#include <sys/socket.h>
#include <netinet/in.h>
-#include "tnl-arp-cache.h"
+#include "tnl-neigh-cache.h"
#include "bfd.h"
#include "bitmap.h"
#include "bond.h"
* which might lead to an infinite loop. This could happen easily
* if a tunnel is marked as 'ip_remote=flow', and the flow does not
* actually set the tun_dst field. */
- ovs_be32 orig_tunnel_ip_dst;
+ struct in6_addr orig_tunnel_ipv6_dst;
/* Stack for the push and pop actions. Each stack element is of type
* "union mf_subvalue". */
* state from the datapath should be honored after recirculation. */
bool conntracked;
+ /* Pointer to an embedded NAT action in a conntrack action, or NULL. */
+ struct ofpact_nat *ct_nat_action;
+
/* OpenFlow 1.1+ action set.
*
* 'action_set' accumulates "struct ofpact"s added by OFPACT_WRITE_ACTIONS.
* datapath actions. */
bool action_set_has_group; /* Action set contains OFPACT_GROUP? */
struct ofpbuf action_set; /* Action set. */
+
+ enum xlate_error error; /* Translation failed. */
};
+const char *xlate_strerror(enum xlate_error error)
+{
+ switch (error) {
+ case XLATE_OK:
+ return "OK";
+ case XLATE_BRIDGE_NOT_FOUND:
+ return "Bridge not found";
+ case XLATE_RECURSION_TOO_DEEP:
+ return "Recursion too deep";
+ case XLATE_TOO_MANY_RESUBMITS:
+ return "Too many resubmits";
+ case XLATE_STACK_TOO_DEEP:
+ return "Stack too deep";
+ case XLATE_NO_RECIRCULATION_CONTEXT:
+ return "No recirculation context";
+ case XLATE_RECIRCULATION_CONFLICT:
+ return "Recirculation conflict";
+ case XLATE_TOO_MANY_MPLS_LABELS:
+ return "Too many MPLS labels";
+ }
+ return "Unknown error";
+}
+
static void xlate_action_set(struct xlate_ctx *ctx);
static void xlate_commit_actions(struct xlate_ctx *ctx);
XC_NORMAL,
XC_FIN_TIMEOUT,
XC_GROUP,
- XC_TNL_ARP,
+ XC_TNL_NEIGH,
};
/* xlate_cache entries hold enough information to perform the side effects of
struct {
char br_name[IFNAMSIZ];
ovs_be32 d_ip;
- } tnl_arp_cache;
+ } tnl_neigh_cache;
} u;
};
}
}
+static struct vlog_rate_limit error_report_rl = VLOG_RATE_LIMIT_INIT(1, 5);
+
+#define XLATE_REPORT_ERROR(CTX, ...) \
+ do { \
+ if (OVS_UNLIKELY((CTX)->xin->report_hook)) { \
+ xlate_report(CTX, __VA_ARGS__); \
+ } else { \
+ VLOG_ERR_RL(&error_report_rl, __VA_ARGS__); \
+ } \
+ } while (0)
+
static inline void
xlate_report_actions(struct xlate_ctx *ctx, const char *title,
const struct ofpact *ofpacts, size_t ofpacts_len)
if (ctx->xin->xcache) {
struct xc_entry *entry;
- entry = xlate_cache_add_entry(ctx->xin->xcache, XC_TNL_ARP);
- 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;
+ entry = xlate_cache_add_entry(ctx->xin->xcache, XC_TNL_NEIGH);
+ ovs_strlcpy(entry->u.tnl_neigh_cache.br_name, out_dev->xbridge->name,
+ sizeof entry->u.tnl_neigh_cache.br_name);
+ entry->u.tnl_neigh_cache.d_ip = d_ip;
}
xlate_report(ctx, "tunneling from "ETH_ADDR_FMT" "IP_FMT
flow->ct_state = 0;
flow->ct_zone = 0;
flow->ct_mark = 0;
+ memset(&flow->ct_label, 0, sizeof flow->ct_label);
}
static void
/* If 'struct flow' gets additional metadata, we'll need to zero it out
* before traversing a patch port. */
- BUILD_ASSERT_DECL(FLOW_WC_SEQ == 34);
+ BUILD_ASSERT_DECL(FLOW_WC_SEQ == 35);
memset(&flow_tnl, 0, sizeof flow_tnl);
if (!xport) {
* recirculated packet! */
ctx->exit = false;
+ /* Peer bridge errors do not propagate back. */
+ ctx->error = XLATE_OK;
+
if (ctx->xin->resubmit_stats) {
netdev_vport_inc_tx(xport->netdev, ctx->xin->resubmit_stats);
netdev_vport_inc_rx(peer->netdev, ctx->xin->resubmit_stats);
}
if (xport->is_tunnel) {
+ ovs_be32 dst;
/* Save tunnel metadata so that changes made due to
* the Logical (tunnel) Port are not visible for any further
* matches, while explicit set actions on tunnel metadata are.
