* [PATCH net-next 8/9] nfp: flower: offload tos and tunnel flags for ipv4 udp tunnels
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, John Hurley
In-Reply-To: <20180630000442.27353-1-jakub.kicinski@netronome.com>
From: John Hurley <john.hurley@netronome.com>
Extract the tos and the tunnel flags from the tunnel key and offload these
action fields. Only the checksum and tunnel key flags are implemented in
fw so reject offloads of other flags. The tunnel key flag is always
considered set in the fw so enforce that it is set in the rule. Note that
the compulsory setting of the tunnel key flag and optional setting of
checksum is inline with how tc currently generates ipv4 udp tunnel
actions.
Signed-off-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
---
drivers/net/ethernet/netronome/nfp/flower/action.c | 9 +++++++++
drivers/net/ethernet/netronome/nfp/flower/cmsg.h | 4 ++--
2 files changed, 11 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/netronome/nfp/flower/action.c b/drivers/net/ethernet/netronome/nfp/flower/action.c
index d421b7fbce96..e56b815a8dc6 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/action.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/action.c
@@ -45,6 +45,8 @@
#include "main.h"
#include "../nfp_net_repr.h"
+#define NFP_FL_SUPPORTED_IPV4_UDP_TUN_FLAGS (TUNNEL_CSUM | TUNNEL_KEY)
+
static void nfp_fl_pop_vlan(struct nfp_fl_pop_vlan *pop_vlan)
{
size_t act_size = sizeof(struct nfp_fl_pop_vlan);
@@ -274,6 +276,13 @@ nfp_fl_set_ipv4_udp_tun(struct nfp_fl_set_ipv4_udp_tun *set_tun,
set_tun->ttl = net->ipv4.sysctl_ip_default_ttl;
}
+ set_tun->tos = ip_tun->key.tos;
+
+ if (!(ip_tun->key.tun_flags & TUNNEL_KEY) ||
+ ip_tun->key.tun_flags & ~NFP_FL_SUPPORTED_IPV4_UDP_TUN_FLAGS)
+ return -EOPNOTSUPP;
+ set_tun->tun_flags = ip_tun->key.tun_flags;
+
/* Complete pre_tunnel action. */
pre_tun->ipv4_dst = ip_tun->key.u.ipv4.dst;
diff --git a/drivers/net/ethernet/netronome/nfp/flower/cmsg.h b/drivers/net/ethernet/netronome/nfp/flower/cmsg.h
index 4a7f3510a296..15f1eacd76b6 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/cmsg.h
+++ b/drivers/net/ethernet/netronome/nfp/flower/cmsg.h
@@ -203,9 +203,9 @@ struct nfp_fl_set_ipv4_udp_tun {
__be16 reserved;
__be64 tun_id __packed;
__be32 tun_type_index;
- __be16 reserved2;
+ __be16 tun_flags;
u8 ttl;
- u8 reserved3;
+ u8 tos;
__be32 extra[2];
};
--
2.17.1
^ permalink raw reply related
* [PATCH net-next 7/9] nfp: flower: extract ipv4 udp tunnel ttl from route
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, John Hurley
In-Reply-To: <20180630000442.27353-1-jakub.kicinski@netronome.com>
From: John Hurley <john.hurley@netronome.com>
Previously the ttl for ipv4 udp tunnels was set to the namespace default.
Modify this to attempt to extract the ttl from a full route lookup on the
tunnel destination. If this is not possible then resort to the default.
Signed-off-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
---
.../ethernet/netronome/nfp/flower/action.c | 19 ++++++++++++++++++-
1 file changed, 18 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/netronome/nfp/flower/action.c b/drivers/net/ethernet/netronome/nfp/flower/action.c
index 61ba8d4f99f1..d421b7fbce96 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/action.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/action.c
@@ -236,9 +236,12 @@ nfp_fl_set_ipv4_udp_tun(struct nfp_fl_set_ipv4_udp_tun *set_tun,
size_t act_size = sizeof(struct nfp_fl_set_ipv4_udp_tun);
struct ip_tunnel_info *ip_tun = tcf_tunnel_info(action);
u32 tmp_set_ip_tun_type_index = 0;
+ struct flowi4 flow = {};
/* Currently support one pre-tunnel so index is always 0. */
int pretun_idx = 0;
+ struct rtable *rt;
struct net *net;
+ int err;
if (ip_tun->options_len)
return -EOPNOTSUPP;
@@ -255,7 +258,21 @@ nfp_fl_set_ipv4_udp_tun(struct nfp_fl_set_ipv4_udp_tun *set_tun,
set_tun->tun_type_index = cpu_to_be32(tmp_set_ip_tun_type_index);
set_tun->tun_id = ip_tun->key.tun_id;
- set_tun->ttl = net->ipv4.sysctl_ip_default_ttl;
+
+ /* Do a route lookup to determine ttl - if fails then use default.
+ * Note that CONFIG_INET is a requirement of CONFIG_NET_SWITCHDEV so
+ * must be defined here.
+ */
+ flow.daddr = ip_tun->key.u.ipv4.dst;
+ flow.flowi4_proto = IPPROTO_UDP;
+ rt = ip_route_output_key(net, &flow);
+ err = PTR_ERR_OR_ZERO(rt);
+ if (!err) {
+ set_tun->ttl = ip4_dst_hoplimit(&rt->dst);
+ ip_rt_put(rt);
+ } else {
+ set_tun->ttl = net->ipv4.sysctl_ip_default_ttl;
+ }
/* Complete pre_tunnel action. */
pre_tun->ipv4_dst = ip_tun->key.u.ipv4.dst;
--
2.17.1
^ permalink raw reply related
* [PATCH net-next 6/9] nfp: flower: ignore checksum actions when performing pedit actions
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, Pieter Jansen van Vuuren
In-Reply-To: <20180630000442.27353-1-jakub.kicinski@netronome.com>
From: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Hardware will automatically update csum in headers when a set action has
been performed. This means we could in the driver ignore the explicit
checksum action when performing a set action.
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
---
.../ethernet/netronome/nfp/flower/action.c | 80 +++++++++++++++++--
1 file changed, 72 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/netronome/nfp/flower/action.c b/drivers/net/ethernet/netronome/nfp/flower/action.c
index 4a6d2db75071..61ba8d4f99f1 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/action.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/action.c
@@ -34,6 +34,7 @@
#include <linux/bitfield.h>
#include <net/pkt_cls.h>
#include <net/switchdev.h>
+#include <net/tc_act/tc_csum.h>
#include <net/tc_act/tc_gact.h>
#include <net/tc_act/tc_mirred.h>
#include <net/tc_act/tc_pedit.h>
@@ -398,8 +399,27 @@ nfp_fl_set_tport(const struct tc_action *action, int idx, u32 off,
return 0;
}
+static u32 nfp_fl_csum_l4_to_flag(u8 ip_proto)
+{
+ switch (ip_proto) {
+ case 0:
+ /* Filter doesn't force proto match,
+ * both TCP and UDP will be updated if encountered
+ */
+ return TCA_CSUM_UPDATE_FLAG_TCP | TCA_CSUM_UPDATE_FLAG_UDP;
+ case IPPROTO_TCP:
+ return TCA_CSUM_UPDATE_FLAG_TCP;
+ case IPPROTO_UDP:
+ return TCA_CSUM_UPDATE_FLAG_UDP;
+ default:
+ /* All other protocols will be ignored by FW */
+ return 0;
+ }
+}
+
static int
-nfp_fl_pedit(const struct tc_action *action, char *nfp_action, int *a_len)
+nfp_fl_pedit(const struct tc_action *action, struct tc_cls_flower_offload *flow,
+ char *nfp_action, int *a_len, u32 *csum_updated)
{
struct nfp_fl_set_ipv6_addr set_ip6_dst, set_ip6_src;
struct nfp_fl_set_ip4_addrs set_ip_addr;
@@ -409,6 +429,7 @@ nfp_fl_pedit(const struct tc_action *action, char *nfp_action, int *a_len)
int idx, nkeys, err;
size_t act_size;
u32 offset, cmd;
+ u8 ip_proto = 0;
memset(&set_ip6_dst, 0, sizeof(set_ip6_dst));
memset(&set_ip6_src, 0, sizeof(set_ip6_src));
@@ -451,6 +472,15 @@ nfp_fl_pedit(const struct tc_action *action, char *nfp_action, int *a_len)
return err;
}
+ if (dissector_uses_key(flow->dissector, FLOW_DISSECTOR_KEY_BASIC)) {
+ struct flow_dissector_key_basic *basic;
+
+ basic = skb_flow_dissector_target(flow->dissector,
+ FLOW_DISSECTOR_KEY_BASIC,
+ flow->key);
+ ip_proto = basic->ip_proto;
+ }
+
if (set_eth.head.len_lw) {
act_size = sizeof(set_eth);
memcpy(nfp_action, &set_eth, act_size);
@@ -459,6 +489,10 @@ nfp_fl_pedit(const struct tc_action *action, char *nfp_action, int *a_len)
act_size = sizeof(set_ip_addr);
memcpy(nfp_action, &set_ip_addr, act_size);
*a_len += act_size;
+
+ /* Hardware will automatically fix IPv4 and TCP/UDP checksum. */
+ *csum_updated |= TCA_CSUM_UPDATE_FLAG_IPV4HDR |
+ nfp_fl_csum_l4_to_flag(ip_proto);
} else if (set_ip6_dst.head.len_lw && set_ip6_src.head.len_lw) {
/* TC compiles set src and dst IPv6 address as a single action,
* the hardware requires this to be 2 separate actions.
