* [PATCH net-next v5 0/2] net/mlx5e: add indr block support in the FT mode
@ 2020-03-23 23:44 wenxu
2020-03-23 23:44 ` [PATCH net-next v5 1/2] net/mlx5e: refactor indr setup block wenxu
2020-03-23 23:44 ` [PATCH net-next v5 2/2] net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support wenxu
0 siblings, 2 replies; 6+ messages in thread
From: wenxu @ 2020-03-23 23:44 UTC (permalink / raw)
To: saeedm; +Cc: paulb, vladbu, netdev
From: wenxu <wenxu@ucloud.cn>
Indr block supported in FT mode can offload the tunnel device in the
flowtables of nftable.
The netfilter patches:
http://patchwork.ozlabs.org/cover/1242812/
Test with mlx driver as following with nft:
ip link add user1 type vrf table 1
ip l set user1 up
ip l set dev mlx_pf0vf0 down
ip l set dev mlx_pf0vf0 master user1
ifconfig mlx_pf0vf0 10.0.0.1/24 up
ifconfig mlx_p0 172.168.152.75/24 up
ip l add dev tun1 type gretap key 1000
ip l set dev tun1 master user1
ifconfig tun1 10.0.1.1/24 up
ip r r 10.0.1.241 encap ip id 1000 dst 172.168.152.241 key dev tun1 table 1
nft add table firewall
nft add chain firewall zones { type filter hook prerouting priority - 300 \; }
nft add rule firewall zones counter ct zone set iif map { "tun1" : 1, "mlx_pf0vf0" : 1 }
nft add chain firewall rule-1000-ingress
nft add rule firewall rule-1000-ingress ct zone 1 ct state established,related counter accept
nft add rule firewall rule-1000-ingress ct zone 1 ct state invalid counter drop
nft add rule firewall rule-1000-ingress ct zone 1 tcp dport 5001 ct state new counter accept
nft add rule firewall rule-1000-ingress ct zone 1 udp dport 5001 ct state new counter accept
nft add rule firewall rule-1000-ingress ct zone 1 tcp dport 22 ct state new counter accept
nft add rule firewall rule-1000-ingress ct zone 1 ip protocol icmp ct state new counter accept
nft add rule firewall rule-1000-ingress counter drop
nft add chain firewall rules-all { type filter hook prerouting priority - 150 \; }
nft add rule firewall rules-all meta iifkind "vrf" counter accept
nft add rule firewall rules-all iif vmap { "tun1" : jump rule-1000-ingress }
nft add flowtable firewall fb1 { hook ingress priority 2 \; devices = { tun1, mlx_pf0vf0 } \; }
nft add chain firewall ftb-all {type filter hook forward priority 0 \; policy accept \; }
nft add rule firewall ftb-all ct zone 1 ip protocol tcp flow offload @fb1
nft add rule firewall ftb-all ct zone 1 ip protocol udp flow offload @fb1
wenxu (2):
net/mlx5e: refactor indr setup block
net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support
drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 94 ++++++++++++++++++------
1 file changed, 73 insertions(+), 21 deletions(-)
--
1.8.3.1
^ permalink raw reply [flat|nested] 6+ messages in thread* [PATCH net-next v5 1/2] net/mlx5e: refactor indr setup block 2020-03-23 23:44 [PATCH net-next v5 0/2] net/mlx5e: add indr block support in the FT mode wenxu @ 2020-03-23 23:44 ` wenxu 2020-03-23 23:44 ` [PATCH net-next v5 2/2] net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support wenxu 1 sibling, 0 replies; 6+ messages in thread From: wenxu @ 2020-03-23 23:44 UTC (permalink / raw) To: saeedm; +Cc: paulb, vladbu, netdev From: wenxu <wenxu@ucloud.cn> Refactor indr setup block for support ft indr setup in the next patch. IThe function mlx5e_rep_indr_offload exposes 'flags' in order set additional flag for FT in next patch. Rename mlx5e_rep_indr_setup_tc_block to mlx5e_rep_indr_setup_block and add flow_setup_cb_t callback parameters in order set the specific callback for FT in next patch. Signed-off-by: wenxu <wenxu@ucloud.cn> --- v5: change the comments drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 42 ++++++++++++------------ 1 file changed, 21 insertions(+), 21 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c index a33d151..057f5f9 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c @@ -694,9 +694,9 @@ static void mlx5e_rep_indr_clean_block_privs(struct mlx5e_rep_priv *rpriv) static int mlx5e_rep_indr_offload(struct net_device *netdev, struct flow_cls_offload *flower, - struct mlx5e_rep_indr_block_priv *indr_priv) + struct mlx5e_rep_indr_block_priv *indr_priv, + unsigned long flags) { - unsigned long flags = MLX5_TC_FLAG(EGRESS) | MLX5_TC_FLAG(ESW_OFFLOAD); struct mlx5e_priv *priv = netdev_priv(indr_priv->rpriv->netdev); int err = 0; @@ -717,20 +717,22 @@ static void mlx5e_rep_indr_clean_block_privs(struct mlx5e_rep_priv *rpriv) return err; } -static int mlx5e_rep_indr_setup_block_cb(enum tc_setup_type type, - void *type_data, void *indr_priv) +static int mlx5e_rep_indr_setup_tc_cb(enum tc_setup_type type, + void *type_data, void *indr_priv) { + unsigned long flags = MLX5_TC_FLAG(EGRESS) | MLX5_TC_FLAG(ESW_OFFLOAD); struct mlx5e_rep_indr_block_priv *priv = indr_priv; switch (type) { case TC_SETUP_CLSFLOWER: - return mlx5e_rep_indr_offload(priv->netdev, type_data, priv); + return mlx5e_rep_indr_offload(priv->netdev, type_data, priv, + flags); default: return -EOPNOTSUPP; } } -static void mlx5e_rep_indr_tc_block_unbind(void *cb_priv) +static void mlx5e_rep_indr_block_unbind(void *cb_priv) { struct mlx5e_rep_indr_block_priv *indr_priv = cb_priv; @@ -741,9 +743,10 @@ static void mlx5e_rep_indr_tc_block_unbind(void *cb_priv) static LIST_HEAD(mlx5e_block_cb_list); static int -mlx5e_rep_indr_setup_tc_block(struct net_device *netdev, - struct mlx5e_rep_priv *rpriv, - struct flow_block_offload *f) +mlx5e_rep_indr_setup_block(struct net_device *netdev, + struct mlx5e_rep_priv *rpriv, + struct flow_block_offload *f, + flow_setup_cb_t *setup_cb) { struct mlx5e_rep_indr_block_priv *indr_priv; struct flow_block_cb *block_cb; @@ -769,9 +772,8 @@ static void mlx5e_rep_indr_tc_block_unbind(void *cb_priv) list_add(&indr_priv->list, &rpriv->uplink_priv.tc_indr_block_priv_list); - block_cb = flow_block_cb_alloc(mlx5e_rep_indr_setup_block_cb, - indr_priv, indr_priv, - mlx5e_rep_indr_tc_block_unbind); + block_cb = flow_block_cb_alloc(setup_cb, indr_priv, indr_priv, + mlx5e_rep_indr_block_unbind); if (IS_ERR(block_cb)) { list_del(&indr_priv->list); kfree(indr_priv); @@ -786,9 +788,7 @@ static void mlx5e_rep_indr_tc_block_unbind(void *cb_priv) if (!indr_priv) return -ENOENT; - block_cb = flow_block_cb_lookup(f->block, - mlx5e_rep_indr_setup_block_cb, - indr_priv); + block_cb = flow_block_cb_lookup(f->block, setup_cb, indr_priv); if (!block_cb) return -ENOENT; @@ -802,13 +802,13 @@ static void mlx5e_rep_indr_tc_block_unbind(void *cb_priv) } static -int mlx5e_rep_indr_setup_tc_cb(struct net_device *netdev, void *cb_priv, - enum tc_setup_type type, void *type_data) +int mlx5e_rep_indr_setup_cb(struct net_device *netdev, void *cb_priv, + enum tc_setup_type type, void *type_data) { switch (type) { case TC_SETUP_BLOCK: - return mlx5e_rep_indr_setup_tc_block(netdev, cb_priv, - type_data); + return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, + mlx5e_rep_indr_setup_tc_cb); default: return -EOPNOTSUPP; } @@ -820,7 +820,7 @@ static int mlx5e_rep_indr_register_block(struct mlx5e_rep_priv *rpriv, int err; err = __flow_indr_block_cb_register(netdev, rpriv, - mlx5e_rep_indr_setup_tc_cb, + mlx5e_rep_indr_setup_cb, rpriv); if (err) { struct mlx5e_priv *priv = netdev_priv(rpriv->netdev); @@ -834,7 +834,7 @@ static int mlx5e_rep_indr_register_block(struct mlx5e_rep_priv *rpriv, static void mlx5e_rep_indr_unregister_block(struct mlx5e_rep_priv *rpriv, struct net_device *netdev) { - __flow_indr_block_cb_unregister(netdev, mlx5e_rep_indr_setup_tc_cb, + __flow_indr_block_cb_unregister(netdev, mlx5e_rep_indr_setup_cb, rpriv); } -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH net-next v5 2/2] net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support 2020-03-23 23:44 [PATCH net-next v5 0/2] net/mlx5e: add indr block support in the FT mode wenxu 2020-03-23 23:44 ` [PATCH net-next v5 1/2] net/mlx5e: refactor indr setup block wenxu @ 2020-03-23 23:44 ` wenxu 2020-03-24 12:05 ` Vlad Buslov 1 sibling, 1 reply; 6+ messages in thread From: wenxu @ 2020-03-23 23:44 UTC (permalink / raw) To: saeedm; +Cc: paulb, vladbu, netdev From: wenxu <wenxu@ucloud.cn> Add mlx5e_rep_indr_setup_ft_cb to support indr block setup in FT mode. Signed-off-by: wenxu <wenxu@ucloud.cn> --- v5: no change drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 52 ++++++++++++++++++++++++ 1 file changed, 52 insertions(+) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c index 057f5f9..30c81c3 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c @@ -732,6 +732,55 @@ static int mlx5e_rep_indr_setup_tc_cb(enum tc_setup_type type, } } +static int mlx5e_rep_indr_setup_ft_cb(enum tc_setup_type type, + void *type_data, void *indr_priv) +{ + struct mlx5e_rep_indr_block_priv *priv = indr_priv; + struct flow_cls_offload *f = type_data; + struct flow_cls_offload tmp; + struct mlx5e_priv *mpriv; + struct mlx5_eswitch *esw; + unsigned long flags; + int err; + + mpriv = netdev_priv(priv->rpriv->netdev); + esw = mpriv->mdev->priv.eswitch; + + flags = MLX5_TC_FLAG(EGRESS) | + MLX5_TC_FLAG(ESW_OFFLOAD) | + MLX5_TC_FLAG(FT_OFFLOAD); + + switch (type) { + case TC_SETUP_CLSFLOWER: + memcpy(&tmp, f, sizeof(*f)); + + if (!mlx5_esw_chains_prios_supported(esw)) + return -EOPNOTSUPP; + + /* Re-use tc offload path by moving the ft flow to the + * reserved ft chain. + * + * FT offload can use prio range [0, INT_MAX], so we normalize + * it to range [1, mlx5_esw_chains_get_prio_range(esw)] + * as with tc, where prio 0 isn't supported. + * + * We only support chain 0 of FT offload. + */ + if (tmp.common.prio >= mlx5_esw_chains_get_prio_range(esw)) + return -EOPNOTSUPP; + if (tmp.common.chain_index != 0) + return -EOPNOTSUPP; + + tmp.common.chain_index = mlx5_esw_chains_get_ft_chain(esw); + tmp.common.prio++; + err = mlx5e_rep_indr_offload(priv->netdev, &tmp, priv, flags); + memcpy(&f->stats, &tmp.stats, sizeof(f->stats)); + return err; + default: + return -EOPNOTSUPP; + } +} + static void mlx5e_rep_indr_block_unbind(void *cb_priv) { struct mlx5e_rep_indr_block_priv *indr_priv = cb_priv; @@ -809,6 +858,9 @@ int mlx5e_rep_indr_setup_cb(struct net_device *netdev, void *cb_priv, case TC_SETUP_BLOCK: return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, mlx5e_rep_indr_setup_tc_cb); + case TC_SETUP_FT: + return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, + mlx5e_rep_indr_setup_ft_cb); default: return -EOPNOTSUPP; } -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH net-next v5 2/2] net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support 2020-03-23 23:44 ` [PATCH net-next v5 2/2] net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support wenxu @ 2020-03-24 12:05 ` Vlad Buslov 2020-03-24 12:49 ` wenxu 0 siblings, 1 reply; 6+ messages in thread From: Vlad Buslov @ 2020-03-24 12:05 UTC (permalink / raw) To: wenxu; +Cc: saeedm, paulb, vladbu, netdev On Tue 24 Mar 2020 at 01:44, wenxu@ucloud.cn wrote: > From: wenxu <wenxu@ucloud.cn> > > Add mlx5e_rep_indr_setup_ft_cb to support indr block setup > in FT mode. > > Signed-off-by: wenxu <wenxu@ucloud.cn> > --- > v5: no change > > drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 52 ++++++++++++++++++++++++ > 1 file changed, 52 insertions(+) > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c > index 057f5f9..30c81c3 100644 > --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c > +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c > @@ -732,6 +732,55 @@ static int mlx5e_rep_indr_setup_tc_cb(enum tc_setup_type type, > } > } > > +static int mlx5e_rep_indr_setup_ft_cb(enum tc_setup_type type, > + void *type_data, void *indr_priv) > +{ > + struct mlx5e_rep_indr_block_priv *priv = indr_priv; > + struct flow_cls_offload *f = type_data; > + struct flow_cls_offload tmp; > + struct mlx5e_priv *mpriv; > + struct mlx5_eswitch *esw; > + unsigned long flags; > + int err; > + > + mpriv = netdev_priv(priv->rpriv->netdev); > + esw = mpriv->mdev->priv.eswitch; > + > + flags = MLX5_TC_FLAG(EGRESS) | > + MLX5_TC_FLAG(ESW_OFFLOAD) | > + MLX5_TC_FLAG(FT_OFFLOAD); > + > + switch (type) { > + case TC_SETUP_CLSFLOWER: > + memcpy(&tmp, f, sizeof(*f)); > + > + if (!mlx5_esw_chains_prios_supported(esw)) > + return -EOPNOTSUPP; > + > + /* Re-use tc offload path by moving the ft flow to the > + * reserved ft chain. > + * > + * FT offload can use prio range [0, INT_MAX], so we normalize > + * it to range [1, mlx5_esw_chains_get_prio_range(esw)] > + * as with tc, where prio 0 isn't supported. > + * > + * We only support chain 0 of FT offload. > + */ > + if (tmp.common.prio >= mlx5_esw_chains_get_prio_range(esw)) > + return -EOPNOTSUPP; > + if (tmp.common.chain_index != 0) > + return -EOPNOTSUPP; > + > + tmp.common.chain_index = mlx5_esw_chains_get_ft_chain(esw); > + tmp.common.prio++; > + err = mlx5e_rep_indr_offload(priv->netdev, &tmp, priv, flags); > + memcpy(&f->stats, &tmp.stats, sizeof(f->stats)); Why do you need to create temporary copy of flow_cls_offload struct and then copy parts of tmp back to the original? Again, this info should probably be in the commit message. > + return err; > + default: > + return -EOPNOTSUPP; > + } > +} > + > static void mlx5e_rep_indr_block_unbind(void *cb_priv) > { > struct mlx5e_rep_indr_block_priv *indr_priv = cb_priv; > @@ -809,6 +858,9 @@ int mlx5e_rep_indr_setup_cb(struct net_device *netdev, void *cb_priv, > case TC_SETUP_BLOCK: > return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, > mlx5e_rep_indr_setup_tc_cb); > + case TC_SETUP_FT: > + return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, > + mlx5e_rep_indr_setup_ft_cb); > default: > return -EOPNOTSUPP; > } ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net-next v5 2/2] net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support 2020-03-24 12:05 ` Vlad Buslov @ 2020-03-24 12:49 ` wenxu 2020-03-24 14:36 ` Vlad Buslov 0 siblings, 1 reply; 6+ messages in thread From: wenxu @ 2020-03-24 12:49 UTC (permalink / raw) To: Vlad Buslov; +Cc: saeedm, paulb, netdev 在 2020/3/24 20:05, Vlad Buslov 写道: > On Tue 24 Mar 2020 at 01:44, wenxu@ucloud.cn wrote: >> From: wenxu <wenxu@ucloud.cn> >> >> Add mlx5e_rep_indr_setup_ft_cb to support indr block setup >> in FT mode. >> >> Signed-off-by: wenxu <wenxu@ucloud.cn> >> --- >> v5: no change >> >> drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 52 ++++++++++++++++++++++++ >> 1 file changed, 52 