xlate_report(ctx, "Tunneling decided against output");
goto out; /* restore flow_nw_tos */
}
- if (flow->tunnel.ip_dst == ctx->orig_tunnel_ip_dst) {
+ dst = in6_addr_get_mapped_ipv4(&ctx->orig_tunnel_ipv6_dst);
+ if (flow->tunnel.ip_dst == dst) {
xlate_report(ctx, "Not tunneling to our own address");
goto out; /* restore flow_nw_tos */
}
static bool
xlate_resubmit_resource_check(struct xlate_ctx *ctx)
{
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
-
if (ctx->recurse >= MAX_RESUBMIT_RECURSION + MAX_INTERNAL_RESUBMITS) {
- VLOG_ERR_RL(&rl, "resubmit actions recursed over %d times",
- MAX_RESUBMIT_RECURSION);
+ XLATE_REPORT_ERROR(ctx, "resubmit actions recursed over %d times",
+ MAX_RESUBMIT_RECURSION);
+ ctx->error = XLATE_RECURSION_TOO_DEEP;
} else if (ctx->resubmits >= MAX_RESUBMITS + MAX_INTERNAL_RESUBMITS) {
- VLOG_ERR_RL(&rl, "over %d resubmit actions", MAX_RESUBMITS);
+ XLATE_REPORT_ERROR(ctx, "over %d resubmit actions", MAX_RESUBMITS);
+ ctx->error = XLATE_TOO_MANY_RESUBMITS;
} else if (ctx->odp_actions->size > UINT16_MAX) {
- VLOG_ERR_RL(&rl, "resubmits yielded over 64 kB of actions");
+ XLATE_REPORT_ERROR(ctx, "resubmits yielded over 64 kB of actions");
+ /* NOT an error, as we'll be slow-pathing the flow in this case? */
+ ctx->exit = true; /* XXX: translation still terminated! */
} else if (ctx->stack.size >= 65536) {
- VLOG_ERR_RL(&rl, "resubmits yielded over 64 kB of stack");
+ XLATE_REPORT_ERROR(ctx, "resubmits yielded over 64 kB of stack");
+ ctx->error = XLATE_STACK_TOO_DEEP;
} else {
return true;
}
ctx->table_id = old_table_id;
return;
}
-
- ctx->exit = true;
}
static void
.ofpacts = ctx->action_set.data,
};
- /* 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(&state);
- 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.
- *
- * This might fail and return 0. 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. */
- id = recirc_find_id(&state);
+ /* Allocate a unique recirc id for the given metadata state in the
+ * flow. An existing id, with a new reference to the corresponding
+ * recirculation context, will be returned if possible.
+ * 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(&state);
+ if (!id) {
+ XLATE_REPORT_ERROR(ctx, "Failed to allocate recirculation id");
+ ctx->error = XLATE_NO_RECIRCULATION_CONTEXT;
+ return;
}
+ recirc_refs_add(&ctx->xout->recircs, id);
nl_msg_put_u32(ctx->odp_actions, OVS_ACTION_ATTR_RECIRC, id);
compose_recirculate_action__(ctx, 0);
}
+/* Fork the pipeline here. The current packet will continue processing the
+ * current action list. A clone of the current packet will recirculate, skip
+ * the remainder of the current action list and asynchronously resume pipeline
+ * processing in 'table' with the current metadata and action set. */
+static void
+compose_recirculate_and_fork(struct xlate_ctx *ctx, uint8_t table)
+{
+ ctx->recirc_action_offset = ctx->action_set.size;
+ compose_recirculate_action__(ctx, table);
+}
+
static void
compose_mpls_push_action(struct xlate_ctx *ctx, struct ofpact_push_mpls *mpls)
{
xlate_commit_actions(ctx);
} else if (n >= FLOW_MAX_MPLS_LABELS) {
if (ctx->xin->packet != NULL) {
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
- VLOG_WARN_RL(&rl, "bridge %s: dropping packet on which an "
+ XLATE_REPORT_ERROR(ctx, "bridge %s: dropping packet on which an "
"MPLS push action can't be performed as it would "
