-/* Copyright (c) 2009, 2010, 2011, 2012, 2013, 2014 Nicira, Inc.
+/* Copyright (c) 2009, 2010, 2011, 2012, 2013, 2014, 2015 Nicira, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
#include "cfm.h"
#include "connmgr.h"
#include "coverage.h"
+#include "dp-packet.h"
#include "dpif.h"
#include "dynamic-string.h"
#include "in-band.h"
#include "lacp.h"
#include "learn.h"
#include "list.h"
+#include "ovs-lldp.h"
#include "mac-learning.h"
#include "mcast-snooping.h"
#include "meta-flow.h"
#include "ofproto/ofproto-dpif-sflow.h"
#include "ofproto/ofproto-dpif.h"
#include "ofproto/ofproto-provider.h"
-#include "packet-dpif.h"
#include "ovs-router.h"
#include "tnl-ports.h"
#include "tunnel.h"
-#include "vlog.h"
+#include "openvswitch/vlog.h"
COVERAGE_DEFINE(xlate_actions);
COVERAGE_DEFINE(xlate_actions_oversize);
COVERAGE_DEFINE(xlate_actions_too_many_output);
-COVERAGE_DEFINE(xlate_actions_mpls_overflow);
VLOG_DEFINE_THIS_MODULE(ofproto_dpif_xlate);
#define MAX_INTERNAL_RESUBMITS 1 /* Max resbmits allowed using rules in
internal table. */
-/* Timeout for internal rules created to handle recirculation */
-#define RECIRC_TIMEOUT 60
-
/* Maximum number of resubmit actions in a flow translation, whether they are
* recursive or not. */
#define MAX_RESUBMITS (MAX_RESUBMIT_RECURSION * MAX_RESUBMIT_RECURSION)
struct hmap_node hmap_node; /* Node in global 'xbridges' map. */
struct ofproto_dpif *ofproto; /* Key in global 'xbridges' map. */
- struct list xbundles; /* Owned xbundles. */
+ struct ovs_list xbundles; /* Owned xbundles. */
struct hmap xports; /* Indexed by ofp_port. */
char *name; /* Name used in log messages. */
bool has_in_band; /* Bridge has in band control? */
bool forward_bpdu; /* Bridge forwards STP BPDUs? */
- /* True if the datapath supports recirculation. */
- bool enable_recirc;
-
- /* True if the datapath supports variable-length
- * OVS_USERSPACE_ATTR_USERDATA in OVS_ACTION_ATTR_USERSPACE actions.
- * False if the datapath supports only 8-byte (or shorter) userdata. */
- bool variable_length_userdata;
-
- /* Number of MPLS label stack entries that the datapath supports
- * in matches. */
- size_t max_mpls_depth;
-
- /* True if the datapath supports masked data in OVS_ACTION_ATTR_SET
- * actions. */
- bool masked_set_action;
+ /* Datapath feature support. */
+ struct dpif_backer_support support;
};
struct xbundle {
struct hmap_node hmap_node; /* In global 'xbundles' map. */
struct ofbundle *ofbundle; /* Key in global 'xbundles' map. */
- struct list list_node; /* In parent 'xbridges' list. */
+ struct ovs_list list_node; /* In parent 'xbridges' list. */
struct xbridge *xbridge; /* Parent xbridge. */
- struct list xports; /* Contains "struct xport"s. */
+ struct ovs_list xports; /* Contains "struct xport"s. */
char *name; /* Name used in log messages. */
struct bond *bond; /* Nonnull iff more than one port. */
odp_port_t odp_port; /* Datapath port number or ODPP_NONE. */
- struct list bundle_node; /* In parent xbundle (if it exists). */
+ struct ovs_list bundle_node; /* In parent xbundle (if it exists). */
struct xbundle *xbundle; /* Parent xbundle or null. */
struct netdev *netdev; /* 'ofport''s netdev. */
struct cfm *cfm; /* CFM handle or null. */
struct bfd *bfd; /* BFD handle or null. */
+ struct lldp *lldp; /* LLDP handle or null. */
};
struct xlate_ctx {
const struct xbridge *xbridge;
+ /* Flow tables version at the beginning of the translation. */
+ cls_version_t tables_version;
+
/* Flow at the last commit. */
struct flow base_flow;
int recurse; /* Current resubmit nesting depth. */
int resubmits; /* Total number of resubmits. */
bool in_group; /* Currently translating ofgroup, if true. */
+ bool in_action_set; /* Currently translating action_set, if true. */
- uint32_t orig_skb_priority; /* Priority when packet arrived. */
uint8_t table_id; /* OpenFlow table ID where flow was found. */
+ ovs_be64 rule_cookie; /* Cookie of the rule being translated. */
+ uint32_t orig_skb_priority; /* Priority when packet arrived. */
uint32_t sflow_n_outputs; /* Number of output ports. */
odp_port_t sflow_odp_port; /* Output port for composing sFlow action. */
uint16_t user_cookie_offset;/* Used for user_action_cookie fixup. */
bool exit; /* No further actions should be processed. */
- bool use_recirc; /* Should generate recirc? */
- struct xlate_recirc recirc; /* Information used for generating
- * recirculation actions */
+ /* These are used for non-bond recirculation. The recirculation IDs are
+ * stored in xout and must be associated with a datapath flow (ukey),
+ * otherwise they will be freed when the xout is uninitialized.
+ *
+ *
+ * Steps in Recirculation Translation
+ * ==================================
+ *
+ * At some point during translation, the code recognizes the need for
+ * recirculation. For example, recirculation is necessary when, after
+ * popping the last MPLS label, an action or a match tries to examine or
+ * modify a field that has been newly revealed following the MPLS label.
+ *
+ * The simplest part of the work to be done is to commit existing changes to
+ * the packet, which produces datapath actions corresponding to the changes,
+ * and after this, add an OVS_ACTION_ATTR_RECIRC datapath action.
+ *
+ * The main problem here is preserving state. When the datapath executes
+ * OVS_ACTION_ATTR_RECIRC, it will upcall to userspace to get a translation
+ * for the post-recirculation actions. At this point userspace has to
+ * resume the translation where it left off, which means that it has to
+ * execute the following:
+ *
+ * - The action that prompted recirculation, and any actions following
+ * it within the same flow.
+ *
+ * - If the action that prompted recirculation was invoked within a
+ * NXAST_RESUBMIT, then any actions following the resubmit. These
+ * "resubmit"s can be nested, so this has to go all the way up the
+ * control stack.
+ *
+ * - The OpenFlow 1.1+ action set.
+ *
+ * State that actions and flow table lookups can depend on, such as the
+ * following, must also be preserved:
+ *
+ * - Metadata fields (input port, registers, OF1.1+ metadata, ...).
+ *
+ * - Action set, stack
+ *
+ * - The table ID and cookie of the flow being translated at each level
+ * of the control stack (since OFPAT_CONTROLLER actions send these to
+ * the controller).
+ *
+ * Translation allows for the control of this state preservation via these
+ * members. When a need for recirculation is identified, the translation
+ * process:
+ *
+ * 1. Sets 'recirc_action_offset' to the current size of 'action_set'. The
+ * action set is part of what needs to be preserved, so this allows the
+ * action set and the additional state to share the 'action_set' buffer.
+ * Later steps can tell that setup for recirculation is in progress from
+ * the nonnegative value of 'recirc_action_offset'.
+ *
+ * 2. Sets 'exit' to true to tell later steps that we're exiting from the
+ * translation process.
+ *
+ * 3. Adds an OFPACT_UNROLL_XLATE action to 'action_set'. This action
+ * holds the current table ID and cookie so that they can be restored
+ * during a post-recirculation upcall translation.
+ *
+ * 4. Adds the action that prompted recirculation and any actions following
+ * it within the same flow to 'action_set', so that they can be executed
+ * during a post-recirculation upcall translation.
+ *
+ * 5. Returns.
+ *
+ * 6. The action that prompted recirculation might be nested in a stack of
+ * nested "resubmit"s that have actions remaining. Each of these notices
+ * that we're exiting (from 'exit') and that recirculation setup is in
+ * progress (from 'recirc_action_offset') and responds by adding more
+ * OFPACT_UNROLL_XLATE actions to 'action_set', as necessary, and any
+ * actions that were yet unprocessed.
+ *
+ * The caller stores all the state produced by this process associated with
+ * the recirculation ID. For post-recirculation upcall translation, the
+ * caller passes it back in for the new translation to execute. The
+ * process yielded a set of ofpacts that can be translated directly, so it
+ * is not much of a special case at that point.
+ */
+ int recirc_action_offset; /* Offset in 'action_set' to actions to be
+ * executed after recirculation, or -1. */
+ int last_unroll_offset; /* Offset in 'action_set' to the latest unroll
+ * action, or -1. */
/* True if a packet was but is no longer MPLS (due to an MPLS pop action).
* This is a trigger for recirculation in cases where translating an action
* 'action_set' accumulates "struct ofpact"s added by OFPACT_WRITE_ACTIONS.
* When translation is otherwise complete, ofpacts_execute_action_set()
* converts it to a set of "struct ofpact"s that can be translated into
- * datapath actions. */
+ * datapath actions. */
bool action_set_has_group; /* Action set contains OFPACT_GROUP? */
struct ofpbuf action_set; /* Action set. */
uint64_t action_set_stub[1024 / 8];
};
+static void xlate_action_set(struct xlate_ctx *ctx);
+
+static void
+ctx_trigger_recirculation(struct xlate_ctx *ctx)
+{
+ ctx->exit = true;
+ ctx->recirc_action_offset = ctx->action_set.size;
+}
+
+static bool
+ctx_first_recirculation_action(const struct xlate_ctx *ctx)
+{
+ return ctx->recirc_action_offset == ctx->action_set.size;
+}
+
+static inline bool
+exit_recirculates(const struct xlate_ctx *ctx)
+{
+ /* When recirculating the 'recirc_action_offset' has a non-negative value.
+ */
+ return ctx->recirc_action_offset >= 0;
+}
+
+static void compose_recirculate_action(struct xlate_ctx *ctx);
+
/* A controller may use OFPP_NONE as the ingress port to indicate that
* it did not arrive on a "real" port. 'ofpp_none_bundle' exists for
* when an input bundle is needed for validation (e.g., mirroring or
static void do_xlate_actions(const struct ofpact *, size_t ofpacts_len,
struct xlate_ctx *);
static void xlate_normal(struct xlate_ctx *);
-static inline void xlate_report(struct xlate_ctx *, const char *);
+static inline void xlate_report(struct xlate_ctx *, const char *, ...)
+ OVS_PRINTF_FORMAT(2, 3);
static void xlate_table_action(struct xlate_ctx *, ofp_port_t in_port,
uint8_t table_id, bool may_packet_in,
bool honor_table_miss);
static uint16_t input_vid_to_vlan(const struct xbundle *, uint16_t vid);
static void output_normal(struct xlate_ctx *, const struct xbundle *,
uint16_t vlan);
-static void compose_output_action(struct xlate_ctx *, ofp_port_t ofp_port);
+
+/* Optional bond recirculation parameter to compose_output_action(). */
+struct xlate_bond_recirc {
+ uint32_t recirc_id; /* !0 Use recirculation instead of output. */
+ uint8_t hash_alg; /* !0 Compute hash for recirc before. */
+ uint32_t hash_basis; /* Compute hash for recirc before. */
+};
+
+static void compose_output_action(struct xlate_ctx *, ofp_port_t ofp_port,
+ const struct xlate_bond_recirc *xr);
static struct xbridge *xbridge_lookup(struct xlate_cfg *,
const struct ofproto_dpif *);
const struct dpif_ipfix *,
const struct netflow *,
bool forward_bpdu, bool has_in_band,
- bool enable_recirc,
- bool variable_length_userdata,
- size_t max_mpls_depth,
- bool masked_set_action);
+ const struct dpif_backer_support *);
static void xlate_xbundle_set(struct xbundle *xbundle,
enum port_vlan_mode vlan_mode, int vlan,
unsigned long *trunks, bool use_priority_tags,
bool floodable);
static void xlate_xport_set(struct xport *xport, odp_port_t odp_port,
const struct netdev *netdev, const struct cfm *cfm,
- const struct bfd *bfd, int stp_port_no,
- const struct rstp_port *rstp_port,
+ const struct bfd *bfd, const struct lldp *lldp,
+ int stp_port_no, const struct rstp_port *rstp_port,
enum ofputil_port_config config,
enum ofputil_port_state state, bool is_tunnel,
bool may_enable);
static void xlate_xcfg_free(struct xlate_cfg *);
static inline void
-xlate_report(struct xlate_ctx *ctx, const char *s)
+xlate_report(struct xlate_ctx *ctx, const char *format, ...)
