route_key key;
- mlxsw_sp_mr_route4_key(mr_table, &key, mfc);
+ mr_table->ops->key_create(mr_table, &key, &mfc->_c);
mr_route = rhashtable_lookup_fast(&mr_table->route_ht, &key,
mlxsw_sp_mr_route_ht_params);
if (mr_route)
@@ -840,6 +806,70 @@ void mlxsw_sp_mr_rif_mtu_update(struct mlxsw_sp_mr_table *mr_table,
}
}
+/* Protocol specific functions */
+static bool
+mlxsw_sp_mr_route4_validate(const struct mlxsw_sp_mr_table *mr_table,
+ const struct mr_mfc *c)
+{
+ struct mfc_cache *mfc = (struct mfc_cache *) c;
+
+ /* If the route is a (*,*) route, abort, as these kind of routes are
+ * used for proxy routes.
+ */
+ if (mfc->mfc_origin == htonl(INADDR_ANY) &&
+ mfc->mfc_mcastgrp == htonl(INADDR_ANY)) {
+ dev_warn(mr_table->mlxsw_sp->bus_info->dev,
+ "Offloading proxy routes is not supported.\n");
+ return false;
+ }
+ return true;
+}
+
+static void mlxsw_sp_mr_route4_key(struct mlxsw_sp_mr_table *mr_table,
+ struct mlxsw_sp_mr_route_key *key,
+ struct mr_mfc *c)
+{
+ const struct mfc_cache *mfc = (struct mfc_cache *) c;
+ bool starg;
+
+ starg = (mfc->mfc_origin == htonl(INADDR_ANY));
+
+ memset(key, 0, sizeof(*key));
+ key->vrid = mr_table->vr_id;
+ key->proto = MLXSW_SP_L3_PROTO_IPV4;
+ key->group.addr4 = mfc->mfc_mcastgrp;
+ key->group_mask.addr4 = htonl(0xffffffff);
+ key->source.addr4 = mfc->mfc_origin;
+ key->source_mask.addr4 = htonl(starg ? 0 : 0xffffffff);
+}
+
+static bool mlxsw_sp_mr_route4_starg(const struct mlxsw_sp_mr_table *mr_table,
+ const struct mlxsw_sp_mr_route *mr_route)
+{
+ return mr_route->key.source_mask.addr4 == htonl(INADDR_ANY);
+}
+
+static bool mlxsw_sp_mr_vif4_is_regular(const struct mlxsw_sp_mr_vif *vif)
+{
+ return !(vif->vif_flags & (VIFF_TUNNEL | VIFF_REGISTER));
+}
+
+static struct
+mlxsw_sp_mr_vif_ops mlxsw_sp_mr_vif_ops_arr[] = {
+ {
+ .is_regular = mlxsw_sp_mr_vif4_is_regular,
+ },
+};
+
+static struct
+mlxsw_sp_mr_table_ops mlxsw_sp_mr_table_ops_arr[] = {
+ {
+ .is_route_valid = mlxsw_sp_mr_route4_validate,
+ .key_create = mlxsw_sp_mr_route4_key,
+ .is_route_starg = mlxsw_sp_mr_route4_starg,
+ },
+};
+
struct mlxsw_sp_mr_table *mlxsw_sp_mr_table_create(struct mlxsw_sp *mlxsw_sp,
u32 vr_id,
enum mlxsw_sp_l3proto proto)
@@ -867,6 +897,7 @@ struct mlxsw_sp_mr_table *mlxsw_sp_mr_table_create(struct mlxsw_sp *mlxsw_sp,
mr_table->vr_id = vr_id;
mr_table->mlxsw_sp = mlxsw_sp;
mr_table->proto = proto;
+ mr_table->ops = &mlxsw_sp_mr_table_ops_arr[proto];
INIT_LIST_HEAD(&mr_table->route_list);
err = rhashtable_init(&mr_table->route_ht,
@@ -877,6 +908,7 @@ struct mlxsw_sp_mr_table *mlxsw_sp_mr_table_create(struct mlxsw_sp *mlxsw_sp,
for (i = 0; i < MAXVIFS; i++) {
INIT_LIST_HEAD(&mr_table->vifs[i].route_evif_list);
INIT_LIST_HEAD(&mr_table->vifs[i].route_ivif_list);
+ mr_table->vifs[i].ops = &mlxsw_sp_mr_vif_ops_arr[proto];
}
err = mr->mr_ops->route_create(mlxsw_sp, mr->priv,
@@ -943,18 +975,10 @@ static void mlxsw_sp_mr_route_stats_update(struct mlxsw_sp *mlxsw_sp,
mr->mr_ops->route_stats(mlxsw_sp, mr_route->route_priv, &packets,
&bytes);
- switch (mr_route->mr_table->proto) {
- case MLXSW_SP_L3_PROTO_IPV4:
- if (mr_route->mfc4->_c.mfc_un.res.pkt != packets)
- mr_route->mfc4->_c.mfc_un.res.lastuse = jiffies;
- mr_route->mfc4->_c.mfc_un.res.pkt = packets;
- mr_route->mfc4->_c.mfc_un.res.bytes = bytes;
- break;
- case MLXSW_SP_L3_PROTO_IPV6:
- /* fall through */
- default:
- WARN_ON_ONCE(1);
- }
+ if (mr_route->mfc->mfc_un.res.pkt != packets)
+ mr_route->mfc->mfc_un.res.lastuse = jiffies;
+ mr_route->mfc->mfc_un.res.pkt = packets;
+ mr_route->mfc->mfc_un.res.bytes = bytes;
}
static void mlxsw_sp_mr_stats_update(struct work_struct *work)
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 10/15] mlxsw: spectrum_router: Support IPv6 multicast to host CPU
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
A step toward offloading IPv6 routing, this adds an additional
multicast routing table meant for IPv6 [with its underlying TCAM
region] and populates the default rule for IPv6 multicast packets.
Following this, ingress IPv6 multicast packets would be trapped and
delivered to the host CPU.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
.../net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c | 105 ++++++++++++++-------
.../net/ethernet/mellanox/mlxsw/spectrum_router.c | 69 +++++++++-----
.../net/ethernet/mellanox/mlxsw/spectrum_router.h | 1 +
3 files changed, 121 insertions(+), 54 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c
index 4c7f32d4288d..4f4c0d311883 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c
@@ -51,7 +51,7 @@ struct mlxsw_sp_mr_tcam_region {
};
struct mlxsw_sp_mr_tcam {
- struct mlxsw_sp_mr_tcam_region ipv4_tcam_region;
+ struct mlxsw_sp_mr_tcam_region tcam_regions[MLXSW_SP_L3_PROTO_MAX];
};
/* This struct maps to one RIGR2 register entry */
@@ -316,20 +316,37 @@ static int mlxsw_sp_mr_tcam_route_replace(struct mlxsw_sp *mlxsw_sp,
mlxsw_afa_block_first_set(afa_block));
break;
case MLXSW_SP_L3_PROTO_IPV6:
- default:
- WARN_ON_ONCE(1);
+ mlxsw_reg_rmft2_ipv6_pack(rmft2_pl, true, parman_item->index,
+ key->vrid,
+ MLXSW_REG_RMFT2_IRIF_MASK_IGNORE, 0,
+ key->group.addr6,
+ key->group_mask.addr6,
+ key->source.addr6,
+ key->source_mask.addr6,
+ mlxsw_afa_block_first_set(afa_block));
}
return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rmft2), rmft2_pl);
}
static int mlxsw_sp_mr_tcam_route_remove(struct mlxsw_sp *mlxsw_sp, int vrid,
+ struct mlxsw_sp_mr_route_key *key,
struct parman_item *parman_item)
{
+ struct in6_addr zero_addr = IN6ADDR_ANY_INIT;
char rmft2_pl[MLXSW_REG_RMFT2_LEN];
- mlxsw_reg_rmft2_ipv4_pack(rmft2_pl, false, parman_item->index, vrid,
- 0, 0, 0, 0, 0, 0, NULL);
+ switch (key->proto) {
+ case MLXSW_SP_L3_PROTO_IPV4:
+ mlxsw_reg_rmft2_ipv4_pack(rmft2_pl, false, parman_item->index,
+ vrid, 0, 0, 0, 0, 0, 0, NULL);
+ break;
+ case MLXSW_SP_L3_PROTO_IPV6:
+ mlxsw_reg_rmft2_ipv6_pack(rmft2_pl, false, parman_item->index,
+ vrid, 0, 0, zero_addr, zero_addr,
+ zero_addr, zero_addr, NULL);
+ break;
+ }
return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rmft2), rmft2_pl);
}
@@ -353,27 +370,30 @@ mlxsw_sp_mr_tcam_erif_populate(struct mlxsw_sp *mlxsw_sp,
return 0;
}
+static struct mlxsw_sp_mr_tcam_region *
+mlxsw_sp_mr_tcam_protocol_region(struct mlxsw_sp_mr_tcam *mr_tcam,
+ enum mlxsw_sp_l3proto proto)
+{
+ return &mr_tcam->tcam_regions[proto];
+}
+
static int
mlxsw_sp_mr_tcam_route_parman_item_add(struct mlxsw_sp_mr_tcam *mr_tcam,
struct mlxsw_sp_mr_tcam_route *route,
enum mlxsw_sp_mr_route_prio prio)
{
- struct parman_prio *parman_prio = NULL;
+ struct mlxsw_sp_mr_tcam_region *tcam_region;
int err;
- switch (route->key.proto) {
- case MLXSW_SP_L3_PROTO_IPV4:
- parman_prio = &mr_tcam->ipv4_tcam_region.parman_prios[prio];
- err = parman_item_add(mr_tcam->ipv4_tcam_region.parman,
- parman_prio, &route->parman_item);
- if (err)
- return err;
- break;
- case MLXSW_SP_L3_PROTO_IPV6:
- default:
- WARN_ON_ONCE(1);
- }
- route->parman_prio = parman_prio;
+ tcam_region = mlxsw_sp_mr_tcam_protocol_region(mr_tcam,
+ route->key.proto);
+ err = parman_item_add(tcam_region->parman,
+ &tcam_region->parman_prios[prio],
+ &route->parman_item);
+ if (err)
+ return err;
+
+ route->parman_prio = &tcam_region->parman_prios[prio];
return 0;
}
@@ -381,15 +401,13 @@ static void
mlxsw_sp_mr_tcam_route_parman_item_remove(struct mlxsw_sp_mr_tcam *mr_tcam,
struct mlxsw_sp_mr_tcam_route *route)
{
- switch (route->key.proto) {
- case MLXSW_SP_L3_PROTO_IPV4:
- parman_item_remove(mr_tcam->ipv4_tcam_region.parman,
- route->parman_prio, &route->parman_item);
- break;
- case MLXSW_SP_L3_PROTO_IPV6:
- default:
- WARN_ON_ONCE(1);
- }
+ struct mlxsw_sp_mr_tcam_region *tcam_region;
+
+ tcam_region = mlxsw_sp_mr_tcam_protocol_region(mr_tcam,
+ route->key.proto);
+
+ parman_item_remove(tcam_region->parman,
+ route->parman_prio, &route->parman_item);
}
static int
@@ -462,7 +480,7 @@ static void mlxsw_sp_mr_tcam_route_destroy(struct mlxsw_sp *mlxsw_sp,
struct mlxsw_sp_mr_tcam *mr_tcam = priv;
mlxsw_sp_mr_tcam_route_remove(mlxsw_sp, route->key.vrid,
- &route->parman_item);
+ &route->key, &route->parman_item);
mlxsw_sp_mr_tcam_route_parman_item_remove(mr_tcam, route);
mlxsw_sp_mr_tcam_afa_block_destroy(route->afa_block);
mlxsw_sp_flow_counter_free(mlxsw_sp, route->counter_index);