"have more MPLS LSEs than the %d supported.",
ctx->xbridge->name, FLOW_MAX_MPLS_LABELS);
}
- ctx->exit = true;
+ ctx->error = XLATE_TOO_MANY_MPLS_LABELS;
return;
}
}
} else if (n >= FLOW_MAX_MPLS_LABELS) {
if (ctx->xin->packet != NULL) {
- static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
- VLOG_WARN_RL(&rl, "bridge %s: dropping packet on which an "
+ XLATE_REPORT_ERROR(ctx, "bridge %s: dropping packet on which an "
"MPLS pop action can't be performed as it has "
"more MPLS LSEs than the %d supported.",
ctx->xbridge->name, FLOW_MAX_MPLS_LABELS);
}
- ctx->exit = true;
+ ctx->error = XLATE_TOO_MANY_MPLS_LABELS;
ofpbuf_clear(ctx->odp_actions);
}
}
case OFPACT_SAMPLE:
case OFPACT_DEBUG_RECIRC:
case OFPACT_CT:
+ case OFPACT_NAT:
break;
/* These need not be copied for restoration. */
}
}
+static void
+put_ct_label(const struct flow *flow, struct flow *base_flow,
+ struct ofpbuf *odp_actions, struct flow_wildcards *wc)
+{
+ if (!ovs_u128_is_zero(&wc->masks.ct_label)
+ && !ovs_u128_equals(&flow->ct_label, &base_flow->ct_label)) {
+ struct {
+ ovs_u128 key;
+ ovs_u128 mask;
+ } *odp_ct_label;
+
+ odp_ct_label = nl_msg_put_unspec_uninit(odp_actions,
+ OVS_CT_ATTR_LABELS,
+ sizeof(*odp_ct_label));
+ odp_ct_label->key = flow->ct_label;
+ odp_ct_label->mask = wc->masks.ct_label;
+ }
+}
+
+static void
+put_ct_helper(struct ofpbuf *odp_actions, struct ofpact_conntrack *ofc)
+{
+ if (ofc->alg) {
+ if (ofc->alg == IPPORT_FTP) {
+ nl_msg_put_string(odp_actions, OVS_CT_ATTR_HELPER, "ftp");
+ } else {
+ VLOG_WARN("Cannot serialize ct_helper %d\n", ofc->alg);
+ }
+ }
+}
+
+static void
+put_ct_nat(struct xlate_ctx *ctx)
+{
+ struct ofpact_nat *ofn = ctx->ct_nat_action;
+ size_t nat_offset;
+
+ if (!ofn) {
+ return;
+ }
+
+ nat_offset = nl_msg_start_nested(ctx->odp_actions, OVS_CT_ATTR_NAT);
+ if (ofn->flags & NX_NAT_F_SRC || ofn->flags & NX_NAT_F_DST) {
+ nl_msg_put_flag(ctx->odp_actions, ofn->flags & NX_NAT_F_SRC
+ ? OVS_NAT_ATTR_SRC : OVS_NAT_ATTR_DST);
+ if (ofn->flags & NX_NAT_F_PERSISTENT) {
+ nl_msg_put_flag(ctx->odp_actions, OVS_NAT_ATTR_PERSISTENT);
+ }
+ if (ofn->flags & NX_NAT_F_PROTO_HASH) {
+ nl_msg_put_flag(ctx->odp_actions, OVS_NAT_ATTR_PROTO_HASH);
+ } else if (ofn->flags & NX_NAT_F_PROTO_RANDOM) {
+ nl_msg_put_flag(ctx->odp_actions, OVS_NAT_ATTR_PROTO_RANDOM);
+ }
+ if (ofn->range_af == AF_INET) {
+ nl_msg_put_be32(ctx->odp_actions, OVS_NAT_ATTR_IP_MIN,
+ ofn->range.addr.ipv4.min);
+ if (ofn->range.addr.ipv4.max &&
+ (ntohl(ofn->range.addr.ipv4.max)
+ > ntohl(ofn->range.addr.ipv4.min))) {
+ nl_msg_put_be32(ctx->odp_actions, OVS_NAT_ATTR_IP_MAX,
+ ofn->range.addr.ipv4.max);
+ }
+ } else if (ofn->range_af == AF_INET6) {
+ nl_msg_put_unspec(ctx->odp_actions, OVS_NAT_ATTR_IP_MIN,
+ &ofn->range.addr.ipv6.min,
+ sizeof ofn->range.addr.ipv6.min);
+ if (!ipv6_mask_is_any(&ofn->range.addr.ipv6.max) &&
+ memcmp(&ofn->range.addr.ipv6.max, &ofn->range.addr.ipv6.min,
+ sizeof ofn->range.addr.ipv6.max) > 0) {
+ nl_msg_put_unspec(ctx->odp_actions, OVS_NAT_ATTR_IP_MAX,
+ &ofn->range.addr.ipv6.max,
+ sizeof ofn->range.addr.ipv6.max);
+ }
+ }
+ if (ofn->range_af != AF_UNSPEC && ofn->range.proto.min) {
+ nl_msg_put_u16(ctx->odp_actions, OVS_NAT_ATTR_PROTO_MIN,
+ ofn->range.proto.min);
+ if (ofn->range.proto.max &&
+ ofn->range.proto.max > ofn->range.proto.min) {
+ nl_msg_put_u16(ctx->odp_actions, OVS_NAT_ATTR_PROTO_MAX,