{
if (OVS_UNLIKELY(ctx->xin->report_hook)) {
- ctx->xin->report_hook(ctx->xin, s, ctx->recurse);
+ va_list args;
+
+ va_start(args, format);
+ ctx->xin->report_hook(ctx->xin, ctx->recurse, format, args);
+ va_end(args);
}
}
const struct dpif_ipfix *ipfix,
const struct netflow *netflow,
bool forward_bpdu, bool has_in_band,
- bool enable_recirc,
- bool variable_length_userdata,
- size_t max_mpls_depth,
- bool masked_set_action)
+ const struct dpif_backer_support *support)
{
if (xbridge->ml != ml) {
mac_learning_unref(xbridge->ml);
xbridge->dpif = dpif;
xbridge->forward_bpdu = forward_bpdu;
xbridge->has_in_band = has_in_band;
- xbridge->enable_recirc = enable_recirc;
- xbridge->variable_length_userdata = variable_length_userdata;
- xbridge->max_mpls_depth = max_mpls_depth;
- xbridge->masked_set_action = masked_set_action;
+ xbridge->support = *support;
}
static void
static void
xlate_xport_set(struct xport *xport, odp_port_t odp_port,
const struct netdev *netdev, const struct cfm *cfm,
- const struct bfd *bfd, int stp_port_no,
+ const struct bfd *bfd, const struct lldp *lldp, int stp_port_no,
const struct rstp_port* rstp_port,
enum ofputil_port_config config, enum ofputil_port_state state,
bool is_tunnel, bool may_enable)
xport->bfd = bfd_ref(bfd);
}
+ if (xport->lldp != lldp) {
+ lldp_unref(xport->lldp);
+ xport->lldp = lldp_ref(lldp);
+ }
+
if (xport->netdev != netdev) {
netdev_close(xport->netdev);
xport->netdev = netdev_ref(netdev);
xbridge->dpif, xbridge->ml, xbridge->stp,
xbridge->rstp, xbridge->ms, xbridge->mbridge,
xbridge->sflow, xbridge->ipfix, xbridge->netflow,
- xbridge->forward_bpdu,
- xbridge->has_in_band, xbridge->enable_recirc,
- xbridge->variable_length_userdata,
- xbridge->max_mpls_depth, xbridge->masked_set_action);
+ xbridge->forward_bpdu, xbridge->has_in_band,
+ &xbridge->support);
LIST_FOR_EACH (xbundle, list_node, &xbridge->xbundles) {
xlate_xbundle_copy(new_xbridge, xbundle);
}
xlate_xport_init(new_xcfg, new_xport);
xlate_xport_set(new_xport, xport->odp_port, xport->netdev, xport->cfm,
- xport->bfd, xport->stp_port_no, xport->rstp_port,
- xport->config, xport->state, xport->is_tunnel,
- xport->may_enable);
+ xport->bfd, xport->lldp, xport->stp_port_no,
+ xport->rstp_port, xport->config, xport->state,
+ xport->is_tunnel, xport->may_enable);
if (xport->peer) {
struct xport *peer = xport_lookup(new_xcfg, xport->peer->ofport);
struct xlate_cfg *xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
ovsrcu_set(&xcfgp, new_xcfg);
- ovsrcu_postpone(xlate_xcfg_free, xcfg);
-
+ ovsrcu_synchronize();
+ xlate_xcfg_free(xcfg);
new_xcfg = NULL;
}
const struct dpif_sflow *sflow,
const struct dpif_ipfix *ipfix,
const struct netflow *netflow,
- bool forward_bpdu, bool has_in_band, bool enable_recirc,
- bool variable_length_userdata, size_t max_mpls_depth,
- bool masked_set_action)
+ bool forward_bpdu, bool has_in_band,
+ const struct dpif_backer_support *support)
{
struct xbridge *xbridge;
xbridge->name = xstrdup(name);
xlate_xbridge_set(xbridge, dpif, ml, stp, rstp, ms, mbridge, sflow, ipfix,
- netflow, forward_bpdu, has_in_band, enable_recirc,
- variable_length_userdata, max_mpls_depth,
- masked_set_action);
+ netflow, forward_bpdu, has_in_band, support);
}
static void
static void
xlate_xbundle_remove(struct xlate_cfg *xcfg, struct xbundle *xbundle)
{
- struct xport *xport, *next;
+ struct xport *xport;
if (!xbundle) {
return;
}
- LIST_FOR_EACH_SAFE (xport, next, bundle_node, &xbundle->xports) {
- list_remove(&xport->bundle_node);
+ LIST_FOR_EACH_POP (xport, bundle_node, &xbundle->xports) {
xport->xbundle = NULL;
}
struct ofport_dpif *ofport, ofp_port_t ofp_port,
odp_port_t odp_port, const struct netdev *netdev,
const struct cfm *cfm, const struct bfd *bfd,
- struct ofport_dpif *peer, int stp_port_no,
- const struct rstp_port *rstp_port,
+ const struct lldp *lldp, struct ofport_dpif *peer,
+ int stp_port_no, const struct rstp_port *rstp_port,
const struct ofproto_port_queue *qdscp_list, size_t n_qdscp,
enum ofputil_port_config config,
enum ofputil_port_state state, bool is_tunnel,
ovs_assert(xport->ofp_port == ofp_port);
- xlate_xport_set(xport, odp_port, netdev, cfm, bfd, stp_port_no,
- rstp_port, config, state, is_tunnel, may_enable);
+ xlate_xport_set(xport, odp_port, netdev, cfm, bfd, lldp,
+ stp_port_no, rstp_port, config, state, is_tunnel,
+ may_enable);
if (xport->peer) {
xport->peer->peer = NULL;
rstp_port_unref(xport->rstp_port);
cfm_unref(xport->cfm);
bfd_unref(xport->bfd);
+ lldp_unref(xport->lldp);
free(xport);
}
xlate_xport_remove(new_xcfg, xport);
}
-/* Given a datapath and flow metadata ('backer', and 'flow' respectively)
- * returns the corresponding struct xport, or NULL if none is found. */
-static struct xport *
-xlate_lookup_xport(const struct dpif_backer *backer, const struct flow *flow)
-{
- struct xlate_cfg *xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
-
- return xport_lookup(xcfg, tnl_port_should_receive(flow)
- ? tnl_port_receive(flow)
- : odp_port_to_ofport(backer, flow->in_port.odp_port));
-}
-
static struct ofproto_dpif *
xlate_lookup_ofproto_(const struct dpif_backer *backer, const struct flow *flow,
ofp_port_t *ofp_in_port, const struct xport **xportp)
{
+ struct xlate_cfg *xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
const struct xport *xport;
- *xportp = xport = xlate_lookup_xport(backer, flow);
-
- if (xport) {
- if (ofp_in_port) {
- *ofp_in_port = xport->ofp_port;
- }
- return xport->xbridge->ofproto;
+ xport = xport_lookup(xcfg, tnl_port_should_receive(flow)
+ ? tnl_port_receive(flow)
+ : odp_port_to_ofport(backer, flow->in_port.odp_port));
+ if (OVS_UNLIKELY(!xport)) {
+ return NULL;
}
-
- return NULL;
+ *xportp = xport;
+ if (ofp_in_port) {
+ *ofp_in_port = xport->ofp_port;
+ }
+ return xport->xbridge->ofproto;
}
/* Given a datapath and flow metadata ('backer', and 'flow' respectively)
* pointers until quiescing, for longer term use additional references must
* be taken.
*
- * '*ofp_in_port' is set to OFPP_NONE if 'flow''s in_port does not exist.
- *
- * Returns 0 if successful, ENODEV if the parsed flow has no associated ofport.
+ * Returns 0 if successful, ENODEV if the parsed flow has no associated ofproto.
*/
int
xlate_lookup(const struct dpif_backer *backer, const struct flow *flow,
ofproto = xlate_lookup_ofproto_(backer, flow, ofp_in_port, &xport);
- if (ofp_in_port && !xport) {
- *ofp_in_port = OFPP_NONE;
- }
-
- if (!xport) {
+ if (!ofproto) {
return ENODEV;
}
}
if (ipfix) {
- *ipfix = xport->xbridge->ipfix;
+ *ipfix = xport ? xport->xbridge->ipfix : NULL;
}
if (sflow) {
- *sflow = xport->xbridge->sflow;
+ *sflow = xport ? xport->xbridge->sflow : NULL;
}
if (netflow) {
- *netflow = xport->xbridge->netflow;
+ *netflow = xport ? xport->xbridge->netflow : NULL;
}
+
return 0;
}
xport_stp_learn_state(const struct xport *xport)
{
struct stp_port *sp = xport_get_stp_port(xport);
- return stp_learn_in_state(sp ? stp_port_get_state(sp) : STP_DISABLED);
+ return sp
+ ? stp_learn_in_state(stp_port_get_state(sp))
+ : true;
}
static bool
xport_stp_forward_state(const struct xport *xport)
{
struct stp_port *sp = xport_get_stp_port(xport);
- return stp_forward_in_state(sp ? stp_port_get_state(sp) : STP_DISABLED);
+ return sp
+ ? stp_forward_in_state(stp_port_get_state(sp))
+ : true;
}
static bool
}
static void
-stp_process_packet(const struct xport *xport, const struct ofpbuf *packet)
+stp_process_packet(const struct xport *xport, const struct dp_packet *packet)
{
struct stp_port *sp = xport_get_stp_port(xport);
- struct ofpbuf payload = *packet;
- struct eth_header *eth = ofpbuf_data(&payload);
+ struct dp_packet payload = *packet;
+ struct eth_header *eth = dp_packet_data(&payload);
/* Sink packets on ports that have STP disabled when the bridge has
* STP enabled. */
}
/* Trim off padding on payload. */
- if (ofpbuf_size(&payload) > ntohs(eth->eth_type) + ETH_HEADER_LEN) {
- ofpbuf_set_size(&payload, ntohs(eth->eth_type) + ETH_HEADER_LEN);
+ if (dp_packet_size(&payload) > ntohs(eth->eth_type) + ETH_HEADER_LEN) {
+ dp_packet_set_size(&payload, ntohs(eth->eth_type) + ETH_HEADER_LEN);
}
- if (ofpbuf_try_pull(&payload, ETH_HEADER_LEN + LLC_HEADER_LEN)) {
- stp_received_bpdu(sp, ofpbuf_data(&payload), ofpbuf_size(&payload));
+ if (dp_packet_try_pull(&payload, ETH_HEADER_LEN + LLC_HEADER_LEN)) {
+ stp_received_bpdu(sp, dp_packet_data(&payload), dp_packet_size(&payload));
}
}
static bool
xport_rstp_learn_state(const struct xport *xport)
{
- return rstp_learn_in_state(xport_get_rstp_port_state(xport));
+ return xport->xbridge->rstp && xport->rstp_port
+ ? rstp_learn_in_state(xport_get_rstp_port_state(xport))
+ : true;
}
static bool
xport_rstp_forward_state(const struct xport *xport)
{
- return rstp_forward_in_state(xport_get_rstp_port_state(xport));
+ return xport->xbridge->rstp && xport->rstp_port
+ ? rstp_forward_in_state(xport_get_rstp_port_state(xport))
+ : true;
}
static bool
}
static void
-rstp_process_packet(const struct xport *xport, const struct ofpbuf *packet)
+rstp_process_packet(const struct xport *xport, const struct dp_packet *packet)
{
- struct ofpbuf payload = *packet;
- struct eth_header *eth = ofpbuf_data(&payload);
+ struct dp_packet payload = *packet;
+ struct eth_header *eth = dp_packet_data(&payload);
/* Sink packets on ports that have no RSTP. */
if (!xport->rstp_port) {
}
/* Trim off padding on payload. */
- if (ofpbuf_size(&payload) > ntohs(eth->eth_type) + ETH_HEADER_LEN) {
- ofpbuf_set_size(&payload, ntohs(eth->eth_type) + ETH_HEADER_LEN);
+ if (dp_packet_size(&payload) > ntohs(eth->eth_type) + ETH_HEADER_LEN) {
+ dp_packet_set_size(&payload, ntohs(eth->eth_type) + ETH_HEADER_LEN);
}
- if (ofpbuf_try_pull(&payload, ETH_HEADER_LEN + LLC_HEADER_LEN)) {
- rstp_port_received_bpdu(xport->rstp_port, ofpbuf_data(&payload),
- ofpbuf_size(&payload));
+ if (dp_packet_try_pull(&payload, ETH_HEADER_LEN + LLC_HEADER_LEN)) {