@@ -471,18 +505,30 @@ nfp_fl_pedit(const struct tc_action *action, char *nfp_action, int *a_len)
memcpy(&nfp_action[sizeof(set_ip6_src)], &set_ip6_dst,
act_size);
*a_len += act_size;
+
+ /* Hardware will automatically fix TCP/UDP checksum. */
+ *csum_updated |= nfp_fl_csum_l4_to_flag(ip_proto);
} else if (set_ip6_dst.head.len_lw) {
act_size = sizeof(set_ip6_dst);
memcpy(nfp_action, &set_ip6_dst, act_size);
*a_len += act_size;
+
+ /* Hardware will automatically fix TCP/UDP checksum. */
+ *csum_updated |= nfp_fl_csum_l4_to_flag(ip_proto);
} else if (set_ip6_src.head.len_lw) {
act_size = sizeof(set_ip6_src);
memcpy(nfp_action, &set_ip6_src, act_size);
*a_len += act_size;
+
+ /* Hardware will automatically fix TCP/UDP checksum. */
+ *csum_updated |= nfp_fl_csum_l4_to_flag(ip_proto);
} else if (set_tport.head.len_lw) {
act_size = sizeof(set_tport);
memcpy(nfp_action, &set_tport, act_size);
*a_len += act_size;
+
+ /* Hardware will automatically fix TCP/UDP checksum. */
+ *csum_updated |= nfp_fl_csum_l4_to_flag(ip_proto);
}
return 0;
@@ -493,12 +539,18 @@ nfp_flower_output_action(struct nfp_app *app, const struct tc_action *a,
struct nfp_fl_payload *nfp_fl, int *a_len,
struct net_device *netdev, bool last,
enum nfp_flower_tun_type *tun_type, int *tun_out_cnt,
- int *out_cnt)
+ int *out_cnt, u32 *csum_updated)
{
struct nfp_flower_priv *priv = app->priv;
struct nfp_fl_output *output;
int err, prelag_size;
+ /* If csum_updated has not been reset by now, it means HW will
+ * incorrectly update csums when they are not requested.
+ */
+ if (*csum_updated)
+ return -EOPNOTSUPP;
+
if (*a_len + sizeof(struct nfp_fl_output) > NFP_FL_MAX_A_SIZ)
return -EOPNOTSUPP;
@@ -529,10 +581,11 @@ nfp_flower_output_action(struct nfp_app *app, const struct tc_action *a,
static int
nfp_flower_loop_action(struct nfp_app *app, const struct tc_action *a,
+ struct tc_cls_flower_offload *flow,
struct nfp_fl_payload *nfp_fl, int *a_len,
struct net_device *netdev,
enum nfp_flower_tun_type *tun_type, int *tun_out_cnt,
- int *out_cnt)
+ int *out_cnt, u32 *csum_updated)
{
struct nfp_fl_set_ipv4_udp_tun *set_tun;
struct nfp_fl_pre_tunnel *pre_tun;
@@ -545,14 +598,14 @@ nfp_flower_loop_action(struct nfp_app *app, const struct tc_action *a,
} else if (is_tcf_mirred_egress_redirect(a)) {
err = nfp_flower_output_action(app, a, nfp_fl, a_len, netdev,
true, tun_type, tun_out_cnt,
- out_cnt);
+ out_cnt, csum_updated);
if (err)
return err;
} else if (is_tcf_mirred_egress_mirror(a)) {
err = nfp_flower_output_action(app, a, nfp_fl, a_len, netdev,
false, tun_type, tun_out_cnt,
- out_cnt);
+ out_cnt, csum_updated);
if (err)
return err;
@@ -602,8 +655,17 @@ nfp_flower_loop_action(struct nfp_app *app, const struct tc_action *a,
/* Tunnel decap is handled by default so accept action. */
return 0;
} else if (is_tcf_pedit(a)) {
- if (nfp_fl_pedit(a, &nfp_fl->action_data[*a_len], a_len))
+ if (nfp_fl_pedit(a, flow, &nfp_fl->action_data[*a_len],
+ a_len, csum_updated))
return -EOPNOTSUPP;
+ } else if (is_tcf_csum(a)) {
+ /* csum action requests recalc of something we have not fixed */
+ if (tcf_csum_update_flags(a) & ~*csum_updated)
+ return -EOPNOTSUPP;
+ /* If we will correctly fix the csum we can remove it from the
+ * csum update list. Which will later be used to check support.
+ */
+ *csum_updated &= ~tcf_csum_update_flags(a);
} else {
/* Currently we do not handle any other actions. */
return -EOPNOTSUPP;
@@ -620,6 +682,7 @@ int nfp_flower_compile_action(struct nfp_app *app,
int act_len, act_cnt, err, tun_out_cnt, out_cnt;
enum nfp_flower_tun_type tun_type;
const struct tc_action *a;
+ u32 csum_updated = 0;
LIST_HEAD(actions);
memset(nfp_flow->action_data, 0, NFP_FL_MAX_A_SIZ);
@@ -632,8 +695,9 @@ int nfp_flower_compile_action(struct nfp_app *app,
tcf_exts_to_list(flow->exts, &actions);
list_for_each_entry(a, &actions, list) {
- err = nfp_flower_loop_action(app, a, nfp_flow, &act_len, netdev,
- &tun_type, &tun_out_cnt, &out_cnt);
+ err = nfp_flower_loop_action(app, a, flow, nfp_flow, &act_len,
+ netdev, &tun_type, &tun_out_cnt,
+ &out_cnt, &csum_updated);
if (err)
return err;
act_cnt++;
--
2.17.1
^ permalink raw reply related
* [PATCH net-next 5/9] nfp: populate bus-info on representors
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, Jakub Kicinski
In-Reply-To: <20180630000442.27353-1-jakub.kicinski@netronome.com>
We used to leave bus-info in ethtool driver info empty for
representors in case multi-PCIe-to-single-host cards make
the association between PCIe device and NFP many to one.
It seems these attempts are futile, we need to link the
representors to one PCIe device in sysfs to get consistent
naming, plus devlink uses one PCIe as a handle, anyway.
The multi-PCIe-to-single-system support won't be clean,
if it ever comes.
Turns out some user space (RHEL tests) likes to read bus-info
so just populate it.
While at it remove unnecessary app NULL-check, representors
are spawned by an app, so it must exist.
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
---
drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c | 8 +++-----
1 file changed, 3 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
index 2aeb4622f1ea..6a79c8e4a7a4 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
@@ -233,12 +233,10 @@ nfp_net_get_drvinfo(struct net_device *netdev, struct ethtool_drvinfo *drvinfo)
static void
nfp_app_get_drvinfo(struct net_device *netdev, struct ethtool_drvinfo *drvinfo)
{
- struct nfp_app *app;
-
- app = nfp_app_from_netdev(netdev);
- if (!app)
- return;
+ struct nfp_app *app = nfp_app_from_netdev(netdev);
+ strlcpy(drvinfo->bus_info, pci_name(app->pdev),
+ sizeof(drvinfo->bus_info));
nfp_get_drvinfo(app, app->pdev, "*", drvinfo);
}
--
2.17.1
^ permalink raw reply related
* [PATCH net-next 4/9] nfp: make use of napi_consume_skb()
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, Jakub Kicinski
In-Reply-To: <20180630000442.27353-1-jakub.kicinski@netronome.com>
Use napi_consume_skb() in nfp_net_tx_complete() to get bulk free.
Pass 0 as budget for ctrl queue completion since it runs out of
a tasklet.
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
---
drivers/net/ethernet/netronome/nfp/nfp_net_common.c | 11 ++++++-----
1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_common.c b/drivers/net/ethernet/netronome/nfp/nfp_net_common.c
index edc6ef682f6d..7df5ca37bfb8 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_common.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_common.c
@@ -945,11 +945,12 @@ static int nfp_net_tx(struct sk_buff *skb, struct net_device *netdev)
/**
* nfp_net_tx_complete() - Handled completed TX packets
- * @tx_ring: TX ring structure
+ * @tx_ring: TX ring structure
+ * @budget: NAPI budget (only used as bool to determine if in NAPI context)
*
* Return: Number of completed TX descriptors
*/
-static void nfp_net_tx_complete(struct nfp_net_tx_ring *tx_ring)
+static void nfp_net_tx_complete(struct nfp_net_tx_ring *tx_ring, int budget)
{
struct nfp_net_r_vector *r_vec = tx_ring->r_vec;
struct nfp_net_dp *dp = &r_vec->nfp_net->dp;
@@ -999,7 +1000,7 @@ static void nfp_net_tx_complete(struct nfp_net_tx_ring *tx_ring)
/* check for last gather fragment */
if (fidx == nr_frags - 1)
- dev_consume_skb_any(skb);
+ napi_consume_skb(skb, budget);
tx_ring->txbufs[idx].dma_addr = 0;
tx_ring->txbufs[idx].skb = NULL;
@@ -1828,7 +1829,7 @@ static int nfp_net_poll(struct napi_struct *napi, int budget)
unsigned int pkts_polled = 0;
if (r_vec->tx_ring)
- nfp_net_tx_complete(r_vec->tx_ring);
+ nfp_net_tx_complete(r_vec->tx_ring, budget);
if (r_vec->rx_ring)
pkts_polled = nfp_net_rx(r_vec->rx_ring, budget);
@@ -2062,7 +2063,7 @@ static void nfp_ctrl_poll(unsigned long arg)
struct nfp_net_r_vector *r_vec = (void *)arg;
spin_lock_bh(&r_vec->lock);
- nfp_net_tx_complete(r_vec->tx_ring);
+ nfp_net_tx_complete(r_vec->tx_ring, 0);
__nfp_ctrl_tx_queued(r_vec);
spin_unlock_bh(&r_vec->lock);
--
2.17.1
^ permalink raw reply related
* [PATCH net-next 3/9] nfp: implement netpoll ndo (thus enabling netconsole)
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, Jakub Kicinski
In-Reply-To: <20180630000442.27353-1-jakub.kicinski@netronome.com>
NFP NAPI handling will only complete the TXed packets when called
with budget of 0, implement ndo_poll_controller by scheduling NAPI
on all TX queues.
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
---
.../ethernet/netronome/nfp/nfp_net_common.c | 18 ++++++++++++++++++
1 file changed, 18 insertions(+)
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_common.c b/drivers/net/ethernet/netronome/nfp/nfp_net_common.c
index d4c27f849f9b..edc6ef682f6d 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_common.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_common.c
@@ -3115,6 +3115,21 @@ nfp_net_vlan_rx_kill_vid(struct net_device *netdev, __be16 proto, u16 vid)
return nfp_net_reconfig_mbox(nn, NFP_NET_CFG_MBOX_CMD_CTAG_FILTER_KILL);
}
+#ifdef CONFIG_NET_POLL_CONTROLLER
+static void nfp_net_netpoll(struct net_device *netdev)
+{
+ struct nfp_net *nn = netdev_priv(netdev);
+ int i;
+
+ /* nfp_net's NAPIs are statically allocated so even if there is a race
+ * with reconfig path this will simply try to schedule some disabled
+ * NAPI instances.
