[ovs-dev] [PATCH v14 6/6] Add incremental proessing to lflow_run and physical_run
Ryan Moats
rmoats at us.ibm.com
Wed Apr 13 13:38:35 UTC 2016
This code changes to allow incremental processing of the
logical flow and physical binding tables whenver possible.
Side Effects:
- Make flow table persistent in ovn controller
- Reset lflow processing when adding/removing patch ports
Note: flows created by physical_run for multicast_groups are
*NOT* handled incrementally due to to be solved issues
with GWs and local routers.
Signed-off-by: Ryan Moats <rmoats at us.ibm.com>
---
ovn/controller/binding.c | 7 +-
ovn/controller/encaps.c | 4 +-
ovn/controller/lflow.c | 177 +++++---
ovn/controller/lflow.h | 3 +
ovn/controller/lport.c | 12 +-
ovn/controller/ovn-controller.c | 1 -
ovn/controller/patch.c | 4 +-
ovn/controller/physical.c | 896 ++++++++++++++++++++++------------------
ovn/controller/physical.h | 2 +
9 files changed, 644 insertions(+), 462 deletions(-)
diff --git a/ovn/controller/binding.c b/ovn/controller/binding.c
index 7f8c26c..87f4362 100644
--- a/ovn/controller/binding.c
+++ b/ovn/controller/binding.c
@@ -15,6 +15,8 @@
#include <config.h>
#include "binding.h"
+#include "lflow.h"
+#include "lport.h"
#include "lib/bitmap.h"
#include "lib/hmap.h"
@@ -163,7 +165,7 @@ remove_local_datapath(struct hmap *local_datapaths, const struct uuid *uuid)
hmap_remove(local_datapaths, &ld->hmap_node);
hmap_remove(&local_datapaths_by_uuid, &ld->uuid_hmap_node);
free(ld);
- //reset_flow_processing();
+ reset_flow_processing();
}
}
@@ -183,7 +185,7 @@ add_local_datapath(struct hmap *local_datapaths,
binding_rec->datapath->tunnel_key);
hmap_insert(&local_datapaths_by_uuid, &ld->uuid_hmap_node,
uuid_hash(uuid));
- //reset_flow_processing();
+ reset_flow_processing();
}
static void
@@ -277,6 +279,7 @@ binding_run(struct controller_ctx *ctx, const struct ovsrec_bridge *br_int,
consider_local_datapath(ctx, &lports, chassis_rec, binding_rec,
local_datapaths);
}
+ flag_rebuild_lport_mcast_indexes();
process_full_binding = false;
} else {
SBREC_PORT_BINDING_FOR_EACH_TRACKED(binding_rec, ctx->ovnsb_idl) {
diff --git a/ovn/controller/encaps.c b/ovn/controller/encaps.c
index 271ddef..ffb8e46 100644
--- a/ovn/controller/encaps.c
+++ b/ovn/controller/encaps.c
@@ -217,7 +217,7 @@ tunnel_add(const struct sbrec_chassis *chassis_rec,
sset_add(&tc.port_names, port_name);
free(port_name);
free(ports);
- // reset_flow_processing();
+ reset_flow_processing();
reset_binding_processing();
process_full_encaps = true;
}
@@ -341,7 +341,7 @@ encaps_run(struct controller_ctx *ctx, const struct ovsrec_bridge *br_int,
hmap_remove(&tc.tunnel_hmap_by_uuid,
&port_hash->uuid_node);
free(port_hash);
- //reset_flow_processing();
+ reset_flow_processing();
reset_binding_processing();
}
continue;
diff --git a/ovn/controller/lflow.c b/ovn/controller/lflow.c
index 7e49e6d..caa16e7 100644
--- a/ovn/controller/lflow.c
+++ b/ovn/controller/lflow.c
@@ -26,6 +26,7 @@
#include "ovn/lib/expr.h"
#include "ovn/lib/ovn-sb-idl.h"
#include "packets.h"
+#include "physical.h"
#include "simap.h"
VLOG_DEFINE_THIS_MODULE(lflow);
@@ -35,6 +36,16 @@ VLOG_DEFINE_THIS_MODULE(lflow);
/* Contains "struct expr_symbol"s for fields supported by OVN lflows. */
static struct shash symtab;
+static bool full_flow_processing = false;
+static bool full_logical_flow_processing = false;
+static bool full_neighbor_flow_processing = false;
+
+void
+reset_flow_processing(void)
+{
+ full_flow_processing = true;
+}
+
static void
add_logical_register(struct shash *symtab, enum mf_field_id id)
{
@@ -193,24 +204,22 @@ is_switch(const struct sbrec_datapath_binding *ldp)
}
-/* Adds the logical flows from the Logical_Flow table to flow tables. */
static void
-add_logical_flows(struct controller_ctx *ctx, const struct lport_index *lports,
- const struct mcgroup_index *mcgroups,
- const struct hmap *local_datapaths,
- const struct hmap *patched_datapaths,
- const struct simap *ct_zones)
+consider_logical_flow(const struct lport_index *lports,
+ const struct mcgroup_index *mcgroups,
+ const struct sbrec_logical_flow *lflow,
+ const struct hmap *local_datapaths,
+ const struct hmap *patched_datapaths,
+ const struct simap *ct_zones,
+ uint32_t *conj_id_ofs_p,
+ bool is_new)
{
- uint32_t conj_id_ofs = 1;
-
- const struct sbrec_logical_flow *lflow;
- SBREC_LOGICAL_FLOW_FOR_EACH (lflow, ctx->ovnsb_idl) {
/* Determine translation of logical table IDs to physical table IDs. */
bool ingress = !strcmp(lflow->pipeline, "ingress");
const struct sbrec_datapath_binding *ldp = lflow->logical_datapath;
if (!ldp) {
- continue;
+ return;
}
if (is_switch(ldp)) {
/* For a logical switch datapath, local_datapaths tells us if there
@@ -243,11 +252,11 @@ add_logical_flows(struct controller_ctx *ctx, const struct lport_index *lports,
if (!get_local_datapath(local_datapaths, ldp->tunnel_key)) {
if (!ingress) {
- continue;
+ return;
}
if (!get_patched_datapath(patched_datapaths,
ldp->tunnel_key)) {
- continue;
+ return;
}
}
}
@@ -293,7 +302,7 @@ add_logical_flows(struct controller_ctx *ctx, const struct lport_index *lports,
VLOG_WARN_RL(&rl, "error parsing actions \"%s\": %s",
lflow->actions, error);
free(error);
- continue;
+ return;
}
/* Translate OVN match into table of OpenFlow matches. */
@@ -315,7 +324,7 @@ add_logical_flows(struct controller_ctx *ctx, const struct lport_index *lports,
expr_destroy(prereqs);
ofpbuf_uninit(&ofpacts);
free(error);
- continue;
+ return;
}
expr = expr_simplify(expr);
@@ -330,11 +339,11 @@ add_logical_flows(struct controller_ctx *ctx, const struct lport_index *lports,
match_set_metadata(&m->match,
htonll(lflow->logical_datapath->tunnel_key));
if (m->match.wc.masks.conj_id) {
- m->match.flow.conj_id += conj_id_ofs;
+ m->match.flow.conj_id += *conj_id_ofs_p;
}
if (!m->n) {