+ rstp_port_received_bpdu(xport->rstp_port, dp_packet_data(&payload),
+ dp_packet_size(&payload));
}
}
const struct group_dpif *group, int depth)
{
struct ofputil_bucket *bucket;
- const struct list *buckets;
+ const struct ovs_list *buckets;
group_dpif_get_buckets(group, &buckets);
LIST_FOR_EACH (bucket, list_node, buckets) {
int i = 0;
struct ofputil_bucket *bucket;
- const struct list *buckets;
+ const struct ovs_list *buckets;
group_dpif_get_buckets(group, &buckets);
LIST_FOR_EACH (bucket, list_node, buckets) {
uint16_t vid;
ovs_be16 tci, old_tci;
struct xport *xport;
+ struct xlate_bond_recirc xr;
+ bool use_recirc = false;
vid = output_vlan_to_vid(out_xbundle, vlan);
if (list_is_empty(&out_xbundle->xports)) {
/* Partially configured bundle with no slaves. Drop the packet. */
return;
} else if (!out_xbundle->bond) {
- ctx->use_recirc = false;
xport = CONTAINER_OF(list_front(&out_xbundle->xports), struct xport,
bundle_node);
} else {
struct xlate_cfg *xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
struct flow_wildcards *wc = &ctx->xout->wc;
- struct xlate_recirc *xr = &ctx->recirc;
struct ofport_dpif *ofport;
- if (ctx->xbridge->enable_recirc) {
- ctx->use_recirc = bond_may_recirc(
- out_xbundle->bond, &xr->recirc_id, &xr->hash_basis);
+ if (ctx->xbridge->support.recirc) {
+ use_recirc = bond_may_recirc(
+ out_xbundle->bond, &xr.recirc_id, &xr.hash_basis);
- if (ctx->use_recirc) {
+ if (use_recirc) {
/* Only TCP mode uses recirculation. */
- xr->hash_alg = OVS_HASH_ALG_L4;
+ xr.hash_alg = OVS_HASH_ALG_L4;
bond_update_post_recirc_rules(out_xbundle->bond, false);
/* Recirculation does not require unmasking hash fields. */
return;
}
- /* If ctx->xout->use_recirc is set, the main thread will handle stats
+ /* If use_recirc is set, the main thread will handle stats
* accounting for this bond. */
- if (!ctx->use_recirc) {
+ if (!use_recirc) {
if (ctx->xin->resubmit_stats) {
bond_account(out_xbundle->bond, &ctx->xin->flow, vid,
ctx->xin->resubmit_stats->n_bytes);
}
*flow_tci = tci;
- compose_output_action(ctx, xport->ofp_port);
+ compose_output_action(ctx, xport->ofp_port, use_recirc ? &xr : NULL);
*flow_tci = old_tci;
}
case BV_DROP_IF_MOVED:
ovs_rwlock_rdlock(&xbridge->ml->rwlock);
mac = mac_learning_lookup(xbridge->ml, flow->dl_src, vlan);
- if (mac && mac->port.p != in_xbundle->ofbundle &&
- (!is_gratuitous_arp(flow, &ctx->xout->wc)
- || mac_entry_is_grat_arp_locked(mac))) {
+ if (mac
+ && mac_entry_get_port(xbridge->ml, mac) != in_xbundle->ofbundle
+ && (!is_gratuitous_arp(flow, &ctx->xout->wc)
+ || mac_entry_is_grat_arp_locked(mac))) {
ovs_rwlock_unlock(&xbridge->ml->rwlock);
xlate_report(ctx, "SLB bond thinks this packet looped back, "
"dropping");
}
}
- return mac->port.p != in_xbundle->ofbundle;
+ return mac_entry_get_port(ml, mac) != in_xbundle->ofbundle;
}
}
}
- if (mac->port.p != in_xbundle->ofbundle) {
+ if (mac_entry_get_port(xbridge->ml, mac) != in_xbundle->ofbundle) {
/* The log messages here could actually be useful in debugging,
* so keep the rate limit relatively high. */
static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(30, 300);
xbridge->name, ETH_ADDR_ARGS(flow->dl_src),
in_xbundle->name, vlan);
- mac->port.p = in_xbundle->ofbundle;
- mac_learning_changed(xbridge->ml);
+ mac_entry_set_port(xbridge->ml, mac, in_xbundle->ofbundle);
}
}
/* Updates multicast snooping table 'ms' given that a packet matching 'flow'
* was received on 'in_xbundle' in 'vlan' and is either Report or Query. */
static void
-update_mcast_snooping_table__(const struct xbridge *xbridge,
- const struct flow *flow,
- struct mcast_snooping *ms,
- ovs_be32 ip4, int vlan,
- struct xbundle *in_xbundle)
+update_mcast_snooping_table4__(const struct xbridge *xbridge,
+ const struct flow *flow,
+ struct mcast_snooping *ms, int vlan,
+ struct xbundle *in_xbundle,
+ const struct dp_packet *packet)
OVS_REQ_WRLOCK(ms->rwlock)
{
static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(60, 30);
+ int count;
+ ovs_be32 ip4 = flow->igmp_group_ip4;
switch (ntohs(flow->tp_src)) {
case IGMP_HOST_MEMBERSHIP_REPORT:
case IGMPV2_HOST_MEMBERSHIP_REPORT:
- if (mcast_snooping_add_group(ms, ip4, vlan, in_xbundle->ofbundle)) {
+ if (mcast_snooping_add_group4(ms, ip4, vlan, in_xbundle->ofbundle)) {
VLOG_DBG_RL(&rl, "bridge %s: multicast snooping learned that "
IP_FMT" is on port %s in VLAN %d",
xbridge->name, IP_ARGS(ip4), in_xbundle->name, vlan);
}
break;
case IGMP_HOST_LEAVE_MESSAGE:
- if (mcast_snooping_leave_group(ms, ip4, vlan, in_xbundle->ofbundle)) {
+ if (mcast_snooping_leave_group4(ms, ip4, vlan, in_xbundle->ofbundle)) {
VLOG_DBG_RL(&rl, "bridge %s: multicast snooping leaving "
IP_FMT" is on port %s in VLAN %d",
xbridge->name, IP_ARGS(ip4), in_xbundle->name, vlan);
in_xbundle->name, vlan);
}
break;
+ case IGMPV3_HOST_MEMBERSHIP_REPORT:
+ if ((count = mcast_snooping_add_report(ms, packet, vlan,
+ in_xbundle->ofbundle))) {
+ VLOG_DBG_RL(&rl, "bridge %s: multicast snooping processed %d "
+ "addresses on port %s in VLAN %d",
+ xbridge->name, count, in_xbundle->name, vlan);
+ }
+ break;
+ }
+}
+
+static void
+update_mcast_snooping_table6__(const struct xbridge *xbridge,
+ const struct flow *flow,
+ struct mcast_snooping *ms, int vlan,
+ struct xbundle *in_xbundle,
+ const struct dp_packet *packet)
+ OVS_REQ_WRLOCK(ms->rwlock)
+{
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(60, 30);
+ int count;
+
+ switch (ntohs(flow->tp_src)) {
+ case MLD_QUERY:
+ if (!ipv6_addr_equals(&flow->ipv6_src, &in6addr_any)
+ && mcast_snooping_add_mrouter(ms, vlan, in_xbundle->ofbundle)) {
+ VLOG_DBG_RL(&rl, "bridge %s: multicast snooping query on port %s"
+ "in VLAN %d",
+ xbridge->name, in_xbundle->name, vlan);
+ }
+ break;
+ case MLD_REPORT:
+ case MLD_DONE:
+ case MLD2_REPORT:
+ count = mcast_snooping_add_mld(ms, packet, vlan, in_xbundle->ofbundle);
+ if (count) {
+ VLOG_DBG_RL(&rl, "bridge %s: multicast snooping processed %d "
+ "addresses on port %s in VLAN %d",
+ xbridge->name, count, in_xbundle->name, vlan);
+ }
+ break;
}
}
static void
update_mcast_snooping_table(const struct xbridge *xbridge,
const struct flow *flow, int vlan,
- struct xbundle *in_xbundle)
+ struct xbundle *in_xbundle,
+ const struct dp_packet *packet)
{
struct mcast_snooping *ms = xbridge->ms;
struct xlate_cfg *xcfg;
struct xbundle *mcast_xbundle;
- struct mcast_fport_bundle *fport;
+ struct mcast_port_bundle *fport;
/* Don't learn the OFPP_NONE port. */
if (in_xbundle == &ofpp_none_bundle) {
mcast_xbundle = NULL;
ovs_rwlock_wrlock(&ms->rwlock);
xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
- LIST_FOR_EACH(fport, fport_node, &ms->fport_list) {
+ LIST_FOR_EACH(fport, node, &ms->fport_list) {
mcast_xbundle = xbundle_lookup(xcfg, fport->port);
if (mcast_xbundle == in_xbundle) {
break;
}
if (!mcast_xbundle || mcast_xbundle != in_xbundle) {
- update_mcast_snooping_table__(xbridge, flow, ms, flow->igmp_group_ip4,
- vlan, in_xbundle);
+ if (flow->dl_type == htons(ETH_TYPE_IP)) {
+ update_mcast_snooping_table4__(xbridge, flow, ms, vlan,
+ in_xbundle, packet);
+ } else {
+ update_mcast_snooping_table6__(xbridge, flow, ms, vlan,
+ in_xbundle, packet);
+ }
}
ovs_rwlock_unlock(&ms->rwlock);
}
OVS_REQ_RDLOCK(ms->rwlock)
{
struct xlate_cfg *xcfg;
- struct mcast_fport_bundle *fport;
+ struct mcast_port_bundle *fport;
struct xbundle *mcast_xbundle;
xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
- LIST_FOR_EACH(fport, fport_node, &ms->fport_list) {
+ LIST_FOR_EACH(fport, node, &ms->fport_list) {
mcast_xbundle = xbundle_lookup(xcfg, fport->port);
if (mcast_xbundle && mcast_xbundle != in_xbundle) {
xlate_report(ctx, "forwarding to mcast flood port");
}
}
+/* forward the Reports to configured ports */
+static void
+xlate_normal_mcast_send_rports(struct xlate_ctx *ctx,
+ struct mcast_snooping *ms,
+ struct xbundle *in_xbundle, uint16_t vlan)
+ OVS_REQ_RDLOCK(ms->rwlock)
+{
+ struct xlate_cfg *xcfg;
+ struct mcast_port_bundle *rport;
+ struct xbundle *mcast_xbundle;
+
+ xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
+ LIST_FOR_EACH(rport, node, &ms->rport_list) {
+ mcast_xbundle = xbundle_lookup(xcfg, rport->port);
+ if (mcast_xbundle && mcast_xbundle != in_xbundle) {
+ xlate_report(ctx, "forwarding Report to mcast flagged port");
+ output_normal(ctx, mcast_xbundle, vlan);
+ } else if (!mcast_xbundle) {
+ xlate_report(ctx, "mcast port is unknown, dropping the Report");
+ } else {
+ xlate_report(ctx, "mcast port is input port, dropping the Report");
+ }
+ }
+}
+
static void
xlate_normal_flood(struct xlate_ctx *ctx, struct xbundle *in_xbundle,
uint16_t vlan)
if (mcast_snooping_enabled(ctx->xbridge->ms)
&& !eth_addr_is_broadcast(flow->dl_dst)
&& eth_addr_is_multicast(flow->dl_dst)
- && flow->dl_type == htons(ETH_TYPE_IP)) {
+ && is_ip_any(flow)) {
struct mcast_snooping *ms = ctx->xbridge->ms;
- struct mcast_group *grp;
+ struct mcast_group *grp = NULL;
- if (flow->nw_proto == IPPROTO_IGMP) {
- if (ctx->xin->may_learn) {
- if (mcast_snooping_is_membership(flow->tp_src) ||
- mcast_snooping_is_query(flow->tp_src)) {
+ if (is_igmp(flow)) {
+ if (mcast_snooping_is_membership(flow->tp_src) ||
+ mcast_snooping_is_query(flow->tp_src)) {
+ if (ctx->xin->may_learn) {
update_mcast_snooping_table(ctx->xbridge, flow, vlan,
- in_xbundle);
- }
+ in_xbundle, ctx->xin->packet);
+ }
+ /*
+ * IGMP packets need to take the slow path, in order to be
+ * processed for mdb updates. That will prevent expires
+ * firing off even after hosts have sent reports.
+ */
+ ctx->xout->slow |= SLOW_ACTION;
}
if (mcast_snooping_is_membership(flow->tp_src)) {
ovs_rwlock_rdlock(&ms->rwlock);
xlate_normal_mcast_send_mrouters(ctx, ms, in_xbundle, vlan);
+ /* RFC4541: section 2.1.1, item 1: A snooping switch should
+ * forward IGMP Membership Reports only to those ports where
+ * multicast routers are attached. Alternatively stated: a
+ * snooping switch should not forward IGMP Membership Reports
+ * to ports on which only hosts are attached.