+ */
+ for (i = 0; i < nn->dp.num_stack_tx_rings; i++)
+ napi_schedule_irqoff(&nn->r_vecs[i].napi);
+}
+#endif
+
static void nfp_net_stat64(struct net_device *netdev,
struct rtnl_link_stats64 *stats)
{
@@ -3482,6 +3497,9 @@ const struct net_device_ops nfp_net_netdev_ops = {
.ndo_get_stats64 = nfp_net_stat64,
.ndo_vlan_rx_add_vid = nfp_net_vlan_rx_add_vid,
.ndo_vlan_rx_kill_vid = nfp_net_vlan_rx_kill_vid,
+#ifdef CONFIG_NET_POLL_CONTROLLER
+ .ndo_poll_controller = nfp_net_netpoll,
+#endif
.ndo_set_vf_mac = nfp_app_set_vf_mac,
.ndo_set_vf_vlan = nfp_app_set_vf_vlan,
.ndo_set_vf_spoofchk = nfp_app_set_vf_spoofchk,
--
2.17.1
^ permalink raw reply related
* [PATCH net-next 2/9] nfp: fail probe if serial or interface id is missing
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, Jakub Kicinski
In-Reply-To: <20180630000442.27353-1-jakub.kicinski@netronome.com>
On some platforms with broken ACPI tables we may not have access
to the Serial Number PCIe capability. This capability is crucial
for us for switchdev operation as we use serial number as switch ID,
and for communication with management FW where interface ID is used.
If we can't determine the Serial Number we have to fail device probe.
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
---
.../netronome/nfp/nfpcore/nfp6000_pcie.c | 16 +++++++++-----
.../ethernet/netronome/nfp/nfpcore/nfp_cpp.h | 4 ++--
.../netronome/nfp/nfpcore/nfp_cppcore.c | 22 ++++++++++++++-----
3 files changed, 28 insertions(+), 14 deletions(-)
diff --git a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp6000_pcie.c b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp6000_pcie.c
index 749655c329b2..c8d0b1016a64 100644
--- a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp6000_pcie.c
+++ b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp6000_pcie.c
@@ -1248,7 +1248,7 @@ static void nfp6000_free(struct nfp_cpp *cpp)
kfree(nfp);
}
-static void nfp6000_read_serial(struct device *dev, u8 *serial)
+static int nfp6000_read_serial(struct device *dev, u8 *serial)
{
struct pci_dev *pdev = to_pci_dev(dev);
int pos;
@@ -1256,25 +1256,29 @@ static void nfp6000_read_serial(struct device *dev, u8 *serial)
pos = pci_find_ext_capability(pdev, PCI_EXT_CAP_ID_DSN);
if (!pos) {
- memset(serial, 0, NFP_SERIAL_LEN);
- return;
+ dev_err(dev, "can't find PCIe Serial Number Capability\n");
+ return -EINVAL;
}
pci_read_config_dword(pdev, pos + 4, ®);
put_unaligned_be16(reg >> 16, serial + 4);
pci_read_config_dword(pdev, pos + 8, ®);
put_unaligned_be32(reg, serial);
+
+ return 0;
}
-static u16 nfp6000_get_interface(struct device *dev)
+static int nfp6000_get_interface(struct device *dev)
{
struct pci_dev *pdev = to_pci_dev(dev);
int pos;
u32 reg;
pos = pci_find_ext_capability(pdev, PCI_EXT_CAP_ID_DSN);
- if (!pos)
- return NFP_CPP_INTERFACE(NFP_CPP_INTERFACE_TYPE_PCI, 0, 0xff);
+ if (!pos) {
+ dev_err(dev, "can't find PCIe Serial Number Capability\n");
+ return -EINVAL;
+ }
pci_read_config_dword(pdev, pos + 4, ®);
diff --git a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_cpp.h b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_cpp.h
index b0da3d436850..c338d539fa96 100644
--- a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_cpp.h
+++ b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_cpp.h
@@ -364,8 +364,8 @@ struct nfp_cpp_operations {
int (*init)(struct nfp_cpp *cpp);
void (*free)(struct nfp_cpp *cpp);
- void (*read_serial)(struct device *dev, u8 *serial);
- u16 (*get_interface)(struct device *dev);
+ int (*read_serial)(struct device *dev, u8 *serial);
+ int (*get_interface)(struct device *dev);
int (*area_init)(struct nfp_cpp_area *area,
u32 dest, unsigned long long address,
diff --git a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_cppcore.c b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_cppcore.c
index ef30597aa319..73de57a09800 100644
--- a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_cppcore.c
+++ b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_cppcore.c
@@ -1163,10 +1163,10 @@ nfp_cpp_from_operations(const struct nfp_cpp_operations *ops,
{
const u32 arm = NFP_CPP_ID(NFP_CPP_TARGET_ARM, NFP_CPP_ACTION_RW, 0);
struct nfp_cpp *cpp;
+ int ifc, err;
u32 mask[2];
u32 xpbaddr;
size_t tgt;
- int err;
cpp = kzalloc(sizeof(*cpp), GFP_KERNEL);
if (!cpp) {
@@ -1176,9 +1176,19 @@ nfp_cpp_from_operations(const struct nfp_cpp_operations *ops,
cpp->op = ops;
cpp->priv = priv;
- cpp->interface = ops->get_interface(parent);
- if (ops->read_serial)
- ops->read_serial(parent, cpp->serial);
+
+ ifc = ops->get_interface(parent);
+ if (ifc < 0) {
+ err = ifc;
+ goto err_free_cpp;
+ }
+ cpp->interface = ifc;
+ if (ops->read_serial) {
+ err = ops->read_serial(parent, cpp->serial);
+ if (err)
+ goto err_free_cpp;
+ }
+
rwlock_init(&cpp->resource_lock);
init_waitqueue_head(&cpp->waitq);
lockdep_set_class(&cpp->resource_lock, &nfp_cpp_resource_lock_key);
@@ -1191,7 +1201,7 @@ nfp_cpp_from_operations(const struct nfp_cpp_operations *ops,
err = device_register(&cpp->dev);
if (err < 0) {
put_device(&cpp->dev);
- goto err_dev;
+ goto err_free_cpp;
}
dev_set_drvdata(&cpp->dev, cpp);
@@ -1238,7 +1248,7 @@ nfp_cpp_from_operations(const struct nfp_cpp_operations *ops,
err_out:
device_unregister(&cpp->dev);
-err_dev:
+err_free_cpp:
kfree(cpp);
err_malloc:
return ERR_PTR(err);
--
2.17.1
^ permalink raw reply related
* [PATCH net-next 1/9] nfp: expose ring stats of inactive rings via ethtool
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, Jakub Kicinski
In-Reply-To: <20180630000442.27353-1-jakub.kicinski@netronome.com>
After user changes the ring count statistics for deactivated
rings disappear from ethtool -S output. This causes loss of
information to the user and means that ethtool stats may not
add up to interface stats. Always expose counters from all
the rings. Note that we allocate at most num_possible_cpus()
rings so number of rings should be reasonable.
The alternative of only listing stats for rings which were
ever in use could be confusing.
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
---
.../ethernet/netronome/nfp/nfp_net_ethtool.c | 50 +++++++------------
1 file changed, 19 insertions(+), 31 deletions(-)
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
index 26d1cc4e2906..2aeb4622f1ea 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
@@ -452,7 +452,7 @@ static unsigned int nfp_vnic_get_sw_stats_count(struct net_device *netdev)
{
struct nfp_net *nn = netdev_priv(netdev);
- return NN_RVEC_GATHER_STATS + nn->dp.num_r_vecs * NN_RVEC_PER_Q_STATS;
+ return NN_RVEC_GATHER_STATS + nn->max_r_vecs * NN_RVEC_PER_Q_STATS;
}
static u8 *nfp_vnic_get_sw_stats_strings(struct net_device *netdev, u8 *data)
@@ -460,7 +460,7 @@ static u8 *nfp_vnic_get_sw_stats_strings(struct net_device *netdev, u8 *data)
struct nfp_net *nn = netdev_priv(netdev);
int i;
- for (i = 0; i < nn->dp.num_r_vecs; i++) {
+ for (i = 0; i < nn->max_r_vecs; i++) {
data = nfp_pr_et(data, "rvec_%u_rx_pkts", i);
data = nfp_pr_et(data, "rvec_%u_tx_pkts", i);
data = nfp_pr_et(data, "rvec_%u_tx_busy", i);
@@ -486,7 +486,7 @@ static u64 *nfp_vnic_get_sw_stats(struct net_device *netdev, u64 *data)
u64 tmp[NN_RVEC_GATHER_STATS];
unsigned int i, j;
- for (i = 0; i < nn->dp.num_r_vecs; i++) {
+ for (i = 0; i < nn->max_r_vecs; i++) {
unsigned int start;
do {
@@ -521,15 +521,13 @@ static u64 *nfp_vnic_get_sw_stats(struct net_device *netdev, u64 *data)
return data;
}
-static unsigned int
-nfp_vnic_get_hw_stats_count(unsigned int rx_rings, unsigned int tx_rings)
+static unsigned int nfp_vnic_get_hw_stats_count(unsigned int num_vecs)
{
- return NN_ET_GLOBAL_STATS_LEN + (rx_rings + tx_rings) * 2;
+ return NN_ET_GLOBAL_STATS_LEN + num_vecs * 4;
}
static u8 *
-nfp_vnic_get_hw_stats_strings(u8 *data, unsigned int rx_rings,
- unsigned int tx_rings, bool repr)
+nfp_vnic_get_hw_stats_strings(u8 *data, unsigned int num_vecs, bool repr)
{
int swap_off, i;
@@ -549,36 +547,29 @@ nfp_vnic_get_hw_stats_strings(u8 *data, unsigned int rx_rings,
for (i = NN_ET_SWITCH_STATS_LEN * 2; i < NN_ET_GLOBAL_STATS_LEN; i++)
data = nfp_pr_et(data, nfp_net_et_stats[i].name);
- for (i = 0; i < tx_rings; i++) {
- data = nfp_pr_et(data, "txq_%u_pkts", i);
- data = nfp_pr_et(data, "txq_%u_bytes", i);
- }
-
- for (i = 0; i < rx_rings; i++) {
+ for (i = 0; i < num_vecs; i++) {
data = nfp_pr_et(data, "rxq_%u_pkts", i);
data = nfp_pr_et(data, "rxq_%u_bytes", i);
+ data = nfp_pr_et(data, "txq_%u_pkts", i);
+ data = nfp_pr_et(data, "txq_%u_bytes", i);
}
return data;
}
static u64 *
-nfp_vnic_get_hw_stats(u64 *data, u8 __iomem *mem,
- unsigned int rx_rings, unsigned int tx_rings)
+nfp_vnic_get_hw_stats(u64 *data, u8 __iomem *mem, unsigned int num_vecs)
{
unsigned int i;
