2 * Copyright (c) 2015, Mellanox Technologies, Ltd. All rights reserved.
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
33 #ifndef __MLX5_ESWITCH_H__
34 #define __MLX5_ESWITCH_H__
36 #include <linux/if_ether.h>
37 #include <linux/if_link.h>
38 #include <net/devlink.h>
39 #include <linux/mlx5/device.h>
40 #include <linux/mlx5/eswitch.h>
41 #include <linux/mlx5/vport.h>
42 #include <linux/mlx5/fs.h>
45 #ifdef CONFIG_MLX5_ESWITCH
47 #define MLX5_MAX_UC_PER_VPORT(dev) \
48 (1 << MLX5_CAP_GEN(dev, log_max_current_uc_list))
50 #define MLX5_MAX_MC_PER_VPORT(dev) \
51 (1 << MLX5_CAP_GEN(dev, log_max_current_mc_list))
53 #define MLX5_MIN_BW_SHARE 1
55 #define MLX5_RATE_TO_BW_SHARE(rate, divider, limit) \
56 min_t(u32, max_t(u32, (rate) / (divider), MLX5_MIN_BW_SHARE), limit)
58 #define mlx5_esw_has_fwd_fdb(dev) \
59 MLX5_CAP_ESW_FLOWTABLE(dev, fdb_multi_path_to_table)
61 #define FDB_MAX_CHAIN 3
62 #define FDB_SLOW_PATH_CHAIN (FDB_MAX_CHAIN + 1)
63 #define FDB_MAX_PRIO 16
65 struct vport_ingress
{
66 struct mlx5_flow_table
*acl
;
67 struct mlx5_flow_group
*allow_untagged_spoofchk_grp
;
68 struct mlx5_flow_group
*allow_spoofchk_only_grp
;
69 struct mlx5_flow_group
*allow_untagged_only_grp
;
70 struct mlx5_flow_group
*drop_grp
;
71 struct mlx5_flow_handle
*allow_rule
;
72 struct mlx5_flow_handle
*drop_rule
;
73 struct mlx5_fc
*drop_counter
;
77 struct mlx5_flow_table
*acl
;
78 struct mlx5_flow_group
*allowed_vlans_grp
;
79 struct mlx5_flow_group
*drop_grp
;
80 struct mlx5_flow_handle
*allowed_vlan
;
81 struct mlx5_flow_handle
*drop_rule
;
82 struct mlx5_fc
*drop_counter
;
85 struct mlx5_vport_drop_stats
{
90 struct mlx5_vport_info
{
103 struct mlx5_core_dev
*dev
;
105 struct hlist_head uc_list
[MLX5_L2_ADDR_HASH_SIZE
];
106 struct hlist_head mc_list
[MLX5_L2_ADDR_HASH_SIZE
];
107 struct mlx5_flow_handle
*promisc_rule
;
108 struct mlx5_flow_handle
*allmulti_rule
;
109 struct work_struct vport_change_handler
;
111 struct vport_ingress ingress
;
112 struct vport_egress egress
;
114 struct mlx5_vport_info info
;
126 enum offloads_fdb_flags
{
127 ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED
= BIT(0),
130 extern const unsigned int ESW_POOLS
[4];
132 #define PRIO_LEVELS 2
133 struct mlx5_eswitch_fdb
{
136 struct mlx5_flow_table
*fdb
;
137 struct mlx5_flow_group
*addr_grp
;
138 struct mlx5_flow_group
*allmulti_grp
;
139 struct mlx5_flow_group
*promisc_grp
;
140 struct mlx5_flow_table
*vepa_fdb
;
141 struct mlx5_flow_handle
*vepa_uplink_rule
;
142 struct mlx5_flow_handle
*vepa_star_rule
;
145 struct offloads_fdb
{
146 struct mlx5_flow_table
*slow_fdb
;
147 struct mlx5_flow_group
*send_to_vport_grp
;
148 struct mlx5_flow_group
*peer_miss_grp
;
149 struct mlx5_flow_handle
**peer_miss_rules
;
150 struct mlx5_flow_group
*miss_grp
;
151 struct mlx5_flow_handle
*miss_rule_uni
;
152 struct mlx5_flow_handle
*miss_rule_multi
;
153 int vlan_push_pop_refcount
;
156 struct mlx5_flow_table