@@ -806,21 +824,42 @@ mlxsw_sp_mr_tcam_region_fini(struct mlxsw_sp_mr_tcam_region *mr_tcam_region)
static int mlxsw_sp_mr_tcam_init(struct mlxsw_sp *mlxsw_sp, void *priv)
{
struct mlxsw_sp_mr_tcam *mr_tcam = priv;
+ struct mlxsw_sp_mr_tcam_region *region = &mr_tcam->tcam_regions[0];
+ u32 rtar_key;
+ int err;
if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MC_ERIF_LIST_ENTRIES) ||
!MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_MAX_TCAM_RULES))
return -EIO;
- return mlxsw_sp_mr_tcam_region_init(mlxsw_sp,
- &mr_tcam->ipv4_tcam_region,
- MLXSW_REG_RTAR_KEY_TYPE_IPV4_MULTICAST);
+ rtar_key = MLXSW_REG_RTAR_KEY_TYPE_IPV4_MULTICAST;
+ err = mlxsw_sp_mr_tcam_region_init(mlxsw_sp,
+ ®ion[MLXSW_SP_L3_PROTO_IPV4],
+ rtar_key);
+ if (err)
+ return err;
+
+ rtar_key = MLXSW_REG_RTAR_KEY_TYPE_IPV6_MULTICAST;
+ err = mlxsw_sp_mr_tcam_region_init(mlxsw_sp,
+ ®ion[MLXSW_SP_L3_PROTO_IPV6],
+ rtar_key);
+ if (err)
+ goto err_ipv6_region_init;
+
+ return 0;
+
+err_ipv6_region_init:
+ mlxsw_sp_mr_tcam_region_fini(®ion[MLXSW_SP_L3_PROTO_IPV4]);
+ return err;
}
static void mlxsw_sp_mr_tcam_fini(void *priv)
{
struct mlxsw_sp_mr_tcam *mr_tcam = priv;
+ struct mlxsw_sp_mr_tcam_region *region = &mr_tcam->tcam_regions[0];
- mlxsw_sp_mr_tcam_region_fini(&mr_tcam->ipv4_tcam_region);
+ mlxsw_sp_mr_tcam_region_fini(®ion[MLXSW_SP_L3_PROTO_IPV6]);
+ mlxsw_sp_mr_tcam_region_fini(®ion[MLXSW_SP_L3_PROTO_IPV4]);
}
const struct mlxsw_sp_mr_ops mlxsw_sp_mr_tcam_ops = {
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index be241c708bb9..caa17e0a39fa 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -467,7 +467,7 @@ struct mlxsw_sp_vr {
unsigned int rif_count;
struct mlxsw_sp_fib *fib4;
struct mlxsw_sp_fib *fib6;
- struct mlxsw_sp_mr_table *mr4_table;
+ struct mlxsw_sp_mr_table *mr_table[MLXSW_SP_L3_PROTO_MAX];
};
static const struct rhashtable_params mlxsw_sp_fib_ht_params;
@@ -711,7 +711,9 @@ static void mlxsw_sp_lpm_fini(struct mlxsw_sp *mlxsw_sp)
static bool mlxsw_sp_vr_is_used(const struct mlxsw_sp_vr *vr)
{
- return !!vr->fib4 || !!vr->fib6 || !!vr->mr4_table;
+ return !!vr->fib4 || !!vr->fib6 ||
+ !!vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] ||
+ !!vr->mr_table[MLXSW_SP_L3_PROTO_IPV6];
}
static struct mlxsw_sp_vr *mlxsw_sp_vr_find_unused(struct mlxsw_sp *mlxsw_sp)
@@ -789,7 +791,7 @@ static struct mlxsw_sp_vr *mlxsw_sp_vr_create(struct mlxsw_sp *mlxsw_sp,
u32 tb_id,
struct netlink_ext_ack *extack)
{
- struct mlxsw_sp_mr_table *mr4_table;
+ struct mlxsw_sp_mr_table *mr4_table, *mr6_table;
struct mlxsw_sp_fib *fib4;
struct mlxsw_sp_fib *fib6;
struct mlxsw_sp_vr *vr;
@@ -812,15 +814,25 @@ static struct mlxsw_sp_vr *mlxsw_sp_vr_create(struct mlxsw_sp *mlxsw_sp,
MLXSW_SP_L3_PROTO_IPV4);
if (IS_ERR(mr4_table)) {
err = PTR_ERR(mr4_table);
- goto err_mr_table_create;
+ goto err_mr4_table_create;
}
+ mr6_table = mlxsw_sp_mr_table_create(mlxsw_sp, vr->id,
+ MLXSW_SP_L3_PROTO_IPV6);
+ if (IS_ERR(mr6_table)) {
+ err = PTR_ERR(mr6_table);
+ goto err_mr6_table_create;
+ }
+
vr->fib4 = fib4;
vr->fib6 = fib6;
- vr->mr4_table = mr4_table;
+ vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] = mr4_table;
+ vr->mr_table[MLXSW_SP_L3_PROTO_IPV6] = mr6_table;
vr->tb_id = tb_id;
return vr;
-err_mr_table_create:
+err_mr6_table_create:
+ mlxsw_sp_mr_table_destroy(mr4_table);
+err_mr4_table_create:
mlxsw_sp_fib_destroy(mlxsw_sp, fib6);
err_fib6_create:
mlxsw_sp_fib_destroy(mlxsw_sp, fib4);
@@ -830,8 +842,10 @@ static struct mlxsw_sp_vr *mlxsw_sp_vr_create(struct mlxsw_sp *mlxsw_sp,
static void mlxsw_sp_vr_destroy(struct mlxsw_sp *mlxsw_sp,
struct mlxsw_sp_vr *vr)
{
- mlxsw_sp_mr_table_destroy(vr->mr4_table);
- vr->mr4_table = NULL;
+ mlxsw_sp_mr_table_destroy(vr->mr_table[MLXSW_SP_L3_PROTO_IPV6]);
+ vr->mr_table[MLXSW_SP_L3_PROTO_IPV6] = NULL;
+ mlxsw_sp_mr_table_destroy(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4]);
+ vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] = NULL;
mlxsw_sp_fib_destroy(mlxsw_sp, vr->fib6);
vr->fib6 = NULL;
mlxsw_sp_fib_destroy(mlxsw_sp, vr->fib4);
@@ -854,7 +868,8 @@ static void mlxsw_sp_vr_put(struct mlxsw_sp *mlxsw_sp, struct mlxsw_sp_vr *vr)
{
if (!vr->rif_count && list_empty(&vr->fib4->node_list) &&
list_empty(&vr->fib6->node_list) &&
- mlxsw_sp_mr_table_empty(vr->mr4_table))
+ mlxsw_sp_mr_table_empty(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4]) &&
+ mlxsw_sp_mr_table_empty(vr->mr_table[MLXSW_SP_L3_PROTO_IPV6]))
mlxsw_sp_vr_destroy(mlxsw_sp, vr);
}
@@ -5391,7 +5406,7 @@ static int mlxsw_sp_router_fibmr_add(struct mlxsw_sp *mlxsw_sp,
if (IS_ERR(vr))
return PTR_ERR(vr);
- return mlxsw_sp_mr_route4_add(vr->mr4_table,
+ return mlxsw_sp_mr_route4_add(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4],
(struct mfc_cache *) men_info->mfc,
replace);
}
@@ -5408,7 +5423,7 @@ static void mlxsw_sp_router_fibmr_del(struct mlxsw_sp *mlxsw_sp,
if (WARN_ON(!vr))
return;
- mlxsw_sp_mr_route4_del(vr->mr4_table,
+ mlxsw_sp_mr_route4_del(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4],
(struct mfc_cache *) men_info->mfc);
mlxsw_sp_vr_put(mlxsw_sp, vr);
}
@@ -5428,7 +5443,8 @@ mlxsw_sp_router_fibmr_vif_add(struct mlxsw_sp *mlxsw_sp,
return PTR_ERR(vr);
rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, ven_info->dev);
- return mlxsw_sp_mr_vif_add(vr->mr4_table, ven_info->dev,
+ return mlxsw_sp_mr_vif_add(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4],
+ ven_info->dev,
ven_info->vif_index,
ven_info->vif_flags, rif);
}
@@ -5446,7 +5462,8 @@ mlxsw_sp_router_fibmr_vif_del(struct mlxsw_sp *mlxsw_sp,
if (WARN_ON(!vr))
return;
- mlxsw_sp_mr_vif_del(vr->mr4_table, ven_info->vif_index);
+ mlxsw_sp_mr_vif_del(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4],
+ ven_info->vif_index);
mlxsw_sp_vr_put(mlxsw_sp, vr);
}
@@ -5538,7 +5555,7 @@ static void mlxsw_sp_vr_fib_flush(struct mlxsw_sp *mlxsw_sp,
static void mlxsw_sp_router_fib_flush(struct mlxsw_sp *mlxsw_sp)
{
- int i;
+ int i, j;
for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS); i++) {
struct mlxsw_sp_vr *vr = &mlxsw_sp->router->vrs[i];
@@ -5546,7 +5563,8 @@ static void mlxsw_sp_router_fib_flush(struct mlxsw_sp *mlxsw_sp)
if (!mlxsw_sp_vr_is_used(vr))
continue;
- mlxsw_sp_mr_table_flush(vr->mr4_table);
+ for (j = 0; j < MLXSW_SP_L3_PROTO_MAX; j++)
+ mlxsw_sp_mr_table_flush(vr->mr_table[j]);
mlxsw_sp_vr_fib_flush(mlxsw_sp, vr, MLXSW_SP_L3_PROTO_IPV4);
/* If virtual router was only used for IPv4, then it's no
@@ -6041,7 +6059,7 @@ mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
struct mlxsw_sp_rif *rif;
struct mlxsw_sp_vr *vr;
u16 rif_index;
- int err;
+ int i, err;
type = mlxsw_sp_dev_rif_type(mlxsw_sp, params->dev);
ops = mlxsw_sp->router->rif_ops_arr[type];
@@ -6081,9 +6099,11 @@ mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
if (err)
goto err_configure;
- err = mlxsw_sp_mr_rif_add(vr->mr4_table, rif);
- if (err)
- goto err_mr_rif_add;
+ for (i = 0; i < MLXSW_SP_L3_PROTO_MAX; i++) {
+ err = mlxsw_sp_mr_rif_add(vr->mr_table[i], rif);
+ if (err)
+ goto err_mr_rif_add;
+ }
mlxsw_sp_rif_counters_alloc(rif);
mlxsw_sp->router->rifs[rif_index] = rif;
@@ -6091,6 +6111,8 @@ mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
return rif;
err_mr_rif_add:
+ for (i--; i >= 0; i--)
+ mlxsw_sp_mr_rif_del(vr->mr_table[i], rif);
ops->deconfigure(rif);
err_configure:
if (fid)
@@ -6110,13 +6132,15 @@ void mlxsw_sp_rif_destroy(struct mlxsw_sp_rif *rif)
struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
struct mlxsw_sp_fid *fid = rif->fid;
struct mlxsw_sp_vr *vr;
+ int i;
mlxsw_sp_router_rif_gone_sync(mlxsw_sp, rif);
vr = &mlxsw_sp->router->vrs[rif->vr_id];
mlxsw_sp->router->rifs[rif->rif_index] = NULL;
mlxsw_sp_rif_counters_free(rif);
- mlxsw_sp_mr_rif_del(vr->mr4_table, rif);
+ for (i = 0; i < MLXSW_SP_L3_PROTO_MAX; i++)
+ mlxsw_sp_mr_rif_del(vr->mr_table[i], rif);
ops->deconfigure(rif);
if (fid)
/* Loopback RIFs are not associated with a FID. */
@@ -6523,13 +6547,16 @@ int mlxsw_sp_netdevice_router_port_event(struct net_device *dev)
if (rif->mtu != dev->mtu) {
struct mlxsw_sp_vr *vr;
+ int i;
/* The RIF is relevant only to its mr_table instance, as unlike
* unicast routing, in multicast routing a RIF cannot be shared
* between several multicast routing tables.