ofctrl_add_flow(ptable, lflow->priority, &m->match, &ofpacts,
- &lflow->header_.uuid, true);
+ &lflow->header_.uuid, is_new);
} else {
uint64_t conj_stubs[64 / 8];
struct ofpbuf conj;
@@ -345,20 +354,54 @@ add_logical_flows(struct controller_ctx *ctx, const struct lport_index *lports,
struct ofpact_conjunction *dst;
dst = ofpact_put_CONJUNCTION(&conj);
- dst->id = src->id + conj_id_ofs;
+ dst->id = src->id + *conj_id_ofs_p;
dst->clause = src->clause;
dst->n_clauses = src->n_clauses;
}
ofctrl_add_flow(ptable, lflow->priority, &m->match, &conj,
- &lflow->header_.uuid, true);
+ &lflow->header_.uuid, is_new);
ofpbuf_uninit(&conj);
}
}
-
/* Clean up. */
expr_matches_destroy(&matches);
ofpbuf_uninit(&ofpacts);
- conj_id_ofs += n_conjs;
+ *conj_id_ofs_p += n_conjs;
+}
+
+/* Adds the logical flows from the Logical_Flow table to 'flow_table'. */
+static void
+add_logical_flows(struct controller_ctx *ctx, const struct lport_index *lports,
+ const struct mcgroup_index *mcgroups,
+ const struct hmap *local_datapaths,
+ const struct hmap *patched_datapaths,
+ const struct simap *ct_zones)
+{
+ uint32_t conj_id_ofs = 1;
+ const struct sbrec_logical_flow *lflow;
+
+ if (full_logical_flow_processing) {
+ SBREC_LOGICAL_FLOW_FOR_EACH (lflow, ctx->ovnsb_idl) {
+ consider_logical_flow(lports, mcgroups, lflow, local_datapaths,
+ patched_datapaths, ct_zones, &conj_id_ofs,
+ true);
+ }
+ full_logical_flow_processing = false;
+ } else {
+ SBREC_LOGICAL_FLOW_FOR_EACH_TRACKED (lflow, ctx->ovnsb_idl) {
+ bool is_deleted = sbrec_logical_flow_row_get_seqno(lflow,
+ OVSDB_IDL_CHANGE_DELETE) > 0;
+ bool is_new = sbrec_logical_flow_row_get_seqno(lflow,
+ OVSDB_IDL_CHANGE_MODIFY) == 0;
+
+ if (is_deleted) {
+ ofctrl_remove_flows(&lflow->header_.uuid);
+ continue;
+ }
+ consider_logical_flow(lports, mcgroups, lflow, local_datapaths,
+ patched_datapaths, ct_zones, &conj_id_ofs,
+ is_new);
+ }
}
}
@@ -375,6 +418,46 @@ put_load(const uint8_t *data, size_t len,
bitwise_one(&sf->mask, sf->field->n_bytes, ofs, n_bits);
}
+unsigned int lflow_mac_binding_seqno = 0;
+
+static void
+consider_neighbor_flow(const struct lport_index *lports,
+ const struct sbrec_mac_binding *b,
+ struct ofpbuf *ofpacts_p,
+ struct match *match_p,
+ bool is_new)
+{
+ const struct sbrec_port_binding *pb
+ = lport_lookup_by_name(lports, b->logical_port);
+ if (!pb) {
+ return;
+ }
+
+ struct eth_addr mac;
+ if (!eth_addr_from_string(b->mac, &mac)) {
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(5, 1);
+ VLOG_WARN_RL(&rl, "bad 'mac' %s", b->mac);
+ return;
+ }
+
+ ovs_be32 ip;
+ if (!ip_parse(b->ip, &ip)) {
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(5, 1);
+ VLOG_WARN_RL(&rl, "bad 'ip' %s", b->ip);
+ return;
+ }
+
+ match_set_metadata(match_p, htonll(pb->datapath->tunnel_key));
+ match_set_reg(match_p, MFF_LOG_OUTPORT - MFF_REG0, pb->tunnel_key);
+ match_set_reg(match_p, 0, ntohl(ip));
+
+ ofpbuf_clear(ofpacts_p);
+ put_load(mac.ea, sizeof mac.ea, MFF_ETH_DST, 0, 48, ofpacts_p);
+
+ ofctrl_add_flow(OFTABLE_MAC_BINDING, 100, match_p, ofpacts_p,
+ &b->header_.uuid, is_new);
+}
+
/* Adds an OpenFlow flow to flow tables for each MAC binding in the OVN
* southbound database, using 'lports' to resolve logical port names to
* numbers. */
@@ -388,36 +471,24 @@ add_neighbor_flows(struct controller_ctx *ctx,
ofpbuf_init(&ofpacts, 0);
const struct sbrec_mac_binding *b;
- SBREC_MAC_BINDING_FOR_EACH (b, ctx->ovnsb_idl) {
- const struct sbrec_port_binding *pb
- = lport_lookup_by_name(lports, b->logical_port);
- if (!pb) {
- continue;
- }
-
- struct eth_addr mac;
- if (!eth_addr_from_string(b->mac, &mac)) {
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(5, 1);
- VLOG_WARN_RL(&rl, "bad 'mac' %s", b->mac);
- continue;
+ if (full_neighbor_flow_processing) {
+ SBREC_MAC_BINDING_FOR_EACH (b, ctx->ovnsb_idl) {
+ consider_neighbor_flow(lports, b, &ofpacts, &match, true);
}
-
- ovs_be32 ip;
- if (!ip_parse(b->ip, &ip)) {
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(5, 1);
- VLOG_WARN_RL(&rl, "bad 'ip' %s", b->ip);
- continue;
+ full_neighbor_flow_processing = false;
+ } else {
+ SBREC_MAC_BINDING_FOR_EACH_TRACKED (b, ctx->ovnsb_idl) {
+ bool is_deleted = sbrec_mac_binding_row_get_seqno(b,
+ OVSDB_IDL_CHANGE_DELETE) > 0;
+ bool is_new = sbrec_mac_binding_row_get_seqno(b,
+ OVSDB_IDL_CHANGE_MODIFY) == 0;
+
+ if (is_deleted) {
+ ofctrl_remove_flows(&b->header_.uuid);
+ continue;
+ }
+ consider_neighbor_flow(lports, b, &ofpacts, &match, is_new);
}
-
- match_set_metadata(&match, htonll(pb->datapath->tunnel_key));
- match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, pb->tunnel_key);
- match_set_reg(&match, 0, ntohl(ip));
-
- ofpbuf_clear(&ofpacts);
- put_load(mac.ea, sizeof mac.ea, MFF_ETH_DST, 0, 48, &ofpacts);
-
- ofctrl_add_flow(OFTABLE_MAC_BINDING, 100, &match, &ofpacts,
- &b->header_.uuid, true);
}
ofpbuf_uninit(&ofpacts);
}
@@ -431,6 +502,14 @@ lflow_run(struct controller_ctx *ctx, const struct lport_index *lports,
const struct hmap *patched_datapaths,
const struct simap *ct_zones)
{
+ if (full_flow_processing) {
+ ovn_flow_table_clear();
+ localvif_to_ofports_clear();
+ tunnels_clear();
+ full_logical_flow_processing = true;
+ full_neighbor_flow_processing = true;
+ full_flow_processing = false;
+ }
add_logical_flows(ctx, lports, mcgroups, local_datapaths,
patched_datapaths, ct_zones);
add_neighbor_flows(ctx, lports);
diff --git a/ovn/controller/lflow.h b/ovn/controller/lflow.h
index 8f8f81a..26cb5e0 100644
--- a/ovn/controller/lflow.h
+++ b/ovn/controller/lflow.h
@@ -65,5 +65,8 @@ void lflow_run(struct controller_ctx *, const struct lport_index *,
const struct hmap *patched_datapaths,
const struct simap *ct_zones);
void lflow_destroy(void);
+void reset_flow_processing(void);
+
+void reset_flow_processing(void);
#endif /* ovn/lflow.h */
diff --git a/ovn/controller/lport.c b/ovn/controller/lport.c
