> The northd creates flows that restrict processing ARP requests:
> If ARP request comes from lswitch that has either localnet or
> l2gateway ports and router's attachment port is L3 dgw port such
> ARP will be allowed only on resident chassis.
> 
> However, same lswitch could also have 'normal' aka virtual lports.
> For these lports restrictions described above should not be applied.
> 
> The fix moves is_chassis_resident check from lrouter to lswitch level.
> The residence check is only applied for ARP requests that were
> originated from localnet/l2gateway ports. If such check is succeeded
> ARP request is allowed otherwise ARP request packet is dropped.
> For ARP requests coming from virtual ports no residence restrictions
> are applied.
> 
> Co-authored-by: Alexandra Rukomoinikova <arukomoinikova@k2.cloud>
> Signed-off-by: Aleksandr Smirnov <alekssmirnov@k2.cloud>
> Signed-off-by: Alexandra Rukomoinikova <arukomoinikova@k2.cloud>

Hi Aleksandr and Alexandra,

thx for the patches. Some comments and questions below.

Regards,
Lorenzo

> 
> ---
> v2: Minor fixes, add co-authored-by
> v3: Apply Dumitru comments, code rework, move part of processing
>     into new I-P node.
> ---
>  lib/stopwatch-names.h    |   1 +
>  northd/automake.mk       |   2 +
>  northd/en-lflow.c        |  29 ++++
>  northd/en-lflow.h        |   2 +
>  northd/en-ls-arp.c       | 284 +++++++++++++++++++++++++++++++++++++++
>  northd/en-ls-arp.h       | 104 ++++++++++++++
>  northd/inc-proc-northd.c |   8 ++
>  northd/northd.c          | 179 ++++++++++++++++++++++--
>  northd/northd.h          |  10 ++
>  northd/ovn-northd.c      |   1 +
>  tests/ovn-northd.at      |  10 +-
>  tests/ovn.at             | 157 ++++++++++++++++++++++
>  12 files changed, 771 insertions(+), 16 deletions(-)
>  create mode 100644 northd/en-ls-arp.c
>  create mode 100644 northd/en-ls-arp.h
> 
> diff --git a/lib/stopwatch-names.h b/lib/stopwatch-names.h
> index f3a931c40..b912e813c 100644
> --- a/lib/stopwatch-names.h
> +++ b/lib/stopwatch-names.h
> @@ -34,6 +34,7 @@
>  #define LR_NAT_RUN_STOPWATCH_NAME "lr_nat_run"
>  #define LR_STATEFUL_RUN_STOPWATCH_NAME "lr_stateful"
>  #define LS_STATEFUL_RUN_STOPWATCH_NAME "ls_stateful"
> +#define LS_ARP_RUN_STOPWATCH_NAME "ls_arp"
>  #define ADVERTISED_ROUTE_SYNC_RUN_STOPWATCH_NAME "advertised_route_sync"
>  #define LEARNED_ROUTE_SYNC_RUN_STOPWATCH_NAME "learned_route_sync"
>  #define DYNAMIC_ROUTES_RUN_STOPWATCH_NAME "dynamic_routes"
> diff --git a/northd/automake.mk b/northd/automake.mk
> index 45ca0337f..8cd4fb3a1 100644
> --- a/northd/automake.mk
> +++ b/northd/automake.mk
> @@ -44,6 +44,8 @@ northd_ovn_northd_SOURCES = \
>       northd/en-lr-stateful.h \
>       northd/en-ls-stateful.c \
>       northd/en-ls-stateful.h \
> +     northd/en-ls-arp.c \
> +     northd/en-ls-arp.h \
>       northd/en-sampling-app.c \
>       northd/en-sampling-app.h \
>       northd/en-acl-ids.c \
> diff --git a/northd/en-lflow.c b/northd/en-lflow.c
> index f903f5e3a..772daf623 100644
> --- a/northd/en-lflow.c
> +++ b/northd/en-lflow.c
> @@ -23,6 +23,7 @@
>  #include "en-lr-nat.h"
>  #include "en-lr-stateful.h"
>  #include "en-ls-stateful.h"
> +#include "en-ls-arp.h"
>  #include "en-multicast.h"
>  #include "en-northd.h"
>  #include "en-meters.h"
> @@ -60,6 +61,8 @@ lflow_get_input_data(struct engine_node *node,
>          engine_get_input_data("lr_stateful", node);
>      struct ed_type_ls_stateful *ls_stateful_data =
>          engine_get_input_data("ls_stateful", node);
> +    struct ed_type_ls_arp *ls_arp_data =
> +        engine_get_input_data("ls_arp", node);
>      struct multicast_igmp_data *multicat_igmp_data =
>          engine_get_input_data("multicast_igmp", node);
>      struct ic_learned_svc_monitors_data *ic_learned_svc_monitors_data =
> @@ -84,6 +87,7 @@ lflow_get_input_data(struct engine_node *node,
>      lflow_input->ls_port_groups = &pg_data->ls_port_groups;
>      lflow_input->lr_stateful_table = &lr_stateful_data->table;
>      lflow_input->ls_stateful_table = &ls_stateful_data->table;
> +    lflow_input->ls_arp_table = &ls_arp_data->table;
>      lflow_input->meter_groups = &sync_meters_data->meter_groups;
>      lflow_input->lb_datapaths_map = &northd_data->lb_datapaths_map;
>      lflow_input->local_svc_monitors_map =
> @@ -215,6 +219,31 @@ lflow_ls_stateful_handler(struct engine_node *node, void 
> *data)
>      return EN_HANDLED_UPDATED;
>  }
>  
> +enum engine_input_handler_result
> +lflow_ls_arp_handler(struct engine_node *node, void *data)
> +{
> +    struct ed_type_ls_arp *ls_arp_data =
> +        engine_get_input_data("ls_arp", node);
> +
> +    if (!ls_arp_has_tracked_data(&ls_arp_data->trk_data)) {
> +        return EN_UNHANDLED;
> +    }
> +
> +    const struct engine_context *eng_ctx = engine_get_context();
> +    struct lflow_data *lflow_data = data;
> +    struct lflow_input lflow_input;
> +
> +    lflow_get_input_data(node, &lflow_input);
> +    if (!lflow_handle_ls_arp_changes(eng_ctx->ovnsb_idl_txn,
> +                                     &ls_arp_data->trk_data,
> +                                     &lflow_input,
> +                                     lflow_data->lflow_table)) {
> +        return EN_UNHANDLED;
> +    }
> +
> +    return EN_HANDLED_UPDATED;
> +}
> +
>  enum engine_input_handler_result
>  lflow_multicast_igmp_handler(struct engine_node *node, void *data)
>  {
> diff --git a/northd/en-lflow.h b/northd/en-lflow.h
> index 99bcfda15..fab778d03 100644
> --- a/northd/en-lflow.h
> +++ b/northd/en-lflow.h
> @@ -25,6 +25,8 @@ lflow_lr_stateful_handler(struct engine_node *, void *data);
>  enum engine_input_handler_result
>  lflow_ls_stateful_handler(struct engine_node *node, void *data);
>  enum engine_input_handler_result
> +lflow_ls_arp_handler(struct engine_node *node, void *data);
> +enum engine_input_handler_result
>  lflow_multicast_igmp_handler(struct engine_node *node, void *data);
>  enum engine_input_handler_result
>  lflow_group_ecmp_route_change_handler(struct engine_node *node, void *data);
> diff --git a/northd/en-ls-arp.c b/northd/en-ls-arp.c
> new file mode 100644
> index 000000000..26f2f815e
> --- /dev/null
> +++ b/northd/en-ls-arp.c
> @@ -0,0 +1,284 @@
> +/*
> +  * Copyright (c) 2024, Red Hat, Inc.