+ ofn->range.proto.max);
+ }
+ }
+ }
+ nl_msg_end_nested(ctx->odp_actions, nat_offset);
+}
+
static void
compose_conntrack_action(struct xlate_ctx *ctx, struct ofpact_conntrack *ofc)
{
+ ovs_u128 old_ct_label = ctx->base_flow.ct_label;
uint32_t old_ct_mark = ctx->base_flow.ct_mark;
size_t ct_offset;
uint16_t zone;
xlate_commit_actions(ctx);
/* Process nested actions first, to populate the key. */
+ ctx->ct_nat_action = NULL;
do_xlate_actions(ofc->actions, ofpact_ct_get_action_len(ofc), ctx);
if (ofc->zone_src.field) {
}
nl_msg_put_u16(ctx->odp_actions, OVS_CT_ATTR_ZONE, zone);
put_ct_mark(&ctx->xin->flow, &ctx->base_flow, ctx->odp_actions, ctx->wc);
+ put_ct_label(&ctx->xin->flow, &ctx->base_flow, ctx->odp_actions, ctx->wc);
+ put_ct_helper(ctx->odp_actions, ofc);
+ put_ct_nat(ctx);
+ ctx->ct_nat_action = NULL;
nl_msg_end_nested(ctx->odp_actions, ct_offset);
/* Restore the original ct fields in the key. These should only be exposed
* after recirculation to another table. */
ctx->base_flow.ct_mark = old_ct_mark;
+ ctx->base_flow.ct_label = old_ct_label;
if (ofc->recirc_table == NX_CT_RECIRC_NONE) {
/* If we do not recirculate as part of this action, hide the results of
} else {
/* Use ct_* fields from datapath during recirculation upcall. */
ctx->conntracked = true;
- ctx_trigger_recirculation(ctx);
- compose_recirculate_action__(ctx, ofc->recirc_table);
+ compose_recirculate_and_fork(ctx, ofc->recirc_table);
}
}
const struct ofpact *a;
if (ovs_native_tunneling_is_on(ctx->xbridge->ofproto)) {
- tnl_arp_snoop(flow, wc, ctx->xbridge->name);
- tnl_nd_snoop(flow, wc, ctx->xbridge->name);
+ tnl_neigh_snoop(flow, wc, ctx->xbridge->name);
}
/* dl_type already in the mask, not set below. */
const struct ofpact_set_field *set_field;
const struct mf_field *mf;
+ if (ctx->error) {
+ break;
+ }
+
if (ctx->exit) {
/* Check if need to store the remaining actions for later
* execution. */
&& !eth_type_mpls(flow->dl_type)) {
break;
}
- /* A flow may wildcard nw_frag. Do nothing if setting a trasport
+ /* A flow may wildcard nw_frag. Do nothing if setting a transport
* header field on a packet that does not have them. */
mf_mask_field_and_prereqs(mf, wc);
if (mf_are_prereqs_ok(mf, flow)) {
compose_conntrack_action(ctx, ofpact_get_CT(a));
break;
+ case OFPACT_NAT:
+ /* This will be processed by compose_conntrack_action(). */
+ ctx->ct_nat_action = ofpact_get_NAT(a);
+ break;
+
case OFPACT_DEBUG_RECIRC:
ctx_trigger_recirculation(ctx);
a = ofpact_next(a);
/* Check if need to store this and the remaining actions for later
* execution. */
- if (ctx->exit && ctx_first_recirculation_action(ctx)) {
+ if (!ctx->error && ctx->exit && ctx_first_recirculation_action(ctx)) {
recirc_unroll_actions(a, OFPACT_ALIGN(ofpacts_len -
((uint8_t *)a -
(uint8_t *)ofpacts)),
xlate_out_uninit(struct xlate_out *xout)
{
if (xout) {
- xlate_out_free_recircs(xout);
+ recirc_refs_unref(&xout->recircs);
}
}
xlate_actions_for_side_effects(struct xlate_in *xin)
{
struct xlate_out xout;
+ enum xlate_error error;
+
+ error = xlate_actions(xin, &xout);
+ if (error) {
+ static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5);
+
+ VLOG_WARN_RL(&rl, "xlate_actions failed (%s)!", xlate_strerror(error));
+ }
- xlate_actions(xin, &xout);
xlate_out_uninit(&xout);
}
\f
/* 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_out_uninit().