+ * An administrative control may be provided to override this
+ * restriction, allowing the report messages to be flooded to
+ * other ports. */
+ xlate_normal_mcast_send_rports(ctx, ms, in_xbundle, vlan);
ovs_rwlock_unlock(&ms->rwlock);
} else {
xlate_report(ctx, "multicast traffic, flooding");
xlate_normal_flood(ctx, in_xbundle, vlan);
}
return;
+ } else if (is_mld(flow)) {
+ ctx->xout->slow |= SLOW_ACTION;
+ if (ctx->xin->may_learn) {
+ update_mcast_snooping_table(ctx->xbridge, flow, vlan,
+ in_xbundle, ctx->xin->packet);
+ }
+ if (is_mld_report(flow)) {
+ ovs_rwlock_rdlock(&ms->rwlock);
+ xlate_normal_mcast_send_mrouters(ctx, ms, in_xbundle, vlan);
+ xlate_normal_mcast_send_rports(ctx, ms, in_xbundle, vlan);
+ ovs_rwlock_unlock(&ms->rwlock);
+ } else {
+ xlate_report(ctx, "MLD query, flooding");
+ xlate_normal_flood(ctx, in_xbundle, vlan);
+ }
} else {
- if (ip_is_local_multicast(flow->nw_dst)) {
+ if ((flow->dl_type == htons(ETH_TYPE_IP)
+ && ip_is_local_multicast(flow->nw_dst))
+ || (flow->dl_type == htons(ETH_TYPE_IPV6)
+ && ipv6_is_all_hosts(&flow->ipv6_dst))) {
/* RFC4541: section 2.1.2, item 2: Packets with a dst IP
* address in the 224.0.0.x range which are not IGMP must
* be forwarded on all ports */
/* forwarding to group base ports */
ovs_rwlock_rdlock(&ms->rwlock);
- grp = mcast_snooping_lookup(ms, flow->nw_dst, vlan);
+ if (flow->dl_type == htons(ETH_TYPE_IP)) {
+ grp = mcast_snooping_lookup4(ms, flow->nw_dst, vlan);
+ } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) {
+ grp = mcast_snooping_lookup(ms, &flow->ipv6_dst, vlan);
+ }
if (grp) {
xlate_normal_mcast_send_group(ctx, ms, grp, in_xbundle, vlan);
xlate_normal_mcast_send_fports(ctx, ms, in_xbundle, vlan);
} else {
ovs_rwlock_rdlock(&ctx->xbridge->ml->rwlock);
mac = mac_learning_lookup(ctx->xbridge->ml, flow->dl_dst, vlan);
- mac_port = mac ? mac->port.p : NULL;
+ mac_port = mac ? mac_entry_get_port(ctx->xbridge->ml, mac) : NULL;
ovs_rwlock_unlock(&ctx->xbridge->ml->rwlock);
if (mac_port) {
static enum slow_path_reason
process_special(struct xlate_ctx *ctx, const struct flow *flow,
- const struct xport *xport, const struct ofpbuf *packet)
+ const struct xport *xport, const struct dp_packet *packet)
{
struct flow_wildcards *wc = &ctx->xout->wc;
const struct xbridge *xbridge = ctx->xbridge;
: rstp_process_packet(xport, packet);
}
return SLOW_STP;
+ } else if (xport->lldp && lldp_should_process_flow(xport->lldp, flow)) {
+ if (packet) {
+ lldp_process_packet(xport->lldp, packet);
+ }
+ return SLOW_LLDP;
} else {
return 0;
}
}
static int
-xlate_flood_packet(struct xbridge *xbridge, struct ofpbuf *packet)
+xlate_flood_packet(struct xbridge *xbridge, struct dp_packet *packet)
{
struct ofpact_output output;
struct flow flow;
ofpact_init(&output.ofpact, OFPACT_OUTPUT, sizeof output);
/* Use OFPP_NONE as the in_port to avoid special packet processing. */
- flow_extract(packet, NULL, &flow);
+ flow_extract(packet, &flow);
flow.in_port.ofp_port = OFPP_NONE;
output.port = OFPP_FLOOD;
output.max_len = 0;
ovs_be32 ip_src, ovs_be32 ip_dst)
{
struct xbridge *xbridge = out_dev->xbridge;
- struct ofpbuf packet;
+ struct dp_packet packet;
- ofpbuf_init(&packet, 0);
- compose_arp(&packet, eth_src, ip_src, ip_dst);
+ dp_packet_init(&packet, 0);
+ compose_arp(&packet, ARP_OP_REQUEST,
+ eth_src, eth_addr_zero, true, ip_src, ip_dst);
xlate_flood_packet(xbridge, &packet);
- ofpbuf_uninit(&packet);
+ dp_packet_uninit(&packet);
}
static int
-build_tunnel_send(const struct xlate_ctx *ctx, const struct xport *xport,
+build_tunnel_send(struct xlate_ctx *ctx, const struct xport *xport,
const struct flow *flow, odp_port_t tunnel_odp_port)
{
struct ovs_action_push_tnl tnl_push_data;
err = tnl_route_lookup_flow(flow, &d_ip, &out_dev);
if (err) {
+ xlate_report(ctx, "native tunnel routing failed");
return err;
}
+ xlate_report(ctx, "tunneling to "IP_FMT" via %s",
+ IP_ARGS(d_ip), netdev_get_name(out_dev->netdev));
/* Use mac addr of bridge port of the peer. */
err = netdev_get_etheraddr(out_dev->netdev, smac);
if (err) {
+ xlate_report(ctx, "tunnel output device lacks Ethernet address");
return err;
}
err = netdev_get_in4(out_dev->netdev, (struct in_addr *) &s_ip, NULL);
if (err) {
+ xlate_report(ctx, "tunnel output device lacks IPv4 address");
return err;
}
err = tnl_arp_lookup(out_dev->xbridge->name, d_ip, dmac);
if (err) {
+ xlate_report(ctx, "ARP cache miss for "IP_FMT" on bridge %s, "
+ "sending ARP request",
+ IP_ARGS(d_ip), out_dev->xbridge->name);
tnl_send_arp_request(out_dev, smac, s_ip, d_ip);
return err;
}
struct xc_entry *entry;
entry = xlate_cache_add_entry(ctx->xin->xcache, XC_TNL_ARP);
- strncpy(entry->u.tnl_arp_cache.br_name, out_dev->xbridge->name, IFNAMSIZ);
+ ovs_strlcpy(entry->u.tnl_arp_cache.br_name, out_dev->xbridge->name,
+ sizeof entry->u.tnl_arp_cache.br_name);
entry->u.tnl_arp_cache.d_ip = d_ip;
}
+
+ xlate_report(ctx, "tunneling from "ETH_ADDR_FMT" "IP_FMT
+ " to "ETH_ADDR_FMT" "IP_FMT,
+ ETH_ADDR_ARGS(smac), IP_ARGS(s_ip),
+ ETH_ADDR_ARGS(dmac), IP_ARGS(d_ip));
err = tnl_port_build_header(xport->ofport, flow,
dmac, smac, s_ip, &tnl_push_data);
if (err) {
static void
compose_output_action__(struct xlate_ctx *ctx, ofp_port_t ofp_port,
- bool check_stp)
+ const struct xlate_bond_recirc *xr, bool check_stp)
{
const struct xport *xport = get_ofp_port(ctx->xbridge, ofp_port);
struct flow_wildcards *wc = &ctx->xout->wc;
/* If 'struct flow' gets additional metadata, we'll need to zero it out
* before traversing a patch port. */
- BUILD_ASSERT_DECL(FLOW_WC_SEQ == 28);
+ BUILD_ASSERT_DECL(FLOW_WC_SEQ == 32);
memset(&flow_tnl, 0, sizeof flow_tnl);
if (!xport) {
if (xport->peer) {
const struct xport *peer = xport->peer;
struct flow old_flow = ctx->xin->flow;
+ bool old_was_mpls = ctx->was_mpls;
+ cls_version_t old_version = ctx->tables_version;
enum slow_path_reason special;
+ struct ofpbuf old_stack = ctx->stack;
+ union mf_subvalue new_stack[1024 / sizeof(union mf_subvalue)];
+ struct ofpbuf old_action_set = ctx->action_set;
+ uint64_t actset_stub[1024 / 8];
+ ofpbuf_use_stub(&ctx->stack, new_stack, sizeof new_stack);
+ ofpbuf_use_stub(&ctx->action_set, actset_stub, sizeof actset_stub);
ctx->xbridge = peer->xbridge;
flow->in_port.ofp_port = peer->ofp_port;
flow->metadata = htonll(0);
memset(flow->regs, 0, sizeof flow->regs);
flow->actset_output = OFPP_UNSET;
+ /* The bridge is now known so obtain its table version. */
+ ctx->tables_version
+ = ofproto_dpif_get_tables_version(ctx->xbridge->ofproto);
+
special = process_special(ctx, &ctx->xin->flow, peer,
ctx->xin->packet);
if (special) {
} else if (may_receive(peer, ctx)) {
if (xport_stp_forward_state(peer) && xport_rstp_forward_state(peer)) {
xlate_table_action(ctx, flow->in_port.ofp_port, 0, true, true);
+ if (ctx->action_set.size) {
+ /* Translate action set only if not dropping the packet and
+ * not recirculating. */
+ if (!exit_recirculates(ctx)) {
+ xlate_action_set(ctx);
+ }
+ }
+ /* Check if need to recirculate. */
+ if (exit_recirculates(ctx)) {
+ compose_recirculate_action(ctx);
+ }
} else {
/* Forwarding is disabled by STP and RSTP. Let OFPP_NORMAL and
* the learning action look at the packet, then drop it. */
struct flow old_base_flow = ctx->base_flow;
- size_t old_size = ofpbuf_size(ctx->xout->odp_actions);
+ size_t old_size = ctx->xout->odp_actions->size;
mirror_mask_t old_mirrors = ctx->xout->mirrors;
+
xlate_table_action(ctx, flow->in_port.ofp_port, 0, true, true);
ctx->xout->mirrors = old_mirrors;
ctx->base_flow = old_base_flow;
- ofpbuf_set_size(ctx->xout->odp_actions, old_size);
+ ctx->xout->odp_actions->size = old_size;
+
+ /* Undo changes that may have been done for recirculation. */
+ if (exit_recirculates(ctx)) {
+ ctx->action_set.size = ctx->recirc_action_offset;
+ ctx->recirc_action_offset = -1;
+ ctx->last_unroll_offset = -1;
+ }
}
}
ctx->xin->flow = old_flow;
ctx->xbridge = xport->xbridge;
+ ofpbuf_uninit(&ctx->action_set);
+ ctx->action_set = old_action_set;
+ ofpbuf_uninit(&ctx->stack);
+ ctx->stack = old_stack;
+
+ /* Restore calling bridge's lookup version. */
+ ctx->tables_version = old_version;
+
+ /* The peer bridge popping MPLS should have no effect on the original
+ * bridge. */
+ ctx->was_mpls = old_was_mpls;
+
+ /* The fact that the peer bridge exits (for any reason) does not mean
+ * that the original bridge should exit. Specifically, if the peer
+ * bridge recirculates (which typically modifies the packet), the
+ * original bridge must continue processing with the original, not the
+ * recirculated packet! */
+ ctx->exit = false;
if (ctx->xin->resubmit_stats) {
netdev_vport_inc_tx(xport->netdev, ctx->xin->resubmit_stats);
}
out_port = odp_port;
if (ovs_native_tunneling_is_on(ctx->xbridge->ofproto)) {
+ xlate_report(ctx, "output to native tunnel");
tnl_push_pop_send = true;
} else {
+ xlate_report(ctx, "output to kernel tunnel");
commit_odp_tunnel_action(flow, &ctx->base_flow,
ctx->xout->odp_actions);
flow->tunnel = flow_tnl; /* Restore tunnel metadata */
}
if (out_port != ODPP_NONE) {
+ bool use_masked = ctx->xbridge->support.masked_set_action;
+
ctx->xout->slow |= commit_odp_actions(flow, &ctx->base_flow,
ctx->xout->odp_actions,
- wc,
- ctx->xbridge->masked_set_action);
+ wc, use_masked);
- if (ctx->use_recirc) {
+ if (xr) {
struct ovs_action_hash *act_hash;
- struct xlate_recirc *xr = &ctx->recirc;
/* Hash action. */
act_hash = nl_msg_put_unspec_uninit(ctx->xout->odp_actions,
}
static void
-compose_output_action(struct xlate_ctx *ctx, ofp_port_t ofp_port)
+compose_output_action(struct xlate_ctx *ctx, ofp_port_t ofp_port,
+ const struct xlate_bond_recirc *xr)
{
- compose_output_action__(ctx, ofp_port, true);
+ compose_output_action__(ctx, ofp_port, xr, true);
}
static void
xlate_recursively(struct xlate_ctx *ctx, struct rule_dpif *rule)
{
struct rule_dpif *old_rule = ctx->rule;
+ ovs_be64 old_cookie = ctx->rule_cookie;
const struct rule_actions *actions;
if (ctx->xin->resubmit_stats) {
ctx->resubmits++;
ctx->recurse++;
ctx->rule = rule;
+ ctx->rule_cookie = rule_dpif_get_flow_cookie(rule);
actions = rule_dpif_get_actions(rule);
do_xlate_actions(actions->ofpacts, actions->ofpacts_len, ctx);
+ ctx->rule_cookie = old_cookie;
ctx->rule = old_rule;
ctx->recurse--;
}
MAX_RESUBMIT_RECURSION);
} else if (ctx->resubmits >= MAX_RESUBMITS + MAX_INTERNAL_RESUBMITS) {
VLOG_ERR_RL(&rl, "over %d resubmit actions", MAX_RESUBMITS);
- } else if (ofpbuf_size(ctx->xout->odp_actions) > UINT16_MAX) {
+ } else if (ctx->xout->odp_actions->size > UINT16_MAX) {
VLOG_ERR_RL(&rl, "resubmits yielded over 64 kB of actions");
- } else if (ofpbuf_size(&ctx->stack) >= 65536) {
+ } else if (ctx->stack.size >= 65536) {
VLOG_ERR_RL(&rl, "resubmits yielded over 64 kB of stack");
} else {
return true;
xlate_table_action(struct xlate_ctx *ctx, ofp_port_t in_port, uint8_t table_id,
bool may_packet_in, bool honor_table_miss)
{
+ /* Check if we need to recirculate before matching in a table. */
+ if (ctx->was_mpls) {
+ ctx_trigger_recirculation(ctx);
+ return;
+ }
if (xlate_resubmit_resource_check(ctx)) {
struct flow_wildcards *wc;
uint8_t old_table_id = ctx->table_id;
wc = (ctx->xin->skip_wildcards) ? NULL : &ctx->xout->wc;
rule = rule_dpif_lookup_from_table(ctx->xbridge->ofproto,
+ ctx->tables_version,
&ctx->xin->flow, wc,
ctx->xin->xcache != NULL,
ctx->xin->resubmit_stats,
{
uint64_t action_list_stub[1024 / 8];
struct ofpbuf action_list, action_set;
+ struct flow old_flow = ctx->xin->flow;
+ bool old_was_mpls = ctx->was_mpls;
ofpbuf_use_const(&action_set, bucket->ofpacts, bucket->ofpacts_len);
ofpbuf_use_stub(&action_list, action_list_stub, sizeof action_list_stub);
ofpacts_execute_action_set(&action_list, &action_set);
ctx->recurse++;
- do_xlate_actions(ofpbuf_data(&action_list), ofpbuf_size(&action_list), ctx);
+ do_xlate_actions(action_list.data, action_list.size, ctx);
ctx->recurse--;
ofpbuf_uninit(&action_set);
ofpbuf_uninit(&action_list);
+
+ /* Check if need to recirculate. */
+ if (exit_recirculates(ctx)) {
+ compose_recirculate_action(ctx);
+ }
+
+ /* Roll back flow to previous state.
+ * This is equivalent to cloning the packet for each bucket.
+ *
+ * As a side effect any subsequently applied actions will
+ * also effectively be applied to a clone of the packet taken
+ * just before applying the all or indirect group.
+ *
+ * Note that group buckets are action sets, hence they cannot modify the
+ * main action set. Also any stack actions are ignored when executing an
+ * action set, so group buckets cannot change the stack either.
