* Re: [Patch net-next v3] tcp: add a tracepoint for tcp retransmission
From: David Ahern @ 2017-10-13 23:13 UTC (permalink / raw)
To: Cong Wang, netdev
Cc: Eric Dumazet, Alexei Starovoitov, Hannes Frederic Sowa,
Brendan Gregg, Neal Cardwell
In-Reply-To: <20171013200316.28561-1-xiyou.wangcong@gmail.com>
On 10/13/17 2:03 PM, Cong Wang wrote:
> diff --git a/include/trace/events/tcp.h b/include/trace/events/tcp.h
> new file mode 100644
> index 000000000000..3d1cbd072b7e
> --- /dev/null
> +++ b/include/trace/events/tcp.h
> @@ -0,0 +1,68 @@
> +#undef TRACE_SYSTEM
> +#define TRACE_SYSTEM tcp
> +
> +#if !defined(_TRACE_TCP_H) || defined(TRACE_HEADER_MULTI_READ)
> +#define _TRACE_TCP_H
> +
> +#include <linux/ipv6.h>
> +#include <linux/tcp.h>
> +#include <linux/tracepoint.h>
> +#include <net/ipv6.h>
> +
> +TRACE_EVENT(tcp_retransmit_skb,
> +
> + TP_PROTO(struct sock *sk, struct sk_buff *skb),
> +
> + TP_ARGS(sk, skb),
> +
> + TP_STRUCT__entry(
> + __field(void *, skbaddr)
> + __field(void *, skaddr)
> + __field(__u16, sport)
> + __field(__u16, dport)
> + __array(__u8, saddr, 4)
> + __array(__u8, daddr, 4)
> + __array(__u8, saddr_v6, 16)
> + __array(__u8, daddr_v6, 16)
> + ),
> +
> + TP_fast_assign(
> + struct ipv6_pinfo *np = inet6_sk(sk);
> + struct inet_sock *inet = inet_sk(sk);
> + struct in6_addr *pin6;
> + __be32 *p32;
> +
> + __entry->skbaddr = skb;
> + __entry->skaddr = sk;
> +
> + __entry->sport = ntohs(inet->inet_sport);
> + __entry->dport = ntohs(inet->inet_dport);
> +
> + p32 = (__be32 *) __entry->saddr;
> + *p32 = inet->inet_saddr;
> +
> + p32 = (__be32 *) __entry->daddr;
> + *p32 = inet->inet_daddr;
> +
> + if (np) {
> + pin6 = (struct in6_addr *)__entry->saddr_v6;
> + *pin6 = np->saddr;
> + pin6 = (struct in6_addr *)__entry->daddr_v6;
> + *pin6 = *(np->daddr_cache);
> + } else {
> + pin6 = (struct in6_addr *)__entry->saddr_v6;
> + ipv6_addr_set_v4mapped(inet->inet_saddr, pin6);
> + pin6 = (struct in6_addr *)__entry->daddr_v6;
> + ipv6_addr_set_v4mapped(inet->inet_daddr, pin6);
> + }
> + ),
> +
> + TP_printk("sport=%hu dport=%hu saddr=%pI4 daddr=%pI4 saddrv6=%pI6 daddrv6=%pI6",
%pI6c is more user friendly for IPv6 addresses.
^ permalink raw reply
* [PATCH net-next 5/5] mlxsw: spectrum_router: Add extack message for RIF and VRF overflow
From: David Ahern @ 2017-10-13 23:02 UTC (permalink / raw)
To: netdev; +Cc: jiri, idosch, kjlx, davem, yoshfuji, David Ahern
In-Reply-To: <1507935733-18950-1-git-send-email-dsahern@gmail.com>
Add extack argument down to mlxsw_sp_rif_create and mlxsw_sp_vr_create
to set an error message on RIF or VR overflow. Now on overflow of
either resource the user gets an informative message as opposed to
failing with EBUSY.
Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
---
.../net/ethernet/mellanox/mlxsw/spectrum_router.c | 114 +++++++++++++--------
1 file changed, 69 insertions(+), 45 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 2a7f066dfab5..9e0b46513ca7 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -731,14 +731,17 @@ static struct mlxsw_sp_fib *mlxsw_sp_vr_fib(const struct mlxsw_sp_vr *vr,
}
static struct mlxsw_sp_vr *mlxsw_sp_vr_create(struct mlxsw_sp *mlxsw_sp,
- u32 tb_id)
+ u32 tb_id,
+ struct netlink_ext_ack *extack)
{
struct mlxsw_sp_vr *vr;
int err;
vr = mlxsw_sp_vr_find_unused(mlxsw_sp);
- if (!vr)
+ if (!vr) {
+ NL_SET_ERR_MSG(extack, "spectrum: Exceeded number of supported virtual routers");
return ERR_PTR(-EBUSY);
+ }
vr->fib4 = mlxsw_sp_fib_create(vr, MLXSW_SP_L3_PROTO_IPV4);
if (IS_ERR(vr->fib4))
return ERR_CAST(vr->fib4);
@@ -775,14 +778,15 @@ static void mlxsw_sp_vr_destroy(struct mlxsw_sp_vr *vr)
vr->fib4 = NULL;
}
-static struct mlxsw_sp_vr *mlxsw_sp_vr_get(struct mlxsw_sp *mlxsw_sp, u32 tb_id)
+static struct mlxsw_sp_vr *mlxsw_sp_vr_get(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
+ struct netlink_ext_ack *extack)
{
struct mlxsw_sp_vr *vr;
tb_id = mlxsw_sp_fix_tb_id(tb_id);
vr = mlxsw_sp_vr_find(mlxsw_sp, tb_id);
if (!vr)
- vr = mlxsw_sp_vr_create(mlxsw_sp, tb_id);
+ vr = mlxsw_sp_vr_create(mlxsw_sp, tb_id, extack);
return vr;
}
@@ -948,7 +952,8 @@ static u32 mlxsw_sp_ipip_dev_ul_tb_id(const struct net_device *ol_dev)
static struct mlxsw_sp_rif *
mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
- const struct mlxsw_sp_rif_params *params);
+ const struct mlxsw_sp_rif_params *params,
+ struct netlink_ext_ack *extack);
static struct mlxsw_sp_rif_ipip_lb *
mlxsw_sp_ipip_ol_ipip_lb_create(struct mlxsw_sp *mlxsw_sp,
@@ -966,7 +971,7 @@ mlxsw_sp_ipip_ol_ipip_lb_create(struct mlxsw_sp *mlxsw_sp,
.lb_config = ipip_ops->ol_loopback_config(mlxsw_sp, ol_dev),
};
- rif = mlxsw_sp_rif_create(mlxsw_sp, &lb_params.common);
+ rif = mlxsw_sp_rif_create(mlxsw_sp, &lb_params.common, NULL);
if (IS_ERR(rif))
return ERR_CAST(rif);
return container_of(rif, struct mlxsw_sp_rif_ipip_lb, common);
@@ -3711,7 +3716,7 @@ mlxsw_sp_fib_node_get(struct mlxsw_sp *mlxsw_sp, u32 tb_id, const void *addr,
struct mlxsw_sp_vr *vr;
int err;
- vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id);
+ vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id, NULL);
if (IS_ERR(vr))
return ERR_CAST(vr);
fib = mlxsw_sp_vr_fib(vr, proto);
@@ -4750,7 +4755,7 @@ static int mlxsw_sp_router_fibmr_add(struct mlxsw_sp *mlxsw_sp,
if (mlxsw_sp->router->aborted)
return 0;
- vr = mlxsw_sp_vr_get(mlxsw_sp, men_info->tb_id);
+ vr = mlxsw_sp_vr_get(mlxsw_sp, men_info->tb_id, NULL);
if (IS_ERR(vr))
return PTR_ERR(vr);
@@ -4783,7 +4788,7 @@ mlxsw_sp_router_fibmr_vif_add(struct mlxsw_sp *mlxsw_sp,
if (mlxsw_sp->router->aborted)
return 0;
- vr = mlxsw_sp_vr_get(mlxsw_sp, ven_info->tb_id);
+ vr = mlxsw_sp_vr_get(mlxsw_sp, ven_info->tb_id, NULL);
if (IS_ERR(vr))
return PTR_ERR(vr);
@@ -5346,7 +5351,8 @@ const struct net_device *mlxsw_sp_rif_dev(const struct mlxsw_sp_rif *rif)
static struct mlxsw_sp_rif *
mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
- const struct mlxsw_sp_rif_params *params)
+ const struct mlxsw_sp_rif_params *params,
+ struct netlink_ext_ack *extack)
{
u32 tb_id = l3mdev_fib_table(params->dev);
const struct mlxsw_sp_rif_ops *ops;
@@ -5360,14 +5366,16 @@ mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
type = mlxsw_sp_dev_rif_type(mlxsw_sp, params->dev);
ops = mlxsw_sp->router->rif_ops_arr[type];
- vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id ? : RT_TABLE_MAIN);
+ vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id ? : RT_TABLE_MAIN, extack);
if (IS_ERR(vr))
return ERR_CAST(vr);
vr->rif_count++;
err = mlxsw_sp_rif_index_alloc(mlxsw_sp, &rif_index);
- if (err)
+ if (err) {
+ NL_SET_ERR_MSG(extack, "spectrum: Exceeded number of supported router interfaces");
goto err_rif_index_alloc;
+ }
rif = mlxsw_sp_rif_alloc(ops->rif_size, rif_index, vr->id, params->dev);
if (!rif) {
@@ -5454,7 +5462,8 @@ mlxsw_sp_rif_subport_params_init(struct mlxsw_sp_rif_params *params,
static int
mlxsw_sp_port_vlan_router_join(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan,
- struct net_device *l3_dev)
+ struct net_device *l3_dev,
+ struct netlink_ext_ack *extack)
{
struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
@@ -5470,7 +5479,7 @@ mlxsw_sp_port_vlan_router_join(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan,
};
mlxsw_sp_rif_subport_params_init(¶ms, mlxsw_sp_port_vlan);
- rif = mlxsw_sp_rif_create(mlxsw_sp, ¶ms);
+ rif = mlxsw_sp_rif_create(mlxsw_sp, ¶ms, extack);
if (IS_ERR(rif))
return PTR_ERR(rif);
}
@@ -5525,7 +5534,8 @@ mlxsw_sp_port_vlan_router_leave(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan)
static int mlxsw_sp_inetaddr_port_vlan_event(struct net_device *l3_dev,
struct net_device *port_dev,
- unsigned long event, u16 vid)
+ unsigned long event, u16 vid,
+ struct netlink_ext_ack *extack)
{
struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(port_dev);
struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
@@ -5537,7 +5547,7 @@ static int mlxsw_sp_inetaddr_port_vlan_event(struct net_device *l3_dev,
switch (event) {
case NETDEV_UP:
return mlxsw_sp_port_vlan_router_join(mlxsw_sp_port_vlan,
- l3_dev);
+ l3_dev, extack);
case NETDEV_DOWN:
mlxsw_sp_port_vlan_router_leave(mlxsw_sp_port_vlan);
break;
@@ -5547,19 +5557,22 @@ static int mlxsw_sp_inetaddr_port_vlan_event(struct net_device *l3_dev,
}
static int mlxsw_sp_inetaddr_port_event(struct net_device *port_dev,
- unsigned long event)
+ unsigned long event,
+ struct netlink_ext_ack *extack)
{
if (netif_is_bridge_port(port_dev) ||
netif_is_lag_port(port_dev) ||
netif_is_ovs_port(port_dev))
return 0;
- return mlxsw_sp_inetaddr_port_vlan_event(port_dev, port_dev, event, 1);
+ return mlxsw_sp_inetaddr_port_vlan_event(port_dev, port_dev, event, 1,
+ extack);
}
static int __mlxsw_sp_inetaddr_lag_event(struct net_device *l3_dev,
struct net_device *lag_dev,
- unsigned long event, u16 vid)
+ unsigned long event, u16 vid,
+ struct netlink_ext_ack *extack)
{
struct net_device *port_dev;
struct list_head *iter;
@@ -5569,7 +5582,8 @@ static int __mlxsw_sp_inetaddr_lag_event(struct net_device *l3_dev,
if (mlxsw_sp_port_dev_check(port_dev)) {
err = mlxsw_sp_inetaddr_port_vlan_event(l3_dev,
port_dev,
- event, vid);
+ event, vid,
+ extack);
if (err)
return err;
}
@@ -5579,16 +5593,19 @@ static int __mlxsw_sp_inetaddr_lag_event(struct net_device *l3_dev,
}
static int mlxsw_sp_inetaddr_lag_event(struct net_device *lag_dev,
- unsigned long event)
+ unsigned long event,
+ struct netlink_ext_ack *extack)
{
if (netif_is_bridge_port(lag_dev))
return 0;
- return __mlxsw_sp_inetaddr_lag_event(lag_dev, lag_dev, event, 1);
+ return __mlxsw_sp_inetaddr_lag_event(lag_dev, lag_dev, event, 1,
+ extack);
}
static int mlxsw_sp_inetaddr_bridge_event(struct net_device *l3_dev,
- unsigned long event)
+ unsigned long event,
+ struct netlink_ext_ack *extack)
{
struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(l3_dev);
struct mlxsw_sp_rif_params params = {
@@ -5598,7 +5615,7 @@ static int mlxsw_sp_inetaddr_bridge_event(struct net_device *l3_dev,
switch (event) {
case NETDEV_UP:
- rif = mlxsw_sp_rif_create(mlxsw_sp, ¶ms);
+ rif = mlxsw_sp_rif_create(mlxsw_sp, ¶ms, extack);
if (IS_ERR(rif))
return PTR_ERR(rif);
break;
@@ -5612,7 +5629,8 @@ static int mlxsw_sp_inetaddr_bridge_event(struct net_device *l3_dev,
}
static int mlxsw_sp_inetaddr_vlan_event(struct net_device *vlan_dev,
- unsigned long event)
+ unsigned long event,
+ struct netlink_ext_ack *extack)
{
struct net_device *real_dev = vlan_dev_real_dev(vlan_dev);
u16 vid = vlan_dev_vlan_id(vlan_dev);
@@ -5622,27 +5640,28 @@ static int mlxsw_sp_inetaddr_vlan_event(struct net_device *vlan_dev,
if (mlxsw_sp_port_dev_check(real_dev))
return mlxsw_sp_inetaddr_port_vlan_event(vlan_dev, real_dev,
- event, vid);
+ event, vid, extack);
else if (netif_is_lag_master(real_dev))
return __mlxsw_sp_inetaddr_lag_event(vlan_dev, real_dev, event,
- vid);
+ vid, extack);
else if (netif_is_bridge_master(real_dev) && br_vlan_enabled(real_dev))
- return mlxsw_sp_inetaddr_bridge_event(vlan_dev, event);
+ return mlxsw_sp_inetaddr_bridge_event(vlan_dev, event, extack);
return 0;
}
static int __mlxsw_sp_inetaddr_event(struct net_device *dev,
- unsigned long event)
+ unsigned long event,
+ struct netlink_ext_ack *extack)
{
if (mlxsw_sp_port_dev_check(dev))
- return mlxsw_sp_inetaddr_port_event(dev, event);
+ return mlxsw_sp_inetaddr_port_event(dev, event, extack);
else if (netif_is_lag_master(dev))
- return mlxsw_sp_inetaddr_lag_event(dev, event);
+ return mlxsw_sp_inetaddr_lag_event(dev, event, extack);
else if (netif_is_bridge_master(dev))
- return mlxsw_sp_inetaddr_bridge_event(dev, event);
+ return mlxsw_sp_inetaddr_bridge_event(dev, event, extack);
else if (is_vlan_dev(dev))
- return mlxsw_sp_inetaddr_vlan_event(dev, event);
+ return mlxsw_sp_inetaddr_vlan_event(dev, event, extack);
else
return 0;
}
@@ -5668,7 +5687,7 @@ int mlxsw_sp_inetaddr_event(struct notifier_block *unused,
if (!mlxsw_sp_rif_should_config(rif, dev, event))
goto out;
- err = __mlxsw_sp_inetaddr_event(dev, event);
+ err = __mlxsw_sp_inetaddr_event(dev, event, NULL);
out:
return notifier_from_errno(err);
}
@@ -5690,7 +5709,7 @@ int mlxsw_sp_inetaddr_valid_event(struct notifier_block *unused,
if (!mlxsw_sp_rif_should_config(rif, dev, event))
goto out;
- err = __mlxsw_sp_inetaddr_event(dev, event);
+ err = __mlxsw_sp_inetaddr_event(dev, event, ivi->extack);
out:
return notifier_from_errno(err);
}
@@ -5719,7 +5738,7 @@ static void mlxsw_sp_inet6addr_event_work(struct work_struct *work)
if (!mlxsw_sp_rif_should_config(rif, dev, event))
goto out;
- __mlxsw_sp_inetaddr_event(dev, event);
+ __mlxsw_sp_inetaddr_event(dev, event, NULL);
out:
rtnl_unlock();
dev_put(dev);
@@ -5771,7 +5790,7 @@ int mlxsw_sp_inet6addr_valid_event(struct notifier_block *unused,
if (!mlxsw_sp_rif_should_config(rif, dev, event))
goto out;
- err = __mlxsw_sp_inetaddr_event(dev, event);
+ err = __mlxsw_sp_inetaddr_event(dev, event, i6vi->extack);
out:
return notifier_from_errno(err);
}
@@ -5848,7 +5867,8 @@ int mlxsw_sp_netdevice_router_port_event(struct net_device *dev)
}
static int mlxsw_sp_port_vrf_join(struct mlxsw_sp *mlxsw_sp,
- struct net_device *l3_dev)
+ struct net_device *l3_dev,
+ struct netlink_ext_ack *extack)
{
struct mlxsw_sp_rif *rif;
@@ -5857,9 +5877,9 @@ static int mlxsw_sp_port_vrf_join(struct mlxsw_sp *mlxsw_sp,
*/
rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, l3_dev);
if (rif)
- __mlxsw_sp_inetaddr_event(l3_dev, NETDEV_DOWN);
+ __mlxsw_sp_inetaddr_event(l3_dev, NETDEV_DOWN, extack);
- return __mlxsw_sp_inetaddr_event(l3_dev, NETDEV_UP);
+ return __mlxsw_sp_inetaddr_event(l3_dev, NETDEV_UP, extack);
}
static void mlxsw_sp_port_vrf_leave(struct mlxsw_sp *mlxsw_sp,
@@ -5870,7 +5890,7 @@ static void mlxsw_sp_port_vrf_leave(struct mlxsw_sp *mlxsw_sp,
rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, l3_dev);
if (!rif)
return;
- __mlxsw_sp_inetaddr_event(l3_dev, NETDEV_DOWN);
+ __mlxsw_sp_inetaddr_event(l3_dev, NETDEV_DOWN, NULL);
}
int mlxsw_sp_netdevice_vrf_event(struct net_device *l3_dev, unsigned long event,
@@ -5886,10 +5906,14 @@ int mlxsw_sp_netdevice_vrf_event(struct net_device *l3_dev, unsigned long event,
case NETDEV_PRECHANGEUPPER:
return 0;
case NETDEV_CHANGEUPPER:
- if (info->linking)
- err = mlxsw_sp_port_vrf_join(mlxsw_sp, l3_dev);
- else
+ if (info->linking) {
+ struct netlink_ext_ack *extack;
+
+ extack = netdev_notifier_info_to_extack(&info->info);
+ err = mlxsw_sp_port_vrf_join(mlxsw_sp, l3_dev, extack);
+ } else {
mlxsw_sp_port_vrf_leave(mlxsw_sp, l3_dev);
+ }
break;
}
@@ -6196,7 +6220,7 @@ mlxsw_sp_rif_ipip_lb_configure(struct mlxsw_sp_rif *rif)
struct mlxsw_sp_vr *ul_vr;
int err;
- ul_vr = mlxsw_sp_vr_get(mlxsw_sp, ul_tb_id);
+ ul_vr = mlxsw_sp_vr_get(mlxsw_sp, ul_tb_id, NULL);
if (IS_ERR(ul_vr))
return PTR_ERR(ul_vr);
--
2.1.4
^ permalink raw reply related
* [PATCH net-next 4/5] mlxsw: spectrum: router: Add support for address validator notifier
From: David Ahern @ 2017-10-13 23:02 UTC (permalink / raw)
To: netdev; +Cc: jiri, idosch, kjlx, davem, yoshfuji, David Ahern
In-Reply-To: <1507935733-18950-1-git-send-email-dsahern@gmail.com>
Add support for inetaddr_validator and inet6addr_validator. The
notifiers provide a means for validating ipv4 and ipv6 addresses
before the addresses are installed and on failure the error
is propagated back to the user.