*fdb
;
158 } fdb_prio
[FDB_MAX_CHAIN
+ 1][FDB_MAX_PRIO
+ 1][PRIO_LEVELS
];
159 /* Protects fdb_prio table */
160 struct mutex fdb_prio_lock
;
162 int fdb_left
[ARRAY_SIZE(ESW_POOLS
)];
168 struct mlx5_esw_offload
{
169 struct mlx5_flow_table
*ft_offloads
;
170 struct mlx5_flow_group
*vport_rx_group
;
171 struct mlx5_eswitch_rep
*vport_reps
;
172 struct list_head peer_flows
;
173 struct mutex peer_mutex
;
174 DECLARE_HASHTABLE(encap_tbl
, 8);
175 DECLARE_HASHTABLE(mod_hdr_tbl
, 8);
176 DECLARE_HASHTABLE(termtbl_tbl
, 8);
177 struct mutex termtbl_mutex
; /* protects termtbl hash */
178 const struct mlx5_eswitch_rep_ops
*rep_ops
[NUM_REP_TYPES
];
184 /* E-Switch MC FDB table hash node */
185 struct esw_mc_addr
{ /* SRIOV only */
186 struct l2addr_node node
;
187 struct mlx5_flow_handle
*uplink_rule
; /* Forward to uplink rule */
191 struct mlx5_host_work
{
192 struct work_struct work
;
193 struct mlx5_eswitch
*esw
;
196 struct mlx5_esw_functions
{
201 struct mlx5_eswitch
{
202 struct mlx5_core_dev
*dev
;
204 struct mlx5_eswitch_fdb fdb_table
;
205 struct hlist_head mc_table
[MLX5_L2_ADDR_HASH_SIZE
];
206 struct workqueue_struct
*work_queue
;
207 struct mlx5_vport
*vports
;
210 /* Synchronize between vport change events
211 * and async SRIOV admin state changes
213 struct mutex state_lock
;
214 struct esw_mc_addr mc_promisc
;
221 struct mlx5_esw_offload offloads
;
225 struct mlx5_esw_functions esw_funcs
;
228 void esw_offloads_cleanup(struct mlx5_eswitch
*esw
);
229 int esw_offloads_init(struct mlx5_eswitch
*esw
, int vf_nvports
,
231 void esw_offloads_cleanup_reps(struct mlx5_eswitch
*esw
);
232 int esw_offloads_init_reps(struct mlx5_eswitch
*esw
);
233 void esw_vport_cleanup_ingress_rules(struct mlx5_eswitch
*esw
,
234 struct mlx5_vport
*vport
);
235 int esw_vport_enable_ingress_acl(struct mlx5_eswitch
*esw
,
236 struct mlx5_vport
*vport
);
237 void esw_vport_cleanup_egress_rules(struct mlx5_eswitch
*esw
,
238 struct mlx5_vport
*vport
);
239 int esw_vport_enable_egress_acl(struct mlx5_eswitch
*esw
,
240 struct mlx5_vport
*vport
);
241 void esw_vport_disable_egress_acl(struct mlx5_eswitch
*esw
,
242 struct mlx5_vport
*vport
);
243 void esw_vport_disable_ingress_acl(struct mlx5_eswitch
*esw
,
244 struct mlx5_vport
*vport
);
247 int mlx5_eswitch_init(struct mlx5_core_dev
*dev
);
248 void mlx5_eswitch_cleanup(struct mlx5_eswitch
*esw
);
249 int mlx5_eswitch_enable_sriov(struct mlx5_eswitch
*esw
, int nvfs
, int mode
);
250 void mlx5_eswitch_disable_sriov(struct mlx5_eswitch
*esw
);
251 int mlx5_eswitch_set_vport_mac(struct mlx5_eswitch
*esw
,
252 u16 vport
, u8 mac
[ETH_ALEN
]);
253 int mlx5_eswitch_set_vport_state(struct mlx5_eswitch
*esw
,
254 u16 vport
, int link_state
);
255 int mlx5_eswitch_set_vport_vlan(struct mlx5_eswitch
*esw
,
256 u16 vport
, u16 vlan
, u8 qos
);
257 int mlx5_eswitch_set_vport_spoofchk(struct mlx5_eswitch
*esw
,
258 u16 vport
, bool spoofchk
);