index 8210bc9..0674212 100644
--- a/ovn/controller/lport.c
+++ b/ovn/controller/lport.c
@@ -75,7 +75,7 @@ lport_index_clear(struct lport_index *lports)
hmap_remove(&lports->by_uuid, &port->uuid_node);
free(port);
}
- //reset_flow_processing();
+ reset_flow_processing();
}
static void
@@ -95,7 +95,7 @@ consider_lport_index(struct lport_index *lports,
uuid_hash(&pb->header_.uuid));
p->uuid = &pb->header_.uuid;
p->pb = pb;
- //reset_flow_processing();
+ reset_flow_processing();
}
void
@@ -115,7 +115,7 @@ lport_index_fill(struct lport_index *lports, struct ovsdb_idl *ovnsb_idl)
if (is_delete) {
lport_index_remove(lports, &pb->header_.uuid);
- //reset_flow_processing();
+ reset_flow_processing();
continue;
}
consider_lport_index(lports, pb);
@@ -213,7 +213,7 @@ mcgroup_index_clear(struct mcgroup_index *mcgroups)
hmap_remove(&mcgroups->by_uuid, &mcgroup->uuid_node);
free(mcgroup);
}
- //reset_flow_processing();
+ reset_flow_processing();
}
static void
@@ -232,7 +232,7 @@ consider_mcgroup_index(struct mcgroup_index *mcgroups,
uuid_hash(&mg->header_.uuid));
m->uuid = &mg->header_.uuid;
m->mg = mg;
- //reset_flow_processing();
+ reset_flow_processing();
}
void
@@ -252,7 +252,7 @@ mcgroup_index_fill(struct mcgroup_index *mcgroups, struct ovsdb_idl *ovnsb_idl)
if (is_delete) {
mcgroup_index_remove(mcgroups, &mg->header_.uuid);
- //reset_flow_processing();
+ reset_flow_processing();
continue;
}
consider_mcgroup_index(mcgroups, mg);
diff --git a/ovn/controller/ovn-controller.c b/ovn/controller/ovn-controller.c
index 5eba924..602ab09 100644
--- a/ovn/controller/ovn-controller.c
+++ b/ovn/controller/ovn-controller.c
@@ -363,7 +363,6 @@ main(int argc, char *argv[])
pinctrl_run(&ctx, &lports, br_int);
- ovn_flow_table_clear();
lflow_run(&ctx, &lports, &mcgroups, &local_datapaths,
&patched_datapaths, &ct_zones);
if (chassis_id) {
diff --git a/ovn/controller/patch.c b/ovn/controller/patch.c
index e07e829..2720f6c 100644
--- a/ovn/controller/patch.c
+++ b/ovn/controller/patch.c
@@ -15,6 +15,7 @@
#include <config.h>
+#include "lflow.h"
#include "patch.h"
#include "hash.h"
@@ -92,7 +93,7 @@ create_patch_port(struct controller_ctx *ctx,
ports[src->n_ports] = port;
ovsrec_bridge_verify_ports(src);
ovsrec_bridge_set_ports(src, ports, src->n_ports + 1);
-
+ reset_flow_processing();
free(ports);
}
@@ -125,6 +126,7 @@ remove_port(struct controller_ctx *ctx,
return;
}
}
+ reset_flow_processing();
}
/* Obtains external-ids:ovn-bridge-mappings from OVSDB and adds patch ports for
diff --git a/ovn/controller/physical.c b/ovn/controller/physical.c
index 8e39e40..3242b3e 100644
--- a/ovn/controller/physical.c
+++ b/ovn/controller/physical.c
@@ -48,8 +48,12 @@ physical_register_ovs_idl(struct ovsdb_idl *ovs_idl)
ovsdb_idl_add_column(ovs_idl, &ovsrec_interface_col_external_ids);
}
-struct uuid *hc_uuid = NULL; // uuid to identify OF flows not associated
- // with ovsdb rows.
+struct uuid *hc_uuid = NULL; // uuid to identify OF flows not associated with
+ // ovsdb rows.
+
+struct simap localvif_to_ofport = SIMAP_INITIALIZER(&localvif_to_ofport);
+struct hmap tunnels = HMAP_INITIALIZER(&tunnels);
+static bool full_binding_processing = false;
/* Maps from a chassis to the OpenFlow port number of the tunnel that can be
* used to reach that chassis. */
@@ -60,6 +64,28 @@ struct chassis_tunnel {
enum chassis_tunnel_type type;
};
+void
+localvif_to_ofports_clear(void)
+{
+ simap_clear(&localvif_to_ofport);
+}
+
+void
+tunnels_clear(void)
+{
+ struct chassis_tunnel *tun, *next;
+ HMAP_FOR_EACH_SAFE (tun, next, hmap_node, &tunnels) {
+ hmap_remove(&tunnels, &tun->hmap_node);
+ free(tun);
+ }
+}
+
+static void
+set_full_processing(void)
+{
+ full_binding_processing = true;
+}
+
static struct chassis_tunnel *
chassis_tunnel_find(struct hmap *tunnels, const char *chassis_id)
{
@@ -146,14 +172,417 @@ get_localnet_port(struct hmap *local_datapaths, int64_t tunnel_key)
return ld ? ld->localnet_port : NULL;
}
+
+static void
+consider_physical(enum mf_field_id mff_ovn_geneve,
+ const struct simap *ct_zones,
+ struct hmap *local_datapaths,
+ struct hmap *patched_datapaths,
+ const struct sbrec_port_binding *binding,
+ struct ofpbuf *ofpacts_p, bool is_new)
+{
+ /* Skip the port binding if the port is on a datapath that is neither
+ * local nor with any logical patch port connected, because local ports
+ * would never need to talk to those ports.
+ *
+ * Even with this approach there could still be unnecessary port
+ * bindings processed. A better approach would be a kind of "flood
+ * fill" algorithm:
+ *
+ * 1. Initialize set S to the logical datapaths that have a port
+ * located on the hypervisor.
+ *
+ * 2. For each patch port P in a logical datapath in S, add the
+ * logical datapath of the remote end of P to S. Iterate
+ * until S reaches a fixed point.
+ *
+ * This can be implemented in northd, which can generate the sets and
+ * save it on each port-binding record in SB, and ovn-controller can
+ * use the information directly. However, there can be update storms
+ * when a pair of patch ports are added/removed to connect/disconnect
+ * large lrouters and lswitches. This need to be studied further.
+ */
+ uint32_t dp_key = binding->datapath->tunnel_key;
+ uint32_t port_key = binding->tunnel_key;
+ if (!get_local_datapath(local_datapaths, dp_key)
+ && !get_patched_datapath(patched_datapaths, dp_key)) {
+ return;
+ }
+
+ /* Find the OpenFlow port for the logical port, as 'ofport'. This is
+ * one of:
+ *
+ * - If the port is a VIF on the chassis we're managing, the
+ * OpenFlow port for the VIF. 'tun' will be NULL.
+ *
+ * The same logic handles logical patch ports, as well as
+ * localnet patch ports.
+ *
+ * For a container nested inside a VM and accessible via a VLAN,
+ * 'tag' is the VLAN ID; otherwise 'tag' is 0.