for (i = 0; i < NN_ET_GLOBAL_STATS_LEN; i++)
*data++ = readq(mem + nfp_net_et_stats[i].off);
- for (i = 0; i < tx_rings; i++) {
- *data++ = readq(mem + NFP_NET_CFG_TXR_STATS(i));
- *data++ = readq(mem + NFP_NET_CFG_TXR_STATS(i) + 8);
- }
-
- for (i = 0; i < rx_rings; i++) {
+ for (i = 0; i < num_vecs; i++) {
*data++ = readq(mem + NFP_NET_CFG_RXR_STATS(i));
*data++ = readq(mem + NFP_NET_CFG_RXR_STATS(i) + 8);
+ *data++ = readq(mem + NFP_NET_CFG_TXR_STATS(i));
+ *data++ = readq(mem + NFP_NET_CFG_TXR_STATS(i) + 8);
}
return data;
@@ -633,8 +624,7 @@ static void nfp_net_get_strings(struct net_device *netdev,
switch (stringset) {
case ETH_SS_STATS:
data = nfp_vnic_get_sw_stats_strings(netdev, data);
- data = nfp_vnic_get_hw_stats_strings(data, nn->dp.num_rx_rings,
- nn->dp.num_tx_rings,
+ data = nfp_vnic_get_hw_stats_strings(data, nn->max_r_vecs,
false);
data = nfp_mac_get_stats_strings(netdev, data);
data = nfp_app_port_get_stats_strings(nn->port, data);
@@ -649,8 +639,7 @@ nfp_net_get_stats(struct net_device *netdev, struct ethtool_stats *stats,
struct nfp_net *nn = netdev_priv(netdev);
data = nfp_vnic_get_sw_stats(netdev, data);
- data = nfp_vnic_get_hw_stats(data, nn->dp.ctrl_bar,
- nn->dp.num_rx_rings, nn->dp.num_tx_rings);
+ data = nfp_vnic_get_hw_stats(data, nn->dp.ctrl_bar, nn->max_r_vecs);
data = nfp_mac_get_stats(netdev, data);
data = nfp_app_port_get_stats(nn->port, data);
}
@@ -662,8 +651,7 @@ static int nfp_net_get_sset_count(struct net_device *netdev, int sset)
switch (sset) {
case ETH_SS_STATS:
return nfp_vnic_get_sw_stats_count(netdev) +
- nfp_vnic_get_hw_stats_count(nn->dp.num_rx_rings,
- nn->dp.num_tx_rings) +
+ nfp_vnic_get_hw_stats_count(nn->max_r_vecs) +
nfp_mac_get_stats_count(netdev) +
nfp_app_port_get_stats_count(nn->port);
default:
@@ -679,7 +667,7 @@ static void nfp_port_get_strings(struct net_device *netdev,
switch (stringset) {
case ETH_SS_STATS:
if (nfp_port_is_vnic(port))
- data = nfp_vnic_get_hw_stats_strings(data, 0, 0, true);
+ data = nfp_vnic_get_hw_stats_strings(data, 0, true);
else
data = nfp_mac_get_stats_strings(netdev, data);
data = nfp_app_port_get_stats_strings(port, data);
@@ -694,7 +682,7 @@ nfp_port_get_stats(struct net_device *netdev, struct ethtool_stats *stats,
struct nfp_port *port = nfp_port_from_netdev(netdev);
if (nfp_port_is_vnic(port))
- data = nfp_vnic_get_hw_stats(data, port->vnic, 0, 0);
+ data = nfp_vnic_get_hw_stats(data, port->vnic, 0);
else
data = nfp_mac_get_stats(netdev, data);
data = nfp_app_port_get_stats(port, data);
@@ -708,7 +696,7 @@ static int nfp_port_get_sset_count(struct net_device *netdev, int sset)
switch (sset) {
case ETH_SS_STATS:
if (nfp_port_is_vnic(port))
- count = nfp_vnic_get_hw_stats_count(0, 0);
+ count = nfp_vnic_get_hw_stats_count(0);
else
count = nfp_mac_get_stats_count(netdev);
count += nfp_app_port_get_stats_count(port);
--
2.17.1
^ permalink raw reply related
* [PATCH net-next 0/9] nfp: flower updates and netconsole
From: Jakub Kicinski @ 2018-06-30 0:04 UTC (permalink / raw)
To: davem; +Cc: oss-drivers, netdev, Jakub Kicinski
Hi!
This set contains assorted updates to driver base and flower.
First patch is a follow up to a fix to calculating counters which
went into net. For ethtool counters we should also make sure
they are visible even after ring reconfiguration. Next patch
is a safety measure in case we are running on a machine with a
broken BIOS we should fail the probe when risk of incorrect
operation is too high. The next two patches add netpoll support
and make use of napi_consume_skb(). Last we populate bus info
on all representors.
Pieter adds support for offload of the checksum action in flower.
John follows up to another fix he's done in net, we set TTL
values on tunnels to stack's default, now Johns does a full
route lookup to get a more precise information, he populates
ToS field as well. Last but not least he follows up on Jiri's
request to enable LAG offload in case the team driver is used
and then hash function is unknown.
Jakub Kicinski (5):
nfp: expose ring stats of inactive rings via ethtool
nfp: fail probe if serial or interface id is missing
nfp: implement netpoll ndo (thus enabling netconsole)
nfp: make use of napi_consume_skb()
nfp: populate bus-info on representors
John Hurley (3):
nfp: flower: extract ipv4 udp tunnel ttl from route
nfp: flower: offload tos and tunnel flags for ipv4 udp tunnels
nfp: flower: enabled offloading of Team LAG
Pieter Jansen van Vuuren (1):
nfp: flower: ignore checksum actions when performing pedit actions
.../ethernet/netronome/nfp/flower/action.c | 108 ++++++++++++++++--
.../net/ethernet/netronome/nfp/flower/cmsg.h | 4 +-
.../ethernet/netronome/nfp/flower/lag_conf.c | 5 +-
.../ethernet/netronome/nfp/nfp_net_common.c | 29 ++++-
.../ethernet/netronome/nfp/nfp_net_ethtool.c | 58 ++++------
.../netronome/nfp/nfpcore/nfp6000_pcie.c | 16 ++-
.../ethernet/netronome/nfp/nfpcore/nfp_cpp.h | 4 +-
.../netronome/nfp/nfpcore/nfp_cppcore.c | 22 +++-
8 files changed, 178 insertions(+), 68 deletions(-)
--
2.17.1
^ permalink raw reply
* Re: [PATCH v2 net-next 0/2] net: preserve sock reference when scrubbing the skb.
From: Cong Wang @ 2018-06-30 0:15 UTC (permalink / raw)
To: David Miller
Cc: Flavio Leitner, Linux Kernel Network Developers, Eric Dumazet,
Paolo Abeni, Florian Westphal, NetFilter
In-Reply-To: <20180629.112235.970317820691624358.davem@davemloft.net>
On Thu, Jun 28, 2018 at 7:22 PM David Miller <davem@davemloft.net> wrote:
>
> From: Cong Wang <xiyou.wangcong@gmail.com>
> Date: Thu, 28 Jun 2018 14:53:09 -0700
>
> > I will send a revert with quote of the above.
>
> And it will go to /dev/null as far as I am concerned. I read it the
> first time, so posting it again will not change my opinion of what you
> have to say.
David, you claim you read it, now tell me, where is "cgroups" or "cpu"
from?
This is the link of my reply you quoted:
https://marc.info/?l=linux-netdev&m=153013948711582&w=2
I did mention cgroups to Eric because of isolation, the softnet_data
is per-CPU, and CPU is not isolated by netns apparently, therefore
sd->input_pkt_queue can't be totally isolated for netns without cpuset.
But this is never the reason why I dislike it, this is why I never even
mentioned it in the link above.
>
> Cong, you really need to calm down and understand that people perhaps
> simply fundamentally disagree with you.
1. Eric's "forwarding to eth0" is missing, never brought up until in his
private reply. Without this information, XPS makes no sense at all in
this patchset. For the record, I provide a different solution for Eric.
2. No one responses to:
https://marc.info/?l=linux-netdev&m=153013948711582&w=2
I never expect you agree with me on all of them, but no one
gives me any response to my concerns.
3. I will write a blog post to draw you some pictures, since it
is so hard to understand the isolation...
^ permalink raw reply
* Re: [PATCH bpf 3/3] bpf: undo prog rejection on read-only lock failure
From: Daniel Borkmann @ 2018-06-29 23:47 UTC (permalink / raw)
To: Kees Cook; +Cc: Alexei Starovoitov, Network Development, Laura Abbott
In-Reply-To: <CAGXu5j+SAurTmBVgtABjOBmbHUczjWbV=DwmTc9xQseTthVwtw@mail.gmail.com>
On 06/29/2018 08:42 PM, Kees Cook wrote:
> On Thu, Jun 28, 2018 at 2:34 PM, Daniel Borkmann <daniel@iogearbox.net> wrote:
>> Kees suggested that if set_memory_*() can fail, we should annotate it with
>> __must_check, and all callers need to deal with it gracefully given those
>> set_memory_*() markings aren't "advisory", but they're expected to actually
>> do what they say. This might be an option worth to move forward in future
>> but would at the same time require that set_memory_*() calls from supporting
>> archs are guaranteed to be "atomic" in that they provide rollback if part
>> of the range fails, once that happened, the transition from RW -> RO could
>> be made more robust that way, while subsequent RO -> RW transition /must/
>> continue guaranteeing to always succeed the undo part.
>
> Does this mean we can have BPF filters that aren't read-only then?
> What's the situation where set_memory_ro() fails? (Can it be induced
> by the user?)
My understanding is that the cpa_process_alias() would attempt to also change
attributes of physmap ranges, and it found that a large page had to be split
for this but failed in doing so thus attributes couldn't be updated there due
to page alloc error. Attempting to change the primary mapping which would be
directly the addr passed to set_memory_ro() was however set to read-only
despite error. While for reproduction I had a toggle on the alloc_pages() in
split_large_page() to have it fail, I only could trigger it occasionally; I
used the selftest suite in a loop to stress test and it hit about or twice
over hours.
Thanks,
Daniel
^ permalink raw reply
* Re: [PATCH net-next] openvswitch: kernel datapath clone action
From: Pravin Shelar @ 2018-06-29 22:08 UTC (permalink / raw)
To: Yifeng Sun; +Cc: Andy Zhou, Linux Kernel Network Developers
In-Reply-To: <1530199208-11687-1-git-send-email-pkusunyifeng@gmail.com>
On Thu, Jun 28, 2018 at 8:20 AM, Yifeng Sun <pkusunyifeng@gmail.com> wrote:
> Add 'clone' action to kernel datapath by using existing functions.
> When actions within clone don't modify the current flow, the flow
> key is not cloned before executing clone actions.
>
> This is a follow up patch for this incomplete work:
> https://patchwork.ozlabs.org/patch/722096/
>
> Signed-off-by: Yifeng Sun <pkusunyifeng@gmail.com>
> Signed-off-by: Andy Zhou <azhou@ovn.org>
> ---
> include/uapi/linux/openvswitch.h | 8 +++++
> net/openvswitch/actions.c | 33 ++++++++++++++++++
> net/openvswitch/flow_netlink.c | 73 ++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 114 insertions(+)
>
> diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
> index 863aaba..5de8583 100644
> --- a/include/uapi/linux/openvswitch.h
> +++ b/include/uapi/linux/openvswitch.h
> @@ -625,6 +625,11 @@ struct sample_arg {
> * 'OVS_SAMPLE_ATTR_PROBABILITY'.
> */
> };
> +
> +#define OVS_CLONE_ATTR_EXEC 0 /* Specify an u32 value. When nonzero,
> + * actions in clone will not change flow
> + * keys. False otherwise.
> + */
> #endif
This symbol is used only in datapath, so we can move it to kernel
headers from uapi.
> diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
> index 30a5df2..4444e31 100644
> --- a/net/openvswitch/actions.c
> +++ b/net/openvswitch/actions.c
> @@ -1057,6 +1057,28 @@ static int sample(struct datapath *dp, struct sk_buff *skb,
> clone_flow_key);
> }
>
> +/* When 'last' is true, clone() should always consume the 'skb'.
> + * Otherwise, clone() should keep 'skb' intact regardless what
> + * actions are executed within clone().
> + */
> +static int clone(struct datapath *dp, struct sk_buff *skb,
> + struct sw_flow_key *key, const struct nlattr *attr,
> + bool last)
> +{
> + struct nlattr *actions;
> + struct nlattr *clone_arg;
> + int rem = nla_len(attr);
> + bool clone_flow_key;
> +
> + /* The first action is always 'OVS_CLONE_ATTR_ARG'. */
> + clone_arg = nla_data(attr);
> + clone_flow_key = !nla_get_u32(clone_arg);
> + actions = nla_next(clone_arg, &rem);
> +
Since OVS_CLONE_ATTR_EXEC means do not clone the key, it can be named
accordingly.
^ permalink raw reply
* Re: [PATCH v3 net-next 7/9] net: ipv4: listified version of ip_rcv
From: kbuild test robot @ 2018-06-29 22:44 UTC (permalink / raw)
To: Edward Cree; +Cc: kbuild-all, davem, netdev
In-Reply-To: <419314b8-1fa0-15e6-794d-5ef8874597a2@solarflare.com>
[-- Attachment #1: Type: text/plain, Size: 2917 bytes --]
Hi Edward,
I love your patch! Yet something to improve:
[auto build test ERROR on net-next/master]
url: https://github.com/0day-ci/linux/commits/Edward-Cree/Handle-multiple-received-packets-at-each-stage/20180630-042204
config: x86_64-randconfig-x003-201825 (attached as .config)
compiler: gcc-7 (Debian 7.3.0-16) 7.3.0
reproduce:
# save the attached .config to linux build tree
make ARCH=x86_64
All errors (new ones prefixed by >>):
net/ipv4/ip_input.c: In function 'ip_sublist_rcv':
>> net/ipv4/ip_input.c:524:14: error: passing argument 6 of 'NF_HOOK_LIST' from incompatible pointer type [-Werror=incompatible-pointer-types]
head, dev, NULL, ip_rcv_finish);
^~~
In file included from include/uapi/linux/netfilter_ipv4.h:9:0,
from include/linux/netfilter_ipv4.h:7,
from net/ipv4/ip_input.c:145:
include/linux/netfilter.h:387:1: note: expected 'struct list_head *' but argument is of type 'struct net_device *'
NF_HOOK_LIST(uint8_t pf, unsigned int hook, struct net *net, struct sock *sk,
^~~~~~~~~~~~
net/ipv4/ip_input.c:524:25: error: passing argument 8 of 'NF_HOOK_LIST' from incompatible pointer type [-Werror=incompatible-pointer-types]
head, dev, NULL, ip_rcv_finish);
^~~~~~~~~~~~~
In file included from include/uapi/linux/netfilter_ipv4.h:9:0,
from include/linux/netfilter_ipv4.h:7,
from net/ipv4/ip_input.c:145:
include/linux/netfilter.h:387:1: note: expected 'struct net_device *' but argument is of type 'int (*)(struct net *, struct sock *, struct sk_buff *)'
NF_HOOK_LIST(uint8_t pf, unsigned int hook, struct net *net, struct sock *sk,
^~~~~~~~~~~~
>> net/ipv4/ip_input.c:523:2: error: too few arguments to function 'NF_HOOK_LIST'
NF_HOOK_LIST(NFPROTO_IPV4, NF_INET_PRE_ROUTING, net, NULL,
^~~~~~~~~~~~
In file included from include/uapi/linux/netfilter_ipv4.h:9:0,
from include/linux/netfilter_ipv4.h:7,
from net/ipv4/ip_input.c:145:
include/linux/netfilter.h:387:1: note: declared here
NF_HOOK_LIST(uint8_t pf, unsigned int hook, struct net *net, struct sock *sk,
^~~~~~~~~~~~
cc1: some warnings being treated as errors
vim +/NF_HOOK_LIST +524 net/ipv4/ip_input.c
517
518 static void ip_sublist_rcv(struct list_head *head, struct net_device *dev,
519 struct net *net)
520 {
521 struct sk_buff *skb, *next;
522
> 523 NF_HOOK_LIST(NFPROTO_IPV4, NF_INET_PRE_ROUTING, net, NULL,
> 524 head, dev, NULL, ip_rcv_finish);
525 list_for_each_entry_safe(skb, next, head, list)
526 ip_rcv_finish(net, NULL, skb);
527 }
528
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 32495 bytes --]
^ permalink raw reply
* Re: [net-next 01/12] net/mlx5e: Add UDP GSO support
From: Willem de Bruijn @ 2018-06-29 22:19 UTC (permalink / raw)
To: Saeed Mahameed
Cc: David Miller, Network Development, ogerlitz, borisp, yossiku,
Alexander Duyck
In-Reply-To: <20180628215103.9141-2-saeedm@mellanox.com>
On Fri, Jun 29, 2018 at 2:24 AM Saeed Mahameed <saeedm@mellanox.com> wrote:
>
> From: Boris Pismenny <borisp@mellanox.com>
>
> This patch enables UDP GSO support. We enable this by using two WQEs
> the first is a UDP LSO WQE for all segments with equal length, and the
> second is for the last segment in case it has different length.
> Due to HW limitation, before sending, we must adjust the packet length fields.
>
> We measure performance between two Intel(R) Xeon(R) CPU E5-2643 v2 @3.50GHz
> machines connected back-to-back with Connectx4-Lx (40Gbps) NICs.
> We compare single stream UDP, UDP GSO and UDP GSO with offload.
> Performance:
> | MSS (bytes) | Throughput (Gbps) | CPU utilization (%)
> UDP GSO offload | 1472 | 35.6 | 8%
> UDP GSO | 1472 | 25.5 | 17%
> UDP | 1472 | 10.2 | 17%
> UDP GSO offload | 1024 | 35.6 | 8%
> UDP GSO | 1024 | 19.2 | 17%
> UDP | 1024 | 5.7 | 17%
> UDP GSO offload | 512 | 33.8 | 16%
> UDP GSO | 512 | 10.4 | 17%
> UDP | 512 | 3.5 | 17%
Very nice results :)
> +static void mlx5e_udp_gso_prepare_last_skb(struct sk_buff *skb,
> + struct sk_buff *nskb,
> + int remaining)
> +{
> + int bytes_needed = remaining, remaining_headlen, remaining_page_offset;
> + int headlen = skb_transport_offset(skb) + sizeof(struct udphdr);
> + int payload_len = remaining + sizeof(struct udphdr);
> + int k = 0, i, j;
> +
> + skb_copy_bits(skb, 0, nskb->data, headlen);
> + nskb->dev = skb->dev;
> + skb_reset_mac_header(nskb);
> + skb_set_network_header(nskb, skb_network_offset(skb));
> + skb_set_transport_header(nskb, skb_transport_offset(skb));
> + skb_set_tail_pointer(nskb, headlen);
> +
> + /* How many frags do we need? */
> + for (i = skb_shinfo(skb)->nr_frags - 1; i >= 0; i--) {
> + bytes_needed -= skb_frag_size(&skb_shinfo(skb)->frags[i]);
> + k++;
> + if (bytes_needed <= 0)
> + break;
> + }
> +
> + /* Fill the first frag and split it if necessary */
> + j = skb_shinfo(skb)->nr_frags - k;
> + remaining_page_offset = -bytes_needed;
> + skb_fill_page_desc(nskb, 0,
> + skb_shinfo(skb)->frags[j].page.p,
> + skb_shinfo(skb)->frags[j].page_offset + remaining_page_offset,
> + skb_shinfo(skb)->frags[j].size - remaining_page_offset);
> +
> + skb_frag_ref(skb, j);
> +
> + /* Fill the rest of the frags */
> + for (i = 1; i < k; i++) {
> + j = skb_shinfo(skb)->nr_frags - k + i;
> +
> + skb_fill_page_desc(nskb, i,
> + skb_shinfo(skb)->frags[j].page.p,
> + skb_shinfo(skb)->frags[j].page_offset,
> + skb_shinfo(skb)->frags[j].size);
> + skb_frag_ref(skb, j);
> + }
> + skb_shinfo(nskb)->nr_frags = k;
> +
> + remaining_headlen = remaining - skb->data_len;
> +
> + /* headlen contains remaining data? */
> + if (remaining_headlen > 0)
> + skb_copy_bits(skb, skb->len - remaining, nskb->data + headlen,
> + remaining_headlen);
> + nskb->len = remaining + headlen;
> + nskb->data_len = payload_len - sizeof(struct udphdr) +
> + max_t(int, 0, remaining_headlen);
> + nskb->protocol = skb->protocol;
> + if (nskb->protocol == htons(ETH_P_IP)) {
> + ip_hdr(nskb)->id = htons(ntohs(ip_hdr(nskb)->id) +
> + skb_shinfo(skb)->gso_segs);
> + ip_hdr(nskb)->tot_len =
> + htons(payload_len + sizeof(struct iphdr));
> + } else {
> + ipv6_hdr(nskb)->payload_len = htons(payload_len);
> + }
> + udp_hdr(nskb)->len = htons(payload_len);
> + skb_shinfo(nskb)->gso_size = 0;
> + nskb->ip_summed = skb->ip_summed;
> + nskb->csum_start = skb->csum_start;
> + nskb->csum_offset = skb->csum_offset;
> + nskb->queue_mapping = skb->queue_mapping;
> +}
> +
> +/* might send skbs and update wqe and pi */
> +struct sk_buff *mlx5e_udp_gso_handle_tx_skb(struct net_device *netdev,
> + struct mlx5e_txqsq *sq,
> + struct sk_buff *skb,
> + struct mlx5e_tx_wqe **wqe,
> + u16 *pi)
> +{
> + int payload_len = skb_shinfo(skb)->gso_size + sizeof(struct udphdr);
> + int headlen = skb_transport_offset(skb) + sizeof(struct udphdr);
> + int remaining = (skb->len - headlen) % skb_shinfo(skb)->gso_size;
> + struct sk_buff *nskb;
> +
> + if (skb->protocol == htons(ETH_P_IP))
> + ip_hdr(skb)->tot_len = htons(payload_len + sizeof(struct iphdr));
> + else
> + ipv6_hdr(skb)->payload_len = htons(payload_len);
> + udp_hdr(skb)->len = htons(payload_len);
> + if (!remaining)
> + return skb;
> +
> + nskb = alloc_skb(max_t(int, headlen, headlen + remaining - skb->data_len), GFP_ATOMIC);
> + if (unlikely(!nskb)) {
> + sq->stats->dropped++;
> + return NULL;
> + }
> +
> + mlx5e_udp_gso_prepare_last_skb(skb, nskb, remaining);
> +
> + skb_shinfo(skb)->gso_segs--;
> + pskb_trim(skb, skb->len - remaining);
> + mlx5e_sq_xmit(sq, skb, *wqe, *pi);
> + mlx5e_sq_fetch_wqe(sq, wqe, pi);
> + return nskb;
> +}
The device driver seems to be implementing the packet split here
similar to NETIF_F_GSO_PARTIAL. When advertising the right flag, the
stack should be able to do that for you and pass two packets to the
driver.
^ permalink raw reply
* Re: [patch net-next v2 0/9] net: sched: introduce chain templates support with offloading to mlxsw
From: Cong Wang @ 2018-06-29 22:18 UTC (permalink / raw)
To: sridhar.samudrala
Cc: Jiri Pirko, David Ahern, Jamal Hadi Salim,
Linux Kernel Network Developers, David Miller, Jakub Kicinski,
Simon Horman, john.hurley, mlxsw
In-Reply-To: <b0c7a083-5a15-336a-c30a-8ee785a06684@intel.com>
On Fri, Jun 29, 2018 at 10:06 AM Samudrala, Sridhar
<sridhar.samudrala@intel.com> wrote:
>
> So instead of introducing 'chaintemplate' object in the kernel, can't we add 'chain'
> object in the kernel that takes the 'template' as an attribute?
This is exactly what I mean above. Making the chain a standalone object
in kernel would benefit:
1. Align with 'tc chain' in iproute2, add/del an object is natural
2. Template is an attribute of this object when creating it:
# tc chain add template ....
# tc chain add ... # non-template chain
3. Easier for sharing by qdiscs:
# tc chain add X block Y ...
# tc filter add ... chain X block Y ...
# tc qdisc add dev eth0 block Y ...
The current 'ingress_block 22 ingress' syntax is ugly.
^ permalink raw reply
* Re: [PATCH] [v2] infiniband: i40iw, nes: don't use wall time for TCP sequence numbers
From: Shiraz Saleem @ 2018-06-29 22:10 UTC (permalink / raw)
To: Arnd Bergmann
Cc: Latif, Faisal, Doug Ledford, Jason Gunthorpe, David S. Miller,
Orosco, Henry, Nikolova, Tatyana E, Ismail, Mustafa,
Bart Van Assche, Yuval Shaia, linux-rdma@vger.kernel.org,
linux-kernel@vger.kernel.org, netdev@vger.kernel.org
In-Reply-To: <20180627132628.915978-1-arnd@arndb.de>
On Wed, Jun 27, 2018 at 07:26:05AM -0600, Arnd Bergmann wrote:
> The nes infiniband driver uses current_kernel_time() to get a nanosecond
> granunarity timestamp to initialize its tcp sequence counters. This is
> one of only a few remaining users of that deprecated function, so we
> should try to get rid of it.
>
> Aside from using a deprecated API, there are several problems I see here:
>
> - Using a CLOCK_REALTIME based time source makes it predictable in
> case the time base is synchronized.
> - Using a coarse timestamp means it only gets updated once per jiffie,
> making it even more predictable in order to avoid having to access
> the hardware clock source
> - The upper 2 bits are always zero because the nanoseconds are at most
> 999999999.
>
> For the Linux TCP implementation, we use secure_tcp_seq(), which appears
> to be appropriate here as well, and solves all the above problems.
>
> i40iw uses a variant of the same code, so I do that same thing there
> for ipv4. Unlike nes, i40e also supports ipv6, which needs to call
> secure_tcpv6_seq instead.
>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
> v2: use secure_tcpv6_seq for IPv6 support as suggested by Shiraz Saleem.
> ---
Looks good.
Acked-by: Shiraz Saleem <shiraz.saleem@intel.com>
^ permalink raw reply
* Anyone know if strongswan works with vrf?
From: Ben Greear @ 2018-06-29 22:10 UTC (permalink / raw)
To: netdev
Hello,
We're trying to create lots of strongswan VPN tunnels on network devices
bound to different VRFs. We are using Fedora-24 on the client side, with a 4.16.15+ kernel
and updated 'ip' package, etc.
So far, no luck getting it to work.
Any idea if this is supported or not?
Thanks,
Ben
--
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc http://www.candelatech.com
^ permalink raw reply
* Re: [PATCH v3 net-next 7/9] net: ipv4: listified version of ip_rcv
From: kbuild test robot @ 2018-06-29 22:08 UTC (permalink / raw)
To: Edward Cree; +Cc: kbuild-all, davem, netdev
In-Reply-To: <419314b8-1fa0-15e6-794d-5ef8874597a2@solarflare.com>
[-- Attachment #1: Type: text/plain, Size: 2754 bytes --]
Hi Edward,
I love your patch! Perhaps something to improve:
[auto build test WARNING on net-next/master]
url: https://github.com/0day-ci/linux/commits/Edward-Cree/Handle-multiple-received-packets-at-each-stage/20180630-042204
config: i386-randconfig-a0-201825 (attached as .config)
compiler: gcc-4.9 (Debian 4.9.4-2) 4.9.4
reproduce:
# save the attached .config to linux build tree
make ARCH=i386
All warnings (new ones prefixed by >>):
net//ipv4/ip_input.c: In function 'ip_sublist_rcv':
>> net//ipv4/ip_input.c:524:14: warning: passing argument 6 of 'NF_HOOK_LIST' from incompatible pointer type
head, dev, NULL, ip_rcv_finish);
^
In file included from include/uapi/linux/netfilter_ipv4.h:9:0,
from include/linux/netfilter_ipv4.h:7,
from net//ipv4/ip_input.c:145:
include/linux/netfilter.h:387:1: note: expected 'struct list_head *' but argument is of type 'struct net_device *'
NF_HOOK_LIST(uint8_t pf, unsigned int hook, struct net *net, struct sock *sk,
^
net//ipv4/ip_input.c:524:25: warning: passing argument 8 of 'NF_HOOK_LIST' from incompatible pointer type
head, dev, NULL, ip_rcv_finish);
^
In file included from include/uapi/linux/netfilter_ipv4.h:9:0,
from include/linux/netfilter_ipv4.h:7,
from net//ipv4/ip_input.c:145:
include/linux/netfilter.h:387:1: note: expected 'struct net_device *' but argument is of type 'int (*)(struct net *, struct sock *, struct sk_buff *)'
NF_HOOK_LIST(uint8_t pf, unsigned int hook, struct net *net, struct sock *sk,
^
net//ipv4/ip_input.c:523:2: error: too few arguments to function 'NF_HOOK_LIST'
NF_HOOK_LIST(NFPROTO_IPV4, NF_INET_PRE_ROUTING, net, NULL,
^
In file included from include/uapi/linux/netfilter_ipv4.h:9:0,
from include/linux/netfilter_ipv4.h:7,
from net//ipv4/ip_input.c:145:
include/linux/netfilter.h:387:1: note: declared here
NF_HOOK_LIST(uint8_t pf, unsigned int hook, struct net *net, struct sock *sk,
^
vim +/NF_HOOK_LIST +524 net//ipv4/ip_input.c
517
518 static void ip_sublist_rcv(struct list_head *head, struct net_device *dev,
519 struct net *net)
520 {
521 struct sk_buff *skb, *next;
522
523 NF_HOOK_LIST(NFPROTO_IPV4, NF_INET_PRE_ROUTING, net, NULL,
> 524 head, dev, NULL, ip_rcv_finish);
525 list_for_each_entry_safe(skb, next, head, list)
526 ip_rcv_finish(net, NULL, skb);
527 }
528
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 27668 bytes --]
^ permalink raw reply
* [PATCH ipsec-next] xfrm: Allow Set Mark to be Updated Using UPDSA
From: Nathan Harold @ 2018-06-29 22:07 UTC (permalink / raw)
To: netdev; +Cc: Nathan Harold
Allow UPDSA to change "set mark" to permit
policy separation of packet routing decisions from
SA keying in systems that use mark-based routing.
The set mark, used as a routing and firewall mark
for outbound packets, is made update-able which
allows routing decisions to be handled independently
of keying/SA creation. To maintain consistency with
other optional attributes, the set mark is only
updated if sent with a non-zero value.
The per-SA lock and the xfrm_state_lock are taken in
that order to avoid a deadlock with
xfrm_timer_handler(), which also takes the locks in
that order.
Signed-off-by: Nathan Harold <nharold@google.com>
Change-Id: Ia05c6733a94c1901cd1e54eb7c7e237704678d71
---
net/xfrm/xfrm_state.c | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c
index e04a510ec992..c9ffcdfa89f6 100644
--- a/net/xfrm/xfrm_state.c
+++ b/net/xfrm/xfrm_state.c
@@ -1562,6 +1562,15 @@ int xfrm_state_update(struct xfrm_state *x)
if (x1->curlft.use_time)
xfrm_state_check_expire(x1);
+ if (x->props.smark.m || x->props.smark.v) {
+ spin_lock_bh(&net->xfrm.xfrm_state_lock);
+
+ x1->props.smark = x->props.smark;
+
+ __xfrm_state_bump_genids(x1);
+ spin_unlock_bh(&net->xfrm.xfrm_state_lock);
+ }
+
err = 0;
x->km.state = XFRM_STATE_DEAD;
__xfrm_state_put(x);
--
2.18.0.399.gad0ab374a1-goog
^ permalink raw reply related
* Re: [PATCH] atmel: using strlcpy() to avoid possible buffer overflows
From: Andy Shevchenko @ 2018-06-29 21:51 UTC (permalink / raw)
To: YueHaibing
Cc: simon, Kalle Valo, Linux Kernel Mailing List, netdev,
open list:TI WILINK WIRELES..., David S. Miller
In-Reply-To: <CAHp75Vet-vTz_Ld=Lturmm8NSpK1etvoj-7Wzn4-h=bM2FQgZg@mail.gmail.com>
On Sat, Jun 30, 2018 at 12:47 AM, Andy Shevchenko
<andy.shevchenko@gmail.com> wrote:
> On Fri, Jun 29, 2018 at 5:51 AM, YueHaibing <yuehaibing@huawei.com> wrote:
>> 'firmware' is a module param which may been longer than firmware_id,
>> so using strlcpy() to guard against overflows
>
> strncat() is against overflow, this does a bit more.
>
>> priv->firmware_id[0] = '\0';
> ...
>> if (firmware) /* module parameter */
>> - strcpy(priv->firmware_id, firmware);
>> + strlcpy(priv->firmware_id, firmware, sizeof(priv->firmware_id));
>
> In either case the above '\0' is not needed.
> But it looks like the intention was to use strncat() / strlcat().
Ah, this is under condition, yes. If no parameter supplied, this needs
to be clean, but
priv is allocated with zeroed memory
https://elixir.bootlin.com/linux/latest/source/net/core/dev.c#L8369
--
With Best Regards,
Andy Shevchenko
^ permalink raw reply
* Re: [PATCH] atmel: using strlcpy() to avoid possible buffer overflows
From: Andy Shevchenko @ 2018-06-29 21:47 UTC (permalink / raw)
To: YueHaibing
Cc: simon, Kalle Valo, Linux Kernel Mailing List, netdev,
open list:TI WILINK WIRELES..., David S. Miller
In-Reply-To: <20180629025135.12236-1-yuehaibing@huawei.com>
On Fri, Jun 29, 2018 at 5:51 AM, YueHaibing <yuehaibing@huawei.com> wrote:
> 'firmware' is a module param which may been longer than firmware_id,
> so using strlcpy() to guard against overflows
strncat() is against overflow, this does a bit more.
> priv->firmware_id[0] = '\0';
...
> if (firmware) /* module parameter */
> - strcpy(priv->firmware_id, firmware);
> + strlcpy(priv->firmware_id, firmware, sizeof(priv->firmware_id));
In either case the above '\0' is not needed.
But it looks like the intention was to use strncat() / strlcat().
--
With Best Regards,
Andy Shevchenko
^ permalink raw reply
* Re: [Patch net] net: use dev_change_tx_queue_len() for SIOCSIFTXQLEN
From: Eric Dumazet @ 2018-06-29 21:36 UTC (permalink / raw)
To: Cong Wang, netdev; +Cc: edumazet
In-Reply-To: <20180629204248.28919-1-xiyou.wangcong@gmail.com>
On 06/29/2018 01:42 PM, Cong Wang wrote:
> As noticed by Eric, we need to switch to the helper
> dev_change_tx_queue_len() for SIOCSIFTXQLEN call path too,
> otheriwse still miss dev_qdisc_change_tx_queue_len().
>
> Fixes: 6a643ddb5624 ("net: introduce helper dev_change_tx_queue_len()")
> Reported-by: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
> ---
>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Thanks !
^ permalink raw reply
* [PATCH net] net: fib_rules: bring back rule_exists to match rule during add
From: Roopa Prabhu @ 2018-06-29 21:32 UTC (permalink / raw)
To: davem; +Cc: netdev, dsa
From: Roopa Prabhu <roopa@cumulusnetworks.com>
After commit f9d4b0c1e969 ("fib_rules: move common handling of newrule
delrule msgs into fib_nl2rule"), rule_exists got replaced by rule_find
for existing rule lookup in both the add and del paths. While this
is good for the delete path, it solves a few problems but opens up
a few invalid key matches in the add path.
$ip -4 rule add table main tos 10 fwmark 1
$ip -4 rule add table main tos 10
RTNETLINK answers: File exists
The problem here is rule_find does not check if the key masks in
the new and old rule are the same and hence ends up matching a more
secific rule. Rule key masks cannot be easily compared today without
an elaborate if-else block. Its best to introduce key masks for easier
and accurate rule comparison in the future. Until then, due to fear of
regressions this patch re-introduces older loose rule_exists during add.
Also fixes both rule_exists and rule_find to cover missing attributes.
Fixes: f9d4b0c1e969 ("fib_rules: move common handling of newrule delrule msgs into fib_nl2rule")
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
---
This does bring back old problems with rule_exists. But at this hour, old problems
are better than new problems!. The pref compare in rule_exists can be fixed, but its best
done in net-next. I will look at adding more coverage in fib rule selftests. Thanks.
net/core/fib_rules.c | 72 +++++++++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 71 insertions(+), 1 deletion(-)
diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index bc8425d..f64aa13 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c
@@ -444,6 +444,9 @@ static struct fib_rule *rule_find(struct fib_rules_ops *ops,
if (rule->ip_proto && r->ip_proto != rule->ip_proto)
continue;
+ if (rule->proto && r->proto != rule->proto)
+ continue;
+
if (fib_rule_port_range_set(&rule->sport_range) &&
!fib_rule_port_range_compare(&r->sport_range,
&rule->sport_range))
@@ -653,6 +656,73 @@ static int fib_nl2rule(struct sk_buff *skb, struct nlmsghdr *nlh,
return err;
}
+static int rule_exists(struct fib_rules_ops *ops, struct fib_rule_hdr *frh,
+ struct nlattr **tb, struct fib_rule *rule)
+{
+ struct fib_rule *r;
+
+ list_for_each_entry(r, &ops->rules_list, list) {
+ if (r->action != rule->action)
+ continue;
+
+ if (r->table != rule->table)
+ continue;
+
+ if (r->pref != rule->pref)
+ continue;
+
+ if (memcmp(r->iifname, rule->iifname, IFNAMSIZ))
+ continue;
+
+ if (memcmp(r->oifname, rule->oifname, IFNAMSIZ))
+ continue;
+
+ if (r->mark != rule->mark)
+ continue;
+
+ if (r->suppress_ifgroup != rule->suppress_ifgroup)
+ continue;
+
+ if (r->suppress_prefixlen != rule->suppress_prefixlen)
+ continue;
+
+ if (r->mark_mask != rule->mark_mask)
+ continue;
+
+ if (r->tun_id != rule->tun_id)
+ continue;
+
+ if (r->fr_net != rule->fr_net)
+ continue;
+
+ if (r->l3mdev != rule->l3mdev)
+ continue;
+
+ if (!uid_eq(r->uid_range.start, rule->uid_range.start) ||
+ !uid_eq(r->uid_range.end, rule->uid_range.end))
+ continue;
+
+ if (r->ip_proto != rule->ip_proto)
+ continue;
+
+ if (r->proto != rule->proto)
+ continue;
+
+ if (!fib_rule_port_range_compare(&r->sport_range,
+ &rule->sport_range))
+ continue;
+
+ if (!fib_rule_port_range_compare(&r->dport_range,
+ &rule->dport_range))
+ continue;
+
+ if (!ops->compare(r, frh, tb))
+ continue;
+ return 1;
+ }
+ return 0;
+}
+
int fib_nl_newrule(struct sk_buff *skb, struct nlmsghdr *nlh,
struct netlink_ext_ack *extack)
{
@@ -687,7 +757,7 @@ int fib_nl_newrule(struct sk_buff *skb, struct nlmsghdr *nlh,
goto errout;
if ((nlh->nlmsg_flags & NLM_F_EXCL) &&
- rule_find(ops, frh, tb, rule, user_priority)) {
+ rule_exists(ops, frh, tb, rule)) {
err = -EEXIST;
goto errout_free;
}
--
2.1.4
^ permalink raw reply related
* Re: [PATCH v6 2/3] drivers core: prepare device_shutdown for multi-threading
From: Pavel Tatashin @ 2018-06-29 21:20 UTC (permalink / raw)
To: Andy Shevchenko
Cc: Steven Sistare, Daniel Jordan, Linux Kernel Mailing List,
Kirsher, Jeffrey T, intel-wired-lan, netdev, Greg Kroah-Hartman,
Alexander Duyck, tobin
In-Reply-To: <CAHp75VdQsrxS5=Jb_+Wfu=JL0-AR1B=p8Jp3psEqH5ApmpewGA@mail.gmail.com>
On 06/29/2018 04:38 PM, Andy Shevchenko wrote:
> On Fri, Jun 29, 2018 at 9:25 PM, Pavel Tatashin
> <pasha.tatashin@oracle.com> wrote:
>> Do all the necessary refactoring to prepare device_shutdown() logic to
>> be multi-threaded.