Signed-off-by: David Ahern <dsahern@gmail.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 15 ++++++-
drivers/net/ethernet/mellanox/mlxsw/spectrum.h | 4 ++
.../net/ethernet/mellanox/mlxsw/spectrum_router.c | 52 ++++++++++++++++++++++
3 files changed, 70 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 321988ac57cc..d51402f98f97 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -4505,9 +4505,16 @@ static struct notifier_block mlxsw_sp_netdevice_nb __read_mostly = {
.notifier_call = mlxsw_sp_netdevice_event,
};
+static struct notifier_block mlxsw_sp_inetaddr_valid_nb __read_mostly = {
+ .notifier_call = mlxsw_sp_inetaddr_valid_event,
+};
+
static struct notifier_block mlxsw_sp_inetaddr_nb __read_mostly = {
.notifier_call = mlxsw_sp_inetaddr_event,
- .priority = 10, /* Must be called before FIB notifier block */
+};
+
+static struct notifier_block mlxsw_sp_inet6addr_valid_nb __read_mostly = {
+ .notifier_call = mlxsw_sp_inet6addr_valid_event,
};
static struct notifier_block mlxsw_sp_inet6addr_nb __read_mostly = {
@@ -4533,7 +4540,9 @@ static int __init mlxsw_sp_module_init(void)
int err;
register_netdevice_notifier(&mlxsw_sp_netdevice_nb);
+ register_inetaddr_validator_notifier(&mlxsw_sp_inetaddr_valid_nb);
register_inetaddr_notifier(&mlxsw_sp_inetaddr_nb);
+ register_inet6addr_validator_notifier(&mlxsw_sp_inet6addr_valid_nb);
register_inet6addr_notifier(&mlxsw_sp_inet6addr_nb);
register_netevent_notifier(&mlxsw_sp_router_netevent_nb);
@@ -4552,7 +4561,9 @@ static int __init mlxsw_sp_module_init(void)
err_core_driver_register:
unregister_netevent_notifier(&mlxsw_sp_router_netevent_nb);
unregister_inet6addr_notifier(&mlxsw_sp_inet6addr_nb);
+ unregister_inet6addr_validator_notifier(&mlxsw_sp_inet6addr_valid_nb);
unregister_inetaddr_notifier(&mlxsw_sp_inetaddr_nb);
+ unregister_inetaddr_validator_notifier(&mlxsw_sp_inetaddr_valid_nb);
unregister_netdevice_notifier(&mlxsw_sp_netdevice_nb);
return err;
}
@@ -4563,7 +4574,9 @@ static void __exit mlxsw_sp_module_exit(void)
mlxsw_core_driver_unregister(&mlxsw_sp_driver);
unregister_netevent_notifier(&mlxsw_sp_router_netevent_nb);
unregister_inet6addr_notifier(&mlxsw_sp_inet6addr_nb);
+ unregister_inet6addr_validator_notifier(&mlxsw_sp_inet6addr_valid_nb);
unregister_inetaddr_notifier(&mlxsw_sp_inetaddr_nb);
+ unregister_inetaddr_validator_notifier(&mlxsw_sp_inetaddr_valid_nb);
unregister_netdevice_notifier(&mlxsw_sp_netdevice_nb);
}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 8e45183dc9bb..4865a6f58c83 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -390,8 +390,12 @@ int mlxsw_sp_router_netevent_event(struct notifier_block *unused,
int mlxsw_sp_netdevice_router_port_event(struct net_device *dev);
int mlxsw_sp_inetaddr_event(struct notifier_block *unused,
unsigned long event, void *ptr);
+int mlxsw_sp_inetaddr_valid_event(struct notifier_block *unused,
+ unsigned long event, void *ptr);
int mlxsw_sp_inet6addr_event(struct notifier_block *unused,
unsigned long event, void *ptr);
+int mlxsw_sp_inet6addr_valid_event(struct notifier_block *unused,
+ unsigned long event, void *ptr);
int mlxsw_sp_netdevice_vrf_event(struct net_device *l3_dev, unsigned long event,
struct netdev_notifier_changeupper_info *info);
void
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 6a356f4b99a3..2a7f066dfab5 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -5656,6 +5656,32 @@ int mlxsw_sp_inetaddr_event(struct notifier_block *unused,
struct mlxsw_sp_rif *rif;
int err = 0;
+ /* NETDEV_UP event is handled by mlxsw_sp_inetaddr_valid_event */
+ if (event == NETDEV_UP)
+ goto out;
+
+ mlxsw_sp = mlxsw_sp_lower_get(dev);
+ if (!mlxsw_sp)
+ goto out;
+
+ rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+ if (!mlxsw_sp_rif_should_config(rif, dev, event))
+ goto out;
+
+ err = __mlxsw_sp_inetaddr_event(dev, event);
+out:
+ return notifier_from_errno(err);
+}
+
+int mlxsw_sp_inetaddr_valid_event(struct notifier_block *unused,
+ unsigned long event, void *ptr)
+{
+ struct in_validator_info *ivi = (struct in_validator_info *) ptr;
+ struct net_device *dev = ivi->ivi_dev->dev;
+ struct mlxsw_sp *mlxsw_sp;
+ struct mlxsw_sp_rif *rif;
+ int err = 0;
+
mlxsw_sp = mlxsw_sp_lower_get(dev);
if (!mlxsw_sp)
goto out;
@@ -5708,6 +5734,10 @@ int mlxsw_sp_inet6addr_event(struct notifier_block *unused,
struct mlxsw_sp_inet6addr_event_work *inet6addr_work;
struct net_device *dev = if6->idev->dev;
+ /* NETDEV_UP event is handled by mlxsw_sp_inet6addr_valid_event */
+ if (event == NETDEV_UP)
+ return NOTIFY_DONE;
+
if (!mlxsw_sp_port_dev_lower_find_rcu(dev))
return NOTIFY_DONE;
@@ -5724,6 +5754,28 @@ int mlxsw_sp_inet6addr_event(struct notifier_block *unused,
return NOTIFY_DONE;
}
+int mlxsw_sp_inet6addr_valid_event(struct notifier_block *unused,
+ unsigned long event, void *ptr)
+{
+ struct in6_validator_info *i6vi = (struct in6_validator_info *) ptr;
+ struct net_device *dev = i6vi->i6vi_dev->dev;
+ struct mlxsw_sp *mlxsw_sp;
+ struct mlxsw_sp_rif *rif;
+ int err = 0;
+
+ mlxsw_sp = mlxsw_sp_lower_get(dev);
+ if (!mlxsw_sp)
+ goto out;
+
+ rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+ if (!mlxsw_sp_rif_should_config(rif, dev, event))
+ goto out;
+
+ err = __mlxsw_sp_inetaddr_event(dev, event);
+out:
+ return notifier_from_errno(err);
+}
+
static int mlxsw_sp_rif_edit(struct mlxsw_sp *mlxsw_sp, u16 rif_index,
const char *mac, int mtu)
{
--
2.1.4
^ permalink raw reply related
* [PATCH net-next 3/5] net: Add extack to validator_info structs used for address notifier
From: David Ahern @ 2017-10-13 23:02 UTC (permalink / raw)
To: netdev; +Cc: jiri, idosch, kjlx, davem, yoshfuji, David Ahern
In-Reply-To: <1507935733-18950-1-git-send-email-dsahern@gmail.com>
Add extack to in_validator_info and in6_validator_info. Update the one
user of each, ipvlan, to return an error message for failures.
Only manual configuration of an address is plumbed in the IPv6 code path.
Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ipvlan/ipvlan_main.c | 10 ++++++++--
include/linux/inetdevice.h | 1 +
include/net/addrconf.h | 1 +
net/ipv4/devinet.c | 8 +++++---
net/ipv6/addrconf.c | 22 ++++++++++++----------
5 files changed, 27 insertions(+), 15 deletions(-)
diff --git a/drivers/net/ipvlan/ipvlan_main.c b/drivers/net/ipvlan/ipvlan_main.c
index 6842739b6679..f0ab55df57f1 100644
--- a/drivers/net/ipvlan/ipvlan_main.c
+++ b/drivers/net/ipvlan/ipvlan_main.c
@@ -847,8 +847,11 @@ static int ipvlan_addr6_validator_event(struct notifier_block *unused,
switch (event) {
case NETDEV_UP:
- if (ipvlan_addr_busy(ipvlan->port, &i6vi->i6vi_addr, true))
+ if (ipvlan_addr_busy(ipvlan->port, &i6vi->i6vi_addr, true)) {
+ NL_SET_ERR_MSG(i6vi->extack,
+ "Address already assigned to an ipvlan device");
return notifier_from_errno(-EADDRINUSE);
+ }
break;
}
@@ -917,8 +920,11 @@ static int ipvlan_addr4_validator_event(struct notifier_block *unused,
switch (event) {
case NETDEV_UP:
- if (ipvlan_addr_busy(ipvlan->port, &ivi->ivi_addr, false))
+ if (ipvlan_addr_busy(ipvlan->port, &ivi->ivi_addr, false)) {
+ NL_SET_ERR_MSG(ivi->extack,
+ "Address already assigned to an ipvlan device");
return notifier_from_errno(-EADDRINUSE);
+ }
break;
}
diff --git a/include/linux/inetdevice.h b/include/linux/inetdevice.h
index 751d051f0bc7..681dff30940b 100644
--- a/include/linux/inetdevice.h
+++ b/include/linux/inetdevice.h
@@ -154,6 +154,7 @@ struct in_ifaddr {
struct in_validator_info {
__be32 ivi_addr;
struct in_device *ivi_dev;
+ struct netlink_ext_ack *extack;
};
int register_inetaddr_notifier(struct notifier_block *nb);
diff --git a/include/net/addrconf.h b/include/net/addrconf.h
index 87981cd63180..b8b16437c6d5 100644
--- a/include/net/addrconf.h
+++ b/include/net/addrconf.h
@@ -55,6 +55,7 @@ struct prefix_info {
struct in6_validator_info {
struct in6_addr i6vi_addr;
struct inet6_dev *i6vi_dev;
+ struct netlink_ext_ack *extack;
};
#define IN6_ADDR_HSIZE_SHIFT 4
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 7ce22a2c07ce..93773e5a80c7 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -444,7 +444,7 @@ static void check_lifetime(struct work_struct *work);
static DECLARE_DELAYED_WORK(check_lifetime_work, check_lifetime);
static int __inet_insert_ifa(struct in_ifaddr *ifa, struct nlmsghdr *nlh,
- u32 portid)
+ u32 portid, struct netlink_ext_ack *extack)
{
struct in_device *in_dev = ifa->ifa_dev;
struct in_ifaddr *ifa1, **ifap, **last_primary;
@@ -489,6 +489,7 @@ static int __inet_insert_ifa(struct in_ifaddr *ifa, struct nlmsghdr *nlh,
*/
ivi.ivi_addr = ifa->ifa_address;
ivi.ivi_dev = ifa->ifa_dev;
+ ivi.extack = extack;
ret = blocking_notifier_call_chain(&inetaddr_validator_chain,
NETDEV_UP, &ivi);
ret = notifier_to_errno(ret);
@@ -521,7 +522,7 @@ static int __inet_insert_ifa(struct in_ifaddr *ifa, struct nlmsghdr *nlh,
static int inet_insert_ifa(struct in_ifaddr *ifa)
{
- return __inet_insert_ifa(ifa, NULL, 0);
+ return __inet_insert_ifa(ifa, NULL, 0, NULL);
}
static int inet_set_ifa(struct net_device *dev, struct in_ifaddr *ifa)
@@ -902,7 +903,8 @@ static int inet_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh,
return ret;
}
}
- return __inet_insert_ifa(ifa, nlh, NETLINK_CB(skb).portid);
+ return __inet_insert_ifa(ifa, nlh, NETLINK_CB(skb).portid,
+ extack);
} else {
inet_free_ifa(ifa);
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 31ff12277bcf..0075dd3fdc57 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -987,7 +987,7 @@ static struct inet6_ifaddr *
ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
const struct in6_addr *peer_addr, int pfxlen,
int scope, u32 flags, u32 valid_lft, u32 prefered_lft,
- bool can_block)
+ bool can_block, struct netlink_ext_ack *extack)
{
gfp_t gfp_flags = can_block ? GFP_KERNEL : GFP_ATOMIC;
struct net *net = dev_net(idev->dev);
@@ -1019,6 +1019,7 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
struct in6_validator_info i6vi = {
.i6vi_addr = *addr,
.i6vi_dev = idev,
+ .extack = extack,
};
err = inet6addr_validator_notifier_call_chain(NETDEV_UP, &i6vi);
@@ -1351,7 +1352,7 @@ static int ipv6_create_tempaddr(struct inet6_ifaddr *ifp, struct inet6_ifaddr *i
ift = ipv6_add_addr(idev, &addr, NULL, tmp_plen,
ipv6_addr_scope(&addr), addr_flags,
- tmp_valid_lft, tmp_prefered_lft, true);
+ tmp_valid_lft, tmp_prefered_lft, true, NULL);
if (IS_ERR(ift)) {
in6_ifa_put(ifp);
in6_dev_put(idev);
@@ -2035,7 +2036,7 @@ void addrconf_dad_failure(struct inet6_ifaddr *ifp)
ifp2 = ipv6_add_addr(idev, &new_addr, NULL, pfxlen,
scope, flags, valid_lft,
- preferred_lft, false);
+ preferred_lft, false, NULL);
if (IS_ERR(ifp2))
goto lock_errdad;
@@ -2493,7 +2494,7 @@ int addrconf_prefix_rcv_add_addr(struct net *net, struct net_device *dev,
pinfo->prefix_len,
addr_type&IPV6_ADDR_SCOPE_MASK,
addr_flags, valid_lft,
- prefered_lft, false);
+ prefered_lft, false, NULL);
if (IS_ERR_OR_NULL(ifp))
return -1;
@@ -2803,7 +2804,8 @@ static int inet6_addr_add(struct net *net, int ifindex,
const struct in6_addr *pfx,
const struct in6_addr *peer_pfx,
unsigned int plen, __u32 ifa_flags,
- __u32 prefered_lft, __u32 valid_lft)
+ __u32 prefered_lft, __u32 valid_lft,
+ struct netlink_ext_ack *extack)
{
struct inet6_ifaddr *ifp;
struct inet6_dev *idev;
@@ -2862,7 +2864,7 @@ static int inet6_addr_add(struct net *net, int ifindex,
}
ifp = ipv6_add_addr(idev, pfx, peer_pfx, plen, scope, ifa_flags,
- valid_lft, prefered_lft, true);
+ valid_lft, prefered_lft, true, extack);
if (!IS_ERR(ifp)) {
if (!(ifa_flags & IFA_F_NOPREFIXROUTE)) {
@@ -2947,7 +2949,7 @@ int addrconf_add_ifaddr(struct net *net, void __user *arg)
rtnl_lock();
err = inet6_addr_add(net, ireq.ifr6_ifindex, &ireq.ifr6_addr, NULL,
ireq.ifr6_prefixlen, IFA_F_PERMANENT,
- INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
+ INFINITY_LIFE_TIME, INFINITY_LIFE_TIME, NULL);
rtnl_unlock();
return err;
}
@@ -2978,7 +2980,7 @@ static void add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
ifp = ipv6_add_addr(idev, addr, NULL, plen,
scope, IFA_F_PERMANENT,
INFINITY_LIFE_TIME, INFINITY_LIFE_TIME,
- true);
+ true, NULL);
if (!IS_ERR(ifp)) {
spin_lock_bh(&ifp->lock);
ifp->flags &= ~IFA_F_TENTATIVE;
@@ -3078,7 +3080,7 @@ void addrconf_add_linklocal(struct inet6_dev *idev,
#endif
ifp = ipv6_add_addr(idev, addr, NULL, 64, IFA_LINK, addr_flags,
- INFINITY_LIFE_TIME, INFINITY_LIFE_TIME, true);
+ INFINITY_LIFE_TIME, INFINITY_LIFE_TIME, true, NULL);
if (!IS_ERR(ifp)) {
addrconf_prefix_route(&ifp->addr, ifp->prefix_len, idev->dev, 0, 0);
addrconf_dad_start(ifp);
@@ -4581,7 +4583,7 @@ inet6_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh,
*/
return inet6_addr_add(net, ifm->ifa_index, pfx, peer_pfx,
ifm->ifa_prefixlen, ifa_flags,
- preferred_lft, valid_lft);
+ preferred_lft, valid_lft, extack);
}
if (nlh->nlmsg_flags & NLM_F_EXCL ||
--
2.1.4
^ permalink raw reply related
* [PATCH net-next 2/5] net: ipv6: Make inet6addr_validator a blocking notifier
From: David Ahern @ 2017-10-13 23:02 UTC (permalink / raw)
To: netdev; +Cc: jiri, idosch, kjlx, davem, yoshfuji, David Ahern
In-Reply-To: <1507935733-18950-1-git-send-email-dsahern@gmail.com>
inet6addr_validator chain was added by commit 3ad7d2468f79f ("Ipvlan
should return an error when an address is already in use") to allow
address validation before changes are committed and to be able to
fail the address change with an error back to the user. The address
validation is not done for addresses received from router
advertisements.
Handling RAs in softirq context is the only reason for the notifier
chain to be atomic versus blocking. Since the only current user, ipvlan,
of the validator chain ignores softirq context, the notifier can be made
blocking and simply not invoked for softirq path.
The blocking option is needed by spectrum for example to validate
resources for an adding an address to an interface.
Signed-off-by: David Ahern <dsahern@gmail.com>
---
drivers/net/ipvlan/ipvlan_main.c | 4 ----
net/ipv6/addrconf.c | 21 ++++++++++++++-------
net/ipv6/addrconf_core.c | 9 +++++----
3 files changed, 19 insertions(+), 15 deletions(-)
diff --git a/drivers/net/ipvlan/ipvlan_main.c b/drivers/net/ipvlan/ipvlan_main.c
index 3cf67db513e2..6842739b6679 100644
--- a/drivers/net/ipvlan/ipvlan_main.c
+++ b/drivers/net/ipvlan/ipvlan_main.c
@@ -808,10 +808,6 @@ static int ipvlan_addr6_event(struct notifier_block *unused,
struct net_device *dev = (struct net_device *)if6->idev->dev;
struct ipvl_dev *ipvlan = netdev_priv(dev);
- /* FIXME IPv6 autoconf calls us from bh without RTNL */
- if (in_softirq())
- return NOTIFY_DONE;
-
if (!netif_is_ipvlan(dev))
return NOTIFY_DONE;
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 80f5fc74f0c4..31ff12277bcf 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -993,7 +993,6 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
struct net *net = dev_net(idev->dev);
struct inet6_ifaddr *ifa = NULL;
struct rt6_info *rt = NULL;
- struct in6_validator_info i6vi;
int err = 0;
int addr_type = ipv6_addr_type(addr);
@@ -1013,12 +1012,20 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
goto out;
}
- i6vi.i6vi_addr = *addr;
- i6vi.i6vi_dev = idev;
- err = inet6addr_validator_notifier_call_chain(NETDEV_UP, &i6vi);
- err = notifier_to_errno(err);
- if (err < 0)
- goto out;
+ /* validator notifier needs to be blocking;
+ * do not call in atomic context
+ */
+ if (can_block) {
+ struct in6_validator_info i6vi = {
+ .i6vi_addr = *addr,
+ .i6vi_dev = idev,
+ };
+
+ err = inet6addr_validator_notifier_call_chain(NETDEV_UP, &i6vi);
+ err = notifier_to_errno(err);
+ if (err < 0)
+ goto out;
+ }
ifa = kzalloc(sizeof(*ifa), gfp_flags);
if (!ifa) {
diff --git a/net/ipv6/addrconf_core.c b/net/ipv6/addrconf_core.c
index 9e3488d50b15..32b564dfd02a 100644
--- a/net/ipv6/addrconf_core.c
+++ b/net/ipv6/addrconf_core.c
@@ -88,7 +88,7 @@ int __ipv6_addr_type(const struct in6_addr *addr)
EXPORT_SYMBOL(__ipv6_addr_type);
static ATOMIC_NOTIFIER_HEAD(inet6addr_chain);
-static ATOMIC_NOTIFIER_HEAD(inet6addr_validator_chain);
+static BLOCKING_NOTIFIER_HEAD(inet6addr_validator_chain);
int register_inet6addr_notifier(struct notifier_block *nb)
{
@@ -110,19 +110,20 @@ EXPORT_SYMBOL(inet6addr_notifier_call_chain);
int register_inet6addr_validator_notifier(struct notifier_block *nb)
{
- return atomic_notifier_chain_register(&inet6addr_validator_chain, nb);
+ return blocking_notifier_chain_register(&inet6addr_validator_chain, nb);
}
EXPORT_SYMBOL(register_inet6addr_validator_notifier);
int unregister_inet6addr_validator_notifier(struct notifier_block *nb)
{
- return atomic_notifier_chain_unregister(&inet6addr_validator_chain, nb);
+ return blocking_notifier_chain_unregister(&inet6addr_validator_chain,
+ nb);
}
EXPORT_SYMBOL(unregister_inet6addr_validator_notifier);
int inet6addr_validator_notifier_call_chain(unsigned long val, void *v)
{
- return atomic_notifier_call_chain(&inet6addr_validator_chain, val, v);
+ return blocking_notifier_call_chain(&inet6addr_validator_chain, val, v);
}
EXPORT_SYMBOL(inet6addr_validator_notifier_call_chain);
--
2.1.4
^ permalink raw reply related
* [PATCH net-next 1/5] ipv6: addrconf: cleanup locking in ipv6_add_addr
From: David Ahern @ 2017-10-13 23:02 UTC (permalink / raw)
To: netdev; +Cc: jiri, idosch, kjlx, davem, yoshfuji, David Ahern
In-Reply-To: <1507935733-18950-1-git-send-email-dsahern@gmail.com>
ipv6_add_addr is called in process context with rtnl lock held
(e.g., manual config of an address) or during softirq processing
(e.g., autoconf and address from a router advertisement).
Currently, ipv6_add_addr calls rcu_read_lock_bh shortly after entry
and does not call unlock until exit, minus the call around the address
validator notifier. Similarly, addrconf_hash_lock is taken after the
validator notifier and held until exit. This forces the allocation of
inet6_ifaddr to always be atomic.
Refactor ipv6_add_addr as follows:
1. add an input boolean to discriminate the call path (process context
or softirq). This new flag controls whether the alloc can be done
with GFP_KERNEL or GFP_ATOMIC.
2. Move the rcu_read_lock_bh and unlock calls only around functions that
do rcu updates.
3. Remove the in6_dev_hold and put added by 3ad7d2468f79f ("Ipvlan should
return an error when an address is already in use."). This was done
presumably because rcu_read_unlock_bh needs to be called before calling
the validator. Since rcu_read_lock is not needed before the validator
runs revert the hold and put added by 3ad7d2468f79f and only do the
hold when setting ifp->idev.
4. move duplicate address check and insertion of new address in the global
address hash into a helper. The helper is called after an ifa is
allocated and filled in.
This allows the ifa for manually configured addresses to be done with
GFP_KERNEL and reduces the overall amount of time with rcu_read_lock held
and hash table spinlock held.
