[ovs-dev] [RFC] bfd: Replace bfd.{c, h} with generic implementation.
Alex Wang
alexw at nicira.com
Tue Mar 4 01:46:41 UTC 2014
This commit replaces the current bfd.{c,h} with a generic library-like
implementation. Wrappers (in bfd_ts.{c,h}) are then added to guarantee
the thread-safety and compatibility with current invocation pattern.
Signed-off-by: Alex Wang <alexw at nicira.com>
---
build-aux/thread-safety-blacklist | 1 -
lib/automake.mk | 2 +
lib/bfd.c | 1327 ++++++++++---------------------------
lib/bfd.h | 229 ++++++-
lib/bfd_ts.c | 620 +++++++++++++++++
lib/bfd_ts.h | 48 ++
ofproto/ofproto-dpif-monitor.c | 16 +-
ofproto/ofproto-dpif-xlate.c | 22 +-
ofproto/ofproto-dpif.c | 19 +-
9 files changed, 1241 insertions(+), 1043 deletions(-)
create mode 100644 lib/bfd_ts.c
create mode 100644 lib/bfd_ts.h
diff --git a/build-aux/thread-safety-blacklist b/build-aux/thread-safety-blacklist
index 42560df..4db5a84 100644
--- a/build-aux/thread-safety-blacklist
+++ b/build-aux/thread-safety-blacklist
@@ -70,7 +70,6 @@
\bputchar_unlocked(
\bputenv(
\bpututxline(
-\brand(
\bsetenv(
\bsetgrent(
\bsetkey(
diff --git a/lib/automake.mk b/lib/automake.mk
index 9c345e7..b712410 100644
--- a/lib/automake.mk
+++ b/lib/automake.mk
@@ -18,6 +18,8 @@ lib_libopenvswitch_la_SOURCES = \
lib/backtrace.h \
lib/bfd.c \
lib/bfd.h \
+ lib/bfd_ts.c \
+ lib/bfd_ts.h \
lib/bitmap.c \
lib/bitmap.h \
lib/bundle.c \
diff --git a/lib/bfd.c b/lib/bfd.c
index 5413105..566214b 100644
--- a/lib/bfd.c
+++ b/lib/bfd.c
@@ -13,363 +13,200 @@
* limitations under the License. */
#include <config.h>
+
#include "bfd.h"
-#include <sys/types.h>
#include <arpa/inet.h>
-#include <netinet/in_systm.h>
-#include <netinet/ip.h>
-
-#include "byte-order.h"
-#include "connectivity.h"
-#include "csum.h"
-#include "dpif.h"
-#include "dynamic-string.h"
-#include "flow.h"
-#include "hash.h"
-#include "hmap.h"
-#include "list.h"
-#include "netdev.h"
-#include "netlink.h"
-#include "odp-util.h"
-#include "ofpbuf.h"
-#include "ovs-thread.h"
-#include "openvswitch/types.h"
-#include "packets.h"
-#include "poll-loop.h"
-#include "random.h"
-#include "seq.h"
-#include "smap.h"
-#include "timeval.h"
-#include "unaligned.h"
-#include "unixctl.h"
-#include "util.h"
-#include "vlog.h"
-
-VLOG_DEFINE_THIS_MODULE(bfd);
-
-/* XXX Finish BFD.
- *
- * The goal of this module is to replace CFM with something both more flexible
- * and standards compliant. In service of this goal, the following needs to be
- * done.
- *
- * - Compliance
- * * Implement Demand mode.
- * * Go through the RFC line by line and verify we comply.
- * * Test against a hardware implementation. Preferably a popular one.
- * * Delete BFD packets with nw_ttl != 255 in the datapath to prevent DOS
- * attacks.
- *
- * - Unit tests.
- *
- * - Set TOS/PCP on the outer tunnel header when encapped.
- *
- * - Sending BFD messages should be in its own thread/process.
- *
- * - Scale testing. How does it operate when there are large number of bfd
- * sessions? Do we ever have random flaps? What's the CPU utilization?
- *
- * - Rely on data traffic for liveness by using BFD demand mode.
- * If we're receiving traffic on a port, we can safely assume it's up (modulo
- * unidrectional failures). BFD has a demand mode in which it can stay quiet
- * unless it feels the need to check the status of the port. Using this, we
- * can implement a strategy in which BFD only sends control messages on dark
- * interfaces.
- *
- * - Depending on how one interprets the spec, it appears that a BFD session
- * can never change bfd.LocalDiag to "No Diagnostic". We should verify that
- * this is what hardware implementations actually do. Seems like "No
- * Diagnostic" should be set once a BFD session state goes UP. */
-
-#define BFD_VERSION 1
-
-enum flags {
- FLAG_MULTIPOINT = 1 << 0,
- FLAG_DEMAND = 1 << 1,
- FLAG_AUTH = 1 << 2,
- FLAG_CTL = 1 << 3,
- FLAG_FINAL = 1 << 4,
- FLAG_POLL = 1 << 5
-};
-
-enum state {
- STATE_ADMIN_DOWN = 0 << 6,
- STATE_DOWN = 1 << 6,
- STATE_INIT = 2 << 6,
- STATE_UP = 3 << 6
-};
-
-enum diag {
- DIAG_NONE = 0, /* No Diagnostic. */
- DIAG_EXPIRED = 1, /* Control Detection Time Expired. */
- DIAG_ECHO_FAILED = 2, /* Echo Function Failed. */
- DIAG_RMT_DOWN = 3, /* Neighbor Signaled Session Down. */
- DIAG_FWD_RESET = 4, /* Forwarding Plane Reset. */
- DIAG_PATH_DOWN = 5, /* Path Down. */
- DIAG_CPATH_DOWN = 6, /* Concatenated Path Down. */
- DIAG_ADMIN_DOWN = 7, /* Administratively Down. */
- DIAG_RCPATH_DOWN = 8 /* Reverse Concatenated Path Down. */
-};
-
-/* RFC 5880 Section 4.1
- * 0 1 2 3
- * 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
- * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
- * |Vers | Diag |Sta|P|F|C|A|D|M| Detect Mult | Length |
- * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
- * | My Discriminator |
- * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
- * | Your Discriminator |
- * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
- * | Desired Min TX Interval |
- * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
- * | Required Min RX Interval |
- * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
- * | Required Min Echo RX Interval |
- * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ */
-struct msg {
- uint8_t vers_diag; /* Version and diagnostic. */
- uint8_t flags; /* 2bit State field followed by flags. */
- uint8_t mult; /* Fault detection multiplier. */
- uint8_t length; /* Length of this BFD message. */
- ovs_be32 my_disc; /* My discriminator. */
- ovs_be32 your_disc; /* Your discriminator. */
- ovs_be32 min_tx; /* Desired minimum tx interval. */
- ovs_be32 min_rx; /* Required minimum rx interval. */
- ovs_be32 min_rx_echo; /* Required minimum echo rx interval. */
-};
-BUILD_ASSERT_DECL(BFD_PACKET_LEN == sizeof(struct msg));
-
-#define DIAG_MASK 0x1f
-#define VERS_SHIFT 5
-#define STATE_MASK 0xC0
-#define FLAGS_MASK 0x3f
-
-struct bfd {
- struct hmap_node node; /* In 'all_bfds'. */
- uint32_t disc; /* bfd.LocalDiscr. Key in 'all_bfds' hmap. */
-
- char *name; /* Name used for logging. */
-
- bool cpath_down; /* Concatenated Path Down. */
- uint8_t mult; /* bfd.DetectMult. */
-
- struct netdev *netdev;
- uint64_t rx_packets; /* Packets received by 'netdev'. */
-
- enum state state; /* bfd.SessionState. */
- enum state rmt_state; /* bfd.RemoteSessionState. */
-
- enum diag diag; /* bfd.LocalDiag. */
- enum diag rmt_diag; /* Remote diagnostic. */
-
- enum flags flags; /* Flags sent on messages. */
- enum flags rmt_flags; /* Flags last received. */
-
- uint32_t rmt_disc; /* bfd.RemoteDiscr. */
-
- uint8_t eth_dst[ETH_ADDR_LEN];/* Ethernet destination address. */
- bool eth_dst_set; /* 'eth_dst' set through database. */
-
- ovs_be32 ip_src; /* IPv4 source address. */
- ovs_be32 ip_dst; /* IPv4 destination address. */
-
- uint16_t udp_src; /* UDP source port. */
-
- /* All timers in milliseconds. */
- long long int rmt_min_rx; /* bfd.RemoteMinRxInterval. */
- long long int rmt_min_tx; /* Remote minimum TX interval. */
-
- long long int cfg_min_tx; /* Configured minimum TX rate. */
- long long int cfg_min_rx; /* Configured required minimum RX rate. */
- long long int poll_min_tx; /* Min TX negotating in a poll sequence. */
- long long int poll_min_rx; /* Min RX negotating in a poll sequence. */
- long long int min_tx; /* bfd.DesiredMinTxInterval. */
- long long int min_rx; /* bfd.RequiredMinRxInterval. */
-
- long long int last_tx; /* Last TX time. */
- long long int next_tx; /* Next TX time. */
- long long int detect_time; /* RFC 5880 6.8.4 Detection time. */
-
- bool last_forwarding; /* Last calculation of forwarding flag. */
- int forwarding_override; /* Manual override of 'forwarding' status. */
-
- atomic_bool check_tnl_key; /* Verify tunnel key of inbound packets? */
- struct ovs_refcount ref_cnt;
-
- /* When forward_if_rx is true, bfd_forwarding() will return
- * true as long as there are incoming packets received.