+ * However, we do allow resubmit actions in group buckets, which could
+ * break the above assumptions. It is up to the controller to not mess up
+ * with the action_set and stack in the tables resubmitted to from
+ * group buckets. */
+ ctx->xin->flow = old_flow;
+
+ /* The group bucket popping MPLS should have no effect after bucket
+ * execution. */
+ ctx->was_mpls = old_was_mpls;
+
+ /* The fact that the group bucket exits (for any reason) does not mean that
+ * the translation after the group action should exit. Specifically, if
+ * the group bucket recirculates (which typically modifies the packet), the
+ * actions after the group action must continue processing with the
+ * original, not the recirculated packet! */
+ ctx->exit = false;
}
static void
xlate_all_group(struct xlate_ctx *ctx, struct group_dpif *group)
{
struct ofputil_bucket *bucket;
- const struct list *buckets;
- struct flow old_flow = ctx->xin->flow;
+ const struct ovs_list *buckets;
group_dpif_get_buckets(group, &buckets);
LIST_FOR_EACH (bucket, list_node, buckets) {
xlate_group_bucket(ctx, bucket);
- /* Roll back flow to previous state.
- * This is equivalent to cloning the packet for each bucket.
- *
- * As a side effect any subsequently applied actions will
- * also effectively be applied to a clone of the packet taken
- * just before applying the all or indirect group. */
- ctx->xin->flow = old_flow;
}
xlate_group_stats(ctx, group, NULL);
}
}
static void
-xlate_select_group(struct xlate_ctx *ctx, struct group_dpif *group)
+xlate_default_select_group(struct xlate_ctx *ctx, struct group_dpif *group)
{
struct flow_wildcards *wc = &ctx->xout->wc;
struct ofputil_bucket *bucket;
uint32_t basis;
basis = flow_hash_symmetric_l4(&ctx->xin->flow, 0);
+ flow_mask_hash_fields(&ctx->xin->flow, wc, NX_HASH_FIELDS_SYMMETRIC_L4);
bucket = group_best_live_bucket(ctx, group, basis);
if (bucket) {
- memset(&wc->masks.dl_dst, 0xff, sizeof wc->masks.dl_dst);
- memset(&wc->masks.dl_src, 0xff, sizeof wc->masks.dl_src);
- memset(&wc->masks.dl_type, 0xff, sizeof wc->masks.dl_type);
- memset(&wc->masks.nw_dst, 0xff, sizeof wc->masks.nw_dst);
- memset(&wc->masks.nw_proto, 0xff, sizeof wc->masks.nw_proto);
- memset(&wc->masks.nw_src, 0xff, sizeof wc->masks.nw_src);
- memset(&wc->masks.tp_dst, 0xff, sizeof wc->masks.tp_dst);
- memset(&wc->masks.tp_src, 0xff, sizeof wc->masks.tp_src);
- memset(&wc->masks.vlan_tci, 0xff, sizeof wc->masks.vlan_tci);
+ xlate_group_bucket(ctx, bucket);
+ xlate_group_stats(ctx, group, bucket);
+ }
+}
+static void
+xlate_hash_fields_select_group(struct xlate_ctx *ctx, struct group_dpif *group)
+{
+ struct mf_bitmap hash_fields = MF_BITMAP_INITIALIZER;
+ struct flow_wildcards *wc = &ctx->xout->wc;
+ const struct field_array *fields;
+ struct ofputil_bucket *bucket;
+ uint32_t basis;
+ int i;
+
+ fields = group_dpif_get_fields(group);
+ basis = hash_uint64(group_dpif_get_selection_method_param(group));
+
+ /* Determine which fields to hash */
+ for (i = 0; i < MFF_N_IDS; i++) {
+ if (bitmap_is_set(fields->used.bm, i)) {
+ const struct mf_field *mf;
+
+ /* If the field is already present in 'hash_fields' then
+ * this loop has already checked that it and its pre-requisites
+ * are present in the flow and its pre-requisites have
+ * already been added to 'hash_fields'. There is nothing more
+ * to do here and as an optimisation the loop can continue. */
+ if (bitmap_is_set(hash_fields.bm, i)) {
+ continue;
+ }
+
+ mf = mf_from_id(i);
+
+ /* Only hash a field if it and its pre-requisites are present
+ * in the flow. */
+ if (!mf_are_prereqs_ok(mf, &ctx->xin->flow)) {
+ continue;
+ }
+
+ /* Hash both the field and its pre-requisites */
+ mf_bitmap_set_field_and_prereqs(mf, &hash_fields);
+ }
+ }
+
+ /* Hash the fields */
+ for (i = 0; i < MFF_N_IDS; i++) {
+ if (bitmap_is_set(hash_fields.bm, i)) {
+ const struct mf_field *mf = mf_from_id(i);
+ union mf_value value;
+ int j;
+
+ mf_get_value(mf, &ctx->xin->flow, &value);
+ /* This seems inefficient but so does apply_mask() */
+ for (j = 0; j < mf->n_bytes; j++) {
+ ((uint8_t *) &value)[j] &= ((uint8_t *) &fields->value[i])[j];
+ }
+ basis = hash_bytes(&value, mf->n_bytes, basis);
+
+ mf_mask_field(mf, &wc->masks);
+ }
+ }
+
+ bucket = group_best_live_bucket(ctx, group, basis);
+ if (bucket) {
xlate_group_bucket(ctx, bucket);
xlate_group_stats(ctx, group, bucket);
}
}
+static void
+xlate_select_group(struct xlate_ctx *ctx, struct group_dpif *group)
+{
+ const char *selection_method = group_dpif_get_selection_method(group);
+
+ if (selection_method[0] == '\0') {
+ xlate_default_select_group(ctx, group);
+ } else if (!strcasecmp("hash", selection_method)) {
+ xlate_hash_fields_select_group(ctx, group);
+ } else {
+ /* Parsing of groups should ensure this never happens */
+ OVS_NOT_REACHED();
+ }
+}
+
static void
xlate_group_action__(struct xlate_ctx *ctx, struct group_dpif *group)
{
+ bool was_in_group = ctx->in_group;
ctx->in_group = true;
switch (group_dpif_get_type(group)) {
}
group_dpif_unref(group);
- ctx->in_group = false;
-}
-
-static bool
-xlate_group_resource_check(struct xlate_ctx *ctx)
-{
- if (!xlate_resubmit_resource_check(ctx)) {
- return false;
- } else if (ctx->in_group) {
- /* Prevent nested translation of OpenFlow groups.
- *
- * OpenFlow allows this restriction. We enforce this restriction only
- * because, with the current architecture, we would otherwise have to
- * take a possibly recursive read lock on the ofgroup rwlock, which is
- * unsafe given that POSIX allows taking a read lock to block if there
- * is a thread blocked on taking the write lock. Other solutions
- * without this restriction are also possible, but seem unwarranted
- * given the current limited use of groups. */
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
-
- VLOG_ERR_RL(&rl, "cannot recursively translate OpenFlow group");
- return false;
- } else {
- return true;
- }
+ ctx->in_group = was_in_group;
}
static bool
xlate_group_action(struct xlate_ctx *ctx, uint32_t group_id)
{
- if (xlate_group_resource_check(ctx)) {
+ if (xlate_resubmit_resource_check(ctx)) {
struct group_dpif *group;
bool got_group;
}
if (all) {
- compose_output_action__(ctx, xport->ofp_port, false);
+ compose_output_action__(ctx, xport->ofp_port, NULL, false);
} else if (!(xport->config & OFPUTIL_PC_NO_FLOOD)) {
- compose_output_action(ctx, xport->ofp_port);
+ compose_output_action(ctx, xport->ofp_port, NULL);
}
}
uint16_t controller_id)
{
struct ofproto_packet_in *pin;
- struct dpif_packet *packet;
+ struct dp_packet *packet;
+ bool use_masked;
ctx->xout->slow |= SLOW_CONTROLLER;
if (!ctx->xin->packet) {
return;
}
- packet = dpif_packet_clone_from_ofpbuf(ctx->xin->packet);
+ packet = dp_packet_clone(ctx->xin->packet);
+ use_masked = ctx->xbridge->support.masked_set_action;
ctx->xout->slow |= commit_odp_actions(&ctx->xin->flow, &ctx->base_flow,
ctx->xout->odp_actions,
- &ctx->xout->wc,
- ctx->xbridge->masked_set_action);
+ &ctx->xout->wc, use_masked);
odp_execute_actions(NULL, &packet, 1, false,
- ofpbuf_data(ctx->xout->odp_actions),
- ofpbuf_size(ctx->xout->odp_actions), NULL);
+ ctx->xout->odp_actions->data,
+ ctx->xout->odp_actions->size, NULL);
pin = xmalloc(sizeof *pin);
- pin->up.packet_len = ofpbuf_size(&packet->ofpbuf);
- pin->up.packet = ofpbuf_steal_data(&packet->ofpbuf);
+ pin->up.packet_len = dp_packet_size(packet);
+ pin->up.packet = dp_packet_steal_data(packet);
pin->up.reason = reason;
pin->up.table_id = ctx->table_id;
- pin->up.cookie = (ctx->rule
- ? rule_dpif_get_flow_cookie(ctx->rule)
- : OVS_BE64_MAX);
+ pin->up.cookie = ctx->rule_cookie;
- flow_get_metadata(&ctx->xin->flow, &pin->up.fmd);
+ flow_get_metadata(&ctx->xin->flow, &pin->up.flow_metadata);
pin->controller_id = controller_id;
pin->send_len = len;
}
}
ofproto_dpif_send_packet_in(ctx->xbridge->ofproto, pin);
- dpif_packet_delete(packet);
+ dp_packet_delete(packet);
}
+/* Called only when ctx->recirc_action_offset is set. */
static void
-compose_recirculate_action(struct xlate_ctx *ctx,
- const struct ofpact *ofpacts_base,
- const struct ofpact *ofpact_current,
- size_t ofpacts_base_len)
+compose_recirculate_action(struct xlate_ctx *ctx)
{
+ struct recirc_metadata md;
+ bool use_masked;
uint32_t id;
- int error;
- unsigned ofpacts_len;
- struct match match;
- struct rule *rule;
- struct ofpbuf ofpacts;
- ctx->exit = true;
+ use_masked = ctx->xbridge->support.masked_set_action;
+ ctx->xout->slow |= commit_odp_actions(&ctx->xin->flow, &ctx->base_flow,
+ ctx->xout->odp_actions,
+ &ctx->xout->wc, use_masked);
- ofpacts_len = ofpacts_base_len -
- ((uint8_t *)ofpact_current - (uint8_t *)ofpacts_base);
+ recirc_metadata_from_flow(&md, &ctx->xin->flow);
- if (ctx->rule) {
- id = rule_dpif_get_recirc_id(ctx->rule);
- } else {
- /* In the case where ctx has no rule then allocate a recirc id.
- * The life-cycle of this recirc id is managed by associating it
- * with the internal rule that is created to to handle
- * recirculation below.
- *
- * The known use-case of this is packet_out which
- * translates actions without a rule */
- id = ofproto_dpif_alloc_recirc_id(ctx->xbridge->ofproto);
- }
- if (!id) {
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
- VLOG_ERR_RL(&rl, "Failed to allocate recirculation id");
- return;
- }
+ ovs_assert(ctx->recirc_action_offset >= 0);
- match_init_catchall(&match);
- match_set_recirc_id(&match, id);
- ofpbuf_use_const(&ofpacts, ofpact_current, ofpacts_len);
- error = ofproto_dpif_add_internal_flow(ctx->xbridge->ofproto, &match,
- RECIRC_RULE_PRIORITY,
- RECIRC_TIMEOUT, &ofpacts, &rule);
- if (error) {
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
- VLOG_ERR_RL(&rl, "Failed to add post recirculation flow %s",
- match_to_string(&match, 0));
- return;
- }
- /* If ctx has no rule then associate the recirc id, which
- * was allocated above, with the internal rule. This allows
- * the recirc id to be released when the internal rule times out. */
- if (!ctx->rule) {
- rule_set_recirc_id(rule, id);
+ /* Only allocate recirculation ID if we have a packet. */
+ if (ctx->xin->packet) {
+ /* Allocate a unique recirc id for the given metadata state in the
+ * flow. The life-cycle of this recirc id is managed by associating it
+ * with the udpif key ('ukey') created for each new datapath flow. */
+ id = recirc_alloc_id_ctx(ctx->xbridge->ofproto, 0, &md, &ctx->stack,
+ ctx->recirc_action_offset,
+ ctx->action_set.size, ctx->action_set.data);
+ if (!id) {
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
+ VLOG_ERR_RL(&rl, "Failed to allocate recirculation id");
+ return;
+ }
+ xlate_out_add_recirc(ctx->xout, id);
+ } else {
+ /* Look up an existing recirc id for the given metadata state in the
+ * flow. No new reference is taken, as the ID is RCU protected and is
+ * only required temporarily for verification. */
+ id = recirc_find_id(ctx->xbridge->ofproto, 0, &md, &ctx->stack,
+ ctx->recirc_action_offset,
+ ctx->action_set.size, ctx->action_set.data);
+ /* We let zero 'id' to be used in the RECIRC action below, which will
+ * fail all revalidations as zero is not a valid recirculation ID. */
}
- ctx->xout->slow |= commit_odp_actions(&ctx->xin->flow, &ctx->base_flow,
- ctx->xout->odp_actions,
- &ctx->xout->wc,
- ctx->xbridge->masked_set_action);
nl_msg_put_u32(ctx->xout->odp_actions, OVS_ACTION_ATTR_RECIRC, id);
+
+ /* Undo changes done by recirculation. */
+ ctx->action_set.size = ctx->recirc_action_offset;
+ ctx->recirc_action_offset = -1;
+ ctx->last_unroll_offset = -1;
}
static void
n = flow_count_mpls_labels(flow, wc);
if (!n) {
+ bool use_masked = ctx->xbridge->support.masked_set_action;
+
ctx->xout->slow |= commit_odp_actions(flow, &ctx->base_flow,
ctx->xout->odp_actions,
- &ctx->xout->wc,
- ctx->xbridge->masked_set_action);
+ &ctx->xout->wc, use_masked);
} else if (n >= FLOW_MAX_MPLS_LABELS) {
if (ctx->xin->packet != NULL) {
static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
}
ctx->exit = true;
return;
- } else if (n >= ctx->xbridge->max_mpls_depth) {
- COVERAGE_INC(xlate_actions_mpls_overflow);
- ctx->xout->slow |= SLOW_ACTION;
}
flow_push_mpls(flow, n, mpls->ethertype, wc);
int n = flow_count_mpls_labels(flow, wc);
if (flow_pop_mpls(flow, n, eth_type, wc)) {
- if (ctx->xbridge->enable_recirc && !eth_type_mpls(eth_type)) {
+ if (ctx->xbridge->support.recirc) {
ctx->was_mpls = true;
}
} else if (n >= FLOW_MAX_MPLS_LABELS) {
switch (port) {
case OFPP_IN_PORT:
- compose_output_action(ctx, ctx->xin->flow.in_port.ofp_port);
+ compose_output_action(ctx, ctx->xin->flow.in_port.ofp_port, NULL);
break;
case OFPP_TABLE:
xlate_table_action(ctx, ctx->xin->flow.in_port.ofp_port,
flood_packets(ctx, true);
break;
case OFPP_CONTROLLER:
- execute_controller_action(ctx, max_len, OFPR_ACTION, 0);
+ execute_controller_action(ctx, max_len,
+ (ctx->in_group ? OFPR_GROUP
+ : ctx->in_action_set ? OFPR_ACTION_SET
+ : OFPR_ACTION),
+ 0);
break;
case OFPP_NONE:
break;
case OFPP_LOCAL:
default:
if (port != ctx->xin->flow.in_port.ofp_port) {
- compose_output_action(ctx, port);
+ compose_output_action(ctx, port, NULL);
} else {
xlate_report(ctx, "skipping output to input port");
}
/* Add datapath actions. */
flow_priority = ctx->xin->flow.skb_priority;
ctx->xin->flow.skb_priority = priority;
- compose_output_action(ctx, ofp_port);
+ compose_output_action(ctx, ofp_port, NULL);
ctx->xin->flow.skb_priority = flow_priority;
/* Update NetFlow output port. */
xlate_sample_action(struct xlate_ctx *ctx,
const struct ofpact_sample *os)
{
- union user_action_cookie cookie;
- /* Scale the probability from 16-bit to 32-bit while representing
- * the same percentage. */
- uint32_t probability = (os->probability << 16) | os->probability;
+ union user_action_cookie cookie;
+ /* Scale the probability from 16-bit to 32-bit while representing
+ * the same percentage. */
+ uint32_t probability = (os->probability << 16) | os->probability;
+ bool use_masked;
- if (!ctx->xbridge->variable_length_userdata) {
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
+ if (!ctx->xbridge->support.variable_length_userdata) {
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
- VLOG_ERR_RL(&rl, "ignoring NXAST_SAMPLE action because datapath "
- "lacks support (needs Linux 3.10+ or kernel module from "
- "OVS 1.11+)");
- return;
- }
+ VLOG_ERR_RL(&rl, "ignoring NXAST_SAMPLE action because datapath "
+ "lacks support (needs Linux 3.10+ or kernel module from "
+ "OVS 1.11+)");
+ return;
+ }
- ctx->xout->slow |= commit_odp_actions(&ctx->xin->flow, &ctx->base_flow,
- ctx->xout->odp_actions,
- &ctx->xout->wc,
- ctx->xbridge->masked_set_action);
+ use_masked = ctx->xbridge->support.masked_set_action;
+ ctx->xout->slow |= commit_odp_actions(&ctx->xin->flow, &ctx->base_flow,
+ ctx->xout->odp_actions,
+ &ctx->xout->wc, use_masked);
- compose_flow_sample_cookie(os->probability, os->collector_set_id,
- os->obs_domain_id, os->obs_point_id, &cookie);
- compose_sample_action(ctx->xbridge, ctx->xout->odp_actions, &ctx->xin->flow,
- probability, &cookie, sizeof cookie.flow_sample,
- ODPP_NONE);
+ compose_flow_sample_cookie(os->probability, os->collector_set_id,
+ os->obs_domain_id, os->obs_point_id, &cookie);
+ compose_sample_action(ctx->xbridge, ctx->xout->odp_actions,
+ &ctx->xin->flow, probability, &cookie,
+ sizeof cookie.flow_sample, ODPP_NONE);
}
static bool
uint64_t action_list_stub[1024 / 64];
struct ofpbuf action_list;
+ ctx->in_action_set = true;
ofpbuf_use_stub(&action_list, action_list_stub, sizeof action_list_stub);
ofpacts_execute_action_set(&action_list, &ctx->action_set);
- do_xlate_actions(ofpbuf_data(&action_list), ofpbuf_size(&action_list), ctx);
+ /* Clear the action set, as it is not needed any more. */
+ ofpbuf_clear(&ctx->action_set);
+ do_xlate_actions(action_list.data, action_list.size, ctx);
+ ctx->in_action_set = false;
ofpbuf_uninit(&action_list);
}
-static bool
-ofpact_needs_recirculation_after_mpls(const struct xlate_ctx *ctx,
- const struct ofpact *a)
+static void
+recirc_put_unroll_xlate(struct xlate_ctx *ctx)
{
- struct flow_wildcards *wc = &ctx->xout->wc;
- struct flow *flow = &ctx->xin->flow;
+ struct ofpact_unroll_xlate *unroll;
- switch (a->type) {
- case OFPACT_OUTPUT:
- case OFPACT_GROUP:
- case OFPACT_CONTROLLER:
- case OFPACT_STRIP_VLAN:
- case OFPACT_SET_VLAN_PCP:
- case OFPACT_SET_VLAN_VID:
- case OFPACT_ENQUEUE:
- case OFPACT_PUSH_VLAN:
- case OFPACT_SET_ETH_SRC:
- case OFPACT_SET_ETH_DST:
- case OFPACT_SET_TUNNEL:
- case OFPACT_SET_QUEUE:
- case OFPACT_POP_QUEUE:
- case OFPACT_POP_MPLS:
- case OFPACT_DEC_MPLS_TTL:
- case OFPACT_SET_MPLS_TTL:
- case OFPACT_SET_MPLS_TC:
- case OFPACT_SET_MPLS_LABEL:
- case OFPACT_NOTE:
- case OFPACT_OUTPUT_REG:
- case OFPACT_EXIT:
- case OFPACT_METER:
- case OFPACT_WRITE_METADATA:
- case OFPACT_WRITE_ACTIONS:
- case OFPACT_CLEAR_ACTIONS:
- case OFPACT_SAMPLE:
- return false;
+ unroll = ctx->last_unroll_offset < 0
+ ? NULL
+ : ALIGNED_CAST(struct ofpact_unroll_xlate *,
+ (char *)ctx->action_set.data + ctx->last_unroll_offset);
- case OFPACT_SET_IPV4_SRC:
- case OFPACT_SET_IPV4_DST:
- case OFPACT_SET_IP_DSCP:
- case OFPACT_SET_IP_ECN:
- case OFPACT_SET_IP_TTL:
- case OFPACT_SET_L4_SRC_PORT:
- case OFPACT_SET_L4_DST_PORT:
- case OFPACT_RESUBMIT:
- case OFPACT_STACK_PUSH:
- case OFPACT_STACK_POP:
- case OFPACT_DEC_TTL:
- case OFPACT_MULTIPATH:
- case OFPACT_BUNDLE:
- case OFPACT_LEARN:
- case OFPACT_FIN_TIMEOUT:
- case OFPACT_GOTO_TABLE:
- return true;
+ /* Restore the table_id and rule cookie for a potential PACKET
+ * IN if needed. */
+ if (!unroll ||
+ (ctx->table_id != unroll->rule_table_id
+ || ctx->rule_cookie != unroll->rule_cookie)) {
- case OFPACT_REG_MOVE:
- return (mf_is_l3_or_higher(ofpact_get_REG_MOVE(a)->dst.field) ||
- mf_is_l3_or_higher(ofpact_get_REG_MOVE(a)->src.field));
+ ctx->last_unroll_offset = ctx->action_set.size;
+ unroll = ofpact_put_UNROLL_XLATE(&ctx->action_set);
+ unroll->rule_table_id = ctx->table_id;
+ unroll->rule_cookie = ctx->rule_cookie;
+ }
+}
- case OFPACT_SET_FIELD:
- return mf_is_l3_or_higher(ofpact_get_SET_FIELD(a)->field);
- case OFPACT_PUSH_MPLS:
- /* Recirculate if it is an IP packet with a zero ttl. This may
- * indicate that the packet was previously MPLS and an MPLS pop action
- * converted it to IP. In this case recirculating should reveal the IP
- * TTL which is used as the basis for a new MPLS LSE. */
- return (!flow_count_mpls_labels(flow, wc)
- && flow->nw_ttl == 0
- && is_ip_any(flow));
- }
+/* Copy remaining actions to the action_set to be executed after recirculation.
+ * UNROLL_XLATE action is inserted, if not already done so, before actions that
+ * may generate PACKET_INs from the current table and without matching another
+ * rule. */
+static void
+recirc_unroll_actions(const struct ofpact *ofpacts, size_t ofpacts_len,
+ struct xlate_ctx *ctx)
+{
+ const struct ofpact *a;
+
+ OFPACT_FOR_EACH (a, ofpacts, ofpacts_len) {
+ switch (a->type) {