Signed-off-by: David Ahern <dsahern@gmail.com>
---
net/ipv6/addrconf.c | 97 +++++++++++++++++++++++++++++------------------------
1 file changed, 54 insertions(+), 43 deletions(-)
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 4603aa488f4f..80f5fc74f0c4 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -957,18 +957,43 @@ static u32 inet6_addr_hash(const struct in6_addr *addr)
return hash_32(ipv6_addr_hash(addr), IN6_ADDR_HSIZE_SHIFT);
}
+static int ipv6_add_addr_hash(struct net_device *dev, struct inet6_ifaddr *ifa)
+{
+ unsigned int hash;
+ int err = 0;
+
+ spin_lock(&addrconf_hash_lock);
+
+ /* Ignore adding duplicate addresses on an interface */
+ if (ipv6_chk_same_addr(dev_net(dev), &ifa->addr, dev)) {
+ ADBG("ipv6_add_addr: already assigned\n");
+ err = -EEXIST;
+ goto out;
+ }
+
+ /* Add to big hash table */
+ hash = inet6_addr_hash(&ifa->addr);
+ hlist_add_head_rcu(&ifa->addr_lst, &inet6_addr_lst[hash]);
+
+out:
+ spin_unlock(&addrconf_hash_lock);
+
+ return err;
+}
+
/* On success it returns ifp with increased reference count */
static struct inet6_ifaddr *
ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
const struct in6_addr *peer_addr, int pfxlen,
- int scope, u32 flags, u32 valid_lft, u32 prefered_lft)
+ int scope, u32 flags, u32 valid_lft, u32 prefered_lft,
+ bool can_block)
{
+ gfp_t gfp_flags = can_block ? GFP_KERNEL : GFP_ATOMIC;
struct net *net = dev_net(idev->dev);
struct inet6_ifaddr *ifa = NULL;
- struct rt6_info *rt;
+ struct rt6_info *rt = NULL;
struct in6_validator_info i6vi;
- unsigned int hash;
int err = 0;
int addr_type = ipv6_addr_type(addr);
@@ -978,42 +1003,24 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
addr_type & IPV6_ADDR_LOOPBACK))
return ERR_PTR(-EADDRNOTAVAIL);
- rcu_read_lock_bh();
-
- in6_dev_hold(idev);
-
if (idev->dead) {
err = -ENODEV; /*XXX*/
- goto out2;
+ goto out;
}
if (idev->cnf.disable_ipv6) {
err = -EACCES;
- goto out2;
+ goto out;
}
i6vi.i6vi_addr = *addr;
i6vi.i6vi_dev = idev;
- rcu_read_unlock_bh();
-
err = inet6addr_validator_notifier_call_chain(NETDEV_UP, &i6vi);
-
- rcu_read_lock_bh();
err = notifier_to_errno(err);
- if (err)
- goto out2;
-
- spin_lock(&addrconf_hash_lock);
-
- /* Ignore adding duplicate addresses on an interface */
- if (ipv6_chk_same_addr(dev_net(idev->dev), addr, idev->dev)) {
- ADBG("ipv6_add_addr: already assigned\n");
- err = -EEXIST;
+ if (err < 0)
goto out;
- }
-
- ifa = kzalloc(sizeof(struct inet6_ifaddr), GFP_ATOMIC);
+ ifa = kzalloc(sizeof(*ifa), gfp_flags);
if (!ifa) {
ADBG("ipv6_add_addr: malloc failed\n");
err = -ENOBUFS;
@@ -1053,16 +1060,21 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
ifa->rt = rt;
ifa->idev = idev;
+ in6_dev_hold(idev);
+
/* For caller */
refcount_set(&ifa->refcnt, 1);
- /* Add to big hash table */
- hash = inet6_addr_hash(addr);
+ rcu_read_lock_bh();
- hlist_add_head_rcu(&ifa->addr_lst, &inet6_addr_lst[hash]);
- spin_unlock(&addrconf_hash_lock);
+ err = ipv6_add_addr_hash(idev->dev, ifa);
+ if (err < 0) {
+ rcu_read_unlock_bh();
+ goto out;
+ }
write_lock(&idev->lock);
+
/* Add to inet6_dev unicast addr list. */
ipv6_link_dev_addr(idev, ifa);
@@ -1073,21 +1085,19 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
in6_ifa_hold(ifa);
write_unlock(&idev->lock);
-out2:
+
rcu_read_unlock_bh();
- if (likely(err == 0))
- inet6addr_notifier_call_chain(NETDEV_UP, ifa);
- else {
+ inet6addr_notifier_call_chain(NETDEV_UP, ifa);
+out:
+ if (unlikely(err < 0)) {
+ if (rt)
+ ip6_rt_put(rt);
kfree(ifa);
- in6_dev_put(idev);
ifa = ERR_PTR(err);
}
return ifa;
-out:
- spin_unlock(&addrconf_hash_lock);
- goto out2;
}
enum cleanup_prefix_rt_t {
@@ -1334,7 +1344,7 @@ static int ipv6_create_tempaddr(struct inet6_ifaddr *ifp, struct inet6_ifaddr *i
ift = ipv6_add_addr(idev, &addr, NULL, tmp_plen,
ipv6_addr_scope(&addr), addr_flags,
- tmp_valid_lft, tmp_prefered_lft);
+ tmp_valid_lft, tmp_prefered_lft, true);
if (IS_ERR(ift)) {
in6_ifa_put(ifp);
in6_dev_put(idev);
@@ -2018,7 +2028,7 @@ void addrconf_dad_failure(struct inet6_ifaddr *ifp)
ifp2 = ipv6_add_addr(idev, &new_addr, NULL, pfxlen,
scope, flags, valid_lft,
- preferred_lft);
+ preferred_lft, false);
if (IS_ERR(ifp2))
goto lock_errdad;
@@ -2476,7 +2486,7 @@ int addrconf_prefix_rcv_add_addr(struct net *net, struct net_device *dev,
pinfo->prefix_len,
addr_type&IPV6_ADDR_SCOPE_MASK,
addr_flags, valid_lft,
- prefered_lft);
+ prefered_lft, false);
if (IS_ERR_OR_NULL(ifp))
return -1;
@@ -2845,7 +2855,7 @@ static int inet6_addr_add(struct net *net, int ifindex,
}
ifp = ipv6_add_addr(idev, pfx, peer_pfx, plen, scope, ifa_flags,
- valid_lft, prefered_lft);
+ valid_lft, prefered_lft, true);
if (!IS_ERR(ifp)) {
if (!(ifa_flags & IFA_F_NOPREFIXROUTE)) {
@@ -2960,7 +2970,8 @@ static void add_addr(struct inet6_dev *idev, const struct in6_addr *addr,
ifp = ipv6_add_addr(idev, addr, NULL, plen,
scope, IFA_F_PERMANENT,
- INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
+ INFINITY_LIFE_TIME, INFINITY_LIFE_TIME,
+ true);
if (!IS_ERR(ifp)) {
spin_lock_bh(&ifp->lock);
ifp->flags &= ~IFA_F_TENTATIVE;
@@ -3060,7 +3071,7 @@ void addrconf_add_linklocal(struct inet6_dev *idev,
#endif
ifp = ipv6_add_addr(idev, addr, NULL, 64, IFA_LINK, addr_flags,
- INFINITY_LIFE_TIME, INFINITY_LIFE_TIME);
+ INFINITY_LIFE_TIME, INFINITY_LIFE_TIME, true);
if (!IS_ERR(ifp)) {
addrconf_prefix_route(&ifp->addr, ifp->prefix_len, idev->dev, 0, 0);
addrconf_dad_start(ifp);
--
2.1.4
^ permalink raw reply related
* [PATCH net-next 0/5] mlxsw: spectrum_router: Add extack messages for RIF and VRF overflow
From: David Ahern @ 2017-10-13 23:02 UTC (permalink / raw)
To: netdev; +Cc: jiri, idosch, kjlx, davem, yoshfuji, David Ahern
Currently, exceeding the number of VRF instances or the number of router
interfaces either fails with a non-intuitive EBUSY:
$ ip li set swp1s1.6 vrf vrf-1s1-6 up
RTNETLINK answers: Device or resource busy
or fails silently (IPv6) since the checks are done in a work queue. This
set adds support for the address validator notifier to spectrum which
allows ext-ack based messages to be returned on failure.
To make that happen the IPv6 version needs to be converted from atomic
to blocking (patch 1), and then support for extack needs to be added
to the notifier (patch 2). Patches 3 and 4 add the validator notifier
to spectrum and then plumb the extack argument.
With this set, VRF overflows fail with:
$ ip li set swp1s1.6 vrf vrf-1s1-6 up
Error: spectrum: Exceeded number of supported VRF.
and RIF overflows fail with:
$ ip addr add dev swp1s2.191 10.12.191.1/24
Error: spectrum: Exceeded number of supported router interfaces.
Changes since RFC
- addressed various comments from Ido
- refactored ipv6_add_addr to allow ifa's to be allocated with
GFP_KERNEL as requested by DaveM
Ido: given the changes in patch 1 and the impact to what is now
patch 2 I dropped your Reviewed-by tag from patch 2.
David Ahern (5):
ipv6: addrconf: cleanup locking in ipv6_add_addr
net: ipv6: Make inet6addr_validator a blocking notifier
net: Add extack to validator_info structs used for address notifier
mlxsw: spectrum: router: Add support for address validator notifier
mlxsw: spectrum_router: Add extack message for RIF and VRF overflow
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 15 +-
drivers/net/ethernet/mellanox/mlxsw/spectrum.h | 4 +
.../net/ethernet/mellanox/mlxsw/spectrum_router.c | 162 +++++++++++++++------
drivers/net/ipvlan/ipvlan_main.c | 14 +-
include/linux/inetdevice.h | 1 +
include/net/addrconf.h | 1 +
net/ipv4/devinet.c | 8 +-
net/ipv6/addrconf.c | 122 +++++++++-------
net/ipv6/addrconf_core.c | 9 +-
9 files changed, 228 insertions(+), 108 deletions(-)
--
2.1.4
^ permalink raw reply
* Re: [next-queue PATCH v7 4/6] net/sched: Introduce Credit Based Shaper (CBS) qdisc
From: Vinicius Costa Gomes @ 2017-10-13 22:54 UTC (permalink / raw)
To: Ivan Khoronzhuk
Cc: netdev, intel-wired-lan, jhs, xiyou.wangcong, jiri, andre.guedes,
ivan.briano, jesus.sanchez-palencia, boon.leong.ong,
richardcochran, henrik, levipearson, rodney.cummings
In-Reply-To: <20171013195924.GA19439@khorivan>
Hi,
Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org> writes:
[...]
>> +
>> +static int cbs_enqueue_soft(struct sk_buff *skb, struct Qdisc *sch)
>> +{
>> + struct cbs_sched_data *q = qdisc_priv(sch);
>> +
>> + if (sch->q.qlen == 0 && q->credits > 0) {
>> + /* We need to stop accumulating credits when there's
>> + * no packet enqueued packets and q->credits is
> no packet -> no
Ugh. Fixed.
>
>> + * positive.
>> + */
>> + q->credits = 0;
>> + q->last = ktime_get_ns();
>> + }
>> +
>> + return qdisc_enqueue_tail(skb, sch);
>> +}
>> +
[...]
>> +static struct sk_buff *cbs_dequeue_soft(struct Qdisc *sch)
>> +{
>> + struct cbs_sched_data *q = qdisc_priv(sch);
>> + s64 now = ktime_get_ns();
>> + struct sk_buff *skb;
>> + s64 credits;
>> + int len;
>> +
>> + if (q->credits < 0) {
>> + credits = timediff_to_credits(now - q->last, q->idleslope);
> Maybe be better to add small optimization by moving some calculations from data
> path, I mean, save idle_slope in bytes instead of kbit and converting it for
> every packet. Both delay_from_credits() and timediff_to_credits() is used only
> once and with idle_slope only...and both of them converting it.
>
> Same for credits_from_len() and send slope, save it in units of port_rate.
>
Done. Thanks.
Cheers,
^ permalink raw reply
* Re: [PATCH net-next] ipv6: check fn before doing FIB6_SUBTREE(fn)
From: Martin KaFai Lau @ 2017-10-13 22:49 UTC (permalink / raw)
To: Wei Wang; +Cc: David Miller, netdev, Eric Dumazet
In-Reply-To: <20171013220108.88710-1-tracywwnj@gmail.com>
On Fri, Oct 13, 2017 at 10:01:08PM +0000, Wei Wang wrote:
> From: Wei Wang <weiwan@google.com>
>
> In fib6_locate(), we need to first make sure fn is not NULL before doing
> FIB6_SUBTREE(fn) to avoid crash.
Acked-by: Martin KaFai Lau <kafai@fb.com>
^ permalink raw reply
* Re: [PATCH] Add -target to clang switch while cross compiling.
From: Abhijit Ayarekar @ 2017-10-13 22:12 UTC (permalink / raw)
To: Alexei Starovoitov; +Cc: Abhijit Ayarekar, ast, daniel, netdev, linux-kernel
In-Reply-To: <20171013221036.jekimnr2i5linzid@ast-mbp>
On Fri, Oct 13, 2017 at 03:10:38PM -0700, Alexei Starovoitov wrote:
> On Fri, Oct 13, 2017 at 12:24:06PM -0700, Abhijit Ayarekar wrote:
> > Update to llvm excludes assembly instructions.
> > llvm git revision is below
> >
> > commit 65fad7c26569 ("bpf: add inline-asm support")
> >
> > This change will be part of llvm release 6.0
> >
> > __ASM_SYSREG_H define is not required for native compile.
> > -target switch includes appropriate target specific files
> > while cross compiling
> >
> > Tested on x86 and arm64.
> >
> > Signed-off-by: Abhijit Ayarekar <abhijit.ayarekar@caviumnetworks.com>
>
> Thanks
> Acked-by: Alexei Starovoitov <ast@kernel.org>
>
Glad i could help :)
^ permalink raw reply
* Re: [PATCH] Add -target to clang switch while cross compiling.
From: Alexei Starovoitov @ 2017-10-13 22:10 UTC (permalink / raw)
To: Abhijit Ayarekar; +Cc: ast, daniel, netdev, linux-kernel
In-Reply-To: <1507922646-19260-1-git-send-email-abhijit.ayarekar@caviumnetworks.com>
On Fri, Oct 13, 2017 at 12:24:06PM -0700, Abhijit Ayarekar wrote:
> Update to llvm excludes assembly instructions.
> llvm git revision is below
>
> commit 65fad7c26569 ("bpf: add inline-asm support")
>
> This change will be part of llvm release 6.0
>
> __ASM_SYSREG_H define is not required for native compile.
> -target switch includes appropriate target specific files
> while cross compiling
>
> Tested on x86 and arm64.
>
> Signed-off-by: Abhijit Ayarekar <abhijit.ayarekar@caviumnetworks.com>
Thanks
Acked-by: Alexei Starovoitov <ast@kernel.org>
^ permalink raw reply
* Re: [Patch net-next v3] tcp: add a tracepoint for tcp retransmission
From: Alexei Starovoitov @ 2017-10-13 22:09 UTC (permalink / raw)
To: Brendan Gregg
Cc: Cong Wang, netdev, Eric Dumazet, Hannes Frederic Sowa,
Neal Cardwell
In-Reply-To: <CAE40pdcdcLw9pdDkVEPFKvCLe0FcQzsV4vU0K-3YbNr1gS0Y6g@mail.gmail.com>
On Fri, Oct 13, 2017 at 01:50:44PM -0700, Brendan Gregg wrote:
> On Fri, Oct 13, 2017 at 1:03 PM, Cong Wang <xiyou.wangcong@gmail.com> wrote:
> > We need a real-time notification for tcp retransmission
> > for monitoring.
> >
> > Of course we could use ftrace to dynamically instrument this
> > kernel function too, however we can't retrieve the connection
> > information at the same time, for example perf-tools [1] reads
> > /proc/net/tcp for socket details, which is slow when we have
> > a lots of connections.
> >
> > Therefore, this patch adds a tracepoint for __tcp_retransmit_skb()
> > and exposes src/dst IP addresses and ports of the connection.
> > This also makes it easier to integrate into perf.
> >
> > Note, I expose both IPv4 and IPv6 addresses at the same time:
> > for a IPv4 socket, v4 mapped address is used as IPv6 addresses,
> > for a IPv6 socket, LOOPBACK4_IPV6 is already filled by kernel.
> > Also, add sk and skb pointers as they are useful for BPF.
>
> Thanks, a TCP retransmit tracepoint would be great. (tcp_set_state
> would be highly useful too, which Alexei already has in his list).
>
> Should skp->__sk_common.skc_state be included in the format string, so
> we don't have to always dig it out of the skaddr? For retransmits I
> always want to know the TCP state, to determine if it is ESTABLISHED
> (packet drop) or SYN_SENT (backlog full) or something else.
let's not expose internal socket fields into tp fields.
Few people still believe that tp fields are abi, so to be safe
no such fields should be exposed.
It's trivial enough to read sk_state from bpf program
with bpf_probe_read().
> We probably need a tracepoint for tcp_send_loss_probe() (TLP) as well,
> for tracing at the same time as retransmits (like my tools do), but
> that can be added later.
hmm. why?
This single tracepoint will cover both cases of retransmits.
For the patch:
Acked-by: Alexei Starovoitov <ast@kernel.org>
I believe it will fit our use case perfectly.
^ permalink raw reply
* [PATCH net-next] ipv6: only update __use and lastusetime once per jiffy at most
From: Wei Wang @ 2017-10-13 22:08 UTC (permalink / raw)
To: David Miller, netdev
Cc: Eric Dumazet, Martin KaFai Lau, Paolo Abeni, Wei Wang
From: Wei Wang <weiwan@google.com>
In order to not dirty the cacheline too often, we try to only update
dst->__use and dst->lastusetime at most once per jiffy.
As dst->lastusetime is only used by ipv6 garbage collector, it should
be good enough time resolution.
And __use is only used in ipv6_route_seq_show() to show how many times a
dst has been used. And as __use is not atomic_t right now, it does not
show the precise number of usage times anyway. So we think it should be
OK to only update it at most once per jiffy.
According to my latest syn flood test on a machine with intel Xeon 6th
gen processor and 2 10G mlx nics bonded together, each with 8 rx queues
on 2 NUMA nodes:
With this patch, the packet process rate increases from ~3.49Mpps to
~3.75Mpps with a 7% increase rate.
Note: dst_use() is being renamed to dst_hold_and_use() to better specify
the purpose of the function.
Signed-off-by: Wei Wang <weiwan@google.com>
Acked-by: Eric Dumazet <edumazet@googl.com>
---
include/net/dst.h | 15 ++++++++-------
net/decnet/dn_route.c | 8 ++++----
2 files changed, 12 insertions(+), 11 deletions(-)
diff --git a/include/net/dst.h b/include/net/dst.h
index 204c19e25456..5047e8053d6c 100644
--- a/include/net/dst.h
+++ b/include/net/dst.h
@@ -255,17 +255,18 @@ static inline void dst_hold(struct dst_entry *dst)
WARN_ON(atomic_inc_not_zero(&dst->__refcnt) == 0);
}
-static inline void dst_use(struct dst_entry *dst, unsigned long time)
+static inline void dst_use_noref(struct dst_entry *dst, unsigned long time)
{
- dst_hold(dst);
- dst->__use++;
- dst->lastuse = time;
+ if (time != dst->lastuse) {
+ dst->__use++;
+ dst->lastuse = time;
+ }
}
-static inline void dst_use_noref(struct dst_entry *dst, unsigned long time)
+static inline void dst_hold_and_use(struct dst_entry *dst, unsigned long time)
{
- dst->__use++;
- dst->lastuse = time;
+ dst_hold(dst);
+ dst_use_noref(dst, time);
}
static inline struct dst_entry *dst_clone(struct dst_entry *dst)
diff --git a/net/decnet/dn_route.c b/net/decnet/dn_route.c
index 0bd3afd01dd2..bff5ab88cdbb 100644
--- a/net/decnet/dn_route.c
+++ b/net/decnet/dn_route.c
@@ -338,7 +338,7 @@ static int dn_insert_route(struct dn_route *rt, unsigned int hash, struct dn_rou
dn_rt_hash_table[hash].chain);
rcu_assign_pointer(dn_rt_hash_table[hash].chain, rth);
- dst_use(&rth->dst, now);
+ dst_hold_and_use(&rth->dst, now);
spin_unlock_bh(&dn_rt_hash_table[hash].lock);
dst_release_immediate(&rt->dst);
@@ -351,7 +351,7 @@ static int dn_insert_route(struct dn_route *rt, unsigned int hash, struct dn_rou
rcu_assign_pointer(rt->dst.dn_next, dn_rt_hash_table[hash].chain);
rcu_assign_pointer(dn_rt_hash_table[hash].chain, rt);
- dst_use(&rt->dst, now);
+ dst_hold_and_use(&rt->dst, now);
spin_unlock_bh(&dn_rt_hash_table[hash].lock);
*rp = rt;
return 0;
@@ -1258,7 +1258,7 @@ static int __dn_route_output_key(struct dst_entry **pprt, const struct flowidn *
(flp->flowidn_mark == rt->fld.flowidn_mark) &&
dn_is_output_route(rt) &&
(rt->fld.flowidn_oif == flp->flowidn_oif)) {
- dst_use(&rt->dst, jiffies);
+ dst_hold_and_use(&rt->dst, jiffies);
rcu_read_unlock_bh();
*pprt = &rt->dst;
return 0;
@@ -1535,7 +1535,7 @@ static int dn_route_input(struct sk_buff *skb)
(rt->fld.flowidn_oif == 0) &&
(rt->fld.flowidn_mark == skb->mark) &&
(rt->fld.flowidn_iif == cb->iif)) {
- dst_use(&rt->dst, jiffies);
+ dst_hold_and_use(&rt->dst, jiffies);
rcu_read_unlock();
skb_dst_set(skb, (struct dst_entry *)rt);
return 0;
--
2.15.0.rc0.271.g36b669edcc-goog
^ permalink raw reply related
* [PATCH net-next] ipv6: check fn before doing FIB6_SUBTREE(fn)
From: Wei Wang @ 2017-10-13 22:01 UTC (permalink / raw)
To: David Miller, netdev; +Cc: Eric Dumazet, Martin KaFai Lau, Wei Wang
From: Wei Wang <weiwan@google.com>
In fib6_locate(), we need to first make sure fn is not NULL before doing
FIB6_SUBTREE(fn) to avoid crash.
This fixes the following static checker warning:
net/ipv6/ip6_fib.c:1462 fib6_locate()
warn: variable dereferenced before check 'fn' (see line 1459)
net/ipv6/ip6_fib.c
1458 if (src_len) {
1459 struct fib6_node *subtree = FIB6_SUBTREE(fn);
^^^^^^^^^^^^^^^^
We shifted this dereference
1460
1461 WARN_ON(saddr == NULL);
1462 if (fn && subtree)
^^
before the check for NULL.