- * Note, forwarding_override still has higher priority. */
- bool forwarding_if_rx;
- long long int forwarding_if_rx_detect_time;
-
- /* BFD decay related variables. */
- bool in_decay; /* True when bfd is in decay. */
- int decay_min_rx; /* min_rx is set to decay_min_rx when */
- /* in decay. */
- int decay_rx_ctl; /* Count bfd packets received within decay */
- /* detect interval. */
- uint64_t decay_rx_packets; /* Packets received by 'netdev'. */
- long long int decay_detect_time; /* Decay detection time. */
-
- uint64_t flap_count; /* Counts bfd forwarding flaps. */
-};
-
-static struct ovs_mutex mutex = OVS_MUTEX_INITIALIZER;
-static struct hmap all_bfds__ = HMAP_INITIALIZER(&all_bfds__);
-static struct hmap *const all_bfds OVS_GUARDED_BY(mutex) = &all_bfds__;
-
-static bool bfd_lookup_ip(const char *host_name, struct in_addr *)
- OVS_REQUIRES(mutex);
-static bool bfd_forwarding__(struct bfd *) OVS_REQUIRES(mutex);
-static bool bfd_in_poll(const struct bfd *) OVS_REQUIRES(mutex);
-static void bfd_poll(struct bfd *bfd) OVS_REQUIRES(mutex);
-static const char *bfd_diag_str(enum diag) OVS_REQUIRES(mutex);
-static const char *bfd_state_str(enum state) OVS_REQUIRES(mutex);
-static long long int bfd_min_tx(const struct bfd *) OVS_REQUIRES(mutex);
-static long long int bfd_tx_interval(const struct bfd *)
- OVS_REQUIRES(mutex);
-static long long int bfd_rx_interval(const struct bfd *)
- OVS_REQUIRES(mutex);
-static void bfd_set_next_tx(struct bfd *) OVS_REQUIRES(mutex);
-static void bfd_set_state(struct bfd *, enum state, enum diag)
- OVS_REQUIRES(mutex);
-static uint32_t generate_discriminator(void) OVS_REQUIRES(mutex);
-static void bfd_put_details(struct ds *, const struct bfd *)
- OVS_REQUIRES(mutex);
-static uint64_t bfd_rx_packets(const struct bfd *) OVS_REQUIRES(mutex);
-static void bfd_try_decay(struct bfd *) OVS_REQUIRES(mutex);
-static void bfd_decay_update(struct bfd *) OVS_REQUIRES(mutex);
-
-static void bfd_forwarding_if_rx_update(struct bfd *) OVS_REQUIRES(mutex);
-static void bfd_unixctl_show(struct unixctl_conn *, int argc,
- const char *argv[], void *aux OVS_UNUSED);
-static void bfd_unixctl_set_forwarding_override(struct unixctl_conn *,
- int argc, const char *argv[],
- void *aux OVS_UNUSED);
-static void log_msg(enum vlog_level, const struct msg *, const char *message,
- const struct bfd *) OVS_REQUIRES(mutex);
-
-static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(20, 20);
-
-/* Returns true if the interface on which 'bfd' is running may be used to
- * forward traffic according to the BFD session state. */
-bool
-bfd_forwarding(struct bfd *bfd) OVS_EXCLUDED(mutex)
+#include <limits.h>
+#include <stdlib.h>
+#include <string.h>
+
+#define MAX(X, Y) ((X) > (Y) ? (X) : (Y))
+#define MIN(X, Y) ((X) < (Y) ? (X) : (Y))
+
+
+static long long int
+bfd_min_tx(const struct bfd *bfd)
{
- bool ret;
+ /* RFC 5880 Section 6.8.3
+ * When bfd.SessionState is not Up, the system MUST set
+ * bfd.DesiredMinTxInterval to a value of not less than one second
+ * (1,000,000 microseconds). This is intended to ensure that the
+ * bandwidth consumed by BFD sessions that are not Up is negligible,
+ * particularly in the case where a neighbor may not be running BFD. */
+ return (bfd->state == STATE_UP ? bfd->min_tx
+ : MAX(bfd->min_tx, 1000));
+}
- ovs_mutex_lock(&mutex);
- ret = bfd_forwarding__(bfd);
- ovs_mutex_unlock(&mutex);
- return ret;
+static long long int
+bfd_tx_interval(const struct bfd *bfd)
+{
+ long long int interval = bfd_min_tx(bfd);
+ return MAX(interval, bfd->rmt_min_rx);
}
-/* When forwarding_if_rx is enabled, if there are packets received,
- * updates forwarding_if_rx_detect_time. */
-void
-bfd_account_rx(struct bfd *bfd, const struct dpif_flow_stats *stats)
+static long long int
+bfd_rx_interval(const struct bfd *bfd)
+{
+ return MAX(bfd->min_rx, bfd->rmt_min_tx);
+}
+
+static void
+bfd_set_next_tx(struct bfd *bfd)
+{
+ long long int interval = bfd_tx_interval(bfd);
+ interval -= interval * (rand() % 26) / 100;
+ bfd->next_tx = bfd->last_tx + interval;
+}
+
+
+static bool
+bfd_forwarding__(const struct bfd *bfd, long long int now)
+{
+ bool should_forward = false;
+
+ if (bfd->forwarding_override) {
+ return bfd->forwarding_override == 1;
+ }
+
+ if (bfd->forward_if_rx_interval) {
+ /* The first time after forward_if_rx is enabled. */
+ if (!bfd->forward_if_rx_detect_time) {
+ should_forward = bfd->state == STATE_UP ? true : false;
+ } else {
+ should_forward = bfd->forward_if_rx_detect_time > now;
+ }
+ }
+
+ return (bfd->state == STATE_UP
+ || (bfd->forward_if_rx_interval && should_forward))
+ && bfd->rmt_diag != DIAG_PATH_DOWN
+ && bfd->rmt_diag != DIAG_CPATH_DOWN
+ && bfd->rmt_diag != DIAG_RCPATH_DOWN;
+}
+
+/* If there is packet received, sets the 'forward_if_rx_detect_time'
+ * to MAX('forward_if_rx_interval', 'rx_int') away from now. */
+static void
+bfd_forward_if_rx(struct bfd *bfd, long long int now)
{
- if (stats->n_packets && bfd->forwarding_if_rx) {
- ovs_mutex_lock(&mutex);
- bfd_forwarding__(bfd);
- bfd_forwarding_if_rx_update(bfd);
- bfd_forwarding__(bfd);
- ovs_mutex_unlock(&mutex);
+ if (bfd->forward_if_rx_data && bfd->forward_if_rx_interval) {
+ uint32_t rx_int = bfd->mult * bfd_rx_interval(bfd);
+
+ bfd->forward_if_rx_detect_time = MAX(bfd->forward_if_rx_interval,
+ rx_int) + now;
+ bfd->forward_if_rx_data = false;
}
}
-/* Returns a 'smap' of key value pairs representing the status of 'bfd'
- * intended for the OVS database. */
-void
-bfd_get_status(const struct bfd *bfd, struct smap *smap)
- OVS_EXCLUDED(mutex)
+/* Increments the 'flap_count' if there is a change in the
+ * forwarding flag value. */
+static void
+bfd_check_forwarding_flap(struct bfd *bfd, long long int now)
{
- ovs_mutex_lock(&mutex);
- smap_add(smap, "forwarding",
- bfd_forwarding__(CONST_CAST(struct bfd *, bfd))
- ? "true" : "false");
- smap_add(smap, "state", bfd_state_str(bfd->state));
- smap_add(smap, "diagnostic", bfd_diag_str(bfd->diag));
- smap_add_format(smap, "flap_count", "%"PRIu64, bfd->flap_count);
-
- if (bfd->state != STATE_DOWN) {
- smap_add(smap, "remote_state", bfd_state_str(bfd->rmt_state));
- smap_add(smap, "remote_diagnostic", bfd_diag_str(bfd->rmt_diag));
+ bool last_forwarding = bfd->last_forwarding;
+
+ bfd->last_forwarding = bfd_forwarding__(bfd, now);
+ if (bfd->last_forwarding != last_forwarding) {
+ bfd->flap_count++;
}
- ovs_mutex_unlock(&mutex);
}
-/* Initializes, destroys, or reconfigures the BFD session 'bfd' (named 'name'),
- * according to the database configuration contained in 'cfg'. Takes ownership
- * of 'bfd', which may be NULL. Returns a BFD object which may be used as a
- * handle for the session, or NULL if BFD is not enabled according to 'cfg'.
- * Also returns NULL if cfg is NULL. */
-struct bfd *
-bfd_configure(struct bfd *bfd, const char *name, const struct smap *cfg,
- struct netdev *netdev) OVS_EXCLUDED(mutex)
+/* Decays the 'bfd->min_rx' to 'bfd->decay_min_rx' when number of packets
+ * received during the 'decay_min_rx' interval is less than two time
+ * number of BFD control packets expected to be received. */
+static void
+bfd_try_decay(struct bfd *bfd, long long int now)
{
- static struct ovsthread_once once = OVSTHREAD_ONCE_INITIALIZER;
- static atomic_uint16_t udp_src = ATOMIC_VAR_INIT(0);
+ if (bfd->state == STATE_UP && bfd->decay_min_rx
+ && now >= bfd->decay_detect_time
+ && bfd->decay_min_rx > bfd->cfg_min_rx) {
+ uint32_t expect_rx = 2 * (bfd->decay_min_rx / bfd->min_rx + 1);
- int decay_min_rx;
- long long int min_tx, min_rx;
- bool need_poll = false;
- bool cfg_min_rx_changed = false;
- bool cpath_down, forwarding_if_rx;
- const char *hwaddr, *ip_src, *ip_dst;
- struct in_addr in_addr;
- uint8_t ea[ETH_ADDR_LEN];
-
- if (ovsthread_once_start(&once)) {
- unixctl_command_register("bfd/show", "[interface]", 0, 1,
- bfd_unixctl_show, NULL);
- unixctl_command_register("bfd/set-forwarding",
- "[interface] normal|false|true", 1, 2,
- bfd_unixctl_set_forwarding_override, NULL);
- ovsthread_once_done(&once);
+ bfd->in_decay = bfd->decay_rx_count < expect_rx;
+ bfd->decay_detect_time = bfd->decay_min_rx + now;
+ bfd->decay_rx_count = 0;
}
+}
- if (!cfg || !smap_get_bool(cfg, "enable", false)) {
- bfd_unref(bfd);
- return NULL;
+static void
+bfd_set_state(struct bfd *bfd, enum bfd_state state, enum bfd_diag diag,
+ long long int now)
+{
+ if (bfd->cpath_down) {
+ diag = DIAG_CPATH_DOWN;
}
- ovs_mutex_lock(&mutex);
- if (!bfd) {
- bfd = xzalloc(sizeof *bfd);
- bfd->name = xstrdup(name);
- bfd->forwarding_override = -1;
- bfd->disc = generate_discriminator();
- hmap_insert(all_bfds, &bfd->node, bfd->disc);
-
- bfd->diag = DIAG_NONE;
- bfd->min_tx = 1000;
- bfd->mult = 3;
- ovs_refcount_init(&bfd->ref_cnt);
- bfd->netdev = netdev_ref(netdev);
- bfd->rx_packets = bfd_rx_packets(bfd);
- bfd->in_decay = false;
- bfd->flap_count = 0;
+ if (bfd->state != state || bfd->diag != diag) {
+ bfd->state = state;
+ bfd->diag = diag;
+
+ if (bfd->state <= STATE_DOWN) {
+ bfd->rmt_state = STATE_DOWN;
+ bfd->rmt_diag = DIAG_NONE;
+ bfd->rmt_min_rx = 1;
+ bfd->rmt_flags = 0;
+ bfd->rmt_disc = 0;
+ bfd->rmt_min_tx = 0;
+ }
+
+ if (bfd->state != STATE_UP && bfd->decay_min_rx) {
+ bfd->min_rx = bfd->cfg_min_rx;
+ bfd->in_decay = false;
+ bfd->decay_rx_count = UINT32_MAX;
+ }
+ }
+
+ bfd_check_forwarding_flap(bfd, now);
+}
+
+static bool
+bfd_in_poll(const struct bfd *bfd)
+{
+ return (bfd->flags & FLAG_POLL) != 0;
+}
- /* RFC 5881 section 4
- * The source port MUST be in the range 49152 through 65535. The same
- * UDP source port number MUST be used for all BFD Control packets
- * associated with a particular session. The source port number SHOULD
- * be unique among all BFD sessions on the system. */
- atomic_add(&udp_src, 1, &bfd->udp_src);
- bfd->udp_src = (bfd->udp_src % 16384) + 49152;
+static void
+bfd_poll(struct bfd *bfd)
+{
+ if (bfd->state > STATE_DOWN && !bfd_in_poll(bfd)
+ && !(bfd->flags & FLAG_FINAL)) {
+ bfd->poll_min_tx = bfd->cfg_min_tx;
+ bfd->poll_min_rx = bfd->in_decay ? bfd->decay_min_rx : bfd->cfg_min_rx;
+ bfd->flags |= FLAG_POLL;
+ bfd->next_tx = 0;
+ }
+}
- bfd_set_state(bfd, STATE_DOWN, DIAG_NONE);
+
+/* Configures 'bfd' using the 'setting'. Returns 0 if successful, a positive
+ * error number otherwise. */
+enum bfd_error
+bfd_configure(struct bfd *bfd, const struct bfd_setting *setting)
+{
+ uint32_t min_tx, min_rx;
+ uint8_t mult;
+ bool min_rx_changed = false;
+ bool need_poll = false;
+
+ if (!bfd || !setting) {
+ return BFD_EINVAL;
+ }
- memcpy(bfd->eth_dst, eth_addr_bfd, ETH_ADDR_LEN);
+ if (bfd->state == STATE_ADMIN_DOWN) {
+ bfd_set_state(bfd, STATE_DOWN, DIAG_NONE, 0);
}
- atomic_store(&bfd->check_tnl_key,
- smap_get_bool(cfg, "check_tnl_key", false));
- min_tx = smap_get_int(cfg, "min_tx", 100);
- min_tx = MAX(min_tx, 100);
+ if (bfd->disc != setting->disc) {
+ bfd->disc = setting->disc;
+ }
+
+ mult = MAX(setting->mult, 3);
+ if (bfd->mult != mult) {
+ bfd->mult = mult;
+ }
+
+ min_tx = MAX(setting->min_tx, 100);
if (bfd->cfg_min_tx != min_tx) {
bfd->cfg_min_tx = min_tx;
if (bfd->state != STATE_UP
@@ -379,220 +216,187 @@ bfd_configure(struct bfd *bfd, const char *name, const struct smap *cfg,
need_poll = true;
}
- min_rx = smap_get_int(cfg, "min_rx", 1000);
- min_rx = MAX(min_rx, 100);
+ min_rx = MAX(setting->min_rx, 100);
if (bfd->cfg_min_rx != min_rx) {
bfd->cfg_min_rx = min_rx;
if (bfd->state != STATE_UP
|| (!bfd_in_poll(bfd) && bfd->cfg_min_rx > bfd->min_rx)) {
bfd->min_rx = bfd->cfg_min_rx;
}
- cfg_min_rx_changed = true;
+ min_rx_changed = true;
need_poll = true;
}
- decay_min_rx = smap_get_int(cfg, "decay_min_rx", 0);
- if (bfd->decay_min_rx != decay_min_rx || cfg_min_rx_changed) {
- if (decay_min_rx > 0 && decay_min_rx < bfd->cfg_min_rx) {
- VLOG_WARN("%s: decay_min_rx cannot be less than %lld ms",
- bfd->name, bfd->cfg_min_rx);
- bfd->decay_min_rx = 0;
- } else {
- bfd->decay_min_rx = decay_min_rx;
- }
- /* Resets decay. */
- bfd->in_decay = false;
- bfd_decay_update(bfd);
- need_poll = true;
- }
-
- cpath_down = smap_get_bool(cfg, "cpath_down", false);
- if (bfd->cpath_down != cpath_down) {
- bfd->cpath_down = cpath_down;
- bfd_set_state(bfd, bfd->state, DIAG_NONE);
+ if (bfd->cpath_down != setting->cpath_down) {
+ bfd->cpath_down = setting->cpath_down;
+ bfd_set_state(bfd, bfd->state, DIAG_NONE, LLONG_MAX);
need_poll = true;
}
- hwaddr = smap_get(cfg, "bfd_dst_mac");
- if (hwaddr && eth_addr_from_string(hwaddr, ea) && !eth_addr_is_zero(ea)) {
- memcpy(bfd->eth_dst, ea, ETH_ADDR_LEN);
- bfd->eth_dst_set = true;
- } else if (bfd->eth_dst_set) {
- memcpy(bfd->eth_dst, eth_addr_bfd, ETH_ADDR_LEN);
- bfd->eth_dst_set = false;
+ if (bfd->forwarding_override != setting->forwarding_override) {
+ bfd->forwarding_override = setting->forwarding_override;
}
- ip_src = smap_get(cfg, "bfd_src_ip");
- if (ip_src && bfd_lookup_ip(ip_src, &in_addr)) {
- memcpy(&bfd->ip_src, &in_addr, sizeof in_addr);
- } else {
- bfd->ip_src = htonl(0xA9FE0100); /* 169.254.1.0. */
+ if (bfd->forward_if_rx_interval != setting->forward_if_rx_interval) {
+ bfd->forward_if_rx_interval = setting->forward_if_rx_interval;
+ bfd->forward_if_rx_detect_time = 0;
}
- ip_dst = smap_get(cfg, "bfd_dst_ip");
- if (ip_dst && bfd_lookup_ip(ip_dst, &in_addr)) {
- memcpy(&bfd->ip_dst, &in_addr, sizeof in_addr);
- } else {
- bfd->ip_dst = htonl(0xA9FE0101); /* 169.254.1.1. */
- }
-
- forwarding_if_rx = smap_get_bool(cfg, "forwarding_if_rx", false);
- if (bfd->forwarding_if_rx != forwarding_if_rx) {
- bfd->forwarding_if_rx = forwarding_if_rx;
- if (bfd->state == STATE_UP && bfd->forwarding_if_rx) {
- bfd_forwarding_if_rx_update(bfd);
- } else {
- bfd->forwarding_if_rx_detect_time = 0;
- }
+ if (bfd->decay_min_rx != setting->decay_min_rx || min_rx_changed) {
+ bfd->decay_min_rx = setting->decay_min_rx;
+ bfd->in_decay = false;
+ bfd->decay_rx_count = UINT32_MAX;
+ bfd->decay_detect_time = 0;
+ need_poll = true;
}
if (need_poll) {
bfd_poll(bfd);
}
- ovs_mutex_unlock(&mutex);
- return bfd;
-}
-
-struct bfd *
-bfd_ref(const struct bfd *bfd_)
-{
- struct bfd *bfd = CONST_CAST(struct bfd *, bfd_);
- if (bfd) {
- ovs_refcount_ref(&bfd->ref_cnt);
- }
- return bfd;
-}
-
-void
-bfd_unref(struct bfd *bfd) OVS_EXCLUDED(mutex)
-{
- if (bfd && ovs_refcount_unref(&bfd->ref_cnt) == 1) {
- ovs_mutex_lock(&mutex);
- hmap_remove(all_bfds, &bfd->node);
- netdev_close(bfd->netdev);
- ovs_refcount_destroy(&bfd->ref_cnt);
- free(bfd->name);
- free(bfd);
- ovs_mutex_unlock(&mutex);
- }
-}
-void
-bfd_wait(const struct bfd *bfd) OVS_EXCLUDED(mutex)
-{
- poll_timer_wait_until(bfd_wake_time(bfd));
+ return BFD_PASS;
}
-/* Returns the next wake up time. */
+/* Returns the wakeup time of the BFD session. */
long long int
-bfd_wake_time(const struct bfd *bfd) OVS_EXCLUDED(mutex)
+bfd_wait(const struct bfd *bfd)
{
- long long int retval;
+ long long int ret;
if (!bfd) {
return LLONG_MAX;
}
- ovs_mutex_lock(&mutex);
if (bfd->flags & FLAG_FINAL) {
- retval = 0;
+ ret = 0;
} else {
- retval = bfd->next_tx;
+ ret = bfd->next_tx;
if (bfd->state > STATE_DOWN) {
- retval = MIN(bfd->detect_time, retval);
+ ret = MIN(bfd->detect_time, ret);
+ }
+ if (bfd->state == STATE_UP && bfd->decay_min_rx) {
+ ret = MIN(bfd->decay_detect_time, ret);
}
}
- ovs_mutex_unlock(&mutex);
- return retval;
+
+ return ret;
}
+/* Updates the BFD sessions status. Checks decay and forward_if_rx.