+ /* May generate PACKET INs. */
+ case OFPACT_OUTPUT_REG:
+ case OFPACT_GROUP:
+ case OFPACT_OUTPUT:
+ case OFPACT_CONTROLLER:
+ case OFPACT_DEC_MPLS_TTL:
+ case OFPACT_DEC_TTL:
+ recirc_put_unroll_xlate(ctx);
+ break;
- OVS_NOT_REACHED();
+ /* These may not generate PACKET INs. */
+ case OFPACT_SET_TUNNEL:
+ case OFPACT_REG_MOVE:
+ case OFPACT_SET_FIELD:
+ case OFPACT_STACK_PUSH:
+ case OFPACT_STACK_POP:
+ case OFPACT_LEARN:
+ case OFPACT_WRITE_METADATA:
+ case OFPACT_RESUBMIT: /* May indirectly generate PACKET INs, */
+ case OFPACT_GOTO_TABLE: /* but from a different table and rule. */
+ case OFPACT_ENQUEUE:
+ case OFPACT_SET_VLAN_VID:
+ case OFPACT_SET_VLAN_PCP:
+ case OFPACT_STRIP_VLAN:
+ case OFPACT_PUSH_VLAN:
+ case OFPACT_SET_ETH_SRC:
+ case OFPACT_SET_ETH_DST:
+ case OFPACT_SET_IPV4_SRC:
+ case OFPACT_SET_IPV4_DST:
+ case OFPACT_SET_IP_DSCP:
+ case OFPACT_SET_IP_ECN:
+ case OFPACT_SET_IP_TTL:
+ case OFPACT_SET_L4_SRC_PORT:
+ case OFPACT_SET_L4_DST_PORT:
+ case OFPACT_SET_QUEUE:
+ case OFPACT_POP_QUEUE:
+ case OFPACT_PUSH_MPLS:
+ case OFPACT_POP_MPLS:
+ case OFPACT_SET_MPLS_LABEL:
+ case OFPACT_SET_MPLS_TC:
+ case OFPACT_SET_MPLS_TTL:
+ case OFPACT_MULTIPATH:
+ case OFPACT_BUNDLE:
+ case OFPACT_EXIT:
+ case OFPACT_UNROLL_XLATE:
+ case OFPACT_FIN_TIMEOUT:
+ case OFPACT_CLEAR_ACTIONS:
+ case OFPACT_WRITE_ACTIONS:
+ case OFPACT_METER:
+ case OFPACT_SAMPLE:
+ break;
+
+ /* These need not be copied for restoration. */
+ case OFPACT_NOTE:
+ case OFPACT_CONJUNCTION:
+ continue;
+ }
+ /* Copy the action over. */
+ ofpbuf_put(&ctx->action_set, a, OFPACT_ALIGN(a->len));
+ }
}
+#define CHECK_MPLS_RECIRCULATION() \
+ if (ctx->was_mpls) { \
+ ctx_trigger_recirculation(ctx); \
+ break; \
+ }
+#define CHECK_MPLS_RECIRCULATION_IF(COND) \
+ if (COND) { \
+ CHECK_MPLS_RECIRCULATION(); \
+ }
+
static void
do_xlate_actions(const struct ofpact *ofpacts, size_t ofpacts_len,
struct xlate_ctx *ctx)
const struct mf_field *mf;
if (ctx->exit) {
+ /* Check if need to store the remaining actions for later
+ * execution. */
+ if (exit_recirculates(ctx)) {
+ recirc_unroll_actions(a, OFPACT_ALIGN(ofpacts_len -
+ ((uint8_t *)a -
+ (uint8_t *)ofpacts)),
+ ctx);
+ }
break;
}
- if (ctx->was_mpls && ofpact_needs_recirculation_after_mpls(ctx, a)) {
- compose_recirculate_action(ctx, ofpacts, a, ofpacts_len);
- return;
- }
-
switch (a->type) {
case OFPACT_OUTPUT:
xlate_output_action(ctx, ofpact_get_OUTPUT(a)->port,
case OFPACT_GROUP:
if (xlate_group_action(ctx, ofpact_get_GROUP(a)->group_id)) {
+ /* Group could not be found. */
return;
}
break;
break;
case OFPACT_SET_IPV4_SRC:
+ CHECK_MPLS_RECIRCULATION();
if (flow->dl_type == htons(ETH_TYPE_IP)) {
memset(&wc->masks.nw_src, 0xff, sizeof wc->masks.nw_src);
flow->nw_src = ofpact_get_SET_IPV4_SRC(a)->ipv4;
break;
case OFPACT_SET_IPV4_DST:
+ CHECK_MPLS_RECIRCULATION();
if (flow->dl_type == htons(ETH_TYPE_IP)) {
memset(&wc->masks.nw_dst, 0xff, sizeof wc->masks.nw_dst);
flow->nw_dst = ofpact_get_SET_IPV4_DST(a)->ipv4;
break;
case OFPACT_SET_IP_DSCP:
+ CHECK_MPLS_RECIRCULATION();
if (is_ip_any(flow)) {
wc->masks.nw_tos |= IP_DSCP_MASK;
flow->nw_tos &= ~IP_DSCP_MASK;
break;
case OFPACT_SET_IP_ECN:
+ CHECK_MPLS_RECIRCULATION();
if (is_ip_any(flow)) {
wc->masks.nw_tos |= IP_ECN_MASK;
flow->nw_tos &= ~IP_ECN_MASK;
break;
case OFPACT_SET_IP_TTL:
+ CHECK_MPLS_RECIRCULATION();
if (is_ip_any(flow)) {
wc->masks.nw_ttl = 0xff;
flow->nw_ttl = ofpact_get_SET_IP_TTL(a)->ttl;
break;
case OFPACT_SET_L4_SRC_PORT:
+ CHECK_MPLS_RECIRCULATION();
if (is_ip_any(flow) && !(flow->nw_frag & FLOW_NW_FRAG_LATER)) {
memset(&wc->masks.nw_proto, 0xff, sizeof wc->masks.nw_proto);
memset(&wc->masks.tp_src, 0xff, sizeof wc->masks.tp_src);
break;
case OFPACT_SET_L4_DST_PORT:
+ CHECK_MPLS_RECIRCULATION();
if (is_ip_any(flow) && !(flow->nw_frag & FLOW_NW_FRAG_LATER)) {
memset(&wc->masks.nw_proto, 0xff, sizeof wc->masks.nw_proto);
memset(&wc->masks.tp_dst, 0xff, sizeof wc->masks.tp_dst);
break;
case OFPACT_REG_MOVE:
+ CHECK_MPLS_RECIRCULATION_IF(
+ mf_is_l3_or_higher(ofpact_get_REG_MOVE(a)->dst.field) ||
+ mf_is_l3_or_higher(ofpact_get_REG_MOVE(a)->src.field));
nxm_execute_reg_move(ofpact_get_REG_MOVE(a), flow, wc);
break;
case OFPACT_SET_FIELD:
+ CHECK_MPLS_RECIRCULATION_IF(
+ mf_is_l3_or_higher(ofpact_get_SET_FIELD(a)->field));
set_field = ofpact_get_SET_FIELD(a);
mf = set_field->field;
break;
case OFPACT_STACK_PUSH:
+ CHECK_MPLS_RECIRCULATION_IF(
+ mf_is_l3_or_higher(ofpact_get_STACK_PUSH(a)->subfield.field));
nxm_execute_stack_push(ofpact_get_STACK_PUSH(a), flow, wc,
&ctx->stack);
break;
case OFPACT_STACK_POP:
+ CHECK_MPLS_RECIRCULATION_IF(
+ mf_is_l3_or_higher(ofpact_get_STACK_POP(a)->subfield.field));
nxm_execute_stack_pop(ofpact_get_STACK_POP(a), flow, wc,
&ctx->stack);
break;
case OFPACT_PUSH_MPLS:
+ /* Recirculate if it is an IP packet with a zero ttl. This may
+ * indicate that the packet was previously MPLS and an MPLS pop
+ * action converted it to IP. In this case recirculating should
+ * reveal the IP TTL which is used as the basis for a new MPLS
+ * LSE. */
+ CHECK_MPLS_RECIRCULATION_IF(
+ !flow_count_mpls_labels(flow, wc)
+ && flow->nw_ttl == 0
+ && is_ip_any(flow));
compose_mpls_push_action(ctx, ofpact_get_PUSH_MPLS(a));
break;
case OFPACT_POP_MPLS:
+ CHECK_MPLS_RECIRCULATION();
compose_mpls_pop_action(ctx, ofpact_get_POP_MPLS(a)->ethertype);
break;
case OFPACT_SET_MPLS_LABEL:
+ CHECK_MPLS_RECIRCULATION();
compose_set_mpls_label_action(
ctx, ofpact_get_SET_MPLS_LABEL(a)->label);
- break;
+ break;
case OFPACT_SET_MPLS_TC:
+ CHECK_MPLS_RECIRCULATION();
compose_set_mpls_tc_action(ctx, ofpact_get_SET_MPLS_TC(a)->tc);
break;
case OFPACT_SET_MPLS_TTL:
+ CHECK_MPLS_RECIRCULATION();
compose_set_mpls_ttl_action(ctx, ofpact_get_SET_MPLS_TTL(a)->ttl);
break;
case OFPACT_DEC_MPLS_TTL:
+ CHECK_MPLS_RECIRCULATION();
if (compose_dec_mpls_ttl_action(ctx)) {
return;
}
break;
case OFPACT_DEC_TTL:
+ CHECK_MPLS_RECIRCULATION();
wc->masks.nw_ttl = 0xff;
if (compose_dec_ttl(ctx, ofpact_get_DEC_TTL(a))) {
return;
break;
case OFPACT_MULTIPATH:
+ CHECK_MPLS_RECIRCULATION();
multipath_execute(ofpact_get_MULTIPATH(a), flow, wc);
break;
case OFPACT_BUNDLE:
+ CHECK_MPLS_RECIRCULATION();
xlate_bundle_action(ctx, ofpact_get_BUNDLE(a));
break;
break;
case OFPACT_LEARN:
+ CHECK_MPLS_RECIRCULATION();
xlate_learn_action(ctx, ofpact_get_LEARN(a));
break;
+ case OFPACT_CONJUNCTION: {
+ /* A flow with a "conjunction" action represents part of a special
+ * kind of "set membership match". Such a flow should not actually
+ * get executed, but it could via, say, a "packet-out", even though
+ * that wouldn't be useful. Log it to help debugging. */
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
+ VLOG_INFO_RL(&rl, "executing no-op conjunction action");
+ break;
+ }
+
case OFPACT_EXIT:
ctx->exit = true;
break;
+ case OFPACT_UNROLL_XLATE: {
+ struct ofpact_unroll_xlate *unroll = ofpact_get_UNROLL_XLATE(a);
+
+ /* Restore translation context data that was stored earlier. */
+ ctx->table_id = unroll->rule_table_id;
+ ctx->rule_cookie = unroll->rule_cookie;
+ break;
+ }
case OFPACT_FIN_TIMEOUT:
+ CHECK_MPLS_RECIRCULATION();
memset(&wc->masks.nw_proto, 0xff, sizeof wc->masks.nw_proto);
ctx->xout->has_fin_timeout = true;
xlate_fin_timeout(ctx, ofpact_get_FIN_TIMEOUT(a));
xlate_sample_action(ctx, ofpact_get_SAMPLE(a));
break;
}
+
+ /* Check if need to store this and the remaining actions for later
+ * execution. */
+ if (ctx->exit && ctx_first_recirculation_action(ctx)) {
+ recirc_unroll_actions(a, OFPACT_ALIGN(ofpacts_len -
+ ((uint8_t *)a -
+ (uint8_t *)ofpacts)),
+ ctx);
+ break;
+ }
}
}
xlate_in_init(struct xlate_in *xin, struct ofproto_dpif *ofproto,
const struct flow *flow, ofp_port_t in_port,
struct rule_dpif *rule, uint16_t tcp_flags,
- const struct ofpbuf *packet)
+ const struct dp_packet *packet)
{
xin->ofproto = ofproto;
xin->flow = *flow;
xin->resubmit_stats = NULL;
xin->skip_wildcards = false;
xin->odp_actions = NULL;
+
+ /* Do recirc lookup. */
+ xin->recirc = flow->recirc_id
+ ? recirc_id_node_find(flow->recirc_id)
+ : NULL;
}
void
xlate_out_uninit(struct xlate_out *xout)
{
- if (xout && xout->odp_actions == &xout->odp_actions_buf) {
- ofpbuf_uninit(xout->odp_actions);
+ if (xout) {
+ if (xout->odp_actions == &xout->odp_actions_buf) {
+ ofpbuf_uninit(xout->odp_actions);
+ }
+ xlate_out_free_recircs(xout);
}
}
dst->odp_actions = &dst->odp_actions_buf;
ofpbuf_use_stub(dst->odp_actions, dst->odp_actions_stub,
sizeof dst->odp_actions_stub);
- ofpbuf_put(dst->odp_actions, ofpbuf_data(src->odp_actions),
- ofpbuf_size(src->odp_actions));
+ ofpbuf_put(dst->odp_actions, src->odp_actions->data, src->odp_actions->size);
}
\f
static struct skb_priority_to_dscp *
const struct nlattr *a;
unsigned int left;
- NL_ATTR_FOR_EACH_UNSAFE (a, left, ofpbuf_data(ctx->xout->odp_actions),
- ofpbuf_size(ctx->xout->odp_actions)) {
+ NL_ATTR_FOR_EACH_UNSAFE (a, left, ctx->xout->odp_actions->data,
+ ctx->xout->odp_actions->size) {
if (nl_attr_type(a) == OVS_ACTION_ATTR_OUTPUT
&& nl_attr_get_odp_port(a) == local_odp_port) {
return true;
size_t left;
int n = 0;
- NL_ATTR_FOR_EACH_UNSAFE (a, left, ofpbuf_data(odp_actions),
- ofpbuf_size(odp_actions)) {
+ NL_ATTR_FOR_EACH_UNSAFE (a, left, odp_actions->data, odp_actions->size) {
if (a->nla_type == OVS_ACTION_ATTR_OUTPUT) {
n++;
}
too_many_output_actions(const struct ofpbuf *odp_actions OVS_UNUSED)
{
#ifdef __linux__
- return (ofpbuf_size(odp_actions) / NL_A_U32_SIZE > netdev_max_backlog()
+ return (odp_actions->size / NL_A_U32_SIZE > netdev_max_backlog()
&& count_output_actions(odp_actions) > netdev_max_backlog());
#else
/* OSes other than Linux might have similar limits, but we don't know how
#endif
}
-/* Translates the 'ofpacts_len' bytes of "struct ofpacts" starting at 'ofpacts'
- * into datapath actions in 'odp_actions', using 'ctx'.
- *
+/* Translates the flow, actions, or rule in 'xin' into datapath actions in
+ * 'xout'.