+ *
+ * For a localnet patch port, if a VLAN ID was configured, 'tag'
+ * is set to that VLAN ID; otherwise 'tag' is 0.
+ *
+ * - If the port is on a remote chassis, the OpenFlow port for a
+ * tunnel to the VIF's remote chassis. 'tun' identifies that
+ * tunnel.
+ */
+
+ int tag = 0;
+ ofp_port_t ofport;
+ bool is_remote = false;
+ if (binding->parent_port && *binding->parent_port) {
+ if (!binding->tag) {
+ return;
+ }
+ ofport = u16_to_ofp(simap_get(&localvif_to_ofport,
+ binding->parent_port));
+ if (ofport) {
+ tag = *binding->tag;
+ }
+ } else {
+ ofport = u16_to_ofp(simap_get(&localvif_to_ofport,
+ binding->logical_port));
+ if (!strcmp(binding->type, "localnet") && ofport && binding->tag) {
+ tag = *binding->tag;
+ }
+ }
+
+ const struct chassis_tunnel *tun = NULL;
+ const struct sbrec_port_binding *localnet_port =
+ get_localnet_port(local_datapaths, dp_key);
+ if (!ofport) {
+ /* It is remote port, may be reached by tunnel or localnet port */
+ is_remote = true;
+ if (!binding->chassis) {
+ return;
+ }
+ if (localnet_port) {
+ ofport = u16_to_ofp(simap_get(&localvif_to_ofport,
+ localnet_port->logical_port));
+ if (!ofport) {
+ return;
+ }
+ } else {
+ tun = chassis_tunnel_find(&tunnels, binding->chassis->name);
+ if (!tun) {
+ return;
+ }
+ ofport = tun->ofport;
+ }
+ }
+
+ struct match match;
+ if (!is_remote) {
+ int zone_id = simap_get(ct_zones, binding->logical_port);
+ /* Packets that arrive from a vif can belong to a VM or
+ * to a container located inside that VM. Packets that
+ * arrive from containers have a tag (vlan) associated with them.
+ */
+
+ /* Table 0, Priority 150 and 100.
+ * ==============================
+ *
+ * Priority 150 is for tagged traffic. This may be containers in a
+ * VM or a VLAN on a local network. For such traffic, match on the
+ * tags and then strip the tag.
+ *
+ * Priority 100 is for traffic belonging to VMs or untagged locally
+ * connected networks.
+ *
+ * For both types of traffic: set MFF_LOG_INPORT to the logical
+ * input port, MFF_LOG_DATAPATH to the logical datapath, and
+ * resubmit into the logical ingress pipeline starting at table
+ * 16. */
+ ofpbuf_clear(ofpacts_p);
+ match_init_catchall(&match);
+ match_set_in_port(&match, ofport);
+
+ /* Match a VLAN tag and strip it, including stripping priority tags
+ * (e.g. VLAN ID 0). In the latter case we'll add a second flow
+ * for frames that lack any 802.1Q header later. */
+ if (tag || !strcmp(binding->type, "localnet")) {
+ match_set_dl_vlan(&match, htons(tag));
+ ofpact_put_STRIP_VLAN(ofpacts_p);
+ }
+
+ /* Remember the size with just strip vlan added so far,
+ * as we're going to remove this with ofpbuf_pull() later. */
+ uint32_t ofpacts_orig_size = ofpacts_p->size;
+
+ if (zone_id) {
+ put_load(zone_id, MFF_LOG_CT_ZONE, 0, 32, ofpacts_p);
+ }
+
+ /* Set MFF_LOG_DATAPATH and MFF_LOG_INPORT. */
+ put_load(dp_key, MFF_LOG_DATAPATH, 0, 64, ofpacts_p);
+ put_load(port_key, MFF_LOG_INPORT, 0, 32, ofpacts_p);
+
+ /* Resubmit to first logical ingress pipeline table. */
+ put_resubmit(OFTABLE_LOG_INGRESS_PIPELINE, ofpacts_p);
+ ofctrl_add_flow(OFTABLE_PHY_TO_LOG,
+ tag ? 150 : 100, &match, ofpacts_p,
+ &binding->header_.uuid, is_new);
+
+ if (!tag && !strcmp(binding->type, "localnet")) {
+ /* Add a second flow for frames that lack any 802.1Q
+ * header. For these, drop the OFPACT_STRIP_VLAN
+ * action. */
+ ofpbuf_pull(ofpacts_p, ofpacts_orig_size);
+ match_set_dl_tci_masked(&match, 0, htons(VLAN_CFI));
+ ofctrl_add_flow(0, 100, &match, ofpacts_p,
+ &binding->header_.uuid, is_new);
+ }
+
+ /* Table 33, priority 100.
+ * =======================
+ *
+ * Implements output to local hypervisor. Each flow matches a
+ * logical output port on the local hypervisor, and resubmits to
+ * table 34.
+ */
+
+ match_init_catchall(&match);
+ ofpbuf_clear(ofpacts_p);
+
+ /* Match MFF_LOG_DATAPATH, MFF_LOG_OUTPORT. */
+ match_set_metadata(&match, htonll(dp_key));
+ match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
+
+ if (zone_id) {
+ put_load(zone_id, MFF_LOG_CT_ZONE, 0, 32, ofpacts_p);
+ }
+
+ /* Resubmit to table 34. */
+ put_resubmit(OFTABLE_DROP_LOOPBACK, ofpacts_p);
+ ofctrl_add_flow(OFTABLE_LOCAL_OUTPUT,
+ 100, &match, ofpacts_p,
+ &binding->header_.uuid, is_new);
+
+ /* Table 34, Priority 100.
+ * =======================
+ *
+ * Drop packets whose logical inport and outport are the same. */
+ match_init_catchall(&match);
+ ofpbuf_clear(ofpacts_p);
+ match_set_metadata(&match, htonll(dp_key));
+ match_set_reg(&match, MFF_LOG_INPORT - MFF_REG0, port_key);
+ match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
+ ofctrl_add_flow(OFTABLE_DROP_LOOPBACK, 100, &match, ofpacts_p,
+ &binding->header_.uuid, is_new);
+
+ /* Table 64, Priority 100.
+ * =======================
+ *
+ * Deliver the packet to the local vif. */
+ match_init_catchall(&match);
+ ofpbuf_clear(ofpacts_p);
+ match_set_metadata(&match, htonll(dp_key));
+ match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
+ if (tag) {
+ /* For containers sitting behind a local vif, tag the packets
+ * before delivering them. */
+ struct ofpact_vlan_vid *vlan_vid;
+ vlan_vid = ofpact_put_SET_VLAN_VID(ofpacts_p);
+ vlan_vid->vlan_vid = tag;
+ vlan_vid->push_vlan_if_needed = true;
+
+ /* A packet might need to hair-pin back into its ingress
+ * OpenFlow port (to a different logical port, which we already
+ * checked back in table 34), so set the in_port to zero. */
+ put_stack(MFF_IN_PORT, ofpact_put_STACK_PUSH(ofpacts_p));
+ put_load(0, MFF_IN_PORT, 0, 16, ofpacts_p);
+ }
+ ofpact_put_OUTPUT(ofpacts_p)->port = ofport;
+ if (tag) {
+ /* Revert the tag added to the packets headed to containers
+ * in the previous step. If we don't do this, the packets
+ * that are to be broadcasted to a VM in the same logical
+ * switch will also contain the tag. Also revert the zero'd
+ * in_port. */
+ ofpact_put_STRIP_VLAN(ofpacts_p);
+ put_stack(MFF_IN_PORT, ofpact_put_STACK_POP(ofpacts_p));
+ }
+ ofctrl_add_flow(OFTABLE_LOG_TO_PHY, 100, &match, ofpacts_p,
+ &binding->header_.uuid, is_new);
+ } else if (!tun) {
+ /* Remote port connected by localnet port */
+ /* Table 33, priority 100.