*/
vr = &mlxsw_sp->router->vrs[rif->vr_id];
- mlxsw_sp_mr_rif_mtu_update(vr->mr4_table, rif, dev->mtu);
+ for (i = 0; i < MLXSW_SP_L3_PROTO_MAX; i++)
+ mlxsw_sp_mr_rif_mtu_update(vr->mr_table[i],
+ rif, dev->mtu);
}
ether_addr_copy(rif->addr, dev->dev_addr);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h
index 1fb82246ce96..a01edcf56797 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h
@@ -41,6 +41,7 @@
enum mlxsw_sp_l3proto {
MLXSW_SP_L3_PROTO_IPV4,
MLXSW_SP_L3_PROTO_IPV6,
+#define MLXSW_SP_L3_PROTO_MAX (MLXSW_SP_L3_PROTO_IPV6 + 1)
};
union mlxsw_sp_l3addr {
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 09/15] mlxsw: spectrum_mr: Pass protocol as part of catchall route params
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Since commit c011ec1bbfd6 ("mlxsw: spectrum: Add the multicast routing
offloading logic") spectrum_mr did not populate the protocol portion of
the catcahall_route_params; mr-tcam logic worked correctly for ipv4
since the enum value for MLXSW_SP_L3_PROTO_IPV4 is '0'.
Explicitly fill the protocol as we'll soon need to differentiate between
ipv4 and ipv6.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
index 51b104ae2eec..a8f4927ff150 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
@@ -848,6 +848,7 @@ struct mlxsw_sp_mr_table *mlxsw_sp_mr_table_create(struct mlxsw_sp *mlxsw_sp,
.prio = MLXSW_SP_MR_ROUTE_PRIO_CATCHALL,
.key = {
.vrid = vr_id,
+ .proto = proto,
},
.value = {
.route_action = MLXSW_SP_MR_ROUTE_ACTION_TRAP,
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 08/15] mlxsw: reg: Add register settings for IPv6 multicast routing
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Add new fields for the rmft register necessary for setting the IPv6
multicast FIB table. Add a matching wrapper function for filling
the register in the IPv6 scenario.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/reg.h | 58 +++++++++++++++++++++++--------
1 file changed, 44 insertions(+), 14 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index 8397bed1f372..6218231e379e 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -6318,30 +6318,34 @@ MLXSW_ITEM32(reg, rmft2, irif_mask, 0x08, 24, 1);
*/
MLXSW_ITEM32(reg, rmft2, irif, 0x08, 0, 16);
-/* reg_rmft2_dip4
- * Destination IPv4 address
+/* reg_rmft2_dip{4,6}
+ * Destination IPv4/6 address
* Access: RW
*/
+MLXSW_ITEM_BUF(reg, rmft2, dip6, 0x10, 16);
MLXSW_ITEM32(reg, rmft2, dip4, 0x1C, 0, 32);
-/* reg_rmft2_dip4_mask
+/* reg_rmft2_dip{4,6}_mask
* A bit that is set directs the TCAM to compare the corresponding bit in key. A
* bit that is clear directs the TCAM to ignore the corresponding bit in key.
* Access: RW
*/
+MLXSW_ITEM_BUF(reg, rmft2, dip6_mask, 0x20, 16);
MLXSW_ITEM32(reg, rmft2, dip4_mask, 0x2C, 0, 32);
-/* reg_rmft2_sip4
- * Source IPv4 address
+/* reg_rmft2_sip{4,6}
+ * Source IPv4/6 address
* Access: RW
*/
+MLXSW_ITEM_BUF(reg, rmft2, sip6, 0x30, 16);
MLXSW_ITEM32(reg, rmft2, sip4, 0x3C, 0, 32);
-/* reg_rmft2_sip4_mask
+/* reg_rmft2_sip{4,6}_mask
* A bit that is set directs the TCAM to compare the corresponding bit in key. A
* bit that is clear directs the TCAM to ignore the corresponding bit in key.
* Access: RW
*/
+MLXSW_ITEM_BUF(reg, rmft2, sip6_mask, 0x40, 16);
MLXSW_ITEM32(reg, rmft2, sip4_mask, 0x4C, 0, 32);
/* reg_rmft2_flexible_action_set
@@ -6359,26 +6363,52 @@ MLXSW_ITEM_BUF(reg, rmft2, flexible_action_set, 0x80,
MLXSW_REG_FLEX_ACTION_SET_LEN);
static inline void
-mlxsw_reg_rmft2_ipv4_pack(char *payload, bool v, u16 offset, u16 virtual_router,
- enum mlxsw_reg_rmft2_irif_mask irif_mask, u16 irif,
- u32 dip4, u32 dip4_mask, u32 sip4, u32 sip4_mask,
- const char *flexible_action_set)
+mlxsw_reg_rmft2_common_pack(char *payload, bool v, u16 offset,
+ u16 virtual_router,
+ enum mlxsw_reg_rmft2_irif_mask irif_mask, u16 irif,
+ const char *flex_action_set)
{
MLXSW_REG_ZERO(rmft2, payload);
mlxsw_reg_rmft2_v_set(payload, v);
- mlxsw_reg_rmft2_type_set(payload, MLXSW_REG_RMFT2_TYPE_IPV4);
mlxsw_reg_rmft2_op_set(payload, MLXSW_REG_RMFT2_OP_READ_WRITE);
mlxsw_reg_rmft2_offset_set(payload, offset);
mlxsw_reg_rmft2_virtual_router_set(payload, virtual_router);
mlxsw_reg_rmft2_irif_mask_set(payload, irif_mask);
mlxsw_reg_rmft2_irif_set(payload, irif);
+ if (flex_action_set)
+ mlxsw_reg_rmft2_flexible_action_set_memcpy_to(payload,
+ flex_action_set);
+}
+
+static inline void
+mlxsw_reg_rmft2_ipv4_pack(char *payload, bool v, u16 offset, u16 virtual_router,
+ enum mlxsw_reg_rmft2_irif_mask irif_mask, u16 irif,
+ u32 dip4, u32 dip4_mask, u32 sip4, u32 sip4_mask,
+ const char *flexible_action_set)
+{
+ mlxsw_reg_rmft2_common_pack(payload, v, offset, virtual_router,
+ irif_mask, irif, flexible_action_set);
+ mlxsw_reg_rmft2_type_set(payload, MLXSW_REG_RMFT2_TYPE_IPV4);
mlxsw_reg_rmft2_dip4_set(payload, dip4);
mlxsw_reg_rmft2_dip4_mask_set(payload, dip4_mask);
mlxsw_reg_rmft2_sip4_set(payload, sip4);
mlxsw_reg_rmft2_sip4_mask_set(payload, sip4_mask);
- if (flexible_action_set)
- mlxsw_reg_rmft2_flexible_action_set_memcpy_to(payload,
- flexible_action_set);
+}
+
+static inline void
+mlxsw_reg_rmft2_ipv6_pack(char *payload, bool v, u16 offset, u16 virtual_router,
+ enum mlxsw_reg_rmft2_irif_mask irif_mask, u16 irif,
+ struct in6_addr dip6, struct in6_addr dip6_mask,
+ struct in6_addr sip6, struct in6_addr sip6_mask,
+ const char *flexible_action_set)
+{
+ mlxsw_reg_rmft2_common_pack(payload, v, offset, virtual_router,
+ irif_mask, irif, flexible_action_set);
+ mlxsw_reg_rmft2_type_set(payload, MLXSW_REG_RMFT2_TYPE_IPV6);
+ mlxsw_reg_rmft2_dip6_memcpy_to(payload, (void *)&dip6);
+ mlxsw_reg_rmft2_dip6_mask_memcpy_to(payload, (void *)&dip6_mask);
+ mlxsw_reg_rmft2_sip6_memcpy_to(payload, (void *)&sip6);
+ mlxsw_reg_rmft2_sip6_mask_memcpy_to(payload, (void *)&sip6_mask);
}
/* MFCR - Management Fan Control Register
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 07/15] mlxsw: reg: Configure RIF to forward IPv6 multicast packets
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Similarly to what was done in commit 4af5964e5888 ("mlxsw: reg:
Configure RIF to forward IPv4 multicast packets by default") we now set
two additional bits to allow IPv6 multicast forwarding.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/reg.h | 16 ++++++++++++++++
1 file changed, 16 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index e002398364c8..8397bed1f372 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -4225,6 +4225,12 @@ MLXSW_ITEM32(reg, ritr, ipv6, 0x00, 28, 1);
*/
MLXSW_ITEM32(reg, ritr, ipv4_mc, 0x00, 27, 1);
+/* reg_ritr_ipv6_mc
+ * IPv6 multicast routing enable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv6_mc, 0x00, 26, 1);
+
enum mlxsw_reg_ritr_if_type {
/* VLAN interface. */
MLXSW_REG_RITR_VLAN_IF,
@@ -4290,6 +4296,14 @@ MLXSW_ITEM32(reg, ritr, ipv6_fe, 0x04, 28, 1);
*/
MLXSW_ITEM32(reg, ritr, ipv4_mc_fe, 0x04, 27, 1);
+/* reg_ritr_ipv6_mc_fe
+ * IPv6 Multicast Forwarding Enable.
+ * When disabled, forwarding is blocked but local traffic (traps and IP to me)
+ * will be enabled.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv6_mc_fe, 0x04, 26, 1);
+
/* reg_ritr_lb_en
* Loop-back filter enable for unicast packets.