+ * Initiates the POLL sequence if needed. */
void
-bfd_run(struct bfd *bfd) OVS_EXCLUDED(mutex)
+bfd_run(struct bfd *bfd, long long int now)
{
- long long int now;
- bool old_in_decay;
+ bool old;
- ovs_mutex_lock(&mutex);
- now = time_msec();
- old_in_decay = bfd->in_decay;
+ if (!bfd) {
+ return;
+ }
if (bfd->state > STATE_DOWN && now >= bfd->detect_time) {
- bfd_set_state(bfd, STATE_DOWN, DIAG_EXPIRED);
+ bfd_set_state(bfd, STATE_DOWN, DIAG_EXPIRED, now);
}
- bfd_forwarding__(bfd);
- /* Decay may only happen when state is STATE_UP, bfd->decay_min_rx is
- * configured, and decay_detect_time is reached. */
- if (bfd->state == STATE_UP && bfd->decay_min_rx > 0
- && now >= bfd->decay_detect_time) {
- bfd_try_decay(bfd);
- }
+ old = bfd->in_decay;
+ bfd_try_decay(bfd, now);
+
+ bfd_forward_if_rx(bfd, now);
+ bfd_check_forwarding_flap(bfd, now);
if (bfd->min_tx != bfd->cfg_min_tx
- || (bfd->min_rx != bfd->cfg_min_rx && bfd->min_rx != bfd->decay_min_rx)
- || bfd->in_decay != old_in_decay) {
+ || (!bfd->in_decay && bfd->min_rx != bfd->cfg_min_rx)
+ || (bfd->in_decay && bfd->min_rx != bfd->decay_min_rx)
+ || bfd->in_decay != old) {
bfd_poll(bfd);
}
- ovs_mutex_unlock(&mutex);
}
+/* Queries the 'bfd''s status, the function will fill in the
+ * 'struct bfd_status'. */
+void
+bfd_get_status(const struct bfd *bfd, struct bfd_status *s)
+{
+ if (!bfd || !s) {
+ return;
+ }
+
+ s->forwarding = bfd->last_forwarding;
+ s->mult = bfd->mult;
+ s->cpath_down = bfd->cpath_down;
+ s->tx_interval = bfd_tx_interval(bfd);
+ s->rx_interval = bfd_rx_interval(bfd);
+
+ s->local_disc = bfd->disc;
+ s->local_min_tx = bfd_min_tx(bfd);
+ s->local_min_rx = bfd->min_rx;
+ s->local_flags = bfd->flags;
+ s->local_state = bfd->state;
+ s->local_diag = bfd->diag;
+ s->detect_time = bfd->detect_time;
+ s->next_tx = bfd->next_tx;
+ s->last_tx = bfd->last_tx;
+
+ s->rmt_disc = bfd->rmt_disc;
+ s->rmt_min_tx = bfd->rmt_min_tx;
+ s->rmt_min_rx = bfd->rmt_min_rx;
+ s->rmt_flags = bfd->rmt_flags;
+ s->rmt_state = bfd->rmt_state;
+ s->rmt_diag = bfd->rmt_diag;
+
+ s->flap_count = bfd->flap_count;
+}
+
+/* Returns true if the interface on which BFD is running may be used to
+ * forward traffic according to the BFD session state. 'now' is the
+ * current time in milliseconds. */
bool
-bfd_should_send_packet(const struct bfd *bfd) OVS_EXCLUDED(mutex)
+bfd_forwarding(const struct bfd *bfd, long long int now)
{
- bool ret;
- ovs_mutex_lock(&mutex);
- ret = bfd->flags & FLAG_FINAL || time_msec() >= bfd->next_tx;
- ovs_mutex_unlock(&mutex);
- return ret;
+ return bfd_forwarding__(bfd, now);
}
+/* Sets the corresponding flags to indicate that packet
+ * is received from this monitored interface. */
void
-bfd_put_packet(struct bfd *bfd, struct ofpbuf *p,
- uint8_t eth_src[ETH_ADDR_LEN]) OVS_EXCLUDED(mutex)
+bfd_account_rx(struct bfd *bfd, uint32_t n_pkt)
+{
+ if (bfd->forward_if_rx_interval && n_pkt) {
+ bfd->forward_if_rx_data = true;
+ }
+
+ if (bfd->decay_min_rx && bfd->decay_rx_count != UINT32_MAX) {
+ bfd->decay_rx_count += n_pkt;
+ }
+}
+
+/* For send/recv BFD control packets. */
+/* Returns true if the BFD control packet should be sent for this BFD
+ * session. e.g. tx timeout or POLL flag is on. */
+bool
+bfd_should_send_packet(const struct bfd *bfd, long long int now)
+{
+ return bfd->flags & FLAG_FINAL || now >= bfd->next_tx;
+}
+
+/* Constructs the BFD control packet in payload. This function assumes that
+ * the payload is properly aligned. */
+enum bfd_error
+bfd_put_packet(struct bfd *bfd, void *p, size_t len, long long int now)
{
long long int min_tx, min_rx;
- struct udp_header *udp;
- struct eth_header *eth;
- struct ip_header *ip;
- struct msg *msg;
-
- ovs_mutex_lock(&mutex);
- if (bfd->next_tx) {
- long long int delay = time_msec() - bfd->next_tx;
- long long int interval = bfd_tx_interval(bfd);
- if (delay > interval * 3 / 2) {
- VLOG_INFO("%s: long delay of %lldms (expected %lldms) sending BFD"
- " control message", bfd->name, delay, interval);
- }
+ struct bfd_msg *msg = p;
+
+ if (!bfd || !p || len < BFD_PACKET_LEN) {
+ return BFD_EINVAL;
}
/* RFC 5880 Section 6.5
* A BFD Control packet MUST NOT have both the Poll (P) and Final (F) bits
* set. */
- ovs_assert(!(bfd->flags & FLAG_POLL) || !(bfd->flags & FLAG_FINAL));
-
- ofpbuf_reserve(p, 2); /* Properly align after the ethernet header. */
- eth = ofpbuf_put_uninit(p, sizeof *eth);
- memcpy(eth->eth_src, eth_src, ETH_ADDR_LEN);
- memcpy(eth->eth_dst, bfd->eth_dst, ETH_ADDR_LEN);
- eth->eth_type = htons(ETH_TYPE_IP);
-
- ip = ofpbuf_put_zeros(p, sizeof *ip);
- ip->ip_ihl_ver = IP_IHL_VER(5, 4);
- ip->ip_tot_len = htons(sizeof *ip + sizeof *udp + sizeof *msg);
- ip->ip_ttl = MAXTTL;
- ip->ip_tos = IPTOS_LOWDELAY | IPTOS_THROUGHPUT;
- ip->ip_proto = IPPROTO_UDP;
- put_16aligned_be32(&ip->ip_src, bfd->ip_src);
- put_16aligned_be32(&ip->ip_dst, bfd->ip_dst);
- ip->ip_csum = csum(ip, sizeof *ip);
-
- udp = ofpbuf_put_zeros(p, sizeof *udp);
- udp->udp_src = htons(bfd->udp_src);
- udp->udp_dst = htons(BFD_DEST_PORT);
- udp->udp_len = htons(sizeof *udp + sizeof *msg);
-
- msg = ofpbuf_put_uninit(p, sizeof *msg);
+ if ((bfd->flags & FLAG_POLL) && (bfd->flags & FLAG_FINAL)) {
+ return BFD_EPOLL;
+ }
+
msg->vers_diag = (BFD_VERSION << 5) | bfd->diag;
msg->flags = (bfd->state & STATE_MASK) | bfd->flags;
@@ -615,69 +419,39 @@ bfd_put_packet(struct bfd *bfd, struct ofpbuf *p,
bfd->flags &= ~FLAG_FINAL;
- log_msg(VLL_DBG, msg, "Sending BFD Message", bfd);
-
- bfd->last_tx = time_msec();
+ bfd->last_tx = now;
bfd_set_next_tx(bfd);
- ovs_mutex_unlock(&mutex);
+
+ return BFD_PASS;
}
+/* Given the packet header entries, check if the packet is BFD control
+ * packet. */
bool
-bfd_should_process_flow(const struct bfd *bfd_, const struct flow *flow,
- struct flow_wildcards *wc)
+bfd_should_process_packet(const __be16 eth_type, const uint8_t ip_proto,
+ const __be16 udp_dst)
{
- struct bfd *bfd = CONST_CAST(struct bfd *, bfd_);
- bool check_tnl_key;
-
- memset(&wc->masks.dl_dst, 0xff, sizeof wc->masks.dl_dst);
- if (bfd->eth_dst_set && memcmp(bfd->eth_dst, flow->dl_dst, ETH_ADDR_LEN)) {
- return false;
- }
-
- memset(&wc->masks.nw_proto, 0xff, sizeof wc->masks.nw_proto);
- memset(&wc->masks.tp_dst, 0xff, sizeof wc->masks.tp_dst);
-
- atomic_read(&bfd->check_tnl_key, &check_tnl_key);
- if (check_tnl_key) {
- memset(&wc->masks.tunnel.tun_id, 0xff, sizeof wc->masks.tunnel.tun_id);
- }
- return (flow->dl_type == htons(ETH_TYPE_IP)
- && flow->nw_proto == IPPROTO_UDP
- && flow->tp_dst == htons(BFD_DEST_PORT)
- && (!check_tnl_key || flow->tunnel.tun_id == htonll(0)));
+ return (eth_type == htons(0x0800) /* IP. */
+ && ip_proto == 17 /* UDP. */
+ && udp_dst == htons(BFD_DEST_PORT));
}
-void
-bfd_process_packet(struct bfd *bfd, const struct flow *flow,
- const struct ofpbuf *p) OVS_EXCLUDED(mutex)
+/* Processes the BFD control packet in payload 'p'. The payload length is
+ * provided. */
+enum bfd_error
+bfd_process_packet(struct bfd *bfd, void *p, size_t len, long long int now)
{
uint32_t rmt_min_rx, pkt_your_disc;
- enum state rmt_state;
- enum flags flags;
+ enum bfd_state rmt_state;
+ enum bfd_flags flags;
uint8_t version;
- struct msg *msg;
+ struct bfd_msg *msg = p;
- /* This function is designed to follow section RFC 5880 6.8.6 closely. */
-
- ovs_mutex_lock(&mutex);
- /* Increments the decay rx counter. */
- bfd->decay_rx_ctl++;
-
- bfd_forwarding__(bfd);
-
- if (flow->nw_ttl != 255) {
- /* XXX Should drop in the kernel to prevent DOS. */
- goto out;
+ if (!bfd || !p || len < BFD_PACKET_LEN) {
+ return BFD_EINVAL;
}
- msg = ofpbuf_at(p, (uint8_t *)p->l7 - (uint8_t *)p->data, BFD_PACKET_LEN);
- if (!msg) {
- VLOG_INFO_RL(&rl, "%s: Received too-short BFD control message (only "
- "%"PRIdPTR" bytes long, at least %d required).",
- bfd->name, (uint8_t *) ofpbuf_tail(p) - (uint8_t *) p->l7,
- BFD_PACKET_LEN);
- goto out;
- }
+ /* This function is designed to follow section RFC 5880 6.8.6 closely. */
/* RFC 5880 Section 6.8.6
* If the Length field is greater than the payload of the encapsulating
@@ -692,61 +466,46 @@ bfd_process_packet(struct bfd *bfd, const struct flow *flow,
rmt_state = msg->flags & STATE_MASK;
version = msg->vers_diag >> VERS_SHIFT;
- log_msg(VLL_DBG, msg, "Received BFD control message", bfd);
-
if (version != BFD_VERSION) {
- log_msg(VLL_WARN, msg, "Incorrect version", bfd);
- goto out;
+ goto err;
}
/* Technically this should happen after the length check. We don't support
* authentication however, so it's simpler to do the check first. */
if (flags & FLAG_AUTH) {
- log_msg(VLL_WARN, msg, "Authenticated control message with"
- " authentication disabled", bfd);
- goto out;
+ goto err;
}
if (msg->length != BFD_PACKET_LEN) {