+ * =======================
+ *
+ * Implements switching to localnet port. Each flow matches a
+ * logical output port on remote hypervisor, switch the output port
+ * to connected localnet port and resubmits to same table.
+ */
+
+ match_init_catchall(&match);
+ ofpbuf_clear(ofpacts_p);
+
+ /* Match MFF_LOG_DATAPATH, MFF_LOG_OUTPORT. */
+ match_set_metadata(&match, htonll(dp_key));
+ match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
+
+ put_load(localnet_port->tunnel_key, MFF_LOG_OUTPORT, 0, 32, ofpacts_p);
+
+ /* Resubmit to table 33. */
+ put_resubmit(OFTABLE_LOCAL_OUTPUT, ofpacts_p);
+ ofctrl_add_flow(OFTABLE_LOCAL_OUTPUT, 100, &match,
+ ofpacts_p, &binding->header_.uuid, is_new);
+ } else {
+ /* Remote port connected by tunnel */
+ /* Table 32, priority 100.
+ * =======================
+ *
+ * Implements output to remote hypervisors. Each flow matches an
+ * output port that includes a logical port on a remote hypervisor,
+ * and tunnels the packet to that hypervisor.
+ */
+
+ match_init_catchall(&match);
+ ofpbuf_clear(ofpacts_p);
+
+ /* Match MFF_LOG_DATAPATH, MFF_LOG_OUTPORT. */
+ match_set_metadata(&match, htonll(dp_key));
+ match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
+
+ put_encapsulation(mff_ovn_geneve, tun, binding->datapath,
+ port_key, ofpacts_p);
+
+ /* Output to tunnel. */
+ ofpact_put_OUTPUT(ofpacts_p)->port = ofport;
+ ofctrl_add_flow(OFTABLE_REMOTE_OUTPUT, 100, &match, ofpacts_p,
+ &binding->header_.uuid, is_new);
+ }
+}
+
+static void
+consider_mc_group(enum mf_field_id mff_ovn_geneve,
+ const struct simap *ct_zones,
+ struct hmap *local_datapaths,
+ const struct sbrec_multicast_group *mc,
+ struct ofpbuf *ofpacts_p,
+ struct ofpbuf *remote_ofpacts_p, bool is_new)
+{
+ struct sset remote_chassis = SSET_INITIALIZER(&remote_chassis);
+ struct match match;
+
+ match_init_catchall(&match);
+ match_set_metadata(&match, htonll(mc->datapath->tunnel_key));
+ match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, mc->tunnel_key);
+
+ /* Go through all of the ports in the multicast group:
+ *
+ * - For remote ports, add the chassis to 'remote_chassis'.
+ *
+ * - For local ports (other than logical patch ports), add actions
+ * to 'ofpacts' to set the output port and resubmit.
+ *
+ * - For logical patch ports, add actions to 'remote_ofpacts'
+ * instead. (If we put them in 'ofpacts', then the output
+ * would happen on every hypervisor in the multicast group,
+ * effectively duplicating the packet.)
+ */
+ ofpbuf_clear(ofpacts_p);
+ ofpbuf_clear(remote_ofpacts_p);
+ for (size_t i = 0; i < mc->n_ports; i++) {
+ struct sbrec_port_binding *port = mc->ports[i];
+
+ if (port->datapath != mc->datapath) {
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(5, 1);
+ VLOG_WARN_RL(&rl, UUID_FMT": multicast group contains ports "
+ "in wrong datapath",
+ UUID_ARGS(&mc->header_.uuid));
+ return;
+ }
+
+ int zone_id = simap_get(ct_zones, port->logical_port);
+ if (zone_id) {
+ put_load(zone_id, MFF_LOG_CT_ZONE, 0, 32, ofpacts_p);
+ }
+
+ if (!strcmp(port->type, "patch")) {
+ put_load(port->tunnel_key, MFF_LOG_OUTPORT, 0, 32,
+ remote_ofpacts_p);
+ put_resubmit(OFTABLE_DROP_LOOPBACK, remote_ofpacts_p);
+ } else if (simap_contains(&localvif_to_ofport,
+ (port->parent_port && *port->parent_port)
+ ? port->parent_port : port->logical_port)) {
+ put_load(port->tunnel_key, MFF_LOG_OUTPORT, 0, 32, ofpacts_p);
+ put_resubmit(OFTABLE_DROP_LOOPBACK, ofpacts_p);
+ } else if (port->chassis && !get_localnet_port(local_datapaths,
+ mc->datapath->tunnel_key)) {
+ /* Add remote chassis only when localnet port not exist,
+ * otherwise multicast will reach remote ports through localnet
+ * port. */
+ sset_add(&remote_chassis, port->chassis->name);
+ }
+ }
+
+ /* Table 33, priority 100.
+ * =======================
+ *
+ * Handle output to the local logical ports in the multicast group, if
+ * any. */
+ bool local_ports = ofpacts_p->size > 0;
+ if (local_ports) {
+ /* Following delivery to local logical ports, restore the multicast
+ * group as the logical output port. */
+ put_load(mc->tunnel_key, MFF_LOG_OUTPORT, 0, 32, ofpacts_p);
+
+ ofctrl_add_flow(OFTABLE_LOCAL_OUTPUT,
+ 100, &match, ofpacts_p,
+ &mc->header_.uuid, is_new);
+ }
+
+ /* Table 32, priority 100.
+ * =======================
+ *
+ * Handle output to the remote chassis in the multicast group, if
+ * any. */
+ if (!sset_is_empty(&remote_chassis) || remote_ofpacts_p->size > 0) {
+ if (remote_ofpacts_p->size > 0) {
+ /* Following delivery to logical patch ports, restore the
+ * multicast group as the logical output port. */
+ put_load(mc->tunnel_key, MFF_LOG_OUTPORT, 0, 32,
+ remote_ofpacts_p);
+ }
+
+ const char *chassis;
+ const struct chassis_tunnel *prev = NULL;
+ SSET_FOR_EACH (chassis, &remote_chassis) {
+ const struct chassis_tunnel *tun
+ = chassis_tunnel_find(&tunnels, chassis);
+ if (!tun) {
+ return;
+ }
+
+ if (!prev || tun->type != prev->type) {
+ put_encapsulation(mff_ovn_geneve, tun, mc->datapath,
+ mc->tunnel_key, remote_ofpacts_p);
+ prev = tun;
+ }
+ ofpact_put_OUTPUT(remote_ofpacts_p)->port = tun->ofport;
+ }
+
+ if (remote_ofpacts_p->size) {
+ if (local_ports) {
+ put_resubmit(OFTABLE_LOCAL_OUTPUT, remote_ofpacts_p);
+ }
+ ofctrl_add_flow(OFTABLE_REMOTE_OUTPUT, 100,
+ &match, remote_ofpacts_p,
+ &mc->header_.uuid, is_new);
+ }
+ }
+ sset_destroy(&remote_chassis);
+}
+
void
physical_run(struct controller_ctx *ctx, enum mf_field_id mff_ovn_geneve,
const struct ovsrec_bridge *br_int, const char *this_chassis_id,
const struct simap *ct_zones,
struct hmap *local_datapaths, struct hmap *patched_datapaths)
{
- struct simap localvif_to_ofport = SIMAP_INITIALIZER(&localvif_to_ofport);
- struct hmap tunnels = HMAP_INITIALIZER(&tunnels);
if (!hc_uuid) {
hc_uuid = xmalloc(sizeof(struct uuid));
uuid_generate(hc_uuid);
@@ -193,11 +622,21 @@ physical_run(struct controller_ctx *ctx, enum mf_field_id mff_ovn_geneve,
bool is_patch = !strcmp(iface_rec->type, "patch");
if (is_patch && localnet) {
/* localnet patch ports can be handled just like VIFs. */
- simap_put(&localvif_to_ofport, localnet, ofport);
+ struct simap_node *old = simap_find(&localvif_to_ofport,
+ localnet);
+ if (!old || old->data != ofport) {
+ simap_put(&localvif_to_ofport, localnet, ofport);
+ set_full_processing();
+ }
break;
} else if (is_patch && logpatch) {
/* Logical patch ports can be handled just like VIFs. */
- simap_put(&localvif_to_ofport, logpatch, ofport);
+ struct simap_node *old = simap_find(&localvif_to_ofport,
+ logpatch);
+ if (!old || old->data != ofport) {
+ simap_put(&localvif_to_ofport, logpatch, ofport);
+ set_full_processing();
+ }
break;
} else if (chassis_id) {
enum chassis_tunnel_type tunnel_type;
@@ -214,18 +653,38 @@ physical_run(struct controller_ctx *ctx, enum mf_field_id mff_ovn_geneve,
continue;
}
- struct chassis_tunnel *tun = xmalloc(sizeof *tun);
- hmap_insert(&tunnels, &tun->hmap_node,
- hash_string(chassis_id, 0));
- tun->chassis_id = chassis_id;
- tun->ofport = u16_to_ofp(ofport);
- tun->type = tunnel_type;
+ struct chassis_tunnel *old = chassis_tunnel_find(&tunnels,
+ chassis_id);
+ if (!old) {
+ struct chassis_tunnel *tun = xmalloc(sizeof *tun);
+ hmap_insert(&tunnels, &tun->hmap_node,
+ hash_string(chassis_id, 0));
+ tun->chassis_id = chassis_id;
+ tun->ofport = u16_to_ofp(ofport);
+ tun->type = tunnel_type;
+ set_full_processing();
+ } else {
+ ofp_port_t new_port = u16_to_ofp(ofport);
+ if (new_port != old->ofport) {
+ old->ofport = new_port;
+ set_full_processing();
+ }
+ if (tunnel_type != old->type) {
+ old->type = tunnel_type;
+ set_full_processing();
+ }
+ }
break;
} else {
const char *iface_id = smap_get(&iface_rec->external_ids,
"iface-id");
if (iface_id) {
- simap_put(&localvif_to_ofport, iface_id, ofport);
+ struct simap_node *old = simap_find(&localvif_to_ofport,
+ iface_id);
+ if (!old || old->data != ofport) {
+ simap_put(&localvif_to_ofport, iface_id, ofport);
+ set_full_processing();
+ }
}
}
}
@@ -237,396 +696,35 @@ physical_run(struct controller_ctx *ctx, enum mf_field_id mff_ovn_geneve,
/* Set up flows in table 0 for physical-to-logical translation and in table
* 64 for logical-to-physical translation. */
const struct sbrec_port_binding *binding;
- SBREC_PORT_BINDING_FOR_EACH (binding, ctx->ovnsb_idl) {
- /* Skip the port binding if the port is on a datapath that is neither
- * local nor with any logical patch port connected, because local ports
- * would never need to talk to those ports.
- *
- * Even with this approach there could still be unnecessary port
- * bindings processed. A better approach would be a kind of "flood
- * fill" algorithm:
- *
- * 1. Initialize set S to the logical datapaths that have a port
- * located on the hypervisor.
- *
- * 2. For each patch port P in a logical datapath in S, add the
- * logical datapath of the remote end of P to S. Iterate
- * until S reaches a fixed point.
- *
- * This can be implemented in northd, which can generate the sets and
- * save it on each port-binding record in SB, and ovn-controller can
- * use the information directly. However, there can be update storms
- * when a pair of patch ports are added/removed to connect/disconnect
- * large lrouters and lswitches. This need to be studied further.
- */
- uint32_t dp_key = binding->datapath->tunnel_key;
- uint32_t port_key = binding->tunnel_key;
- if (!get_local_datapath(local_datapaths, dp_key)
- && !get_patched_datapath(patched_datapaths, dp_key)) {
- continue;
+ if (full_binding_processing) {
+ SBREC_PORT_BINDING_FOR_EACH (binding, ctx->ovnsb_idl) {
+ consider_physical(mff_ovn_geneve, ct_zones, local_datapaths,
+ patched_datapaths, binding, &ofpacts, true);
}
-
- /* Find the OpenFlow port for the logical port, as 'ofport'. This is
- * one of:
- *
- * - If the port is a VIF on the chassis we're managing, the
- * OpenFlow port for the VIF. 'tun' will be NULL.
- *
- * The same logic handles logical patch ports, as well as
- * localnet patch ports.
- *
- * For a container nested inside a VM and accessible via a VLAN,
- * 'tag' is the VLAN ID; otherwise 'tag' is 0.
- *
- * For a localnet patch port, if a VLAN ID was configured, 'tag'
- * is set to that VLAN ID; otherwise 'tag' is 0.
- *
- * - If the port is on a remote chassis, the OpenFlow port for a
- * tunnel to the VIF's remote chassis. 'tun' identifies that
- * tunnel.
- */
-
- int tag = 0;
- ofp_port_t ofport;
- bool is_remote = false;
- if (binding->parent_port && *binding->parent_port) {
- if (!binding->tag) {
+ full_binding_processing = false;
+ } else {
+ SBREC_PORT_BINDING_FOR_EACH_TRACKED (binding, ctx->ovnsb_idl) {
+ bool is_deleted = sbrec_port_binding_row_get_seqno(binding,
+ OVSDB_IDL_CHANGE_DELETE) > 0;
+ bool is_new = sbrec_port_binding_row_get_seqno(binding,
+ OVSDB_IDL_CHANGE_MODIFY) == 0;
+
+ if (is_deleted) {
+ ofctrl_remove_flows(&binding->header_.uuid);
continue;
}
- ofport = u16_to_ofp(simap_get(&localvif_to_ofport,
- binding->parent_port));
- if (ofport) {
- tag = *binding->tag;
- }
- } else {
- ofport = u16_to_ofp(simap_get(&localvif_to_ofport,
- binding->logical_port));
- if (!strcmp(binding->type, "localnet") && ofport && binding->tag) {
- tag = *binding->tag;
- }
- }
- const struct chassis_tunnel *tun = NULL;
- const struct sbrec_port_binding *localnet_port =
- get_localnet_port(local_datapaths, dp_key);
- if (!ofport) {
- /* It is remote port, may be reached by tunnel or localnet port */
- is_remote = true;
- if (!binding->chassis) {
- continue;
- }
- if (localnet_port) {
- ofport = u16_to_ofp(simap_get(&localvif_to_ofport,
- localnet_port->logical_port));
- if (!ofport) {
- continue;
- }
- } else {
- tun = chassis_tunnel_find(&tunnels, binding->chassis->name);
- if (!tun) {
- continue;
- }
- ofport = tun->ofport;
- }
- }
-
- struct match match;
- if (!is_remote) {
- int zone_id = simap_get(ct_zones, binding->logical_port);
- /* Packets that arrive from a vif can belong to a VM or
- * to a container located inside that VM. Packets that
- * arrive from containers have a tag (vlan) associated with them.
- */
-
- /* Table 0, Priority 150 and 100.
- * ==============================
- *
- * Priority 150 is for tagged traffic. This may be containers in a
- * VM or a VLAN on a local network. For such traffic, match on the
- * tags and then strip the tag.
- *
- * Priority 100 is for traffic belonging to VMs or untagged locally
- * connected networks.
- *
- * For both types of traffic: set MFF_LOG_INPORT to the logical
- * input port, MFF_LOG_DATAPATH to the logical datapath, and
- * resubmit into the logical ingress pipeline starting at table
- * 16. */
- ofpbuf_clear(&ofpacts);
- match_init_catchall(&match);
- match_set_in_port(&match, ofport);
-
- /* Match a VLAN tag and strip it, including stripping priority tags
- * (e.g. VLAN ID 0). In the latter case we'll add a second flow
- * for frames that lack any 802.1Q header later. */
- if (tag || !strcmp(binding->type, "localnet")) {
- match_set_dl_vlan(&match, htons(tag));
- ofpact_put_STRIP_VLAN(&ofpacts);
- }
-
- /* Remember the size with just strip vlan added so far,
- * as we're going to remove this with ofpbuf_pull() later. */
- uint32_t ofpacts_orig_size = ofpacts.size;
-
- if (zone_id) {
- put_load(zone_id, MFF_LOG_CT_ZONE, 0, 32, &ofpacts);
- }
-
- /* Set MFF_LOG_DATAPATH and MFF_LOG_INPORT. */
- put_load(dp_key, MFF_LOG_DATAPATH, 0, 64, &ofpacts);
- put_load(port_key, MFF_LOG_INPORT, 0, 32, &ofpacts);
-
- /* Resubmit to first logical ingress pipeline table. */
- put_resubmit(OFTABLE_LOG_INGRESS_PIPELINE, &ofpacts);
- ofctrl_add_flow(OFTABLE_PHY_TO_LOG,
- tag ? 150 : 100, &match, &ofpacts,
- &binding->header_.uuid, true);
-
- if (!tag && !strcmp(binding->type, "localnet")) {
- /* Add a second flow for frames that lack any 802.1Q
- * header. For these, drop the OFPACT_STRIP_VLAN
- * action. */
- ofpbuf_pull(&ofpacts, ofpacts_orig_size);
- match_set_dl_tci_masked(&match, 0, htons(VLAN_CFI));
- ofctrl_add_flow(0, 100, &match, &ofpacts,
- &binding->header_.uuid, true);
- }
-
- /* Table 33, priority 100.
- * =======================
- *
- * Implements output to local hypervisor. Each flow matches a
- * logical output port on the local hypervisor, and resubmits to
- * table 34.
- */
-
- match_init_catchall(&match);
- ofpbuf_clear(&ofpacts);
-
- /* Match MFF_LOG_DATAPATH, MFF_LOG_OUTPORT. */
- match_set_metadata(&match, htonll(dp_key));
- match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
-
- if (zone_id) {
- put_load(zone_id, MFF_LOG_CT_ZONE, 0, 32, &ofpacts);
- }
-
- /* Resubmit to table 34. */
- put_resubmit(OFTABLE_DROP_LOOPBACK, &ofpacts);
- ofctrl_add_flow(OFTABLE_LOCAL_OUTPUT, 100, &match, &ofpacts,
- &binding->header_.uuid, true);
-
- /* Table 34, Priority 100.
- * =======================
- *
- * Drop packets whose logical inport and outport are the same. */
- match_init_catchall(&match);
- ofpbuf_clear(&ofpacts);
- match_set_metadata(&match, htonll(dp_key));
- match_set_reg(&match, MFF_LOG_INPORT - MFF_REG0, port_key);
- match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
- ofctrl_add_flow(OFTABLE_DROP_LOOPBACK, 100, &match, &ofpacts,
- &binding->header_.uuid, true);
-
- /* Table 64, Priority 100.
- * =======================
- *
- * Deliver the packet to the local vif. */
- match_init_catchall(&match);
- ofpbuf_clear(&ofpacts);
- match_set_metadata(&match, htonll(dp_key));
- match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
- if (tag) {
- /* For containers sitting behind a local vif, tag the packets
- * before delivering them. */
- struct ofpact_vlan_vid *vlan_vid;
- vlan_vid = ofpact_put_SET_VLAN_VID(&ofpacts);
- vlan_vid->vlan_vid = tag;
- vlan_vid->push_vlan_if_needed = true;
-
- /* A packet might need to hair-pin back into its ingress
- * OpenFlow port (to a different logical port, which we already
- * checked back in table 34), so set the in_port to zero. */
- put_stack(MFF_IN_PORT, ofpact_put_STACK_PUSH(&ofpacts));
- put_load(0, MFF_IN_PORT, 0, 16, &ofpacts);
- }
- ofpact_put_OUTPUT(&ofpacts)->port = ofport;
- if (tag) {
- /* Revert the tag added to the packets headed to containers
- * in the previous step. If we don't do this, the packets
- * that are to be broadcasted to a VM in the same logical
- * switch will also contain the tag. Also revert the zero'd
- * in_port. */
- ofpact_put_STRIP_VLAN(&ofpacts);
- put_stack(MFF_IN_PORT, ofpact_put_STACK_POP(&ofpacts));
- }
- ofctrl_add_flow(OFTABLE_LOG_TO_PHY, 100, &match, &ofpacts,
- &binding->header_.uuid, true);
- } else if (!tun) {
- /* Remote port connected by localnet port */
- /* Table 33, priority 100.
- * =======================
- *
- * Implements switching to localnet port. Each flow matches a
- * logical output port on remote hypervisor, switch the output port
- * to connected localnet port and resubmits to same table.
- */
-
- match_init_catchall(&match);
- ofpbuf_clear(&ofpacts);
-
- /* Match MFF_LOG_DATAPATH, MFF_LOG_OUTPORT. */
- match_set_metadata(&match, htonll(dp_key));
- match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
-
- put_load(localnet_port->tunnel_key, MFF_LOG_OUTPORT, 0, 32, &ofpacts);
-
- /* Resubmit to table 33. */
- put_resubmit(OFTABLE_LOCAL_OUTPUT, &ofpacts);
- ofctrl_add_flow(OFTABLE_LOCAL_OUTPUT, 100, &match, &ofpacts,
- &binding->header_.uuid, true);
- } else {
- /* Remote port connected by tunnel */
- /* Table 32, priority 100.
- * =======================
- *
- * Implements output to remote hypervisors. Each flow matches an
- * output port that includes a logical port on a remote hypervisor,
- * and tunnels the packet to that hypervisor.
- */
-
- match_init_catchall(&match);
- ofpbuf_clear(&ofpacts);
-
- /* Match MFF_LOG_DATAPATH, MFF_LOG_OUTPORT. */
- match_set_metadata(&match, htonll(dp_key));
- match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, port_key);
-
- put_encapsulation(mff_ovn_geneve, tun, binding->datapath,
- port_key, &ofpacts);
-
- /* Output to tunnel. */
- ofpact_put_OUTPUT(&ofpacts)->port = ofport;
- ofctrl_add_flow(OFTABLE_REMOTE_OUTPUT, 100, &match, &ofpacts,
- &binding->header_.uuid, true);
+ consider_physical(mff_ovn_geneve, ct_zones, local_datapaths,
+ patched_datapaths, binding, &ofpacts, is_new);
}
}
-
/* Handle output to multicast groups, in tables 32 and 33. */
const struct sbrec_multicast_group *mc;
struct ofpbuf remote_ofpacts;
ofpbuf_init(&remote_ofpacts, 0);
SBREC_MULTICAST_GROUP_FOR_EACH (mc, ctx->ovnsb_idl) {
- struct sset remote_chassis = SSET_INITIALIZER(&remote_chassis);
- struct match match;
-
- match_init_catchall(&match);
- match_set_metadata(&match, htonll(mc->datapath->tunnel_key));
- match_set_reg(&match, MFF_LOG_OUTPORT - MFF_REG0, mc->tunnel_key);
-
- /* Go through all of the ports in the multicast group:
- *
- * - For remote ports, add the chassis to 'remote_chassis'.
- *
- * - For local ports (other than logical patch ports), add actions
- * to 'ofpacts' to set the output port and resubmit.
- *
- * - For logical patch ports, add actions to 'remote_ofpacts'
- * instead. (If we put them in 'ofpacts', then the output
- * would happen on every hypervisor in the multicast group,
- * effectively duplicating the packet.)
- */
- ofpbuf_clear(&ofpacts);
- ofpbuf_clear(&remote_ofpacts);
- for (size_t i = 0; i < mc->n_ports; i++) {
- struct sbrec_port_binding *port = mc->ports[i];
-
- if (port->datapath != mc->datapath) {
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(5, 1);
- VLOG_WARN_RL(&rl, UUID_FMT": multicast group contains ports "
- "in wrong datapath",
- UUID_ARGS(&mc->header_.uuid));
- continue;
- }
-
- int zone_id = simap_get(ct_zones, port->logical_port);
- if (zone_id) {
- put_load(zone_id, MFF_LOG_CT_ZONE, 0, 32, &ofpacts);
- }
-
- if (!strcmp(port->type, "patch")) {
- put_load(port->tunnel_key, MFF_LOG_OUTPORT, 0, 32,
- &remote_ofpacts);
- put_resubmit(OFTABLE_DROP_LOOPBACK, &remote_ofpacts);
- } else if (simap_contains(&localvif_to_ofport,
- (port->parent_port && *port->parent_port)
- ? port->parent_port : port->logical_port)) {
- put_load(port->tunnel_key, MFF_LOG_OUTPORT, 0, 32, &ofpacts);
- put_resubmit(OFTABLE_DROP_LOOPBACK, &ofpacts);
- } else if (port->chassis && !get_localnet_port(local_datapaths,
- mc->datapath->tunnel_key)) {
- /* Add remote chassis only when localnet port not exist,
- * otherwise multicast will reach remote ports through localnet
- * port. */
- sset_add(&remote_chassis, port->chassis->name);
- }
- }
-
- /* Table 33, priority 100.
- * =======================
- *
- * Handle output to the local logical ports in the multicast group, if
- * any. */
- bool local_ports = ofpacts.size > 0;
- if (local_ports) {
- /* Following delivery to local logical ports, restore the multicast
- * group as the logical output port. */
- put_load(mc->tunnel_key, MFF_LOG_OUTPORT, 0, 32, &ofpacts);
-
- ofctrl_add_flow(OFTABLE_LOCAL_OUTPUT, 100, &match, &ofpacts,
- &mc->header_.uuid, true);
- }
-
- /* Table 32, priority 100.
- * =======================
- *
- * Handle output to the remote chassis in the multicast group, if
- * any. */
- if (!sset_is_empty(&remote_chassis) || remote_ofpacts.size > 0) {
- if (remote_ofpacts.size > 0) {
- /* Following delivery to logical patch ports, restore the
- * multicast group as the logical output port. */
- put_load(mc->tunnel_key, MFF_LOG_OUTPORT, 0, 32,
- &remote_ofpacts);
- }
-
- const char *chassis;
- const struct chassis_tunnel *prev = NULL;
- SSET_FOR_EACH (chassis, &remote_chassis) {
- const struct chassis_tunnel *tun
- = chassis_tunnel_find(&tunnels, chassis);
- if (!tun) {
- continue;
- }
-
- if (!prev || tun->type != prev->type) {
- put_encapsulation(mff_ovn_geneve, tun, mc->datapath,
- mc->tunnel_key, &remote_ofpacts);
- prev = tun;
- }
- ofpact_put_OUTPUT(&remote_ofpacts)->port = tun->ofport;
- }
-
- if (remote_ofpacts.size) {
- if (local_ports) {
- put_resubmit(OFTABLE_LOCAL_OUTPUT, &remote_ofpacts);
- }
- ofctrl_add_flow(OFTABLE_REMOTE_OUTPUT, 100,
- &match, &remote_ofpacts,
- &mc->header_.uuid, true);
- }
- }
- sset_destroy(&remote_chassis);
+ consider_mc_group(mff_ovn_geneve, ct_zones, local_datapaths,
+ mc, &ofpacts, &remote_ofpacts, true);
}
ofpbuf_uninit(&remote_ofpacts);
@@ -687,17 +785,20 @@ physical_run(struct controller_ctx *ctx, enum mf_field_id mff_ovn_geneve,
SBREC_PORT_BINDING_FOR_EACH (binding, ctx->ovnsb_idl) {
struct match match = MATCH_CATCHALL_INITIALIZER;
+ uint32_t dp_key = binding->datapath->tunnel_key;
+ uint32_t port_key = binding->tunnel_key;
+
if (!binding->chassis ||
strcmp(tun->chassis_id, binding->chassis->name)) {
continue;
}
match_set_in_port(&match, tun->ofport);
- match_set_tun_id(&match, htonll(binding->datapath->tunnel_key));
+ match_set_tun_id(&match, htonll(dp_key));
ofpbuf_clear(&ofpacts);
put_move(MFF_TUN_ID, 0, MFF_LOG_DATAPATH, 0, 24, &ofpacts);
- put_load(binding->tunnel_key, MFF_LOG_INPORT, 0, 15, &ofpacts);
+ put_load(port_key, MFF_LOG_INPORT, 0, 15, &ofpacts);
put_resubmit(OFTABLE_LOG_INGRESS_PIPELINE, &ofpacts);
ofctrl_add_flow(OFTABLE_PHY_TO_LOG, 100, &match, &ofpacts,
@@ -733,11 +834,4 @@ physical_run(struct controller_ctx *ctx, enum mf_field_id mff_ovn_geneve,
hc_uuid, true);
ofpbuf_uninit(&ofpacts);
- simap_destroy(&localvif_to_ofport);
- struct chassis_tunnel *tun_next;
- HMAP_FOR_EACH_SAFE (tun, tun_next, hmap_node, &tunnels) {
- hmap_remove(&tunnels, &tun->hmap_node);
- free(tun);
- }
- hmap_destroy(&tunnels);
}
diff --git a/ovn/controller/physical.h b/ovn/controller/physical.h
index 658b49f..be9fb74 100644
--- a/ovn/controller/physical.h
+++ b/ovn/controller/physical.h
@@ -45,5 +45,7 @@ void physical_run(struct controller_ctx *, enum mf_field_id mff_ovn_geneve,
const struct ovsrec_bridge *br_int, const char *chassis_id,
const struct simap *ct_zones,
struct hmap *local_datapaths, struct hmap *patched_datapaths);
+void localvif_to_ofports_clear(void);
+void tunnels_clear(void);
#endif /* ovn/physical.h */
--
1.9.1
More information about the dev
mailing list