* If the flag is set then loop-back filter for unicast packets is
@@ -4513,12 +4527,14 @@ static inline void mlxsw_reg_ritr_pack(char *payload, bool enable,
mlxsw_reg_ritr_ipv4_set(payload, 1);
mlxsw_reg_ritr_ipv6_set(payload, 1);
mlxsw_reg_ritr_ipv4_mc_set(payload, 1);
+ mlxsw_reg_ritr_ipv6_mc_set(payload, 1);
mlxsw_reg_ritr_type_set(payload, type);
mlxsw_reg_ritr_op_set(payload, op);
mlxsw_reg_ritr_rif_set(payload, rif);
mlxsw_reg_ritr_ipv4_fe_set(payload, 1);
mlxsw_reg_ritr_ipv6_fe_set(payload, 1);
mlxsw_reg_ritr_ipv4_mc_fe_set(payload, 1);
+ mlxsw_reg_ritr_ipv6_mc_fe_set(payload, 1);
mlxsw_reg_ritr_lb_en_set(payload, 1);
mlxsw_reg_ritr_virtual_router_set(payload, vr_id);
mlxsw_reg_ritr_mtu_set(payload, mtu);
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 06/15] ip6mr: Add refcounting to mfc
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Since ipmr and ip6mr are using the same mr_mfc struct at their core, we
can now refactor the ipmr_cache_{hold,put} logic and apply refcounting
to both ipmr and ip6mr.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c | 6 +++---
drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c | 6 +++---
include/linux/mroute.h | 19 -------------------
include/linux/mroute_base.h | 13 +++++++++++++
net/ipv4/ipmr.c | 8 ++++----
net/ipv6/ip6mr.c | 6 ++++--
6 files changed, 27 insertions(+), 31 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
index 978a3c70653a..51b104ae2eec 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
@@ -360,7 +360,7 @@ mlxsw_sp_mr_route4_create(struct mlxsw_sp_mr_table *mr_table,
/* Find min_mtu and link iVIF and eVIFs */
mr_route->min_mtu = ETH_MAX_MTU;
- ipmr_cache_hold(mfc);
+ mr_cache_hold(&mfc->_c);
mr_route->mfc4 = mfc;
mr_route->mr_table = mr_table;
for (i = 0; i < MAXVIFS; i++) {
@@ -380,7 +380,7 @@ mlxsw_sp_mr_route4_create(struct mlxsw_sp_mr_table *mr_table,
mr_route->route_action = mlxsw_sp_mr_route_action(mr_route);
return mr_route;
err:
- ipmr_cache_put(mfc);
+ mr_cache_put(&mfc->_c);
list_for_each_entry_safe(rve, tmp, &mr_route->evif_list, route_node)
mlxsw_sp_mr_route_evif_unlink(rve);
kfree(mr_route);
@@ -393,7 +393,7 @@ static void mlxsw_sp_mr_route4_destroy(struct mlxsw_sp_mr_table *mr_table,
struct mlxsw_sp_mr_route_vif_entry *rve, *tmp;
mlxsw_sp_mr_route_ivif_unlink(mr_route);
- ipmr_cache_put(mr_route->mfc4);
+ mr_cache_put((struct mr_mfc *)mr_route->mfc4);
list_for_each_entry_safe(rve, tmp, &mr_route->evif_list, route_node)
mlxsw_sp_mr_route_evif_unlink(rve);
kfree(mr_route);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 8d067de924cf..be241c708bb9 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -5685,11 +5685,11 @@ static void mlxsw_sp_router_fibmr_event_work(struct work_struct *work)
replace);
if (err)
mlxsw_sp_router_fib_abort(mlxsw_sp);
- ipmr_cache_put((struct mfc_cache *) fib_work->men_info.mfc);
+ mr_cache_put(fib_work->men_info.mfc);
break;
case FIB_EVENT_ENTRY_DEL:
mlxsw_sp_router_fibmr_del(mlxsw_sp, &fib_work->men_info);
- ipmr_cache_put((struct mfc_cache *) fib_work->men_info.mfc);
+ mr_cache_put(fib_work->men_info.mfc);
break;
case FIB_EVENT_VIF_ADD:
err = mlxsw_sp_router_fibmr_vif_add(mlxsw_sp,
@@ -5769,7 +5769,7 @@ mlxsw_sp_router_fibmr_event(struct mlxsw_sp_fib_event_work *fib_work,
case FIB_EVENT_ENTRY_ADD: /* fall through */
case FIB_EVENT_ENTRY_DEL:
memcpy(&fib_work->men_info, info, sizeof(fib_work->men_info));
- ipmr_cache_hold((struct mfc_cache *) fib_work->men_info.mfc);
+ mr_cache_hold(fib_work->men_info.mfc);
break;
case FIB_EVENT_VIF_ADD: /* fall through */
case FIB_EVENT_VIF_DEL:
diff --git a/include/linux/mroute.h b/include/linux/mroute.h
index c855d80b51f7..9a36fad9e068 100644
--- a/include/linux/mroute.h
+++ b/include/linux/mroute.h
@@ -84,23 +84,4 @@ struct rtmsg;
int ipmr_get_route(struct net *net, struct sk_buff *skb,
__be32 saddr, __be32 daddr,
struct rtmsg *rtm, u32 portid);
-
-#ifdef CONFIG_IP_MROUTE
-void ipmr_cache_free(struct mfc_cache *mfc_cache);
-#else
-static inline void ipmr_cache_free(struct mfc_cache *mfc_cache)
-{
-}
-#endif
-
-static inline void ipmr_cache_put(struct mfc_cache *c)
-{
- if (refcount_dec_and_test(&c->_c.mfc_un.res.refcount))
- ipmr_cache_free(c);
-}
-static inline void ipmr_cache_hold(struct mfc_cache *c)
-{
- refcount_inc(&c->_c.mfc_un.res.refcount);
-}
-
#endif
diff --git a/include/linux/mroute_base.h b/include/linux/mroute_base.h
index 289eb5aa7b5d..d617fe45543e 100644
--- a/include/linux/mroute_base.h
+++ b/include/linux/mroute_base.h
@@ -125,6 +125,7 @@ enum {
* @refcount: reference count for this entry
* @list: global entry list
* @rcu: used for entry destruction
+ * @free: Operation used for freeing an entry under RCU
*/
struct mr_mfc {
struct rhlist_head mnode;
@@ -150,8 +151,20 @@ struct mr_mfc {
} mfc_un;
struct list_head list;
struct rcu_head rcu;
+ void (*free)(struct rcu_head *head);
};
+static inline void mr_cache_put(struct mr_mfc *c)
+{
+ if (refcount_dec_and_test(&c->mfc_un.res.refcount))
+ call_rcu(&c->rcu, c->free);
+}
+
+static inline void mr_cache_hold(struct mr_mfc *c)
+{
+ refcount_inc(&c->mfc_un.res.refcount);
+}
+
struct mfc_entry_notifier_info {
struct fib_notifier_info info;
struct mr_mfc *mfc;
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index 24c340021aba..e79211a8537c 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -732,11 +732,10 @@ static void ipmr_cache_free_rcu(struct rcu_head *head)
kmem_cache_free(mrt_cachep, (struct mfc_cache *)c);
}
-void ipmr_cache_free(struct mfc_cache *c)
+static void ipmr_cache_free(struct mfc_cache *c)
{
call_rcu(&c->_c.rcu, ipmr_cache_free_rcu);
}
-EXPORT_SYMBOL(ipmr_cache_free);
/* Destroy an unresolved cache entry, killing queued skbs
* and reporting error to netlink readers.
@@ -987,6 +986,7 @@ static struct mfc_cache *ipmr_cache_alloc(void)
if (c) {
c->_c.mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1;
c->_c.mfc_un.res.minvif = MAXVIFS;
+ c->_c.free = ipmr_cache_free_rcu;
refcount_set(&c->_c.mfc_un.res.refcount, 1);
}
return c;
@@ -1206,7 +1206,7 @@ static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc, int parent)
list_del_rcu(&c->_c.list);
call_ipmr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_DEL, c, mrt->id);
mroute_netlink_event(mrt, c, RTM_DELROUTE);
- ipmr_cache_put(c);
+ mr_cache_put(&c->_c);
return 0;
}
@@ -1318,7 +1318,7 @@ static void mroute_clean_tables(struct mr_table *mrt, bool all)
call_ipmr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_DEL, cache,
mrt->id);
mroute_netlink_event(mrt, cache, RTM_DELROUTE);
- ipmr_cache_put(cache);
+ mr_cache_put(c);
}
if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index a187c523a95f..1c8fa29d155a 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -989,6 +989,8 @@ static struct mfc6_cache *ip6mr_cache_alloc(void)
return NULL;
c->_c.mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1;
c->_c.mfc_un.res.minvif = MAXMIFS;
+ c->_c.free = ip6mr_cache_free_rcu;
+ refcount_set(&c->_c.mfc_un.res.refcount, 1);
return c;
}
@@ -1227,7 +1229,7 @@ static int ip6mr_mfc_delete(struct mr_table *mrt, struct mf6cctl *mfc,
call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
FIB_EVENT_ENTRY_DEL, c, mrt->id);
mr6_netlink_event(mrt, c, RTM_DELROUTE);
- ip6mr_cache_free(c);
+ mr_cache_put(&c->_c);
return 0;
}
@@ -1516,7 +1518,7 @@ static void mroute_clean_tables(struct mr_table *mrt, bool all)
rhltable_remove(&mrt->mfc_hash, &c->mnode, ip6mr_rht_params);
list_del_rcu(&c->list);
mr6_netlink_event(mrt, (struct mfc6_cache *)c, RTM_DELROUTE);
- ip6mr_cache_free((struct mfc6_cache *)c);
+ mr_cache_put(c);
}
if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 05/15] ip6mr: Add API for default_rule fib
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Add the ability to discern whether a given FIB rule notification relates
to the default rule inserted when registering ip6mr or a different one.