> + *
> + * Licensed under the Apache License, Version 2.0 (the "License");
> + * you may not use this file except in compliance with the License.
> + * You may obtain a copy of the License at:
> + *
> + *     http://www.apache.org/licenses/LICENSE-2.0
> + *
> + * Unless required by applicable law or agreed to in writing, software
> + * distributed under the License is distributed on an "AS IS" BASIS,
> + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
> + * See the License for the specific language governing permissions and
> + * limitations under the License.
> + */
> +
> +#include <config.h>
> +
> +/* OVS includes */
> +#include "include/openvswitch/hmap.h"
> +#include "openvswitch/util.h"
> +#include "openvswitch/vlog.h"
> +#include "stopwatch.h"
> +
> +/* OVN includes */
> +#include "en-lr-nat.h"
> +#include "en-ls-arp.h"
> +#include "lib/inc-proc-eng.h"
> +#include "lib/ovn-nb-idl.h"
> +#include "lib/ovn-sb-idl.h"
> +#include "lib/ovn-util.h"
> +#include "lib/stopwatch-names.h"
> +#include "lflow-mgr.h"
> +#include "northd.h"
> +
> +VLOG_DEFINE_THIS_MODULE(en_ls_arp);
> +
> +/* static functions. */
> +
> +struct ls_arp_input {
> +    const struct ovn_datapaths *ls_datapaths;
> +    const struct hmap *lr_nats;
> +};
> +
> +static struct ls_arp_input
> +ls_arp_get_input_data(struct engine_node *node)
> +{
> +    const struct northd_data *northd_data =
> +        engine_get_input_data("northd", node);
> +    struct ed_type_lr_nat_data *lr_nat_data =
> +        engine_get_input_data("lr_nat", node);
> +
> +    return (struct ls_arp_input) {
> +        .ls_datapaths = &northd_data->ls_datapaths,
> +        .lr_nats = &lr_nat_data->lr_nats.entries,
> +    };
> +}
> +
> +static void
> +ls_arp_table_clear(struct ls_arp_table *table)
> +{
> +    struct ls_arp_record *ar;
> +    HMAP_FOR_EACH_POP (ar, key_node, &table->entries) {
> +        lflow_ref_destroy(ar->lflow_ref);
> +        hmapx_destroy(&ar->nat_records);
> +        free(ar);
> +    }
> +}
> +
> +/*
> + *   Return hmapx (odmap) of datapaths, assumed lswitches,
> + *   that are gateways for given nat.
> + */
> +static void
> +nat_odmap_create(struct lr_nat_record *nr,
> +                 struct hmapx *odmap)
> +{
> +    for (size_t i = 0; i < nr->n_nat_entries; i++) {
> +        struct ovn_nat *ent = &nr->nat_entries[i];
> +
> +        if (ent->is_valid
> +            && ent->l3dgw_port
> +            && ent->l3dgw_port->peer
> +            && ent->l3dgw_port->peer->od
> +            && !ent->is_distributed) {
> +            hmapx_add(odmap, ent->l3dgw_port->peer->od);
> +        }
> +    }
> +}
> +
> +static bool
> +ods_find_by_index(struct hmapx *odmap, size_t index)
> +{
> +    struct hmapx_node *hmapx_node;
> +    HMAPX_FOR_EACH (hmapx_node, odmap) {
> +        struct ovn_datapath *od = hmapx_node->data;
> +
> +        if (od->index == index) {
> +            return true;
> +        }
> +    }
> +
> +    return false;
> +}
> +
> +static struct ls_arp_record*
> +ars_find_by_index(const struct ls_arp_table *table, size_t index)
> +{
> +    struct ls_arp_record *ar;
> +    HMAP_FOR_EACH (ar, key_node, &table->entries) {
> +        if (ar->ls_index == index) {
> +            return ar;
> +        }
> +    }
> +
> +    return NULL;
> +}
> +
> +/*
> + * Fill nat_records in given ls_arp_record with nat records that have
> + * lswitch, owned by arp record, as nat gateway.
> + */
> +static void
> +ls_arp_record_set_nats(struct ls_arp_record *ar,
> +                       const struct hmap *nats)
> +{
> +    hmapx_init(&ar->nat_records);
> +
> +    struct lr_nat_record *nr;
> +    HMAP_FOR_EACH (nr, key_node, nats) {
> +        struct hmapx ods = HMAPX_INITIALIZER(&ods);
> +
> +        nat_odmap_create(nr, &ods);
> +
> +        if (ods_find_by_index(&ods, ar->ls_index)) {
> +            hmapx_add(&ar->nat_records, nr);
> +        }
> +
> +        hmapx_destroy(&ods);
> +    }
> +}
> +
> +static struct ls_arp_record *
> +ls_arp_record_create(struct ls_arp_table *table,
> +                     const struct ovn_datapath *od,
> +                     const struct hmap *lr_nats)
> +{
> +    struct ls_arp_record *ar = xzalloc(sizeof *ar);
> +
> +    ar->ls_index = od->index;
> +    ls_arp_record_set_nats(ar, lr_nats);
> +    ar->lflow_ref = lflow_ref_create();
> +
> +    hmap_insert(&table->entries, &ar->key_node,
> +                uuid_hash(&od->nbs->header_.uuid));
> +
> +    return ar;
> +}
> +
> +/* public functions. */
> +void*
> +en_ls_arp_init(struct engine_node *node OVS_UNUSED,
> +               struct engine_arg *arg OVS_UNUSED)
> +{
> +    struct ed_type_ls_arp *data = xzalloc(sizeof *data);
> +
> +    hmap_init(&data->table.entries);
> +    hmapx_init(&data->trk_data.crupdated);
> +
> +    return data;
> +}
> +
> +void
> +en_ls_arp_cleanup(void *data_)
> +{
> +    struct ed_type_ls_arp *data = data_;
> +
> +    ls_arp_table_clear(&data->table);
> +    hmap_destroy(&data->table.entries);
> +    hmapx_destroy(&data->trk_data.crupdated);
> +}
> +
> +void
> +en_ls_arp_clear_tracked_data(void *data_)
> +{
> +    struct ed_type_ls_arp *data = data_;
> +    hmapx_clear(&data->trk_data.crupdated);
> +}
> +
> +enum engine_node_state
> +en_ls_arp_run(struct engine_node *node, void *data_)
> +{
> +    struct ls_arp_input input_data = ls_arp_get_input_data(node);
> +    struct ed_type_ls_arp *data = data_;
> +
> +    stopwatch_start(LS_ARP_RUN_STOPWATCH_NAME, time_msec());
> +
> +    ls_arp_table_clear(&data->table);
> +
> +    const struct ovn_datapath *od;
> +    HMAP_FOR_EACH (od, key_node, &input_data.ls_datapaths->datapaths) {
> +        ls_arp_record_create(&data->table, od, input_data.lr_nats);
> +    }
> +
> +    stopwatch_stop(LS_ARP_RUN_STOPWATCH_NAME, time_msec());
> +
> +    return EN_UPDATED;
> +}
> +
> +/* Handler functions. */
> +
> +enum engine_input_handler_result
> +ls_arp_northd_handler(struct engine_node *node, void *x OVS_UNUSED)
> +{
> +    struct northd_data *northd_data = engine_get_input_data("northd", node);
> +    if (!northd_has_tracked_data(&northd_data->trk_data)) {
> +        return EN_UNHANDLED;
> +    }

Now we support IP for logical-switch creation/deletion. Should we take care of
it too?