1463 fn = fib6_locate_1(subtree, saddr, src_len,
1464 offsetof(struct rt6_info, rt6i_src)
Fixes: 66f5d6ce53e6 ("ipv6: replace rwlock with rcu and spinlock in fib6_table")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Wei Wang <weiwan@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
---
net/ipv6/ip6_fib.c | 11 +++++++----
1 file changed, 7 insertions(+), 4 deletions(-)
diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
index c2ecd5ec638a..548af48212fc 100644
--- a/net/ipv6/ip6_fib.c
+++ b/net/ipv6/ip6_fib.c
@@ -1456,13 +1456,16 @@ struct fib6_node *fib6_locate(struct fib6_node *root,
#ifdef CONFIG_IPV6_SUBTREES
if (src_len) {
- struct fib6_node *subtree = FIB6_SUBTREE(fn);
-
WARN_ON(saddr == NULL);
- if (fn && subtree)
- fn = fib6_locate_1(subtree, saddr, src_len,
+ if (fn) {
+ struct fib6_node *subtree = FIB6_SUBTREE(fn);
+
+ if (subtree) {
+ fn = fib6_locate_1(subtree, saddr, src_len,
offsetof(struct rt6_info, rt6i_src),
exact_match);
+ }
+ }
}
#endif
--
2.15.0.rc0.271.g36b669edcc-goog
^ permalink raw reply related
* [net-next 7/9] i40e: make const array patterns static, reduces object code size
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Colin Ian King, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Colin Ian King <colin.king@canonical.com>
Don't populate const array patterns on the stack, instead make it
static. Makes the object code smaller by over 60 bytes:
Before:
text data bss dec hex filename
1953 496 0 2449 991 i40e_diag.o
After:
text data bss dec hex filename
1798 584 0 2382 94e i40e_diag.o
(gcc 6.3.0, x86-64)
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/i40e/i40e_diag.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e_diag.c b/drivers/net/ethernet/intel/i40e/i40e_diag.c
index f141e78d409e..76ed56641864 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_diag.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_diag.c
@@ -36,7 +36,9 @@
static i40e_status i40e_diag_reg_pattern_test(struct i40e_hw *hw,
u32 reg, u32 mask)
{
- const u32 patterns[] = {0x5A5A5A5A, 0xA5A5A5A5, 0x00000000, 0xFFFFFFFF};
+ static const u32 patterns[] = {
+ 0x5A5A5A5A, 0xA5A5A5A5, 0x00000000, 0xFFFFFFFF
+ };
u32 pat, val, orig_val;
int i;
--
2.14.2
^ permalink raw reply related
* [net-next 1/9] mqprio: Introduce new hardware offload mode and shaper in mqprio
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Amritha Nambiar, netdev, nhorman, sassmann, jogreene,
Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Amritha Nambiar <amritha.nambiar@intel.com>
The offload types currently supported in mqprio are 0 (no offload) and
1 (offload only TCs) by setting these values for the 'hw' option. If
offloads are supported by setting the 'hw' option to 1, the default
offload mode is 'dcb' where only the TC values are offloaded to the
device. This patch introduces a new hardware offload mode called
'channel' with 'hw' set to 1 in mqprio which makes full use of the
mqprio options, the TCs, the queue configurations and the QoS parameters
for the TCs. This is achieved through a new netlink attribute for the
'mode' option which takes values such as 'dcb' (default) and 'channel'.
The 'channel' mode also supports QoS attributes for traffic class such as
minimum and maximum values for bandwidth rate limits.
This patch enables configuring additional HW shaper attributes associated
with a traffic class. Currently the shaper for bandwidth rate limiting is
supported which takes options such as minimum and maximum bandwidth rates
and are offloaded to the hardware in the 'channel' mode. The min and max
limits for bandwidth rates are provided by the user along with the TCs
and the queue configurations when creating the mqprio qdisc. The interface
can be extended to support new HW shapers in future through the 'shaper'
attribute.
Introduces a new data structure 'tc_mqprio_qopt_offload' for offloading
mqprio queue options and use this to be shared between the kernel and
device driver. This contains a copy of the existing data structure
for mqprio queue options. This new data structure can be extended when
adding new attributes for traffic class such as mode, shaper, shaper
parameters (bandwidth rate limits). The existing data structure for mqprio
queue options will be shared between the kernel and userspace.
Example:
queues 4@0 4@4 hw 1 mode channel shaper bw_rlimit\
min_rate 1Gbit 2Gbit max_rate 4Gbit 5Gbit
To dump the bandwidth rates:
qdisc mqprio 804a: root tc 2 map 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0
queues:(0:3) (4:7)
mode:channel
shaper:bw_rlimit min_rate:1Gbit 2Gbit max_rate:4Gbit 5Gbit
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
include/net/pkt_cls.h | 9 ++
include/uapi/linux/pkt_sched.h | 32 +++++++
net/sched/sch_mqprio.c | 183 +++++++++++++++++++++++++++++++++++++++--
3 files changed, 215 insertions(+), 9 deletions(-)
diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
index f5263743076b..60d39789e4f0 100644
--- a/include/net/pkt_cls.h
+++ b/include/net/pkt_cls.h
@@ -546,6 +546,15 @@ struct tc_cls_bpf_offload {
u32 gen_flags;
};
+struct tc_mqprio_qopt_offload {
+ /* struct tc_mqprio_qopt must always be the first element */
+ struct tc_mqprio_qopt qopt;
+ u16 mode;
+ u16 shaper;
+ u32 flags;
+ u64 min_rate[TC_QOPT_MAX_QUEUE];
+ u64 max_rate[TC_QOPT_MAX_QUEUE];
+};
/* This structure holds cookie structure that is passed from user
* to the kernel for actions and classifiers
diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
index 099bf5528fed..e95b5c9b9fad 100644
--- a/include/uapi/linux/pkt_sched.h
+++ b/include/uapi/linux/pkt_sched.h
@@ -625,6 +625,22 @@ enum {
#define TC_MQPRIO_HW_OFFLOAD_MAX (__TC_MQPRIO_HW_OFFLOAD_MAX - 1)
+enum {
+ TC_MQPRIO_MODE_DCB,
+ TC_MQPRIO_MODE_CHANNEL,
+ __TC_MQPRIO_MODE_MAX
+};
+
+#define __TC_MQPRIO_MODE_MAX (__TC_MQPRIO_MODE_MAX - 1)
+
+enum {
+ TC_MQPRIO_SHAPER_DCB,
+ TC_MQPRIO_SHAPER_BW_RATE, /* Add new shapers below */
+ __TC_MQPRIO_SHAPER_MAX
+};
+
+#define __TC_MQPRIO_SHAPER_MAX (__TC_MQPRIO_SHAPER_MAX - 1)
+
struct tc_mqprio_qopt {
__u8 num_tc;
__u8 prio_tc_map[TC_QOPT_BITMASK + 1];
@@ -633,6 +649,22 @@ struct tc_mqprio_qopt {
__u16 offset[TC_QOPT_MAX_QUEUE];
};
+#define TC_MQPRIO_F_MODE 0x1
+#define TC_MQPRIO_F_SHAPER 0x2
+#define TC_MQPRIO_F_MIN_RATE 0x4
+#define TC_MQPRIO_F_MAX_RATE 0x8
+
+enum {
+ TCA_MQPRIO_UNSPEC,
+ TCA_MQPRIO_MODE,
+ TCA_MQPRIO_SHAPER,
+ TCA_MQPRIO_MIN_RATE64,
+ TCA_MQPRIO_MAX_RATE64,
+ __TCA_MQPRIO_MAX,
+};
+
+#define TCA_MQPRIO_MAX (__TCA_MQPRIO_MAX - 1)
+
/* SFB */
enum {
diff --git a/net/sched/sch_mqprio.c b/net/sched/sch_mqprio.c
index 6bcdfe6e7b63..f1ae9be83934 100644
--- a/net/sched/sch_mqprio.c
+++ b/net/sched/sch_mqprio.c
@@ -18,10 +18,16 @@
#include <net/netlink.h>
#include <net/pkt_sched.h>
#include <net/sch_generic.h>
+#include <net/pkt_cls.h>
struct mqprio_sched {
struct Qdisc **qdiscs;
+ u16 mode;
+ u16 shaper;
int hw_offload;
+ u32 flags;
+ u64 min_rate[TC_QOPT_MAX_QUEUE];
+ u64 max_rate[TC_QOPT_MAX_QUEUE];
};
static void mqprio_destroy(struct Qdisc *sch)
@@ -39,9 +45,17 @@ static void mqprio_destroy(struct Qdisc *sch)
}
if (priv->hw_offload && dev->netdev_ops->ndo_setup_tc) {
- struct tc_mqprio_qopt mqprio = {};
+ struct tc_mqprio_qopt_offload mqprio = { { 0 } };
- dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_MQPRIO, &mqprio);
+ switch (priv->mode) {
+ case TC_MQPRIO_MODE_DCB:
+ case TC_MQPRIO_MODE_CHANNEL:
+ dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_MQPRIO,
+ &mqprio);
+ break;
+ default:
+ return;
+ }
} else {
netdev_set_num_tc(dev, 0);
}
@@ -97,6 +111,26 @@ static int mqprio_parse_opt(struct net_device *dev, struct tc_mqprio_qopt *qopt)
return 0;
}
+static const struct nla_policy mqprio_policy[TCA_MQPRIO_MAX + 1] = {
+ [TCA_MQPRIO_MODE] = { .len = sizeof(u16) },
+ [TCA_MQPRIO_SHAPER] = { .len = sizeof(u16) },
+ [TCA_MQPRIO_MIN_RATE64] = { .type = NLA_NESTED },
+ [TCA_MQPRIO_MAX_RATE64] = { .type = NLA_NESTED },
+};
+
+static int parse_attr(struct nlattr *tb[], int maxtype, struct nlattr *nla,
+ const struct nla_policy *policy, int len)
+{
+ int nested_len = nla_len(nla) - NLA_ALIGN(len);
+
+ if (nested_len >= nla_attr_size(0))
+ return nla_parse(tb, maxtype, nla_data(nla) + NLA_ALIGN(len),
+ nested_len, policy, NULL);
+
+ memset(tb, 0, sizeof(struct nlattr *) * (maxtype + 1));
+ return 0;
+}
+
static int mqprio_init(struct Qdisc *sch, struct nlattr *opt)
{
struct net_device *dev = qdisc_dev(sch);
@@ -105,6 +139,10 @@ static int mqprio_init(struct Qdisc *sch, struct nlattr *opt)
struct Qdisc *qdisc;
int i, err = -EOPNOTSUPP;
struct tc_mqprio_qopt *qopt = NULL;
+ struct nlattr *tb[TCA_MQPRIO_MAX + 1];
+ struct nlattr *attr;
+ int rem;
+ int len = nla_len(opt) - NLA_ALIGN(sizeof(*qopt));
BUILD_BUG_ON(TC_MAX_QUEUE != TC_QOPT_MAX_QUEUE);
BUILD_BUG_ON(TC_BITMASK != TC_QOPT_BITMASK);
@@ -122,6 +160,58 @@ static int mqprio_init(struct Qdisc *sch, struct nlattr *opt)
if (mqprio_parse_opt(dev, qopt))
return -EINVAL;
+ if (len > 0) {
+ err = parse_attr(tb, TCA_MQPRIO_MAX, opt, mqprio_policy,
+ sizeof(*qopt));
+ if (err < 0)
+ return err;
+
+ if (!qopt->hw)
+ return -EINVAL;
+
+ if (tb[TCA_MQPRIO_MODE]) {
+ priv->flags |= TC_MQPRIO_F_MODE;
+ priv->mode = *(u16 *)nla_data(tb[TCA_MQPRIO_MODE]);
+ }
+
+ if (tb[TCA_MQPRIO_SHAPER]) {
+ priv->flags |= TC_MQPRIO_F_SHAPER;
+ priv->shaper = *(u16 *)nla_data(tb[TCA_MQPRIO_SHAPER]);
+ }
+
+ if (tb[TCA_MQPRIO_MIN_RATE64]) {
+ if (priv->shaper != TC_MQPRIO_SHAPER_BW_RATE)
+ return -EINVAL;
+ i = 0;
+ nla_for_each_nested(attr, tb[TCA_MQPRIO_MIN_RATE64],
+ rem) {
+ if (nla_type(attr) != TCA_MQPRIO_MIN_RATE64)
+ return -EINVAL;
+ if (i >= qopt->num_tc)
+ break;
+ priv->min_rate[i] = *(u64 *)nla_data(attr);
+ i++;
+ }
+ priv->flags |= TC_MQPRIO_F_MIN_RATE;
+ }
+
+ if (tb[TCA_MQPRIO_MAX_RATE64]) {
+ if (priv->shaper != TC_MQPRIO_SHAPER_BW_RATE)
+ return -EINVAL;
+ i = 0;
+ nla_for_each_nested(attr, tb[TCA_MQPRIO_MAX_RATE64],
+ rem) {
+ if (nla_type(attr) != TCA_MQPRIO_MAX_RATE64)
+ return -EINVAL;
+ if (i >= qopt->num_tc)
+ break;
+ priv->max_rate[i] = *(u64 *)nla_data(attr);
+ i++;
+ }
+ priv->flags |= TC_MQPRIO_F_MAX_RATE;
+ }
+ }
+
/* pre-allocate qdisc, attachment can't fail */
priv->qdiscs = kcalloc(dev->num_tx_queues, sizeof(priv->qdiscs[0]),
GFP_KERNEL);
@@ -146,14 +236,36 @@ static int mqprio_init(struct Qdisc *sch, struct nlattr *opt)
* supplied and verified mapping
*/
if (qopt->hw) {
- struct tc_mqprio_qopt mqprio = *qopt;
+ struct tc_mqprio_qopt_offload mqprio = {.qopt = *qopt};
- err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_MQPRIO,
+ switch (priv->mode) {
+ case TC_MQPRIO_MODE_DCB:
+ if (priv->shaper != TC_MQPRIO_SHAPER_DCB)
+ return -EINVAL;
+ break;
+ case TC_MQPRIO_MODE_CHANNEL:
+ mqprio.flags = priv->flags;
+ if (priv->flags & TC_MQPRIO_F_MODE)
+ mqprio.mode = priv->mode;
+ if (priv->flags & TC_MQPRIO_F_SHAPER)
+ mqprio.shaper = priv->shaper;
+ if (priv->flags & TC_MQPRIO_F_MIN_RATE)
+ for (i = 0; i < mqprio.qopt.num_tc; i++)
+ mqprio.min_rate[i] = priv->min_rate[i];
+ if (priv->flags & TC_MQPRIO_F_MAX_RATE)
+ for (i = 0; i < mqprio.qopt.num_tc; i++)
+ mqprio.max_rate[i] = priv->max_rate[i];
+ break;
+ default:
+ return -EINVAL;
+ }
+ err = dev->netdev_ops->ndo_setup_tc(dev,
+ TC_SETUP_MQPRIO,
&mqprio);
if (err)
return err;
- priv->hw_offload = mqprio.hw;
+ priv->hw_offload = mqprio.qopt.hw;
} else {
netdev_set_num_tc(dev, qopt->num_tc);
for (i = 0; i < qopt->num_tc; i++)
@@ -223,11 +335,51 @@ static int mqprio_graft(struct Qdisc *sch, unsigned long cl, struct Qdisc *new,
return 0;
}
+static int dump_rates(struct mqprio_sched *priv,
+ struct tc_mqprio_qopt *opt, struct sk_buff *skb)
+{
+ struct nlattr *nest;
+ int i;
+
+ if (priv->flags & TC_MQPRIO_F_MIN_RATE) {
+ nest = nla_nest_start(skb, TCA_MQPRIO_MIN_RATE64);
+ if (!nest)
+ goto nla_put_failure;
+
+ for (i = 0; i < opt->num_tc; i++) {
+ if (nla_put(skb, TCA_MQPRIO_MIN_RATE64,
+ sizeof(priv->min_rate[i]),
+ &priv->min_rate[i]))
+ goto nla_put_failure;
+ }
+ nla_nest_end(skb, nest);
+ }
+
+ if (priv->flags & TC_MQPRIO_F_MAX_RATE) {
+ nest = nla_nest_start(skb, TCA_MQPRIO_MAX_RATE64);
+ if (!nest)
+ goto nla_put_failure;
+
+ for (i = 0; i < opt->num_tc; i++) {
+ if (nla_put(skb, TCA_MQPRIO_MAX_RATE64,
+ sizeof(priv->max_rate[i]),
+ &priv->max_rate[i]))
+ goto nla_put_failure;
+ }
+ nla_nest_end(skb, nest);
+ }
+ return 0;
+
+nla_put_failure:
+ nla_nest_cancel(skb, nest);
+ return -1;
+}
+
static int mqprio_dump(struct Qdisc *sch, struct sk_buff *skb)
{
struct net_device *dev = qdisc_dev(sch);
struct mqprio_sched *priv = qdisc_priv(sch);
- unsigned char *b = skb_tail_pointer(skb);
+ struct nlattr *nla = (struct nlattr *)skb_tail_pointer(skb);
struct tc_mqprio_qopt opt = { 0 };
struct Qdisc *qdisc;
unsigned int i;
@@ -258,12 +410,25 @@ static int mqprio_dump(struct Qdisc *sch, struct sk_buff *skb)
opt.offset[i] = dev->tc_to_txq[i].offset;
}
- if (nla_put(skb, TCA_OPTIONS, sizeof(opt), &opt))
+ if (nla_put(skb, TCA_OPTIONS, NLA_ALIGN(sizeof(opt)), &opt))
+ goto nla_put_failure;
+
+ if ((priv->flags & TC_MQPRIO_F_MODE) &&
+ nla_put_u16(skb, TCA_MQPRIO_MODE, priv->mode))
+ goto nla_put_failure;
+
+ if ((priv->flags & TC_MQPRIO_F_SHAPER) &&
+ nla_put_u16(skb, TCA_MQPRIO_SHAPER, priv->shaper))
+ goto nla_put_failure;
+
+ if ((priv->flags & TC_MQPRIO_F_MIN_RATE ||
+ priv->flags & TC_MQPRIO_F_MAX_RATE) &&
+ (dump_rates(priv, &opt, skb) != 0))
goto nla_put_failure;
- return skb->len;
+ return nla_nest_end(skb, nla);
nla_put_failure:
- nlmsg_trim(skb, b);
+ nlmsg_trim(skb, nla);
return -1;
}
--
2.14.2
^ permalink raw reply related
* [net-next 2/9] i40e: Add macro for PF reset bit
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Amritha Nambiar, netdev, nhorman, sassmann, jogreene,
Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Amritha Nambiar <amritha.nambiar@intel.com>
Introduce a macro for the bit setting the PF reset flag and
update its usages. This makes it easier to use this flag
in functions to be introduced in future without encountering
checkpatch issues related to alignment and line over 80
characters.
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/i40e/i40e.h | 2 ++
drivers/net/ethernet/intel/i40e/i40e_debugfs.c | 3 +--
drivers/net/ethernet/intel/i40e/i40e_main.c | 9 ++++-----
drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 5 ++---
4 files changed, 9 insertions(+), 10 deletions(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e.h b/drivers/net/ethernet/intel/i40e/i40e.h
index 8139b4ee1dc3..e7c7a853cf7f 100644
--- a/drivers/net/ethernet/intel/i40e/i40e.h
+++ b/drivers/net/ethernet/intel/i40e/i40e.h
@@ -157,6 +157,8 @@ enum i40e_state_t {
__I40E_STATE_SIZE__,
};
+#define I40E_PF_RESET_FLAG BIT_ULL(__I40E_PF_RESET_REQUESTED)
+
/* VSI state flags */
enum i40e_vsi_state_t {
__I40E_VSI_DOWN,
diff --git a/drivers/net/ethernet/intel/i40e/i40e_debugfs.c b/drivers/net/ethernet/intel/i40e/i40e_debugfs.c
index 6f2725fc50a1..2b8bbc84e34f 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_debugfs.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_debugfs.c
@@ -798,8 +798,7 @@ static ssize_t i40e_dbg_command_write(struct file *filp,
*/
if (!(pf->flags & I40E_FLAG_VEB_MODE_ENABLED)) {
pf->flags |= I40E_FLAG_VEB_MODE_ENABLED;
- i40e_do_reset_safe(pf,
- BIT_ULL(__I40E_PF_RESET_REQUESTED));
+ i40e_do_reset_safe(pf, I40E_PF_RESET_FLAG);
}
vsi = i40e_vsi_setup(pf, I40E_VSI_VMDQ2, vsi_seid, 0);
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index 4de52001a2b9..6190257eecfe 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -5747,7 +5747,7 @@ int i40e_vsi_open(struct i40e_vsi *vsi)
err_setup_tx:
i40e_vsi_free_tx_resources(vsi);
if (vsi == pf->vsi[pf->lan_vsi])
- i40e_do_reset(pf, BIT_ULL(__I40E_PF_RESET_REQUESTED), true);
+ i40e_do_reset(pf, I40E_PF_RESET_FLAG, true);
return err;
}
@@ -5875,7 +5875,7 @@ void i40e_do_reset(struct i40e_pf *pf, u32 reset_flags, bool lock_acquired)
wr32(&pf->hw, I40E_GLGEN_RTRIG, val);
i40e_flush(&pf->hw);
- } else if (reset_flags & BIT_ULL(__I40E_PF_RESET_REQUESTED)) {
+ } else if (reset_flags & I40E_PF_RESET_FLAG) {
/* Request a PF Reset
*
@@ -9223,7 +9223,7 @@ static int i40e_set_features(struct net_device *netdev,
need_reset = i40e_set_ntuple(pf, features);
if (need_reset)
- i40e_do_reset(pf, BIT_ULL(__I40E_PF_RESET_REQUESTED), true);
+ i40e_do_reset(pf, I40E_PF_RESET_FLAG, true);
return 0;
}
@@ -9475,8 +9475,7 @@ static int i40e_ndo_bridge_setlink(struct net_device *dev,
pf->flags |= I40E_FLAG_VEB_MODE_ENABLED;
else
pf->flags &= ~I40E_FLAG_VEB_MODE_ENABLED;
- i40e_do_reset(pf, BIT_ULL(__I40E_PF_RESET_REQUESTED),
- true);
+ i40e_do_reset(pf, I40E_PF_RESET_FLAG, true);
break;
}
}
diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
index 0c4fa225c7be..e7f98e306554 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
@@ -1425,8 +1425,7 @@ int i40e_pci_sriov_configure(struct pci_dev *pdev, int num_vfs)
if (num_vfs) {
if (!(pf->flags & I40E_FLAG_VEB_MODE_ENABLED)) {
pf->flags |= I40E_FLAG_VEB_MODE_ENABLED;
- i40e_do_reset_safe(pf,
- BIT_ULL(__I40E_PF_RESET_REQUESTED));
+ i40e_do_reset_safe(pf, I40E_PF_RESET_FLAG);
}
return i40e_pci_sriov_enable(pdev, num_vfs);
}
@@ -1434,7 +1433,7 @@ int i40e_pci_sriov_configure(struct pci_dev *pdev, int num_vfs)
if (!pci_vfs_assigned(pf->pdev)) {
i40e_free_vfs(pf);
pf->flags &= ~I40E_FLAG_VEB_MODE_ENABLED;
- i40e_do_reset_safe(pf, BIT_ULL(__I40E_PF_RESET_REQUESTED));
+ i40e_do_reset_safe(pf, I40E_PF_RESET_FLAG);
} else {
dev_warn(&pdev->dev, "Unable to free VFs because some are assigned to VMs.\n");
return -EINVAL;
--
2.14.2
^ permalink raw reply related
* [net-next 6/9] i40e: Add support setting TC max bandwidth rates
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Amritha Nambiar, netdev, nhorman, sassmann, jogreene,
Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Amritha Nambiar <amritha.nambiar@intel.com>
This patch enables setting up maximum Tx rates for the traffic
classes in i40e. The maximum rate is offloaded to the hardware through
the mqprio framework by specifying the mode option as 'channel' and
shaper option as 'bw_rlimit' and is configured for the VSI. Configuring
minimum Tx rate limit is not supported in the device. The minimum
usable value for Tx rate is 50Mbps.