- log_msg(VLL_WARN, msg, "Unexpected length", bfd);
if (msg->length < BFD_PACKET_LEN) {
- goto out;
+ goto err;
}
}
if (!msg->mult) {
- log_msg(VLL_WARN, msg, "Zero multiplier", bfd);
- goto out;
+ goto err;
}
if (flags & FLAG_MULTIPOINT) {
- log_msg(VLL_WARN, msg, "Unsupported multipoint flag", bfd);
- goto out;
+ goto err;
}
if (!msg->my_disc) {
- log_msg(VLL_WARN, msg, "NULL my_disc", bfd);
- goto out;
+ goto err;
}
pkt_your_disc = ntohl(msg->your_disc);
if (pkt_your_disc) {
/* Technically, we should use the your discriminator field to figure
* out which 'struct bfd' this packet is destined towards. That way a
- * bfd session could migrate from one interface to another
+ * BFD session could migrate from one interface to another
* transparently. This doesn't fit in with the OVS structure very
* well, so in this respect, we are not compliant. */
if (pkt_your_disc != bfd->disc) {
- log_msg(VLL_WARN, msg, "Incorrect your_disc", bfd);
- goto out;
+ goto err;
}
} else if (rmt_state > STATE_DOWN) {
- log_msg(VLL_WARN, msg, "Null your_disc", bfd);
- goto out;
- }
-
- if (bfd->rmt_state != rmt_state) {
- seq_change(connectivity_seq_get());
+ goto err;
}
bfd->rmt_disc = ntohl(msg->my_disc);
@@ -758,7 +517,6 @@ bfd_process_packet(struct bfd *bfd, const struct flow *flow,
bfd->min_tx = bfd->poll_min_tx;
bfd->min_rx = bfd->poll_min_rx;
bfd->flags &= ~FLAG_POLL;
- log_msg(VLL_INFO, msg, "Poll sequence terminated", bfd);
}
if (flags & FLAG_POLL) {
@@ -777,209 +535,107 @@ bfd_process_packet(struct bfd *bfd, const struct flow *flow,
if (bfd->next_tx) {
bfd_set_next_tx(bfd);
}
- log_msg(VLL_INFO, msg, "New remote min_rx", bfd);
}
bfd->rmt_min_tx = MAX(ntohl(msg->min_tx) / 1000, 1);
- bfd->detect_time = bfd_rx_interval(bfd) * bfd->mult + time_msec();
+ bfd->detect_time = bfd_rx_interval(bfd) * bfd->mult + now;
if (bfd->state == STATE_ADMIN_DOWN) {
- VLOG_DBG_RL(&rl, "Administratively down, dropping control message.");
goto out;
}
if (rmt_state == STATE_ADMIN_DOWN) {
if (bfd->state != STATE_DOWN) {
- bfd_set_state(bfd, STATE_DOWN, DIAG_RMT_DOWN);
+ bfd_set_state(bfd, STATE_DOWN, DIAG_RMT_DOWN, now);
}
} else {
switch (bfd->state) {
case STATE_DOWN:
if (rmt_state == STATE_DOWN) {
- bfd_set_state(bfd, STATE_INIT, bfd->diag);
+ bfd_set_state(bfd, STATE_INIT, bfd->diag, now);
} else if (rmt_state == STATE_INIT) {
- bfd_set_state(bfd, STATE_UP, bfd->diag);
+ bfd_set_state(bfd, STATE_UP, bfd->diag, now);
}
break;
case STATE_INIT:
if (rmt_state > STATE_DOWN) {
- bfd_set_state(bfd, STATE_UP, bfd->diag);
+ bfd_set_state(bfd, STATE_UP, bfd->diag, now);
}
break;
case STATE_UP:
if (rmt_state <= STATE_DOWN) {
- bfd_set_state(bfd, STATE_DOWN, DIAG_RMT_DOWN);
- log_msg(VLL_INFO, msg, "Remote signaled STATE_DOWN", bfd);
+ bfd_set_state(bfd, STATE_DOWN, DIAG_RMT_DOWN, now);
}
break;
case STATE_ADMIN_DOWN:
default:
- OVS_NOT_REACHED();
+ break;
}
}
- /* XXX: RFC 5880 Section 6.8.6 Demand mode related calculations here. */
out:
- bfd_forwarding__(bfd);
- ovs_mutex_unlock(&mutex);
-}
+ return BFD_PASS;
-/* Must be called when the netdev owned by 'bfd' should change. */
-void
-bfd_set_netdev(struct bfd *bfd, const struct netdev *netdev)
- OVS_EXCLUDED(mutex)
-{
- ovs_mutex_lock(&mutex);
- if (bfd->netdev != netdev) {
- netdev_close(bfd->netdev);
- bfd->netdev = netdev_ref(netdev);
- if (bfd->decay_min_rx && bfd->state == STATE_UP) {
- bfd_decay_update(bfd);
- }
- if (bfd->forwarding_if_rx && bfd->state == STATE_UP) {
- bfd_forwarding_if_rx_update(bfd);
- }
- bfd->rx_packets = bfd_rx_packets(bfd);
- }
- ovs_mutex_unlock(&mutex);
-}
-
-
-/* Updates the forwarding flag. If override is not configured and
- * the forwarding flag value changes, increments the flap count.
- *
- * Note this function may be called multiple times in a function
- * (e.g. bfd_account_rx) before and after the bfd state or status
- * change. This is to capture any forwarding flag flap. */
-static bool
-bfd_forwarding__(struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- long long int time;
- bool last_forwarding = bfd->last_forwarding;
-
- if (bfd->forwarding_override != -1) {
- return bfd->forwarding_override == 1;
- }
-
- time = bfd->forwarding_if_rx_detect_time;
- bfd->last_forwarding = (bfd->state == STATE_UP
- || (bfd->forwarding_if_rx && time > time_msec()))
- && bfd->rmt_diag != DIAG_PATH_DOWN
- && bfd->rmt_diag != DIAG_CPATH_DOWN
- && bfd->rmt_diag != DIAG_RCPATH_DOWN;
- if (bfd->last_forwarding != last_forwarding) {
- bfd->flap_count++;
- seq_change(connectivity_seq_get());
- }
- return bfd->last_forwarding;
+err:
+ return BFD_EMSG;
}
/* Helpers. */
-static bool
-bfd_lookup_ip(const char *host_name, struct in_addr *addr)
-{
- if (!inet_pton(AF_INET, host_name, addr)) {
- VLOG_ERR_RL(&rl, "\"%s\" is not a valid IP address", host_name);
- return false;
- }
- return true;
-}
-
-static bool
-bfd_in_poll(const struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- return (bfd->flags & FLAG_POLL) != 0;
-}
-
-static void
-bfd_poll(struct bfd *bfd) OVS_REQUIRES(mutex)
+/* Converts the BFD error code to string. */
+const char *
+bfd_error_to_str(enum bfd_error error)
{
- if (bfd->state > STATE_DOWN && !bfd_in_poll(bfd)
- && !(bfd->flags & FLAG_FINAL)) {
- bfd->poll_min_tx = bfd->cfg_min_tx;
- bfd->poll_min_rx = bfd->in_decay ? bfd->decay_min_rx : bfd->cfg_min_rx;
- bfd->flags |= FLAG_POLL;
- bfd->next_tx = 0;
- VLOG_INFO_RL(&rl, "%s: Initiating poll sequence", bfd->name);
+ switch (error) {
+ case BFD_PASS: return "No Error";
+ case BFD_EINVAL: return "Invalid Arguments";
+ case BFD_EPOLL: return "Both POLL And FINAL Set";
+ case BFD_EMSG: return "Bad Control Packet";
+ default: return "Not An Error Code";
}
}
-static long long int
-bfd_min_tx(const struct bfd *bfd) OVS_REQUIRES(mutex)
+/* Converts the BFD flags to string. */
+const char *
+bfd_flag_to_str(enum bfd_flags flags)
{
- /* RFC 5880 Section 6.8.3
- * When bfd.SessionState is not Up, the system MUST set
- * bfd.DesiredMinTxInterval to a value of not less than one second
- * (1,000,000 microseconds). This is intended to ensure that the
- * bandwidth consumed by BFD sessions that are not Up is negligible,
- * particularly in the case where a neighbor may not be running BFD. */
- return (bfd->state == STATE_UP ? bfd->min_tx : MAX(bfd->min_tx, 1000));
-}
-
-static long long int
-bfd_tx_interval(const struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- long long int interval = bfd_min_tx(bfd);
- return MAX(interval, bfd->rmt_min_rx);
-}
-
-static long long int
-bfd_rx_interval(const struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- return MAX(bfd->min_rx, bfd->rmt_min_tx);
-}
-
-static void
-bfd_set_next_tx(struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- long long int interval = bfd_tx_interval(bfd);
- interval -= interval * random_range(26) / 100;
- bfd->next_tx = bfd->last_tx + interval;
-}
-
-static const char *
-bfd_flag_str(enum flags flags)
-{
- struct ds ds = DS_EMPTY_INITIALIZER;
static char flag_str[128];
if (!flags) {
return "none";
}
+ memset(flag_str, 0, sizeof *flag_str);
+
if (flags & FLAG_MULTIPOINT) {
- ds_put_cstr(&ds, "multipoint ");
+ strcat(flag_str, "multipoint ");
}
if (flags & FLAG_DEMAND) {
- ds_put_cstr(&ds, "demand ");
+ strcat(flag_str, "demand ");
}
if (flags & FLAG_AUTH) {
- ds_put_cstr(&ds, "auth ");
+ strcat(flag_str, "auth");
}
if (flags & FLAG_CTL) {
- ds_put_cstr(&ds, "ctl ");
+ strcat(flag_str, "ctl");
}
if (flags & FLAG_FINAL) {
- ds_put_cstr(&ds, "final ");
+ strcat(flag_str, "final");
}
if (flags & FLAG_POLL) {
- ds_put_cstr(&ds, "poll ");
+ strcat(flag_str, "poll");
}
- /* Do not copy the trailing whitespace. */
- ds_chomp(&ds, ' ');
- ovs_strlcpy(flag_str, ds_cstr(&ds), sizeof flag_str);
- ds_destroy(&ds);
return flag_str;
}
-static const char *
-bfd_state_str(enum state state)
+/* Converts the BFD state code to string. */
+const char *
+bfd_state_to_str(enum bfd_state state)
{
switch (state) {
case STATE_ADMIN_DOWN: return "admin_down";
@@ -990,8 +646,9 @@ bfd_state_str(enum state state)
}
}
-static const char *
-bfd_diag_str(enum diag diag) {
+/* Converts the BFD diag to string. */
+const char *
+bfd_diag_to_str(enum bfd_diag diag) {
switch (diag) {
case DIAG_NONE: return "No Diagnostic";
case DIAG_EXPIRED: return "Control Detection Time Expired";
@@ -1004,286 +661,4 @@ bfd_diag_str(enum diag diag) {
case DIAG_RCPATH_DOWN: return "Reverse Concatenated Path Down";
default: return "Invalid Diagnostic";
}
-};
-
-static void
-log_msg(enum vlog_level level, const struct msg *p, const char *message,
- const struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- struct ds ds = DS_EMPTY_INITIALIZER;
-
- if (vlog_should_drop(THIS_MODULE, level, &rl)) {
- return;
- }
-
- ds_put_format(&ds,
- "%s: %s."