> +
> +    return EN_HANDLED_UNCHANGED;
> +}
> +
> +enum engine_input_handler_result
> +ls_arp_lr_nat_handler(struct engine_node *node, void *data_)
> +{
> +    struct ed_type_lr_nat_data *lr_nat_data =
> +        engine_get_input_data("lr_nat", node);
> +
> +    if (!lr_nat_has_tracked_data(&lr_nat_data->trk_data)) {
> +        return EN_UNHANDLED;
> +    }
> +
> +    struct ed_type_ls_arp *data = data_;
> +
> +    struct hmapx_node *hmapx_node;
> +    HMAPX_FOR_EACH (hmapx_node, &lr_nat_data->trk_data.crupdated) {
> +        struct lr_nat_record *nr_cur = hmapx_node->data;
> +        struct hmapx ods = HMAPX_INITIALIZER(&ods);
> +
> +        /* Collect all ods (lswitch) that are gateways for given nat */
> +        nat_odmap_create(nr_cur, &ods);
> +
> +        struct ls_arp_record *ar;
> +        LS_ARP_TABLE_FOR_EACH (ar, &data->table) {
> +            struct hmapx_node *nr_node = hmapx_find(&ar->nat_records, 
> nr_cur);
> +
> +            /* current nat record is already registered for given arp_record 
> */
> +            if (nr_node) {
> +                /* trigger this arp_record to regenerate od lflow */
> +                hmapx_add(&data->trk_data.crupdated, ar);
> +
> +                /* ... but not part of affected ods anymore,
> +                   i.e. the change of the nat removes this gateway */
> +                if (!ods_find_by_index(&ods, ar->ls_index)) {
> +                    hmapx_delete(&ar->nat_records, nr_node);
> +                }
> +            }
> +        }
> +
> +        /* Process gateways that are new */
> +        struct hmapx_node *hmapx_node2;
> +        HMAPX_FOR_EACH (hmapx_node2, &ods) {
> +            struct ovn_datapath *od = hmapx_node2->data;
> +
> +            /* Determine which arp_record is affected */
> +            ar = ars_find_by_index(&data->table, od->index);
> +            ovs_assert(ar);
> +
> +            /* new gateway triggers lflow regeneration for this arp_records 
> */
> +            hmapx_add(&data->trk_data.crupdated, ar);
> +            hmapx_add(&ar->nat_records, nr_cur);
> +        }
> +
> +        hmapx_destroy(&ods);
> +    }
> +
> +
> +    if (ls_arp_has_tracked_data(&data->trk_data)) {
> +        return EN_HANDLED_UPDATED;
> +    }
> +
> +    return EN_HANDLED_UNCHANGED;
> +}
> +
> diff --git a/northd/en-ls-arp.h b/northd/en-ls-arp.h
> new file mode 100644
> index 000000000..ecc39eb77
> --- /dev/null
> +++ b/northd/en-ls-arp.h
> @@ -0,0 +1,104 @@
> +/*
> + * Copyright (c) 2024, Red Hat, Inc.
> + *
> + * Licensed under the Apache License, Version 2.0 (the "License");
> + * you may not use this file except in compliance with the License.
> + * You may obtain a copy of the License at:
> + *
> + *     http://www.apache.org/licenses/LICENSE-2.0
> + *
> + * Unless required by applicable law or agreed to in writing, software
> + * distributed under the License is distributed on an "AS IS" BASIS,
> + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
> + * See the License for the specific language governing permissions and
> + * limitations under the License.
> + */
> +#ifndef EN_LS_ARP_H
> +#define EN_LS_ARP_H 1
> +
> +/* OVS includes. */
> +#include "lib/hmapx.h"
> +#include "openvswitch/hmap.h"
> +
> +/* OVN includes. */
> +#include "lib/inc-proc-eng.h"
> +#include "lib/ovn-nb-idl.h"
> +#include "lib/ovn-sb-idl.h"
> +#include "lib/ovn-util.h"
> +#include "lib/stopwatch-names.h"
> +
> +struct lflow_ref;
> +
> +struct ls_arp_record {
> +    struct hmap_node key_node;
> +
> +    /* Unique id of the logical switch.  Note : This id is assigned
> +     * by the northd engine node for each logical switch. */
> +    size_t ls_index;
> +
> +    /* 'lflow_ref' is used to reference logical flows generated for
> +     * this ls_arp record.
> +     *
> +     * This data is initialized and destroyed by the en_ls_arp node,
> +     * but populated and used only by the en_lflow node. Ideally this data
> +     * should be maintained as part of en_lflow's data.  However, it would
> +     * be less efficient and more complex:
> +     *
> +     * 1. It would require an extra search (using the index) to find the
> +     * lflows.
> +     *
> +     * 2. Building the index needs to be thread-safe, using either a global
> +     * lock which is obviously less efficient, or hash-based lock array which
> +     * is more complex.
> +     *
> +     * Adding the lflow_ref here is more straightforward. The drawback is 
> that
> +     * we need to keep in mind that this data belongs to en_lflow node, so
> +     * never access it from any other nodes.
> +     *
> +     * Note: lflow_ref is not thread safe.  Only one thread should
> +     * access ls_arp_record->lflow_ref at any given time.