Would later be used by drivers wishing to offload ipv6 multicast routes
but unable to offload rules other than the default one.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
include/linux/mroute6.h | 10 ++++++++++
net/ipv6/ip6mr.c | 7 +++++++
2 files changed, 17 insertions(+)
diff --git a/include/linux/mroute6.h b/include/linux/mroute6.h
index 1ac38e6819f5..c4a45859f586 100644
--- a/include/linux/mroute6.h
+++ b/include/linux/mroute6.h
@@ -8,6 +8,7 @@
#include <net/net_namespace.h>
#include <uapi/linux/mroute6.h>
#include <linux/mroute_base.h>
+#include <net/fib_rules.h>
#ifdef CONFIG_IPV6_MROUTE
static inline int ip6_mroute_opt(int opt)
@@ -63,6 +64,15 @@ static inline void ip6_mr_cleanup(void)
}
#endif
+#ifdef CONFIG_IPV6_MROUTE_MULTIPLE_TABLES
+bool ip6mr_rule_default(const struct fib_rule *rule);
+#else
+static inline bool ip6mr_rule_default(const struct fib_rule *rule)
+{
+ return true;
+}
+#endif
+
#define VIFF_STATIC 0x8000
struct mfc6_cache_cmp_arg {
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 0be2f333e168..a187c523a95f 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -268,6 +268,13 @@ static unsigned int ip6mr_rules_seq_read(struct net *net)
{
return fib_rules_seq_read(net, RTNL_FAMILY_IP6MR);
}
+
+bool ip6mr_rule_default(const struct fib_rule *rule)
+{
+ return fib_rule_matchall(rule) && rule->action == FR_ACT_TO_TBL &&
+ rule->table == RT6_TABLE_DFLT && !rule->l3mdev;
+}
+EXPORT_SYMBOL(ip6mr_rule_default);
#else
#define ip6mr_for_each_table(mrt, net) \
for (mrt = net->ipv6.mrt6; mrt; mrt = NULL)
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 04/15] ip6mr: Support fib notifications
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
In similar fashion to ipmr, support fib notifications for ip6mr mfc and
vif related events. This would later allow drivers to react to said
notifications and offload the IPv6 mroutes.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
include/net/netns/ipv6.h | 2 +
net/ipv6/ip6mr.c | 112 +++++++++++++++++++++++++++++++++++++++++++----
2 files changed, 106 insertions(+), 8 deletions(-)
diff --git a/include/net/netns/ipv6.h b/include/net/netns/ipv6.h
index 5b51110435fc..c29f09cfc9d7 100644
--- a/include/net/netns/ipv6.h
+++ b/include/net/netns/ipv6.h
@@ -96,6 +96,8 @@ struct netns_ipv6 {
atomic_t fib6_sernum;
struct seg6_pernet_data *seg6_data;
struct fib_notifier_ops *notifier_ops;
+ struct fib_notifier_ops *ip6mr_notifier_ops;
+ unsigned int ipmr_seq; /* protected by rtnl_mutex */
struct {
struct hlist_head head;
spinlock_t lock;
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 7345bd6c4b7d..0be2f333e168 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -258,6 +258,16 @@ static void __net_exit ip6mr_rules_exit(struct net *net)
fib_rules_unregister(net->ipv6.mr6_rules_ops);
rtnl_unlock();
}
+
+static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb)
+{
+ return fib_rules_dump(net, nb, RTNL_FAMILY_IP6MR);
+}
+
+static unsigned int ip6mr_rules_seq_read(struct net *net)
+{
+ return fib_rules_seq_read(net, RTNL_FAMILY_IP6MR);
+}
#else
#define ip6mr_for_each_table(mrt, net) \
for (mrt = net->ipv6.mrt6; mrt; mrt = NULL)
@@ -295,6 +305,16 @@ static void __net_exit ip6mr_rules_exit(struct net *net)
net->ipv6.mrt6 = NULL;
rtnl_unlock();
}
+
+static int ip6mr_rules_dump(struct net *net, struct notifier_block *nb)
+{
+ return 0;
+}
+
+static unsigned int ip6mr_rules_seq_read(struct net *net)
+{
+ return 0;
+}
#endif
static int ip6mr_hash_cmp(struct rhashtable_compare_arg *arg,
@@ -653,10 +673,25 @@ static struct net_device *ip6mr_reg_vif(struct net *net, struct mr_table *mrt)
}
#endif
-/*
- * Delete a VIF entry
- */
+static int call_ip6mr_vif_entry_notifiers(struct net *net,
+ enum fib_event_type event_type,
+ struct vif_device *vif,
+ mifi_t vif_index, u32 tb_id)
+{
+ return mr_call_vif_notifiers(net, RTNL_FAMILY_IP6MR, event_type,
+ vif, vif_index, tb_id,
+ &net->ipv6.ipmr_seq);
+}
+static int call_ip6mr_mfc_entry_notifiers(struct net *net,
+ enum fib_event_type event_type,
+ struct mfc6_cache *mfc, u32 tb_id)
+{
+ return mr_call_mfc_notifiers(net, RTNL_FAMILY_IP6MR, event_type,
+ &mfc->_c, tb_id, &net->ipv6.ipmr_seq);
+}
+
+/* Delete a VIF entry */
static int mif6_delete(struct mr_table *mrt, int vifi, int notify,
struct list_head *head)
{
@@ -669,6 +704,11 @@ static int mif6_delete(struct mr_table *mrt, int vifi, int notify,
v = &mrt->vif_table[vifi];
+ if (VIF_EXISTS(mrt, vifi))
+ call_ip6mr_vif_entry_notifiers(read_pnet(&mrt->net),
+ FIB_EVENT_VIF_DEL, v, vifi,
+ mrt->id);
+
write_lock_bh(&mrt_lock);
dev = v->dev;
v->dev = NULL;
@@ -887,6 +927,8 @@ static int mif6_add(struct net *net, struct mr_table *mrt,
if (vifi + 1 > mrt->maxvif)
mrt->maxvif = vifi + 1;
write_unlock_bh(&mrt_lock);
+ call_ip6mr_vif_entry_notifiers(net, FIB_EVENT_VIF_ADD,
+ v, vifi, mrt->id);
return 0;
}
@@ -1175,6 +1217,8 @@ static int ip6mr_mfc_delete(struct mr_table *mrt, struct mf6cctl *mfc,
rhltable_remove(&mrt->mfc_hash, &c->_c.mnode, ip6mr_rht_params);
list_del_rcu(&c->_c.list);
+ call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
+ FIB_EVENT_ENTRY_DEL, c, mrt->id);
mr6_netlink_event(mrt, c, RTM_DELROUTE);
ip6mr_cache_free(c);
return 0;
@@ -1203,21 +1247,63 @@ static int ip6mr_device_event(struct notifier_block *this,
return NOTIFY_DONE;
}
+static unsigned int ip6mr_seq_read(struct net *net)
+{
+ ASSERT_RTNL();
+
+ return net->ipv6.ipmr_seq + ip6mr_rules_seq_read(net);
+}
+
+static int ip6mr_dump(struct net *net, struct notifier_block *nb)
+{
+ return mr_dump(net, nb, RTNL_FAMILY_IP6MR, ip6mr_rules_dump,
+ ip6mr_mr_table_iter, &mrt_lock);
+}
+
static struct notifier_block ip6_mr_notifier = {
.notifier_call = ip6mr_device_event
};
-/*
- * Setup for IP multicast routing
- */
+static const struct fib_notifier_ops ip6mr_notifier_ops_template = {
+ .family = RTNL_FAMILY_IP6MR,
+ .fib_seq_read = ip6mr_seq_read,
+ .fib_dump = ip6mr_dump,
+ .owner = THIS_MODULE,
+};
+
+static int __net_init ip6mr_notifier_init(struct net *net)
+{
+ struct fib_notifier_ops *ops;
+
+ net->ipv6.ipmr_seq = 0;
+ ops = fib_notifier_ops_register(&ip6mr_notifier_ops_template, net);
+ if (IS_ERR(ops))
+ return PTR_ERR(ops);
+
+ net->ipv6.ip6mr_notifier_ops = ops;
+
+ return 0;
+}
+
+static void __net_exit ip6mr_notifier_exit(struct net *net)
+{
+ fib_notifier_ops_unregister(net->ipv6.ip6mr_notifier_ops);
+ net->ipv6.ip6mr_notifier_ops = NULL;
+}
+
+/* Setup for IP multicast routing */
static int __net_init ip6mr_net_init(struct net *net)
{
int err;
+ err = ip6mr_notifier_init(net);
+ if (err)
+ return err;
+
err = ip6mr_rules_init(net);
if (err < 0)
- goto fail;
+ goto ip6mr_rules_fail;
#ifdef CONFIG_PROC_FS
err = -ENOMEM;
@@ -1235,7 +1321,8 @@ static int __net_init ip6mr_net_init(struct net *net)
proc_vif_fail:
ip6mr_rules_exit(net);
#endif
-fail:
+ip6mr_rules_fail:
+ ip6mr_notifier_exit(net);
return err;
}
@@ -1246,6 +1333,7 @@ static void __net_exit ip6mr_net_exit(struct net *net)
remove_proc_entry("ip6_mr_vif", net->proc_net);
#endif
ip6mr_rules_exit(net);
+ ip6mr_notifier_exit(net);
}
static struct pernet_operations ip6mr_net_ops = {
@@ -1337,6 +1425,8 @@ static int ip6mr_mfc_add(struct net *net, struct mr_table *mrt,
if (!mrtsock)
c->_c.mfc_flags |= MFC_STATIC;
write_unlock_bh(&mrt_lock);
+ call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_REPLACE,
+ c, mrt->id);
mr6_netlink_event(mrt, c, RTM_NEWROUTE);
return 0;
}
@@ -1388,6 +1478,8 @@ static int ip6mr_mfc_add(struct net *net, struct mr_table *mrt,
ip6mr_cache_resolve(net, mrt, uc, c);
ip6mr_cache_free(uc);
}
+ call_ip6mr_mfc_entry_notifiers(net, FIB_EVENT_ENTRY_ADD,
+ c, mrt->id);
mr6_netlink_event(mrt, c, RTM_NEWROUTE);
return 0;
}
@@ -1424,6 +1516,10 @@ static void mroute_clean_tables(struct mr_table *mrt, bool all)
spin_lock_bh(&mfc_unres_lock);
list_for_each_entry_safe(c, tmp, &mrt->mfc_unres_queue, list) {
list_del(&c->list);
+ call_ip6mr_mfc_entry_notifiers(read_pnet(&mrt->net),
+ FIB_EVENT_ENTRY_DEL,
+ (struct mfc6_cache *)c,
+ mrt->id);
mr6_netlink_event(mrt, (struct mfc6_cache *)c,
RTM_DELROUTE);
ip6mr_destroy_unres(mrt, (struct mfc6_cache *)c);
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 03/15] ipmr: Make ipmr_dump() common
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Since all the primitive elements used for the notification done by ipmr
are now common [mr_table, mr_mfc, vif_device] we can refactor the logic
for dumping them to a common file.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
include/linux/mroute_base.h | 18 +++++++++++++++
net/ipv4/ipmr.c | 53 ++-------------------------------------------
net/ipv4/ipmr_base.c | 42 +++++++++++++++++++++++++++++++++++
3 files changed, 62 insertions(+), 51 deletions(-)
diff --git a/include/linux/mroute_base.h b/include/linux/mroute_base.h
index 2c594686c05e..289eb5aa7b5d 100644
--- a/include/linux/mroute_base.h
+++ b/include/linux/mroute_base.h
@@ -277,6 +277,13 @@ int mr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb,
u32 portid, u32 seq, struct mr_mfc *c,
int cmd, int flags),
spinlock_t *lock);
+
+int mr_dump(struct net *net, struct notifier_block *nb, unsigned short family,
+ int (*rules_dump)(struct net *net,
+ struct notifier_block *nb),
+ struct mr_table *(*mr_iter)(struct net *net,
+ struct mr_table *mrt),
+ rwlock_t *mrt_lock);
#else
static inline void vif_device_init(struct vif_device *v,
struct net_device *dev,
@@ -333,6 +340,17 @@ mr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb,
{
return -EINVAL;
}
+
+static inline int mr_dump(struct net *net, struct notifier_block *nb,
+ unsigned short family,
+ int (*rules_dump)(struct net *net,
+ struct notifier_block *nb),
+ struct mr_table *(*mr_iter)(struct net *net,
+ struct mr_table *mrt),
+ rwlock_t *mrt_lock)
+{
+ return -EINVAL;
+}
#endif
static inline void *mr_mfc_find(struct mr_table *mrt, void *hasharg)
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index 470956d2d8ad..24c340021aba 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -644,16 +644,6 @@ static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt)
}
#endif
-static int call_ipmr_vif_entry_notifier(struct notifier_block *nb,
- struct net *net,
- enum fib_event_type event_type,
- struct vif_device *vif,
- vifi_t vif_index, u32 tb_id)
-{
- return mr_call_vif_notifier(nb, net, RTNL_FAMILY_IPMR, event_type,
- vif, vif_index, tb_id);
-}
-
static int call_ipmr_vif_entry_notifiers(struct net *net,
enum fib_event_type event_type,
struct vif_device *vif,