Example:
# tc qdisc add dev eth0 root mqprio num_tc 2 map 0 0 0 0 1 1 1 1\
queues 4@0 4@4 hw 1 mode channel shaper bw_rlimit\
max_rate 4Gbit 5Gbit
To dump the bandwidth rates:
# tc qdisc show dev eth0
qdisc mqprio 804a: root tc 2 map 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0
queues:(0:3) (4:7)
mode:channel
shaper:bw_rlimit max_rate:4Gbit 5Gbit
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/i40e/i40e.h | 2 +
drivers/net/ethernet/intel/i40e/i40e_main.c | 100 +++++++++++++++++++++++++---
2 files changed, 93 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e.h b/drivers/net/ethernet/intel/i40e/i40e.h
index 524aa06a9e0e..266e1dc5e786 100644
--- a/drivers/net/ethernet/intel/i40e/i40e.h
+++ b/drivers/net/ethernet/intel/i40e/i40e.h
@@ -359,6 +359,8 @@ struct i40e_channel {
u8 enabled_tc;
struct i40e_aqc_vsi_properties_data info;
+ u64 max_tx_rate;
+
/* track this channel belongs to which VSI */
struct i40e_vsi *parent_vsi;
};
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index fc6eaf44d87c..bb31d53c4923 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -5196,9 +5196,16 @@ static int i40e_vsi_configure_bw_alloc(struct i40e_vsi *vsi, u8 enabled_tc,
i40e_status ret;
int i;
- if ((vsi->back->flags & I40E_FLAG_TC_MQPRIO) ||
- !vsi->mqprio_qopt.qopt.hw)
+ if (vsi->back->flags & I40E_FLAG_TC_MQPRIO)
return 0;
+ if (!vsi->mqprio_qopt.qopt.hw) {
+ ret = i40e_set_bw_limit(vsi, vsi->seid, 0);
+ if (ret)
+ dev_info(&vsi->back->pdev->dev,
+ "Failed to reset tx rate for vsi->seid %u\n",
+ vsi->seid);
+ return ret;
+ }
bw_data.tc_valid_bits = enabled_tc;
for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++)
bw_data.tc_bw_credits[i] = bw_share[i];
@@ -5505,6 +5512,13 @@ static void i40e_remove_queue_channels(struct i40e_vsi *vsi)
rx_ring->ch = NULL;
}
+ /* Reset BW configured for this VSI via mqprio */
+ ret = i40e_set_bw_limit(vsi, ch->seid, 0);
+ if (ret)
+ dev_info(&vsi->back->pdev->dev,
+ "Failed to reset tx rate for ch->seid %u\n",
+ ch->seid);
+
/* delete VSI from FW */
ret = i40e_aq_delete_element(&vsi->back->hw, ch->seid,
NULL);
@@ -6047,6 +6061,17 @@ int i40e_create_queue_channel(struct i40e_vsi *vsi,
"Setup channel (id:%u) utilizing num_queues %d\n",
ch->seid, ch->num_queue_pairs);
+ /* configure VSI for BW limit */
+ if (ch->max_tx_rate) {
+ if (i40e_set_bw_limit(vsi, ch->seid, ch->max_tx_rate))
+ return -EINVAL;
+
+ dev_dbg(&pf->pdev->dev,
+ "Set tx rate of %llu Mbps (count of 50Mbps %llu) for vsi->seid %u\n",
+ ch->max_tx_rate,
+ ch->max_tx_rate / I40E_BW_CREDIT_DIVISOR, ch->seid);
+ }
+
/* in case of VF, this will be main SRIOV VSI */
ch->parent_vsi = vsi;
@@ -6082,6 +6107,12 @@ static int i40e_configure_queue_channels(struct i40e_vsi *vsi)
ch->base_queue =
vsi->tc_config.tc_info[i].qoffset;
+ /* Bandwidth limit through tc interface is in bytes/s,
+ * change to Mbit/s
+ */
+ ch->max_tx_rate =
+ vsi->mqprio_qopt.max_rate[i] / (1000000 / 8);
+
list_add_tail(&ch->list, &vsi->ch_list);
ret = i40e_create_queue_channel(vsi, ch);
@@ -6508,6 +6539,7 @@ void i40e_down(struct i40e_vsi *vsi)
static int i40e_validate_mqprio_qopt(struct i40e_vsi *vsi,
struct tc_mqprio_qopt_offload *mqprio_qopt)
{
+ u64 sum_max_rate = 0;
int i;
if (mqprio_qopt->qopt.offset[0] != 0 ||
@@ -6517,8 +6549,13 @@ static int i40e_validate_mqprio_qopt(struct i40e_vsi *vsi,
for (i = 0; ; i++) {
if (!mqprio_qopt->qopt.count[i])
return -EINVAL;
- if (mqprio_qopt->min_rate[i] || mqprio_qopt->max_rate[i])
+ if (mqprio_qopt->min_rate[i]) {
+ dev_err(&vsi->back->pdev->dev,
+ "Invalid min tx rate (greater than 0) specified\n");
return -EINVAL;
+ }
+ sum_max_rate += (mqprio_qopt->max_rate[i] / (1000000 / 8));
+
if (i >= mqprio_qopt->qopt.num_tc - 1)
break;
if (mqprio_qopt->qopt.offset[i + 1] !=
@@ -6529,6 +6566,11 @@ static int i40e_validate_mqprio_qopt(struct i40e_vsi *vsi,
(mqprio_qopt->qopt.offset[i] + mqprio_qopt->qopt.count[i])) {
return -EINVAL;
}
+ if (sum_max_rate > i40e_get_link_speed(vsi)) {
+ dev_err(&vsi->back->pdev->dev,
+ "Invalid max tx rate specified\n");
+ return -EINVAL;
+ }
return 0;
}
@@ -6655,6 +6697,21 @@ static int i40e_setup_tc(struct net_device *netdev, void *type_data)
}
if (pf->flags & I40E_FLAG_TC_MQPRIO) {
+ if (vsi->mqprio_qopt.max_rate[0]) {
+ u64 max_tx_rate = vsi->mqprio_qopt.max_rate[0] /
+ (1000000 / 8);
+ ret = i40e_set_bw_limit(vsi, vsi->seid, max_tx_rate);
+ if (!ret) {
+ dev_dbg(&vsi->back->pdev->dev,
+ "Set tx rate of %llu Mbps (count of 50Mbps %llu) for vsi->seid %u\n",
+ max_tx_rate,
+ max_tx_rate / I40E_BW_CREDIT_DIVISOR,
+ vsi->seid);
+ } else {
+ need_reset = true;
+ goto exit;
+ }
+ }
ret = i40e_configure_queue_channels(vsi);
if (ret) {
netdev_info(netdev,
@@ -8108,6 +8165,17 @@ static int i40e_rebuild_channels(struct i40e_vsi *vsi)
vsi->uplink_seid);
return ret;
}
+ if (ch->max_tx_rate) {
+ if (i40e_set_bw_limit(vsi, ch->seid,
+ ch->max_tx_rate))
+ return -EINVAL;
+
+ dev_dbg(&vsi->back->pdev->dev,
+ "Set tx rate of %llu Mbps (count of 50Mbps %llu) for vsi->seid %u\n",
+ ch->max_tx_rate,
+ ch->max_tx_rate / I40E_BW_CREDIT_DIVISOR,
+ ch->seid);
+ }
}
return 0;
}
@@ -8248,6 +8316,7 @@ static int i40e_reset(struct i40e_pf *pf)
**/
static void i40e_rebuild(struct i40e_pf *pf, bool reinit, bool lock_acquired)
{
+ struct i40e_vsi *vsi = pf->vsi[pf->lan_vsi];
struct i40e_hw *hw = &pf->hw;
u8 set_fc_aq_fail = 0;
i40e_status ret;
@@ -8330,7 +8399,7 @@ static void i40e_rebuild(struct i40e_pf *pf, bool reinit, bool lock_acquired)
* If there were VEBs but the reconstitution failed, we'll try
* try to recover minimal use by getting the basic PF VSI working.
*/
- if (pf->vsi[pf->lan_vsi]->uplink_seid != pf->mac_seid) {
+ if (vsi->uplink_seid != pf->mac_seid) {
dev_dbg(&pf->pdev->dev, "attempting to rebuild switch\n");
/* find the one VEB connected to the MAC, and find orphans */
for (v = 0; v < I40E_MAX_VEB; v++) {
@@ -8354,8 +8423,7 @@ static void i40e_rebuild(struct i40e_pf *pf, bool reinit, bool lock_acquired)
dev_info(&pf->pdev->dev,
"rebuild of switch failed: %d, will try to set up simple PF connection\n",
ret);
- pf->vsi[pf->lan_vsi]->uplink_seid
- = pf->mac_seid;
+ vsi->uplink_seid = pf->mac_seid;
break;
} else if (pf->veb[v]->uplink_seid == 0) {
dev_info(&pf->pdev->dev,
@@ -8366,10 +8434,10 @@ static void i40e_rebuild(struct i40e_pf *pf, bool reinit, bool lock_acquired)
}
}
- if (pf->vsi[pf->lan_vsi]->uplink_seid == pf->mac_seid) {
+ if (vsi->uplink_seid == pf->mac_seid) {
dev_dbg(&pf->pdev->dev, "attempting to rebuild PF VSI\n");
/* no VEB, so rebuild only the Main VSI */
- ret = i40e_add_vsi(pf->vsi[pf->lan_vsi]);
+ ret = i40e_add_vsi(vsi);
if (ret) {
dev_info(&pf->pdev->dev,
"rebuild of Main VSI failed: %d\n", ret);
@@ -8377,10 +8445,24 @@ static void i40e_rebuild(struct i40e_pf *pf, bool reinit, bool lock_acquired)
}
}
+ if (vsi->mqprio_qopt.max_rate[0]) {
+ u64 max_tx_rate = vsi->mqprio_qopt.max_rate[0] / (1000000 / 8);
+
+ ret = i40e_set_bw_limit(vsi, vsi->seid, max_tx_rate);
+ if (!ret)
+ dev_dbg(&vsi->back->pdev->dev,
+ "Set tx rate of %llu Mbps (count of 50Mbps %llu) for vsi->seid %u\n",
+ max_tx_rate,
+ max_tx_rate / I40E_BW_CREDIT_DIVISOR,
+ vsi->seid);
+ else
+ goto end_unlock;
+ }
+
/* PF Main VSI is rebuild by now, go ahead and rebuild channel VSIs
* for this main VSI if they exist
*/
- ret = i40e_rebuild_channels(pf->vsi[pf->lan_vsi]);
+ ret = i40e_rebuild_channels(vsi);
if (ret)
goto end_unlock;
--
2.14.2
^ permalink raw reply related
* [net-next 5/9] i40e: Refactor VF BW rate limiting
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Amritha Nambiar, netdev, nhorman, sassmann, jogreene,
Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Amritha Nambiar <amritha.nambiar@intel.com>
This patch refactors the BW rate limiting for Tx traffic
on the VF to be reused in the next patch for rate limiting Tx
traffic for the VSIs on the PF as well.
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/i40e/i40e.h | 5 ++
drivers/net/ethernet/intel/i40e/i40e_main.c | 64 ++++++++++++++++++++++
drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 45 +--------------
3 files changed, 71 insertions(+), 43 deletions(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e.h b/drivers/net/ethernet/intel/i40e/i40e.h
index 024c88474951..524aa06a9e0e 100644
--- a/drivers/net/ethernet/intel/i40e/i40e.h
+++ b/drivers/net/ethernet/intel/i40e/i40e.h
@@ -128,6 +128,10 @@
/* default to trying for four seconds */
#define I40E_TRY_LINK_TIMEOUT (4 * HZ)
+/* BW rate limiting */
+#define I40E_BW_CREDIT_DIVISOR 50 /* 50Mbps per BW credit */
+#define I40E_MAX_BW_INACTIVE_ACCUM 4 /* accumulate 4 credits max */
+
/* driver state flags */
enum i40e_state_t {
__I40E_TESTING,
@@ -1039,4 +1043,5 @@ static inline bool i40e_enabled_xdp_vsi(struct i40e_vsi *vsi)
}
int i40e_create_queue_channel(struct i40e_vsi *vsi, struct i40e_channel *ch);
+int i40e_set_bw_limit(struct i40e_vsi *vsi, u16 seid, u64 max_tx_rate);
#endif /* _I40E_H_ */
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index e803aa1552c6..fc6eaf44d87c 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -5399,6 +5399,70 @@ static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
return ret;
}
+/**
+ * i40e_get_link_speed - Returns link speed for the interface
+ * @vsi: VSI to be configured
+ *
+ **/
+int i40e_get_link_speed(struct i40e_vsi *vsi)
+{
+ struct i40e_pf *pf = vsi->back;
+
+ switch (pf->hw.phy.link_info.link_speed) {
+ case I40E_LINK_SPEED_40GB:
+ return 40000;
+ case I40E_LINK_SPEED_25GB:
+ return 25000;
+ case I40E_LINK_SPEED_20GB:
+ return 20000;
+ case I40E_LINK_SPEED_10GB:
+ return 10000;
+ case I40E_LINK_SPEED_1GB:
+ return 1000;
+ default:
+ return -EINVAL;
+ }
+}
+
+/**
+ * i40e_set_bw_limit - setup BW limit for Tx traffic based on max_tx_rate
+ * @vsi: VSI to be configured
+ * @seid: seid of the channel/VSI
+ * @max_tx_rate: max TX rate to be configured as BW limit
+ *
+ * Helper function to set BW limit for a given VSI
+ **/
+int i40e_set_bw_limit(struct i40e_vsi *vsi, u16 seid, u64 max_tx_rate)
+{
+ struct i40e_pf *pf = vsi->back;
+ int speed = 0;
+ int ret = 0;
+
+ speed = i40e_get_link_speed(vsi);
+ if (max_tx_rate > speed) {
+ dev_err(&pf->pdev->dev,
+ "Invalid max tx rate %llu specified for VSI seid %d.",
+ max_tx_rate, seid);
+ return -EINVAL;
+ }
+ if (max_tx_rate && max_tx_rate < 50) {
+ dev_warn(&pf->pdev->dev,
+ "Setting max tx rate to minimum usable value of 50Mbps.\n");
+ max_tx_rate = 50;
+ }
+
+ /* Tx rate credits are in values of 50Mbps, 0 is disabled */
+ ret = i40e_aq_config_vsi_bw_limit(&pf->hw, seid,
+ max_tx_rate / I40E_BW_CREDIT_DIVISOR,
+ I40E_MAX_BW_INACTIVE_ACCUM, NULL);
+ if (ret)
+ dev_err(&pf->pdev->dev,
+ "Failed set tx rate (%llu Mbps) for vsi->seid %u, err %s aq_err %s\n",
+ max_tx_rate, seid, i40e_stat_str(&pf->hw, ret),
+ i40e_aq_str(&pf->hw, pf->hw.aq.asq_last_status));
+ return ret;
+}
+
/**
* i40e_remove_queue_channels - Remove queue channels for the TCs
* @vsi: VSI to be configured
diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
index e7f98e306554..ce0981e2f605 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
@@ -3117,8 +3117,6 @@ int i40e_ndo_set_vf_port_vlan(struct net_device *netdev, int vf_id,
return ret;
}
-#define I40E_BW_CREDIT_DIVISOR 50 /* 50Mbps per BW credit */
-#define I40E_MAX_BW_INACTIVE_ACCUM 4 /* device can accumulate 4 credits max */
/**
* i40e_ndo_set_vf_bw
* @netdev: network interface device structure
@@ -3134,7 +3132,6 @@ int i40e_ndo_set_vf_bw(struct net_device *netdev, int vf_id, int min_tx_rate,
struct i40e_pf *pf = np->vsi->back;
struct i40e_vsi *vsi;
struct i40e_vf *vf;
- int speed = 0;
int ret = 0;
/* validate the request */
@@ -3159,48 +3156,10 @@ int i40e_ndo_set_vf_bw(struct net_device *netdev, int vf_id, int min_tx_rate,
goto error;
}
- switch (pf->hw.phy.link_info.link_speed) {
- case I40E_LINK_SPEED_40GB:
- speed = 40000;
- break;
- case I40E_LINK_SPEED_25GB:
- speed = 25000;
- break;
- case I40E_LINK_SPEED_20GB:
- speed = 20000;
- break;
- case I40E_LINK_SPEED_10GB:
- speed = 10000;
- break;
- case I40E_LINK_SPEED_1GB:
- speed = 1000;
- break;
- default:
- break;
- }
-
- if (max_tx_rate > speed) {
- dev_err(&pf->pdev->dev, "Invalid max tx rate %d specified for VF %d.\n",
- max_tx_rate, vf->vf_id);
- ret = -EINVAL;
+ ret = i40e_set_bw_limit(vsi, vsi->seid, max_tx_rate);
+ if (ret)
goto error;
- }
- if ((max_tx_rate < 50) && (max_tx_rate > 0)) {
- dev_warn(&pf->pdev->dev, "Setting max Tx rate to minimum usable value of 50Mbps.\n");
- max_tx_rate = 50;
- }
-
- /* Tx rate credits are in values of 50Mbps, 0 is disabled*/
- ret = i40e_aq_config_vsi_bw_limit(&pf->hw, vsi->seid,
- max_tx_rate / I40E_BW_CREDIT_DIVISOR,
- I40E_MAX_BW_INACTIVE_ACCUM, NULL);
- if (ret) {
- dev_err(&pf->pdev->dev, "Unable to set max tx rate, error code %d.\n",
- ret);
- ret = -EIO;
- goto error;
- }
vf->tx_rate = max_tx_rate;
error:
return ret;
--
2.14.2
^ permalink raw reply related
* [net-next 9/9] i40e/i40evf: don't trust VF to reset itself
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Alan Brady, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Alan Brady <alan.brady@intel.com>
When using 'ethtool -L' on a VF to change number of requested queues
from PF, we shouldn't trust the VF to reset itself after making the
request. Doing it that way opens the door for a potentially malicious
VF to do nasty things to the PF which should never be the case.
This makes it such that after VF makes a successful request, PF will
then reset the VF to institute required changes. Only if the request
fails will PF send a message back to VF letting it know the request was
unsuccessful.
Testing-hints:
There should be no real functional changes. This is simply hardening
against a potentially malicious VF.
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 9 +++++++--
drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c | 7 +++----
include/linux/avf/virtchnl.h | 4 ++--
3 files changed, 12 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
index ce0981e2f605..f8a794b72462 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
@@ -2045,8 +2045,9 @@ static int i40e_vc_disable_queues_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
* @msglen: msg length
*
* VFs get a default number of queues but can use this message to request a
- * different number. Will respond with either the number requested or the
- * maximum we can support.
+ * different number. If the request is successful, PF will reset the VF and
+ * return 0. If unsuccessful, PF will send message informing VF of number of
+ * available queues and return result of sending VF a message.
**/
static int i40e_vc_request_queues_msg(struct i40e_vf *vf, u8 *msg, int msglen)
{
@@ -2077,7 +2078,11 @@ static int i40e_vc_request_queues_msg(struct i40e_vf *vf, u8 *msg, int msglen)
pf->queues_left);
vfres->num_queue_pairs = pf->queues_left + cur_pairs;
} else {
+ /* successful request */
vf->num_req_queues = req_pairs;
+ i40e_vc_notify_vf_reset(vf);
+ i40e_reset_vf(vf, false);
+ return 0;
}
return i40e_vc_send_msg_to_vf(vf, VIRTCHNL_OP_REQUEST_QUEUES, 0,
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
index 2bb81c39d85f..46c8b8a3907c 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
@@ -407,6 +407,7 @@ int i40evf_request_queues(struct i40evf_adapter *adapter, int num)
vfres.num_queue_pairs = num;
adapter->current_op = VIRTCHNL_OP_REQUEST_QUEUES;
+ adapter->flags |= I40EVF_FLAG_REINIT_ITR_NEEDED;
return i40evf_send_pf_msg(adapter, VIRTCHNL_OP_REQUEST_QUEUES,
(u8 *)&vfres, sizeof(vfres));
}
@@ -1098,15 +1099,13 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
case VIRTCHNL_OP_REQUEST_QUEUES: {
struct virtchnl_vf_res_request *vfres =
(struct virtchnl_vf_res_request *)msg;
- if (vfres->num_queue_pairs == adapter->num_req_queues) {
- adapter->flags |= I40EVF_FLAG_REINIT_ITR_NEEDED;
- i40evf_schedule_reset(adapter);
- } else {
+ if (vfres->num_queue_pairs != adapter->num_req_queues) {
dev_info(&adapter->pdev->dev,
"Requested %d queues, PF can support %d\n",
adapter->num_req_queues,
vfres->num_queue_pairs);
adapter->num_req_queues = 0;
+ adapter->flags &= ~I40EVF_FLAG_REINIT_ITR_NEEDED;
}
}
break;
diff --git a/include/linux/avf/virtchnl.h b/include/linux/avf/virtchnl.h
index 60e5d90cb18a..3ce61342fa31 100644
--- a/include/linux/avf/virtchnl.h
+++ b/include/linux/avf/virtchnl.h
@@ -333,8 +333,8 @@ struct virtchnl_vsi_queue_config_info {
* additional queues must be negotiated. This is a best effort request as it
* is possible the PF does not have enough queues left to support the request.
* If the PF cannot support the number requested it will respond with the
- * maximum number it is able to support; otherwise it will respond with the
- * number requested.
+ * maximum number it is able to support. If the request is successful, PF will
+ * then reset the VF to institute required changes.
*/
/* VF resource request */
--
2.14.2
^ permalink raw reply related
* [net-next 4/9] i40e: Enable 'channel' mode in mqprio for TC configs
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Amritha Nambiar, netdev, nhorman, sassmann, jogreene,
Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Amritha Nambiar <amritha.nambiar@intel.com>
The i40e driver is modified to enable the new mqprio hardware
offload mode and factor the TCs and queue configuration by
creating channel VSIs. In this mode, the priority to traffic
class mapping and the user specified queue ranges are used
to configure the traffic classes by setting the mode option to
'channel'.
Example:
map 0 0 0 0 1 2 2 3 queues 2@0 2@2 1@4 1@5\
hw 1 mode channel
qdisc mqprio 8038: root tc 4 map 0 0 0 0 1 2 2 3 0 0 0 0 0 0 0 0
queues:(0:1) (2:3) (4:4) (5:5)
mode:channel
shaper:dcb
The HW channels created are removed and all the queue configuration
is set to default when the qdisc is detached from the root of the
device.
This patch also disables setting up channels via ethtool (ethtool -L)
when the TCs are configured using mqprio scheduler.
The patch also limits setting ethtool Rx flow hash indirection
(ethtool -X eth0 equal N) to max queues configured via mqprio.
The Rx flow hash indirection input through ethtool should be
validated so that it is within in the queue range configured via
tc/mqprio. The bound checking is achieved by reporting the current
rss size to the kernel when queues are configured via mqprio.