> +     */
> +    struct lflow_ref *lflow_ref;
> +
> +    /* lr_nat_record ptrs that trigger this od to rebuild lflow */
> +    struct hmapx nat_records;
> +};
> +
> +struct ls_arp_table {
> +    struct hmap entries;
> +};
> +
> +#define LS_ARP_TABLE_FOR_EACH(LS_ARP_REC, TABLE) \
> +    HMAP_FOR_EACH (LS_ARP_REC, key_node, \
> +                   &(TABLE)->entries)
> +
> +#define LS_ARP_TABLE_FOR_EACH_IN_P(LS_ARP_REC, JOBID, TABLE) \
> +    HMAP_FOR_EACH_IN_PARALLEL (LS_ARP_REC, key_node, JOBID, \
> +                               &(TABLE)->entries)
> +
> +struct ls_arp_tracked_data {
> +    struct hmapx crupdated; /* Stores 'struct ls_arp_record'. */
> +};
> +
> +struct ed_type_ls_arp {
> +    struct ls_arp_table table;
> +    struct ls_arp_tracked_data trk_data;
> +};
> +
> +void *en_ls_arp_init(struct engine_node *, struct engine_arg *);
> +void en_ls_arp_cleanup(void *data);
> +void en_ls_arp_clear_tracked_data(void *data);
> +enum engine_node_state en_ls_arp_run(struct engine_node *, void *data);
> +
> +enum engine_input_handler_result
> +ls_arp_lr_nat_handler(struct engine_node *node, void *data);
> +enum engine_input_handler_result
> +ls_arp_northd_handler(struct engine_node *node, void *data);
> +
> +static inline bool
> +ls_arp_has_tracked_data(struct ls_arp_tracked_data *trk_data) {
> +    return !hmapx_is_empty(&trk_data->crupdated);
> +}
> +
> +#endif /* EN_LS_ARP_H */
> +
> diff --git a/northd/inc-proc-northd.c b/northd/inc-proc-northd.c
> index bfb893a1a..8068123f0 100644
> --- a/northd/inc-proc-northd.c
> +++ b/northd/inc-proc-northd.c
> @@ -34,6 +34,7 @@
>  #include "en-lr-stateful.h"
>  #include "en-lr-nat.h"
>  #include "en-ls-stateful.h"
> +#include "en-ls-arp.h"
>  #include "en-multicast.h"
>  #include "en-northd.h"
>  #include "en-lflow.h"
> @@ -171,6 +172,7 @@ static ENGINE_NODE(lb_data, CLEAR_TRACKED_DATA);
>  static ENGINE_NODE(lr_nat, CLEAR_TRACKED_DATA);
>  static ENGINE_NODE(lr_stateful, CLEAR_TRACKED_DATA);
>  static ENGINE_NODE(ls_stateful, CLEAR_TRACKED_DATA);
> +static ENGINE_NODE(ls_arp, CLEAR_TRACKED_DATA);
>  static ENGINE_NODE(route_policies);
>  static ENGINE_NODE(routes);
>  static ENGINE_NODE(bfd);
> @@ -296,6 +298,9 @@ void inc_proc_northd_init(struct ovsdb_idl_loop *nb,
>                       ls_stateful_port_group_handler);
>      engine_add_input(&en_ls_stateful, &en_nb_acl, ls_stateful_acl_handler);
>  
> +    engine_add_input(&en_ls_arp, &en_lr_nat, ls_arp_lr_nat_handler);
> +    engine_add_input(&en_ls_arp, &en_northd, ls_arp_northd_handler);
> +
>      engine_add_input(&en_mac_binding_aging, &en_sb_mac_binding, NULL);
>      engine_add_input(&en_mac_binding_aging, &en_northd, NULL);
>      engine_add_input(&en_mac_binding_aging, &en_mac_binding_aging_waker, 
> NULL);
> @@ -390,6 +395,9 @@ void inc_proc_northd_init(struct ovsdb_idl_loop *nb,
>      engine_add_input(&en_lflow, &en_port_group, engine_noop_handler);
>      engine_add_input(&en_lflow, &en_lr_stateful, lflow_lr_stateful_handler);
>      engine_add_input(&en_lflow, &en_ls_stateful, lflow_ls_stateful_handler);
> +
> +    engine_add_input(&en_lflow, &en_ls_arp, lflow_ls_arp_handler);
> +
>      engine_add_input(&en_lflow, &en_multicast_igmp,
>                       lflow_multicast_igmp_handler);
>      engine_add_input(&en_lflow, &en_sb_acl_id, NULL);
> diff --git a/northd/northd.c b/northd/northd.c
> index e0a329a17..33a3ff9b6 100644
> --- a/northd/northd.c
> +++ b/northd/northd.c
> @@ -50,6 +50,7 @@
>  #include "en-lr-nat.h"
>  #include "en-lr-stateful.h"
>  #include "en-ls-stateful.h"
> +#include "en-ls-arp.h"
>  #include "en-multicast.h"
>  #include "en-sampling-app.h"
>  #include "en-datapath-logical-switch.h"
> @@ -135,6 +136,7 @@ static bool vxlan_mode;
>  #define REGBIT_IP_FRAG            "reg0[19]"
>  #define REGBIT_ACL_PERSIST_ID     "reg0[20]"
>  #define REGBIT_ACL_HINT_ALLOW_PERSISTED "reg0[21]"
> +#define REGBIT_EXT_ARP            "reg0[22]"
>  
>  /* Register definitions for switches and routers. */
>  
> @@ -532,6 +534,7 @@ ovn_datapath_create(struct hmap *datapaths, const struct 
> uuid *key,
>      hmap_insert(datapaths, &od->key_node, uuid_hash(&od->key));
>      od->lr_group = NULL;
>      hmap_init(&od->ports);
> +    hmapx_init(&od->ph_ports);
>      sset_init(&od->router_ips);
>      od->ls_peers = VECTOR_EMPTY_INITIALIZER(struct ovn_datapath *);
>      od->router_ports = VECTOR_EMPTY_INITIALIZER(struct ovn_port *);
> @@ -569,6 +572,7 @@ ovn_datapath_destroy(struct hmap *datapaths, struct 
> ovn_datapath *od)
>          vector_destroy(&od->l3dgw_ports);
>          destroy_mcast_info_for_datapath(od);
>          destroy_ports_for_datapath(od);
> +        hmapx_destroy(&od->ph_ports);
>          sset_destroy(&od->router_ips);
>          free(od);
>      }
> @@ -1177,6 +1181,12 @@ lsp_is_vtep(const struct nbrec_logical_switch_port 
> *nbsp)
>      return !strcmp(nbsp->type, "vtep");
>  }
>  
> +static bool
> +lsp_is_l2gw(const struct nbrec_logical_switch_port *nbsp)
> +{
> +    return !strcmp(nbsp->type, "l2gateway");
> +}
> +
>  static bool
>  localnet_can_learn_mac(const struct nbrec_logical_switch_port *nbsp)
>  {
> @@ -1563,6 +1573,10 @@ join_logical_ports_lsp(struct hmap *ports,
>          od->has_vtep_lports = true;
>      }
>  
> +    if (lsp_is_localnet(nbsp) || lsp_is_l2gw(nbsp)) {
> +        hmapx_add(&od->ph_ports, op);
> +    }
> +
>      parse_lsp_addrs(op);
>  
>      op->od = od;
> @@ -1956,6 +1970,15 @@ join_logical_ports(const struct 
> sbrec_port_binding_table *sbrec_pb_table,
>      }
>  }
>  
> +static bool
> +is_nat_distributed(const struct nbrec_nat *nat,
> +                   const struct ovn_datapath *od)
> +{
> +    return !vector_is_empty(&od->l3dgw_ports)
> +           && nat->logical_port && nat->external_mac
> +           && !strcmp(nat->type, "dnat_and_snat");
> +}
> +
>  /* Returns an array of strings, each consisting of a MAC address followed
>   * by one or more IP addresses, and if the port is a distributed gateway
>   * port, followed by 'is_chassis_resident("LPORT_NAME")', where the
> @@ -2014,9 +2037,7 @@ get_nat_addresses(const struct ovn_port *op, size_t *n, 
> bool routable_only,
>  
>          /* Determine whether this NAT rule satisfies the conditions for
>           * distributed NAT processing. */
> -        if (!vector_is_empty(&op->od->l3dgw_ports) &&
> -            !strcmp(nat->type, "dnat_and_snat") &&
> -            nat->logical_port && nat->external_mac) {
> +        if (is_nat_distributed(nat, op->od)) {
>              /* Distributed NAT rule. */
>              if (eth_addr_from_string(nat->external_mac, &mac)) {
>                  struct ds address = DS_EMPTY_INITIALIZER;
> @@ -9210,6 +9231,90 @@ 
> build_drop_arp_nd_flows_for_unbound_router_ports(struct ovn_port *op,
>      ds_destroy(&match);
>  }
>  
> +/*
> + * Create ARP filtering flow for od, assumed logical switch,
> + * for the following condition:
> + * Given lswitch has either localnet or l2gateway ports and
> + * router connection ports that requires chassis residence.