* The caller must take responsibility for eventually freeing 'xout', with
* xlate_out_uninit(). */
void
xlate_actions(struct xlate_in *xin, struct xlate_out *xout)
{
struct xlate_cfg *xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
- struct flow_wildcards *wc = &xout->wc;
+ struct flow_wildcards *wc = NULL;
struct flow *flow = &xin->flow;
struct rule_dpif *rule = NULL;
enum slow_path_reason special;
const struct ofpact *ofpacts;
+ struct xbridge *xbridge;
struct xport *in_port;
struct flow orig_flow;
struct xlate_ctx ctx;
ctx.xout->has_fin_timeout = false;
ctx.xout->nf_output_iface = NF_OUT_DROP;
ctx.xout->mirrors = 0;
+ ctx.xout->n_recircs = 0;
xout->odp_actions = xin->odp_actions;
if (!xout->odp_actions) {
}
ofpbuf_reserve(xout->odp_actions, NL_A_U32_SIZE);
- ctx.xbridge = xbridge_lookup(xcfg, xin->ofproto);
- if (!ctx.xbridge) {
+ xbridge = xbridge_lookup(xcfg, xin->ofproto);
+ if (!xbridge) {
return;
}
-
+ /* 'ctx.xbridge' may be changed by action processing, whereas 'xbridge'
+ * will remain set on the original input bridge. */
+ ctx.xbridge = xbridge;
ctx.rule = xin->rule;
ctx.base_flow = *flow;
memset(&ctx.base_flow.tunnel, 0, sizeof ctx.base_flow.tunnel);
ctx.orig_tunnel_ip_dst = flow->tunnel.ip_dst;
- flow_wildcards_init_catchall(wc);
- memset(&wc->masks.in_port, 0xff, sizeof wc->masks.in_port);
- memset(&wc->masks.dl_type, 0xff, sizeof wc->masks.dl_type);
- if (is_ip_any(flow)) {
- wc->masks.nw_frag |= FLOW_NW_FRAG_MASK;
+ if (!xin->skip_wildcards) {
+ wc = &xout->wc;
+ flow_wildcards_init_catchall(wc);
+ memset(&wc->masks.in_port, 0xff, sizeof wc->masks.in_port);
+ memset(&wc->masks.dl_type, 0xff, sizeof wc->masks.dl_type);
+ if (is_ip_any(flow)) {
+ wc->masks.nw_frag |= FLOW_NW_FRAG_MASK;
+ }
+ if (xbridge->support.recirc) {
+ /* Always exactly match recirc_id when datapath supports
+ * recirculation. */
+ wc->masks.recirc_id = UINT32_MAX;
+ }
+ if (xbridge->netflow) {
+ netflow_mask_wc(flow, wc);
+ }
}
is_icmp = is_icmpv4(flow) || is_icmpv6(flow);
tnl_may_send = tnl_xlate_init(&ctx.base_flow, flow, wc);
- if (ctx.xbridge->netflow) {
- netflow_mask_wc(flow, wc);
- }
ctx.recurse = 0;
ctx.resubmits = 0;
ctx.in_group = false;
+ ctx.in_action_set = false;
ctx.orig_skb_priority = flow->skb_priority;
ctx.table_id = 0;
+ ctx.rule_cookie = OVS_BE64_MAX;
ctx.exit = false;
- ctx.use_recirc = false;
ctx.was_mpls = false;
+ ctx.recirc_action_offset = -1;
+ ctx.last_unroll_offset = -1;
+
+ ctx.action_set_has_group = false;
+ ofpbuf_use_stub(&ctx.action_set,
+ ctx.action_set_stub, sizeof ctx.action_set_stub);
+
+ ofpbuf_use_stub(&ctx.stack, ctx.init_stack, sizeof ctx.init_stack);
+
+ /* The in_port of the original packet before recirculation. */
+ in_port = get_ofp_port(xbridge, flow->in_port.ofp_port);
+
+ if (xin->recirc) {
+ const struct recirc_id_node *recirc = xin->recirc;
+
+ if (xin->ofpacts_len > 0 || ctx.rule) {
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
+
+ VLOG_WARN_RL(&rl, "Recirculation conflict (%s)!",
+ xin->ofpacts_len > 0
+ ? "actions"
+ : "rule");
+ return;
+ }
+
+ /* Set the bridge for post-recirculation processing if needed. */
+ if (ctx.xbridge->ofproto != recirc->ofproto) {
+ struct xlate_cfg *xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
+ const struct xbridge *new_bridge = xbridge_lookup(xcfg,
+ recirc->ofproto);
+
+ if (OVS_UNLIKELY(!new_bridge)) {
+ /* Drop the packet if the bridge cannot be found. */
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
+ VLOG_WARN_RL(&rl, "Recirculation bridge no longer exists.");
+ return;
+ }
+ ctx.xbridge = new_bridge;
+ }
+
+ /* Set the post-recirculation table id. Note: A table lookup is done
+ * only if there are no post-recirculation actions. */
+ ctx.table_id = recirc->table_id;
+
+ /* Restore pipeline metadata. May change flow's in_port and other
+ * metadata to the values that existed when recirculation was
+ * triggered. */
+ recirc_metadata_to_flow(&recirc->metadata, flow);
+
+ /* Restore stack, if any. */
+ if (recirc->stack) {
+ ofpbuf_put(&ctx.stack, recirc->stack->data, recirc->stack->size);
+ }
+
+ /* Restore action set, if any. */
+ if (recirc->action_set_len) {
+ const struct ofpact *a;
+
+ ofpbuf_put(&ctx.action_set, recirc->ofpacts,
+ recirc->action_set_len);
+
+ OFPACT_FOR_EACH(a, recirc->ofpacts, recirc->action_set_len) {
+ if (a->type == OFPACT_GROUP) {
+ ctx.action_set_has_group = true;
+ break;
+ }
+ }
+ }
+
+ /* Restore recirculation actions. If there are no actions, processing
+ * will start with a lookup in the table set above. */
+ if (recirc->ofpacts_len > recirc->action_set_len) {
+ xin->ofpacts_len = recirc->ofpacts_len - recirc->action_set_len;
+ xin->ofpacts = recirc->ofpacts +
+ recirc->action_set_len / sizeof *recirc->ofpacts;
+ }
+ } else if (OVS_UNLIKELY(flow->recirc_id)) {
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
+
+ VLOG_WARN_RL(&rl, "Recirculation context not found for ID %"PRIx32,
+ flow->recirc_id);
+ return;
+ }
+ /* The bridge is now known so obtain its table version. */
+ ctx.tables_version = ofproto_dpif_get_tables_version(ctx.xbridge->ofproto);
if (!xin->ofpacts && !ctx.rule) {
- rule = rule_dpif_lookup(ctx.xbridge->ofproto, flow,
- xin->skip_wildcards ? NULL : wc,
- ctx.xin->xcache != NULL,
- ctx.xin->resubmit_stats, &ctx.table_id);
+ rule = rule_dpif_lookup_from_table(ctx.xbridge->ofproto,
+ ctx.tables_version, flow, wc,
+ ctx.xin->xcache != NULL,
+ ctx.xin->resubmit_stats,
+ &ctx.table_id,
+ flow->in_port.ofp_port, true, true);
if (ctx.xin->resubmit_stats) {
rule_dpif_credit_stats(rule, ctx.xin->resubmit_stats);
}
ofpacts = actions->ofpacts;
ofpacts_len = actions->ofpacts_len;
+
+ ctx.rule_cookie = rule_dpif_get_flow_cookie(ctx.rule);
} else {
OVS_NOT_REACHED();
}
- ofpbuf_use_stub(&ctx.stack, ctx.init_stack, sizeof ctx.init_stack);
-
- ctx.action_set_has_group = false;
- ofpbuf_use_stub(&ctx.action_set,
- ctx.action_set_stub, sizeof ctx.action_set_stub);
-
- if (mbridge_has_mirrors(ctx.xbridge->mbridge)) {
+ if (mbridge_has_mirrors(xbridge->mbridge)) {
/* Do this conditionally because the copy is expensive enough that it
* shows up in profiles. */
orig_flow = *flow;
}
- in_port = get_ofp_port(ctx.xbridge, flow->in_port.ofp_port);
- if (in_port && in_port->is_tunnel) {
+ /* Tunnel stats only for non-recirculated packets. */
+ if (!xin->recirc && in_port && in_port->is_tunnel) {
if (ctx.xin->resubmit_stats) {
netdev_vport_inc_rx(in_port->netdev, ctx.xin->resubmit_stats);
if (in_port->bfd) {
}
}
- special = process_special(&ctx, flow, in_port, ctx.xin->packet);
- if (special) {
+ /* Do not perform special processing on recirculated packets,
+ * as recirculated packets are not really received by the bridge. */
+ if (!xin->recirc &&
+ (special = process_special(&ctx, flow, in_port, ctx.xin->packet))) {
ctx.xout->slow |= special;
} else {
size_t sample_actions_len;
if (flow->in_port.ofp_port
- != vsp_realdev_to_vlandev(ctx.xbridge->ofproto,
+ != vsp_realdev_to_vlandev(xbridge->ofproto,
flow->in_port.ofp_port,
flow->vlan_tci)) {
ctx.base_flow.vlan_tci = 0;
}
- add_sflow_action(&ctx);
- add_ipfix_action(&ctx);
- sample_actions_len = ofpbuf_size(ctx.xout->odp_actions);
+ /* Sampling is done only for packets really received by the bridge. */
+ if (!xin->recirc) {
+ add_sflow_action(&ctx);
+ add_ipfix_action(&ctx);
+ sample_actions_len = ctx.xout->odp_actions->size;
+ } else {
+ sample_actions_len = 0;
+ }
if (tnl_may_send && (!in_port || may_receive(in_port, &ctx))) {
do_xlate_actions(ofpacts, ofpacts_len, &ctx);
* packet, so drop it now if forwarding is disabled. */
if (in_port && (!xport_stp_forward_state(in_port) ||
!xport_rstp_forward_state(in_port))) {
- ofpbuf_set_size(ctx.xout->odp_actions, sample_actions_len);
+ /* Drop all actions added by do_xlate_actions() above. */
+ ctx.xout->odp_actions->size = sample_actions_len;
+
+ /* Undo changes that may have been done for recirculation. */
+ if (exit_recirculates(&ctx)) {
+ ctx.action_set.size = ctx.recirc_action_offset;
+ ctx.recirc_action_offset = -1;
+ ctx.last_unroll_offset = -1;
+ }
+ } else if (ctx.action_set.size) {
+ /* Translate action set only if not dropping the packet and
+ * not recirculating. */
+ if (!exit_recirculates(&ctx)) {
+ xlate_action_set(&ctx);
+ }
+ }
+ /* Check if need to recirculate. */
+ if (exit_recirculates(&ctx)) {
+ compose_recirculate_action(&ctx);
}
}
- if (ofpbuf_size(&ctx.action_set)) {
- xlate_action_set(&ctx);
- }
-
- if (ctx.xbridge->has_in_band
+ /* Output only fully processed packets. */
+ if (!exit_recirculates(&ctx)
+ && xbridge->has_in_band
&& in_band_must_output_to_local_port(flow)
&& !actions_output_to_local_port(&ctx)) {
- compose_output_action(&ctx, OFPP_LOCAL);
+ compose_output_action(&ctx, OFPP_LOCAL, NULL);
}
- fix_sflow_action(&ctx);
-
- if (mbridge_has_mirrors(ctx.xbridge->mbridge)) {
+ if (!xin->recirc) {
+ fix_sflow_action(&ctx);
+ }
+ /* Only mirror fully processed packets. */
+ if (!exit_recirculates(&ctx)
+ && mbridge_has_mirrors(xbridge->mbridge)) {
add_mirror_actions(&ctx, &orig_flow);
}
}
- if (nl_attr_oversized(ofpbuf_size(ctx.xout->odp_actions))) {
+ if (nl_attr_oversized(ctx.xout->odp_actions->size)) {
/* These datapath actions are too big for a Netlink attribute, so we
* can't hand them to the kernel directly. dpif_execute() can execute
* them one by one with help, so just mark the result as SLOW_ACTION to
ctx.xout->slow |= SLOW_ACTION;
}
- if (mbridge_has_mirrors(ctx.xbridge->mbridge)) {
+ /* Update mirror stats only for packets really received by the bridge. */
+ if (!xin->recirc && mbridge_has_mirrors(xbridge->mbridge)) {
if (ctx.xin->resubmit_stats) {
- mirror_update_stats(ctx.xbridge->mbridge, xout->mirrors,
+ mirror_update_stats(xbridge->mbridge, xout->mirrors,
ctx.xin->resubmit_stats->n_packets,
ctx.xin->resubmit_stats->n_bytes);
}
struct xc_entry *entry;
entry = xlate_cache_add_entry(ctx.xin->xcache, XC_MIRROR);
- entry->u.mirror.mbridge = mbridge_ref(ctx.xbridge->mbridge);
+ entry->u.mirror.mbridge = mbridge_ref(xbridge->mbridge);
entry->u.mirror.mirrors = xout->mirrors;
}
}
- if (ctx.xbridge->netflow) {
+ /* Do netflow only for packets really received by the bridge. */
+ if (!xin->recirc && xbridge->netflow) {
/* Only update netflow if we don't have controller flow. We don't
* report NetFlow expiration messages for such facets because they
* are just part of the control logic for the network, not real
|| ofpacts->type != OFPACT_CONTROLLER
|| ofpact_next(ofpacts) < ofpact_end(ofpacts, ofpacts_len)) {
if (ctx.xin->resubmit_stats) {
- netflow_flow_update(ctx.xbridge->netflow, flow,
+ netflow_flow_update(xbridge->netflow, flow,
xout->nf_output_iface,
ctx.xin->resubmit_stats);
}
struct xc_entry *entry;
entry = xlate_cache_add_entry(ctx.xin->xcache, XC_NETFLOW);
- entry->u.nf.netflow = netflow_ref(ctx.xbridge->netflow);
+ entry->u.nf.netflow = netflow_ref(xbridge->netflow);
entry->u.nf.flow = xmemdup(flow, sizeof *flow);
entry->u.nf.iface = xout->nf_output_iface;
}
ofpbuf_uninit(&ctx.stack);
ofpbuf_uninit(&ctx.action_set);
- /* Clear the metadata and register wildcard masks, because we won't
- * use non-header fields as part of the cache. */
- flow_wildcards_clear_non_packet_fields(wc);
-
- /* ICMPv4 and ICMPv6 have 8-bit "type" and "code" fields. struct flow uses
- * the low 8 bits of the 16-bit tp_src and tp_dst members to represent
- * these fields. The datapath interface, on the other hand, represents
- * them with just 8 bits each. This means that if the high 8 bits of the
- * masks for these fields somehow become set, then they will get chopped
- * off by a round trip through the datapath, and revalidation will spot
- * that as an inconsistency and delete the flow. Avoid the problem here by
- * making sure that only the low 8 bits of either field can be unwildcarded
- * for ICMP.
- */
- if (is_icmp) {
- wc->masks.tp_src &= htons(UINT8_MAX);
- wc->masks.tp_dst &= htons(UINT8_MAX);
+ if (wc) {
+ /* Clear the metadata and register wildcard masks, because we won't
+ * use non-header fields as part of the cache. */
+ flow_wildcards_clear_non_packet_fields(wc);
+
+ /* ICMPv4 and ICMPv6 have 8-bit "type" and "code" fields. struct flow
+ * uses the low 8 bits of the 16-bit tp_src and tp_dst members to
+ * represent these fields. The datapath interface, on the other hand,
+ * represents them with just 8 bits each. This means that if the high
+ * 8 bits of the masks for these fields somehow become set, then they
+ * will get chopped off by a round trip through the datapath, and
+ * revalidation will spot that as an inconsistency and delete the flow.
+ * Avoid the problem here by making sure that only the low 8 bits of
+ * either field can be unwildcarded for ICMP.
+ */
+ if (is_icmp) {
+ wc->masks.tp_src &= htons(UINT8_MAX);
+ wc->masks.tp_dst &= htons(UINT8_MAX);
+ }
+ /* VLAN_TCI CFI bit must be matched if any of the TCI is matched. */
+ if (wc->masks.vlan_tci) {
+ wc->masks.vlan_tci |= htons(VLAN_CFI);
+ }
}
}
* May modify 'packet'.
* Returns 0 if successful, otherwise a positive errno value. */
int
-xlate_send_packet(const struct ofport_dpif *ofport, struct ofpbuf *packet)
+xlate_send_packet(const struct ofport_dpif *ofport, struct dp_packet *packet)
{
struct xlate_cfg *xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
struct xport *xport;
ofpact_init(&output.ofpact, OFPACT_OUTPUT, sizeof output);
/* Use OFPP_NONE as the in_port to avoid special packet processing. */
- flow_extract(packet, NULL, &flow);
+ flow_extract(packet, &flow);
flow.in_port.ofp_port = OFPP_NONE;
xport = xport_lookup(xcfg, ofport);
projects / cascardo / ovs.git / blobdiff