- "\n\tvers:%"PRIu8" diag:\"%s\" state:%s mult:%"PRIu8
- " length:%"PRIu8
- "\n\tflags: %s"
- "\n\tmy_disc:0x%"PRIx32" your_disc:0x%"PRIx32
- "\n\tmin_tx:%"PRIu32"us (%"PRIu32"ms)"
- "\n\tmin_rx:%"PRIu32"us (%"PRIu32"ms)"
- "\n\tmin_rx_echo:%"PRIu32"us (%"PRIu32"ms)",
- bfd->name, message, p->vers_diag >> VERS_SHIFT,
- bfd_diag_str(p->vers_diag & DIAG_MASK),
- bfd_state_str(p->flags & STATE_MASK),
- p->mult, p->length, bfd_flag_str(p->flags & FLAGS_MASK),
- ntohl(p->my_disc), ntohl(p->your_disc),
- ntohl(p->min_tx), ntohl(p->min_tx) / 1000,
- ntohl(p->min_rx), ntohl(p->min_rx) / 1000,
- ntohl(p->min_rx_echo), ntohl(p->min_rx_echo) / 1000);
- bfd_put_details(&ds, bfd);
- VLOG(level, "%s", ds_cstr(&ds));
- ds_destroy(&ds);
-}
-
-static void
-bfd_set_state(struct bfd *bfd, enum state state, enum diag diag)
- OVS_REQUIRES(mutex)
-{
- if (bfd->cpath_down) {
- diag = DIAG_CPATH_DOWN;
- }
-
- if (bfd->state != state || bfd->diag != diag) {
- if (!VLOG_DROP_INFO(&rl)) {
- struct ds ds = DS_EMPTY_INITIALIZER;
-
- ds_put_format(&ds, "%s: BFD state change: %s->%s"
- " \"%s\"->\"%s\".\n",
- bfd->name, bfd_state_str(bfd->state),
- bfd_state_str(state), bfd_diag_str(bfd->diag),
- bfd_diag_str(diag));
- bfd_put_details(&ds, bfd);
- VLOG_INFO("%s", ds_cstr(&ds));
- ds_destroy(&ds);
- }
-
- bfd->state = state;
- bfd->diag = diag;
-
- if (bfd->state <= STATE_DOWN) {
- bfd->rmt_state = STATE_DOWN;
- bfd->rmt_diag = DIAG_NONE;
- bfd->rmt_min_rx = 1;
- bfd->rmt_flags = 0;
- bfd->rmt_disc = 0;
- bfd->rmt_min_tx = 0;
- /* Resets the min_rx if in_decay. */
- if (bfd->in_decay) {
- bfd->min_rx = bfd->cfg_min_rx;
- bfd->in_decay = false;
- }
- }
- /* Resets the decay when state changes to STATE_UP
- * and decay_min_rx is configured. */
- if (bfd->state == STATE_UP && bfd->decay_min_rx) {
- bfd_decay_update(bfd);
- }
-
- seq_change(connectivity_seq_get());
- }
-}
-
-static uint64_t
-bfd_rx_packets(const struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- struct netdev_stats stats;
-
- if (!netdev_get_stats(bfd->netdev, &stats)) {
- return stats.rx_packets;
- } else {
- return 0;
- }
-}
-
-/* Decays the bfd->min_rx to bfd->decay_min_rx when 'diff' is less than
- * the 'expect' value. */
-static void
-bfd_try_decay(struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- int64_t diff, expect;
-
- /* The 'diff' is the difference between current interface rx_packets
- * stats and last-time check. The 'expect' is the recorded number of
- * bfd control packets received within an approximately decay_min_rx
- * (2000 ms if decay_min_rx is less than 2000 ms) interval.
- *
- * Since the update of rx_packets stats at interface happens
- * asynchronously to the bfd_rx_packets() function, the 'diff' value
- * can be jittered. Thusly, we double the decay_rx_ctl to provide
- * more wiggle room. */
- diff = bfd_rx_packets(bfd) - bfd->decay_rx_packets;
- expect = 2 * MAX(bfd->decay_rx_ctl, 1);
- bfd->in_decay = diff <= expect ? true : false;
- bfd_decay_update(bfd);
-}
-
-/* Updates the rx_packets, decay_rx_ctl and decay_detect_time. */
-static void
-bfd_decay_update(struct bfd * bfd) OVS_REQUIRES(mutex)
-{
- bfd->decay_rx_packets = bfd_rx_packets(bfd);
- bfd->decay_rx_ctl = 0;
- bfd->decay_detect_time = MAX(bfd->decay_min_rx, 2000) + time_msec();
-}
-
-static void
-bfd_forwarding_if_rx_update(struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- int64_t incr = bfd_rx_interval(bfd) * bfd->mult;
- bfd->forwarding_if_rx_detect_time = MAX(incr, 2000) + time_msec();
-}
-
-static uint32_t
-generate_discriminator(void)
-{
- uint32_t disc = 0;
-
- /* RFC 5880 Section 6.8.1
- * It SHOULD be set to a random (but still unique) value to improve
- * security. The value is otherwise outside the scope of this
- * specification. */
-
- while (!disc) {
- struct bfd *bfd;
-
- /* 'disc' is by definition random, so there's no reason to waste time
- * hashing it. */
- disc = random_uint32();
- HMAP_FOR_EACH_IN_BUCKET (bfd, node, disc, all_bfds) {
- if (bfd->disc == disc) {
- disc = 0;
- break;
- }
- }
- }
-
- return disc;
-}
-
-static struct bfd *
-bfd_find_by_name(const char *name) OVS_REQUIRES(mutex)
-{
- struct bfd *bfd;
-
- HMAP_FOR_EACH (bfd, node, all_bfds) {
- if (!strcmp(bfd->name, name)) {
- return bfd;
- }
- }
- return NULL;
-}
-
-static void
-bfd_put_details(struct ds *ds, const struct bfd *bfd) OVS_REQUIRES(mutex)
-{
- ds_put_format(ds, "\tForwarding: %s\n",
- bfd_forwarding__(CONST_CAST(struct bfd *, bfd))
- ? "true" : "false");
- ds_put_format(ds, "\tDetect Multiplier: %d\n", bfd->mult);
- ds_put_format(ds, "\tConcatenated Path Down: %s\n",
- bfd->cpath_down ? "true" : "false");
- ds_put_format(ds, "\tTX Interval: Approx %lldms\n", bfd_tx_interval(bfd));
- ds_put_format(ds, "\tRX Interval: Approx %lldms\n", bfd_rx_interval(bfd));
- ds_put_format(ds, "\tDetect Time: now %+lldms\n",
- time_msec() - bfd->detect_time);
- ds_put_format(ds, "\tNext TX Time: now %+lldms\n",
- time_msec() - bfd->next_tx);
- ds_put_format(ds, "\tLast TX Time: now %+lldms\n",
- time_msec() - bfd->last_tx);
-
- ds_put_cstr(ds, "\n");
-
- ds_put_format(ds, "\tLocal Flags: %s\n", bfd_flag_str(bfd->flags));
- ds_put_format(ds, "\tLocal Session State: %s\n",
- bfd_state_str(bfd->state));
- ds_put_format(ds, "\tLocal Diagnostic: %s\n", bfd_diag_str(bfd->diag));
- ds_put_format(ds, "\tLocal Discriminator: 0x%"PRIx32"\n", bfd->disc);
- ds_put_format(ds, "\tLocal Minimum TX Interval: %lldms\n",
- bfd_min_tx(bfd));
- ds_put_format(ds, "\tLocal Minimum RX Interval: %lldms\n", bfd->min_rx);
-
- ds_put_cstr(ds, "\n");
-
- ds_put_format(ds, "\tRemote Flags: %s\n", bfd_flag_str(bfd->rmt_flags));
- ds_put_format(ds, "\tRemote Session State: %s\n",
- bfd_state_str(bfd->rmt_state));
- ds_put_format(ds, "\tRemote Diagnostic: %s\n",
- bfd_diag_str(bfd->rmt_diag));
- ds_put_format(ds, "\tRemote Discriminator: 0x%"PRIx32"\n", bfd->rmt_disc);
- ds_put_format(ds, "\tRemote Minimum TX Interval: %lldms\n",
- bfd->rmt_min_tx);
- ds_put_format(ds, "\tRemote Minimum RX Interval: %lldms\n",
- bfd->rmt_min_rx);
-}
-
-static void
-bfd_unixctl_show(struct unixctl_conn *conn, int argc, const char *argv[],
- void *aux OVS_UNUSED) OVS_EXCLUDED(mutex)
-{
- struct ds ds = DS_EMPTY_INITIALIZER;
- struct bfd *bfd;
-
- ovs_mutex_lock(&mutex);
- if (argc > 1) {
- bfd = bfd_find_by_name(argv[1]);
- if (!bfd) {
- unixctl_command_reply_error(conn, "no such bfd object");
- goto out;
- }
- bfd_put_details(&ds, bfd);
- } else {
- HMAP_FOR_EACH (bfd, node, all_bfds) {
- ds_put_format(&ds, "---- %s ----\n", bfd->name);
- bfd_put_details(&ds, bfd);
- }
- }
- unixctl_command_reply(conn, ds_cstr(&ds));
- ds_destroy(&ds);
-
-out:
- ovs_mutex_unlock(&mutex);
-}
-
-
-static void
-bfd_unixctl_set_forwarding_override(struct unixctl_conn *conn, int argc,
- const char *argv[], void *aux OVS_UNUSED)
- OVS_EXCLUDED(mutex)
-{
- const char *forward_str = argv[argc - 1];
- int forwarding_override;
- struct bfd *bfd;
-
- ovs_mutex_lock(&mutex);
- if (!strcasecmp("true", forward_str)) {
- forwarding_override = 1;
- } else if (!strcasecmp("false", forward_str)) {
- forwarding_override = 0;
- } else if (!strcasecmp("normal", forward_str)) {
- forwarding_override = -1;
- } else {
- unixctl_command_reply_error(conn, "unknown fault string");
- goto out;
- }
-
- if (argc > 2) {
- bfd = bfd_find_by_name(argv[1]);
- if (!bfd) {
- unixctl_command_reply_error(conn, "no such BFD object");
- goto out;
- }
- bfd->forwarding_override = forwarding_override;
- } else {
- HMAP_FOR_EACH (bfd, node, all_bfds) {
- bfd->forwarding_override = forwarding_override;
- }
- }
-
- unixctl_command_reply(conn, "OK");
-
-out:
- ovs_mutex_unlock(&mutex);
}
diff --git a/lib/bfd.h b/lib/bfd.h
index 4e7d4cb..99f7006 100644
--- a/lib/bfd.h
+++ b/lib/bfd.h
@@ -1,4 +1,4 @@
-/* Copyright (c) 2012 Nicira, Inc.