> + * ARP requests coming from localnet/l2gateway ports
> + * allowed for processing on resident chassis only.
> + */
> +static void
> +build_lswitch_arp_chassis_resident(const struct ovn_datapath *od,
> +                                   struct lflow_table *lflows,
> +                                   const struct ls_arp_record *ar)
> +{
> +    struct sset port_check;
> +    struct sset resident_check;
> +
> +    sset_init(&port_check);
> +    sset_init(&resident_check);
> +
> +    struct ds match = DS_EMPTY_INITIALIZER;
> +
> +    struct hmapx_node *node;
> +    HMAPX_FOR_EACH (node, &od->ph_ports) {
> +        struct ovn_port *op = node->data;
> +
> +        ds_clear(&match);
> +        ds_put_format(&match, "(arp.op == 1 || arp.op == 2) && inport == %s",
> +                      op->json_key);
> +        sset_add(&port_check, ds_cstr(&match));

Have you done any evaluation of the impact of these new flows at scale?

> +    }
> +
> +    struct ovn_port *op;
> +    VECTOR_FOR_EACH (&od->router_ports, op) {
> +        struct ovn_port *op_r = op->peer;
> +
> +        if (lrp_is_l3dgw(op_r)) {
> +            ds_clear(&match);
> +            ds_put_format(&match,
> +                          REGBIT_EXT_ARP" == 1 && is_chassis_resident(%s)",
> +                          op_r->cr_port->json_key);
> +            sset_add(&resident_check, ds_cstr(&match));
> +        }
> +    }
> +
> +    struct hmapx_node *hmapx_node;
> +    HMAPX_FOR_EACH (hmapx_node, &ar->nat_records) {
> +        struct lr_nat_record *nr = hmapx_node->data;
> +
> +        for (size_t i = 0; i < nr->n_nat_entries; i++) {
> +            struct ovn_nat *ent = &nr->nat_entries[i];
> +            if (ent->is_valid && ent->is_distributed) {
> +                ds_clear(&match);
> +                ds_put_format(&match,
> +                              REGBIT_EXT_ARP
> +                              " == 1 && is_chassis_resident(\"%s\")",
> +                              ent->nb->logical_port);
> +                sset_add(&resident_check, ds_cstr(&match));
> +            }
> +        }
> +    }
> +
> +    if (!sset_is_empty(&port_check) && !sset_is_empty(&resident_check)) {
> +        const char *item;
> +
> +        SSET_FOR_EACH (item, &port_check) {
> +            ovn_lflow_add(lflows, od, S_SWITCH_IN_CHECK_PORT_SEC, 75,
> +                          item, REGBIT_EXT_ARP" = 1; next;",
> +                          ar->lflow_ref);
> +        }
> +
> +        SSET_FOR_EACH (item, &resident_check) {
> +            ovn_lflow_add(lflows, od, S_SWITCH_IN_APPLY_PORT_SEC, 75,
> +                          item, "next;", ar->lflow_ref);
> +        }
> +
> +        ovn_lflow_add(lflows, od, S_SWITCH_IN_APPLY_PORT_SEC, 70,
> +                      REGBIT_EXT_ARP" == 1", "drop;", ar->lflow_ref);
> +    }
> +
> +    ds_destroy(&match);
> +    sset_destroy(&port_check);
> +    sset_destroy(&resident_check);
> +}
> +
>  static bool
>  is_vlan_transparent(const struct ovn_datapath *od)
>  {
> @@ -13594,10 +13699,6 @@ build_neigh_learning_flows_for_lrouter_port(
>                            op->lrp_networks.ipv4_addrs[i].network_s,
>                            op->lrp_networks.ipv4_addrs[i].plen,
>                            op->lrp_networks.ipv4_addrs[i].addr_s);
> -            if (lrp_is_l3dgw(op)) {
> -                ds_put_format(match, " && is_chassis_resident(%s)",
> -                              op->cr_port->json_key);

Have you done any evaluation of the impact of these new flows added to OvS at 
scale?
(the same apply for each flow where you removed 'is_chassis_resident()')

> -            }
>              const char *actions_s = REGBIT_LOOKUP_NEIGHBOR_RESULT
>                                " = lookup_arp(inport, arp.spa, arp.sha); "
>                                REGBIT_LOOKUP_NEIGHBOR_IP_RESULT" = 1;"
> @@ -13614,10 +13715,6 @@ build_neigh_learning_flows_for_lrouter_port(
>                        op->json_key,
>                        op->lrp_networks.ipv4_addrs[i].network_s,
>                        op->lrp_networks.ipv4_addrs[i].plen);
> -        if (lrp_is_l3dgw(op)) {
> -            ds_put_format(match, " && is_chassis_resident(%s)",
> -                          op->cr_port->json_key);
> -        }
>          ds_clear(actions);
>          ds_put_format(actions, REGBIT_LOOKUP_NEIGHBOR_RESULT
>                        " = lookup_arp(inport, arp.spa, arp.sha); %snext;",
> @@ -17532,6 +17629,7 @@ struct lswitch_flow_build_info {
>      const struct ls_port_group_table *ls_port_groups;
>      const struct lr_stateful_table *lr_stateful_table;
>      const struct ls_stateful_table *ls_stateful_table;
> +    const struct ls_arp_table *ls_arp_table;
>      struct lflow_table *lflows;
>      const struct shash *meter_groups;
>      const struct hmap *lb_dps_map;
> @@ -17714,6 +17812,7 @@ build_lflows_thread(void *arg)
>      struct worker_control *control = (struct worker_control *) arg;
>      const struct lr_stateful_record *lr_stateful_rec;
>      const struct ls_stateful_record *ls_stateful_rec;
> +    const struct ls_arp_record *ls_arp_rec;
>      struct lswitch_flow_build_info *lsi;
>      struct ovn_lb_datapaths *lb_dps;
>      struct ovn_datapath *od;
> @@ -17870,6 +17969,19 @@ build_lflows_thread(void *arg)
>                  }
>              }
>  
> +            for (bnum = control->id;
> +                    bnum <= lsi->ls_arp_table->entries.mask;
> +                    bnum += control->pool->size)
> +            {
> +                LS_ARP_TABLE_FOR_EACH_IN_P (ls_arp_rec, bnum,
> +                                            lsi->ls_arp_table) {
> +                    od = ovn_datapaths_find_by_index(
> +                        lsi->ls_datapaths, ls_arp_rec->ls_index);