Example:
map 0 0 0 1 0 2 3 0 queues 2@0 4@2 8@6 11@14\
hw 1 mode channel
Cannot set RX flow hash configuration: Invalid argument
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/i40e/i40e.h | 3 +
drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 8 +-
drivers/net/ethernet/intel/i40e/i40e_main.c | 457 +++++++++++++++++++------
3 files changed, 362 insertions(+), 106 deletions(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e.h b/drivers/net/ethernet/intel/i40e/i40e.h
index bde982541772..024c88474951 100644
--- a/drivers/net/ethernet/intel/i40e/i40e.h
+++ b/drivers/net/ethernet/intel/i40e/i40e.h
@@ -54,6 +54,7 @@
#include <linux/clocksource.h>
#include <linux/net_tstamp.h>
#include <linux/ptp_clock_kernel.h>
+#include <net/pkt_cls.h>
#include "i40e_type.h"
#include "i40e_prototype.h"
#include "i40e_client.h"
@@ -700,6 +701,7 @@ struct i40e_vsi {
enum i40e_vsi_type type; /* VSI type, e.g., LAN, FCoE, etc */
s16 vf_id; /* Virtual function ID for SRIOV VSIs */
+ struct tc_mqprio_qopt_offload mqprio_qopt; /* queue parameters */
struct i40e_tc_configuration tc_config;
struct i40e_aqc_vsi_properties_data info;
@@ -725,6 +727,7 @@ struct i40e_vsi {
u16 cnt_q_avail; /* num of queues available for channel usage */
u16 orig_rss_size;
u16 current_rss_size;
+ bool reconfig_rss;
u16 next_base_queue; /* next queue to be used for channel setup */
diff --git a/drivers/net/ethernet/intel/i40e/i40e_ethtool.c b/drivers/net/ethernet/intel/i40e/i40e_ethtool.c
index afd3ca8d9851..72d5f2cdf419 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_ethtool.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_ethtool.c
@@ -2652,7 +2652,7 @@ static int i40e_get_rxnfc(struct net_device *netdev, struct ethtool_rxnfc *cmd,
switch (cmd->cmd) {
case ETHTOOL_GRXRINGS:
- cmd->data = vsi->num_queue_pairs;
+ cmd->data = vsi->rss_size;
ret = 0;
break;
case ETHTOOL_GRXFH:
@@ -3897,6 +3897,12 @@ static int i40e_set_channels(struct net_device *dev,
if (vsi->type != I40E_VSI_MAIN)
return -EINVAL;
+ /* We do not support setting channels via ethtool when TCs are
+ * configured through mqprio
+ */
+ if (pf->flags & I40E_FLAG_TC_MQPRIO)
+ return -EINVAL;
+
/* verify they are not requesting separate vectors */
if (!count || ch->rx_count || ch->tx_count)
return -EINVAL;
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index e23105bee6d1..e803aa1552c6 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -1588,6 +1588,170 @@ static int i40e_set_mac(struct net_device *netdev, void *p)
return 0;
}
+/**
+ * i40e_config_rss_aq - Prepare for RSS using AQ commands
+ * @vsi: vsi structure
+ * @seed: RSS hash seed
+ **/
+static int i40e_config_rss_aq(struct i40e_vsi *vsi, const u8 *seed,
+ u8 *lut, u16 lut_size)
+{
+ struct i40e_pf *pf = vsi->back;
+ struct i40e_hw *hw = &pf->hw;
+ int ret = 0;
+
+ if (seed) {
+ struct i40e_aqc_get_set_rss_key_data *seed_dw =
+ (struct i40e_aqc_get_set_rss_key_data *)seed;
+ ret = i40e_aq_set_rss_key(hw, vsi->id, seed_dw);
+ if (ret) {
+ dev_info(&pf->pdev->dev,
+ "Cannot set RSS key, err %s aq_err %s\n",
+ i40e_stat_str(hw, ret),
+ i40e_aq_str(hw, hw->aq.asq_last_status));
+ return ret;
+ }
+ }
+ if (lut) {
+ bool pf_lut = vsi->type == I40E_VSI_MAIN ? true : false;
+
+ ret = i40e_aq_set_rss_lut(hw, vsi->id, pf_lut, lut, lut_size);
+ if (ret) {
+ dev_info(&pf->pdev->dev,
+ "Cannot set RSS lut, err %s aq_err %s\n",
+ i40e_stat_str(hw, ret),
+ i40e_aq_str(hw, hw->aq.asq_last_status));
+ return ret;
+ }
+ }
+ return ret;
+}
+
+/**
+ * i40e_vsi_config_rss - Prepare for VSI(VMDq) RSS if used
+ * @vsi: VSI structure
+ **/
+static int i40e_vsi_config_rss(struct i40e_vsi *vsi)
+{
+ struct i40e_pf *pf = vsi->back;
+ u8 seed[I40E_HKEY_ARRAY_SIZE];
+ u8 *lut;
+ int ret;
+
+ if (!(pf->hw_features & I40E_HW_RSS_AQ_CAPABLE))
+ return 0;
+ if (!vsi->rss_size)
+ vsi->rss_size = min_t(int, pf->alloc_rss_size,
+ vsi->num_queue_pairs);
+ if (!vsi->rss_size)
+ return -EINVAL;
+ lut = kzalloc(vsi->rss_table_size, GFP_KERNEL);
+ if (!lut)
+ return -ENOMEM;
+
+ /* Use the user configured hash keys and lookup table if there is one,
+ * otherwise use default
+ */
+ if (vsi->rss_lut_user)
+ memcpy(lut, vsi->rss_lut_user, vsi->rss_table_size);
+ else
+ i40e_fill_rss_lut(pf, lut, vsi->rss_table_size, vsi->rss_size);
+ if (vsi->rss_hkey_user)
+ memcpy(seed, vsi->rss_hkey_user, I40E_HKEY_ARRAY_SIZE);
+ else
+ netdev_rss_key_fill((void *)seed, I40E_HKEY_ARRAY_SIZE);
+ ret = i40e_config_rss_aq(vsi, seed, lut, vsi->rss_table_size);
+ kfree(lut);
+ return ret;
+}
+
+/**
+ * i40e_vsi_setup_queue_map_mqprio - Prepares mqprio based tc_config
+ * @vsi: the VSI being configured,
+ * @ctxt: VSI context structure
+ * @enabled_tc: number of traffic classes to enable
+ *
+ * Prepares VSI tc_config to have queue configurations based on MQPRIO options.
+ **/
+static int i40e_vsi_setup_queue_map_mqprio(struct i40e_vsi *vsi,
+ struct i40e_vsi_context *ctxt,
+ u8 enabled_tc)
+{
+ u16 qcount = 0, max_qcount, qmap, sections = 0;
+ int i, override_q, pow, num_qps, ret;
+ u8 netdev_tc = 0, offset = 0;
+
+ if (vsi->type != I40E_VSI_MAIN)
+ return -EINVAL;
+ sections = I40E_AQ_VSI_PROP_QUEUE_MAP_VALID;
+ sections |= I40E_AQ_VSI_PROP_SCHED_VALID;
+ vsi->tc_config.numtc = vsi->mqprio_qopt.qopt.num_tc;
+ vsi->tc_config.enabled_tc = enabled_tc ? enabled_tc : 1;
+ num_qps = vsi->mqprio_qopt.qopt.count[0];
+
+ /* find the next higher power-of-2 of num queue pairs */
+ pow = ilog2(num_qps);
+ if (!is_power_of_2(num_qps))
+ pow++;
+ qmap = (offset << I40E_AQ_VSI_TC_QUE_OFFSET_SHIFT) |
+ (pow << I40E_AQ_VSI_TC_QUE_NUMBER_SHIFT);
+
+ /* Setup queue offset/count for all TCs for given VSI */
+ max_qcount = vsi->mqprio_qopt.qopt.count[0];
+ for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+ /* See if the given TC is enabled for the given VSI */
+ if (vsi->tc_config.enabled_tc & BIT(i)) {
+ offset = vsi->mqprio_qopt.qopt.offset[i];
+ qcount = vsi->mqprio_qopt.qopt.count[i];
+ if (qcount > max_qcount)
+ max_qcount = qcount;
+ vsi->tc_config.tc_info[i].qoffset = offset;
+ vsi->tc_config.tc_info[i].qcount = qcount;
+ vsi->tc_config.tc_info[i].netdev_tc = netdev_tc++;
+ } else {
+ /* TC is not enabled so set the offset to
+ * default queue and allocate one queue
+ * for the given TC.
+ */
+ vsi->tc_config.tc_info[i].qoffset = 0;
+ vsi->tc_config.tc_info[i].qcount = 1;
+ vsi->tc_config.tc_info[i].netdev_tc = 0;
+ }
+ }
+
+ /* Set actual Tx/Rx queue pairs */
+ vsi->num_queue_pairs = offset + qcount;
+
+ /* Setup queue TC[0].qmap for given VSI context */
+ ctxt->info.tc_mapping[0] = cpu_to_le16(qmap);
+ ctxt->info.mapping_flags |= cpu_to_le16(I40E_AQ_VSI_QUE_MAP_CONTIG);
+ ctxt->info.queue_mapping[0] = cpu_to_le16(vsi->base_queue);
+ ctxt->info.valid_sections |= cpu_to_le16(sections);
+
+ /* Reconfigure RSS for main VSI with max queue count */
+ vsi->rss_size = max_qcount;
+ ret = i40e_vsi_config_rss(vsi);
+ if (ret) {
+ dev_info(&vsi->back->pdev->dev,
+ "Failed to reconfig rss for num_queues (%u)\n",
+ max_qcount);
+ return ret;
+ }
+ vsi->reconfig_rss = true;
+ dev_dbg(&vsi->back->pdev->dev,
+ "Reconfigured rss with num_queues (%u)\n", max_qcount);
+
+ /* Find queue count available for channel VSIs and starting offset
+ * for channel VSIs
+ */
+ override_q = vsi->mqprio_qopt.qopt.count[0];
+ if (override_q && override_q < vsi->num_queue_pairs) {
+ vsi->cnt_q_avail = vsi->num_queue_pairs - override_q;
+ vsi->next_base_queue = override_q;
+ }
+ return 0;
+}
+
/**
* i40e_vsi_setup_queue_map - Setup a VSI queue map based on enabled_tc
* @vsi: the VSI being setup
@@ -1626,7 +1790,7 @@ static void i40e_vsi_setup_queue_map(struct i40e_vsi *vsi,
numtc = 1;
}
} else {
- /* At least TC0 is enabled in case of non-DCB case */
+ /* At least TC0 is enabled in non-DCB, non-MQPRIO case */
numtc = 1;
}
@@ -3158,6 +3322,7 @@ static void i40e_vsi_config_dcb_rings(struct i40e_vsi *vsi)
rx_ring->dcb_tc = 0;
tx_ring->dcb_tc = 0;
}
+ return;
}
for (n = 0; n < I40E_MAX_TRAFFIC_CLASS; n++) {
@@ -4873,6 +5038,24 @@ static u8 i40e_dcb_get_enabled_tc(struct i40e_dcbx_config *dcbcfg)
return enabled_tc;
}
+/**
+ * i40e_mqprio_get_enabled_tc - Get enabled traffic classes
+ * @pf: PF being queried
+ *
+ * Query the current MQPRIO configuration and return the number of
+ * traffic classes enabled.
+ **/
+static u8 i40e_mqprio_get_enabled_tc(struct i40e_pf *pf)
+{
+ struct i40e_vsi *vsi = pf->vsi[pf->lan_vsi];
+ u8 num_tc = vsi->mqprio_qopt.qopt.num_tc;
+ u8 enabled_tc = 1, i;
+
+ for (i = 1; i < num_tc; i++)
+ enabled_tc |= BIT(i);
+ return enabled_tc;
+}
+
/**
* i40e_pf_get_num_tc - Get enabled traffic classes for PF
* @pf: PF being queried
@@ -4886,7 +5069,10 @@ static u8 i40e_pf_get_num_tc(struct i40e_pf *pf)
u8 num_tc = 0;
struct i40e_dcbx_config *dcbcfg = &hw->local_dcbx_config;
- /* If DCB is not enabled then always in single TC */
+ if (pf->flags & I40E_FLAG_TC_MQPRIO)
+ return pf->vsi[pf->lan_vsi]->mqprio_qopt.qopt.num_tc;
+
+ /* If neither MQPRIO nor DCB is enabled, then always use single TC */
if (!(pf->flags & I40E_FLAG_DCB_ENABLED))
return 1;
@@ -4915,7 +5101,12 @@ static u8 i40e_pf_get_num_tc(struct i40e_pf *pf)
**/
static u8 i40e_pf_get_tc_map(struct i40e_pf *pf)
{
- /* If DCB is not enabled for this PF then just return default TC */
+ if (pf->flags & I40E_FLAG_TC_MQPRIO)
+ return i40e_mqprio_get_enabled_tc(pf);
+
+ /* If neither MQPRIO nor DCB is enabled for this PF then just return
+ * default TC
+ */
if (!(pf->flags & I40E_FLAG_DCB_ENABLED))
return I40E_DEFAULT_TRAFFIC_CLASS;
@@ -5005,6 +5196,9 @@ static int i40e_vsi_configure_bw_alloc(struct i40e_vsi *vsi, u8 enabled_tc,
i40e_status ret;
int i;
+ if ((vsi->back->flags & I40E_FLAG_TC_MQPRIO) ||
+ !vsi->mqprio_qopt.qopt.hw)
+ return 0;
bw_data.tc_valid_bits = enabled_tc;
for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++)
bw_data.tc_bw_credits[i] = bw_share[i];
@@ -5067,6 +5261,9 @@ static void i40e_vsi_config_netdev_tc(struct i40e_vsi *vsi, u8 enabled_tc)
vsi->tc_config.tc_info[i].qoffset);
}
+ if (pf->flags & I40E_FLAG_TC_MQPRIO)
+ return;
+
/* Assign UP2TC map for the VSI */
for (i = 0; i < I40E_MAX_USER_PRIORITY; i++) {
/* Get the actual TC# for the UP */
@@ -5117,7 +5314,8 @@ static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
int i;
/* Check if enabled_tc is same as existing or new TCs */
- if (vsi->tc_config.enabled_tc == enabled_tc)
+ if (vsi->tc_config.enabled_tc == enabled_tc &&
+ vsi->mqprio_qopt.mode != TC_MQPRIO_MODE_CHANNEL)
return ret;
/* Enable ETS TCs with equal BW Share for now across all VSIs */
@@ -5140,15 +5338,37 @@ static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
ctxt.vf_num = 0;
ctxt.uplink_seid = vsi->uplink_seid;
ctxt.info = vsi->info;
- i40e_vsi_setup_queue_map(vsi, &ctxt, enabled_tc, false);
+ if (vsi->back->flags & I40E_FLAG_TC_MQPRIO) {
+ ret = i40e_vsi_setup_queue_map_mqprio(vsi, &ctxt, enabled_tc);
+ if (ret)
+ goto out;
+ } else {
+ i40e_vsi_setup_queue_map(vsi, &ctxt, enabled_tc, false);
+ }
+ /* On destroying the qdisc, reset vsi->rss_size, as number of enabled
+ * queues changed.
+ */
+ if (!vsi->mqprio_qopt.qopt.hw && vsi->reconfig_rss) {
+ vsi->rss_size = min_t(int, vsi->back->alloc_rss_size,
+ vsi->num_queue_pairs);
+ ret = i40e_vsi_config_rss(vsi);
+ if (ret) {
+ dev_info(&vsi->back->pdev->dev,
+ "Failed to reconfig rss for num_queues\n");
+ return ret;
+ }
+ vsi->reconfig_rss = false;
+ }
if (vsi->back->flags & I40E_FLAG_IWARP_ENABLED) {
ctxt.info.valid_sections |=
cpu_to_le16(I40E_AQ_VSI_PROP_QUEUE_OPT_VALID);
ctxt.info.queueing_opt_flags |= I40E_AQ_VSI_QUE_OPT_TCP_ENA;
}
- /* Update the VSI after updating the VSI queue-mapping information */
+ /* Update the VSI after updating the VSI queue-mapping
+ * information
+ */
ret = i40e_aq_update_vsi_params(&vsi->back->hw, &ctxt, NULL);
if (ret) {
dev_info(&vsi->back->pdev->dev,
@@ -6216,54 +6436,157 @@ void i40e_down(struct i40e_vsi *vsi)
}
+/**
+ * i40e_validate_mqprio_qopt- validate queue mapping info
+ * @vsi: the VSI being configured
+ * @mqprio_qopt: queue parametrs
+ **/
+static int i40e_validate_mqprio_qopt(struct i40e_vsi *vsi,
+ struct tc_mqprio_qopt_offload *mqprio_qopt)
+{
+ int i;
+
+ if (mqprio_qopt->qopt.offset[0] != 0 ||
+ mqprio_qopt->qopt.num_tc < 1 ||
+ mqprio_qopt->qopt.num_tc > I40E_MAX_TRAFFIC_CLASS)
+ return -EINVAL;
+ for (i = 0; ; i++) {
+ if (!mqprio_qopt->qopt.count[i])
+ return -EINVAL;
+ if (mqprio_qopt->min_rate[i] || mqprio_qopt->max_rate[i])
+ return -EINVAL;
+ if (i >= mqprio_qopt->qopt.num_tc - 1)
+ break;
+ if (mqprio_qopt->qopt.offset[i + 1] !=
+ (mqprio_qopt->qopt.offset[i] + mqprio_qopt->qopt.count[i]))
+ return -EINVAL;
+ }
+ if (vsi->num_queue_pairs <
+ (mqprio_qopt->qopt.offset[i] + mqprio_qopt->qopt.count[i])) {
+ return -EINVAL;
+ }
+ return 0;
+}
+
+/**
+ * i40e_vsi_set_default_tc_config - set default values for tc configuration
+ * @vsi: the VSI being configured
+ **/
+static void i40e_vsi_set_default_tc_config(struct i40e_vsi *vsi)
+{
+ u16 qcount;
+ int i;
+
+ /* Only TC0 is enabled */
+ vsi->tc_config.numtc = 1;
+ vsi->tc_config.enabled_tc = 1;
+ qcount = min_t(int, vsi->alloc_queue_pairs,
+ i40e_pf_get_max_q_per_tc(vsi->back));
+ for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+ /* For the TC that is not enabled set the offset to to default
+ * queue and allocate one queue for the given TC.