insertions(+) >> >> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c >> index 057f5f9..30c81c3 100644 >> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c >> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c >> @@ -732,6 +732,55 @@ static int mlx5e_rep_indr_setup_tc_cb(enum tc_setup_type type, >> } >> } >> >> +static int mlx5e_rep_indr_setup_ft_cb(enum tc_setup_type type, >> + void *type_data, void *indr_priv) >> +{ >> + struct mlx5e_rep_indr_block_priv *priv = indr_priv; >> + struct flow_cls_offload *f = type_data; >> + struct flow_cls_offload tmp; >> + struct mlx5e_priv *mpriv; >> + struct mlx5_eswitch *esw; >> + unsigned long flags; >> + int err; >> + >> + mpriv = netdev_priv(priv->rpriv->netdev); >> + esw = mpriv->mdev->priv.eswitch; >> + >> + flags = MLX5_TC_FLAG(EGRESS) | >> + MLX5_TC_FLAG(ESW_OFFLOAD) | >> + MLX5_TC_FLAG(FT_OFFLOAD); >> + >> + switch (type) { >> + case TC_SETUP_CLSFLOWER: >> + memcpy(&tmp, f, sizeof(*f)); >> + >> + if (!mlx5_esw_chains_prios_supported(esw)) >> + return -EOPNOTSUPP; >> + >> + /* Re-use tc offload path by moving the ft flow to the >> + * reserved ft chain. >> + * >> + * FT offload can use prio range [0, INT_MAX], so we normalize >> + * it to range [1, mlx5_esw_chains_get_prio_range(esw)] >> + * as with tc, where prio 0 isn't supported. >> + * >> + * We only support chain 0 of FT offload. >> + */ >> + if (tmp.common.prio >= mlx5_esw_chains_get_prio_range(esw)) >> + return -EOPNOTSUPP; >> + if (tmp.common.chain_index != 0) >> + return -EOPNOTSUPP; >> + >> + tmp.common.chain_index = mlx5_esw_chains_get_ft_chain(esw); >> + tmp.common.prio++; >> + err = mlx5e_rep_indr_offload(priv->netdev, &tmp, priv, flags); >> + memcpy(&f->stats, &tmp.stats, sizeof(f->stats)); > Why do you need to create temporary copy of flow_cls_offload struct and > then copy parts of tmp back to the original? Again, this info should > probably be in the commit message. This FT mode using the specficed chain_index which changed by driver adnd using only in driver. It will move the flow table rules to their steering domain. This scenario just follow the mlx5e_rep_setup_ft_cb which did offload in FT mode. So it's an old scenario, Is it necessary to add to the commit message? > >> + return err; >> + default: >> + return -EOPNOTSUPP; >> + } >> +} >> + >> static void mlx5e_rep_indr_block_unbind(void *cb_priv) >> { >> struct mlx5e_rep_indr_block_priv *indr_priv = cb_priv; >> @@ -809,6 +858,9 @@ int mlx5e_rep_indr_setup_cb(struct net_device *netdev, void *cb_priv, >> case TC_SETUP_BLOCK: >> return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, >> mlx5e_rep_indr_setup_tc_cb); >> + case TC_SETUP_FT: >> + return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, >> + mlx5e_rep_indr_setup_ft_cb); >> default: >> return -EOPNOTSUPP; >> } ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net-next v5 2/2] net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support 2020-03-24 12:49 ` wenxu @ 2020-03-24 14:36 ` Vlad Buslov 0 siblings, 0 replies; 6+ messages in thread From: Vlad Buslov @ 2020-03-24 14:36 UTC (permalink / raw) To: wenxu; +Cc: Vlad Buslov, saeedm, paulb, netdev On Tue 24 Mar 2020 at 14:49, wenxu <wenxu@ucloud.cn> wrote: > 在 2020/3/24 20:05, Vlad Buslov 写道: >> On Tue 24 Mar 2020 at 01:44, wenxu@ucloud.cn wrote: >>> From: wenxu <wenxu@ucloud.cn> >>> >>> Add mlx5e_rep_indr_setup_ft_cb to support indr block setup >>> in FT mode. >>> >>> Signed-off-by: wenxu <wenxu@ucloud.cn> >>> --- >>> v5: no change >>> >>> drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 52 ++++++++++++++++++++++++ >>> 1 file changed, 52 