> +                    build_lswitch_arp_chassis_resident(od, lsi->lflows,
> +                                                       ls_arp_rec);
> +                }
> +            }
> +
>              lsi->thread_lflow_counter = thread_lflow_counter;
>          }
>          post_completed_work(control);
> @@ -17918,6 +18030,7 @@ build_lswitch_and_lrouter_flows(
>      const struct ls_port_group_table *ls_pgs,
>      const struct lr_stateful_table *lr_stateful_table,
>      const struct ls_stateful_table *ls_stateful_table,
> +    const struct ls_arp_table *ls_arp_table,
>      struct lflow_table *lflows,
>      const struct shash *meter_groups,
>      const struct hmap *lb_dps_map,
> @@ -17954,6 +18067,7 @@ build_lswitch_and_lrouter_flows(
>              lsiv[index].ls_port_groups = ls_pgs;
>              lsiv[index].lr_stateful_table = lr_stateful_table;
>              lsiv[index].ls_stateful_table = ls_stateful_table;
> +            lsiv[index].ls_arp_table = ls_arp_table;
>              lsiv[index].meter_groups = meter_groups;
>              lsiv[index].lb_dps_map = lb_dps_map;
>              lsiv[index].local_svc_monitor_map =
> @@ -17988,6 +18102,7 @@ build_lswitch_and_lrouter_flows(
>      } else {
>          const struct lr_stateful_record *lr_stateful_rec;
>          const struct ls_stateful_record *ls_stateful_rec;
> +        const struct ls_arp_record *ls_arp_rec;
>          struct ovn_lb_datapaths *lb_dps;
>          struct ovn_datapath *od;
>          struct ovn_port *op;
> @@ -18000,6 +18115,7 @@ build_lswitch_and_lrouter_flows(
>              .ls_port_groups = ls_pgs,
>              .lr_stateful_table = lr_stateful_table,
>              .ls_stateful_table = ls_stateful_table,
> +            .ls_arp_table = ls_arp_table,
>              .lflows = lflows,
>              .meter_groups = meter_groups,
>              .lb_dps_map = lb_dps_map,
> @@ -18095,6 +18211,12 @@ build_lswitch_and_lrouter_flows(
>                                      lsi.sbrec_acl_id_table);
>          }
>  
> +        LS_ARP_TABLE_FOR_EACH (ls_arp_rec, ls_arp_table) {
> +            od = ovn_datapaths_find_by_index(lsi.ls_datapaths,
> +                                             ls_arp_rec->ls_index);
> +            build_lswitch_arp_chassis_resident(od, lsi.lflows, ls_arp_rec);
> +        }
> +
>          ds_destroy(&lsi.match);
>          ds_destroy(&lsi.actions);
>      }
> @@ -18178,6 +18300,7 @@ void build_lflows(struct ovsdb_idl_txn *ovnsb_txn,
>                                      input_data->ls_port_groups,
>                                      input_data->lr_stateful_table,
>                                      input_data->ls_stateful_table,
> +                                    input_data->ls_arp_table,
>                                      lflows,
>                                      input_data->meter_groups,
>                                      input_data->lb_datapaths_map,
> @@ -18603,6 +18726,40 @@ lflow_handle_ls_stateful_changes(struct 
> ovsdb_idl_txn *ovnsb_txn,
>      return true;
>  }
>  
> +bool
> +lflow_handle_ls_arp_changes(struct ovsdb_idl_txn *ovnsb_txn,
> +                            struct ls_arp_tracked_data *trk_data,
> +                            struct lflow_input *lflow_input,
> +                            struct lflow_table *lflows)
> +{
> +    struct hmapx_node *hmapx_node;
> +
> +    HMAPX_FOR_EACH (hmapx_node, &trk_data->crupdated) {
> +        const struct ls_arp_record *ar = hmapx_node->data;
> +        const struct ovn_datapath *od =
> +            ovn_datapaths_find_by_index(lflow_input->ls_datapaths,
> +                                        ar->ls_index);
> +        lflow_ref_unlink_lflows(ar->lflow_ref);
> +
> +        /* Generate new lflows. */
> +        build_lswitch_arp_chassis_resident(od, lflows, ar);
> +
> +        /* Sync the new flows to SB. */
> +        bool handled = lflow_ref_sync_lflows(
> +            ar->lflow_ref, lflows, ovnsb_txn,
> +            lflow_input->ls_datapaths,
> +            lflow_input->lr_datapaths,
> +            lflow_input->ovn_internal_version_changed,
> +            lflow_input->sbrec_logical_flow_table,
> +            lflow_input->sbrec_logical_dp_group_table);
> +        if (!handled) {
> +            return false;
> +        }
> +    }
> +
> +    return true;
> +}
> +
>  static bool
>  mirror_needs_update(const struct nbrec_mirror *nb_mirror,
>                      const struct sbrec_mirror *sb_mirror)
> diff --git a/northd/northd.h b/northd/northd.h
> index 8f865e8b3..d75883612 100644
> --- a/northd/northd.h
> +++ b/northd/northd.h
> @@ -20,12 +20,14 @@
>  #include "lib/ovn-util.h"
>  #include "lib/ovs-atomic.h"
>  #include "lib/sset.h"
> +#include "lib/hmapx.h"
>  #include "northd/en-port-group.h"
>  #include "northd/ipam.h"
>  #include "openvswitch/hmap.h"
>  #include "simap.h"
>  #include "ovs-thread.h"
>  #include "en-lr-stateful.h"
> +#include "en-ls-arp.h"
>  #include "vec.h"
>  #include "datapath-sync.h"
>  
> @@ -253,6 +255,7 @@ struct lflow_input {
>      const struct ls_port_group_table *ls_port_groups;
>      const struct lr_stateful_table *lr_stateful_table;
>      const struct ls_stateful_table *ls_stateful_table;
> +    const struct ls_arp_table *ls_arp_table;
>      const struct shash *meter_groups;
>      const struct hmap *lb_datapaths_map;
>      const struct sset *bfd_ports;
> @@ -445,6 +448,9 @@ struct ovn_datapath {
>       * Valid only if it is logical router datapath. NULL otherwise. */
>      struct lrouter_group *lr_group;
>  
> +    /* Set of localnet or l2gw ports. */
> +    struct hmapx ph_ports;
> +
>      /* Map of ovn_port objects belonging to this datapath.