+ */
+ vsi->tc_config.tc_info[i].qoffset = 0;
+ if (i == 0)
+ vsi->tc_config.tc_info[i].qcount = qcount;
+ else
+ vsi->tc_config.tc_info[i].qcount = 1;
+ vsi->tc_config.tc_info[i].netdev_tc = 0;
+ }
+}
+
/**
* i40e_setup_tc - configure multiple traffic classes
* @netdev: net device to configure
- * @tc: number of traffic classes to enable
+ * @type_data: tc offload data
**/
-static int i40e_setup_tc(struct net_device *netdev, u8 tc)
+static int i40e_setup_tc(struct net_device *netdev, void *type_data)
{
+ struct tc_mqprio_qopt_offload *mqprio_qopt = type_data;
struct i40e_netdev_priv *np = netdev_priv(netdev);
struct i40e_vsi *vsi = np->vsi;
struct i40e_pf *pf = vsi->back;
- u8 enabled_tc = 0;
+ u8 enabled_tc = 0, num_tc, hw;
+ bool need_reset = false;
int ret = -EINVAL;
+ u16 mode;
int i;
- /* Check if DCB enabled to continue */
- if (!(pf->flags & I40E_FLAG_DCB_ENABLED)) {
- netdev_info(netdev, "DCB is not enabled for adapter\n");
- goto exit;
+ num_tc = mqprio_qopt->qopt.num_tc;
+ hw = mqprio_qopt->qopt.hw;
+ mode = mqprio_qopt->mode;
+ if (!hw) {
+ pf->flags &= ~I40E_FLAG_TC_MQPRIO;
+ memcpy(&vsi->mqprio_qopt, mqprio_qopt, sizeof(*mqprio_qopt));
+ goto config_tc;
}
/* Check if MFP enabled */
if (pf->flags & I40E_FLAG_MFP_ENABLED) {
- netdev_info(netdev, "Configuring TC not supported in MFP mode\n");
- goto exit;
+ netdev_info(netdev,
+ "Configuring TC not supported in MFP mode\n");
+ return ret;
}
+ switch (mode) {
+ case TC_MQPRIO_MODE_DCB:
+ pf->flags &= ~I40E_FLAG_TC_MQPRIO;
- /* Check whether tc count is within enabled limit */
- if (tc > i40e_pf_get_num_tc(pf)) {
- netdev_info(netdev, "TC count greater than enabled on link for adapter\n");
- goto exit;
+ /* Check if DCB enabled to continue */
+ if (!(pf->flags & I40E_FLAG_DCB_ENABLED)) {
+ netdev_info(netdev,
+ "DCB is not enabled for adapter\n");
+ return ret;
+ }
+
+ /* Check whether tc count is within enabled limit */
+ if (num_tc > i40e_pf_get_num_tc(pf)) {
+ netdev_info(netdev,
+ "TC count greater than enabled on link for adapter\n");
+ return ret;
+ }
+ break;
+ case TC_MQPRIO_MODE_CHANNEL:
+ if (pf->flags & I40E_FLAG_DCB_ENABLED) {
+ netdev_info(netdev,
+ "Full offload of TC Mqprio options is not supported when DCB is enabled\n");
+ return ret;
+ }
+ if (!(pf->flags & I40E_FLAG_MSIX_ENABLED))
+ return ret;
+ ret = i40e_validate_mqprio_qopt(vsi, mqprio_qopt);
+ if (ret)
+ return ret;
+ memcpy(&vsi->mqprio_qopt, mqprio_qopt,
+ sizeof(*mqprio_qopt));
+ pf->flags |= I40E_FLAG_TC_MQPRIO;
+ pf->flags &= ~I40E_FLAG_DCB_ENABLED;
+ break;
+ default:
+ return -EINVAL;
}
+config_tc:
/* Generate TC map for number of tc requested */
- for (i = 0; i < tc; i++)
+ for (i = 0; i < num_tc; i++)
enabled_tc |= BIT(i);
/* Requesting same TC configuration as already enabled */
- if (enabled_tc == vsi->tc_config.enabled_tc)
+ if (enabled_tc == vsi->tc_config.enabled_tc &&
+ mode != TC_MQPRIO_MODE_CHANNEL)
return 0;
/* Quiesce VSI queues */
i40e_quiesce_vsi(vsi);
+ if (!hw && !(pf->flags & I40E_FLAG_TC_MQPRIO))
+ i40e_remove_queue_channels(vsi);
+
/* Configure VSI for enabled TCs */
ret = i40e_vsi_config_tc(vsi, enabled_tc);
if (ret) {
netdev_info(netdev, "Failed configuring TC for VSI seid=%d\n",
vsi->seid);
+ need_reset = true;
goto exit;
}
@@ -6272,11 +6595,18 @@ static int i40e_setup_tc(struct net_device *netdev, u8 tc)
if (ret) {
netdev_info(netdev,
"Failed configuring queue channels\n");
+ need_reset = true;
goto exit;
}
}
exit:
+ /* Reset the configuration data to defaults, only TC0 is enabled */
+ if (need_reset) {
+ i40e_vsi_set_default_tc_config(vsi);
+ need_reset = false;
+ }
+
/* Unquiesce VSI */
i40e_unquiesce_vsi(vsi);
return ret;
@@ -6285,14 +6615,10 @@ static int i40e_setup_tc(struct net_device *netdev, u8 tc)
static int __i40e_setup_tc(struct net_device *netdev, enum tc_setup_type type,
void *type_data)
{
- struct tc_mqprio_qopt *mqprio = type_data;
-
if (type != TC_SETUP_MQPRIO)
return -EOPNOTSUPP;
- mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
-
- return i40e_setup_tc(netdev, mqprio->num_tc);
+ return i40e_setup_tc(netdev, type_data);
}
/**
@@ -9153,45 +9479,6 @@ static int i40e_setup_misc_vector(struct i40e_pf *pf)
return err;
}
-/**
- * i40e_config_rss_aq - Prepare for RSS using AQ commands
- * @vsi: vsi structure
- * @seed: RSS hash seed
- **/
-static int i40e_config_rss_aq(struct i40e_vsi *vsi, const u8 *seed,
- u8 *lut, u16 lut_size)
-{
- struct i40e_pf *pf = vsi->back;
- struct i40e_hw *hw = &pf->hw;
- int ret = 0;
-
- if (seed) {
- struct i40e_aqc_get_set_rss_key_data *seed_dw =
- (struct i40e_aqc_get_set_rss_key_data *)seed;
- ret = i40e_aq_set_rss_key(hw, vsi->id, seed_dw);
- if (ret) {
- dev_info(&pf->pdev->dev,
- "Cannot set RSS key, err %s aq_err %s\n",
- i40e_stat_str(hw, ret),
- i40e_aq_str(hw, hw->aq.asq_last_status));
- return ret;
- }
- }
- if (lut) {
- bool pf_lut = vsi->type == I40E_VSI_MAIN ? true : false;
-
- ret = i40e_aq_set_rss_lut(hw, vsi->id, pf_lut, lut, lut_size);
- if (ret) {
- dev_info(&pf->pdev->dev,
- "Cannot set RSS lut, err %s aq_err %s\n",
- i40e_stat_str(hw, ret),
- i40e_aq_str(hw, hw->aq.asq_last_status));
- return ret;
- }
- }
- return ret;
-}
-
/**
* i40e_get_rss_aq - Get RSS keys and lut by using AQ commands
* @vsi: Pointer to vsi structure
@@ -9238,46 +9525,6 @@ static int i40e_get_rss_aq(struct i40e_vsi *vsi, const u8 *seed,
return ret;
}
-/**
- * i40e_vsi_config_rss - Prepare for VSI(VMDq) RSS if used
- * @vsi: VSI structure
- **/
-static int i40e_vsi_config_rss(struct i40e_vsi *vsi)
-{
- u8 seed[I40E_HKEY_ARRAY_SIZE];
- struct i40e_pf *pf = vsi->back;
- u8 *lut;
- int ret;
-
- if (!(pf->hw_features & I40E_HW_RSS_AQ_CAPABLE))
- return 0;
-
- if (!vsi->rss_size)
- vsi->rss_size = min_t(int, pf->alloc_rss_size,
- vsi->num_queue_pairs);
- if (!vsi->rss_size)
- return -EINVAL;
-
- lut = kzalloc(vsi->rss_table_size, GFP_KERNEL);
- if (!lut)
- return -ENOMEM;
- /* Use the user configured hash keys and lookup table if there is one,
- * otherwise use default
- */
- if (vsi->rss_lut_user)
- memcpy(lut, vsi->rss_lut_user, vsi->rss_table_size);
- else
- i40e_fill_rss_lut(pf, lut, vsi->rss_table_size, vsi->rss_size);
- if (vsi->rss_hkey_user)
- memcpy(seed, vsi->rss_hkey_user, I40E_HKEY_ARRAY_SIZE);
- else
- netdev_rss_key_fill((void *)seed, I40E_HKEY_ARRAY_SIZE);
- ret = i40e_config_rss_aq(vsi, seed, lut, vsi->rss_table_size);
- kfree(lut);
-
- return ret;
-}
-
/**
* i40e_config_rss_reg - Configure RSS keys and lut by writing registers
* @vsi: Pointer to vsi structure
--
2.14.2
^ permalink raw reply related
* [net-next 8/9] i40e: fix link reporting
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Alan Brady, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Alan Brady <alan.brady@intel.com>
When querying the NVM for supported phy_types, on some firmware
versions, we were failing to actually fill out the phy_types which means
ethtool wouldn't report any link types.
Testing-hints:
Check 'ethtool <iface>' if you have the right (wrong?) firmware.
Without this patch, no link modes will be reported.
Signed-off-by: Alan Brady <alan.brady@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/i40e/i40e_common.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e_common.c b/drivers/net/ethernet/intel/i40e/i40e_common.c
index 53aad378d49c..aeb497258f20 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_common.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_common.c
@@ -1611,8 +1611,13 @@ i40e_status i40e_aq_get_phy_capabilities(struct i40e_hw *hw,
if (report_init) {
if (hw->mac.type == I40E_MAC_XL710 &&
hw->aq.api_maj_ver == I40E_FW_API_VERSION_MAJOR &&
- hw->aq.api_min_ver >= I40E_MINOR_VER_GET_LINK_INFO_XL710)
+ hw->aq.api_min_ver >= I40E_MINOR_VER_GET_LINK_INFO_XL710) {
status = i40e_aq_get_link_info(hw, true, NULL, NULL);
+ } else {
+ hw->phy.phy_types = le32_to_cpu(abilities->phy_type);
+ hw->phy.phy_types |=
+ ((u64)abilities->phy_type_ext << 32);
+ }
}
return status;
--
2.14.2
^ permalink raw reply related
* [net-next 3/9] i40e: Add infrastructure for queue channel support
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem
Cc: Amritha Nambiar, netdev, nhorman, sassmann, jogreene, Kiran Patil,
Jeff Kirsher
In-Reply-To: <20171013215249.61286-1-jeffrey.t.kirsher@intel.com>
From: Amritha Nambiar <amritha.nambiar@intel.com>
This patch sets up the infrastructure for offloading TCs and
queue configurations to the hardware by creating HW channels(VSI).
A new channel is created for each of the traffic class
configuration offloaded via mqprio framework except for the first TC
(TC0). TC0 for the main VSI is also reconfigured as per user provided
queue parameters. Queue counts that are not power-of-2 are handled by
reconfiguring RSS by reprogramming LUTs using the queue count value.
This patch also handles configuring the TX rings for the channels,
setting up the RX queue map for channel.
Also, the channels so created are removed and all the queue
configuration is set to default when the qdisc is detached from the
root of the device.
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/i40e/i40e.h | 32 ++
drivers/net/ethernet/intel/i40e/i40e_main.c | 718 +++++++++++++++++++++++++++-
drivers/net/ethernet/intel/i40e/i40e_txrx.h | 2 +
3 files changed, 743 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e.h b/drivers/net/ethernet/intel/i40e/i40e.h
index e7c7a853cf7f..bde982541772 100644
--- a/drivers/net/ethernet/intel/i40e/i40e.h
+++ b/drivers/net/ethernet/intel/i40e/i40e.h
@@ -87,6 +87,7 @@
#define I40E_AQ_LEN 256
#define I40E_AQ_WORK_LIMIT 66 /* max number of VFs + a little */
#define I40E_MAX_USER_PRIORITY 8
+#define I40E_MAX_QUEUES_PER_CH 64
#define I40E_DEFAULT_TRAFFIC_CLASS BIT(0)
#define I40E_DEFAULT_MSG_ENABLE 4
#define I40E_QUEUE_WAIT_RETRY_LIMIT 10
@@ -340,6 +341,23 @@ struct i40e_flex_pit {
u8 pit_index;
};
+struct i40e_channel {
+ struct list_head list;
+ bool initialized;
+ u8 type;
+ u16 vsi_number; /* Assigned VSI number from AQ 'Add VSI' response */
+ u16 stat_counter_idx;
+ u16 base_queue;
+ u16 num_queue_pairs; /* Requested by user */
+ u16 seid;
+
+ u8 enabled_tc;
+ struct i40e_aqc_vsi_properties_data info;
+
+ /* track this channel belongs to which VSI */
+ struct i40e_vsi *parent_vsi;
+};
+
/* struct that defines the Ethernet device */
struct i40e_pf {
struct pci_dev *pdev;
@@ -456,6 +474,7 @@ struct i40e_pf {
#define I40E_FLAG_CLIENT_RESET BIT(26)
#define I40E_FLAG_LINK_DOWN_ON_CLOSE_ENABLED BIT(27)
#define I40E_FLAG_SOURCE_PRUNING_DISABLED BIT(28)
+#define I40E_FLAG_TC_MQPRIO BIT(29)
struct i40e_client_instance *cinst;
bool stat_offsets_loaded;
@@ -536,6 +555,8 @@ struct i40e_pf {
u32 ioremap_len;
u32 fd_inv;
u16 phy_led_val;
+
+ u16 override_q_count;
};
/**
@@ -700,6 +721,15 @@ struct i40e_vsi {
bool current_isup; /* Sync 'link up' logging */
enum i40e_aq_link_speed current_speed; /* Sync link speed logging */
+ /* channel specific fields */
+ u16 cnt_q_avail; /* num of queues available for channel usage */
+ u16 orig_rss_size;
+ u16 current_rss_size;
+
+ u16 next_base_queue; /* next queue to be used for channel setup */
+
+ struct list_head ch_list;
+
void *priv; /* client driver data reference. */
/* VSI specific handlers */
@@ -1004,4 +1034,6 @@ static inline bool i40e_enabled_xdp_vsi(struct i40e_vsi *vsi)
{
return !!vsi->xdp_prog;
}
+
+int i40e_create_queue_channel(struct i40e_vsi *vsi, struct i40e_channel *ch);
#endif /* _I40E_H_ */
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index 6190257eecfe..e23105bee6d1 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -2881,7 +2881,7 @@ static void i40e_config_xps_tx_ring(struct i40e_ring *ring)
{
int cpu;
- if (!ring->q_vector || !ring->netdev)
+ if (!ring->q_vector || !ring->netdev || ring->ch)
return;
/* We only initialize XPS once, so as not to overwrite user settings */
@@ -2944,7 +2944,14 @@ static int i40e_configure_tx_ring(struct i40e_ring *ring)
* initialization. This has to be done regardless of
* DCB as by default everything is mapped to TC0.
*/
- tx_ctx.rdylist = le16_to_cpu(vsi->info.qs_handle[ring->dcb_tc]);
+
+ if (ring->ch)
+ tx_ctx.rdylist =
+ le16_to_cpu(ring->ch->info.qs_handle[ring->dcb_tc]);
+
+ else
+ tx_ctx.rdylist = le16_to_cpu(vsi->info.qs_handle[ring->dcb_tc]);
+
tx_ctx.rdylist_act = 0;
/* clear the context in the HMC */
@@ -2966,12 +2973,23 @@ static int i40e_configure_tx_ring(struct i40e_ring *ring)
}
/* Now associate this queue with this PCI function */
- if (vsi->type == I40E_VSI_VMDQ2) {
- qtx_ctl = I40E_QTX_CTL_VM_QUEUE;
- qtx_ctl |= ((vsi->id) << I40E_QTX_CTL_VFVM_INDX_SHIFT) &
- I40E_QTX_CTL_VFVM_INDX_MASK;
+ if (ring->ch) {
+ if (ring->ch->type == I40E_VSI_VMDQ2)
+ qtx_ctl = I40E_QTX_CTL_VM_QUEUE;
+ else
+ return -EINVAL;
+
+ qtx_ctl |= (ring->ch->vsi_number <<
+ I40E_QTX_CTL_VFVM_INDX_SHIFT) &
+ I40E_QTX_CTL_VFVM_INDX_MASK;
} else {
- qtx_ctl = I40E_QTX_CTL_PF_QUEUE;
+ if (vsi->type == I40E_VSI_VMDQ2) {
+ qtx_ctl = I40E_QTX_CTL_VM_QUEUE;
+ qtx_ctl |= ((vsi->id) << I40E_QTX_CTL_VFVM_INDX_SHIFT) &
+ I40E_QTX_CTL_VFVM_INDX_MASK;
+ } else {
+ qtx_ctl = I40E_QTX_CTL_PF_QUEUE;
+ }
}
qtx_ctl |= ((hw->pf_id << I40E_QTX_CTL_PF_INDX_SHIFT) &
@@ -5161,6 +5179,643 @@ static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
return ret;
}
+/**
+ * i40e_remove_queue_channels - Remove queue channels for the TCs
+ * @vsi: VSI to be configured
+ *
+ * Remove queue channels for the TCs
+ **/
+static void i40e_remove_queue_channels(struct i40e_vsi *vsi)
+{
+ struct i40e_channel *ch, *ch_tmp;
+ int ret, i;
+
+ /* Reset rss size that was stored when reconfiguring rss for
+ * channel VSIs with non-power-of-2 queue count.
+ */
+ vsi->current_rss_size = 0;
+
+ /* perform cleanup for channels if they exist */
+ if (list_empty(&vsi->ch_list))
+ return;
+
+ list_for_each_entry_safe(ch, ch_tmp, &vsi->ch_list, list) {
+ struct i40e_vsi *p_vsi;
+
+ list_del(&ch->list);
+ p_vsi = ch->parent_vsi;
+ if (!p_vsi || !ch->initialized) {
+ kfree(ch);
+ continue;
+ }
+ /* Reset queue contexts */
+ for (i = 0; i < ch->num_queue_pairs; i++) {
+ struct i40e_ring *tx_ring, *rx_ring;
+ u16 pf_q;
+
+ pf_q = ch->base_queue + i;
+ tx_ring = vsi->tx_rings[pf_q];
+ tx_ring->ch = NULL;
+
+ rx_ring = vsi->rx_rings[pf_q];
+ rx_ring->ch = NULL;
+ }
+
+ /* delete VSI from FW */
+ ret = i40e_aq_delete_element(&vsi->back->hw, ch->seid,
+ NULL);
+ if (ret)
+ dev_err(&vsi->back->pdev->dev,
+ "unable to remove channel (%d) for parent VSI(%d)\n",
+ ch->seid, p_vsi->seid);
+ kfree(ch);
+ }
+ INIT_LIST_HEAD(&vsi->ch_list);
+}
+
+/**
+ * i40e_is_any_channel - channel exist or not
+ * @vsi: ptr to VSI to which channels are associated with
+ *
+ * Returns true or false if channel(s) exist for associated VSI or not
+ **/
+static bool i40e_is_any_channel(struct i40e_vsi *vsi)
+{
+ struct i40e_channel *ch, *ch_tmp;
+
+ list_for_each_entry_safe(ch, ch_tmp, &vsi->ch_list, list) {
+ if (ch->initialized)
+ return true;
+ }
+
+ return false;
+}
+
+/**
+ * i40e_get_max_queues_for_channel
+ * @vsi: ptr to VSI to which channels are associated with
+ *
+ * Helper function which returns max value among the queue counts set on the
+ * channels/TCs created.
+ **/
+static int i40e_get_max_queues_for_channel(struct i40e_vsi *vsi)
+{
+ struct i40e_channel *ch, *ch_tmp;
+ int max = 0;
+
+ list_for_each_entry_safe(ch, ch_tmp, &vsi->ch_list, list) {
+ if (!ch->initialized)
+ continue;
+ if (ch->num_queue_pairs > max)
+ max = ch->num_queue_pairs;
+ }
+
+ return max;
+}
+
+/**
+ * i40e_validate_num_queues - validate num_queues w.r.t channel
+ * @pf: ptr to PF device
+ * @num_queues: number of queues
+ * @vsi: the parent VSI
+ * @reconfig_rss: indicates should the RSS be reconfigured or not
+ *
+ * This function validates number of queues in the context of new channel
+ * which is being established and determines if RSS should be reconfigured
+ * or not for parent VSI.
+ **/
+static int i40e_validate_num_queues(struct i40e_pf *pf, int num_queues,
+ struct i40e_vsi *vsi, bool *reconfig_rss)
+{
+ int max_ch_queues;
+
+ if (!reconfig_rss)
+ return -EINVAL;
+
+ *reconfig_rss = false;
+
+ if (num_queues > I40E_MAX_QUEUES_PER_CH) {
+ dev_err(&pf->pdev->dev,
+ "Failed to create VMDq VSI. User requested num_queues (%d) > I40E_MAX_QUEUES_PER_VSI (%u)\n",
+ num_queues, I40E_MAX_QUEUES_PER_CH);
+ return -EINVAL;
+ }
+
+ if (vsi->current_rss_size) {
+ if (num_queues > vsi->current_rss_size) {
+ dev_dbg(&pf->pdev->dev,
+ "Error: num_queues (%d) > vsi's current_size(%d)\n",
+ num_queues, vsi->current_rss_size);
+ return -EINVAL;
+ } else if ((num_queues < vsi->current_rss_size) &&
+ (!is_power_of_2(num_queues))) {
+ dev_dbg(&pf->pdev->dev,
+ "Error: num_queues (%d) < vsi's current_size(%d), but not power of 2\n",
+ num_queues, vsi->current_rss_size);
+ return -EINVAL;
+ }
+ }
+
+ if (!is_power_of_2(num_queues)) {
+ /* Find the max num_queues configured for channel if channel
+ * exist.
+ * if channel exist, then enforce 'num_queues' to be more than
+ * max ever queues configured for channel.
+ */
+ max_ch_queues = i40e_get_max_queues_for_channel(vsi);
+ if (num_queues < max_ch_queues) {
+ dev_dbg(&pf->pdev->dev,
+ "Error: num_queues (%d) < max queues configured for channel(%d)\n",
+ num_queues, max_ch_queues);
+ return -EINVAL;
+ }
+ *reconfig_rss = true;
+ }
+
+ return 0;
+}
+
+/**
+ * i40e_vsi_reconfig_rss - reconfig RSS based on specified rss_size
+ * @vsi: the VSI being setup
+ * @rss_size: size of RSS, accordingly LUT gets reprogrammed
+ *
+ * This function reconfigures RSS by reprogramming LUTs using 'rss_size'
+ **/
+static int i40e_vsi_reconfig_rss(struct i40e_vsi *vsi, u16 rss_size)
+{
+ struct i40e_pf *pf = vsi->back;
+ u8 seed[I40E_HKEY_ARRAY_SIZE];
+ struct i40e_hw *hw = &pf->hw;
+ int local_rss_size;
+ u8 *lut;
+ int ret;
+
+ if (!vsi->rss_size)
+ return -EINVAL;
+
+ if (rss_size > vsi->rss_size)
+ return -EINVAL;
+
+ local_rss_size = min_t(int, vsi->rss_size, rss_size);
+ lut = kzalloc(vsi->rss_table_size, GFP_KERNEL);
+ if (!lut)
+ return -ENOMEM;
+
+ /* Ignoring user configured lut if there is one */
+ i40e_fill_rss_lut(pf, lut, vsi->rss_table_size, local_rss_size);
+
+ /* Use user configured hash key if there is one, otherwise
+ * use default.