insertions(+) >>> >>> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c >>> index 057f5f9..30c81c3 100644 >>> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c >>> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c >>> @@ -732,6 +732,55 @@ static int mlx5e_rep_indr_setup_tc_cb(enum tc_setup_type type, >>> } >>> } >>> >>> +static int mlx5e_rep_indr_setup_ft_cb(enum tc_setup_type type, >>> + void *type_data, void *indr_priv) >>> +{ >>> + struct mlx5e_rep_indr_block_priv *priv = indr_priv; >>> + struct flow_cls_offload *f = type_data; >>> + struct flow_cls_offload tmp; >>> + struct mlx5e_priv *mpriv; >>> + struct mlx5_eswitch *esw; >>> + unsigned long flags; >>> + int err; >>> + >>> + mpriv = netdev_priv(priv->rpriv->netdev); >>> + esw = mpriv->mdev->priv.eswitch; >>> + >>> + flags = MLX5_TC_FLAG(EGRESS) | >>> + MLX5_TC_FLAG(ESW_OFFLOAD) | >>> + MLX5_TC_FLAG(FT_OFFLOAD); >>> + >>> + switch (type) { >>> + case TC_SETUP_CLSFLOWER: >>> + memcpy(&tmp, f, sizeof(*f)); >>> + >>> + if (!mlx5_esw_chains_prios_supported(esw)) >>> + return -EOPNOTSUPP; >>> + >>> + /* Re-use tc offload path by moving the ft flow to the >>> + * reserved ft chain. >>> + * >>> + * FT offload can use prio range [0, INT_MAX], so we normalize >>> + * it to range [1, mlx5_esw_chains_get_prio_range(esw)] >>> + * as with tc, where prio 0 isn't supported. >>> + * >>> + * We only support chain 0 of FT offload. >>> + */ >>> + if (tmp.common.prio >= mlx5_esw_chains_get_prio_range(esw)) >>> + return -EOPNOTSUPP; >>> + if (tmp.common.chain_index != 0) >>> + return -EOPNOTSUPP; >>> + >>> + tmp.common.chain_index = mlx5_esw_chains_get_ft_chain(esw); >>> + tmp.common.prio++; >>> + err = mlx5e_rep_indr_offload(priv->netdev, &tmp, priv, flags); >>> + memcpy(&f->stats, &tmp.stats, sizeof(f->stats)); >> Why do you need to create temporary copy of flow_cls_offload struct and >> then copy parts of tmp back to the original? Again, this info should >> probably be in the commit message. > > This FT mode using the specficed chain_index which changed by driver adnd > > using only in driver. It will move the flow table rules to their steering > domain. > > This scenario just follow the mlx5e_rep_setup_ft_cb which did offload in FT > mode. > > So it's an old scenario, Is it necessary to add to the commit message? Well, without knowing the reasoning it is hard to understand if the original scenario is applicable in this case. > >> >>> + return err; >>> + default: >>> + return -EOPNOTSUPP; >>> + } >>> +} >>> + >>> static void mlx5e_rep_indr_block_unbind(void *cb_priv) >>> { >>> struct mlx5e_rep_indr_block_priv *indr_priv = cb_priv; >>> @@ -809,6 +858,9 @@ int mlx5e_rep_indr_setup_cb(struct net_device *netdev, void *cb_priv, >>> case TC_SETUP_BLOCK: >>> return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, >>> mlx5e_rep_indr_setup_tc_cb); >>> + case TC_SETUP_FT: >>> + return mlx5e_rep_indr_setup_block(netdev, cb_priv, type_data, >>> + mlx5e_rep_indr_setup_ft_cb); >>> default: >>> return -EOPNOTSUPP; >>> } ^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2020-03-24 14:36 UTC | newest] Thread overview: 6+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2020-03-23 23:44 [PATCH net-next v5 0/2] net/mlx5e: add indr block support in the FT mode wenxu 2020-03-23 23:44 ` [PATCH net-next v5 1/2] net/mlx5e: refactor indr setup block wenxu 2020-03-23 23:44 ` [PATCH net-next v5 2/2] net/mlx5e: add mlx5e_rep_indr_setup_ft_cb support wenxu 2020-03-24 12:05 ` Vlad Buslov 2020-03-24 12:49 ` wenxu 2020-03-24 14:36 ` Vlad Buslov
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).