>       * This map doesn't include derived ports. */
>      struct hmap ports;
> @@ -922,6 +928,10 @@ bool lflow_handle_ls_stateful_changes(struct 
> ovsdb_idl_txn *,
>                                        struct ls_stateful_tracked_data *,
>                                        struct lflow_input *,
>                                        struct lflow_table *lflows);
> +bool lflow_handle_ls_arp_changes(struct ovsdb_idl_txn *,
> +                                 struct ls_arp_tracked_data *,
> +                                 struct lflow_input *,
> +                                 struct lflow_table *lflows);
>  bool northd_handle_sb_port_binding_changes(
>      const struct sbrec_port_binding_table *, struct hmap *ls_ports,
>      struct hmap *lr_ports);
> diff --git a/northd/ovn-northd.c b/northd/ovn-northd.c
> index 62da4a3aa..ab9339369 100644
> --- a/northd/ovn-northd.c
> +++ b/northd/ovn-northd.c
> @@ -989,6 +989,7 @@ main(int argc, char *argv[])
>      stopwatch_create(LR_NAT_RUN_STOPWATCH_NAME, SW_MS);
>      stopwatch_create(LR_STATEFUL_RUN_STOPWATCH_NAME, SW_MS);
>      stopwatch_create(LS_STATEFUL_RUN_STOPWATCH_NAME, SW_MS);
> +    stopwatch_create(LS_ARP_RUN_STOPWATCH_NAME, SW_MS);
>      stopwatch_create(ADVERTISED_ROUTE_SYNC_RUN_STOPWATCH_NAME, SW_MS);
>      stopwatch_create(LEARNED_ROUTE_SYNC_RUN_STOPWATCH_NAME, SW_MS);
>      stopwatch_create(DYNAMIC_ROUTES_RUN_STOPWATCH_NAME, SW_MS);
> diff --git a/tests/ovn-northd.at b/tests/ovn-northd.at
> index b2b9f092c..57372a1e6 100644
> --- a/tests/ovn-northd.at
> +++ b/tests/ovn-northd.at
> @@ -7274,10 +7274,10 @@ AT_CHECK([grep lr_in_admission lrflows | grep cr-DR | 
> ovn_strip_lflows], [0], [d
>    table=??(lr_in_admission    ), priority=50   , match=(eth.dst == 
> 04:ac:10:01:00:01 && inport == "DR-S3" && is_chassis_resident("cr-DR-S3")), 
> action=(xreg0[[0..47]] = 04:ac:10:01:00:01; next;)
>  ])
>  # Check the flows in lr_in_lookup_neighbor stage
> -AT_CHECK([grep lr_in_lookup_neighbor lrflows | grep cr-DR | 
> ovn_strip_lflows], [0], [dnl
> -  table=??(lr_in_lookup_neighbor), priority=100  , match=(inport == "DR-S1" 
> && arp.spa == 172.16.1.0/24 && arp.op == 1 && 
> is_chassis_resident("cr-DR-S1")), action=(reg9[[2]] = lookup_arp(inport, 
> arp.spa, arp.sha); next;)
> -  table=??(lr_in_lookup_neighbor), priority=100  , match=(inport == "DR-S2" 
> && arp.spa == 172.16.2.0/24 && arp.op == 1 && 
> is_chassis_resident("cr-DR-S2")), action=(reg9[[2]] = lookup_arp(inport, 
> arp.spa, arp.sha); next;)
> -  table=??(lr_in_lookup_neighbor), priority=100  , match=(inport == "DR-S3" 
> && arp.spa == 172.16.3.0/24 && arp.op == 1 && 
> is_chassis_resident("cr-DR-S3")), action=(reg9[[2]] = lookup_arp(inport, 
> arp.spa, arp.sha); next;)
> +AT_CHECK([grep lr_in_lookup_neighbor lrflows | grep DR-S. | 
> ovn_strip_lflows], [0], [dnl
> +  table=??(lr_in_lookup_neighbor), priority=100  , match=(inport == "DR-S1" 
> && arp.spa == 172.16.1.0/24 && arp.op == 1), action=(reg9[[2]] = 
> lookup_arp(inport, arp.spa, arp.sha); next;)
> +  table=??(lr_in_lookup_neighbor), priority=100  , match=(inport == "DR-S2" 
> && arp.spa == 172.16.2.0/24 && arp.op == 1), action=(reg9[[2]] = 
> lookup_arp(inport, arp.spa, arp.sha); next;)
> +  table=??(lr_in_lookup_neighbor), priority=100  , match=(inport == "DR-S3" 
> && arp.spa == 172.16.3.0/24 && arp.op == 1), action=(reg9[[2]] = 
> lookup_arp(inport, arp.spa, arp.sha); next;)
>  ])
>  # Check the flows in lr_in_gw_redirect stage
>  AT_CHECK([grep lr_in_gw_redirect lrflows | grep cr-DR | ovn_strip_lflows], 
> [0], [dnl
> @@ -13902,7 +13902,7 @@ AT_CHECK([grep -Fe "172.168.0.110" -e "172.168.0.120" 
> -e "10.0.0.3" -e "20.0.0.3
>    table=??(lr_out_undnat      ), priority=100  , match=(ip && ip4.src == 
> 20.0.0.3 && outport == "lr0-public" && is_chassis_resident("cr-lr0-public")), 
> action=(ct_dnat;)
>  ])
>  
> -AT_CHECK([grep -Fe "172.168.0.110" -e "172.168.0.120" -e "10.0.0.3" -e 
> "20.0.0.3" -e "30:54:00:00:00:03"  -e "sw0-port1" publicflows | 
> ovn_strip_lflows], [0], [dnl
> +AT_CHECK([grep -Fe "172.168.0.110" -e "172.168.0.120" -e "10.0.0.3" -e 
> "20.0.0.3" -e "30:54:00:00:00:03"  -e "sw0-port1" publicflows | 
> ovn_strip_lflows | grep -v "reg0.*22"], [0], [dnl
>    table=??(ls_in_l2_lkup      ), priority=50   , match=(eth.dst == 
> 30:54:00:00:00:03 && is_chassis_resident("sw0-port1")), action=(outport = 
> "public-lr0"; output;)
>    table=??(ls_in_l2_lkup      ), priority=75   , match=(eth.src == 
> {00:00:00:00:ff:02, 30:54:00:00:00:03} && eth.dst == ff:ff:ff:ff:ff:ff && 
> (arp.op == 1 || rarp.op == 3 || nd_ns)), action=(outport = "_MC_flood_l2"; 
> output;)
>    table=??(ls_in_l2_lkup      ), priority=80   , match=(flags[[1]] == 0 && 
> arp.op == 1 && arp.tpa == 172.168.0.110), action=(clone {outport = 
> "public-lr0"; output; }; outport = "_MC_flood_l2"; output;)
> diff --git a/tests/ovn.at b/tests/ovn.at
> index 292ca0dae..926efde08 100644
> --- a/tests/ovn.at
> +++ b/tests/ovn.at
> @@ -43872,3 +43872,160 @@ wait_for_ports_up sw0-vir
>  OVN_CLEANUP([hv1])
>  AT_CLEANUP
>  ])
> +
> +AT_SETUP([GARP delivery: gw and extenal ports])
> +ovn_start
> +#
> +# Configure initial environment
> +# LR1: down_link <-> LS1: up_link
> +# set lr_down: gateway port (chassis redirect) bound to hv1
> +# LS1: down_vif1 - vif port bound to hv1
> +#      down_vif2 - vif port bound to hv2
> +#      down_ext - outher port will be iteratred as localnet, l2gateway
> +#
> +# Test: throw ARP annonce request from vitrual ports (down_vif1, down_vif2)
> +#       ensure mac_binding is always updated.
> +#       (Fixing the issue: mac_binding is only updated for packets came from
> +#        down_link's resident chassis)
> +#       throw ARP annonce request from from localnet.