>>
>> Which includes:
>> 1. Change the direction of traversing the list instead of going backward,
>> we now go forward.
>> 2. Children are shutdown recursively for each root device from bottom-up.
>> 3. Functions that can be multi-threaded have _task() in their name.
>
>> +/*
>> + * device_children_count - device children count
>> + * @parent: parent struct device.
>> + *
>> + * Returns number of children for this device or 0 if none.
>> + */
>> +static int device_children_count(struct device *parent)
>> +{
>> + struct klist_iter i;
>> + int children = 0;
>> +
>> + if (!parent->p)
>> + return 0;
>> +
>> + klist_iter_init(&parent->p->klist_children, &i);
>> + while (next_device(&i))
>> + children++;
>> + klist_iter_exit(&i);
>> +
>> + return children;
>> +}
>
> Looking at another patch in LKML (for gluedir children counting) I
> would suggest to consider to cache children value.
> Would it make sense?
>
> (So, basically you would update that value on children registered / removed)
>
Hi Andy,
Thank you for looking at this patch.
For this particular series, this function is called only once per device during shutdown, when nothing else is running on the machine. So, caching won't really provide us any improvement.
If, in the future this function is going to be called from other sites during run-time of the machine, then yes, that would mean that caching might help.
Thank you,
Pavel
^ permalink raw reply
* [PATCH net] hv_netvsc: split sub-channel setup into async and sync
From: Haiyang Zhang @ 2018-06-29 21:07 UTC (permalink / raw)
To: davem, netdev
Cc: haiyangz, kys, sthemmin, olaf, vkuznets, devel, linux-kernel
From: Stephen Hemminger <sthemmin@microsoft.com>
When doing device hotplug the sub channel must be async to avoid
deadlock issues because device is discovered in softirq context.
When doing changes to MTU and number of channels, the setup
must be synchronous to avoid races such as when MTU and device
settings are done in a single ip command.
Reported-by: Thomas Walker <Thomas.Walker@twosigma.com>
Fixes: 8195b1396ec8 ("hv_netvsc: fix deadlock on hotplug")
Fixes: 732e49850c5e ("netvsc: fix race on sub channel creation")
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
---
drivers/net/hyperv/hyperv_net.h | 2 +-
drivers/net/hyperv/netvsc.c | 37 ++++++++++++++++++-
drivers/net/hyperv/netvsc_drv.c | 17 ++++++++-
drivers/net/hyperv/rndis_filter.c | 61 ++++++-------------------------
4 files changed, 65 insertions(+), 52 deletions(-)
diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h
index 1a924b867b07..4b6e308199d2 100644
--- a/drivers/net/hyperv/hyperv_net.h
+++ b/drivers/net/hyperv/hyperv_net.h
@@ -210,7 +210,7 @@ int netvsc_recv_callback(struct net_device *net,
void netvsc_channel_cb(void *context);
int netvsc_poll(struct napi_struct *napi, int budget);
-void rndis_set_subchannel(struct work_struct *w);
+int rndis_set_subchannel(struct net_device *ndev, struct netvsc_device *nvdev);
int rndis_filter_open(struct netvsc_device *nvdev);
int rndis_filter_close(struct netvsc_device *nvdev);
struct netvsc_device *rndis_filter_device_add(struct hv_device *dev,
diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c
index 5d5bd513847f..8e9d0ee1572b 100644
--- a/drivers/net/hyperv/netvsc.c
+++ b/drivers/net/hyperv/netvsc.c
@@ -65,6 +65,41 @@ void netvsc_switch_datapath(struct net_device *ndev, bool vf)
VM_PKT_DATA_INBAND, 0);
}
+/* Worker to setup sub channels on initial setup
+ * Initial hotplug event occurs in softirq context
+ * and can't wait for channels.
+ */
+static void netvsc_subchan_work(struct work_struct *w)
+{
+ struct netvsc_device *nvdev =
+ container_of(w, struct netvsc_device, subchan_work);
+ struct rndis_device *rdev;
+ int i, ret;
+
+ /* Avoid deadlock with device removal already under RTNL */
+ if (!rtnl_trylock()) {
+ schedule_work(w);
+ return;
+ }
+
+ rdev = nvdev->extension;
+ if (rdev) {
+ ret = rndis_set_subchannel(rdev->ndev, nvdev);
+ if (ret == 0) {
+ netif_device_attach(rdev->ndev);
+ } else {
+ /* fallback to only primary channel */
+ for (i = 1; i < nvdev->num_chn; i++)
+ netif_napi_del(&nvdev->chan_table[i].napi);
+
+ nvdev->max_chn = 1;
+ nvdev->num_chn = 1;
+ }
+ }
+
+ rtnl_unlock();
+}
+
static struct netvsc_device *alloc_net_device(void)
{
struct netvsc_device *net_device;
@@ -81,7 +116,7 @@ static struct netvsc_device *alloc_net_device(void)
init_completion(&net_device->channel_init_wait);
init_waitqueue_head(&net_device->subchan_open);
- INIT_WORK(&net_device->subchan_work, rndis_set_subchannel);
+ INIT_WORK(&net_device->subchan_work, netvsc_subchan_work);
return net_device;
}
diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c
index fe2256bf1d13..dd1d6e115145 100644
--- a/drivers/net/hyperv/netvsc_drv.c
+++ b/drivers/net/hyperv/netvsc_drv.c
@@ -905,8 +905,20 @@ static int netvsc_attach(struct net_device *ndev,
if (IS_ERR(nvdev))
return PTR_ERR(nvdev);
- /* Note: enable and attach happen when sub-channels setup */
+ if (nvdev->num_chn > 1) {
+ ret = rndis_set_subchannel(ndev, nvdev);
+
+ /* if unavailable, just proceed with one queue */
+ if (ret) {
+ nvdev->max_chn = 1;
+ nvdev->num_chn = 1;
+ }
+ }
+
+ /* In any case device is now ready */
+ netif_device_attach(ndev);
+ /* Note: enable and attach happen when sub-channels setup */
netif_carrier_off(ndev);
if (netif_running(ndev)) {
@@ -2089,6 +2101,9 @@ static int netvsc_probe(struct hv_device *dev,
memcpy(net->dev_addr, device_info.mac_adr, ETH_ALEN);
+ if (nvdev->num_chn > 1)
+ schedule_work(&nvdev->subchan_work);
+
/* hw_features computed in rndis_netdev_set_hwcaps() */
net->features = net->hw_features |
NETIF_F_HIGHDMA | NETIF_F_SG |
diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c
index 5428bb261102..9b4e3c3787e5 100644
--- a/drivers/net/hyperv/rndis_filter.c
+++ b/drivers/net/hyperv/rndis_filter.c
@@ -1062,29 +1062,15 @@ static void netvsc_sc_open(struct vmbus_channel *new_sc)
* This breaks overlap of processing the host message for the
* new primary channel with the initialization of sub-channels.
*/
-void rndis_set_subchannel(struct work_struct *w)
+int rndis_set_subchannel(struct net_device *ndev, struct netvsc_device *nvdev)
{
- struct netvsc_device *nvdev
- = container_of(w, struct netvsc_device, subchan_work);
struct nvsp_message *init_packet = &nvdev->channel_init_pkt;
- struct net_device_context *ndev_ctx;
- struct rndis_device *rdev;
- struct net_device *ndev;
- struct hv_device *hv_dev;
+ struct net_device_context *ndev_ctx = netdev_priv(ndev);
+ struct hv_device *hv_dev = ndev_ctx->device_ctx;
+ struct rndis_device *rdev = nvdev->extension;
int i, ret;
- if (!rtnl_trylock()) {
- schedule_work(w);
- return;
- }
-
- rdev = nvdev->extension;
- if (!rdev)
- goto unlock; /* device was removed */
-
- ndev = rdev->ndev;
- ndev_ctx = netdev_priv(ndev);
- hv_dev = ndev_ctx->device_ctx;
+ ASSERT_RTNL();
memset(init_packet, 0, sizeof(struct nvsp_message));
init_packet->hdr.msg_type = NVSP_MSG5_TYPE_SUBCHANNEL;
@@ -1100,13 +1086,13 @@ void rndis_set_subchannel(struct work_struct *w)
VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED);
if (ret) {
netdev_err(ndev, "sub channel allocate send failed: %d\n", ret);
- goto failed;
+ return ret;
}
wait_for_completion(&nvdev->channel_init_wait);
if (init_packet->msg.v5_msg.subchn_comp.status != NVSP_STAT_SUCCESS) {
netdev_err(ndev, "sub channel request failed\n");
- goto failed;
+ return -EIO;
}
nvdev->num_chn = 1 +
@@ -1125,21 +1111,7 @@ void rndis_set_subchannel(struct work_struct *w)
for (i = 0; i < VRSS_SEND_TAB_SIZE; i++)
ndev_ctx->tx_table[i] = i % nvdev->num_chn;
- netif_device_attach(ndev);
- rtnl_unlock();
- return;
-
-failed:
- /* fallback to only primary channel */
- for (i = 1; i < nvdev->num_chn; i++)
- netif_napi_del(&nvdev->chan_table[i].napi);
-
- nvdev->max_chn = 1;
- nvdev->num_chn = 1;
-
- netif_device_attach(ndev);
-unlock:
- rtnl_unlock();
+ return 0;
}
static int rndis_netdev_set_hwcaps(struct rndis_device *rndis_device,
@@ -1360,21 +1332,12 @@ struct netvsc_device *rndis_filter_device_add(struct hv_device *dev,
netif_napi_add(net, &net_device->chan_table[i].napi,
netvsc_poll, NAPI_POLL_WEIGHT);
- if (net_device->num_chn > 1)
- schedule_work(&net_device->subchan_work);
+ return net_device;
out:
- /* if unavailable, just proceed with one queue */
- if (ret) {
- net_device->max_chn = 1;
- net_device->num_chn = 1;
- }
-
- /* No sub channels, device is ready */
- if (net_device->num_chn == 1)
- netif_device_attach(net);
-
- return net_device;
+ /* setting up multiple channels failed */
+ net_device->max_chn = 1;
+ net_device->num_chn = 1;
err_dev_remv:
rndis_filter_device_remove(dev, net_device);
--
2.17.1
^ 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