+/* Copyright (c) 2014 Nicira, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
@@ -10,46 +10,203 @@
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
- * limitations under the License. */
+ * limitations under the License. */
#ifndef BFD_H
#define BFD_H 1
+#include <linux/types.h>
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdint.h>
+
+#define BFD_VERSION 1
#define BFD_PACKET_LEN 24
#define BFD_DEST_PORT 3784
-#include <stdbool.h>
-#include <inttypes.h>
-
-struct bfd;
-struct dpif_flow_stats;
-struct flow;
-struct flow_wildcards;
-struct netdev;
-struct ofpbuf;
-struct smap;
-
-void bfd_wait(const struct bfd *);
-void bfd_run(struct bfd *);
-
-bool bfd_should_send_packet(const struct bfd *);
-void bfd_put_packet(struct bfd *bfd, struct ofpbuf *packet,
- uint8_t eth_src[6]);
-
-bool bfd_should_process_flow(const struct bfd *, const struct flow *,
- struct flow_wildcards *);
-void bfd_process_packet(struct bfd *, const struct flow *,
- const struct ofpbuf *);
-
-struct bfd *bfd_configure(struct bfd *, const char *name,
- const struct smap *smap,
- struct netdev *netdev);
-struct bfd *bfd_ref(const struct bfd *);
-void bfd_unref(struct bfd *);
-
-void bfd_account_rx(struct bfd *, const struct dpif_flow_stats *);
-bool bfd_forwarding(struct bfd *);
-void bfd_get_status(const struct bfd *, struct smap *);
-void bfd_set_netdev(struct bfd *, const struct netdev *);
-long long int bfd_wake_time(const struct bfd *);
+#define VERS_SHIFT 5
+#define DIAG_MASK 0x1f
+#define STATE_MASK 0xC0
+#define FLAGS_MASK 0x3f
+
+/* RFC 5880 Section 4.1
+ * 0 1 2 3
+ * 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ * |Vers | Diag |Sta|P|F|C|A|D|M| Detect Mult | Length |
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ * | My Discriminator |
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ * | Your Discriminator |
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ * | Desired Min TX Interval |
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ * | Required Min RX Interval |
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ * | Required Min Echo RX Interval |
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ */
+struct bfd_msg {
+ uint8_t vers_diag; /* Version and diagnostic. */
+ uint8_t flags; /* 2bit State field followed by flags. */
+ uint8_t mult; /* Fault detection multiplier. */
+ uint8_t length; /* Length of this BFD message. */
+ __be32 my_disc; /* My discriminator. */
+ __be32 your_disc; /* Your discriminator. */
+ __be32 min_tx; /* Desired minimum tx interval. */
+ __be32 min_rx; /* Required minimum rx interval. */
+ __be32 min_rx_echo; /* Required minimum echo rx interval. */
+};
+
+enum bfd_flags {
+ FLAG_MULTIPOINT = 1 << 0,
+ FLAG_DEMAND = 1 << 1,
+ FLAG_AUTH = 1 << 2,
+ FLAG_CTL = 1 << 3,
+ FLAG_FINAL = 1 << 4,
+ FLAG_POLL = 1 << 5
+};
+
+enum bfd_state {
+ STATE_ADMIN_DOWN = 0 << 6,
+ STATE_DOWN = 1 << 6,
+ STATE_INIT = 2 << 6,
+ STATE_UP = 3 << 6
+};
+
+enum bfd_diag {
+ DIAG_NONE = 0, /* No Diagnostic. */
+ DIAG_EXPIRED = 1, /* Control Detection Time Expired. */
+ DIAG_ECHO_FAILED = 2, /* Echo Function Failed. */
+ DIAG_RMT_DOWN = 3, /* Neighbor Signaled Session Down. */
+ DIAG_FWD_RESET = 4, /* Forwarding Plane Reset. */
+ DIAG_PATH_DOWN = 5, /* Path Down. */
+ DIAG_CPATH_DOWN = 6, /* Concatenated Path Down. */
+ DIAG_ADMIN_DOWN = 7, /* Administratively Down. */
+ DIAG_RCPATH_DOWN = 8 /* Reverse Concatenated Path Down. */
+};
+
+enum bfd_error {
+ BFD_PASS = 0, /* No error. */
+ BFD_EINVAL = 1, /* Invalid arguments. */
+ BFD_EPOLL = 2, /* bfd poll and final flags are both on. */
+ BFD_EMSG = 3 /* bfd control packet error. */
+};
+
+/* Used to configure a BFD session. */
+struct bfd_setting {
+ /* Local state variables. */
+ uint32_t disc; /* bfd.LocalDiscr. */
+ uint8_t mult; /* bfd.DetectMult. */
+ uint32_t min_tx; /* bfd.DesiredMinTxInterval. */
+ uint32_t min_rx; /* bfd.RequiredMinRxInterval. */
+
+ /* Open Vswitch specific settings. */
+ bool cpath_down; /* Set Concatenated Path Down. */
+ bool forwarding_override; /* Manual override of 'forwarding' status. */
+ int forward_if_rx_interval; /* How often to detect forward_if_rx. */
+ int decay_min_rx; /* bfd.min_rx is set to decay_min_rx when */
+ /* in decay. */
+};
+
+/* BFD status. */
+struct bfd_status {
+ bool forwarding; /* The liveness of bfd session. */
+ uint8_t mult; /* bfd.DetectMult. */
+ bool cpath_down; /* If cpath_down enabled. */
+ uint32_t tx_interval; /* tx interval in use. */
+ uint32_t rx_interval; /* rx interval in use. */
+
+ uint32_t local_disc; /* bfd.LocalDiscr. */
+ uint32_t local_min_tx; /* bfd.DesiredMinTxInterval */
+ uint32_t local_min_rx; /* bfd.DesiredMinRxInterval*/
+ enum bfd_flags local_flags; /* Flags sent on messages. */
+ enum bfd_state local_state; /* bfd.SessionState. */
+ enum bfd_diag local_diag; /* bfd.LocalDiag. */
+ long long int detect_time; /* RFC 5880 6.8.4 Detection time. */
+ long long int last_tx; /* Last TX time. */
+ long long int next_tx; /* Next TX time. */
+
+ uint32_t rmt_disc; /* bfd.RemoteDiscr. */
+ uint32_t rmt_min_tx; /* bfd.RemoteMinTxInterval */
+ uint32_t rmt_min_rx; /* bfd.RemoteMinRxInterval*/
+ enum bfd_flags rmt_flags; /* Flags last received. */
+ enum bfd_state rmt_state; /* bfd.RemoteSessionState. */
+ enum bfd_diag rmt_diag; /* Remote diagnostic. */
+
+ uint64_t flap_count; /* Flap count of forwarding. */
+};
+
+/* A BFD session. Users are not permitted to directly access the variable
+ * of this struct. For BFD configuration, use the bfd_configure(). For
+ * BFD status extraction, use the bfd_get_status(). */
+struct bfd {
+ /* Local state variables. */
+ uint32_t disc; /* bfd.LocalDiscr. */
+ uint8_t mult; /* bfd.DetectMult. */
+ enum bfd_state state; /* bfd.SessionState. */
+ enum bfd_diag diag; /* bfd.LocalDiag. */
+ enum bfd_flags flags; /* Flags sent on messages. */
+ uint32_t min_tx; /* bfd.DesiredMinTxInterval. */
+ uint32_t min_rx; /* bfd.RequiredMinRxInterval. */
+ uint32_t cfg_min_tx; /* Configured minimum TX rate. */
+ uint32_t cfg_min_rx; /* Configured required minimum RX rate. */
+ long long int detect_time; /* RFC 5880 6.8.4 Detection time. */
+ long long int last_tx; /* Last TX time. */
+ long long int next_tx; /* Next TX time. */
+
+ /* Remote side state variables. */
+ uint32_t rmt_disc; /* bfd.RemoteDiscr. */
+ enum bfd_state rmt_state; /* bfd.RemoteSessionState. */
+ enum bfd_diag rmt_diag; /* Remote diagnostic. */
+ enum bfd_flags rmt_flags; /* Flags last received. */
+ long long int rmt_min_rx; /* bfd.RemoteMinRxInterval. */
+ long long int rmt_min_tx; /* Remote minimum TX interval. */
+
+ /* POLL sequence. */
+ uint32_t poll_min_tx; /* min_tx in POLL sequence. */
+ uint32_t poll_min_rx; /* min_rx in POLL sequence. */
+
+ /* Open Vswitch specific features. */
+ bool cpath_down; /* Set Concatenated Path Down. */
+
+ int forwarding_override; /* Manual override of 'forwarding' status. */
+
+ /* Equivalent to bfd demand mode. */
+ bool last_forwarding; /* Last calculation of forwarding flag. */
+ int forward_if_rx_interval; /* How often to detect forward_if_rx. */
+ long long int forward_if_rx_detect_time;
+ bool forward_if_rx_data; /* Data packet received in last interval. */
+
+ /* BFD decay feature is for reducing the */
+ bool in_decay; /* True when bfd is in decay. */
+ int decay_min_rx; /* bfd.min_rx is set to decay_min_rx when */
+ /* in decay. */
+ long long int decay_detect_time; /* Next decay detect time. */
+ uint32_t decay_rx_count; /* Count of data packets received. */
+
+ uint64_t flap_count; /* Counts bfd forwarding flaps. */
+};
+
+enum bfd_error bfd_configure(struct bfd *, const struct bfd_setting *);
+long long int bfd_wait(const struct bfd *);
+
+void bfd_run(struct bfd *, long long int now);
+void bfd_get_status(const struct bfd *, struct bfd_status *);
+bool bfd_forwarding(const struct bfd *, long long int now);
+void bfd_account_rx(struct bfd *, uint32_t n_pkt);
+
+bool bfd_should_send_packet(const struct bfd *, long long int now);
+enum bfd_error bfd_put_packet(struct bfd *, void *, size_t len,
+ long long int now);
+bool bfd_should_process_packet(const __be16 eth_type, const uint8_t ip_proto,
+ const __be16 udp_dst);
+enum bfd_error bfd_process_packet(struct bfd *, void *, size_t len,
+ long long now);
+
+/* Helpers. */
+const char * bfd_error_to_str(enum bfd_error error);
+const char * bfd_flag_to_str(enum bfd_flags flags);
+const char * bfd_state_to_str(enum bfd_state state);
+const char * bfd_diag_to_str(enum bfd_diag diag);
+
#endif /* bfd.h */
diff --git a/lib/bfd_ts.c b/lib/bfd_ts.c
new file mode 100644
index 0000000..f05e0b0
--- /dev/null
+++ b/lib/bfd_ts.c
@@ -0,0 +1,620 @@
+/* Copyright (c) 2014 Nicira, Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License. */
+
+#include <config.h>
+
+#include "bfd.h"
+#include "bfd_ts.h"
+
+#include <sys/types.h>
+#include <arpa/inet.h>
+#include <netinet/in_systm.h>
+#include <netinet/ip.h>
+#include <errno.h>
+
+#include "byte-order.h"
+#include "connectivity.h"
+#include "csum.h"
+#include "dpif.h"
+#include "dynamic-string.h"
+#include "flow.h"
+#include "hmap.h"
+#include "odp-util.h"
+#include "ofpbuf.h"
+#include "ovs-thread.h"
+#include "openvswitch/types.h"
+#include "packets.h"
+#include "poll-loop.h"
+#include "random.h"
+#include "seq.h"
+#include "smap.h"
+#include "timeval.h"
+#include "unaligned.h"
+#include "unixctl.h"
+#include "util.h"
+#include "vlog.h"
+
+VLOG_DEFINE_THIS_MODULE(bfd_ts);
+
+static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(20, 20);
+
+static struct ovs_mutex mutex = OVS_MUTEX_INITIALIZER;
+static struct hmap all_bfds__ = HMAP_INITIALIZER(&all_bfds__);
+static struct hmap *const all_bfds OVS_GUARDED_BY(mutex) = &all_bfds__;
+
+/* Thread-safe wrapper for bfd.{c,h}.
+ * The 'struct bfd_ts' is the thread-safe wrapper for 'struct bfd'. It
+ * contains the additional information (e.g. monitored interface name)
+ * and header fields needed for construction and parse of the BFD control
+ * packet. Each 'struct bfd' should be created by creating the
+ * 'struct bfd_ts'. Thusly, the argument 'struct bfd *' can be converted to
+ * the corresponding 'struct bfd_ts *'. */
+struct bfd_ts {
+ struct hmap_node node; /* In 'all_bfds'. */
+ struct bfd bfd; /* The BFD session. */
+ char *name; /* Name of the monitored interface. */
+ uint32_t disc; /* bfd.Discriminator. */
+ uint64_t last_flap_count; /* Last check of bfd.flap_count value. */
+ atomic_int ref_cnt; /* Reference counter. */
+
+ bool check_tnl_key; /* Verify tunnel key of inbound packets? */
+ uint8_t eth_dst[ETH_ADDR_LEN];/* Ethernet destination address. */
+ bool eth_dst_set; /* 'eth_dst' set through database. */
+ ovs_be32 ip_src; /* IPv4 source address. */
+ ovs_be32 ip_dst; /* IPv4 destination address. */
+ uint32_t udp_src; /* UDP source port. */
+};
+
+
+static void
+bfd_ts_check_flap(struct bfd *bfd) OVS_REQUIRES(mutex)
+{
+ struct bfd_ts *ts = CONTAINER_OF(bfd, struct bfd_ts, bfd);
+ struct bfd_status s;
+
+ bfd_get_status(bfd, &s);
+
+ if (ts->last_flap_count != s.flap_count) {
+ seq_change(connectivity_seq_get());
+ ts->last_flap_count = s.flap_count;
+ }
+}
+
+static uint32_t
+bfd_ts_generate_discriminator(void) OVS_REQUIRES(mutex)
+{
+ uint32_t disc = 0;
+
+ /* RFC 5880 Section 6.8.1
+ * It SHOULD be set to a random (but still unique) value to improve
+ * security. The value is otherwise outside the scope of this
+ * specification. */
+ while (!disc) {
+ struct bfd_ts *ts;
+
+ /* 'disc' is by definition random, so there's no reason to waste time
+ * hashing it. */
+ disc = random_uint32();
+ HMAP_FOR_EACH_IN_BUCKET (ts, node, disc, all_bfds) {
+ if (ts->disc == disc) {
+ disc = 0;
+ break;
+ }
+ }
+ }
+
+ return disc;
+}
+
+static bool
+bfd_ts_lookup_ip(const char *host_name, struct in_addr *addr)
+ OVS_REQUIRES(mutex)
+{
+ if (!inet_pton(AF_INET, host_name, addr)) {
+ VLOG_ERR("\"%s\" is not a valid IP address", host_name);
+
+ return false;
+ }
+
+ return true;
+}
+
+static struct bfd *
+bfd_ts_find_by_name(const char *name)
+ OVS_REQUIRES(mutex)
+{
+ struct bfd_ts *ts;
+
+ HMAP_FOR_EACH (ts, node, all_bfds) {
+ if (!strcmp(ts->name, name)) {
+ return &ts->bfd;
+ }
+ }
+
+ return NULL;
+}
+
+static void
+bfd_ts_put_details(struct ds *ds, const struct bfd *bfd)
+ OVS_REQUIRES(mutex)
+{
+ struct bfd_status status;
+ bfd_get_status(bfd, &status);
+
+ ds_put_format(ds, "\tForwarding: %s\n",
+ bfd_forwarding(bfd, time_msec()) ? "true" : "false");
+ ds_put_format(ds, "\tDetect Multiplier: %d\n", status.mult);
+ ds_put_format(ds, "\tConcatenated Path Down: %s\n",
+ status.cpath_down ? "true" : "false");
+ ds_put_format(ds, "\tTX Interval: Approx %ums\n", status.tx_interval);
+ ds_put_format(ds, "\tRX Interval: Approx %ums\n", status.rx_interval);
+ ds_put_format(ds, "\tDetect Time: now %+lldms\n",
+ time_msec() - status.detect_time);
+ ds_put_format(ds, "\tNext TX Time: now %+lldms\n",
+ time_msec() - status.next_tx);
+ ds_put_format(ds, "\tLast TX Time: now %+lldms\n",
+ time_msec() - status.last_tx);
+
+ ds_put_cstr(ds, "\n");
+
+ ds_put_format(ds, "\tLocal Flags: %s\n",
+ bfd_flag_to_str(status.local_flags));
+ ds_put_format(ds, "\tLocal Session State: %s\n",
+ bfd_state_to_str(status.local_state));
+ ds_put_format(ds, "\tLocal Diagnostic: %s\n",
+ bfd_diag_to_str(status.local_diag));
+ ds_put_format(ds, "\tLocal Discriminator: 0x%"PRIx32"\n",
+ status.local_disc);
+ ds_put_format(ds, "\tLocal Minimum TX Interval: %dms\n",
+ status.local_min_tx);
+ ds_put_format(ds, "\tLocal Minimum RX Interval: %ums\n",
+ status.local_min_rx);
+
+ ds_put_cstr(ds, "\n");
+
+ ds_put_format(ds, "\tRemote Flags: %s\n",
+ bfd_flag_to_str(status.rmt_flags));
+ ds_put_format(ds, "\tRemote Session State: %s\n",
+ bfd_state_to_str(status.rmt_state));
+ ds_put_format(ds, "\tRemote Diagnostic: %s\n",
+ bfd_diag_to_str(status.rmt_diag));
+ ds_put_format(ds, "\tRemote Discriminator: 0x%"PRIx32"\n",
+ status.rmt_disc);
+ ds_put_format(ds, "\tRemote Minimum TX Interval: %ums\n",
+ status.rmt_min_tx);
+ ds_put_format(ds, "\tRemote Minimum RX Interval: %ums\n",
+ status.rmt_min_rx);
+}
+
+static void
+bfd_ts_unixctl_show(struct unixctl_conn *conn, int argc, const char *argv[],
+ void *aux OVS_UNUSED)
+ OVS_EXCLUDED(mutex)
+{
+ struct ds ds = DS_EMPTY_INITIALIZER;
+
+ ovs_mutex_lock(&mutex);
+ if (argc > 1) {
+ struct bfd *bfd;
+
+ bfd = bfd_ts_find_by_name(argv[1]);
+ if (!bfd) {
+ unixctl_command_reply_error(conn, "no such bfd object");
+ goto out;
+ }
+ bfd_ts_put_details(&ds, bfd);
+ } else {
+ struct bfd_ts *ts;
+
+ HMAP_FOR_EACH (ts, node, all_bfds) {
+ ds_put_format(&ds, "---- %s ----\n", ts->name);
+ bfd_ts_put_details(&ds, &ts->bfd);
+ }
+ }
+ unixctl_command_reply(conn, ds_cstr(&ds));
+ ds_destroy(&ds);
+
+out:
+ ovs_mutex_unlock(&mutex);
+}
+
+static void
+bfd_ts_unixctl_set_forwarding_override(struct unixctl_conn *conn, int argc,
+ const char *argv[], void *aux OVS_UNUSED)
+ OVS_EXCLUDED(mutex)
+{
+ const char *forward_str = argv[argc - 1];
+ int forwarding_override;
+
+ ovs_mutex_lock(&mutex);
+ if (!strcasecmp("true", forward_str)) {
+ forwarding_override = 1;
+ } else if (!strcasecmp("false", forward_str)) {
+ forwarding_override = -1;
+ } else if (!strcasecmp("normal", forward_str)) {
+ forwarding_override = 0;
+ } else {
+ unixctl_command_reply_error(conn, "unknown fault string");
+ goto out;
+ }
+
+ if (argc > 2) {
+ struct bfd *bfd;
+
+ bfd = bfd_ts_find_by_name(argv[1]);
+ if (!bfd) {
+ unixctl_command_reply_error(conn, "no such BFD object");
+ goto out;
+ }
+ bfd->forwarding_override = forwarding_override;
+ } else {
+ struct bfd_ts *ts;
+
+ HMAP_FOR_EACH (ts, node, all_bfds) {
+ ts->bfd.forwarding_override = forwarding_override;
+ }
+ }
+
+ unixctl_command_reply(conn, "OK");
+
+out:
+ ovs_mutex_unlock(&mutex);
+}
+
+
+/* Initializes, destroys, or reconfigures the BFD session 'bfd' and the wrapper
+ * 'struct bfd_ts' (named 'name'), according to the database configuration
+ * contained in 'cfg'. Takes ownership of 'bfd', which may be NULL. Returns
+ * a BFD object which may be used as a handle for the session, or NULL if BFD
+ * is not enabled according to 'cfg'. Also returns NULL if cfg is NULL. */
+struct bfd *
+bfd_ts_configure(struct bfd *bfd, const char *name, const struct smap *cfg)
+{
+ static struct ovsthread_once once = OVSTHREAD_ONCE_INITIALIZER;
+ static atomic_uint16_t udp_src = ATOMIC_VAR_INIT(0);
+ struct bfd_ts *ts;
+ struct bfd_setting setting;
+ const char *hwaddr, *ip_src, *ip_dst;
+ struct in_addr in_addr;
+ uint8_t ea[ETH_ADDR_LEN];
+
+ if (ovsthread_once_start(&once)) {
+ unixctl_command_register("bfd/show", "[interface]", 0, 1,
+ bfd_ts_unixctl_show, NULL);
+ unixctl_command_register("bfd/set-forwarding",
+ "[interface] normal|false|true", 1, 2,
+ bfd_ts_unixctl_set_forwarding_override, NULL);
+ ovsthread_once_done(&once);
+ }
+
+ if (!cfg || !smap_get_bool(cfg, "enable", false)) {
+ bfd_ts_unref(bfd);
+ return NULL;
+ }
+
+ ovs_mutex_lock(&mutex);
+ if (!bfd) {
+ ts = xzalloc(sizeof *ts);
+
+ ts->name = xstrdup(name);
+ ts->disc = bfd_ts_generate_discriminator();
+ atomic_init(&ts->ref_cnt, 1);
+
+ /* RFC 5881 section 4
+ * The source port MUST be in the range 49152 through 65535. The same
+ * UDP source port number MUST be used for all BFD Control packets
+ * associated with a particular session. The source port number SHOULD
+ * be unique among all BFD sessions on the system. */
+ atomic_add(&udp_src, 1, &ts->udp_src);
+ ts->udp_src = (ts->udp_src % 16384) + 49152;
+
+ hmap_insert(all_bfds, &ts->node, ts->disc);
+ } else {
+ ts = CONTAINER_OF(bfd, struct bfd_ts, bfd);
+ }
+
+ /* Congfiures the 'struct bfd_ts'. */
+ ts->check_tnl_key = smap_get_bool(cfg, "check_tnl_key", false);
+
+ hwaddr = smap_get(cfg, "bfd_dst_mac");
+ if (hwaddr && eth_addr_from_string(hwaddr, ea) && !eth_addr_is_zero(ea)) {
+ memcpy(ts->eth_dst, ea, ETH_ADDR_LEN);
+ ts->eth_dst_set = true;
+ } else if (ts->eth_dst_set) {
+ memcpy(ts->eth_dst, eth_addr_bfd, ETH_ADDR_LEN);
+ ts->eth_dst_set = false;
+ }
+
+ ip_src = smap_get(cfg, "bfd_src_ip");
+ if (ip_src && bfd_ts_lookup_ip(ip_src, &in_addr)) {
+ memcpy(&ts->ip_src, &in_addr, sizeof in_addr);
+ } else {
+ ts->ip_src = htonl(0xA9FE0100); /* 169.254.1.0. */
+ }
+
+ ip_dst = smap_get(cfg, "bfd_dst_ip");
+ if (ip_dst && bfd_ts_lookup_ip(ip_dst, &in_addr)) {
+ memcpy(&ts->ip_dst, &in_addr, sizeof in_addr);
+ } else {
+ ts->ip_dst = htonl(0xA9FE0101); /* 169.254.1.1. */
+ }
+
+ /* Configures the 'struct bfd'. */
+ setting.disc = ts->disc;
+ setting.mult = 3;
+ setting.min_tx = smap_get_int(cfg, "min_tx", 100);
+ setting.min_rx = smap_get_int(cfg, "min_rx", 1000);
+ setting.cpath_down = smap_get_bool(cfg, "cpath_down", false);
+ setting.forward_if_rx_interval = smap_get_bool(cfg, "forwarding_if_rx",
+ 0) ? 3000 : 0;
+ setting.decay_min_rx = smap_get_int(cfg, "decay_min_rx", 0);
+
+ /* Since forwarding_override can only be set via 'ovs-appctl'
+ * command, use this hack to retain its configured value. */
+ setting.forwarding_override = ts->bfd.forwarding_override;
+ bfd_configure(&ts->bfd, &setting);
+
+ ovs_mutex_unlock(&mutex);
+
+ return &ts->bfd;
+}
+
+/* Called when accounting the number of packets received from the interface. */
+void
+bfd_ts_account_rx(struct bfd *bfd, uint32_t n_pkt)
+{
+ ovs_mutex_lock(&mutex);
+ bfd_account_rx(bfd, n_pkt);
+ ovs_mutex_unlock(&mutex);
+}
+
+/* References the 'bfd' in the corresponding 'strcut bfd_ts'. */
+struct bfd *
+bfd_ts_ref(const struct bfd *bfd_)
+{
+ struct bfd *bfd = CONST_CAST(struct bfd *, bfd_);
+ struct bfd_ts *ts;
+
+ if (bfd) {
+ int orig;
+
+ ts = CONTAINER_OF(bfd, struct bfd_ts, bfd);
+ atomic_add(&ts->ref_cnt, 1, &orig);
+ ovs_assert(orig > 0);
+ }
+ return bfd;
+}
+
+/* Un-references the 'bfd' in the corresponding 'strcut bfd_ts'. Frees
+ * the memory if the new 'ref_count' is 0. */
+void
+bfd_ts_unref(struct bfd *bfd)
+{
+ if (bfd) {
+ int orig;
+ struct bfd_ts *ts = CONTAINER_OF(bfd, struct bfd_ts, bfd);
+
+ atomic_sub(&ts->ref_cnt, 1, &orig);
+ ovs_assert(orig > 0);
+ if (orig == 1) {
+ ovs_mutex_lock(&mutex);
+ hmap_remove(all_bfds, &ts->node);
+ free(ts->name);
+ free(ts);
+ ovs_mutex_unlock(&mutex);
+ }
+ }
+}
+
+/* Returns a 'smap' of key value pairs representing the status of 'bfd'
+ * intended for the OVS database. */
+int
+bfd_ts_get_status(struct bfd *bfd, struct smap *smap)
+{
+ struct bfd_status status;
+
+ ovs_mutex_lock(&mutex);
+ if (!bfd) {
+ ovs_mutex_unlock(&mutex);
+ return ENOENT;
+ }
+
+ bfd_get_status(bfd, &status);
+
+ smap_add(smap, "forwarding", bfd_forwarding(bfd, time_msec())
+ ? "true" : "false");
+ smap_add(smap, "state", bfd_state_to_str(status.local_state));
+ smap_add(smap, "diagnostic", bfd_diag_to_str(status.local_diag));
+ smap_add_format(smap, "flap_count", "%"PRIu64, status.flap_count);
+
+ if (status.local_state != STATE_DOWN) {
+ smap_add(smap, "remote_state", bfd_state_to_str(status.rmt_state));
+ smap_add(smap, "remote_diagnostic", bfd_diag_to_str(status.rmt_diag));
+ }
+
+ ovs_mutex_unlock(&mutex);
+ return 0;
+}
+
+/* Queries the bfd.forwarding flag. */
+bool
+bfd_ts_forwarding(struct bfd *bfd)
+{
+ bool ret;
+
+ ovs_mutex_lock(&mutex);
+ ret = bfd_forwarding(bfd, time_msec());
+ ovs_mutex_unlock(&mutex);
+
+ return ret;
+}
+
+/* Called to check the BFD sessions status, decay and forwarding_if_rx. */
+void
+bfd_ts_run(struct bfd *bfd)
+{
+ ovs_mutex_lock(&mutex);
+ bfd_run(bfd, time_msec());
+ bfd_ts_check_flap(bfd);
+ ovs_mutex_unlock(&mutex);
+}
+