259 int mlx5_eswitch_set_vport_trust(struct mlx5_eswitch
*esw
,
260 u16 vport_num
, bool setting
);
261 int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch
*esw
, u16 vport
,
262 u32 max_rate
, u32 min_rate
);
263 int mlx5_eswitch_set_vepa(struct mlx5_eswitch
*esw
, u8 setting
);
264 int mlx5_eswitch_get_vepa(struct mlx5_eswitch
*esw
, u8
*setting
);
265 int mlx5_eswitch_get_vport_config(struct mlx5_eswitch
*esw
,
266 u16 vport
, struct ifla_vf_info
*ivi
);
267 int mlx5_eswitch_get_vport_stats(struct mlx5_eswitch
*esw
,
269 struct ifla_vf_stats
*vf_stats
);
270 void mlx5_eswitch_del_send_to_vport_rule(struct mlx5_flow_handle
*rule
);
272 struct mlx5_flow_spec
;
273 struct mlx5_esw_flow_attr
;
274 struct mlx5_termtbl_handle
;
277 mlx5_eswitch_termtbl_required(struct mlx5_eswitch
*esw
,
278 struct mlx5_flow_act
*flow_act
,
279 struct mlx5_flow_spec
*spec
);
281 struct mlx5_flow_handle
*
282 mlx5_eswitch_add_termtbl_rule(struct mlx5_eswitch
*esw
,
283 struct mlx5_flow_table
*ft
,
284 struct mlx5_flow_spec
*spec
,
285 struct mlx5_esw_flow_attr
*attr
,
286 struct mlx5_flow_act
*flow_act
,
287 struct mlx5_flow_destination
*dest
,
291 mlx5_eswitch_termtbl_put(struct mlx5_eswitch
*esw
,
292 struct mlx5_termtbl_handle
*tt
);
294 struct mlx5_flow_handle
*
295 mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch
*esw
,
296 struct mlx5_flow_spec
*spec
,
297 struct mlx5_esw_flow_attr
*attr
);
298 struct mlx5_flow_handle
*
299 mlx5_eswitch_add_fwd_rule(struct mlx5_eswitch
*esw
,
300 struct mlx5_flow_spec
*spec
,
301 struct mlx5_esw_flow_attr
*attr
);
303 mlx5_eswitch_del_offloaded_rule(struct mlx5_eswitch
*esw
,
304 struct mlx5_flow_handle
*rule
,
305 struct mlx5_esw_flow_attr
*attr
);
307 mlx5_eswitch_del_fwd_rule(struct mlx5_eswitch
*esw
,
308 struct mlx5_flow_handle
*rule
,
309 struct mlx5_esw_flow_attr
*attr
);
312 mlx5_eswitch_prios_supported(struct mlx5_eswitch
*esw
);
315 mlx5_eswitch_get_prio_range(struct mlx5_eswitch
*esw
);
318 mlx5_eswitch_get_chain_range(struct mlx5_eswitch
*esw
);
320 struct mlx5_flow_handle
*
321 mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch
*esw
, u16 vport
,
322 struct mlx5_flow_destination
*dest
);
325 SET_VLAN_STRIP
= BIT(0),
326 SET_VLAN_INSERT
= BIT(1)
329 enum mlx5_flow_match_level
{
330 MLX5_MATCH_NONE
= MLX5_INLINE_MODE_NONE
,
331 MLX5_MATCH_L2
= MLX5_INLINE_MODE_L2
,
332 MLX5_MATCH_L3
= MLX5_INLINE_MODE_IP
,
333 MLX5_MATCH_L4
= MLX5_INLINE_MODE_TCP_UDP
,
336 /* current maximum for flow based vport multicasting */
337 #define MLX5_MAX_FLOW_FWD_VPORTS 2
340 MLX5_ESW_DEST_ENCAP
= BIT(0),
341 MLX5_ESW_DEST_ENCAP_VALID
= BIT(1),
344 struct mlx5_esw_flow_attr
{
345 struct mlx5_eswitch_rep
*in_rep
;
346 struct mlx5_core_dev
*in_mdev
;
347 struct mlx5_core_dev
*counter_dev
;
353 __be16 vlan_proto
[MLX5_FS_VLAN_DEPTH
];
354 u16 vlan_vid
[MLX5_FS_VLAN_DEPTH
];
355 u8 vlan_prio
[MLX5_FS_VLAN_DEPTH
];
360 struct mlx5_eswitch_rep
*rep
;
361 struct mlx5_core_dev
*mdev
;
363 struct mlx5_termtbl_handle
*termtbl
;
364 } dests
[MLX5_MAX_FLOW_FWD_VPORTS
];
367 u8 tunnel_match_level
;
368 struct mlx5_fc
*counter
;
372 struct mlx5e_tc_flow_parse_attr
*parse_attr
;
375 int mlx5_devlink_eswitch_mode_set(struct devlink
*devlink
, u16 mode
,
376 struct netlink_ext_ack
*extack
);
377 int mlx5_devlink_eswitch_mode_get(struct devlink
*devlink
, u16
*mode
);
378 int mlx5_devlink_eswitch_inline_mode_set(struct devlink
*devlink
, u8 mode
,
379 struct netlink_ext_ack
*extack
);
380 int mlx5_devlink_eswitch_inline_mode_get(struct devlink
*devlink
, u8
*mode
);
381 int mlx5_eswitch_inline_mode_get(struct mlx5_eswitch
*esw
, int nvfs
, u8
*mode
);
382 int mlx5_devlink_eswitch_encap_mode_set(struct devlink
*devlink
, u8 encap
,
383 struct netlink_ext_ack
*extack
);
384 int mlx5_devlink_eswitch_encap_mode_get(struct devlink
*devlink
, u8
*encap
);
385 void *mlx5_eswitch_get_uplink_priv(struct mlx5_eswitch
*esw
, u8 rep_type
);
387 int mlx5_eswitch_add_vlan_action(struct mlx5_eswitch
*esw
,
388 struct mlx5_esw_flow_attr
*attr
);
389 int mlx5_eswitch_del_vlan_action(struct mlx5_eswitch
*esw
,
390 struct mlx5_esw_flow_attr
*attr
);
391 int __mlx5_eswitch_set_vport_vlan(struct mlx5_eswitch
*esw
,
392 u16 vport
, u16 vlan
, u8 qos
, u8 set_flags
);
394 static inline bool mlx5_eswitch_vlan_actions_supported(struct mlx5_core_dev
*dev
,
397 bool ret
= MLX5_CAP_ESW_FLOWTABLE_FDB(dev
, pop_vlan
) &&
398 MLX5_CAP_ESW_FLOWTABLE_FDB(dev
, push_vlan
);
403 return ret
&& MLX5_CAP_ESW_FLOWTABLE_FDB(dev
, pop_vlan_2
) &&
404 MLX5_CAP_ESW_FLOWTABLE_FDB(dev
, push_vlan_2
);
407 bool mlx5_esw_lag_prereq(struct mlx5_core_dev
*dev0
,
408 struct mlx5_core_dev
*dev1
);
409 bool mlx5_esw_multipath_prereq(struct mlx5_core_dev
*dev0
,
410 struct mlx5_core_dev
*dev1
);
412 int mlx5_esw_query_functions(struct mlx5_core_dev
*dev
, u16
*num_vfs
);
414 #define MLX5_DEBUG_ESWITCH_MASK BIT(3)
416 #define esw_info(__dev, format, ...) \
417 dev_info((__dev)->device, "E-Switch: " format, ##__VA_ARGS__)
419 #define esw_warn(__dev, format, ...) \
420 dev_warn((__dev)->device, "E-Switch: " format, ##__VA_ARGS__)
422 #define esw_debug(dev, format, ...) \
423 mlx5_core_dbg_mask(dev, MLX5_DEBUG_ESWITCH_MASK, format, ##__VA_ARGS__)
425 /* The returned number is valid only when the dev is eswitch manager. */
426 static inline u16
mlx5_eswitch_manager_vport(struct mlx5_core_dev
*dev
)
428 return mlx5_core_is_ecpf_esw_manager(dev
) ?
429 MLX5_VPORT_ECPF
: MLX5_VPORT_PF
;
432 static inline bool mlx5_eswitch_is_funcs_handler(struct mlx5_core_dev
*dev
)
434 /* Ideally device should have the functions changed supported
435 * capability regardless of it being ECPF or PF wherever such
436 * event should be processed such as on eswitch manager device.
437 * However, some ECPF based device might not have this capability
438 * set. Hence OR for ECPF check to cover such device.
440 return MLX5_CAP_ESW(dev
, esw_functions_changed
) ||
441 mlx5_core_is_ecpf_esw_manager(dev
);
444 static inline int mlx5_eswitch_uplink_idx(struct mlx5_eswitch
*esw
)
446 /* Uplink always locate at the last element of the array.*/
447 return esw
->total_vports
- 1;
450 static inline int mlx5_eswitch_ecpf_idx(struct mlx5_eswitch
*esw
)
452 return esw
->total_vports
- 2;
455 static inline int mlx5_eswitch_vport_num_to_index(struct mlx5_eswitch
*esw
,
458 if (vport_num
== MLX5_VPORT_ECPF
) {
459 if (!mlx5_ecpf_vport_exists(esw
->dev
))
460 esw_warn(esw
->dev
, "ECPF vport doesn't exist!\n");
461 return mlx5_eswitch_ecpf_idx(esw
);
464 if (vport_num
== MLX5_VPORT_UPLINK
)
465 return mlx5_eswitch_uplink_idx(esw
);
470 static inline u16
mlx5_eswitch_index_to_vport_num(struct mlx5_eswitch
*esw
,
473 if (index
== mlx5_eswitch_ecpf_idx(esw
) &&
474 mlx5_ecpf_vport_exists(esw
->dev
))
475 return MLX5_VPORT_ECPF
;
477 if (index
== mlx5_eswitch_uplink_idx(esw
))
478 return MLX5_VPORT_UPLINK
;
483 /* TODO: This mlx5e_tc function shouldn't be called by eswitch */
484 void mlx5e_tc_clean_fdb_peer_flows(struct mlx5_eswitch
*esw
);
486 /* The vport getter/iterator are only valid after esw->total_vports
487 * and vport->vport are initialized in mlx5_eswitch_init.
489 #define mlx5_esw_for_all_vports(esw, i, vport) \
490 for ((i) = MLX5_VPORT_PF; \
491 (vport) = &(esw)->vports[i], \
492 (i) < (esw)->total_vports; (i)++)
494 #define mlx5_esw_for_each_vf_vport(esw, i, vport, nvfs) \
495 for ((i) = MLX5_VPORT_FIRST_VF; \
496 (vport) = &(esw)->vports[(i)], \
497 (i) <= (nvfs); (i)++)
499 #define mlx5_esw_for_each_vf_vport_reverse(esw, i, vport, nvfs) \
501 (vport) = &(esw)->vports[(i)], \
502 (i) >= MLX5_VPORT_FIRST_VF; (i)--)
504 /* The rep getter/iterator are only valid after esw->total_vports
505 * and vport->vport are initialized in mlx5_eswitch_init.
507 #define mlx5_esw_for_all_reps(esw, i, rep) \
508 for ((i) = MLX5_VPORT_PF; \
509 (rep) = &(esw)->offloads.vport_reps[i], \
510 (i) < (esw)->total_vports; (i)++)
512 #define mlx5_esw_for_each_vf_rep(esw, i, rep, nvfs) \
513 for ((i) = MLX5_VPORT_FIRST_VF; \
514 (rep) = &(esw)->offloads.vport_reps[i], \
515 (i) <= (nvfs); (i)++)
517 #define mlx5_esw_for_each_vf_rep_reverse(esw, i, rep, nvfs) \
519 (rep) = &(esw)->offloads.vport_reps[i], \
520 (i) >= MLX5_VPORT_FIRST_VF; (i)--)
522 #define mlx5_esw_for_each_vf_vport_num(esw, vport, nvfs) \
523 for ((vport) = MLX5_VPORT_FIRST_VF; (vport) <= (nvfs); (vport)++)
525 #define mlx5_esw_for_each_vf_vport_num_reverse(esw, vport, nvfs) \
526 for ((vport) = (nvfs); (vport) >= MLX5_VPORT_FIRST_VF; (vport)--)
528 struct mlx5_vport
*__must_check
529 mlx5_eswitch_get_vport(struct mlx5_eswitch
*esw
, u16 vport_num
);
531 #else /* CONFIG_MLX5_ESWITCH */
532 /* eswitch API stubs */
533 static inline int mlx5_eswitch_init(struct mlx5_core_dev
*dev
) { return 0; }
534 static inline void mlx5_eswitch_cleanup(struct mlx5_eswitch
*esw
) {}
535 static inline int mlx5_eswitch_enable_sriov(struct mlx5_eswitch
*esw
, int nvfs
, int mode
) { return 0; }
536 static inline void mlx5_eswitch_disable_sriov(struct mlx5_eswitch
*esw
) {}
537 static inline bool mlx5_esw_lag_prereq(struct mlx5_core_dev
*dev0
, struct mlx5_core_dev
*dev1
) { return true; }
538 static inline bool mlx5_eswitch_is_funcs_handler(struct mlx5_core_dev
*dev
) { return false; }
540 #define FDB_MAX_CHAIN 1
541 #define FDB_SLOW_PATH_CHAIN (FDB_MAX_CHAIN + 1)
542 #define FDB_MAX_PRIO 1
544 #endif /* CONFIG_MLX5_ESWITCH */
546 #endif /* __MLX5_ESWITCH_H__ */