@@ -664,15 +654,6 @@ static int call_ipmr_vif_entry_notifiers(struct net *net,
&net->ipv4.ipmr_seq);
}
-static int call_ipmr_mfc_entry_notifier(struct notifier_block *nb,
- struct net *net,
- enum fib_event_type event_type,
- struct mfc_cache *mfc, u32 tb_id)
-{
- return mr_call_mfc_notifier(nb, net, RTNL_FAMILY_IPMR,
- event_type, &mfc->_c, tb_id);
-}
-
static int call_ipmr_mfc_entry_notifiers(struct net *net,
enum fib_event_type event_type,
struct mfc_cache *mfc, u32 tb_id)
@@ -2950,38 +2931,8 @@ static unsigned int ipmr_seq_read(struct net *net)
static int ipmr_dump(struct net *net, struct notifier_block *nb)
{
- struct mr_table *mrt;
- int err;
-
- err = ipmr_rules_dump(net, nb);
- if (err)
- return err;
-
- ipmr_for_each_table(mrt, net) {
- struct vif_device *v = &mrt->vif_table[0];
- struct mr_mfc *mfc;
- int vifi;
-
- /* Notifiy on table VIF entries */
- read_lock(&mrt_lock);
- for (vifi = 0; vifi < mrt->maxvif; vifi++, v++) {
- if (!v->dev)
- continue;
-
- call_ipmr_vif_entry_notifier(nb, net, FIB_EVENT_VIF_ADD,
- v, vifi, mrt->id);
- }
- read_unlock(&mrt_lock);
-
- /* Notify on table MFC entries */
- list_for_each_entry_rcu(mfc, &mrt->mfc_cache_list, list)
- call_ipmr_mfc_entry_notifier(nb, net,
- FIB_EVENT_ENTRY_ADD,
- (struct mfc_cache *)mfc,
- mrt->id);
- }
-
- return 0;
+ return mr_dump(net, nb, RTNL_FAMILY_IPMR, ipmr_rules_dump,
+ ipmr_mr_table_iter, &mrt_lock);
}
static const struct fib_notifier_ops ipmr_notifier_ops_template = {
diff --git a/net/ipv4/ipmr_base.c b/net/ipv4/ipmr_base.c
index 8ba55bfda817..4fe97723b53f 100644
--- a/net/ipv4/ipmr_base.c
+++ b/net/ipv4/ipmr_base.c
@@ -321,3 +321,45 @@ int mr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb,
return skb->len;
}
EXPORT_SYMBOL(mr_rtm_dumproute);
+
+int mr_dump(struct net *net, struct notifier_block *nb, unsigned short family,
+ int (*rules_dump)(struct net *net,
+ struct notifier_block *nb),
+ struct mr_table *(*mr_iter)(struct net *net,
+ struct mr_table *mrt),
+ rwlock_t *mrt_lock)
+{
+ struct mr_table *mrt;
+ int err;
+
+ err = rules_dump(net, nb);
+ if (err)
+ return err;
+
+ for (mrt = mr_iter(net, NULL); mrt; mrt = mr_iter(net, mrt)) {
+ struct vif_device *v = &mrt->vif_table[0];
+ struct mr_mfc *mfc;
+ int vifi;
+
+ /* Notifiy on table VIF entries */
+ read_lock(mrt_lock);
+ for (vifi = 0; vifi < mrt->maxvif; vifi++, v++) {
+ if (!v->dev)
+ continue;
+
+ mr_call_vif_notifier(nb, net, family,
+ FIB_EVENT_VIF_ADD,
+ v, vifi, mrt->id);
+ }
+ read_unlock(mrt_lock);
+
+ /* Notify on table MFC entries */
+ list_for_each_entry_rcu(mfc, &mrt->mfc_cache_list, list)
+ mr_call_mfc_notifier(nb, net, family,
+ FIB_EVENT_ENTRY_ADD,
+ mfc, mrt->id);
+ }
+
+ return 0;
+}
+EXPORT_SYMBOL(mr_dump);
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 02/15] ipmr: Make MFC fib notifiers common
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
Like vif notifications, move the notifier struct for MFC as well as its
helpers into a common file; Currently they're only used by ipmr.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
.../net/ethernet/mellanox/mlxsw/spectrum_router.c | 13 ++++---
include/linux/mroute.h | 6 ---
include/linux/mroute_base.h | 44 ++++++++++++++++++++++
net/ipv4/ipmr.c | 26 ++-----------
4 files changed, 56 insertions(+), 33 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 921bd1075edf..8d067de924cf 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -5391,7 +5391,9 @@ static int mlxsw_sp_router_fibmr_add(struct mlxsw_sp *mlxsw_sp,
if (IS_ERR(vr))
return PTR_ERR(vr);
- return mlxsw_sp_mr_route4_add(vr->mr4_table, men_info->mfc, replace);
+ return mlxsw_sp_mr_route4_add(vr->mr4_table,
+ (struct mfc_cache *) men_info->mfc,
+ replace);
}
static void mlxsw_sp_router_fibmr_del(struct mlxsw_sp *mlxsw_sp,
@@ -5406,7 +5408,8 @@ static void mlxsw_sp_router_fibmr_del(struct mlxsw_sp *mlxsw_sp,
if (WARN_ON(!vr))
return;
- mlxsw_sp_mr_route4_del(vr->mr4_table, men_info->mfc);
+ mlxsw_sp_mr_route4_del(vr->mr4_table,
+ (struct mfc_cache *) men_info->mfc);
mlxsw_sp_vr_put(mlxsw_sp, vr);
}
@@ -5682,11 +5685,11 @@ static void mlxsw_sp_router_fibmr_event_work(struct work_struct *work)
replace);
if (err)
mlxsw_sp_router_fib_abort(mlxsw_sp);
- ipmr_cache_put(fib_work->men_info.mfc);
+ ipmr_cache_put((struct mfc_cache *) fib_work->men_info.mfc);
break;
case FIB_EVENT_ENTRY_DEL:
mlxsw_sp_router_fibmr_del(mlxsw_sp, &fib_work->men_info);
- ipmr_cache_put(fib_work->men_info.mfc);
+ ipmr_cache_put((struct mfc_cache *) fib_work->men_info.mfc);
break;
case FIB_EVENT_VIF_ADD:
err = mlxsw_sp_router_fibmr_vif_add(mlxsw_sp,
@@ -5766,7 +5769,7 @@ mlxsw_sp_router_fibmr_event(struct mlxsw_sp_fib_event_work *fib_work,
case FIB_EVENT_ENTRY_ADD: /* fall through */
case FIB_EVENT_ENTRY_DEL:
memcpy(&fib_work->men_info, info, sizeof(fib_work->men_info));
- ipmr_cache_hold(fib_work->men_info.mfc);
+ ipmr_cache_hold((struct mfc_cache *) fib_work->men_info.mfc);
break;
case FIB_EVENT_VIF_ADD: /* fall through */
case FIB_EVENT_VIF_DEL:
diff --git a/include/linux/mroute.h b/include/linux/mroute.h
index 3f70a04a5879..c855d80b51f7 100644
--- a/include/linux/mroute.h
+++ b/include/linux/mroute.h
@@ -80,12 +80,6 @@ struct mfc_cache {
};
};
-struct mfc_entry_notifier_info {
- struct fib_notifier_info info;
- struct mfc_cache *mfc;
- u32 tb_id;
-};
-
struct rtmsg;
int ipmr_get_route(struct net *net, struct sk_buff *skb,
__be32 saddr, __be32 daddr,
diff --git a/include/linux/mroute_base.h b/include/linux/mroute_base.h
index 23326f5402f3..2c594686c05e 100644
--- a/include/linux/mroute_base.h
+++ b/include/linux/mroute_base.h
@@ -152,6 +152,50 @@ struct mr_mfc {
struct rcu_head rcu;
};
+struct mfc_entry_notifier_info {
+ struct fib_notifier_info info;
+ struct mr_mfc *mfc;
+ u32 tb_id;
+};
+
+static inline int mr_call_mfc_notifier(struct notifier_block *nb,
+ struct net *net,
+ unsigned short family,
+ enum fib_event_type event_type,
+ struct mr_mfc *mfc, u32 tb_id)
+{
+ struct mfc_entry_notifier_info info = {
+ .info = {
+ .family = family,
+ .net = net,
+ },
+ .mfc = mfc,
+ .tb_id = tb_id
+ };
+
+ return call_fib_notifier(nb, net, event_type, &info.info);
+}
+
+static inline int mr_call_mfc_notifiers(struct net *net,
+ unsigned short family,
+ enum fib_event_type event_type,
+ struct mr_mfc *mfc, u32 tb_id,
+ unsigned int *ipmr_seq)
+{
+ struct mfc_entry_notifier_info info = {
+ .info = {
+ .family = family,
+ .net = net,
+ },
+ .mfc = mfc,
+ .tb_id = tb_id
+ };
+
+ ASSERT_RTNL();
+ (*ipmr_seq)++;
+ return call_fib_notifiers(net, event_type, &info.info);
+}
+
struct mr_table;
/**
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index bb1a0655f8e4..470956d2d8ad 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -669,34 +669,16 @@ static int call_ipmr_mfc_entry_notifier(struct notifier_block *nb,
enum fib_event_type event_type,
struct mfc_cache *mfc, u32 tb_id)
{
- struct mfc_entry_notifier_info info = {
- .info = {
- .family = RTNL_FAMILY_IPMR,
- .net = net,
- },
- .mfc = mfc,
- .tb_id = tb_id
- };
-
- return call_fib_notifier(nb, net, event_type, &info.info);
+ return mr_call_mfc_notifier(nb, net, RTNL_FAMILY_IPMR,
+ event_type, &mfc->_c, tb_id);
}
static int call_ipmr_mfc_entry_notifiers(struct net *net,
enum fib_event_type event_type,
struct mfc_cache *mfc, u32 tb_id)
{
- struct mfc_entry_notifier_info info = {
- .info = {
- .family = RTNL_FAMILY_IPMR,
- .net = net,
- },
- .mfc = mfc,
- .tb_id = tb_id
- };
-
- ASSERT_RTNL();
- net->ipv4.ipmr_seq++;
- return call_fib_notifiers(net, event_type, &info.info);
+ return mr_call_mfc_notifiers(net, RTNL_FAMILY_IPMR, event_type,
+ &mfc->_c, tb_id, &net->ipv4.ipmr_seq);
}
/**
--
2.14.3
^ permalink raw reply related
* [PATCH net-next 00/15] mlxsw: Offload IPv6 multicast routes
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
Yuval says:
The series is intended to allow offloading IPv6 multicast routes
and is split into two parts:
- First half of the patches continue extending ip6mr [& refactor ipmr]
with missing bits necessary for the offloading - fib-notifications,
mfc refcounting and default rule identification.
- Second half of the patches extend functionality inside mlxsw,
beginning with extending lower-parts to support IPv6 mroutes
to host and later extending the router/mr internal APIs within
the driver to accommodate support in ipv6 configurations.
Lastly it adds support in the RTNL_FAMILY_IP6MR notifications,
allowing driver to react and offload related routes.
Yuval Mintz (15):
ipmr: Make vif fib notifiers common
ipmr: Make MFC fib notifiers common
ipmr: Make ipmr_dump() common
ip6mr: Support fib notifications
ip6mr: Add API for default_rule fib
ip6mr: Add refcounting to mfc
mlxsw: reg: Configure RIF to forward IPv6 multicast packets
mlxsw: reg: Add register settings for IPv6 multicast routing
mlxsw: spectrum_mr: Pass protocol as part of catchall route params
mlxsw: spectrum_router: Support IPv6 multicast to host CPU
mlxsw: spectrum_mr: Convert into using mr_mfc
mlxsw: spectrum_router: Make IPMR-related APIs family agnostic
mlxsw: spectrum_mr: Add ipv6 specific operations
mlxsw: spectrum_router: Process IP6MR fib notification
mlxsw: spectrum: Add multicast router trap for PIMv6
drivers/net/ethernet/mellanox/mlxsw/reg.h | 74 +++++-
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 1 +
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c | 293 +++++++++++++--------
drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h | 9 +-
.../net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c | 105 +++++---
.../net/ethernet/mellanox/mlxsw/spectrum_router.c | 98 +++++--
.../net/ethernet/mellanox/mlxsw/spectrum_router.h | 1 +
drivers/net/ethernet/mellanox/mlxsw/trap.h | 1 +
include/linux/mroute.h | 33 ---
include/linux/mroute6.h | 10 +
include/linux/mroute_base.h | 128 +++++++++
include/net/netns/ipv6.h | 2 +
net/ipv4/ipmr.c | 110 +-------
net/ipv4/ipmr_base.c | 42 +++
net/ipv6/ip6mr.c | 125 ++++++++-
15 files changed, 701 insertions(+), 331 deletions(-)
--
2.14.3
^ permalink raw reply
* [PATCH net-next 01/15] ipmr: Make vif fib notifiers common
From: Ido Schimmel @ 2018-03-26 12:01 UTC (permalink / raw)
To: netdev; +Cc: davem, yuvalm, jiri, nikolay, Ido Schimmel
In-Reply-To: <20180326120145.11752-1-idosch@mellanox.com>
From: Yuval Mintz <yuvalm@mellanox.com>
The fib-notifiers are tightly coupled with the vif_device which is
already common. Move the notifier struct definition and helpers to the
common file; Currently they're only used by ipmr.
Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
include/linux/mroute.h | 8 -------
include/linux/mroute_base.h | 53 +++++++++++++++++++++++++++++++++++++++++++++
net/ipv4/ipmr.c | 31 +++++---------------------
3 files changed, 58 insertions(+), 34 deletions(-)
diff --git a/include/linux/mroute.h b/include/linux/mroute.h
index 7ed82e4f11b3..3f70a04a5879 100644
--- a/include/linux/mroute.h
+++ b/include/linux/mroute.h
@@ -55,14 +55,6 @@ static inline bool ipmr_rule_default(const struct fib_rule *rule)
}
#endif
-struct vif_entry_notifier_info {
- struct fib_notifier_info info;
- struct net_device *dev;
- vifi_t vif_index;
- unsigned short vif_flags;
- u32 tb_id;
-};
-
#define VIFF_STATIC 0x8000
struct mfc_cache_cmp_arg {
diff --git a/include/linux/mroute_base.h b/include/linux/mroute_base.h
index c2560cb50f1d..23326f5402f3 100644
--- a/include/linux/mroute_base.h
+++ b/include/linux/mroute_base.h
@@ -6,6 +6,7 @@
#include <linux/spinlock.h>
#include <net/net_namespace.h>
#include <net/sock.h>
+#include <net/fib_notifier.h>
/**
* struct vif_device - interface representor for multicast routing
@@ -36,6 +37,58 @@ struct vif_device {
__be32 local, remote;
};
+struct vif_entry_notifier_info {
+ struct fib_notifier_info info;
+ struct net_device *dev;
+ unsigned short vif_index;
+ unsigned short vif_flags;
+ u32 tb_id;
+};
+
+static inline int mr_call_vif_notifier(struct notifier_block *nb,
+ struct net *net,
+ unsigned short family,
+ enum fib_event_type event_type,
+ struct vif_device *vif,
+ unsigned short vif_index, u32 tb_id)
+{
+ struct vif_entry_notifier_info info = {
+ .info = {
+ .family = family,
+ .net = net,
+ },
+ .dev = vif->dev,
+ .vif_index = vif_index,
+ .vif_flags = vif->flags,
+ .tb_id = tb_id,
+ };
+
+ return call_fib_notifier(nb, net, event_type, &info.info);
+}
+
+static inline int mr_call_vif_notifiers(struct net *net,
+ unsigned short family,
+ enum fib_event_type event_type,
+ struct vif_device *vif,
+ unsigned short vif_index, u32 tb_id,
+ unsigned int *ipmr_seq)
+{
+ struct vif_entry_notifier_info info = {
+ .info = {
+ .family = family,
+ .net = net,
+ },
+ .dev = vif->dev,
+ .vif_index = vif_index,
+ .vif_flags = vif->flags,
+ .tb_id = tb_id,
+ };
+
+ ASSERT_RTNL();
+ (*ipmr_seq)++;
+ return call_fib_notifiers(net, event_type, &info.info);
+}
+
#ifndef MAXVIFS
/* This one is nasty; value is defined in uapi using different symbols for
* mroute and morute6 but both map into same 32.
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index f6be5db16da2..bb1a0655f8e4 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -650,18 +650,8 @@ static int call_ipmr_vif_entry_notifier(struct notifier_block *nb,
struct vif_device *vif,
vifi_t vif_index, u32 tb_id)
{
- struct vif_entry_notifier_info info = {
- .info = {
- .family = RTNL_FAMILY_IPMR,
- .net = net,
- },
- .dev = vif->dev,
- .vif_index = vif_index,
- .vif_flags = vif->flags,
- .tb_id = tb_id,
- };
-
- return call_fib_notifier(nb, net, event_type, &info.info);
+ return mr_call_vif_notifier(nb, net, RTNL_FAMILY_IPMR, event_type,
+ vif, vif_index, tb_id);
}
static int call_ipmr_vif_entry_notifiers(struct net *net,
@@ -669,20 +659,9 @@ static int call_ipmr_vif_entry_notifiers(struct net *net,
struct vif_device *vif,
vifi_t vif_index, u32 tb_id)
{
- struct vif_entry_notifier_info info = {
- .info = {
- .family = RTNL_FAMILY_IPMR,
- .net = net,
- },
- .dev = vif->dev,
- .vif_index = vif_index,
- .vif_flags = vif->flags,
- .tb_id = tb_id,
- };
-
- ASSERT_RTNL();
- net->ipv4.ipmr_seq++;
- return call_fib_notifiers(net, event_type, &info.info);
+ return mr_call_vif_notifiers(net, RTNL_FAMILY_IPMR, event_type,
+ vif, vif_index, tb_id,
+ &net->ipv4.ipmr_seq);
}
static int call_ipmr_mfc_entry_notifier(struct notifier_block *nb,
--
2.14.3
^ permalink raw reply related
* Re: [bpf-next V5 PATCH 14/15] xdp: transition into using xdp_frame for return API
From: Jesper Dangaard Brouer @ 2018-03-26 11:42 UTC (permalink / raw)
To: Alexander Duyck
Cc: Netdev, BjörnTöpel, Karlsson, Magnus, Eugenia Emantayev,
Jason Wang, John Fastabend, Eran Ben Elisha, Saeed Mahameed,
Gal Pressman, Daniel Borkmann, Alexei Starovoitov, Tariq Toukan,
brouer
In-Reply-To: <CAKgT0UfqqTt0J-6ssQJ4-drRHTMmOdGyEgadUAt9vcgfN-Enxg@mail.gmail.com>
On Fri, 23 Mar 2018 10:29:29 -0700 Alexander Duyck <alexander.duyck@gmail.com> wrote:
> On Fri, Mar 23, 2018 at 5:19 AM, Jesper Dangaard Brouer
> <brouer@redhat.com> wrote:
> > Changing API xdp_return_frame() to take struct xdp_frame as argument,
> > seems like a natural choice. But there are some subtle performance
> > details here that needs extra care, which is a deliberate choice.
> >
> > When de-referencing xdp_frame on a remote CPU during DMA-TX
> > completion, result in the cache-line is change to "Shared"
> > state. Later when the page is reused for RX, then this xdp_frame
> > cache-line is written, which change the state to "Modified".
> >
> > This situation already happens (naturally) for, virtio_net, tun and
> > cpumap as the xdp_frame pointer is the queued object. In tun and
> > cpumap, the ptr_ring is used for efficiently transferring cache-lines
> > (with pointers) between CPUs. Thus, the only option is to
> > de-referencing xdp_frame.
> >
> > It is only the ixgbe driver that had an optimization, in which it can
> > avoid doing the de-reference of xdp_frame. The driver already have
> > TX-ring queue, which (in case of remote DMA-TX completion) have to be
> > transferred between CPUs anyhow. In this data area, we stored a
> > struct xdp_mem_info and a data pointer, which allowed us to avoid
> > de-referencing xdp_frame.
> >
> > To compensate for this, a prefetchw is used for telling the cache
> > coherency protocol about our access pattern. My benchmarks show that
> > this prefetchw is enough to compensate the ixgbe driver.
> >
> > Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
>
> I'm really not a fan of this patch. It seems like it is adding a ton
> of overhead for one case that is going to penalize all of the other
> use cases for XDP. Just the extra prefetch is going to have a
> significant impact on things like the XDP_DROP case.
>
[...]
> > diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> > index ff069597fccf..e6e9b28ecfba 100644
> > --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> > +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> > @@ -1207,7 +1207,7 @@ static bool ixgbe_clean_tx_irq(struct ixgbe_q_vector *q_vector,
> >
> > /* free the skb */
> > if (ring_is_xdp(tx_ring))
> > - xdp_return_frame(tx_buffer->data, &tx_buffer->xdp_mem);
> > + xdp_return_frame(tx_buffer->xdpf);
> > else
> > napi_consume_skb(tx_buffer->skb, napi_budget);
> >
> > @@ -2376,6 +2376,7 @@ static int ixgbe_clean_rx_irq(struct ixgbe_q_vector *q_vector,
> > xdp.data_hard_start = xdp.data -
> > ixgbe_rx_offset(rx_ring);
> > xdp.data_end = xdp.data + size;
> > + prefetchw(xdp.data_hard_start); /* xdp_frame write */
> >
> > skb = ixgbe_run_xdp(adapter, rx_ring, &xdp);
> > }
>
> Do we really need to be prefetching yet another cache line? This is
> going to hurt general performance since for most cases you are now
> prefetching a cache line that will never be used.
My intuition were also that this would hurt performance. But I've done
many tests, and they all show no performance regression from this change!
XDP_DROP testing with xdp1 on ixgbe:
Baseline: 14,703,344 pkt/s
Patched: 14,711,574 pkt/s
For people reproducing, notice that it requires tuning on the generator
side (with pktgen) to reach these extreme speeds.
As you might have noticed, the prefetchw is only active when a XDP
program is loaded. Thus, I created a benchmark[1] that loads an XDP
program and always returns XDP_PASS (via --action)
Then, I drop packets in iptables raw table:
Baseline: 5,783,509 pps
Patched: 5,789,195 pps
Then unload netfilter modules and let packets reach UDP step
"UdpNoPorts" listening stage:
Baseline: 3,832,956 pps
Patched: 3,855,470 pps
Then, add a udp_sink (--recvmsg) to allow packets to travel deeper into
the network stack (and force udp_sink to run on another CPU):
Baseline: 2,278,855 pps
Patched: 2,270,373 pps
By measurements, it should be clear that this patchset does not add "a
ton of overhead" and does not "penalize all of the other use cases for
XDP".
For the XDP_REDIRECT use-case, I actually find the prefetchw() quite
beautiful, because xdp_buff (which is stack variable) is used by the
bpf_prog, and the prefetch have time to fetch the memory area that the
conversion of xdp_buff to xdp_frame goes into, and xdp_buff is known to
be hot.
--
Best regards,
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
LinkedIn: http://www.linkedin.com/in/brouer
[1] https://github.com/netoptimizer/prototype-kernel/blob/master/kernel/samples/bpf/xdp_bench01_mem_access_cost_kern.c
^ permalink raw reply
* Cześć słodka
From: Wesley @ 2018-03-26 11:29 UTC (permalink / raw)
Am Wes ze Stanów Zjednoczonych, ale obecnie przebywa w Syrii na misji pokojowej. Obecnie szukam przyjaźni, która doprowadzi do związku, w którym znowu czuję się kochana ...
Chcę cię lepiej poznać, jeśli mogę być odważny. Uważam się za łatwego człowieka ..............
Proszę wybaczyć moje maniery nie są dobre, jeśli chodzi o Internet, ponieważ to nie jest moja dziedzina. Tutaj w Syrii nie wolno nam wychodzić, co sprawia, że bardzo się nudzę, więc myślę, że potrzebuję przyjaciela do rozmowy z zewnątrz, żeby mnie utrzymać ...
Chciałbym poznać "prawdziwego" ciebie jako przyjaciela. Twoje polubienia, nielubienia, twoje zainteresowania ...... co cię wyróżnia.
Mój ulubiony kolor to niebieski. Moje ulubione jedzenie to BACON, mogłem z łatwością zostać wegetarianinem, gdyby nie było to na bekonie !!
Mam nadzieję, że możesz mi powiedzieć więcej szczegółów na temat twojej pracy, związku i przeszłości .........
Mam nadzieję, że wkrótce skontaktuję się z Tobą .........
Wes.
^ permalink raw reply
* [PATCH 2/2] af_key: Always verify length of provided sadb_key
From: Kevin Easton @ 2018-03-26 11:39 UTC (permalink / raw)
To: Steffen Klassert, Herbert Xu, David S. Miller, netdev,
linux-kernel
In-Reply-To: <cover.1522063171.git.kevin@guarana.org>
Key extensions (struct sadb_key) include a user-specified number of key
bits. The kernel uses that number to determine how much key data to copy
out of the message in pfkey_msg2xfrm_state().
The length of the sadb_key message must be verified to be long enough,
even in the case of SADB_X_AALG_NULL. Furthermore, the sadb_key_len value
must be long enough to include both the key data and the struct sadb_key
itself.
Introduce a helper function verify_key_len(), and call it from
parse_exthdrs() where other exthdr types are similarly checked for
correctness.
Signed-off-by: Kevin Easton <kevin@guarana.org>
Reported-by: syzbot+5022a34ca5a3d49b84223653fab632dfb7b4cf37@syzkaller.appspotmail.com
---
net/key/af_key.c | 45 +++++++++++++++++++++++++++++++++++----------
1 file changed, 35 insertions(+), 10 deletions(-)
diff --git a/net/key/af_key.c b/net/key/af_key.c
index 911b68d..f3ebb84 100644
--- a/net/key/af_key.c
+++ b/net/key/af_key.c
@@ -437,6 +437,24 @@ static int verify_address_len(const void *p)
return 0;
}
+static inline int sadb_key_len(const struct sadb_key *key)
+{
+ int key_bytes = DIV_ROUND_UP(key->sadb_key_bits, 8);
+
+ return DIV_ROUND_UP(sizeof(struct sadb_key) + key_bytes,
+ sizeof(uint64_t));
+}
+
+static int verify_key_len(const void *p)
+{
+ const struct sadb_key *key = p;
+
+ if (sadb_key_len(key) > key->sadb_key_len)
+ return -EINVAL;
+
+ return 0;
+}
+
static inline int pfkey_sec_ctx_len(const struct sadb_x_sec_ctx *sec_ctx)
{
return DIV_ROUND_UP(sizeof(struct sadb_x_sec_ctx) +
@@ -533,16 +551,25 @@ static int parse_exthdrs(struct sk_buff *skb, const struct sadb_msg *hdr, void *
return -EINVAL;
if (ext_hdrs[ext_type-1] != NULL)
return -EINVAL;
- if (ext_type == SADB_EXT_ADDRESS_SRC ||
- ext_type == SADB_EXT_ADDRESS_DST ||
- ext_type == SADB_EXT_ADDRESS_PROXY ||
- ext_type == SADB_X_EXT_NAT_T_OA) {
+ switch (ext_type) {
+ case SADB_EXT_ADDRESS_SRC:
+ case SADB_EXT_ADDRESS_DST:
+ case SADB_EXT_ADDRESS_PROXY:
+ case SADB_X_EXT_NAT_T_OA:
if (verify_address_len(p))
return -EINVAL;
- }
- if (ext_type == SADB_X_EXT_SEC_CTX) {
+ break;
+ case SADB_X_EXT_SEC_CTX:
if (verify_sec_ctx_len(p))
return -EINVAL;
+ break;
+ case SADB_EXT_KEY_AUTH:
+ case SADB_EXT_KEY_ENCRYPT:
+ if (verify_key_len(p))
+ return -EINVAL;
+ break;
+ default:
+ break;
}
ext_hdrs[ext_type-1] = (void *) p;
}
@@ -1105,14 +1132,12 @@ static struct xfrm_state * pfkey_msg2xfrm_state(struct net *net,
key = ext_hdrs[SADB_EXT_KEY_AUTH - 1];
if (key != NULL &&
sa->sadb_sa_auth != SADB_X_AALG_NULL &&
- (DIV_ROUND_UP(key->sadb_key_bits, 8) == 0 ||
- DIV_ROUND_UP(key->sadb_key_bits, 8) > key->sadb_key_len * sizeof(uint64_t)))
+ key->sadb_key_bits == 0)
return ERR_PTR(-EINVAL);
key = ext_hdrs[SADB_EXT_KEY_ENCRYPT-1];
if (key != NULL &&
sa->sadb_sa_encrypt != SADB_EALG_NULL &&
- (DIV_ROUND_UP(key->sadb_key_bits, 8) == 0 ||
- DIV_ROUND_UP(key->sadb_key_bits, 8) > key->sadb_key_len * sizeof(uint64_t)))
+ key->sadb_key_bits == 0)
return ERR_PTR(-EINVAL);
x = xfrm_state_alloc(net);
--
2.8.1
^ permalink raw reply related
* [PATCH 1/2] af_key: Use DIV_ROUND_UP() instead of open-coded equivalent
From: Kevin Easton @ 2018-03-26 11:39 UTC (permalink / raw)
To: Steffen Klassert, Herbert Xu, David S. Miller, netdev,
linux-kernel
In-Reply-To: <cover.1522063171.git.kevin@guarana.org>
Several places use (x + 7) / 8 to convert from a number of bits to a number
of bytes. Replace those with DIV_ROUND_UP(x, 8) instead, for consistency
with other parts of the same file.
Signed-off-by: Kevin Easton <kevin@guarana.org>
---
net/key/af_key.c | 21 +++++++++++----------
1 file changed, 11 insertions(+), 10 deletions(-)
diff --git a/net/key/af_key.c b/net/key/af_key.c
index 7e2e718..911b68d 100644
--- a/net/key/af_key.c
+++ b/net/key/af_key.c
@@ -795,12 +795,12 @@ static struct sk_buff *__pfkey_xfrm_state2msg(const struct xfrm_state *x,
if (add_keys) {
if (x->aalg && x->aalg->alg_key_len) {
auth_key_size =
- PFKEY_ALIGN8((x->aalg->alg_key_len + 7) / 8);
+ PFKEY_ALIGN8(DIV_ROUND_UP(x->aalg->alg_key_len, 8));
size += sizeof(struct sadb_key) + auth_key_size;
}
if (x->ealg && x->ealg->alg_key_len) {
encrypt_key_size =
- PFKEY_ALIGN8((x->ealg->alg_key_len+7) / 8);
+ PFKEY_ALIGN8(DIV_ROUND_UP(x->ealg->alg_key_len, 8));
size += sizeof(struct sadb_key) + encrypt_key_size;
}
}
@@ -960,7 +960,8 @@ static struct sk_buff *__pfkey_xfrm_state2msg(const struct xfrm_state *x,
key->sadb_key_exttype = SADB_EXT_KEY_AUTH;
key->sadb_key_bits = x->aalg->alg_key_len;
key->sadb_key_reserved = 0;
- memcpy(key + 1, x->aalg->alg_key, (x->aalg->alg_key_len+7)/8);
+ memcpy(key + 1, x->aalg->alg_key,
+ DIV_ROUND_UP(x->aalg->alg_key_len, 8));
}
/* encrypt key */
if (add_keys && encrypt_key_size) {
@@ -971,7 +972,7 @@ static struct sk_buff *__pfkey_xfrm_state2msg(const struct xfrm_state *x,
key->sadb_key_bits = x->ealg->alg_key_len;
key->sadb_key_reserved = 0;
memcpy(key + 1, x->ealg->alg_key,
- (x->ealg->alg_key_len+7)/8);
+ DIV_ROUND_UP(x->ealg->alg_key_len, 8));
}
/* sa */
@@ -1104,14 +1105,14 @@ static struct xfrm_state * pfkey_msg2xfrm_state(struct net *net,
key = ext_hdrs[SADB_EXT_KEY_AUTH - 1];
if (key != NULL &&
sa->sadb_sa_auth != SADB_X_AALG_NULL &&
- ((key->sadb_key_bits+7) / 8 == 0 ||
- (key->sadb_key_bits+7) / 8 > key->sadb_key_len * sizeof(uint64_t)))
+ (DIV_ROUND_UP(key->sadb_key_bits, 8) == 0 ||
+ DIV_ROUND_UP(key->sadb_key_bits, 8) > key->sadb_key_len * sizeof(uint64_t)))
return ERR_PTR(-EINVAL);
key = ext_hdrs[SADB_EXT_KEY_ENCRYPT-1];
if (key != NULL &&
sa->sadb_sa_encrypt != SADB_EALG_NULL &&
- ((key->sadb_key_bits+7) / 8 == 0 ||
- (key->sadb_key_bits+7) / 8 > key->sadb_key_len * sizeof(uint64_t)))
+ (DIV_ROUND_UP(key->sadb_key_bits, 8) == 0 ||
+ DIV_ROUND_UP(key->sadb_key_bits, 8) > key->sadb_key_len * sizeof(uint64_t)))
return ERR_PTR(-EINVAL);
x = xfrm_state_alloc(net);
@@ -1168,7 +1169,7 @@ static struct xfrm_state * pfkey_msg2xfrm_state(struct net *net,
goto out;
}
if (key)
- keysize = (key->sadb_key_bits + 7) / 8;
+ keysize = DIV_ROUND_UP(key->sadb_key_bits, 8);
x->aalg = kmalloc(sizeof(*x->aalg) + keysize, GFP_KERNEL);
if (!x->aalg) {
err = -ENOMEM;
@@ -1207,7 +1208,7 @@ static struct xfrm_state * pfkey_msg2xfrm_state(struct net *net,
}
key = (struct sadb_key*) ext_hdrs[SADB_EXT_KEY_ENCRYPT-1];
if (key)
- keysize = (key->sadb_key_bits + 7) / 8;
+ keysize = DIV_ROUND_UP(key->sadb_key_bits, 8);
x->ealg = kmalloc(sizeof(*x->ealg) + keysize, GFP_KERNEL);
if (!x->ealg) {
err = -ENOMEM;
--
2.8.1
^ permalink raw reply related
* [PATCH 0/2] af_key: Fix for sadb_key memcpy read overrun
From: Kevin Easton @ 2018-03-26 11:39 UTC (permalink / raw)
To: Steffen Klassert, Herbert Xu, David S. Miller, netdev,
linux-kernel
As found by syzbot, af_key does not properly validate the key length in
sadb_key messages from userspace. This can result in copying from beyond
the end of the sadb_key part of the message, or indeed beyond the end of
the entire packet.
Kevin Easton (2):
af_key: Use DIV_ROUND_UP() instead of open-coded equivalent
af_key: Always verify length of provided sadb_key
net/key/af_key.c | 58 ++++++++++++++++++++++++++++++++++++++++----------------
1 file changed, 42 insertions(+), 16 deletions(-)
--
2.8.1
^ permalink raw reply
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