+/* Registers the next wakeup time with the poll loop. */
+void
+bfd_ts_wait(const struct bfd *bfd)
+{
+ poll_timer_wait_until(bfd_ts_wake_time(bfd));
+}
+
+/* Returns the next wakeup time of the BFD session. */
+long long int
+bfd_ts_wake_time(const struct bfd *bfd)
+{
+ long long int ret;
+
+ ovs_mutex_lock(&mutex);
+ ret = bfd_wait(bfd);
+ ovs_mutex_unlock(&mutex);
+
+ return ret;
+}
+
+/* Checks if the BFD control packet should be sent. */
+bool
+bfd_ts_should_send_packet(struct bfd *bfd)
+{
+ bool ret;
+
+ ovs_mutex_lock(&mutex);
+ ret = bfd_should_send_packet(bfd, time_msec());
+ ovs_mutex_unlock(&mutex);
+
+ return ret;
+}
+
+/* Constructs the entire BFD control packet. */
+void
+bfd_ts_put_packet(struct bfd *bfd, struct ofpbuf *p,
+ uint8_t eth_src[ETH_ADDR_LEN])
+{
+ struct udp_header *udp;
+ struct eth_header *eth;
+ struct ip_header *ip;
+ struct bfd_msg *msg;
+ struct bfd_ts *ts = CONTAINER_OF(bfd, struct bfd_ts, bfd);
+ struct bfd_status s;
+
+ ovs_mutex_lock(&mutex);
+ bfd_get_status(bfd, &s);
+ if (s.next_tx) {
+ long long int delay = time_msec() - s.next_tx;
+ if (delay > s.tx_interval * 3 / 2) {
+ VLOG_INFO("%s: long delay of %lldms (expected %ums) sending BFD"
+ " control message", ts->name, delay, s.tx_interval);
+ }
+ }
+
+ /* RFC 5880 Section 6.5
+ * A BFD Control packet MUST NOT have both the Poll (P) and Final (F) bits
+ * set. */
+ ovs_assert(!(s.local_flags & FLAG_POLL) || !(s.local_flags & FLAG_FINAL));
+
+ ofpbuf_reserve(p, 2); /* Properly align after the ethernet header. */
+ eth = ofpbuf_put_uninit(p, sizeof *eth);
+ memcpy(eth->eth_src, eth_src, ETH_ADDR_LEN);
+ memcpy(eth->eth_dst, ts->eth_dst, ETH_ADDR_LEN);
+ eth->eth_type = htons(ETH_TYPE_IP);
+
+ ip = ofpbuf_put_zeros(p, sizeof *ip);
+ ip->ip_ihl_ver = IP_IHL_VER(5, 4);
+ ip->ip_tot_len = htons(sizeof *ip + sizeof *udp + sizeof *msg);
+ ip->ip_ttl = MAXTTL;
+ ip->ip_tos = IPTOS_LOWDELAY | IPTOS_THROUGHPUT;
+ ip->ip_proto = IPPROTO_UDP;
+ put_16aligned_be32(&ip->ip_src, ts->ip_src);
+ put_16aligned_be32(&ip->ip_dst, ts->ip_dst);
+ ip->ip_csum = csum(ip, sizeof *ip);
+
+ udp = ofpbuf_put_zeros(p, sizeof *udp);
+ udp->udp_src = htons(ts->udp_src);
+ udp->udp_dst = htons(BFD_DEST_PORT);
+ udp->udp_len = htons(sizeof *udp + sizeof *msg);
+
+ bfd_put_packet(bfd, ofpbuf_put_uninit(p, sizeof *msg),
+ BFD_PACKET_LEN, time_msec());
+ ovs_mutex_unlock(&mutex);
+}
+
+/* Checks if the received packet should be processed as BFD control
+ * packet. */
+bool
+bfd_ts_should_process_packet(struct bfd *bfd, const struct flow *flow,
+ struct flow_wildcards *wc)
+{
+ struct bfd_ts *ts = CONTAINER_OF(bfd, struct bfd_ts, bfd);
+ bool ret = false;
+
+ ovs_mutex_lock(&mutex);
+ memset(&wc->masks.dl_dst, 0xff, sizeof wc->masks.dl_dst);
+ if (ts->eth_dst_set && memcmp(ts->eth_dst, flow->dl_dst, ETH_ADDR_LEN)) {
+ goto out;
+ }
+
+ memset(&wc->masks.nw_proto, 0xff, sizeof wc->masks.nw_proto);
+ memset(&wc->masks.tp_dst, 0xff, sizeof wc->masks.tp_dst);
+
+ if (ts->check_tnl_key) {
+ memset(&wc->masks.tunnel.tun_id, 0xff, sizeof wc->masks.tunnel.tun_id);
+ }
+
+ ret = (!ts->check_tnl_key || flow->tunnel.tun_id == htonll(0))
+ && bfd_should_process_packet(flow->dl_type, flow->nw_proto,
+ flow->tp_dst);
+out:
+ ovs_mutex_unlock(&mutex);
+
+ return ret;
+}
+
+
+/* Processes the BFD control packet. Updates the corresponding BFD status. */
+int
+bfd_ts_process_packet(struct bfd *bfd, const struct flow *flow,
+ const struct ofpbuf *p)
+{
+ struct bfd_ts *ts = CONTAINER_OF(bfd, struct bfd_ts, bfd);
+ void *msg;
+
+ ovs_mutex_lock(&mutex);
+ if (flow->nw_ttl != 255) {
+ /* XXX Should drop in the kernel to prevent DOS. */
+ goto out;
+ }
+
+ msg = ofpbuf_at(p, (uint8_t *)p->l7 - (uint8_t *)p->data, BFD_PACKET_LEN);
+ if (!msg) {
+ VLOG_INFO_RL(&rl, "%s: Received too-short BFD control message (only "
+ "%"PRIdPTR" bytes long, at least %d required).",
+ ts->name, (uint8_t *) ofpbuf_tail(p) - (uint8_t *) p->l7,
+ BFD_PACKET_LEN);
+ goto out;
+ }
+ bfd_process_packet(bfd, msg, BFD_PACKET_LEN, time_msec());
+
+ bfd_ts_check_flap(bfd);
+out:
+ ovs_mutex_unlock(&mutex);
+
+ return 0;
+}
diff --git a/lib/bfd_ts.h b/lib/bfd_ts.h
new file mode 100644
index 0000000..86f62e4
--- /dev/null
+++ b/lib/bfd_ts.h
@@ -0,0 +1,48 @@
+/* Copyright (c) 2014 Nicira, Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License. */
+
+#ifndef BFD_TS_H
+#define BFD_TS_H 1
+
+#include <stdbool.h>
+#include <inttypes.h>
+
+#include "packets.h"
+
+/* A thread-safe wrapper for bfd.{c,h}. */
+struct bfd;
+struct flow;
+struct ofpbuf;
+struct smap;
+
+struct bfd * bfd_ts_configure(struct bfd *, const char *,
+ const struct smap *);
+void bfd_ts_account_rx(struct bfd *, uint32_t n_pkt);
+struct bfd * bfd_ts_ref(const struct bfd *);
+void bfd_ts_unref(struct bfd *);
+int bfd_ts_get_status(struct bfd *, struct smap *);
+bool bfd_ts_forwarding(struct bfd *);
+void bfd_ts_run(struct bfd *);
+void bfd_ts_wait(const struct bfd *);
+long long int bfd_ts_wake_time(const struct bfd *);
+
+bool bfd_ts_should_send_packet(struct bfd *);
+void bfd_ts_put_packet(struct bfd *, struct ofpbuf *,
+ uint8_t eth_src[ETH_ADDR_LEN]);
+bool bfd_ts_should_process_packet(struct bfd *, const struct flow *,
+ struct flow_wildcards *);
+int bfd_ts_process_packet(struct bfd *, const struct flow *,
+ const struct ofpbuf *);
+
+#endif /* bfd_thread-safe.h */
diff --git a/ofproto/ofproto-dpif-monitor.c b/ofproto/ofproto-dpif-monitor.c
index b521735..aa20547 100644
--- a/ofproto/ofproto-dpif-monitor.c
+++ b/ofproto/ofproto-dpif-monitor.c
@@ -19,7 +19,7 @@
#include <string.h>
-#include "bfd.h"
+#include "bfd_ts.h"
#include "cfm.h"
#include "hash.h"
#include "heap.h"
@@ -140,8 +140,8 @@ mport_update(struct mport *mport, struct bfd *bfd, struct cfm *cfm,
mport->cfm = cfm_ref(cfm);
}
if (mport->bfd != bfd) {
- bfd_unref(mport->bfd);
- mport->bfd = bfd_ref(bfd);
+ bfd_ts_unref(mport->bfd);
+ mport->bfd = bfd_ts_ref(bfd);
}
if (hw_addr && memcmp(mport->hw_addr, hw_addr, ETH_ADDR_LEN)) {
memcpy(mport->hw_addr, hw_addr, ETH_ADDR_LEN);
@@ -198,9 +198,9 @@ monitor_run(void)
cfm_compose_ccm(mport->cfm, &packet, mport->hw_addr);
ofproto_dpif_send_packet(mport->ofport, &packet);
}
- if (mport->bfd && bfd_should_send_packet(mport->bfd)) {
+ if (mport->bfd && bfd_ts_should_send_packet(mport->bfd)) {
ofpbuf_clear(&packet);
- bfd_put_packet(mport->bfd, &packet, mport->hw_addr);
+ bfd_ts_put_packet(mport->bfd, &packet, mport->hw_addr);
ofproto_dpif_send_packet(mport->ofport, &packet);
}
if (mport->cfm) {
@@ -208,11 +208,11 @@ monitor_run(void)
cfm_wait(mport->cfm);
}
if (mport->bfd) {
- bfd_run(mport->bfd);
- bfd_wait(mport->bfd);
+ bfd_ts_run(mport->bfd);
+ bfd_ts_wait(mport->bfd);
}
/* Computes the next wakeup time for this mport. */
- next_wake_time = MIN(bfd_wake_time(mport->bfd),
+ next_wake_time = MIN(bfd_ts_wake_time(mport->bfd),
cfm_wake_time(mport->cfm));
heap_change(&monitor_heap, &mport->heap_node,
MSEC_TO_PRIO(next_wake_time));
diff --git a/ofproto/ofproto-dpif-xlate.c b/ofproto/ofproto-dpif-xlate.c
index eb4931e..edc4dd1 100644
--- a/ofproto/ofproto-dpif-xlate.c
+++ b/ofproto/ofproto-dpif-xlate.c
@@ -18,7 +18,7 @@
#include <errno.h>
-#include "bfd.h"
+#include "bfd_ts.h"
#include "bitmap.h"
#include "bond.h"
#include "bundle.h"
@@ -451,8 +451,8 @@ xlate_ofport_set(struct ofproto_dpif *ofproto, struct ofbundle *ofbundle,
}
if (xport->bfd != bfd) {
- bfd_unref(xport->bfd);
- xport->bfd = bfd_ref(bfd);
+ bfd_ts_unref(xport->bfd);
+ xport->bfd = bfd_ts_ref(bfd);
}
if (xport->peer) {
@@ -515,7 +515,7 @@ xlate_ofport_remove(struct ofport_dpif *ofport)
netdev_close(xport->netdev);
cfm_unref(xport->cfm);
- bfd_unref(xport->bfd);
+ bfd_ts_unref(xport->bfd);
free(xport);
}
@@ -1641,11 +1641,13 @@ process_special(struct xlate_ctx *ctx, const struct flow *flow,
cfm_process_heartbeat(xport->cfm, packet);
}
return SLOW_CFM;
- } else if (xport->bfd && bfd_should_process_flow(xport->bfd, flow, wc)) {
+ } else if (xport->bfd
+ && bfd_ts_should_process_packet(xport->bfd, flow, wc)) {
if (packet) {
- bfd_process_packet(xport->bfd, flow, packet);
+ bfd_ts_process_packet(xport->bfd, flow, packet);
+
/* If POLL received, immediately sends FINAL back. */
- if (bfd_should_send_packet(xport->bfd)) {
+ if (bfd_ts_should_send_packet(xport->bfd)) {
if (xport->peer) {
ofproto_dpif_monitor_port_send_soon(xport->ofport);
} else {
@@ -1741,7 +1743,8 @@ compose_output_action__(struct xlate_ctx *ctx, ofp_port_t ofp_port,
netdev_vport_inc_tx(xport->netdev, ctx->xin->resubmit_stats);
netdev_vport_inc_rx(peer->netdev, ctx->xin->resubmit_stats);
if (peer->bfd) {
- bfd_account_rx(peer->bfd, ctx->xin->resubmit_stats);
+ bfd_ts_account_rx(peer->bfd,
+ ctx->xin->resubmit_stats->n_packets);
}
}
@@ -3066,7 +3069,8 @@ xlate_actions__(struct xlate_in *xin, struct xlate_out *xout)
if (in_port && in_port->is_tunnel && ctx.xin->resubmit_stats) {
netdev_vport_inc_rx(in_port->netdev, ctx.xin->resubmit_stats);
if (in_port->bfd) {
- bfd_account_rx(in_port->bfd, ctx.xin->resubmit_stats);
+ bfd_ts_account_rx(in_port->bfd,
+ ctx.xin->resubmit_stats->n_packets);
}
}
diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
index 8c43ee9..60d50df 100644
--- a/ofproto/ofproto-dpif.c
+++ b/ofproto/ofproto-dpif.c
@@ -21,7 +21,7 @@
#include <errno.h>
-#include "bfd.h"
+#include "bfd_ts.h"
#include "bond.h"
#include "bundle.h"
#include "byte-order.h"
@@ -77,6 +77,7 @@ enum { N_TABLES = 255 };
enum { TBL_INTERNAL = N_TABLES - 1 }; /* Used for internal hidden rules. */
BUILD_ASSERT_DECL(N_TABLES >= 2 && N_TABLES <= 255);
+struct bfd;
struct flow_miss;
struct rule_dpif {
@@ -1573,9 +1574,6 @@ port_modified(struct ofport *port_)
cfm_set_netdev(port->cfm, port->up.netdev);
}
- if (port->bfd) {
- bfd_set_netdev(port->bfd, port->up.netdev);
- }
ofproto_dpif_monitor_port_update(port, port->bfd, port->cfm,
port->up.pp.hw_addr);
@@ -1722,8 +1720,8 @@ set_bfd(struct ofport *ofport_, const struct smap *cfg)
struct bfd *old;
old = ofport->bfd;
- ofport->bfd = bfd_configure(old, netdev_get_name(ofport->up.netdev),
- cfg, ofport->up.netdev);
+ ofport->bfd = bfd_ts_configure(old, netdev_get_name(ofport->up.netdev),
+ cfg);
if (ofport->bfd != old) {
ofproto->backer->need_revalidate = REV_RECONFIGURE;
}
@@ -1737,12 +1735,7 @@ get_bfd_status(struct ofport *ofport_, struct smap *smap)
{
struct ofport_dpif *ofport = ofport_dpif_cast(ofport_);
- if (ofport->bfd) {
- bfd_get_status(ofport->bfd, smap);
- return 0;
- } else {
- return ENOENT;
- }
+ return bfd_ts_get_status(ofport->bfd, smap);
}
/* Spanning Tree. */
@@ -2648,7 +2641,7 @@ port_run(struct ofport_dpif *ofport)
}
if (ofport->bfd) {
- bfd_enable = bfd_forwarding(ofport->bfd);
+ bfd_enable = bfd_ts_forwarding(ofport->bfd);
}
if (ofport->bfd || ofport->cfm) {
--
1.7.9.5
More information about the dev
mailing list