+ * Returns 'XLATE_OK' if translation was successful. In case of an error an
+ * empty set of actions will be returned in 'xin->odp_actions' (if non-NULL),
+ * so that most callers may ignore the return value and transparently install a
+ * drop flow when the translation fails. */
+enum xlate_error
xlate_actions(struct xlate_in *xin, struct xlate_out *xout)
{
*xout = (struct xlate_out) {
.slow = 0,
.fail_open = false,
- .n_recircs = 0,
+ .recircs = RECIRC_REFS_EMPTY_INITIALIZER,
};
struct xlate_cfg *xcfg = ovsrcu_get(struct xlate_cfg *, &xcfgp);
struct xbridge *xbridge = xbridge_lookup(xcfg, xin->ofproto);
if (!xbridge) {
- return;
+ return XLATE_BRIDGE_NOT_FOUND;
}
struct flow *flow = &xin->flow;
.xin = xin,
.xout = xout,
.base_flow = *flow,
- .orig_tunnel_ip_dst = flow->tunnel.ip_dst,
.xbridge = xbridge,
.stack = OFPBUF_STUB_INITIALIZER(stack_stub),
.rule = xin->rule,
.sflow_odp_port = 0,
.nf_output_iface = NF_OUT_DROP,
.exit = false,
+ .error = XLATE_OK,
.mirrors = 0,
.recirc_action_offset = -1,
.was_mpls = false,
.conntracked = false,
+ .ct_nat_action = NULL,
+
.action_set_has_group = false,
.action_set = OFPBUF_STUB_INITIALIZER(action_set_stub),
};
+ in6_addr_set_mapped_ipv4(&ctx.orig_tunnel_ipv6_dst, flow->tunnel.ip_dst);
/* 'base_flow' reflects the packet as it came in, but we need it to reflect
* the packet as the datapath will treat it for output actions:
VLOG_WARN_RL(&rl, "Recirculation conflict (%s)!", conflict);
xlate_report(&ctx, "- Recirculation conflict (%s)!", conflict);
+ ctx.error = XLATE_RECIRCULATION_CONFLICT;
goto exit;
}
static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 1);
VLOG_WARN_RL(&rl, "Recirculation bridge no longer exists.");
xlate_report(&ctx, "- Recirculation bridge no longer exists.");
+ ctx.error = XLATE_BRIDGE_NOT_FOUND;
goto exit;
}
ctx.xbridge = new_bridge;
VLOG_WARN_RL(&rl, "Recirculation context not found for ID %"PRIx32,
flow->recirc_id);
+ ctx.error = XLATE_NO_RECIRCULATION_CONTEXT;
goto exit;
}
/* The bridge is now known so obtain its table version. */
mirror_ingress_packet(&ctx);
do_xlate_actions(ofpacts, ofpacts_len, &ctx);
+ if (ctx.error) {
+ goto exit;
+ }
/* We've let OFPP_NORMAL and the learning action look at the
* packet, so drop it now if forwarding is disabled. */
ofpbuf_uninit(&ctx.stack);
ofpbuf_uninit(&ctx.action_set);
ofpbuf_uninit(&scratch_actions);
+
+ /* Make sure we return a "drop flow" in case of an error. */
+ if (ctx.error) {
+ xout->slow = 0;
+ if (xin->odp_actions) {
+ ofpbuf_clear(xin->odp_actions);
+ }
+ }
+ return ctx.error;
}
/* Sends 'packet' out 'ofport'.
group_dpif_credit_stats(entry->u.group.group, entry->u.group.bucket,
stats);
break;
- case XC_TNL_ARP:
- /* Lookup arp to avoid arp timeout. */
- tnl_arp_lookup(entry->u.tnl_arp_cache.br_name,
- entry->u.tnl_arp_cache.d_ip, &dmac);
+ case XC_TNL_NEIGH:
+ /* Lookup neighbor to avoid timeout. */
+ tnl_arp_lookup(entry->u.tnl_neigh_cache.br_name,
+ entry->u.tnl_neigh_cache.d_ip, &dmac);
break;
default:
OVS_NOT_REACHED();
case XC_GROUP:
group_dpif_unref(entry->u.group.group);
break;
- case XC_TNL_ARP:
+ case XC_TNL_NEIGH:
break;
default:
OVS_NOT_REACHED();