diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-28 09:53:30 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-28 09:53:30 +0000 |
commit | 2c7cac91ed6e7db0f6937923d2b57f97dbdbc337 (patch) | |
tree | c05dc0f8e6aa3accc84e3e5cffc933ed94941383 /zebra/zebra_evpn_mh.c | |
parent | Initial commit. (diff) | |
download | frr-upstream.tar.xz frr-upstream.zip |
Adding upstream version 8.4.4.upstream/8.4.4upstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'zebra/zebra_evpn_mh.c')
-rw-r--r-- | zebra/zebra_evpn_mh.c | 4039 |
1 files changed, 4039 insertions, 0 deletions
diff --git a/zebra/zebra_evpn_mh.c b/zebra/zebra_evpn_mh.c new file mode 100644 index 0000000..6f6ba25 --- /dev/null +++ b/zebra/zebra_evpn_mh.c @@ -0,0 +1,4039 @@ +/* + * Zebra EVPN multihoming code + * + * Copyright (C) 2019 Cumulus Networks, Inc. + * Anuradha Karuppiah + * + * This file is part of FRR. + * + * FRR is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * FRR is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#include <zebra.h> + +#include "command.h" +#include "hash.h" +#include "if.h" +#include "jhash.h" +#include "linklist.h" +#include "log.h" +#include "memory.h" +#include "prefix.h" +#include "stream.h" +#include "table.h" +#include "vlan.h" +#include "vxlan.h" + +#include "zebra/zebra_router.h" +#include "zebra/debug.h" +#include "zebra/interface.h" +#include "zebra/rib.h" +#include "zebra/rt.h" +#include "zebra/rt_netlink.h" +#include "zebra/if_netlink.h" +#include "zebra/zebra_errors.h" +#include "zebra/zebra_l2.h" +#include "zebra/zebra_ns.h" +#include "zebra/zebra_vrf.h" +#include "zebra/zebra_vxlan.h" +#include "zebra/zebra_evpn.h" +#include "zebra/zebra_evpn_mac.h" +#include "zebra/zebra_vxlan_private.h" +#include "zebra/zebra_router.h" +#include "zebra/zebra_evpn_mh.h" +#include "zebra/zebra_nhg.h" + +DEFINE_MTYPE_STATIC(ZEBRA, ZACC_BD, "Access Broadcast Domain"); +DEFINE_MTYPE_STATIC(ZEBRA, ZES, "Ethernet Segment"); +DEFINE_MTYPE_STATIC(ZEBRA, ZES_EVI, "ES info per-EVI"); +DEFINE_MTYPE_STATIC(ZEBRA, ZMH_INFO, "MH global info"); +DEFINE_MTYPE_STATIC(ZEBRA, ZES_VTEP, "VTEP attached to the ES"); +DEFINE_MTYPE_STATIC(ZEBRA, L2_NH, "L2 nexthop"); + +static void zebra_evpn_es_get_one_base_evpn(void); +static int zebra_evpn_es_evi_send_to_client(struct zebra_evpn_es *es, + struct zebra_evpn *zevpn, bool add); +static void zebra_evpn_local_es_del(struct zebra_evpn_es **esp); +static int zebra_evpn_local_es_update(struct zebra_if *zif, esi_t *esi); +static bool zebra_evpn_es_br_port_dplane_update(struct zebra_evpn_es *es, + const char *caller); +static void zebra_evpn_mh_uplink_cfg_update(struct zebra_if *zif, bool set); +static void zebra_evpn_mh_update_protodown_es(struct zebra_evpn_es *es, + bool resync_dplane); +static void zebra_evpn_mh_clear_protodown_es(struct zebra_evpn_es *es); +static void zebra_evpn_mh_startup_delay_timer_start(const char *rc); + +esi_t zero_esi_buf, *zero_esi = &zero_esi_buf; + +/*****************************************************************************/ +/* Ethernet Segment to EVI association - + * 1. The ES-EVI entry is maintained as a RB tree per L2-VNI + * (struct zebra_evpn.es_evi_rb_tree). + * 2. Each local ES-EVI entry is sent to BGP which advertises it as an + * EAD-EVI (Type-1 EVPN) route + * 3. Local ES-EVI setup is re-evaluated on the following triggers - + * a. When an ESI is set or cleared on an access port. + * b. When an access port associated with an ESI is deleted. + * c. When VLAN member ship changes on an access port. + * d. When a VXLAN_IF is set or cleared on an access broadcast domain. + * e. When a L2-VNI is added or deleted for a VxLAN_IF. + * 4. Currently zebra doesn't remote ES-EVIs. Those are managed and maintained + * entirely in BGP which consolidates them into a remote ES. The remote ES + * is then sent to zebra which allocates a NHG for it. + */ + +/* compare ES-IDs for the ES-EVI RB tree maintained per-EVPN */ +static int zebra_es_evi_rb_cmp(const struct zebra_evpn_es_evi *es_evi1, + const struct zebra_evpn_es_evi *es_evi2) +{ + return memcmp(&es_evi1->es->esi, &es_evi2->es->esi, ESI_BYTES); +} +RB_GENERATE(zebra_es_evi_rb_head, zebra_evpn_es_evi, + rb_node, zebra_es_evi_rb_cmp); + +/* allocate a new ES-EVI and insert it into the per-L2-VNI and per-ES + * tables. + */ +static struct zebra_evpn_es_evi *zebra_evpn_es_evi_new(struct zebra_evpn_es *es, + struct zebra_evpn *zevpn) +{ + struct zebra_evpn_es_evi *es_evi; + + es_evi = XCALLOC(MTYPE_ZES_EVI, sizeof(struct zebra_evpn_es_evi)); + + es_evi->es = es; + es_evi->zevpn = zevpn; + + /* insert into the EVPN-ESI rb tree */ + RB_INSERT(zebra_es_evi_rb_head, &zevpn->es_evi_rb_tree, es_evi); + + /* add to the ES's VNI list */ + listnode_init(&es_evi->es_listnode, es_evi); + listnode_add(es->es_evi_list, &es_evi->es_listnode); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s evi %d new", + es_evi->es->esi_str, es_evi->zevpn->vni); + + return es_evi; +} + +/* Evaluate if the es_evi is ready to be sent BGP - + * 1. If it is ready an add is sent to BGP + * 2. If it is not ready a del is sent (if the ES had been previously added + * to BGP). + */ +static void zebra_evpn_es_evi_re_eval_send_to_client( + struct zebra_evpn_es_evi *es_evi) +{ + bool old_ready; + bool new_ready; + + old_ready = !!(es_evi->flags & ZEBRA_EVPNES_EVI_READY_FOR_BGP); + + /* ES and L2-VNI have to be individually ready for BGP */ + if ((es_evi->flags & ZEBRA_EVPNES_EVI_LOCAL) && + (es_evi->es->flags & ZEBRA_EVPNES_READY_FOR_BGP) && + zebra_evpn_send_to_client_ok(es_evi->zevpn)) + es_evi->flags |= ZEBRA_EVPNES_EVI_READY_FOR_BGP; + else + es_evi->flags &= ~ZEBRA_EVPNES_EVI_READY_FOR_BGP; + + new_ready = !!(es_evi->flags & ZEBRA_EVPNES_EVI_READY_FOR_BGP); + + if (old_ready == new_ready) + return; + + if (new_ready) + zebra_evpn_es_evi_send_to_client(es_evi->es, es_evi->zevpn, + true /* add */); + else + zebra_evpn_es_evi_send_to_client(es_evi->es, es_evi->zevpn, + false /* add */); +} + +/* remove the ES-EVI from the per-L2-VNI and per-ES tables and free + * up the memory. + */ +static void zebra_evpn_es_evi_free(struct zebra_evpn_es_evi *es_evi) +{ + struct zebra_evpn_es *es = es_evi->es; + struct zebra_evpn *zevpn = es_evi->zevpn; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s evi %d free", + es_evi->es->esi_str, es_evi->zevpn->vni); + + /* remove from the ES's VNI list */ + list_delete_node(es->es_evi_list, &es_evi->es_listnode); + + /* remove from the VNI-ESI rb tree */ + RB_REMOVE(zebra_es_evi_rb_head, &zevpn->es_evi_rb_tree, es_evi); + + /* remove from the VNI-ESI rb tree */ + XFREE(MTYPE_ZES_EVI, es_evi); +} + +/* find the ES-EVI in the per-L2-VNI RB tree */ +struct zebra_evpn_es_evi *zebra_evpn_es_evi_find(struct zebra_evpn_es *es, + struct zebra_evpn *zevpn) +{ + struct zebra_evpn_es_evi es_evi; + + es_evi.es = es; + + return RB_FIND(zebra_es_evi_rb_head, &zevpn->es_evi_rb_tree, &es_evi); +} + +/* Tell BGP about an ES-EVI deletion and then delete it */ +static void zebra_evpn_local_es_evi_do_del(struct zebra_evpn_es_evi *es_evi) +{ + if (!(es_evi->flags & ZEBRA_EVPNES_EVI_LOCAL)) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("local es %s evi %d del", + es_evi->es->esi_str, es_evi->zevpn->vni); + + if (es_evi->flags & ZEBRA_EVPNES_EVI_READY_FOR_BGP) { + /* send a del only if add was sent for it earlier */ + zebra_evpn_es_evi_send_to_client(es_evi->es, + es_evi->zevpn, false /* add */); + } + + /* delete it from the EVPN's local list */ + list_delete_node(es_evi->zevpn->local_es_evi_list, + &es_evi->l2vni_listnode); + + es_evi->flags &= ~ZEBRA_EVPNES_EVI_LOCAL; + zebra_evpn_es_evi_free(es_evi); +} +static void zebra_evpn_local_es_evi_del(struct zebra_evpn_es *es, + struct zebra_evpn *zevpn) +{ + struct zebra_evpn_es_evi *es_evi; + + es_evi = zebra_evpn_es_evi_find(es, zevpn); + if (es_evi) + zebra_evpn_local_es_evi_do_del(es_evi); +} + +/* If there are any existing MAC entries for this es/zevpn we need + * to install it in the dataplane. + * + * Note: primary purpose of this is to handle es del/re-add windows where + * sync MAC entries may be added by bgpd before the es-evi membership is + * created in the dataplane and in zebra + */ +static void zebra_evpn_es_evi_mac_install(struct zebra_evpn_es_evi *es_evi) +{ + struct zebra_mac *mac; + struct listnode *node; + struct zebra_evpn_es *es = es_evi->es; + + if (listcount(es->mac_list) && IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("dp-mac install on es %s evi %d add", es->esi_str, + es_evi->zevpn->vni); + + for (ALL_LIST_ELEMENTS_RO(es->mac_list, node, mac)) { + if (mac->zevpn != es_evi->zevpn) + continue; + + if (!CHECK_FLAG(mac->flags, ZEBRA_MAC_LOCAL)) + continue; + + zebra_evpn_sync_mac_dp_install(mac, false, false, __func__); + } +} + +/* Create an ES-EVI if it doesn't already exist and tell BGP */ +static void zebra_evpn_local_es_evi_add(struct zebra_evpn_es *es, + struct zebra_evpn *zevpn) +{ + struct zebra_evpn_es_evi *es_evi; + + es_evi = zebra_evpn_es_evi_find(es, zevpn); + if (!es_evi) { + es_evi = zebra_evpn_es_evi_new(es, zevpn); + if (!es_evi) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("local es %s evi %d add", + es_evi->es->esi_str, es_evi->zevpn->vni); + es_evi->flags |= ZEBRA_EVPNES_EVI_LOCAL; + /* add to the EVPN's local list */ + listnode_init(&es_evi->l2vni_listnode, es_evi); + listnode_add(zevpn->local_es_evi_list, &es_evi->l2vni_listnode); + + zebra_evpn_es_evi_re_eval_send_to_client(es_evi); + + zebra_evpn_es_evi_mac_install(es_evi); + } +} + +static void zebra_evpn_es_evi_show_entry(struct vty *vty, + struct zebra_evpn_es_evi *es_evi, + json_object *json_array) +{ + char type_str[4]; + + if (json_array) { + json_object *json; + json_object *json_types; + + /* Separate JSON object for each es-evi entry */ + json = json_object_new_object(); + + json_object_string_add(json, "esi", es_evi->es->esi_str); + json_object_int_add(json, "vni", es_evi->zevpn->vni); + if (es_evi->flags & ZEBRA_EVPNES_EVI_LOCAL) { + json_types = json_object_new_array(); + if (es_evi->flags & ZEBRA_EVPNES_EVI_LOCAL) + json_array_string_add(json_types, "local"); + json_object_object_add(json, "type", json_types); + } + + /* Add es-evi entry to json array */ + json_object_array_add(json_array, json); + } else { + type_str[0] = '\0'; + if (es_evi->flags & ZEBRA_EVPNES_EVI_LOCAL) + strlcat(type_str, "L", sizeof(type_str)); + + vty_out(vty, "%-8d %-30s %-4s\n", + es_evi->zevpn->vni, es_evi->es->esi_str, + type_str); + } +} + +static void +zebra_evpn_es_evi_show_entry_detail(struct vty *vty, + struct zebra_evpn_es_evi *es_evi, + json_object *json_array) +{ + char type_str[4]; + + if (json_array) { + json_object *json; + json_object *json_flags; + + /* Separate JSON object for each es-evi entry */ + json = json_object_new_object(); + + json_object_string_add(json, "esi", es_evi->es->esi_str); + json_object_int_add(json, "vni", es_evi->zevpn->vni); + if (es_evi->flags + & (ZEBRA_EVPNES_EVI_LOCAL + | ZEBRA_EVPNES_EVI_READY_FOR_BGP)) { + json_flags = json_object_new_array(); + if (es_evi->flags & ZEBRA_EVPNES_EVI_LOCAL) + json_array_string_add(json_flags, "local"); + if (es_evi->flags & ZEBRA_EVPNES_EVI_READY_FOR_BGP) + json_array_string_add(json_flags, + "readyForBgp"); + json_object_object_add(json, "flags", json_flags); + } + + /* Add es-evi entry to json array */ + json_object_array_add(json_array, json); + } else { + type_str[0] = '\0'; + if (es_evi->flags & ZEBRA_EVPNES_EVI_LOCAL) + strlcat(type_str, "L", sizeof(type_str)); + + vty_out(vty, "VNI %d ESI: %s\n", + es_evi->zevpn->vni, es_evi->es->esi_str); + vty_out(vty, " Type: %s\n", type_str); + vty_out(vty, " Ready for BGP: %s\n", + (es_evi->flags & + ZEBRA_EVPNES_EVI_READY_FOR_BGP) ? + "yes" : "no"); + vty_out(vty, "\n"); + } +} + +static void zebra_evpn_es_evi_show_one_evpn(struct zebra_evpn *zevpn, + struct vty *vty, + json_object *json_array, int detail) +{ + struct zebra_evpn_es_evi *es_evi; + + RB_FOREACH(es_evi, zebra_es_evi_rb_head, &zevpn->es_evi_rb_tree) { + if (detail) + zebra_evpn_es_evi_show_entry_detail(vty, es_evi, + json_array); + else + zebra_evpn_es_evi_show_entry(vty, es_evi, json_array); + } +} + +struct evpn_mh_show_ctx { + struct vty *vty; + json_object *json; + int detail; +}; + +static void zebra_evpn_es_evi_show_one_evpn_hash_cb(struct hash_bucket *bucket, + void *ctxt) +{ + struct zebra_evpn *zevpn = (struct zebra_evpn *)bucket->data; + struct evpn_mh_show_ctx *wctx = (struct evpn_mh_show_ctx *)ctxt; + + zebra_evpn_es_evi_show_one_evpn(zevpn, wctx->vty, + wctx->json, wctx->detail); +} + +void zebra_evpn_es_evi_show(struct vty *vty, bool uj, int detail) +{ + json_object *json_array = NULL; + struct zebra_vrf *zvrf; + struct evpn_mh_show_ctx wctx; + + zvrf = zebra_vrf_get_evpn(); + if (uj) + json_array = json_object_new_array(); + + memset(&wctx, 0, sizeof(wctx)); + wctx.vty = vty; + wctx.json = json_array; + wctx.detail = detail; + + if (!detail && !json_array) { + vty_out(vty, "Type: L local, R remote\n"); + vty_out(vty, "%-8s %-30s %-4s\n", "VNI", "ESI", "Type"); + } + /* Display all L2-VNIs */ + hash_iterate(zvrf->evpn_table, zebra_evpn_es_evi_show_one_evpn_hash_cb, + &wctx); + + if (uj) + vty_json(vty, json_array); +} + +void zebra_evpn_es_evi_show_vni(struct vty *vty, bool uj, vni_t vni, int detail) +{ + json_object *json_array = NULL; + struct zebra_evpn *zevpn; + + zevpn = zebra_evpn_lookup(vni); + if (uj) + json_array = json_object_new_array(); + + if (zevpn) { + if (!detail && !json_array) { + vty_out(vty, "Type: L local, R remote\n"); + vty_out(vty, "%-8s %-30s %-4s\n", "VNI", "ESI", "Type"); + } + zebra_evpn_es_evi_show_one_evpn(zevpn, vty, json_array, detail); + } else { + if (!uj) + vty_out(vty, "VNI %d doesn't exist\n", vni); + } + + if (uj) + vty_json(vty, json_array); +} + +/* Initialize the ES tables maintained per-L2_VNI */ +void zebra_evpn_es_evi_init(struct zebra_evpn *zevpn) +{ + /* Initialize the ES-EVI RB tree */ + RB_INIT(zebra_es_evi_rb_head, &zevpn->es_evi_rb_tree); + + /* Initialize the local and remote ES lists maintained for quick + * walks by type + */ + zevpn->local_es_evi_list = list_new(); + listset_app_node_mem(zevpn->local_es_evi_list); +} + +/* Cleanup the ES info maintained per- EVPN */ +void zebra_evpn_es_evi_cleanup(struct zebra_evpn *zevpn) +{ + struct zebra_evpn_es_evi *es_evi; + struct zebra_evpn_es_evi *es_evi_next; + + RB_FOREACH_SAFE(es_evi, zebra_es_evi_rb_head, + &zevpn->es_evi_rb_tree, es_evi_next) { + zebra_evpn_local_es_evi_do_del(es_evi); + } + + list_delete(&zevpn->local_es_evi_list); + zebra_evpn_es_clear_base_evpn(zevpn); +} + +/* called when the oper state or bridge membership changes for the + * vxlan device + */ +void zebra_evpn_update_all_es(struct zebra_evpn *zevpn) +{ + struct zebra_evpn_es_evi *es_evi; + struct listnode *node; + struct interface *vlan_if; + struct interface *vxlan_if; + struct zebra_if *vxlan_zif; + + /* the EVPN is now elgible as a base for EVPN-MH */ + if (zebra_evpn_send_to_client_ok(zevpn)) + zebra_evpn_es_set_base_evpn(zevpn); + else + zebra_evpn_es_clear_base_evpn(zevpn); + + for (ALL_LIST_ELEMENTS_RO(zevpn->local_es_evi_list, node, es_evi)) + zebra_evpn_es_evi_re_eval_send_to_client(es_evi); + + /* reinstall SVI MAC */ + vxlan_if = zevpn->vxlan_if; + if (vxlan_if) { + vxlan_zif = vxlan_if->info; + if (if_is_operative(vxlan_if) + && vxlan_zif->brslave_info.br_if) { + vlan_if = zvni_map_to_svi( + vxlan_zif->l2info.vxl.access_vlan, + vxlan_zif->brslave_info.br_if); + if (vlan_if) + zebra_evpn_acc_bd_svi_mac_add(vlan_if); + } + } +} + +/*****************************************************************************/ +/* Access broadcast domains (BD) + * 1. These broadcast domains can be VLAN aware (in which case + * the key is VID) or VLAN unaware (in which case the key is + * 2. A VID-BD is created when a VLAN is associated with an access port or + * when the VLAN is associated with VXLAN_IF + * 3. A BD is translated into ES-EVI entries when a VNI is associated + * with the broadcast domain + */ +/* Hash key for VLAN based broadcast domains */ +static unsigned int zebra_evpn_acc_vl_hash_keymake(const void *p) +{ + const struct zebra_evpn_access_bd *acc_bd = p; + + return jhash_1word(acc_bd->vid, 0); +} + +/* Compare two VLAN based broadcast domains */ +static bool zebra_evpn_acc_vl_cmp(const void *p1, const void *p2) +{ + const struct zebra_evpn_access_bd *acc_bd1 = p1; + const struct zebra_evpn_access_bd *acc_bd2 = p2; + + if (acc_bd1 == NULL && acc_bd2 == NULL) + return true; + + if (acc_bd1 == NULL || acc_bd2 == NULL) + return false; + + return (acc_bd1->vid == acc_bd2->vid); +} + +/* Lookup VLAN based broadcast domain */ +static struct zebra_evpn_access_bd *zebra_evpn_acc_vl_find(vlanid_t vid) +{ + struct zebra_evpn_access_bd *acc_bd; + struct zebra_evpn_access_bd tmp; + + tmp.vid = vid; + acc_bd = hash_lookup(zmh_info->evpn_vlan_table, &tmp); + + return acc_bd; +} + +/* A new broadcast domain can be created when a VLAN member or VLAN<=>VxLAN_IF + * mapping is added. + */ +static struct zebra_evpn_access_bd * +zebra_evpn_acc_vl_new(vlanid_t vid, struct interface *br_if) +{ + struct zebra_evpn_access_bd *acc_bd; + struct interface *vlan_if; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("access vlan %d add", vid); + + acc_bd = XCALLOC(MTYPE_ZACC_BD, sizeof(struct zebra_evpn_access_bd)); + + acc_bd->vid = vid; + + /* Initialize the mbr list */ + acc_bd->mbr_zifs = list_new(); + + /* Add to hash */ + (void)hash_get(zmh_info->evpn_vlan_table, acc_bd, hash_alloc_intern); + + /* check if an svi exists for the vlan */ + if (br_if) { + vlan_if = zvni_map_to_svi(vid, br_if); + if (vlan_if) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("vlan %d SVI %s set", vid, + vlan_if->name); + acc_bd->vlan_zif = vlan_if->info; + } + } + return acc_bd; +} + +/* Free VLAN based broadcast domain - + * This just frees appropriate memory, caller should have taken other + * needed actions. + */ +static void zebra_evpn_acc_vl_free(struct zebra_evpn_access_bd *acc_bd) +{ + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("access vlan %d del", acc_bd->vid); + + if (acc_bd->vlan_zif && acc_bd->zevpn && acc_bd->zevpn->mac_table) + zebra_evpn_mac_svi_del(acc_bd->vlan_zif->ifp, acc_bd->zevpn); + + /* cleanup resources maintained against the ES */ + list_delete(&acc_bd->mbr_zifs); + + /* remove EVI from various tables */ + hash_release(zmh_info->evpn_vlan_table, acc_bd); + + XFREE(MTYPE_ZACC_BD, acc_bd); +} + +static void zebra_evpn_acc_vl_cleanup_all(struct hash_bucket *bucket, void *arg) +{ + struct zebra_evpn_access_bd *acc_bd = bucket->data; + + zebra_evpn_acc_vl_free(acc_bd); +} + +/* called when a bd mbr is removed or VxLAN_IF is diassociated from the access + * VLAN + */ +static void zebra_evpn_acc_bd_free_on_deref(struct zebra_evpn_access_bd *acc_bd) +{ + if (!list_isempty(acc_bd->mbr_zifs) || acc_bd->vxlan_zif) + return; + + /* if there are no references free the EVI */ + zebra_evpn_acc_vl_free(acc_bd); +} + +/* called when a SVI is goes up/down */ +void zebra_evpn_acc_bd_svi_set(struct zebra_if *vlan_zif, + struct zebra_if *br_zif, bool is_up) +{ + struct zebra_evpn_access_bd *acc_bd; + struct zebra_l2info_bridge *br; + uint16_t vid; + struct zebra_if *tmp_br_zif = br_zif; + + if (!tmp_br_zif) { + if (!vlan_zif->link || !vlan_zif->link->info) + return; + + tmp_br_zif = vlan_zif->link->info; + } + + br = &tmp_br_zif->l2info.br; + /* ignore vlan unaware bridges */ + if (!br->vlan_aware) + return; + + vid = vlan_zif->l2info.vl.vid; + acc_bd = zebra_evpn_acc_vl_find(vid); + if (!acc_bd) + return; + + if (is_up) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("vlan %d SVI %s set", vid, + vlan_zif->ifp->name); + + acc_bd->vlan_zif = vlan_zif; + if (acc_bd->zevpn) + zebra_evpn_mac_svi_add(acc_bd->vlan_zif->ifp, + acc_bd->zevpn); + } else if (acc_bd->vlan_zif) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("vlan %d SVI clear", vid); + acc_bd->vlan_zif = NULL; + if (acc_bd->zevpn && acc_bd->zevpn->mac_table) + zebra_evpn_mac_svi_del(vlan_zif->ifp, acc_bd->zevpn); + } +} + +/* On some events macs are force-flushed. This api can be used to reinstate + * the svi-mac after such cleanup-events. + */ +void zebra_evpn_acc_bd_svi_mac_add(struct interface *vlan_if) +{ + zebra_evpn_acc_bd_svi_set(vlan_if->info, NULL, + if_is_operative(vlan_if)); +} + +/* called when a EVPN-L2VNI is set or cleared against a BD */ +static void zebra_evpn_acc_bd_evpn_set(struct zebra_evpn_access_bd *acc_bd, + struct zebra_evpn *zevpn, + struct zebra_evpn *old_zevpn) +{ + struct zebra_if *zif; + struct listnode *node; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("access vlan %d l2-vni %u set", + acc_bd->vid, zevpn ? zevpn->vni : 0); + + for (ALL_LIST_ELEMENTS_RO(acc_bd->mbr_zifs, node, zif)) { + if (!zif->es_info.es) + continue; + + if (zevpn) + zebra_evpn_local_es_evi_add(zif->es_info.es, zevpn); + else if (old_zevpn) + zebra_evpn_local_es_evi_del(zif->es_info.es, old_zevpn); + } + + if (acc_bd->vlan_zif) { + if (zevpn) + zebra_evpn_mac_svi_add(acc_bd->vlan_zif->ifp, + acc_bd->zevpn); + else if (old_zevpn && old_zevpn->mac_table) + zebra_evpn_mac_svi_del(acc_bd->vlan_zif->ifp, + old_zevpn); + } +} + +/* handle VLAN->VxLAN_IF association */ +void zebra_evpn_vl_vxl_ref(uint16_t vid, struct zebra_if *vxlan_zif) +{ + struct zebra_evpn_access_bd *acc_bd; + struct zebra_if *old_vxlan_zif; + struct zebra_evpn *old_zevpn; + + if (!vid) + return; + + acc_bd = zebra_evpn_acc_vl_find(vid); + if (!acc_bd) + acc_bd = zebra_evpn_acc_vl_new(vid, + vxlan_zif->brslave_info.br_if); + + old_vxlan_zif = acc_bd->vxlan_zif; + acc_bd->vxlan_zif = vxlan_zif; + if (vxlan_zif == old_vxlan_zif) + return; + + old_zevpn = acc_bd->zevpn; + acc_bd->zevpn = zebra_evpn_lookup(vxlan_zif->l2info.vxl.vni); + if (acc_bd->zevpn == old_zevpn) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("access vlan %d vni %u ref", + acc_bd->vid, vxlan_zif->l2info.vxl.vni); + + if (old_zevpn) + zebra_evpn_acc_bd_evpn_set(acc_bd, NULL, old_zevpn); + + if (acc_bd->zevpn) + zebra_evpn_acc_bd_evpn_set(acc_bd, acc_bd->zevpn, NULL); +} + +/* handle VLAN->VxLAN_IF deref */ +void zebra_evpn_vl_vxl_deref(uint16_t vid, struct zebra_if *vxlan_zif) +{ + struct zebra_evpn_access_bd *acc_bd; + + if (!vid) + return; + + acc_bd = zebra_evpn_acc_vl_find(vid); + if (!acc_bd) + return; + + /* clear vxlan_if only if it matches */ + if (acc_bd->vxlan_zif != vxlan_zif) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("access vlan %d vni %u deref", + acc_bd->vid, vxlan_zif->l2info.vxl.vni); + + if (acc_bd->zevpn) + zebra_evpn_acc_bd_evpn_set(acc_bd, NULL, acc_bd->zevpn); + + acc_bd->zevpn = NULL; + acc_bd->vxlan_zif = NULL; + + /* if there are no other references the access_bd can be freed */ + zebra_evpn_acc_bd_free_on_deref(acc_bd); +} + +/* handle EVPN add/del */ +void zebra_evpn_vxl_evpn_set(struct zebra_if *zif, struct zebra_evpn *zevpn, + bool set) +{ + struct zebra_l2info_vxlan *vxl; + struct zebra_evpn_access_bd *acc_bd; + + if (!zif) + return; + + /* locate access_bd associated with the vxlan device */ + vxl = &zif->l2info.vxl; + acc_bd = zebra_evpn_acc_vl_find(vxl->access_vlan); + if (!acc_bd) + return; + + if (set) { + zebra_evpn_es_set_base_evpn(zevpn); + if (acc_bd->zevpn != zevpn) { + acc_bd->zevpn = zevpn; + zebra_evpn_acc_bd_evpn_set(acc_bd, zevpn, NULL); + } + } else { + if (acc_bd->zevpn) { + struct zebra_evpn *old_zevpn = acc_bd->zevpn; + acc_bd->zevpn = NULL; + zebra_evpn_acc_bd_evpn_set(acc_bd, NULL, old_zevpn); + } + } +} + +/* handle addition of new VLAN members */ +void zebra_evpn_vl_mbr_ref(uint16_t vid, struct zebra_if *zif) +{ + struct zebra_evpn_access_bd *acc_bd; + + if (!vid) + return; + + acc_bd = zebra_evpn_acc_vl_find(vid); + if (!acc_bd) + acc_bd = zebra_evpn_acc_vl_new(vid, zif->brslave_info.br_if); + + if (listnode_lookup(acc_bd->mbr_zifs, zif)) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("access vlan %d mbr %s ref", + vid, zif->ifp->name); + + listnode_add(acc_bd->mbr_zifs, zif); + if (acc_bd->zevpn && zif->es_info.es) + zebra_evpn_local_es_evi_add(zif->es_info.es, acc_bd->zevpn); +} + +/* handle deletion of VLAN members */ +void zebra_evpn_vl_mbr_deref(uint16_t vid, struct zebra_if *zif) +{ + struct zebra_evpn_access_bd *acc_bd; + struct listnode *node; + + if (!vid) + return; + + acc_bd = zebra_evpn_acc_vl_find(vid); + if (!acc_bd) + return; + + node = listnode_lookup(acc_bd->mbr_zifs, zif); + if (!node) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("access vlan %d mbr %s deref", + vid, zif->ifp->name); + + list_delete_node(acc_bd->mbr_zifs, node); + + if (acc_bd->zevpn && zif->es_info.es) + zebra_evpn_local_es_evi_del(zif->es_info.es, acc_bd->zevpn); + + /* if there are no other references the access_bd can be freed */ + zebra_evpn_acc_bd_free_on_deref(acc_bd); +} + +static void zebra_evpn_acc_vl_adv_svi_mac_cb(struct hash_bucket *bucket, + void *ctxt) +{ + struct zebra_evpn_access_bd *acc_bd = bucket->data; + + if (acc_bd->vlan_zif && acc_bd->zevpn) + zebra_evpn_mac_svi_add(acc_bd->vlan_zif->ifp, acc_bd->zevpn); +} + +/* called when advertise SVI MAC is enabled on the switch */ +static void zebra_evpn_acc_vl_adv_svi_mac_all(void) +{ + hash_iterate(zmh_info->evpn_vlan_table, + zebra_evpn_acc_vl_adv_svi_mac_cb, NULL); +} + +static void zebra_evpn_acc_vl_json_fill(struct zebra_evpn_access_bd *acc_bd, + json_object *json, bool detail) +{ + json_object_int_add(json, "vlan", acc_bd->vid); + if (acc_bd->vxlan_zif) + json_object_string_add(json, "vxlanIf", + acc_bd->vxlan_zif->ifp->name); + if (acc_bd->zevpn) + json_object_int_add(json, "vni", acc_bd->zevpn->vni); + if (acc_bd->mbr_zifs) + json_object_int_add(json, "memberIfCount", + listcount(acc_bd->mbr_zifs)); + + if (detail) { + json_object *json_mbrs; + json_object *json_mbr; + struct zebra_if *zif; + struct listnode *node; + + + json_mbrs = json_object_new_array(); + for (ALL_LIST_ELEMENTS_RO(acc_bd->mbr_zifs, node, zif)) { + json_mbr = json_object_new_object(); + json_object_string_add(json_mbr, "ifName", + zif->ifp->name); + json_object_array_add(json_mbrs, json_mbr); + } + json_object_object_add(json, "members", json_mbrs); + } +} + +static void zebra_evpn_acc_vl_show_entry_detail(struct vty *vty, + struct zebra_evpn_access_bd *acc_bd, json_object *json) +{ + struct zebra_if *zif; + struct listnode *node; + + if (json) { + zebra_evpn_acc_vl_json_fill(acc_bd, json, true); + } else { + vty_out(vty, "VLAN: %u\n", acc_bd->vid); + vty_out(vty, " VxLAN Interface: %s\n", + acc_bd->vxlan_zif ? + acc_bd->vxlan_zif->ifp->name : "-"); + vty_out(vty, " SVI: %s\n", + acc_bd->vlan_zif ? acc_bd->vlan_zif->ifp->name : "-"); + vty_out(vty, " L2-VNI: %d\n", + acc_bd->zevpn ? acc_bd->zevpn->vni : 0); + vty_out(vty, " Member Count: %d\n", + listcount(acc_bd->mbr_zifs)); + vty_out(vty, " Members: \n"); + for (ALL_LIST_ELEMENTS_RO(acc_bd->mbr_zifs, node, zif)) + vty_out(vty, " %s\n", zif->ifp->name); + vty_out(vty, "\n"); + } +} + +static void zebra_evpn_acc_vl_show_entry(struct vty *vty, + struct zebra_evpn_access_bd *acc_bd, json_object *json) +{ + if (json) { + zebra_evpn_acc_vl_json_fill(acc_bd, json, false); + } else { + vty_out(vty, "%-5u %-15s %-8d %-15s %u\n", acc_bd->vid, + acc_bd->vlan_zif ? acc_bd->vlan_zif->ifp->name : "-", + acc_bd->zevpn ? acc_bd->zevpn->vni : 0, + acc_bd->vxlan_zif ? acc_bd->vxlan_zif->ifp->name : "-", + listcount(acc_bd->mbr_zifs)); + } +} + +static void zebra_evpn_acc_vl_show_hash(struct hash_bucket *bucket, void *ctxt) +{ + struct evpn_mh_show_ctx *wctx = ctxt; + struct zebra_evpn_access_bd *acc_bd = bucket->data; + json_object *json = NULL; + + if (wctx->json) + json = json_object_new_object(); + if (wctx->detail) + zebra_evpn_acc_vl_show_entry_detail(wctx->vty, acc_bd, json); + else + zebra_evpn_acc_vl_show_entry(wctx->vty, acc_bd, json); + if (json) + json_object_array_add(wctx->json, json); +} + +void zebra_evpn_acc_vl_show(struct vty *vty, bool uj) +{ + struct evpn_mh_show_ctx wctx; + json_object *json_array = NULL; + + if (uj) + json_array = json_object_new_array(); + + memset(&wctx, 0, sizeof(wctx)); + wctx.vty = vty; + wctx.json = json_array; + wctx.detail = false; + + if (!uj) + vty_out(vty, "%-5s %-15s %-8s %-15s %s\n", "VLAN", "SVI", + "L2-VNI", "VXLAN-IF", "# Members"); + + hash_iterate(zmh_info->evpn_vlan_table, zebra_evpn_acc_vl_show_hash, + &wctx); + + if (uj) + vty_json(vty, json_array); +} + +void zebra_evpn_acc_vl_show_detail(struct vty *vty, bool uj) +{ + struct evpn_mh_show_ctx wctx; + json_object *json_array = NULL; + + if (uj) + json_array = json_object_new_array(); + memset(&wctx, 0, sizeof(wctx)); + wctx.vty = vty; + wctx.json = json_array; + wctx.detail = true; + + hash_iterate(zmh_info->evpn_vlan_table, zebra_evpn_acc_vl_show_hash, + &wctx); + + if (uj) + vty_json(vty, json_array); +} + +void zebra_evpn_acc_vl_show_vid(struct vty *vty, bool uj, vlanid_t vid) +{ + json_object *json = NULL; + struct zebra_evpn_access_bd *acc_bd; + + if (uj) + json = json_object_new_object(); + + acc_bd = zebra_evpn_acc_vl_find(vid); + if (acc_bd) { + zebra_evpn_acc_vl_show_entry_detail(vty, acc_bd, json); + } else { + if (!json) + vty_out(vty, "VLAN %u not present\n", vid); + } + + if (uj) + vty_json(vty, json); +} + +/* Initialize VLAN member bitmap on an interface. Although VLAN membership + * is independent of EVPN we only process it if its of interest to EVPN-MH + * i.e. on access ports that can be setup as Ethernet Segments. And that is + * intended as an optimization. + */ +void zebra_evpn_if_init(struct zebra_if *zif) +{ + if (!zebra_evpn_is_if_es_capable(zif)) + return; + + if (!bf_is_inited(zif->vlan_bitmap)) + bf_init(zif->vlan_bitmap, IF_VLAN_BITMAP_MAX); + + /* if an es_id and sysmac are already present against the interface + * activate it + */ + zebra_evpn_local_es_update(zif, &zif->es_info.esi); +} + +/* handle deletion of an access port by removing it from all associated + * broadcast domains. + */ +void zebra_evpn_if_cleanup(struct zebra_if *zif) +{ + vlanid_t vid; + struct zebra_evpn_es *es; + + if (bf_is_inited(zif->vlan_bitmap)) { + bf_for_each_set_bit(zif->vlan_bitmap, vid, IF_VLAN_BITMAP_MAX) + { + zebra_evpn_vl_mbr_deref(vid, zif); + } + + bf_free(zif->vlan_bitmap); + } + + /* Delete associated Ethernet Segment */ + es = zif->es_info.es; + if (es) + zebra_evpn_local_es_del(&es); +} + +/***************************************************************************** + * L2 NH/NHG Management + * A L2 NH entry is programmed in the kernel for every ES-VTEP entry. This + * NH is then added to the L2-ECMP-NHG associated with the ES. + */ +static uint32_t zebra_evpn_nhid_alloc(struct zebra_evpn_es *es) +{ + uint32_t id; + uint32_t nh_id; + + bf_assign_index(zmh_info->nh_id_bitmap, id); + + if (!id) + return 0; + + if (es) { + nh_id = id | EVPN_NHG_ID_TYPE_BIT; + /* Add to NHG hash */ + es->nhg_id = nh_id; + (void)hash_get(zmh_info->nhg_table, es, hash_alloc_intern); + } else { + nh_id = id | EVPN_NH_ID_TYPE_BIT; + } + + return nh_id; +} + +static void zebra_evpn_nhid_free(uint32_t nh_id, struct zebra_evpn_es *es) +{ + uint32_t id = (nh_id & EVPN_NH_ID_VAL_MASK); + + if (!id) + return; + + if (es) { + hash_release(zmh_info->nhg_table, es); + es->nhg_id = 0; + } + + bf_release_index(zmh_info->nh_id_bitmap, id); +} + +static unsigned int zebra_evpn_nh_ip_hash_keymake(const void *p) +{ + const struct zebra_evpn_l2_nh *nh = p; + + return jhash_1word(nh->vtep_ip.s_addr, 0); +} + +static bool zebra_evpn_nh_ip_cmp(const void *p1, const void *p2) +{ + const struct zebra_evpn_l2_nh *nh1 = p1; + const struct zebra_evpn_l2_nh *nh2 = p2; + + if (nh1 == NULL && nh2 == NULL) + return true; + + if (nh1 == NULL || nh2 == NULL) + return false; + + return (nh1->vtep_ip.s_addr == nh2->vtep_ip.s_addr); +} + +static unsigned int zebra_evpn_nhg_hash_keymake(const void *p) +{ + const struct zebra_evpn_es *es = p; + + return jhash_1word(es->nhg_id, 0); +} + +static bool zebra_evpn_nhg_cmp(const void *p1, const void *p2) +{ + const struct zebra_evpn_es *es1 = p1; + const struct zebra_evpn_es *es2 = p2; + + if (es1 == NULL && es2 == NULL) + return true; + + if (es1 == NULL || es2 == NULL) + return false; + + return (es1->nhg_id == es2->nhg_id); +} + +/* Lookup ES using the NHG id associated with it */ +static struct zebra_evpn_es *zebra_evpn_nhg_find(uint32_t nhg_id) +{ + struct zebra_evpn_es *es; + struct zebra_evpn_es tmp; + + tmp.nhg_id = nhg_id; + es = hash_lookup(zmh_info->nhg_table, &tmp); + + return es; +} + +/* Returns TRUE if the NHG is associated with a local ES */ +bool zebra_evpn_nhg_is_local_es(uint32_t nhg_id, + struct zebra_evpn_es **local_es) +{ + struct zebra_evpn_es *es; + + es = zebra_evpn_nhg_find(nhg_id); + if (es && (es->flags & ZEBRA_EVPNES_LOCAL)) { + *local_es = es; + return true; + } + + *local_es = NULL; + return false; +} + +/* update remote macs associated with the ES */ +static void zebra_evpn_nhg_mac_update(struct zebra_evpn_es *es) +{ + struct zebra_mac *mac; + struct listnode *node; + bool local_via_nw; + + local_via_nw = zebra_evpn_es_local_mac_via_network_port(es); + if (IS_ZEBRA_DEBUG_EVPN_MH_ES || IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug("mac update on es %s nhg %s", es->esi_str, + (es->flags & ZEBRA_EVPNES_NHG_ACTIVE) + ? "activate" + : "de-activate"); + + for (ALL_LIST_ELEMENTS_RO(es->mac_list, node, mac)) { + if (CHECK_FLAG(mac->flags, ZEBRA_MAC_REMOTE) + || (local_via_nw && CHECK_FLAG(mac->flags, ZEBRA_MAC_LOCAL) + && zebra_evpn_mac_is_static(mac))) { + if (es->flags & ZEBRA_EVPNES_NHG_ACTIVE) { + if (IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug( + "%smac %pEA install via es %s nhg 0x%x", + (mac->flags & ZEBRA_MAC_REMOTE) + ? "rem" + : "local-nw", + &mac->macaddr, es->esi_str, + es->nhg_id); + zebra_evpn_rem_mac_install( + mac->zevpn, mac, false /*was_static*/); + } else { + if (IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug( + "%smac %pEA un-install es %s", + (mac->flags & ZEBRA_MAC_REMOTE) + ? "rem" + : "local-nw", + &mac->macaddr, es->esi_str); + zebra_evpn_rem_mac_uninstall(mac->zevpn, mac, + true /*force*/); + } + } + } +} + +/* The MAC ECMP group is activated on the first VTEP */ +static void zebra_evpn_nhg_update(struct zebra_evpn_es *es) +{ + uint32_t nh_cnt = 0; + struct nh_grp nh_ids[ES_VTEP_MAX_CNT]; + struct zebra_evpn_es_vtep *es_vtep; + struct listnode *node; + + if (!es->nhg_id) + return; + + for (ALL_LIST_ELEMENTS_RO(es->es_vtep_list, node, es_vtep)) { + if (!es_vtep->nh) + continue; + + if (nh_cnt >= ES_VTEP_MAX_CNT) + break; + + memset(&nh_ids[nh_cnt], 0, sizeof(struct nh_grp)); + nh_ids[nh_cnt].id = es_vtep->nh->nh_id; + ++nh_cnt; + } + + if (nh_cnt) { + if (IS_ZEBRA_DEBUG_EVPN_MH_NH) { + char nh_str[ES_VTEP_LIST_STR_SZ]; + uint32_t i; + char nh_buf[16]; + + nh_str[0] = '\0'; + for (i = 0; i < nh_cnt; ++i) { + snprintf(nh_buf, sizeof(nh_buf), "%u ", + nh_ids[i].id); + strlcat(nh_str, nh_buf, sizeof(nh_str)); + } + zlog_debug("es %s nhg %u add %s", es->esi_str, + es->nhg_id, nh_str); + } + + kernel_upd_mac_nhg(es->nhg_id, nh_cnt, nh_ids); + if (!(es->flags & ZEBRA_EVPNES_NHG_ACTIVE)) { + es->flags |= ZEBRA_EVPNES_NHG_ACTIVE; + /* add backup NHG to the br-port */ + if ((es->flags & ZEBRA_EVPNES_LOCAL)) + zebra_evpn_es_br_port_dplane_update(es, + __func__); + zebra_evpn_nhg_mac_update(es); + } + } else { + if (es->flags & ZEBRA_EVPNES_NHG_ACTIVE) { + if (IS_ZEBRA_DEBUG_EVPN_MH_NH) + zlog_debug("es %s nhg %u del", es->esi_str, + es->nhg_id); + es->flags &= ~ZEBRA_EVPNES_NHG_ACTIVE; + /* remove backup NHG from the br-port */ + if ((es->flags & ZEBRA_EVPNES_LOCAL)) + zebra_evpn_es_br_port_dplane_update(es, + __func__); + zebra_evpn_nhg_mac_update(es); + kernel_del_mac_nhg(es->nhg_id); + } + } + +} + +static void zebra_evpn_es_l2_nh_show_entry(struct zebra_evpn_l2_nh *nh, + struct vty *vty, + json_object *json_array) +{ + if (json_array) { + json_object *json = NULL; + + json = json_object_new_object(); + json_object_string_addf(json, "vtep", "%pI4", &nh->vtep_ip); + json_object_int_add(json, "nhId", nh->nh_id); + json_object_int_add(json, "refCnt", nh->ref_cnt); + + json_object_array_add(json_array, json); + } else { + vty_out(vty, "%-16pI4 %-10u %u\n", &nh->vtep_ip, nh->nh_id, + nh->ref_cnt); + } +} + +static void zebra_evpn_l2_nh_show_cb(struct hash_bucket *bucket, void *ctxt) +{ + struct zebra_evpn_l2_nh *nh = (struct zebra_evpn_l2_nh *)bucket->data; + struct evpn_mh_show_ctx *wctx = (struct evpn_mh_show_ctx *)ctxt; + + zebra_evpn_es_l2_nh_show_entry(nh, wctx->vty, wctx->json); +} + +void zebra_evpn_l2_nh_show(struct vty *vty, bool uj) +{ + struct evpn_mh_show_ctx wctx; + json_object *json_array = NULL; + + if (uj) { + json_array = json_object_new_array(); + } else { + vty_out(vty, "%-16s %-10s %s\n", "VTEP", "NH id", "#ES"); + } + + memset(&wctx, 0, sizeof(wctx)); + wctx.vty = vty; + wctx.json = json_array; + + hash_iterate(zmh_info->nh_ip_table, zebra_evpn_l2_nh_show_cb, &wctx); + + if (uj) + vty_json(vty, json_array); +} + +static struct zebra_evpn_l2_nh *zebra_evpn_l2_nh_find(struct in_addr vtep_ip) +{ + struct zebra_evpn_l2_nh *nh; + struct zebra_evpn_l2_nh tmp; + + tmp.vtep_ip.s_addr = vtep_ip.s_addr; + nh = hash_lookup(zmh_info->nh_ip_table, &tmp); + + return nh; +} + +static struct zebra_evpn_l2_nh *zebra_evpn_l2_nh_alloc(struct in_addr vtep_ip) +{ + struct zebra_evpn_l2_nh *nh; + + nh = XCALLOC(MTYPE_L2_NH, sizeof(*nh)); + nh->vtep_ip = vtep_ip; + (void)hash_get(zmh_info->nh_ip_table, nh, hash_alloc_intern); + + nh->nh_id = zebra_evpn_nhid_alloc(NULL); + if (!nh->nh_id) { + hash_release(zmh_info->nh_ip_table, nh); + XFREE(MTYPE_L2_NH, nh); + return NULL; + } + + /* install the NH in the dataplane */ + kernel_upd_mac_nh(nh->nh_id, nh->vtep_ip); + + return nh; +} + +static void zebra_evpn_l2_nh_free(struct zebra_evpn_l2_nh *nh) +{ + /* delete the NH from the dataplane */ + kernel_del_mac_nh(nh->nh_id); + + zebra_evpn_nhid_free(nh->nh_id, NULL); + hash_release(zmh_info->nh_ip_table, nh); + XFREE(MTYPE_L2_NH, nh); +} + +static void zebra_evpn_l2_nh_es_vtep_ref(struct zebra_evpn_es_vtep *es_vtep) +{ + if (es_vtep->nh) + return; + + es_vtep->nh = zebra_evpn_l2_nh_find(es_vtep->vtep_ip); + if (!es_vtep->nh) + es_vtep->nh = zebra_evpn_l2_nh_alloc(es_vtep->vtep_ip); + + if (!es_vtep->nh) { + zlog_warn("es %s vtep %pI4 nh ref failed", es_vtep->es->esi_str, + &es_vtep->vtep_ip); + return; + } + + ++es_vtep->nh->ref_cnt; + + if (IS_ZEBRA_DEBUG_EVPN_MH_NH) + zlog_debug("es %s vtep %pI4 nh %u ref %u", es_vtep->es->esi_str, + &es_vtep->vtep_ip, es_vtep->nh->nh_id, + es_vtep->nh->ref_cnt); + + /* add the NH to the parent NHG */ + zebra_evpn_nhg_update(es_vtep->es); +} + +static void zebra_evpn_l2_nh_es_vtep_deref(struct zebra_evpn_es_vtep *es_vtep) +{ + struct zebra_evpn_l2_nh *nh = es_vtep->nh; + + if (!nh) + return; + + es_vtep->nh = NULL; + if (nh->ref_cnt) + --nh->ref_cnt; + + if (IS_ZEBRA_DEBUG_EVPN_MH_NH) + zlog_debug("es %s vtep %pI4 nh %u deref %u", + es_vtep->es->esi_str, &es_vtep->vtep_ip, nh->nh_id, + nh->ref_cnt); + + /* remove the NH from the parent NHG */ + zebra_evpn_nhg_update(es_vtep->es); + + /* uninstall the NH */ + if (!nh->ref_cnt) + zebra_evpn_l2_nh_free(nh); +} + +/*****************************************************************************/ +/* Ethernet Segment Management + * 1. Ethernet Segment is a collection of links attached to the same + * server (MHD) or switch (MHN) + * 2. An Ethernet Segment can span multiple PEs and is identified by the + * 10-byte ES-ID. + * 3. Zebra manages the local ESI configuration. + * 4. It also maintains the aliasing that maps an ESI (local or remote) + * to one or more PEs/VTEPs. + * 5. remote ESs are added by BGP (on rxing EAD Type-1 routes) + */ +/* A list of remote VTEPs is maintained for each ES. This list includes - + * 1. VTEPs for which we have imported the ESR i.e. ES-peers + * 2. VTEPs that have an "active" ES-EVI VTEP i.e. EAD-per-ES and EAD-per-EVI + * have been imported into one or more EVPNs + */ +static int zebra_evpn_es_vtep_cmp(void *p1, void *p2) +{ + const struct zebra_evpn_es_vtep *es_vtep1 = p1; + const struct zebra_evpn_es_vtep *es_vtep2 = p2; + + return es_vtep1->vtep_ip.s_addr - es_vtep2->vtep_ip.s_addr; +} + +static struct zebra_evpn_es_vtep *zebra_evpn_es_vtep_new( + struct zebra_evpn_es *es, struct in_addr vtep_ip) +{ + struct zebra_evpn_es_vtep *es_vtep; + + es_vtep = XCALLOC(MTYPE_ZES_VTEP, sizeof(*es_vtep)); + + es_vtep->es = es; + es_vtep->vtep_ip.s_addr = vtep_ip.s_addr; + listnode_init(&es_vtep->es_listnode, es_vtep); + listnode_add_sort(es->es_vtep_list, &es_vtep->es_listnode); + + return es_vtep; +} + +static void zebra_evpn_es_vtep_free(struct zebra_evpn_es_vtep *es_vtep) +{ + struct zebra_evpn_es *es = es_vtep->es; + + list_delete_node(es->es_vtep_list, &es_vtep->es_listnode); + /* update the L2-NHG associated with the ES */ + zebra_evpn_l2_nh_es_vtep_deref(es_vtep); + XFREE(MTYPE_ZES_VTEP, es_vtep); +} + + +/* check if VTEP is already part of the list */ +static struct zebra_evpn_es_vtep *zebra_evpn_es_vtep_find( + struct zebra_evpn_es *es, struct in_addr vtep_ip) +{ + struct listnode *node = NULL; + struct zebra_evpn_es_vtep *es_vtep; + + for (ALL_LIST_ELEMENTS_RO(es->es_vtep_list, node, es_vtep)) { + if (es_vtep->vtep_ip.s_addr == vtep_ip.s_addr) + return es_vtep; + } + return NULL; +} + +/* flush all the dataplane br-port info associated with the ES */ +static bool zebra_evpn_es_br_port_dplane_clear(struct zebra_evpn_es *es) +{ + struct in_addr sph_filters[ES_VTEP_MAX_CNT]; + + if (!(es->flags & ZEBRA_EVPNES_BR_PORT)) + return false; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s br-port dplane clear", es->esi_str); + + memset(&sph_filters, 0, sizeof(sph_filters)); + dplane_br_port_update(es->zif->ifp, false /* non_df */, 0, sph_filters, + 0 /* backup_nhg_id */); + return true; +} + +static inline bool +zebra_evpn_es_br_port_dplane_update_needed(struct zebra_evpn_es *es) +{ + return (es->flags & ZEBRA_EVPNES_NON_DF) + || (es->flags & ZEBRA_EVPNES_NHG_ACTIVE) + || listcount(es->es_vtep_list); +} + +/* returns TRUE if dplane entry was updated */ +static bool zebra_evpn_es_br_port_dplane_update(struct zebra_evpn_es *es, + const char *caller) +{ + uint32_t backup_nhg_id; + struct in_addr sph_filters[ES_VTEP_MAX_CNT]; + struct listnode *node = NULL; + struct zebra_evpn_es_vtep *es_vtep; + uint32_t sph_filter_cnt = 0; + + if (!(es->flags & ZEBRA_EVPNES_LOCAL)) + return zebra_evpn_es_br_port_dplane_clear(es); + + /* If the ES is not a bridge port there is nothing + * in the dataplane + */ + if (!(es->flags & ZEBRA_EVPNES_BR_PORT)) + return false; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s br-port dplane update by %s", es->esi_str, + caller); + backup_nhg_id = (es->flags & ZEBRA_EVPNES_NHG_ACTIVE) ? es->nhg_id : 0; + + memset(&sph_filters, 0, sizeof(sph_filters)); + if (es->flags & ZEBRA_EVPNES_BYPASS) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug( + "es %s SPH filter disabled as it is in bypass", + es->esi_str); + } else { + if (listcount(es->es_vtep_list) > ES_VTEP_MAX_CNT) { + zlog_warn("es %s vtep count %d exceeds filter cnt %d", + es->esi_str, listcount(es->es_vtep_list), + ES_VTEP_MAX_CNT); + } else { + for (ALL_LIST_ELEMENTS_RO(es->es_vtep_list, node, + es_vtep)) { + if (es_vtep->flags + & ZEBRA_EVPNES_VTEP_DEL_IN_PROG) + continue; + sph_filters[sph_filter_cnt] = es_vtep->vtep_ip; + ++sph_filter_cnt; + } + } + } + + dplane_br_port_update(es->zif->ifp, !!(es->flags & ZEBRA_EVPNES_NON_DF), + sph_filter_cnt, sph_filters, backup_nhg_id); + + return true; +} + +/* returns TRUE if dplane entry was updated */ +static bool zebra_evpn_es_df_change(struct zebra_evpn_es *es, bool new_non_df, + const char *caller, const char *reason) +{ + bool old_non_df; + + old_non_df = !!(es->flags & ZEBRA_EVPNES_NON_DF); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("df-change es %s %s to %s; %s: %s", es->esi_str, + old_non_df ? "non-df" : "df", + new_non_df ? "non-df" : "df", caller, reason); + + if (old_non_df == new_non_df) + return false; + + if (new_non_df) + es->flags |= ZEBRA_EVPNES_NON_DF; + else + es->flags &= ~ZEBRA_EVPNES_NON_DF; + + /* update non-DF block filter in the dataplane */ + return zebra_evpn_es_br_port_dplane_update(es, __func__); +} + + +/* returns TRUE if dplane entry was updated */ +static bool zebra_evpn_es_run_df_election(struct zebra_evpn_es *es, + const char *caller) +{ + struct listnode *node = NULL; + struct zebra_evpn_es_vtep *es_vtep; + bool new_non_df = false; + + /* If the ES is not ready (i.e. not completely configured) there + * is no need to setup the BUM block filter + */ + if (!(es->flags & ZEBRA_EVPNES_LOCAL) + || (es->flags & ZEBRA_EVPNES_BYPASS) + || !zmh_info->es_originator_ip.s_addr) + return zebra_evpn_es_df_change(es, new_non_df, caller, + "not-ready"); + + /* if oper-state is down DF filtering must be on. when the link comes + * up again dataplane should block BUM till FRR has had the chance + * to run DF election again + */ + if (!(es->flags & ZEBRA_EVPNES_OPER_UP)) { + new_non_df = true; + return zebra_evpn_es_df_change(es, new_non_df, caller, + "oper-down"); + } + + /* ES was just created; we need to wait for the peers to rx the + * our Type-4 routes and for the switch to import the peers' Type-4 + * routes + */ + if (es->df_delay_timer) { + new_non_df = true; + return zebra_evpn_es_df_change(es, new_non_df, caller, + "df-delay"); + } + + for (ALL_LIST_ELEMENTS_RO(es->es_vtep_list, node, es_vtep)) { + /* Only VTEPs that have advertised the ESR can participate + * in DF election + */ + if (!(es_vtep->flags & ZEBRA_EVPNES_VTEP_RXED_ESR)) + continue; + + /* If the DF alg is not the same we should fall back to + * service-carving. But as service-carving is not supported + * we will stop forwarding BUM + */ + if (es_vtep->df_alg != EVPN_MH_DF_ALG_PREF) { + new_non_df = true; + break; + } + + /* Peer VTEP wins DF election if - + * the peer-VTEP has higher preference (or) + * the pref is the same but peer's IP address is lower + */ + if ((es_vtep->df_pref > es->df_pref) + || ((es_vtep->df_pref == es->df_pref) + && (es_vtep->vtep_ip.s_addr + < zmh_info->es_originator_ip.s_addr))) { + new_non_df = true; + break; + } + } + + return zebra_evpn_es_df_change(es, new_non_df, caller, "elected"); +} + +static void zebra_evpn_es_vtep_add(struct zebra_evpn_es *es, + struct in_addr vtep_ip, bool esr_rxed, + uint8_t df_alg, uint16_t df_pref) +{ + struct zebra_evpn_es_vtep *es_vtep; + bool old_esr_rxed; + bool dplane_updated = false; + + es_vtep = zebra_evpn_es_vtep_find(es, vtep_ip); + + if (!es_vtep) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s vtep %pI4 add", + es->esi_str, &vtep_ip); + es_vtep = zebra_evpn_es_vtep_new(es, vtep_ip); + /* update the L2-NHG associated with the ES */ + zebra_evpn_l2_nh_es_vtep_ref(es_vtep); + } + + old_esr_rxed = !!(es_vtep->flags & ZEBRA_EVPNES_VTEP_RXED_ESR); + if ((old_esr_rxed != esr_rxed) || (es_vtep->df_alg != df_alg) + || (es_vtep->df_pref != df_pref)) { + /* If any of the DF election params changed we need to re-run + * DF election + */ + if (esr_rxed) + es_vtep->flags |= ZEBRA_EVPNES_VTEP_RXED_ESR; + else + es_vtep->flags &= ~ZEBRA_EVPNES_VTEP_RXED_ESR; + es_vtep->df_alg = df_alg; + es_vtep->df_pref = df_pref; + dplane_updated = zebra_evpn_es_run_df_election(es, __func__); + } + /* add the vtep to the SPH list */ + if (!dplane_updated && (es->flags & ZEBRA_EVPNES_LOCAL)) + zebra_evpn_es_br_port_dplane_update(es, __func__); +} + +static void zebra_evpn_es_vtep_del(struct zebra_evpn_es *es, + struct in_addr vtep_ip) +{ + struct zebra_evpn_es_vtep *es_vtep; + bool dplane_updated = false; + + es_vtep = zebra_evpn_es_vtep_find(es, vtep_ip); + + if (es_vtep) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s vtep %pI4 del", + es->esi_str, &vtep_ip); + es_vtep->flags |= ZEBRA_EVPNES_VTEP_DEL_IN_PROG; + if (es_vtep->flags & ZEBRA_EVPNES_VTEP_RXED_ESR) { + es_vtep->flags &= ~ZEBRA_EVPNES_VTEP_RXED_ESR; + dplane_updated = + zebra_evpn_es_run_df_election(es, __func__); + } + /* remove the vtep from the SPH list */ + if (!dplane_updated && (es->flags & ZEBRA_EVPNES_LOCAL)) + zebra_evpn_es_br_port_dplane_update(es, __func__); + zebra_evpn_es_vtep_free(es_vtep); + } +} + +/* compare ES-IDs for the global ES RB tree */ +static int zebra_es_rb_cmp(const struct zebra_evpn_es *es1, + const struct zebra_evpn_es *es2) +{ + return memcmp(&es1->esi, &es2->esi, ESI_BYTES); +} +RB_GENERATE(zebra_es_rb_head, zebra_evpn_es, rb_node, zebra_es_rb_cmp); + +/* Lookup ES */ +struct zebra_evpn_es *zebra_evpn_es_find(const esi_t *esi) +{ + struct zebra_evpn_es tmp; + + memcpy(&tmp.esi, esi, sizeof(esi_t)); + return RB_FIND(zebra_es_rb_head, &zmh_info->es_rb_tree, &tmp); +} + +/* A new local es is created when a local-es-id and sysmac is configured + * against an interface. + */ +static struct zebra_evpn_es *zebra_evpn_es_new(const esi_t *esi) +{ + struct zebra_evpn_es *es; + + if (!memcmp(esi, zero_esi, sizeof(esi_t))) + return NULL; + + es = XCALLOC(MTYPE_ZES, sizeof(struct zebra_evpn_es)); + + /* fill in ESI */ + memcpy(&es->esi, esi, sizeof(esi_t)); + esi_to_str(&es->esi, es->esi_str, sizeof(es->esi_str)); + + /* Add to rb_tree */ + RB_INSERT(zebra_es_rb_head, &zmh_info->es_rb_tree, es); + + /* Initialise the ES-EVI list */ + es->es_evi_list = list_new(); + listset_app_node_mem(es->es_evi_list); + + /* Initialise the VTEP list */ + es->es_vtep_list = list_new(); + listset_app_node_mem(es->es_vtep_list); + es->es_vtep_list->cmp = zebra_evpn_es_vtep_cmp; + + /* mac entries associated with the ES */ + es->mac_list = list_new(); + listset_app_node_mem(es->mac_list); + + /* reserve a NHG */ + es->nhg_id = zebra_evpn_nhid_alloc(es); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s nhg %u new", es->esi_str, es->nhg_id); + + return es; +} + +/* Free a given ES - + * This just frees appropriate memory, caller should have taken other + * needed actions. + */ +static void zebra_evpn_es_free(struct zebra_evpn_es **esp) +{ + struct zebra_evpn_es *es = *esp; + + /* If the ES has a local or remote reference it cannot be freed. + * Free is also prevented if there are MAC entries referencing + * it. + */ + if ((es->flags & (ZEBRA_EVPNES_LOCAL | ZEBRA_EVPNES_REMOTE)) || + listcount(es->mac_list)) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s free", es->esi_str); + + /* If the NHG is still installed uninstall it and free the id */ + if (es->flags & ZEBRA_EVPNES_NHG_ACTIVE) { + es->flags &= ~ZEBRA_EVPNES_NHG_ACTIVE; + kernel_del_mac_nhg(es->nhg_id); + } + zebra_evpn_nhid_free(es->nhg_id, es); + + /* cleanup resources maintained against the ES */ + list_delete(&es->es_evi_list); + list_delete(&es->es_vtep_list); + list_delete(&es->mac_list); + + /* remove from the VNI-ESI rb tree */ + RB_REMOVE(zebra_es_rb_head, &zmh_info->es_rb_tree, es); + + XFREE(MTYPE_ZES, es); + + *esp = NULL; +} + +/* Inform BGP about local ES addition */ +static int zebra_evpn_es_send_add_to_client(struct zebra_evpn_es *es) +{ + struct zserv *client; + struct stream *s; + uint8_t oper_up; + bool bypass; + + client = zserv_find_client(ZEBRA_ROUTE_BGP, 0); + /* BGP may not be running. */ + if (!client) + return 0; + + s = stream_new(ZEBRA_MAX_PACKET_SIZ); + + zclient_create_header(s, ZEBRA_LOCAL_ES_ADD, zebra_vrf_get_evpn_id()); + stream_put(s, &es->esi, sizeof(esi_t)); + stream_put_ipv4(s, zmh_info->es_originator_ip.s_addr); + oper_up = !!(es->flags & ZEBRA_EVPNES_OPER_UP); + stream_putc(s, oper_up); + stream_putw(s, es->df_pref); + bypass = !!(es->flags & ZEBRA_EVPNES_BYPASS); + stream_putc(s, bypass); + + /* Write packet size. */ + stream_putw_at(s, 0, stream_get_endp(s)); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug( + "send add local es %s %pI4 active %u df_pref %u%s to %s", + es->esi_str, &zmh_info->es_originator_ip, oper_up, + es->df_pref, bypass ? " bypass" : "", + zebra_route_string(client->proto)); + + client->local_es_add_cnt++; + return zserv_send_message(client, s); +} + +/* Inform BGP about local ES deletion */ +static int zebra_evpn_es_send_del_to_client(struct zebra_evpn_es *es) +{ + struct zserv *client; + struct stream *s; + + client = zserv_find_client(ZEBRA_ROUTE_BGP, 0); + /* BGP may not be running. */ + if (!client) + return 0; + + s = stream_new(ZEBRA_MAX_PACKET_SIZ); + stream_reset(s); + + zclient_create_header(s, ZEBRA_LOCAL_ES_DEL, zebra_vrf_get_evpn_id()); + stream_put(s, &es->esi, sizeof(esi_t)); + + /* Write packet size. */ + stream_putw_at(s, 0, stream_get_endp(s)); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("send del local es %s to %s", es->esi_str, + zebra_route_string(client->proto)); + + client->local_es_del_cnt++; + return zserv_send_message(client, s); +} + +static void zebra_evpn_es_re_eval_send_to_client(struct zebra_evpn_es *es, + bool es_evi_re_reval) +{ + bool old_ready; + bool new_ready; + struct listnode *node; + struct zebra_evpn_es_evi *es_evi; + + old_ready = !!(es->flags & ZEBRA_EVPNES_READY_FOR_BGP); + + if ((es->flags & ZEBRA_EVPNES_LOCAL) && + zmh_info->es_originator_ip.s_addr) + es->flags |= ZEBRA_EVPNES_READY_FOR_BGP; + else + es->flags &= ~ZEBRA_EVPNES_READY_FOR_BGP; + + new_ready = !!(es->flags & ZEBRA_EVPNES_READY_FOR_BGP); + if (old_ready == new_ready) + return; + + if (new_ready) + zebra_evpn_es_send_add_to_client(es); + else + zebra_evpn_es_send_del_to_client(es); + + /* re-eval associated EVIs */ + if (es_evi_re_reval) { + for (ALL_LIST_ELEMENTS_RO(es->es_evi_list, node, es_evi)) { + if (!(es_evi->flags & ZEBRA_EVPNES_EVI_LOCAL)) + continue; + zebra_evpn_es_evi_re_eval_send_to_client(es_evi); + } + } +} + +void zebra_evpn_es_send_all_to_client(bool add) +{ + struct listnode *es_node; + struct listnode *evi_node; + struct zebra_evpn_es *es; + struct zebra_evpn_es_evi *es_evi; + + if (!zmh_info) + return; + + for (ALL_LIST_ELEMENTS_RO(zmh_info->local_es_list, es_node, es)) { + if (es->flags & ZEBRA_EVPNES_READY_FOR_BGP) { + if (add) + zebra_evpn_es_send_add_to_client(es); + for (ALL_LIST_ELEMENTS_RO(es->es_evi_list, + evi_node, es_evi)) { + if (!(es_evi->flags & + ZEBRA_EVPNES_EVI_READY_FOR_BGP)) + continue; + + if (add) + zebra_evpn_es_evi_send_to_client( + es, es_evi->zevpn, + true /* add */); + else + zebra_evpn_es_evi_send_to_client( + es, es_evi->zevpn, + false /* add */); + } + if (!add) + zebra_evpn_es_send_del_to_client(es); + } + } +} + +/* walk the vlan bitmap associated with the zif and create or delete + * es_evis for all vlans associated with a VNI. + * XXX: This API is really expensive. optimize later if possible. + */ +static void zebra_evpn_es_setup_evis(struct zebra_evpn_es *es) +{ + struct zebra_if *zif = es->zif; + uint16_t vid; + struct zebra_evpn_access_bd *acc_bd; + + if (!bf_is_inited(zif->vlan_bitmap)) + return; + + bf_for_each_set_bit(zif->vlan_bitmap, vid, IF_VLAN_BITMAP_MAX) { + acc_bd = zebra_evpn_acc_vl_find(vid); + if (acc_bd->zevpn) + zebra_evpn_local_es_evi_add(es, acc_bd->zevpn); + } +} + +static void zebra_evpn_flush_local_mac(struct zebra_mac *mac, + struct interface *ifp) +{ + struct zebra_if *zif; + struct interface *br_ifp; + vlanid_t vid; + + zif = ifp->info; + br_ifp = zif->brslave_info.br_if; + if (!br_ifp) + return; + + if (mac->zevpn->vxlan_if) { + zif = mac->zevpn->vxlan_if->info; + vid = zif->l2info.vxl.access_vlan; + } else { + vid = 0; + } + + /* delete the local mac from the dataplane */ + dplane_local_mac_del(ifp, br_ifp, vid, &mac->macaddr); + /* delete the local mac in zebra */ + zebra_evpn_del_local_mac(mac->zevpn, mac, true); +} + +static void zebra_evpn_es_flush_local_macs(struct zebra_evpn_es *es, + struct interface *ifp, bool add) +{ + struct zebra_mac *mac; + struct listnode *node; + struct listnode *nnode; + + for (ALL_LIST_ELEMENTS(es->mac_list, node, nnode, mac)) { + if (!CHECK_FLAG(mac->flags, ZEBRA_MAC_LOCAL)) + continue; + + /* If ES is being attached/detached from the access port we + * need to clear local activity and peer activity and start + * over */ + if (IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug("VNI %u mac %pEA update; local ES %s %s", + mac->zevpn->vni, + &mac->macaddr, + es->esi_str, add ? "add" : "del"); + zebra_evpn_flush_local_mac(mac, ifp); + } +} + +void zebra_evpn_es_local_br_port_update(struct zebra_if *zif) +{ + struct zebra_evpn_es *es = zif->es_info.es; + bool old_br_port = !!(es->flags & ZEBRA_EVPNES_BR_PORT); + bool new_br_port; + + if (zif->brslave_info.bridge_ifindex != IFINDEX_INTERNAL) + es->flags |= ZEBRA_EVPNES_BR_PORT; + else + es->flags &= ~ZEBRA_EVPNES_BR_PORT; + + new_br_port = !!(es->flags & ZEBRA_EVPNES_BR_PORT); + if (old_br_port == new_br_port) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s br_port change old %u new %u", es->esi_str, + old_br_port, new_br_port); + + /* update the dataplane br_port attrs */ + if (new_br_port && zebra_evpn_es_br_port_dplane_update_needed(es)) + zebra_evpn_es_br_port_dplane_update(es, __func__); +} + +/* On config of first local-ES turn off DAD */ +static void zebra_evpn_mh_dup_addr_detect_off(void) +{ + struct zebra_vrf *zvrf; + bool old_detect; + bool new_detect; + + if (zmh_info->flags & ZEBRA_EVPN_MH_DUP_ADDR_DETECT_OFF) + return; + + zvrf = zebra_vrf_get_evpn(); + old_detect = zebra_evpn_do_dup_addr_detect(zvrf); + zmh_info->flags |= ZEBRA_EVPN_MH_DUP_ADDR_DETECT_OFF; + new_detect = zebra_evpn_do_dup_addr_detect(zvrf); + + if (old_detect && !new_detect) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug( + "evpn-mh config caused DAD addr detect chg from %s to %s", + old_detect ? "on" : "off", + new_detect ? "on" : "off"); + zebra_vxlan_clear_dup_detect_vni_all(zvrf); + } +} + +/* On config of first local-ES turn off advertisement of STALE/DELAY/PROBE + * neighbors + */ +static void zebra_evpn_mh_advertise_reach_neigh_only(void) +{ + if (zmh_info->flags & ZEBRA_EVPN_MH_ADV_REACHABLE_NEIGH_ONLY) + return; + + zmh_info->flags |= ZEBRA_EVPN_MH_ADV_REACHABLE_NEIGH_ONLY; + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("evpn-mh: only REACHABLE neigh advertised"); + + /* XXX - if STALE/DELAY/PROBE neighs were previously advertised we + * need to withdraw them + */ +} + +/* On config of first local-ES turn on advertisement of local SVI-MAC */ +static void zebra_evpn_mh_advertise_svi_mac(void) +{ + if (zmh_info->flags & ZEBRA_EVPN_MH_ADV_SVI_MAC) + return; + + zmh_info->flags |= ZEBRA_EVPN_MH_ADV_SVI_MAC; + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("evpn-mh: advertise SVI MAC"); + + /* walk through all SVIs and see if we need to advertise the MAC */ + zebra_evpn_acc_vl_adv_svi_mac_all(); +} + +static void zebra_evpn_es_df_delay_exp_cb(struct thread *t) +{ + struct zebra_evpn_es *es; + + es = THREAD_ARG(t); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s df-delay expired", es->esi_str); + + zebra_evpn_es_run_df_election(es, __func__); +} + +/* currently there is no global config to turn on MH instead we use + * the addition of the first local Ethernet Segment as the trigger to + * init MH specific processing + */ +static void zebra_evpn_mh_on_first_local_es(void) +{ + zebra_evpn_mh_dup_addr_detect_off(); + zebra_evpn_mh_advertise_reach_neigh_only(); + zebra_evpn_mh_advertise_svi_mac(); +} + +static void zebra_evpn_es_local_info_set(struct zebra_evpn_es *es, + struct zebra_if *zif) +{ + if (es->flags & ZEBRA_EVPNES_LOCAL) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("local es %s add; nhg %u if %s", es->esi_str, + es->nhg_id, zif->ifp->name); + + zebra_evpn_mh_on_first_local_es(); + + es->flags |= ZEBRA_EVPNES_LOCAL; + listnode_init(&es->local_es_listnode, es); + listnode_add(zmh_info->local_es_list, &es->local_es_listnode); + + /* attach es to interface */ + zif->es_info.es = es; + es->df_pref = zif->es_info.df_pref ? zif->es_info.df_pref + : EVPN_MH_DF_PREF_DEFAULT; + + /* attach interface to es */ + es->zif = zif; + if (if_is_operative(zif->ifp)) + es->flags |= ZEBRA_EVPNES_OPER_UP; + + if (zif->brslave_info.bridge_ifindex != IFINDEX_INTERNAL) + es->flags |= ZEBRA_EVPNES_BR_PORT; + + /* inherit the bypass flag from the interface */ + if (zif->flags & ZIF_FLAG_LACP_BYPASS) + es->flags |= ZEBRA_EVPNES_BYPASS; + + /* setup base-vni if one doesn't already exist; the ES will get sent + * to BGP as a part of that process + */ + if (!zmh_info->es_base_evpn) + zebra_evpn_es_get_one_base_evpn(); + else + /* send notification to bgp */ + zebra_evpn_es_re_eval_send_to_client(es, + false /* es_evi_re_reval */); + + /* Start the DF delay timer on the local ES */ + if (!es->df_delay_timer) + thread_add_timer(zrouter.master, zebra_evpn_es_df_delay_exp_cb, + es, ZEBRA_EVPN_MH_DF_DELAY_TIME, + &es->df_delay_timer); + + /* See if the local VTEP can function as DF on the ES */ + if (!zebra_evpn_es_run_df_election(es, __func__)) { + /* check if the dplane entry needs to be re-programmed as a + * result of some thing other than DF status change + */ + if (zebra_evpn_es_br_port_dplane_update_needed(es)) + zebra_evpn_es_br_port_dplane_update(es, __func__); + } + + + /* Setup ES-EVIs for all VxLAN stretched VLANs associated with + * the zif + */ + zebra_evpn_es_setup_evis(es); + /* if there any local macs referring to the ES as dest we + * need to clear the contents and start over + */ + zebra_evpn_es_flush_local_macs(es, zif->ifp, true); + + /* inherit EVPN protodown flags on the access port */ + zebra_evpn_mh_update_protodown_es(es, true /*resync_dplane*/); +} + +static void zebra_evpn_es_local_info_clear(struct zebra_evpn_es **esp) +{ + struct zebra_if *zif; + struct zebra_evpn_es *es = *esp; + bool dplane_updated = false; + + if (!(es->flags & ZEBRA_EVPNES_LOCAL)) + return; + + zif = es->zif; + + /* if there any local macs referring to the ES as dest we + * need to clear the contents and start over + */ + zebra_evpn_es_flush_local_macs(es, zif->ifp, false); + + es->flags &= ~(ZEBRA_EVPNES_LOCAL | ZEBRA_EVPNES_READY_FOR_BGP); + + THREAD_OFF(es->df_delay_timer); + + /* clear EVPN protodown flags on the access port */ + zebra_evpn_mh_clear_protodown_es(es); + + /* remove the DF filter */ + dplane_updated = zebra_evpn_es_run_df_election(es, __func__); + + /* flush the BUM filters and backup NHG */ + if (!dplane_updated) + zebra_evpn_es_br_port_dplane_clear(es); + + /* clear the es from the parent interface */ + zif->es_info.es = NULL; + es->zif = NULL; + + /* clear all local flags associated with the ES */ + es->flags &= ~(ZEBRA_EVPNES_OPER_UP | ZEBRA_EVPNES_BR_PORT + | ZEBRA_EVPNES_BYPASS); + + /* remove from the ES list */ + list_delete_node(zmh_info->local_es_list, &es->local_es_listnode); + + /* free up the ES if there is no remote reference */ + zebra_evpn_es_free(esp); +} + +/* Delete an ethernet segment and inform BGP */ +static void zebra_evpn_local_es_del(struct zebra_evpn_es **esp) +{ + struct zebra_evpn_es_evi *es_evi; + struct listnode *node = NULL; + struct listnode *nnode = NULL; + struct zebra_if *zif; + struct zebra_evpn_es *es = *esp; + + if (!CHECK_FLAG(es->flags, ZEBRA_EVPNES_LOCAL)) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) { + zif = es->zif; + zlog_debug("local es %s del; nhg %u if %s", es->esi_str, + es->nhg_id, zif ? zif->ifp->name : "-"); + } + + /* remove all ES-EVIs associated with the ES */ + for (ALL_LIST_ELEMENTS(es->es_evi_list, node, nnode, es_evi)) + zebra_evpn_local_es_evi_do_del(es_evi); + + /* send a del if the ES had been sent to BGP earlier */ + if (es->flags & ZEBRA_EVPNES_READY_FOR_BGP) + zebra_evpn_es_send_del_to_client(es); + + zebra_evpn_es_local_info_clear(esp); +} + +/* eval remote info associated with the ES */ +static void zebra_evpn_es_remote_info_re_eval(struct zebra_evpn_es **esp) +{ + struct zebra_evpn_es *es = *esp; + + /* if there are remote VTEPs the ES-EVI is classified as "remote" */ + if (listcount(es->es_vtep_list)) { + if (!(es->flags & ZEBRA_EVPNES_REMOTE)) { + es->flags |= ZEBRA_EVPNES_REMOTE; + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("remote es %s add; nhg %u", + es->esi_str, es->nhg_id); + } + } else { + if (es->flags & ZEBRA_EVPNES_REMOTE) { + es->flags &= ~ZEBRA_EVPNES_REMOTE; + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("remote es %s del; nhg %u", + es->esi_str, es->nhg_id); + zebra_evpn_es_free(esp); + } + } +} + +/* A new local es is created when a local-es-id and sysmac is configured + * against an interface. + */ +static int zebra_evpn_local_es_update(struct zebra_if *zif, esi_t *esi) +{ + struct zebra_evpn_es *old_es = zif->es_info.es; + struct zebra_evpn_es *es; + + if (old_es && !memcmp(&old_es->esi, esi, sizeof(*esi))) + /* dup - nothing to be done */ + return 0; + + /* release the old_es against the zif */ + if (old_es) + zebra_evpn_local_es_del(&old_es); + + es = zebra_evpn_es_find(esi); + if (es) { + /* if it exists against another interface flag an error */ + if (es->zif && es->zif != zif) + return -1; + } else { + /* create new es */ + es = zebra_evpn_es_new(esi); + } + + memcpy(&zif->es_info.esi, esi, sizeof(*esi)); + if (es) + zebra_evpn_es_local_info_set(es, zif); + + return 0; +} + +static int zebra_evpn_type3_esi_update(struct zebra_if *zif, uint32_t lid, + struct ethaddr *sysmac) +{ + struct zebra_evpn_es *old_es = zif->es_info.es; + esi_t esi; + int offset = 0; + int field_bytes = 0; + + /* Complete config of the ES-ID bootstraps the ES */ + if (!lid || is_zero_mac(sysmac)) { + /* clear old esi */ + memset(&zif->es_info.esi, 0, sizeof(zif->es_info.esi)); + /* if in ES is attached to zif delete it */ + if (old_es) + zebra_evpn_local_es_del(&old_es); + return 0; + } + + /* build 10-byte type-3-ESI - + * Type(1-byte), MAC(6-bytes), ES-LID (3-bytes) + */ + field_bytes = 1; + esi.val[offset] = ESI_TYPE_MAC; + offset += field_bytes; + + field_bytes = ETH_ALEN; + memcpy(&esi.val[offset], (uint8_t *)sysmac, field_bytes); + offset += field_bytes; + + esi.val[offset++] = (uint8_t)(lid >> 16); + esi.val[offset++] = (uint8_t)(lid >> 8); + esi.val[offset++] = (uint8_t)lid; + + return zebra_evpn_local_es_update(zif, &esi); +} + +int zebra_evpn_remote_es_del(const esi_t *esi, struct in_addr vtep_ip) +{ + char buf[ESI_STR_LEN]; + struct zebra_evpn_es *es; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("remote es %s vtep %pI4 del", + esi_to_str(esi, buf, sizeof(buf)), &vtep_ip); + + es = zebra_evpn_es_find(esi); + if (!es) { + zlog_warn("remote es %s vtep %pI4 del failed, es missing", + esi_to_str(esi, buf, sizeof(buf)), &vtep_ip); + return -1; + } + + zebra_evpn_es_vtep_del(es, vtep_ip); + zebra_evpn_es_remote_info_re_eval(&es); + + return 0; +} + +/* force delete a remote ES on the way down */ +static void zebra_evpn_remote_es_flush(struct zebra_evpn_es **esp) +{ + struct zebra_evpn_es_vtep *es_vtep; + struct listnode *node; + struct listnode *nnode; + struct zebra_evpn_es *es = *esp; + + for (ALL_LIST_ELEMENTS(es->es_vtep_list, node, nnode, es_vtep)) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s vtep %pI4 flush", + es->esi_str, + &es_vtep->vtep_ip); + zebra_evpn_es_vtep_free(es_vtep); + } + zebra_evpn_es_remote_info_re_eval(esp); +} + +int zebra_evpn_remote_es_add(const esi_t *esi, struct in_addr vtep_ip, + bool esr_rxed, uint8_t df_alg, uint16_t df_pref) +{ + char buf[ESI_STR_LEN]; + struct zebra_evpn_es *es; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("remote es %s vtep %pI4 add %s df_alg %d df_pref %d", + esi_to_str(esi, buf, sizeof(buf)), + &vtep_ip, esr_rxed ? "esr" : "", df_alg, + df_pref); + + es = zebra_evpn_es_find(esi); + if (!es) { + es = zebra_evpn_es_new(esi); + if (!es) { + zlog_warn( + "remote es %s vtep %pI4 add failed, es missing", + esi_to_str(esi, buf, sizeof(buf)), &vtep_ip); + return -1; + } + } + + if (df_alg != EVPN_MH_DF_ALG_PREF) + zlog_warn( + "remote es %s vtep %pI4 add %s with unsupported df_alg %d", + esi_to_str(esi, buf, sizeof(buf)), &vtep_ip, + esr_rxed ? "esr" : "", df_alg); + + zebra_evpn_es_vtep_add(es, vtep_ip, esr_rxed, df_alg, df_pref); + zebra_evpn_es_remote_info_re_eval(&es); + + return 0; +} + +void zebra_evpn_proc_remote_es(ZAPI_HANDLER_ARGS) +{ + struct stream *s; + struct in_addr vtep_ip; + esi_t esi; + + if (!is_evpn_enabled()) { + zlog_debug( + "%s: EVPN not enabled yet we received a es_add zapi call", + __func__); + return; + } + + memset(&esi, 0, sizeof(esi_t)); + s = msg; + + STREAM_GET(&esi, s, sizeof(esi_t)); + STREAM_GET(&vtep_ip.s_addr, s, sizeof(vtep_ip.s_addr)); + + if (hdr->command == ZEBRA_REMOTE_ES_VTEP_ADD) { + uint32_t zapi_flags; + uint8_t df_alg; + uint16_t df_pref; + bool esr_rxed; + + STREAM_GETL(s, zapi_flags); + esr_rxed = (zapi_flags & ZAPI_ES_VTEP_FLAG_ESR_RXED) ? true + : false; + STREAM_GETC(s, df_alg); + STREAM_GETW(s, df_pref); + zebra_rib_queue_evpn_rem_es_add(&esi, &vtep_ip, esr_rxed, + df_alg, df_pref); + } else { + zebra_rib_queue_evpn_rem_es_del(&esi, &vtep_ip); + } + +stream_failure: + return; +} + +void zebra_evpn_es_mac_deref_entry(struct zebra_mac *mac) +{ + struct zebra_evpn_es *es = mac->es; + + mac->es = NULL; + if (!es) + return; + + list_delete_node(es->mac_list, &mac->es_listnode); + if (!listcount(es->mac_list)) + zebra_evpn_es_free(&es); +} + +/* Associate a MAC entry with a local or remote ES. Returns false if there + * was no ES change. + */ +bool zebra_evpn_es_mac_ref_entry(struct zebra_mac *mac, + struct zebra_evpn_es *es) +{ + if (mac->es == es) + return false; + + if (mac->es) + zebra_evpn_es_mac_deref_entry(mac); + + if (!es) + return true; + + mac->es = es; + listnode_init(&mac->es_listnode, mac); + listnode_add(es->mac_list, &mac->es_listnode); + + return true; +} + +bool zebra_evpn_es_mac_ref(struct zebra_mac *mac, const esi_t *esi) +{ + struct zebra_evpn_es *es; + + es = zebra_evpn_es_find(esi); + if (!es) { + /* If non-zero esi implicitly create a new ES */ + if (memcmp(esi, zero_esi, sizeof(esi_t))) { + es = zebra_evpn_es_new(esi); + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("auto es %s add on mac ref", + es->esi_str); + } + } + + return zebra_evpn_es_mac_ref_entry(mac, es); +} + +/* Inform BGP about local ES-EVI add or del */ +static int zebra_evpn_es_evi_send_to_client(struct zebra_evpn_es *es, + struct zebra_evpn *zevpn, bool add) +{ + struct zserv *client; + struct stream *s; + + client = zserv_find_client(ZEBRA_ROUTE_BGP, 0); + /* BGP may not be running. */ + if (!client) + return 0; + + s = stream_new(ZEBRA_MAX_PACKET_SIZ); + + zclient_create_header(s, + add ? ZEBRA_LOCAL_ES_EVI_ADD : ZEBRA_LOCAL_ES_EVI_DEL, + zebra_vrf_get_evpn_id()); + stream_put(s, &es->esi, sizeof(esi_t)); + stream_putl(s, zevpn->vni); + + /* Write packet size. */ + stream_putw_at(s, 0, stream_get_endp(s)); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("send %s local es %s evi %u to %s", + add ? "add" : "del", + es->esi_str, zevpn->vni, + zebra_route_string(client->proto)); + + client->local_es_add_cnt++; + return zserv_send_message(client, s); +} + +/* sysmac part of a local ESI has changed */ +static int zebra_evpn_es_sys_mac_update(struct zebra_if *zif, + struct ethaddr *sysmac) +{ + int rv; + + rv = zebra_evpn_type3_esi_update(zif, zif->es_info.lid, sysmac); + if (!rv) + memcpy(&zif->es_info.sysmac, sysmac, sizeof(struct ethaddr)); + + return rv; +} + +/* local-ID part of ESI has changed */ +static int zebra_evpn_es_lid_update(struct zebra_if *zif, uint32_t lid) +{ + int rv; + + rv = zebra_evpn_type3_esi_update(zif, lid, &zif->es_info.sysmac); + if (!rv) + zif->es_info.lid = lid; + + return rv; +} + +/* type-0 esi has changed */ +static int zebra_evpn_es_type0_esi_update(struct zebra_if *zif, esi_t *esi) +{ + int rv; + + rv = zebra_evpn_local_es_update(zif, esi); + + /* clear the old es_lid, es_sysmac - type-0 is being set so old + * type-3 params need to be flushed + */ + memset(&zif->es_info.sysmac, 0, sizeof(struct ethaddr)); + zif->es_info.lid = 0; + + return rv; +} + +void zebra_evpn_es_cleanup(void) +{ + struct zebra_evpn_es *es; + struct zebra_evpn_es *es_next; + + RB_FOREACH_SAFE(es, zebra_es_rb_head, + &zmh_info->es_rb_tree, es_next) { + zebra_evpn_local_es_del(&es); + if (es) + zebra_evpn_remote_es_flush(&es); + } +} + +static void zebra_evpn_es_df_pref_update(struct zebra_if *zif, uint16_t df_pref) +{ + struct zebra_evpn_es *es; + uint16_t tmp_pref; + + if (zif->es_info.df_pref == df_pref) + return; + + zif->es_info.df_pref = df_pref; + es = zif->es_info.es; + + if (!es) + return; + + tmp_pref = zif->es_info.df_pref ? zif->es_info.df_pref + : EVPN_MH_DF_PREF_DEFAULT; + + if (es->df_pref == tmp_pref) + return; + + es->df_pref = tmp_pref; + /* run df election */ + zebra_evpn_es_run_df_election(es, __func__); + /* notify bgp */ + if (es->flags & ZEBRA_EVPNES_READY_FOR_BGP) + zebra_evpn_es_send_add_to_client(es); +} + +/* If bypass mode on an es changed we set all local macs to + * inactive and drop the sync info + */ +static void zebra_evpn_es_bypass_update_macs(struct zebra_evpn_es *es, + struct interface *ifp, bool bypass) +{ + struct zebra_mac *mac; + struct listnode *node; + struct listnode *nnode; + struct zebra_if *zif; + + /* Flush all MACs linked to the ES */ + for (ALL_LIST_ELEMENTS(es->mac_list, node, nnode, mac)) { + if (!CHECK_FLAG(mac->flags, ZEBRA_MAC_LOCAL)) + continue; + + if (IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug("VNI %u mac %pEA %s update es %s", + mac->zevpn->vni, + &mac->macaddr, + bypass ? "bypass" : "non-bypass", + es->esi_str); + zebra_evpn_flush_local_mac(mac, ifp); + } + + /* While in bypass-mode locally learnt MACs are linked + * to the access port instead of the ES + */ + zif = ifp->info; + if (!zif->mac_list) + return; + + for (ALL_LIST_ELEMENTS(zif->mac_list, node, nnode, mac)) { + if (!CHECK_FLAG(mac->flags, ZEBRA_MAC_LOCAL)) + continue; + + if (IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug("VNI %u mac %pEA %s update ifp %s", + mac->zevpn->vni, + &mac->macaddr, + bypass ? "bypass" : "non-bypass", ifp->name); + zebra_evpn_flush_local_mac(mac, ifp); + } +} + +void zebra_evpn_es_bypass_update(struct zebra_evpn_es *es, + struct interface *ifp, bool bypass) +{ + bool old_bypass; + bool dplane_updated; + + old_bypass = !!(es->flags & ZEBRA_EVPNES_BYPASS); + if (old_bypass == bypass) + return; + + if (bypass) + es->flags |= ZEBRA_EVPNES_BYPASS; + else + es->flags &= ~ZEBRA_EVPNES_BYPASS; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("bond %s es %s lacp bypass changed to %s", ifp->name, + es->esi_str, bypass ? "on" : "off"); + + /* send bypass update to BGP */ + if (es->flags & ZEBRA_EVPNES_READY_FOR_BGP) + zebra_evpn_es_send_add_to_client(es); + + zebra_evpn_es_bypass_update_macs(es, ifp, bypass); + + /* re-run DF election */ + dplane_updated = zebra_evpn_es_run_df_election(es, __func__); + + /* disable SPH filter */ + if (!dplane_updated && (es->flags & ZEBRA_EVPNES_LOCAL) + && (listcount(es->es_vtep_list) > ES_VTEP_MAX_CNT)) + zebra_evpn_es_br_port_dplane_update(es, __func__); +} + +static void zebra_evpn_es_bypass_cfg_update(struct zebra_if *zif, bool bypass) +{ + bool old_bypass = !!(zif->es_info.flags & ZIF_CFG_ES_FLAG_BYPASS); + + if (old_bypass == bypass) + return; + + if (bypass) + zif->es_info.flags |= ZIF_CFG_ES_FLAG_BYPASS; + else + zif->es_info.flags &= ~ZIF_CFG_ES_FLAG_BYPASS; + + + if (zif->es_info.es) + zebra_evpn_es_bypass_update(zif->es_info.es, zif->ifp, bypass); +} + + +/* Only certain types of access ports can be setup as an Ethernet Segment */ +bool zebra_evpn_is_if_es_capable(struct zebra_if *zif) +{ + if (zif->zif_type == ZEBRA_IF_BOND) + return true; + + /* relax the checks to allow config to be applied in zebra + * before interface is rxed from the kernel + */ + if (zif->ifp->ifindex == IFINDEX_INTERNAL) + return true; + + /* XXX: allow swpX i.e. a regular ethernet port to be an ES link too */ + return false; +} + +void zebra_evpn_if_es_print(struct vty *vty, json_object *json, + struct zebra_if *zif) +{ + char buf[ETHER_ADDR_STRLEN]; + char esi_buf[ESI_STR_LEN]; + + if (json) { + json_object *json_evpn; + + json_evpn = json_object_new_object(); + json_object_object_add(json, "evpnMh", json_evpn); + + if (zif->es_info.lid || !is_zero_mac(&zif->es_info.sysmac)) { + json_object_int_add(json_evpn, "esId", + zif->es_info.lid); + json_object_string_add( + json_evpn, "esSysmac", + prefix_mac2str(&zif->es_info.sysmac, buf, + sizeof(buf))); + } else if (memcmp(&zif->es_info.esi, zero_esi, + sizeof(*zero_esi))) { + json_object_string_add(json_evpn, "esId", + esi_to_str(&zif->es_info.esi, + esi_buf, + sizeof(esi_buf))); + } + + if (zif->flags & ZIF_FLAG_EVPN_MH_UPLINK) + json_object_string_add( + json_evpn, "uplink", + CHECK_FLAG(zif->flags, + ZIF_FLAG_EVPN_MH_UPLINK_OPER_UP) + ? "up" + : "down"); + } else { + char mh_buf[80]; + bool vty_print = false; + + mh_buf[0] = '\0'; + strlcat(mh_buf, " EVPN-MH:", sizeof(mh_buf)); + if (zif->es_info.lid || !is_zero_mac(&zif->es_info.sysmac)) { + vty_print = true; + snprintf(mh_buf + strlen(mh_buf), + sizeof(mh_buf) - strlen(mh_buf), + " ES id %u ES sysmac %s", zif->es_info.lid, + prefix_mac2str(&zif->es_info.sysmac, buf, + sizeof(buf))); + } else if (memcmp(&zif->es_info.esi, zero_esi, + sizeof(*zero_esi))) { + vty_print = true; + snprintf(mh_buf + strnlen(mh_buf, sizeof(mh_buf)), + sizeof(mh_buf) + - strnlen(mh_buf, sizeof(mh_buf)), + " ES id %s", + esi_to_str(&zif->es_info.esi, esi_buf, + sizeof(esi_buf))); + } + + if (zif->flags & ZIF_FLAG_EVPN_MH_UPLINK) { + vty_print = true; + if (zif->flags & ZIF_FLAG_EVPN_MH_UPLINK_OPER_UP) + strlcat(mh_buf, " uplink (up)", sizeof(mh_buf)); + else + strlcat(mh_buf, " uplink (down)", + sizeof(mh_buf)); + } + + if (vty_print) + vty_out(vty, "%s\n", mh_buf); + } +} + +static void zebra_evpn_local_mac_oper_state_change(struct zebra_evpn_es *es) +{ + struct zebra_mac *mac; + struct listnode *node; + + /* If fast-failover is supported by the dataplane via the use + * of an ES backup NHG there is nothing to be done in the + * control plane + */ + if (!(zmh_info->flags & ZEBRA_EVPN_MH_REDIRECT_OFF)) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES || IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug("mac slow-fail on es %s %s ", es->esi_str, + (es->flags & ZEBRA_EVPNES_OPER_UP) ? "up" : "down"); + + for (ALL_LIST_ELEMENTS_RO(es->mac_list, node, mac)) { + if (!(mac->flags & ZEBRA_MAC_LOCAL) + || !zebra_evpn_mac_is_static(mac)) + continue; + + if (es->flags & ZEBRA_EVPNES_OPER_UP) { + if (IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug( + "VNI %u mac %pEA move to acc %s es %s %s ", + mac->zevpn->vni, + &mac->macaddr, + es->zif->ifp->name, es->esi_str, + (es->flags & ZEBRA_EVPNES_OPER_UP) + ? "up" + : "down"); + /* switch the local macs to access port */ + if (zebra_evpn_sync_mac_dp_install( + mac, false /*set_inactive*/, + false /*force_clear_static*/, __func__) + < 0) + /* if the local mac install fails get rid of the + * old rem entry + */ + zebra_evpn_rem_mac_uninstall(mac->zevpn, mac, + true /*force*/); + } else { + /* switch the local macs to network port. if there + * is no active NHG we don't bother deleting the MAC; + * that is left up to the dataplane to handle. + */ + if (!(es->flags & ZEBRA_EVPNES_NHG_ACTIVE)) + continue; + if (IS_ZEBRA_DEBUG_EVPN_MH_MAC) + zlog_debug( + "VNI %u mac %pEA move to nhg %u es %s %s ", + mac->zevpn->vni, + &mac->macaddr, + es->nhg_id, es->esi_str, + (es->flags & ZEBRA_EVPNES_OPER_UP) + ? "up" + : "down"); + zebra_evpn_rem_mac_install(mac->zevpn, mac, + true /*was_static*/); + } + } +} + +void zebra_evpn_es_if_oper_state_change(struct zebra_if *zif, bool up) +{ + struct zebra_evpn_es *es = zif->es_info.es; + bool old_up = !!(es->flags & ZEBRA_EVPNES_OPER_UP); + + if (old_up == up) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es %s state changed to %s ", + es->esi_str, + up ? "up" : "down"); + if (up) + es->flags |= ZEBRA_EVPNES_OPER_UP; + else + es->flags &= ~ZEBRA_EVPNES_OPER_UP; + + zebra_evpn_es_run_df_election(es, __func__); + zebra_evpn_local_mac_oper_state_change(es); + + /* inform BGP of the ES oper state change */ + if (es->flags & ZEBRA_EVPNES_READY_FOR_BGP) + zebra_evpn_es_send_add_to_client(es); +} + +static char *zebra_evpn_es_vtep_str(char *vtep_str, struct zebra_evpn_es *es, + uint8_t vtep_str_size) +{ + struct zebra_evpn_es_vtep *zvtep; + struct listnode *node; + bool first = true; + char ip_buf[INET6_ADDRSTRLEN]; + + vtep_str[0] = '\0'; + for (ALL_LIST_ELEMENTS_RO(es->es_vtep_list, node, zvtep)) { + if (first) { + first = false; + strlcat(vtep_str, + inet_ntop(AF_INET, &zvtep->vtep_ip, ip_buf, + sizeof(ip_buf)), + vtep_str_size); + } else { + strlcat(vtep_str, ",", vtep_str_size); + strlcat(vtep_str, + inet_ntop(AF_INET, &zvtep->vtep_ip, ip_buf, + sizeof(ip_buf)), + vtep_str_size); + } + } + return vtep_str; +} + +static void zebra_evpn_es_json_vtep_fill(struct zebra_evpn_es *es, + json_object *json_vteps) +{ + struct zebra_evpn_es_vtep *es_vtep; + struct listnode *node; + json_object *json_vtep_entry; + char alg_buf[EVPN_DF_ALG_STR_LEN]; + + for (ALL_LIST_ELEMENTS_RO(es->es_vtep_list, node, es_vtep)) { + json_vtep_entry = json_object_new_object(); + json_object_string_addf(json_vtep_entry, "vtep", "%pI4", + &es_vtep->vtep_ip); + if (es_vtep->flags & ZEBRA_EVPNES_VTEP_RXED_ESR) { + json_object_string_add( + json_vtep_entry, "dfAlgorithm", + evpn_es_df_alg2str(es_vtep->df_alg, alg_buf, + sizeof(alg_buf))); + json_object_int_add(json_vtep_entry, "dfPreference", + es_vtep->df_pref); + } + if (es_vtep->nh) + json_object_int_add(json_vtep_entry, "nexthopId", + es_vtep->nh->nh_id); + json_object_array_add(json_vteps, json_vtep_entry); + } +} + +static void zebra_evpn_es_show_entry(struct vty *vty, struct zebra_evpn_es *es, + json_object *json_array) +{ + char type_str[5]; + char vtep_str[ES_VTEP_LIST_STR_SZ]; + + if (json_array) { + json_object *json = NULL; + json_object *json_vteps; + json_object *json_flags; + + json = json_object_new_object(); + json_object_string_add(json, "esi", es->esi_str); + + if (es->flags + & (ZEBRA_EVPNES_LOCAL | ZEBRA_EVPNES_REMOTE + | ZEBRA_EVPNES_NON_DF)) { + json_flags = json_object_new_array(); + if (es->flags & ZEBRA_EVPNES_LOCAL) + json_array_string_add(json_flags, "local"); + if (es->flags & ZEBRA_EVPNES_REMOTE) + json_array_string_add(json_flags, "remote"); + if (es->flags & ZEBRA_EVPNES_NON_DF) + json_array_string_add(json_flags, "nonDF"); + if (es->flags & ZEBRA_EVPNES_BYPASS) + json_array_string_add(json_flags, "bypass"); + json_object_object_add(json, "flags", json_flags); + } + + if (es->zif) + json_object_string_add(json, "accessPort", + es->zif->ifp->name); + + if (listcount(es->es_vtep_list)) { + json_vteps = json_object_new_array(); + zebra_evpn_es_json_vtep_fill(es, json_vteps); + json_object_object_add(json, "vteps", json_vteps); + } + json_object_array_add(json_array, json); + } else { + type_str[0] = '\0'; + if (es->flags & ZEBRA_EVPNES_LOCAL) + strlcat(type_str, "L", sizeof(type_str)); + if (es->flags & ZEBRA_EVPNES_REMOTE) + strlcat(type_str, "R", sizeof(type_str)); + if (es->flags & ZEBRA_EVPNES_NON_DF) + strlcat(type_str, "N", sizeof(type_str)); + if (es->flags & ZEBRA_EVPNES_BYPASS) + strlcat(type_str, "B", sizeof(type_str)); + + zebra_evpn_es_vtep_str(vtep_str, es, sizeof(vtep_str)); + + vty_out(vty, "%-30s %-4s %-21s %s\n", + es->esi_str, type_str, + es->zif ? es->zif->ifp->name : "-", + vtep_str); + } +} + +static void zebra_evpn_es_show_entry_detail(struct vty *vty, + struct zebra_evpn_es *es, json_object *json) +{ + char type_str[80]; + char alg_buf[EVPN_DF_ALG_STR_LEN]; + struct zebra_evpn_es_vtep *es_vtep; + struct listnode *node; + char thread_buf[THREAD_TIMER_STRLEN]; + + if (json) { + json_object *json_vteps; + json_object *json_flags; + + json_object_string_add(json, "esi", es->esi_str); + if (es->zif) + json_object_string_add(json, "accessPort", + es->zif->ifp->name); + + + if (es->flags) { + json_flags = json_object_new_array(); + if (es->flags & ZEBRA_EVPNES_LOCAL) + json_array_string_add(json_flags, "local"); + if (es->flags & ZEBRA_EVPNES_REMOTE) + json_array_string_add(json_flags, "remote"); + if (es->flags & ZEBRA_EVPNES_NON_DF) + json_array_string_add(json_flags, "nonDF"); + if (es->flags & ZEBRA_EVPNES_BYPASS) + json_array_string_add(json_flags, "bypass"); + if (es->flags & ZEBRA_EVPNES_READY_FOR_BGP) + json_array_string_add(json_flags, + "readyForBgp"); + if (es->flags & ZEBRA_EVPNES_BR_PORT) + json_array_string_add(json_flags, "bridgePort"); + if (es->flags & ZEBRA_EVPNES_OPER_UP) + json_array_string_add(json_flags, "operUp"); + if (es->flags & ZEBRA_EVPNES_NHG_ACTIVE) + json_array_string_add(json_flags, + "nexthopGroupActive"); + json_object_object_add(json, "flags", json_flags); + } + + json_object_int_add(json, "vniCount", + listcount(es->es_evi_list)); + json_object_int_add(json, "macCount", listcount(es->mac_list)); + json_object_int_add(json, "dfPreference", es->df_pref); + if (es->df_delay_timer) + json_object_string_add( + json, "dfDelayTimer", + thread_timer_to_hhmmss(thread_buf, + sizeof(thread_buf), + es->df_delay_timer)); + json_object_int_add(json, "nexthopGroup", es->nhg_id); + if (listcount(es->es_vtep_list)) { + json_vteps = json_object_new_array(); + zebra_evpn_es_json_vtep_fill(es, json_vteps); + json_object_object_add(json, "vteps", json_vteps); + } + } else { + type_str[0] = '\0'; + if (es->flags & ZEBRA_EVPNES_LOCAL) + strlcat(type_str, "Local", sizeof(type_str)); + if (es->flags & ZEBRA_EVPNES_REMOTE) { + if (strnlen(type_str, sizeof(type_str))) + strlcat(type_str, ",", sizeof(type_str)); + strlcat(type_str, "Remote", sizeof(type_str)); + } + + vty_out(vty, "ESI: %s\n", es->esi_str); + vty_out(vty, " Type: %s\n", type_str); + vty_out(vty, " Interface: %s\n", + (es->zif) ? + es->zif->ifp->name : "-"); + if (es->flags & ZEBRA_EVPNES_LOCAL) { + vty_out(vty, " State: %s\n", + (es->flags & ZEBRA_EVPNES_OPER_UP) ? "up" + : "down"); + vty_out(vty, " Bridge port: %s\n", + (es->flags & ZEBRA_EVPNES_BR_PORT) ? "yes" + : "no"); + } + vty_out(vty, " Ready for BGP: %s\n", + (es->flags & ZEBRA_EVPNES_READY_FOR_BGP) ? + "yes" : "no"); + if (es->flags & ZEBRA_EVPNES_BYPASS) + vty_out(vty, " LACP bypass: on\n"); + vty_out(vty, " VNI Count: %d\n", listcount(es->es_evi_list)); + vty_out(vty, " MAC Count: %d\n", listcount(es->mac_list)); + if (es->flags & ZEBRA_EVPNES_LOCAL) + vty_out(vty, " DF status: %s \n", + (es->flags & ZEBRA_EVPNES_NON_DF) ? "non-df" + : "df"); + if (es->df_delay_timer) + vty_out(vty, " DF delay: %s\n", + thread_timer_to_hhmmss(thread_buf, + sizeof(thread_buf), + es->df_delay_timer)); + vty_out(vty, " DF preference: %u\n", es->df_pref); + vty_out(vty, " Nexthop group: %u\n", es->nhg_id); + vty_out(vty, " VTEPs:\n"); + for (ALL_LIST_ELEMENTS_RO(es->es_vtep_list, node, es_vtep)) { + vty_out(vty, " %pI4", + &es_vtep->vtep_ip); + if (es_vtep->flags & ZEBRA_EVPNES_VTEP_RXED_ESR) + vty_out(vty, " df_alg: %s df_pref: %d", + evpn_es_df_alg2str(es_vtep->df_alg, + alg_buf, + sizeof(alg_buf)), + es_vtep->df_pref); + vty_out(vty, " nh: %u\n", + es_vtep->nh ? es_vtep->nh->nh_id : 0); + } + + vty_out(vty, "\n"); + } +} + +void zebra_evpn_es_show(struct vty *vty, bool uj) +{ + struct zebra_evpn_es *es; + json_object *json_array = NULL; + + if (uj) { + json_array = json_object_new_array(); + } else { + vty_out(vty, "Type: B bypass, L local, R remote, N non-DF\n"); + vty_out(vty, "%-30s %-4s %-21s %s\n", + "ESI", "Type", "ES-IF", "VTEPs"); + } + + RB_FOREACH(es, zebra_es_rb_head, &zmh_info->es_rb_tree) + zebra_evpn_es_show_entry(vty, es, json_array); + + if (uj) + vty_json(vty, json_array); +} + +void zebra_evpn_es_show_detail(struct vty *vty, bool uj) +{ + struct zebra_evpn_es *es; + json_object *json_array = NULL; + + if (uj) + json_array = json_object_new_array(); + + RB_FOREACH (es, zebra_es_rb_head, &zmh_info->es_rb_tree) { + json_object *json = NULL; + + if (uj) + json = json_object_new_object(); + zebra_evpn_es_show_entry_detail(vty, es, json); + if (uj) + json_object_array_add(json_array, json); + } + + if (uj) + vty_json(vty, json_array); +} + +void zebra_evpn_es_show_esi(struct vty *vty, bool uj, esi_t *esi) +{ + struct zebra_evpn_es *es; + char esi_str[ESI_STR_LEN]; + json_object *json = NULL; + + if (uj) + json = json_object_new_object(); + + es = zebra_evpn_es_find(esi); + + if (es) { + zebra_evpn_es_show_entry_detail(vty, es, json); + } else { + if (!uj) { + esi_to_str(esi, esi_str, sizeof(esi_str)); + vty_out(vty, "ESI %s does not exist\n", esi_str); + } + } + + if (uj) + vty_json(vty, json); +} + +int zebra_evpn_mh_if_write(struct vty *vty, struct interface *ifp) +{ + struct zebra_if *zif = ifp->info; + char buf[ETHER_ADDR_STRLEN]; + bool type_3_esi = false; + char esi_buf[ESI_STR_LEN]; + + if (zif->es_info.lid) { + vty_out(vty, " evpn mh es-id %u\n", zif->es_info.lid); + type_3_esi = true; + } + + if (!is_zero_mac(&zif->es_info.sysmac)) { + vty_out(vty, " evpn mh es-sys-mac %s\n", + prefix_mac2str(&zif->es_info.sysmac, + buf, sizeof(buf))); + type_3_esi = true; + } + + if (!type_3_esi + && memcmp(&zif->es_info.esi, zero_esi, sizeof(*zero_esi))) + vty_out(vty, " evpn mh es-id %s\n", + esi_to_str(&zif->es_info.esi, esi_buf, sizeof(esi_buf))); + + if (zif->es_info.df_pref) + vty_out(vty, " evpn mh es-df-pref %u\n", zif->es_info.df_pref); + + if (zif->flags & ZIF_FLAG_EVPN_MH_UPLINK) + vty_out(vty, " evpn mh uplink\n"); + + return 0; +} + +#ifndef VTYSH_EXTRACT_PL +#include "zebra/zebra_evpn_mh_clippy.c" +#endif +/* CLI for setting an ES in bypass mode */ +DEFPY_HIDDEN(zebra_evpn_es_bypass, zebra_evpn_es_bypass_cmd, + "[no] evpn mh bypass", + NO_STR "EVPN\n" EVPN_MH_VTY_STR "set bypass mode\n") +{ + VTY_DECLVAR_CONTEXT(interface, ifp); + struct zebra_if *zif; + + zif = ifp->info; + + if (no) { + zebra_evpn_es_bypass_cfg_update(zif, false); + } else { + if (!zebra_evpn_is_if_es_capable(zif)) { + vty_out(vty, + "%% DF bypass cannot be associated with this interface type\n"); + return CMD_WARNING; + } + zebra_evpn_es_bypass_cfg_update(zif, true); + } + return CMD_SUCCESS; +} + +/* CLI for configuring DF preference part for an ES */ +DEFPY(zebra_evpn_es_pref, zebra_evpn_es_pref_cmd, + "[no$no] evpn mh es-df-pref [(1-65535)$df_pref]", + NO_STR "EVPN\n" EVPN_MH_VTY_STR + "preference value used for DF election\n" + "pref\n") +{ + VTY_DECLVAR_CONTEXT(interface, ifp); + struct zebra_if *zif; + + zif = ifp->info; + + if (no) { + zebra_evpn_es_df_pref_update(zif, 0); + } else { + if (!zebra_evpn_is_if_es_capable(zif)) { + vty_out(vty, + "%% DF preference cannot be associated with this interface type\n"); + return CMD_WARNING; + } + zebra_evpn_es_df_pref_update(zif, df_pref); + } + return CMD_SUCCESS; +} + +/* CLI for setting up sysmac part of ESI on an access port */ +DEFPY(zebra_evpn_es_sys_mac, + zebra_evpn_es_sys_mac_cmd, + "[no$no] evpn mh es-sys-mac [X:X:X:X:X:X$mac]", + NO_STR + "EVPN\n" + EVPN_MH_VTY_STR + "Ethernet segment system MAC\n" + MAC_STR +) +{ + VTY_DECLVAR_CONTEXT(interface, ifp); + struct zebra_if *zif; + int ret = 0; + + zif = ifp->info; + + if (no) { + static struct ethaddr zero_mac; + + ret = zebra_evpn_es_sys_mac_update(zif, &zero_mac); + if (ret == -1) { + vty_out(vty, "%% Failed to clear ES sysmac\n"); + return CMD_WARNING; + } + } else { + + if (!zebra_evpn_is_if_es_capable(zif)) { + vty_out(vty, + "%% ESI cannot be associated with this interface type\n"); + return CMD_WARNING; + } + + if (!mac || is_zero_mac(&mac->eth_addr)) { + vty_out(vty, "%% ES sysmac value is invalid\n"); + return CMD_WARNING; + } + + ret = zebra_evpn_es_sys_mac_update(zif, &mac->eth_addr); + if (ret == -1) { + vty_out(vty, + "%% ESI already exists on a different interface\n"); + return CMD_WARNING; + } + } + return CMD_SUCCESS; +} + +/* CLI for setting up local-ID part of ESI on an access port */ +DEFPY(zebra_evpn_es_id, + zebra_evpn_es_id_cmd, + "[no$no] evpn mh es-id [(1-16777215)$es_lid | NAME$esi_str]", + NO_STR + "EVPN\n" + EVPN_MH_VTY_STR + "Ethernet segment identifier\n" + "local discriminator\n" + "10-byte ID - 00:AA:BB:CC:DD:EE:FF:GG:HH:II\n" +) +{ + VTY_DECLVAR_CONTEXT(interface, ifp); + struct zebra_if *zif; + int ret = 0; + esi_t esi; + + zif = ifp->info; + + if (no) { + if (zif->es_info.lid) + ret = zebra_evpn_es_lid_update(zif, 0); + else if (memcmp(&zif->es_info.esi, zero_esi, sizeof(*zero_esi))) + ret = zebra_evpn_es_type0_esi_update(zif, zero_esi); + + if (ret == -1) { + vty_out(vty, + "%% Failed to clear ES local id or ESI name\n"); + return CMD_WARNING; + } + } else { + if (!zebra_evpn_is_if_es_capable(zif)) { + vty_out(vty, + "%% ESI cannot be associated with this interface type\n"); + return CMD_WARNING; + } + + if (esi_str) { + if (!str_to_esi(esi_str, &esi)) { + vty_out(vty, "%% Malformed ESI name\n"); + return CMD_WARNING; + } + ret = zebra_evpn_es_type0_esi_update(zif, &esi); + } else { + if (!es_lid) { + vty_out(vty, + "%% Specify ES local id or ESI name\n"); + return CMD_WARNING; + } + ret = zebra_evpn_es_lid_update(zif, es_lid); + } + + if (ret == -1) { + vty_out(vty, + "%% ESI already exists on a different interface\n"); + return CMD_WARNING; + } + } + return CMD_SUCCESS; +} + +/* CLI for tagging an interface as an uplink */ +DEFPY(zebra_evpn_mh_uplink, zebra_evpn_mh_uplink_cmd, "[no] evpn mh uplink", + NO_STR "EVPN\n" EVPN_MH_VTY_STR "uplink to the VxLAN core\n") +{ + VTY_DECLVAR_CONTEXT(interface, ifp); + struct zebra_if *zif; + + zif = ifp->info; + zebra_evpn_mh_uplink_cfg_update(zif, no ? false : true); + + return CMD_SUCCESS; +} + +void zebra_evpn_mh_json(json_object *json) +{ + json_object *json_array; + char thread_buf[THREAD_TIMER_STRLEN]; + + json_object_int_add(json, "macHoldtime", zmh_info->mac_hold_time); + json_object_int_add(json, "neighHoldtime", zmh_info->neigh_hold_time); + json_object_int_add(json, "startupDelay", zmh_info->startup_delay_time); + json_object_string_add( + json, "startupDelayTimer", + thread_timer_to_hhmmss(thread_buf, sizeof(thread_buf), + zmh_info->startup_delay_timer)); + json_object_int_add(json, "uplinkConfigCount", + zmh_info->uplink_cfg_cnt); + json_object_int_add(json, "uplinkActiveCount", + zmh_info->uplink_oper_up_cnt); + + if (zmh_info->protodown_rc) { + json_array = json_object_new_array(); + if (CHECK_FLAG(zmh_info->protodown_rc, + ZEBRA_PROTODOWN_EVPN_STARTUP_DELAY)) + json_object_array_add( + json_array, + json_object_new_string("startupDelay")); + if (CHECK_FLAG(zmh_info->protodown_rc, + ZEBRA_PROTODOWN_EVPN_UPLINK_DOWN)) + json_object_array_add( + json_array, + json_object_new_string("uplinkDown")); + json_object_object_add(json, "protodownReasons", json_array); + } +} + +void zebra_evpn_mh_print(struct vty *vty) +{ + char pd_buf[ZEBRA_PROTODOWN_RC_STR_LEN]; + char thread_buf[THREAD_TIMER_STRLEN]; + + vty_out(vty, "EVPN MH:\n"); + vty_out(vty, " mac-holdtime: %ds, neigh-holdtime: %ds\n", + zmh_info->mac_hold_time, zmh_info->neigh_hold_time); + vty_out(vty, " startup-delay: %ds, start-delay-timer: %s\n", + zmh_info->startup_delay_time, + thread_timer_to_hhmmss(thread_buf, sizeof(thread_buf), + zmh_info->startup_delay_timer)); + vty_out(vty, " uplink-cfg-cnt: %u, uplink-active-cnt: %u\n", + zmh_info->uplink_cfg_cnt, zmh_info->uplink_oper_up_cnt); + if (zmh_info->protodown_rc) + vty_out(vty, " protodown reasons: %s\n", + zebra_protodown_rc_str(zmh_info->protodown_rc, pd_buf, + sizeof(pd_buf))); +} + +/*****************************************************************************/ +/* A base L2-VNI is maintained to derive parameters such as ES originator-IP. + * XXX: once single vxlan device model becomes available this will not be + * necessary + */ +/* called when a new vni is added or becomes oper up or becomes a bridge port */ +void zebra_evpn_es_set_base_evpn(struct zebra_evpn *zevpn) +{ + struct listnode *node; + struct zebra_evpn_es *es; + + if (zmh_info->es_base_evpn) { + if (zmh_info->es_base_evpn != zevpn) { + /* unrelated EVPN; ignore it */ + return; + } + /* check if the local vtep-ip has changed */ + } else { + /* check if the EVPN can be used as base EVPN */ + if (!zebra_evpn_send_to_client_ok(zevpn)) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es base vni set to %d", + zevpn->vni); + zmh_info->es_base_evpn = zevpn; + } + + /* update local VTEP-IP */ + if (zmh_info->es_originator_ip.s_addr == + zmh_info->es_base_evpn->local_vtep_ip.s_addr) + return; + + zmh_info->es_originator_ip.s_addr = + zmh_info->es_base_evpn->local_vtep_ip.s_addr; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es originator ip set to %pI4", + &zmh_info->es_base_evpn->local_vtep_ip); + + /* if originator ip changes we need to update bgp */ + for (ALL_LIST_ELEMENTS_RO(zmh_info->local_es_list, node, es)) { + zebra_evpn_es_run_df_election(es, __func__); + + if (es->flags & ZEBRA_EVPNES_READY_FOR_BGP) + zebra_evpn_es_send_add_to_client(es); + else + zebra_evpn_es_re_eval_send_to_client(es, + true /* es_evi_re_reval */); + } +} + +/* called when a vni is removed or becomes oper down or is removed from a + * bridge + */ +void zebra_evpn_es_clear_base_evpn(struct zebra_evpn *zevpn) +{ + struct listnode *node; + struct zebra_evpn_es *es; + + if (zmh_info->es_base_evpn != zevpn) + return; + + zmh_info->es_base_evpn = NULL; + /* lost current base EVPN; try to find a new one */ + zebra_evpn_es_get_one_base_evpn(); + + /* couldn't locate an eligible base evpn */ + if (!zmh_info->es_base_evpn && zmh_info->es_originator_ip.s_addr) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("es originator ip cleared"); + + zmh_info->es_originator_ip.s_addr = 0; + /* lost originator ip */ + for (ALL_LIST_ELEMENTS_RO(zmh_info->local_es_list, node, es)) { + zebra_evpn_es_re_eval_send_to_client(es, + true /* es_evi_re_reval */); + } + } +} + +/* Locate an "eligible" L2-VNI to follow */ +static int zebra_evpn_es_get_one_base_evpn_cb(struct hash_bucket *b, void *data) +{ + struct zebra_evpn *zevpn = b->data; + + zebra_evpn_es_set_base_evpn(zevpn); + + if (zmh_info->es_base_evpn) + return HASHWALK_ABORT; + + return HASHWALK_CONTINUE; +} + +/* locate a base_evpn to follow for the purposes of common params like + * originator IP + */ +static void zebra_evpn_es_get_one_base_evpn(void) +{ + struct zebra_vrf *zvrf; + + zvrf = zebra_vrf_get_evpn(); + hash_walk(zvrf->evpn_table, zebra_evpn_es_get_one_base_evpn_cb, NULL); +} + +/***************************************************************************** + * local ethernet segments can be error-disabled if the switch is not + * ready to start transmitting traffic via the VxLAN overlay + */ +bool zebra_evpn_is_es_bond(struct interface *ifp) +{ + struct zebra_if *zif = ifp->info; + + return !!(struct zebra_if *)zif->es_info.es; +} + +bool zebra_evpn_is_es_bond_member(struct interface *ifp) +{ + struct zebra_if *zif = ifp->info; + + return IS_ZEBRA_IF_BOND_SLAVE(zif->ifp) && zif->bondslave_info.bond_if + && ((struct zebra_if *)zif->bondslave_info.bond_if->info) + ->es_info.es; +} + +void zebra_evpn_mh_update_protodown_bond_mbr(struct zebra_if *zif, bool clear, + const char *caller) +{ + bool new_protodown; + uint32_t old_protodown_rc = 0; + uint32_t new_protodown_rc = 0; + uint32_t protodown_rc = 0; + + if (!clear) { + struct zebra_if *bond_zif; + + bond_zif = zif->bondslave_info.bond_if->info; + protodown_rc = bond_zif->protodown_rc; + } + + old_protodown_rc = zif->protodown_rc; + new_protodown_rc = (old_protodown_rc & ~ZEBRA_PROTODOWN_EVPN_ALL); + new_protodown_rc |= (protodown_rc & ZEBRA_PROTODOWN_EVPN_ALL); + new_protodown = !!new_protodown_rc; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES && (new_protodown_rc != old_protodown_rc)) + zlog_debug( + "%s bond mbr %s protodown_rc changed; old 0x%x new 0x%x", + caller, zif->ifp->name, old_protodown_rc, + new_protodown_rc); + + if (zebra_if_update_protodown_rc(zif->ifp, new_protodown, + new_protodown_rc) == 0) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("%s protodown %s", zif->ifp->name, + new_protodown ? "on" : "off"); + } +} + +/* The bond members inherit the protodown reason code from the bond */ +static void zebra_evpn_mh_update_protodown_bond(struct zebra_if *bond_zif) +{ + struct zebra_if *zif; + struct listnode *node; + + if (!bond_zif->bond_info.mbr_zifs) + return; + + for (ALL_LIST_ELEMENTS_RO(bond_zif->bond_info.mbr_zifs, node, zif)) { + zebra_evpn_mh_update_protodown_bond_mbr(zif, false /*clear*/, + __func__); + } +} + +/* The global EVPN MH protodown rc is applied to all local ESs */ +static void zebra_evpn_mh_update_protodown_es(struct zebra_evpn_es *es, + bool resync_dplane) +{ + struct zebra_if *zif; + uint32_t old_protodown_rc; + + zif = es->zif; + /* if the reason code is the same bail unless it is a new + * ES bond in that case we would need to ensure that the + * dplane is really in sync with zebra + */ + if (!resync_dplane + && (zif->protodown_rc & ZEBRA_PROTODOWN_EVPN_ALL) + == (zmh_info->protodown_rc & ZEBRA_PROTODOWN_EVPN_ALL)) + return; + + old_protodown_rc = zif->protodown_rc; + zif->protodown_rc &= ~ZEBRA_PROTODOWN_EVPN_ALL; + zif->protodown_rc |= + (zmh_info->protodown_rc & ZEBRA_PROTODOWN_EVPN_ALL); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES + && (old_protodown_rc != zif->protodown_rc)) + zlog_debug( + "es %s ifp %s protodown_rc changed; old 0x%x new 0x%x", + es->esi_str, zif->ifp->name, old_protodown_rc, + zif->protodown_rc); + + /* update dataplane with the new protodown setting */ + zebra_evpn_mh_update_protodown_bond(zif); +} + +static void zebra_evpn_mh_clear_protodown_es(struct zebra_evpn_es *es) +{ + struct zebra_if *zif; + uint32_t old_protodown_rc; + + zif = es->zif; + if (!(zif->protodown_rc & ZEBRA_PROTODOWN_EVPN_ALL)) + return; + + old_protodown_rc = zif->protodown_rc; + zif->protodown_rc &= ~ZEBRA_PROTODOWN_EVPN_ALL; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug( + "clear: es %s ifp %s protodown_rc cleared; old 0x%x new 0x%x", + es->esi_str, zif->ifp->name, old_protodown_rc, + zif->protodown_rc); + + /* update dataplane with the new protodown setting */ + zebra_evpn_mh_update_protodown_bond(zif); +} + +static void zebra_evpn_mh_update_protodown_es_all(void) +{ + struct listnode *node; + struct zebra_evpn_es *es; + + for (ALL_LIST_ELEMENTS_RO(zmh_info->local_es_list, node, es)) + zebra_evpn_mh_update_protodown_es(es, false /*resync_dplane*/); +} + +static void zebra_evpn_mh_update_protodown(uint32_t protodown_rc, bool set) +{ + uint32_t old_protodown_rc = zmh_info->protodown_rc; + + if (set) { + if ((protodown_rc & zmh_info->protodown_rc) == protodown_rc) + return; + + zmh_info->protodown_rc |= protodown_rc; + } else { + if (!(protodown_rc & zmh_info->protodown_rc)) + return; + zmh_info->protodown_rc &= ~protodown_rc; + } + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("mh protodown_rc changed; old 0x%x new 0x%x", + old_protodown_rc, zmh_info->protodown_rc); + zebra_evpn_mh_update_protodown_es_all(); +} + +static inline bool zebra_evpn_mh_is_all_uplinks_down(void) +{ + return zmh_info->uplink_cfg_cnt && !zmh_info->uplink_oper_up_cnt; +} + +static void zebra_evpn_mh_uplink_oper_flags_update(struct zebra_if *zif, + bool set) +{ + if (set && if_is_operative(zif->ifp)) { + if (!(zif->flags & ZIF_FLAG_EVPN_MH_UPLINK_OPER_UP)) { + zif->flags |= ZIF_FLAG_EVPN_MH_UPLINK_OPER_UP; + ++zmh_info->uplink_oper_up_cnt; + } + } else { + if (zif->flags & ZIF_FLAG_EVPN_MH_UPLINK_OPER_UP) { + zif->flags &= ~ZIF_FLAG_EVPN_MH_UPLINK_OPER_UP; + if (zmh_info->uplink_oper_up_cnt) + --zmh_info->uplink_oper_up_cnt; + } + } +} + +static void zebra_evpn_mh_uplink_cfg_update(struct zebra_if *zif, bool set) +{ + bool old_protodown = zebra_evpn_mh_is_all_uplinks_down(); + bool new_protodown; + + if (set) { + if (zif->flags & ZIF_FLAG_EVPN_MH_UPLINK) + return; + + zif->flags |= ZIF_FLAG_EVPN_MH_UPLINK; + ++zmh_info->uplink_cfg_cnt; + } else { + if (!(zif->flags & ZIF_FLAG_EVPN_MH_UPLINK)) + return; + + zif->flags &= ~ZIF_FLAG_EVPN_MH_UPLINK; + if (zmh_info->uplink_cfg_cnt) + --zmh_info->uplink_cfg_cnt; + } + + zebra_evpn_mh_uplink_oper_flags_update(zif, set); + new_protodown = zebra_evpn_mh_is_all_uplinks_down(); + if (old_protodown == new_protodown) + return; + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug( + "mh-uplink-cfg-chg on if %s/%d %s uplinks cfg %u up %u", + zif->ifp->name, zif->ifp->ifindex, set ? "set" : "down", + zmh_info->uplink_cfg_cnt, zmh_info->uplink_oper_up_cnt); + + zebra_evpn_mh_update_protodown(ZEBRA_PROTODOWN_EVPN_UPLINK_DOWN, + new_protodown); +} + +void zebra_evpn_mh_uplink_oper_update(struct zebra_if *zif) +{ + bool old_protodown = zebra_evpn_mh_is_all_uplinks_down(); + bool new_protodown; + + zebra_evpn_mh_uplink_oper_flags_update(zif, true /*set*/); + + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug( + "mh-uplink-oper-chg on if %s/%d %s; uplinks cfg %u up %u", + zif->ifp->name, zif->ifp->ifindex, + if_is_operative(zif->ifp) ? "up" : "down", + zmh_info->uplink_cfg_cnt, zmh_info->uplink_oper_up_cnt); + + new_protodown = zebra_evpn_mh_is_all_uplinks_down(); + if (old_protodown == new_protodown) + return; + + /* if protodown_rc XXX_UPLINK_DOWN is about to be cleared + * fire up the start-up delay timer to allow the EVPN network + * to converge (Type-2 routes need to be advertised and processed) + */ + if (!new_protodown && (zmh_info->uplink_oper_up_cnt == 1)) + zebra_evpn_mh_startup_delay_timer_start("uplink-up"); + + zebra_evpn_mh_update_protodown(ZEBRA_PROTODOWN_EVPN_UPLINK_DOWN, + new_protodown); +} + +static void zebra_evpn_mh_startup_delay_exp_cb(struct thread *t) +{ + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("startup-delay expired"); + + zebra_evpn_mh_update_protodown(ZEBRA_PROTODOWN_EVPN_STARTUP_DELAY, + false /* set */); +} + +static void zebra_evpn_mh_startup_delay_timer_start(const char *rc) +{ + if (zmh_info->startup_delay_timer) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("startup-delay timer cancelled"); + THREAD_OFF(zmh_info->startup_delay_timer); + } + + if (zmh_info->startup_delay_time) { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug( + "startup-delay timer started for %d sec on %s", + zmh_info->startup_delay_time, rc); + thread_add_timer(zrouter.master, + zebra_evpn_mh_startup_delay_exp_cb, NULL, + zmh_info->startup_delay_time, + &zmh_info->startup_delay_timer); + zebra_evpn_mh_update_protodown( + ZEBRA_PROTODOWN_EVPN_STARTUP_DELAY, true /* set */); + } else { + zebra_evpn_mh_update_protodown( + ZEBRA_PROTODOWN_EVPN_STARTUP_DELAY, false /* set */); + } +} + +/***************************************************************************** + * Nexthop management: nexthops associated with Type-2 routes that have + * an ES as destination are consolidated by BGP into a per-VRF nh->rmac + * mapping which is the installed as a remote neigh/fdb entry with a + * dummy (type-1) prefix referencing it. + * This handling is needed because Type-2 routes with ES as dest use NHG + * that are setup using EAD routes (i.e. such NHGs do not include the + * RMAC info). + ****************************************************************************/ +void zebra_evpn_proc_remote_nh(ZAPI_HANDLER_ARGS) +{ + struct stream *s; + vrf_id_t vrf_id; + struct ipaddr nh; + struct ethaddr rmac; + struct prefix_evpn dummy_prefix; + size_t min_len = 4 + sizeof(nh); + + s = msg; + + /* + * Ensure that the stream sent to us is long enough + */ + if (hdr->command == ZEBRA_EVPN_REMOTE_NH_ADD) + min_len += sizeof(rmac); + if (hdr->length < min_len) + return; + + vrf_id = stream_getl(s); + stream_get(&nh, s, sizeof(nh)); + + memset(&dummy_prefix, 0, sizeof(dummy_prefix)); + dummy_prefix.family = AF_EVPN; + dummy_prefix.prefixlen = (sizeof(struct evpn_addr) * 8); + dummy_prefix.prefix.route_type = 1; /* XXX - fixup to type-1 def */ + dummy_prefix.prefix.ead_addr.ip.ipa_type = nh.ipa_type; + + if (hdr->command == ZEBRA_EVPN_REMOTE_NH_ADD) { + stream_get(&rmac, s, sizeof(rmac)); + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug( + "evpn remote nh %d %pIA rmac %pEA add pfx %pFX", + vrf_id, &nh, &rmac, &dummy_prefix); + zebra_rib_queue_evpn_route_add(vrf_id, &rmac, &nh, + (struct prefix *)&dummy_prefix); + } else { + if (IS_ZEBRA_DEBUG_EVPN_MH_ES) + zlog_debug("evpn remote nh %d %pIA del pfx %pFX", + vrf_id, &nh, &dummy_prefix); + zebra_rib_queue_evpn_route_del(vrf_id, &nh, + (struct prefix *)&dummy_prefix); + } +} + +/*****************************************************************************/ +void zebra_evpn_mh_config_write(struct vty *vty) +{ + if (zmh_info->mac_hold_time != ZEBRA_EVPN_MH_MAC_HOLD_TIME_DEF) + vty_out(vty, "evpn mh mac-holdtime %d\n", + zmh_info->mac_hold_time); + + if (zmh_info->neigh_hold_time != ZEBRA_EVPN_MH_NEIGH_HOLD_TIME_DEF) + vty_out(vty, "evpn mh neigh-holdtime %d\n", + zmh_info->neigh_hold_time); + + if (zmh_info->startup_delay_time != ZEBRA_EVPN_MH_STARTUP_DELAY_DEF) + vty_out(vty, "evpn mh startup-delay %d\n", + zmh_info->startup_delay_time); + + if (zmh_info->flags & ZEBRA_EVPN_MH_REDIRECT_OFF) + vty_out(vty, "evpn mh redirect-off\n"); +} + +int zebra_evpn_mh_neigh_holdtime_update(struct vty *vty, + uint32_t duration, bool set_default) +{ + if (set_default) + duration = ZEBRA_EVPN_MH_NEIGH_HOLD_TIME_DEF; + + zmh_info->neigh_hold_time = duration; + + return 0; +} + +int zebra_evpn_mh_mac_holdtime_update(struct vty *vty, + uint32_t duration, bool set_default) +{ + if (set_default) + duration = ZEBRA_EVPN_MH_MAC_HOLD_TIME_DEF; + + zmh_info->mac_hold_time = duration; + + return 0; +} + +int zebra_evpn_mh_startup_delay_update(struct vty *vty, uint32_t duration, + bool set_default) +{ + if (set_default) + duration = ZEBRA_EVPN_MH_STARTUP_DELAY_DEF; + + zmh_info->startup_delay_time = duration; + + /* if startup_delay_timer is running allow it to be adjusted + * up or down + */ + if (zmh_info->startup_delay_timer) + zebra_evpn_mh_startup_delay_timer_start("config"); + + return 0; +} + +int zebra_evpn_mh_redirect_off(struct vty *vty, bool redirect_off) +{ + /* This knob needs to be set before ESs are configured + * i.e. cannot be changed on the fly + */ + if (redirect_off) + zmh_info->flags |= ZEBRA_EVPN_MH_REDIRECT_OFF; + else + zmh_info->flags &= ~ZEBRA_EVPN_MH_REDIRECT_OFF; + + return 0; +} + +void zebra_evpn_interface_init(void) +{ + install_element(INTERFACE_NODE, &zebra_evpn_es_id_cmd); + install_element(INTERFACE_NODE, &zebra_evpn_es_sys_mac_cmd); + install_element(INTERFACE_NODE, &zebra_evpn_es_pref_cmd); + install_element(INTERFACE_NODE, &zebra_evpn_es_bypass_cmd); + install_element(INTERFACE_NODE, &zebra_evpn_mh_uplink_cmd); +} + +void zebra_evpn_mh_init(void) +{ + zrouter.mh_info = XCALLOC(MTYPE_ZMH_INFO, sizeof(*zrouter.mh_info)); + + zmh_info->mac_hold_time = ZEBRA_EVPN_MH_MAC_HOLD_TIME_DEF; + zmh_info->neigh_hold_time = ZEBRA_EVPN_MH_NEIGH_HOLD_TIME_DEF; + /* setup ES tables */ + RB_INIT(zebra_es_rb_head, &zmh_info->es_rb_tree); + zmh_info->local_es_list = list_new(); + listset_app_node_mem(zmh_info->local_es_list); + + bf_init(zmh_info->nh_id_bitmap, EVPN_NH_ID_MAX); + bf_assign_zero_index(zmh_info->nh_id_bitmap); + zmh_info->nhg_table = hash_create(zebra_evpn_nhg_hash_keymake, + zebra_evpn_nhg_cmp, "l2 NHG table"); + zmh_info->nh_ip_table = + hash_create(zebra_evpn_nh_ip_hash_keymake, zebra_evpn_nh_ip_cmp, + "l2 NH IP table"); + + /* setup broadcast domain tables */ + zmh_info->evpn_vlan_table = hash_create(zebra_evpn_acc_vl_hash_keymake, + zebra_evpn_acc_vl_cmp, "access VLAN hash table"); + + zmh_info->startup_delay_time = ZEBRA_EVPN_MH_STARTUP_DELAY_DEF; + zebra_evpn_mh_startup_delay_timer_start("init"); +} + +void zebra_evpn_mh_terminate(void) +{ + list_delete(&zmh_info->local_es_list); + + hash_iterate(zmh_info->evpn_vlan_table, + zebra_evpn_acc_vl_cleanup_all, NULL); + hash_free(zmh_info->evpn_vlan_table); + hash_free(zmh_info->nhg_table); + hash_free(zmh_info->nh_ip_table); + bf_free(zmh_info->nh_id_bitmap); + + XFREE(MTYPE_ZMH_INFO, zrouter.mh_info); +} |