> +#       ensure mac_binding is updated only if localnet bound to same hv as 
> l3dgw
> +#
> +
> +check ovn-nbctl lr-add lr1
> +check ovn-nbctl lrp-add lr1 down_link f0:00:00:00:00:f1 192.168.1.1/24
> +
> +check ovn-nbctl ls-add ls1
> +check ovn-nbctl lsp-add ls1 up_link
> +check ovn-nbctl lsp-add ls1 down_vif1
> +check ovn-nbctl lsp-add ls1 down_vif2
> +check ovn-nbctl lsp-add ls1 down_ext
> +
> +check ovn-nbctl set Logical_Switch_Port up_link \
> +    type=router \
> +    options:router-port=down_link \
> +    addresses=router
> +
> +check ovn-nbctl lsp-set-addresses down_vif1 'f0:00:00:00:00:01 192.168.1.101'
> +check ovn-nbctl lsp-set-addresses down_vif2 'f0:00:00:00:00:02 192.168.1.102'
> +
> +check ovn-nbctl lsp-set-type down_ext localnet
> +check ovn-nbctl lsp-set-options down_ext network_name=physnet1
> +check ovn-nbctl lrp-set-gateway-chassis down_link hv1
> +
> +net_add n1
> +
> +# Create hypervisor hv1 connected to n1
> +sim_add hv1
> +as hv1
> +ovs-vsctl add-br br-phys
> +ovn_attach n1 br-phys 192.168.0.1
> +ovs-vsctl add-port br-int vif1 -- \
> +    set Interface vif1 external-ids:iface-id=down_vif1 \
> +    options:tx_pcap=hv1/vif1-tx.pcap options:rxq_pcap=hv1/vif1-rx.pcap \
> +    ofport_request=1
> +
> +# Create hypervisor hv2 connected to n1, add localnet here
> +sim_add hv2
> +as hv2
> +ovs-vsctl add-br br-phys
> +ovs-vsctl add-br br-eth0
> +ovn_attach n1 br-phys 192.168.0.2
> +ovs-vsctl add-port br-int vif2 -- \
> +    set Interface vif2 external-ids:iface-id=down_vif2 \
> +    options:tx_pcap=hv2/vif2-tx.pcap options:rxq_pcap=hv2/vif2-rx.pcap \
> +    ofport_request=1
> +
> +ovs-vsctl set Open_vSwitch . 
> external_ids:ovn-bridge-mappings="physnet1:br-eth0"
> +
> +ovs-vsctl add-port br-eth0 vif_ext -- \
> +    set Interface vif_ext options:tx_pcap=hv2/vif_ext-tx.pcap \
> +    options:rxq_pcap=hv2/vif_ext-rx.pcap
> +
> +# Pre-populate the hypervisors' ARP tables so that we don't lose any
> +# packets for ARP resolution (native tunneling doesn't queue packets
> +# for ARP resolution).
> +OVN_POPULATE_ARP
> +
> +wait_for_ports_up
> +check ovn-nbctl --wait=hv sync
> +
> +#
> +# Annonce 192.168.1.222 from localnet in hv2
> +# result: drop, hv2 is not gateway chassis for down_link
> +#
> +sha=0200000000ee
> +tha=000000000000
> +spa=`ip_to_hex 192 168 1 222`
> +tpa=$spa
> +garp=ffffffffffff${sha}08060001080006040001${sha}${spa}${tha}${tpa}

Packets are supposed to be created with 'fmt_pkt'.

> +as hv2 ovs-appctl netdev-dummy/receive vif_ext $garp
> +
> +#
> +# Make hv2 gateway chassis
> +# Annonce 192.168.1.223 from localnet in hv2
> +# result: ok, hv2 is gateway chassis for down_link
> +#
> +check ovn-nbctl lrp-set-gateway-chassis down_link hv2
> +
> +wait_row_count Port_Binding 1 logical_port=cr-down_link 'chassis!=[[]]'
> +
> +sha=0200000000ee
> +tha=000000000000
> +spa=`ip_to_hex 192 168 1 223`
> +tpa=$spa
> +garp=ffffffffffff${sha}08060001080006040001${sha}${spa}${tha}${tpa}
> +as hv2 ovs-appctl netdev-dummy/receive vif_ext $garp
> +
> +#
> +# Annonce 192.168.1.111, 112 from vif1, vif2 in hv1, hv2
> +# result: ok, vif1, vif2 are virtual ports, restrictions are not applied.
> +#
> +sha=f00000000001
> +tha=000000000000
> +spa=`ip_to_hex 192 168 1 111`
> +tpa=000000000000
> +garp=ffffffffffff${sha}08060001080006040001${sha}${spa}${tha}${tpa}
> +as hv1 ovs-appctl netdev-dummy/receive vif1 $garp
> +
> +sha=f00000000002
> +tha=000000000000
> +spa=`ip_to_hex 192 168 1 112`
> +tpa=000000000000
> +garp=ffffffffffff${sha}08060001080006040001${sha}${spa}${tha}${tpa}
> +as hv2 ovs-appctl netdev-dummy/receive vif2 $garp
> +
> +#
> +# Set down_ext type to l2gateway
> +# Annonce 192.168.1.113, 114 from vif1, vif2 in hv1, hv2
> +# result: ok, vif1, vif2 are virtual ports, restrictions are not applied.
> +#
> +check ovn-nbctl --wait=hv lsp-set-type down_ext l2gateway
> +
> +sha=f00000000001
> +tha=000000000000
> +spa=`ip_to_hex 192 168 1 113`
> +tpa=000000000000
> +garp=ffffffffffff${sha}08060001080006040001${sha}${spa}${tha}${tpa}
> +as hv1 ovs-appctl netdev-dummy/receive vif1 $garp
> +
> +sha=f00000000002
> +tha=000000000000
> +spa=`ip_to_hex 192 168 1 114`
> +tpa=000000000000
> +garp=ffffffffffff${sha}08060001080006040001${sha}${spa}${tha}${tpa}
> +as hv2 ovs-appctl netdev-dummy/receive vif2 $garp
> +
> +wait_row_count MAC_Binding 1 ip="192.168.1.111" mac='"f0:00:00:00:00:01"' 
> logical_port='"down_link"'
> +wait_row_count MAC_Binding 1 ip="192.168.1.112" mac='"f0:00:00:00:00:02"' 
> logical_port='"down_link"'
> +wait_row_count MAC_Binding 1 ip="192.168.1.113" mac='"f0:00:00:00:00:01"' 
> logical_port='"down_link"'
> +wait_row_count MAC_Binding 1 ip="192.168.1.114" mac='"f0:00:00:00:00:02"' 
> logical_port='"down_link"'
> +wait_row_count MAC_Binding 1 ip="192.168.1.223" mac='"02:00:00:00:00:ee"' 
> logical_port='"down_link"'
> +wait_row_count MAC_Binding 0 ip="192.168.1.222" mac='"02:00:00:00:00:ee"' 
> logical_port='"down_link"'
> +
> +check ovn-nbctl --wait=hv lsp-set-type down_ext localnet
> +
> +OVN_CLEANUP([hv1],[hv2])
> +
> +AT_CLEANUP
> +])
> +
> -- 
> 2.49.0
> 
> _______________________________________________
> dev mailing list
> d...@openvswitch.org
> https://mail.openvswitch.org/mailman/listinfo/ovs-dev
> 
_______________________________________________
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to