+ */
+ if (vsi->rss_hkey_user)
+ memcpy(seed, vsi->rss_hkey_user, I40E_HKEY_ARRAY_SIZE);
+ else
+ netdev_rss_key_fill((void *)seed, I40E_HKEY_ARRAY_SIZE);
+
+ ret = i40e_config_rss(vsi, seed, lut, vsi->rss_table_size);
+ if (ret) {
+ dev_info(&pf->pdev->dev,
+ "Cannot set RSS lut, err %s aq_err %s\n",
+ i40e_stat_str(hw, ret),
+ i40e_aq_str(hw, hw->aq.asq_last_status));
+ kfree(lut);
+ return ret;
+ }
+ kfree(lut);
+
+ /* Do the update w.r.t. storing rss_size */
+ if (!vsi->orig_rss_size)
+ vsi->orig_rss_size = vsi->rss_size;
+ vsi->current_rss_size = local_rss_size;
+
+ return ret;
+}
+
+/**
+ * i40e_channel_setup_queue_map - Setup a channel queue map
+ * @pf: ptr to PF device
+ * @vsi: the VSI being setup
+ * @ctxt: VSI context structure
+ * @ch: ptr to channel structure
+ *
+ * Setup queue map for a specific channel
+ **/
+static void i40e_channel_setup_queue_map(struct i40e_pf *pf,
+ struct i40e_vsi_context *ctxt,
+ struct i40e_channel *ch)
+{
+ u16 qcount, qmap, sections = 0;
+ u8 offset = 0;
+ int pow;
+
+ sections = I40E_AQ_VSI_PROP_QUEUE_MAP_VALID;
+ sections |= I40E_AQ_VSI_PROP_SCHED_VALID;
+
+ qcount = min_t(int, ch->num_queue_pairs, pf->num_lan_msix);
+ ch->num_queue_pairs = qcount;
+
+ /* find the next higher power-of-2 of num queue pairs */
+ pow = ilog2(qcount);
+ if (!is_power_of_2(qcount))
+ pow++;
+
+ qmap = (offset << I40E_AQ_VSI_TC_QUE_OFFSET_SHIFT) |
+ (pow << I40E_AQ_VSI_TC_QUE_NUMBER_SHIFT);
+
+ /* Setup queue TC[0].qmap for given VSI context */
+ ctxt->info.tc_mapping[0] = cpu_to_le16(qmap);
+
+ ctxt->info.up_enable_bits = 0x1; /* TC0 enabled */
+ ctxt->info.mapping_flags |= cpu_to_le16(I40E_AQ_VSI_QUE_MAP_CONTIG);
+ ctxt->info.queue_mapping[0] = cpu_to_le16(ch->base_queue);
+ ctxt->info.valid_sections |= cpu_to_le16(sections);
+}
+
+/**
+ * i40e_add_channel - add a channel by adding VSI
+ * @pf: ptr to PF device
+ * @uplink_seid: underlying HW switching element (VEB) ID
+ * @ch: ptr to channel structure
+ *
+ * Add a channel (VSI) using add_vsi and queue_map
+ **/
+static int i40e_add_channel(struct i40e_pf *pf, u16 uplink_seid,
+ struct i40e_channel *ch)
+{
+ struct i40e_hw *hw = &pf->hw;
+ struct i40e_vsi_context ctxt;
+ u8 enabled_tc = 0x1; /* TC0 enabled */
+ int ret;
+
+ if (ch->type != I40E_VSI_VMDQ2) {
+ dev_info(&pf->pdev->dev,
+ "add new vsi failed, ch->type %d\n", ch->type);
+ return -EINVAL;
+ }
+
+ memset(&ctxt, 0, sizeof(ctxt));
+ ctxt.pf_num = hw->pf_id;
+ ctxt.vf_num = 0;
+ ctxt.uplink_seid = uplink_seid;
+ ctxt.connection_type = I40E_AQ_VSI_CONN_TYPE_NORMAL;
+ if (ch->type == I40E_VSI_VMDQ2)
+ ctxt.flags = I40E_AQ_VSI_TYPE_VMDQ2;
+
+ if (pf->flags & I40E_FLAG_VEB_MODE_ENABLED) {
+ ctxt.info.valid_sections |=
+ cpu_to_le16(I40E_AQ_VSI_PROP_SWITCH_VALID);
+ ctxt.info.switch_id =
+ cpu_to_le16(I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB);
+ }
+
+ /* Set queue map for a given VSI context */
+ i40e_channel_setup_queue_map(pf, &ctxt, ch);
+
+ /* Now time to create VSI */
+ ret = i40e_aq_add_vsi(hw, &ctxt, NULL);
+ if (ret) {
+ dev_info(&pf->pdev->dev,
+ "add new vsi failed, err %s aq_err %s\n",
+ i40e_stat_str(&pf->hw, ret),
+ i40e_aq_str(&pf->hw,
+ pf->hw.aq.asq_last_status));
+ return -ENOENT;
+ }
+
+ /* Success, update channel */
+ ch->enabled_tc = enabled_tc;
+ ch->seid = ctxt.seid;
+ ch->vsi_number = ctxt.vsi_number;
+ ch->stat_counter_idx = cpu_to_le16(ctxt.info.stat_counter_idx);
+
+ /* copy just the sections touched not the entire info
+ * since not all sections are valid as returned by
+ * update vsi params
+ */
+ ch->info.mapping_flags = ctxt.info.mapping_flags;
+ memcpy(&ch->info.queue_mapping,
+ &ctxt.info.queue_mapping, sizeof(ctxt.info.queue_mapping));
+ memcpy(&ch->info.tc_mapping, ctxt.info.tc_mapping,
+ sizeof(ctxt.info.tc_mapping));
+
+ return 0;
+}
+
+static int i40e_channel_config_bw(struct i40e_vsi *vsi, struct i40e_channel *ch,
+ u8 *bw_share)
+{
+ struct i40e_aqc_configure_vsi_tc_bw_data bw_data;
+ i40e_status ret;
+ int i;
+
+ bw_data.tc_valid_bits = ch->enabled_tc;
+ for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++)
+ bw_data.tc_bw_credits[i] = bw_share[i];
+
+ ret = i40e_aq_config_vsi_tc_bw(&vsi->back->hw, ch->seid,
+ &bw_data, NULL);
+ if (ret) {
+ dev_info(&vsi->back->pdev->dev,
+ "Config VSI BW allocation per TC failed, aq_err: %d for new_vsi->seid %u\n",
+ vsi->back->hw.aq.asq_last_status, ch->seid);
+ return -EINVAL;
+ }
+
+ for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++)
+ ch->info.qs_handle[i] = bw_data.qs_handles[i];
+
+ return 0;
+}
+
+/**
+ * i40e_channel_config_tx_ring - config TX ring associated with new channel
+ * @pf: ptr to PF device
+ * @vsi: the VSI being setup
+ * @ch: ptr to channel structure
+ *
+ * Configure TX rings associated with channel (VSI) since queues are being
+ * from parent VSI.
+ **/
+static int i40e_channel_config_tx_ring(struct i40e_pf *pf,
+ struct i40e_vsi *vsi,
+ struct i40e_channel *ch)
+{
+ i40e_status ret;
+ int i;
+ u8 bw_share[I40E_MAX_TRAFFIC_CLASS] = {0};
+
+ /* Enable ETS TCs with equal BW Share for now across all VSIs */
+ for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+ if (ch->enabled_tc & BIT(i))
+ bw_share[i] = 1;
+ }
+
+ /* configure BW for new VSI */
+ ret = i40e_channel_config_bw(vsi, ch, bw_share);
+ if (ret) {
+ dev_info(&vsi->back->pdev->dev,
+ "Failed configuring TC map %d for channel (seid %u)\n",
+ ch->enabled_tc, ch->seid);
+ return ret;
+ }
+
+ for (i = 0; i < ch->num_queue_pairs; i++) {
+ struct i40e_ring *tx_ring, *rx_ring;
+ u16 pf_q;
+
+ pf_q = ch->base_queue + i;
+
+ /* Get to TX ring ptr of main VSI, for re-setup TX queue
+ * context
+ */
+ tx_ring = vsi->tx_rings[pf_q];
+ tx_ring->ch = ch;
+
+ /* Get the RX ring ptr */
+ rx_ring = vsi->rx_rings[pf_q];
+ rx_ring->ch = ch;
+ }
+
+ return 0;
+}
+
+/**
+ * i40e_setup_hw_channel - setup new channel
+ * @pf: ptr to PF device
+ * @vsi: the VSI being setup
+ * @ch: ptr to channel structure
+ * @uplink_seid: underlying HW switching element (VEB) ID
+ * @type: type of channel to be created (VMDq2/VF)
+ *
+ * Setup new channel (VSI) based on specified type (VMDq2/VF)
+ * and configures TX rings accordingly
+ **/
+static inline int i40e_setup_hw_channel(struct i40e_pf *pf,
+ struct i40e_vsi *vsi,
+ struct i40e_channel *ch,
+ u16 uplink_seid, u8 type)
+{
+ int ret;
+
+ ch->initialized = false;
+ ch->base_queue = vsi->next_base_queue;
+ ch->type = type;
+
+ /* Proceed with creation of channel (VMDq2) VSI */
+ ret = i40e_add_channel(pf, uplink_seid, ch);
+ if (ret) {
+ dev_info(&pf->pdev->dev,
+ "failed to add_channel using uplink_seid %u\n",
+ uplink_seid);
+ return ret;
+ }
+
+ /* Mark the successful creation of channel */
+ ch->initialized = true;
+
+ /* Reconfigure TX queues using QTX_CTL register */
+ ret = i40e_channel_config_tx_ring(pf, vsi, ch);
+ if (ret) {
+ dev_info(&pf->pdev->dev,
+ "failed to configure TX rings for channel %u\n",
+ ch->seid);
+ return ret;
+ }
+
+ /* update 'next_base_queue' */
+ vsi->next_base_queue = vsi->next_base_queue + ch->num_queue_pairs;
+ dev_dbg(&pf->pdev->dev,
+ "Added channel: vsi_seid %u, vsi_number %u, stat_counter_idx %u, num_queue_pairs %u, pf->next_base_queue %d\n",
+ ch->seid, ch->vsi_number, ch->stat_counter_idx,
+ ch->num_queue_pairs,
+ vsi->next_base_queue);
+ return ret;
+}
+
+/**
+ * i40e_setup_channel - setup new channel using uplink element
+ * @pf: ptr to PF device
+ * @type: type of channel to be created (VMDq2/VF)
+ * @uplink_seid: underlying HW switching element (VEB) ID
+ * @ch: ptr to channel structure
+ *
+ * Setup new channel (VSI) based on specified type (VMDq2/VF)
+ * and uplink switching element (uplink_seid)
+ **/
+static bool i40e_setup_channel(struct i40e_pf *pf, struct i40e_vsi *vsi,
+ struct i40e_channel *ch)
+{
+ u8 vsi_type;
+ u16 seid;
+ int ret;
+
+ if (vsi->type == I40E_VSI_MAIN) {
+ vsi_type = I40E_VSI_VMDQ2;
+ } else {
+ dev_err(&pf->pdev->dev, "unsupported parent vsi type(%d)\n",
+ vsi->type);
+ return false;
+ }
+
+ /* underlying switching element */
+ seid = pf->vsi[pf->lan_vsi]->uplink_seid;
+
+ /* create channel (VSI), configure TX rings */
+ ret = i40e_setup_hw_channel(pf, vsi, ch, seid, vsi_type);
+ if (ret) {
+ dev_err(&pf->pdev->dev, "failed to setup hw_channel\n");
+ return false;
+ }
+
+ return ch->initialized ? true : false;
+}
+
+/**
+ * i40e_create_queue_channel - function to create channel
+ * @vsi: VSI to be configured
+ * @ch: ptr to channel (it contains channel specific params)
+ *
+ * This function creates channel (VSI) using num_queues specified by user,
+ * reconfigs RSS if needed.
+ **/
+int i40e_create_queue_channel(struct i40e_vsi *vsi,
+ struct i40e_channel *ch)
+{
+ struct i40e_pf *pf = vsi->back;
+ bool reconfig_rss;
+ int err;
+
+ if (!ch)
+ return -EINVAL;
+
+ if (!ch->num_queue_pairs) {
+ dev_err(&pf->pdev->dev, "Invalid num_queues requested: %d\n",
+ ch->num_queue_pairs);
+ return -EINVAL;
+ }
+
+ /* validate user requested num_queues for channel */
+ err = i40e_validate_num_queues(pf, ch->num_queue_pairs, vsi,
+ &reconfig_rss);
+ if (err) {
+ dev_info(&pf->pdev->dev, "Failed to validate num_queues (%d)\n",
+ ch->num_queue_pairs);
+ return -EINVAL;
+ }
+
+ /* By default we are in VEPA mode, if this is the first VF/VMDq
+ * VSI to be added switch to VEB mode.
+ */
+ if ((!(pf->flags & I40E_FLAG_VEB_MODE_ENABLED)) ||
+ (!i40e_is_any_channel(vsi))) {
+ if (!is_power_of_2(vsi->tc_config.tc_info[0].qcount)) {
+ dev_dbg(&pf->pdev->dev,
+ "Failed to create channel. Override queues (%u) not power of 2\n",
+ vsi->tc_config.tc_info[0].qcount);
+ return -EINVAL;
+ }
+
+ if (!(pf->flags & I40E_FLAG_VEB_MODE_ENABLED)) {
+ pf->flags |= I40E_FLAG_VEB_MODE_ENABLED;
+
+ if (vsi->type == I40E_VSI_MAIN) {
+ if (pf->flags & I40E_FLAG_TC_MQPRIO)
+ i40e_do_reset(pf, I40E_PF_RESET_FLAG,
+ true);
+ else
+ i40e_do_reset_safe(pf,
+ I40E_PF_RESET_FLAG);
+ }
+ }
+ /* now onwards for main VSI, number of queues will be value
+ * of TC0's queue count
+ */
+ }
+
+ /* By this time, vsi->cnt_q_avail shall be set to non-zero and
+ * it should be more than num_queues
+ */
+ if (!vsi->cnt_q_avail || vsi->cnt_q_avail < ch->num_queue_pairs) {
+ dev_dbg(&pf->pdev->dev,
+ "Error: cnt_q_avail (%u) less than num_queues %d\n",
+ vsi->cnt_q_avail, ch->num_queue_pairs);
+ return -EINVAL;
+ }
+
+ /* reconfig_rss only if vsi type is MAIN_VSI */
+ if (reconfig_rss && (vsi->type == I40E_VSI_MAIN)) {
+ err = i40e_vsi_reconfig_rss(vsi, ch->num_queue_pairs);
+ if (err) {
+ dev_info(&pf->pdev->dev,
+ "Error: unable to reconfig rss for num_queues (%u)\n",
+ ch->num_queue_pairs);
+ return -EINVAL;
+ }
+ }
+
+ if (!i40e_setup_channel(pf, vsi, ch)) {
+ dev_info(&pf->pdev->dev, "Failed to setup channel\n");
+ return -EINVAL;
+ }
+
+ dev_info(&pf->pdev->dev,
+ "Setup channel (id:%u) utilizing num_queues %d\n",
+ ch->seid, ch->num_queue_pairs);
+
+ /* in case of VF, this will be main SRIOV VSI */
+ ch->parent_vsi = vsi;
+
+ /* and update main_vsi's count for queue_available to use */
+ vsi->cnt_q_avail -= ch->num_queue_pairs;
+
+ return 0;
+}
+
+/**
+ * i40e_configure_queue_channels - Add queue channel for the given TCs
+ * @vsi: VSI to be configured
+ *
+ * Configures queue channel mapping to the given TCs
+ **/
+static int i40e_configure_queue_channels(struct i40e_vsi *vsi)
+{
+ struct i40e_channel *ch;
+ int ret = 0, i;
+
+ /* Create app vsi with the TCs. Main VSI with TC0 is already set up */
+ for (i = 1; i < I40E_MAX_TRAFFIC_CLASS; i++) {
+ if (vsi->tc_config.enabled_tc & BIT(i)) {
+ ch = kzalloc(sizeof(*ch), GFP_KERNEL);
+ if (!ch) {
+ ret = -ENOMEM;
+ goto err_free;
+ }
+
+ INIT_LIST_HEAD(&ch->list);
+ ch->num_queue_pairs =
+ vsi->tc_config.tc_info[i].qcount;
+ ch->base_queue =
+ vsi->tc_config.tc_info[i].qoffset;
+
+ list_add_tail(&ch->list, &vsi->ch_list);
+
+ ret = i40e_create_queue_channel(vsi, ch);
+ if (ret) {
+ dev_err(&vsi->back->pdev->dev,
+ "Failed creating queue channel with TC%d: queues %d\n",
+ i, ch->num_queue_pairs);
+ goto err_free;
+ }
+ }
+ }
+ return ret;
+
+err_free:
+ i40e_remove_queue_channels(vsi);
+ return ret;
+}
+
/**
* i40e_veb_config_tc - Configure TCs for given VEB
* @veb: given VEB
@@ -5612,10 +6267,18 @@ static int i40e_setup_tc(struct net_device *netdev, u8 tc)
goto exit;
}
- /* Unquiesce VSI */
- i40e_unquiesce_vsi(vsi);
+ if (pf->flags & I40E_FLAG_TC_MQPRIO) {
+ ret = i40e_configure_queue_channels(vsi);
+ if (ret) {
+ netdev_info(netdev,
+ "Failed configuring queue channels\n");
+ goto exit;
+ }
+ }
exit:
+ /* Unquiesce VSI */
+ i40e_unquiesce_vsi(vsi);
return ret;
}
@@ -7030,6 +7693,35 @@ static void i40e_fdir_teardown(struct i40e_pf *pf)
i40e_vsi_release(vsi);
}
+/**
+ * i40e_rebuild_channels - Rebuilds channel VSIs if they existed before reset
+ * @vsi: PF main vsi
+ *
+ * Rebuilds channel VSIs if they existed before reset
+ **/
+static int i40e_rebuild_channels(struct i40e_vsi *vsi)
+{
+ struct i40e_channel *ch, *ch_tmp;
+ i40e_status ret;
+
+ if (list_empty(&vsi->ch_list))
+ return 0;
+
+ list_for_each_entry_safe(ch, ch_tmp, &vsi->ch_list, list) {
+ if (!ch->initialized)
+ break;
+ /* Proceed with creation of channel (VMDq2) VSI */
+ ret = i40e_add_channel(vsi->back, vsi->uplink_seid, ch);
+ if (ret) {
+ dev_info(&vsi->back->pdev->dev,
+ "failed to rebuild channels using uplink_seid %u\n",
+ vsi->uplink_seid);
+ return ret;
+ }
+ }
+ return 0;
+}
+
/**
* i40e_prep_for_reset - prep for the core to reset
* @pf: board private structure
@@ -7295,6 +7987,13 @@ static void i40e_rebuild(struct i40e_pf *pf, bool reinit, bool lock_acquired)
}
}
+ /* PF Main VSI is rebuild by now, go ahead and rebuild channel VSIs
+ * for this main VSI if they exist
+ */
+ ret = i40e_rebuild_channels(pf->vsi[pf->lan_vsi]);
+ if (ret)
+ goto end_unlock;
+
/* Reconfigure hardware for allowing smaller MSS in the case
* of TSO, so that we avoid the MDD being fired and causing
* a reset in the case of small MSS+TSO.
@@ -11608,6 +12307,7 @@ static int i40e_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
dev_info(&pdev->dev, "setup_pf_switch failed: %d\n", err);
goto err_vsis;
}
+ INIT_LIST_HEAD(&pf->vsi[pf->lan_vsi]->ch_list);
/* Make sure flow control is set according to current settings */
err = i40e_set_fc(hw, &set_fc_aq_fail, true);
diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.h b/drivers/net/ethernet/intel/i40e/i40e_txrx.h
index ff57ae451524..fbae1182e2ea 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_txrx.h
+++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.h
@@ -426,6 +426,8 @@ struct i40e_ring {
* i40e_clean_rx_ring_irq() is called
* for this ring.
*/
+
+ struct i40e_channel *ch;
} ____cacheline_internodealigned_in_smp;
static inline bool ring_uses_build_skb(struct i40e_ring *ring)
--
2.14.2
^ permalink raw reply related
* [net-next 0/9][pull request] 40GbE Intel Wired LAN Driver Updates 2017-10-13
From: Jeff Kirsher @ 2017-10-13 21:52 UTC (permalink / raw)
To: davem; +Cc: Jeff Kirsher, netdev, nhorman, sassmann, jogreene
This series contains updates to mqprio and i40e.
Amritha introduces a new hardware offload mode in tc/mqprio where the TCs,
the queue configurations and bandwidth rate limits are offloaded to the
hardware. The existing mqprio framework is extended to configure the queue
counts and layout and also added support for rate limiting. This is
achieved through new netlink attributes for the 'mode' option which takes
values such as 'dcb' (default) and 'channel' and a 'shaper' option for
QoS attributes such as bandwidth rate limits in hw mode 1. Legacy devices
can fall back to the existing setup supporting hw mode 1 without these
additional options where only the TCs are offloaded and then the 'mode'
and 'shaper' options defaults to DCB support. The i40e driver enables the
new mqprio hardware offload mechanism factoring the TCs, queue
configuration and bandwidth rates by creating HW channel VSIs.
In this new mode, the priority to traffic class mapping and the user
specified queue ranges are used to configure the traffic class when the
'mode' option is set to 'channel'. This is achieved by creating HW
channels(VSI). A new channel is created for each of the traffic class
configuration offloaded via mqprio framework except for the first TC (TC0)
which is for the main VSI. TC0 for the main VSI is also reconfigured as
per user provided queue parameters. Finally, bandwidth rate limits are set
on these traffic classes through the shaper attribute by sending these
rates in addition to the number of TCs and the queue configurations.
Colin Ian King makes an array of constant values "constant".
Alan fixes and issue where on some firmware versions, we were failing to
actually fill out the phy_types which caused ethtool to not report any
link types. Also hardened against a potentially malicious VF by not
letting the VF to reset itself after requesting to change the number of
queues (via ethtool), let the PF reset the VF to institute the requested
changes.
The following are changes since commit a00344bd1bbea2ba40719ae0eb3b6da7fae08cf2:
Merge branch 'tipc-comm-groups'
and are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queue 40GbE
Alan Brady (2):
i40e: fix link reporting
i40e/i40evf: don't trust VF to reset itself
Amritha Nambiar (6):
mqprio: Introduce new hardware offload mode and shaper in mqprio
i40e: Add macro for PF reset bit
i40e: Add infrastructure for queue channel support
i40e: Enable 'channel' mode in mqprio for TC configs
i40e: Refactor VF BW rate limiting
i40e: Add support setting TC max bandwidth rates
Colin Ian King (1):
i40e: make const array patterns static, reduces object code size
drivers/net/ethernet/intel/i40e/i40e.h | 44 +
drivers/net/ethernet/intel/i40e/i40e_common.c | 7 +-
drivers/net/ethernet/intel/i40e/i40e_debugfs.c | 3 +-
drivers/net/ethernet/intel/i40e/i40e_diag.c | 4 +-
drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 8 +-
drivers/net/ethernet/intel/i40e/i40e_main.c | 1462 +++++++++++++++++---
drivers/net/ethernet/intel/i40e/i40e_txrx.h | 2 +
drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 59 +-
.../net/ethernet/intel/i40evf/i40evf_virtchnl.c | 7 +-
include/linux/avf/virtchnl.h | 4 +-
include/net/pkt_cls.h | 9 +
include/uapi/linux/pkt_sched.h | 32 +
net/sched/sch_mqprio.c | 183 ++-
13 files changed, 1571 insertions(+), 253 deletions(-)
--
2.14.2
^ permalink raw reply
* [PATCH net-next v2 4/4] tc-testing: fix the -l argument bug in tdc.py
From: Lucas Bates @ 2017-10-13 21:51 UTC (permalink / raw)
To: davem; +Cc: netdev, xiyou.wangcong, jiri, jhs, aring, mrv, Lucas Bates
In-Reply-To: <1507931485-1814-1-git-send-email-lucasb@mojatatu.com>
This patch fixes a bug in the tdc script, where executing tdc
with the -l argument would cause the tests to start running
as opposed to listing all the known test cases.
Signed-off-by: Lucas Bates <lucasb@mojatatu.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
---
tools/testing/selftests/tc-testing/tdc.py | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/tools/testing/selftests/tc-testing/tdc.py b/tools/testing/selftests/tc-testing/tdc.py
index cd61b78..d2391df 100755
--- a/tools/testing/selftests/tc-testing/tdc.py
+++ b/tools/testing/selftests/tc-testing/tdc.py
@@ -49,7 +49,7 @@ def exec_cmd(command, nsonly=True):
stderr=subprocess.PIPE)
(rawout, serr) = proc.communicate()
- if proc.returncode != 0:
+ if proc.returncode != 0 and len(serr) > 0:
foutput = serr.decode("utf-8")
else:
foutput = rawout.decode("utf-8")
@@ -203,7 +203,7 @@ def set_args(parser):
help='Run tests only from the specified category, or if no category is specified, list known categories.')
parser.add_argument('-f', '--file', type=str,
help='Run tests from the specified file')
- parser.add_argument('-l', '--list', type=str, nargs='?', const="", metavar='CATEGORY',
+ parser.add_argument('-l', '--list', type=str, nargs='?', const="++", metavar='CATEGORY',
help='List all test cases, or those only within the specified category')
parser.add_argument('-s', '--show', type=str, nargs=1, metavar='ID', dest='showID',
help='Display the test case with specified id')
@@ -357,10 +357,10 @@ def set_operation_mode(args):
testcases = get_categorized_testlist(alltests, ucat)
if args.list:
- if (len(args.list) == 0):
+ if (args.list == "++"):
list_test_cases(alltests)
exit(0)
- elif(len(args.list > 0)):
+ elif(len(args.list) > 0):
if (args.list not in ucat):
print("Unknown category " + args.list)
print("Available categories:")
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox