core-lightning/lightningd/peer_control.c
Rusty Russell b3a2a0623c lightningd: set up reconnect timer if we don't want to forget peer.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2017-05-25 14:24:47 +09:30

1973 lines
56 KiB
C

#include "lightningd.h"
#include "peer_control.h"
#include "subd.h"
#include <bitcoin/script.h>
#include <bitcoin/tx.h>
#include <ccan/crypto/ripemd160/ripemd160.h>
#include <ccan/io/io.h>
#include <ccan/noerr/noerr.h>
#include <ccan/take/take.h>
#include <ccan/tal/str/str.h>
#include <daemon/chaintopology.h>
#include <daemon/dns.h>
#include <daemon/invoice.h>
#include <daemon/jsonrpc.h>
#include <daemon/log.h>
#include <daemon/timeout.h>
#include <errno.h>
#include <inttypes.h>
#include <lightningd/build_utxos.h>
#include <lightningd/channel.h>
#include <lightningd/channel/gen_channel_wire.h>
#include <lightningd/funding_tx.h>
#include <lightningd/gen_peer_state_names.h>
#include <lightningd/gossip/gen_gossip_wire.h>
#include <lightningd/hsm/gen_hsm_wire.h>
#include <lightningd/key_derive.h>
#include <lightningd/new_connection.h>
#include <lightningd/opening/gen_opening_wire.h>
#include <lightningd/pay.h>
#include <lightningd/sphinx.h>
#include <netinet/in.h>
#include <overflows.h>
#include <sys/socket.h>
#include <sys/types.h>
#include <wally_bip32.h>
#include <wire/gen_onion_wire.h>
#include <wire/gen_peer_wire.h>
struct channel_info {
secp256k1_ecdsa_signature commit_sig;
struct channel_config their_config;
struct pubkey remote_fundingkey;
struct basepoints theirbase;
struct pubkey their_per_commit_point;
};
static void destroy_peer(struct peer *peer)
{
list_del_from(&peer->ld->peers, &peer->list);
if (peer->fd >= 0)
close(peer->fd);
}
static struct peer *peer_by_pubkey(struct lightningd *ld, const struct pubkey *id)
{
struct peer *peer;
list_for_each(&ld->peers, peer, list) {
if (pubkey_cmp(id, peer->id) == 0)
return peer;
}
return NULL;
}
/* Mutual recursion, sets timer. */
static void peer_reconnect(struct peer *peer);
static void reconnect_failed(struct lightningd_state *dstate,
struct connection *c)
{
/* Figure out what peer, set reconnect timer. */
struct lightningd *ld = ld_from_dstate(dstate);
struct peer *peer = peer_by_pubkey(ld, connection_known_id(c));
tal_free(c);
peer_reconnect(peer);
}
static void try_reconnect(struct peer *peer)
{
struct connection *c;
struct netaddr *addrs;
/* We may already be reconnected (another incoming connection) */
if (peer->fd != -1) {
log_debug(peer->log, "try_reconnect: already reconnected");
return;
}
c = new_connection(peer, peer->ld, NULL, peer->id);
/* FIXME: Combine known address with gossip addresses and possibly
* DNS seed addresses. */
addrs = tal_dup_arr(c, struct netaddr, &peer->netaddr, 1, 0);
multiaddress_connect(&peer->ld->dstate, addrs,
connection_out, reconnect_failed, c);
}
static void peer_reconnect(struct peer *peer)
{
new_reltimer(&peer->ld->dstate.timers,
peer, peer->ld->dstate.config.poll_time,
try_reconnect, peer);
}
void peer_fail(struct peer *peer, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
log_info(peer->log, "Peer failure in %s: ",
peer_state_name(peer->state));
logv(peer->log, -1, fmt, ap);
va_end(ap);
/* If we haven't reached awaiting locked, we don't need to reconnect */
if (!peer_persists(peer)) {
log_info(peer->log, "Only reached state %s: forgetting",
peer_state_name(peer->state));
goto dont_talk;
}
/* Reconnect unless we've dropped to chain. */
if (!peer_on_chain(peer)) {
peer_reconnect(peer);
return;
}
dont_talk:
/* In case we reconnected in the meantime. */
if (peer->fd != -1) {
/* FIXME: We should retransmit error if this happens. */
close(peer->fd);
}
tal_free(peer);
return;
}
void peer_set_condition(struct peer *peer, enum peer_state old_state,
enum peer_state state)
{
log_info(peer->log, "state: %s -> %s",
peer_state_name(peer->state), peer_state_name(state));
if (peer->state != old_state)
fatal("peer state %s should be %s",
peer_state_name(peer->state), peer_state_name(old_state));
peer->state = state;
}
static bool peer_start_channeld_hsmfd(struct subd *hsm, const u8 *resp,
const int *fds,
struct peer *peer);
/* Returns true if we consider this a reconnection. */
static bool peer_reconnected(struct lightningd *ld,
const struct pubkey *id,
int fd,
const struct crypto_state *cs)
{
struct peer *peer = peer_by_id(ld, id);
if (!peer)
return false;
log_info(peer->log, "Peer has reconnected, state %s",
peer_state_name(peer->state));
/* Always copy cryptostate; at worst we'll throw it away. */
tal_free(peer->cs);
peer->cs = tal_dup(peer, struct crypto_state, cs);
switch (peer->state) {
/* This can't happen. */
case UNINITIALIZED:
abort();
case GOSSIPD:
/* Tell gossipd to kick that one out, will call peer_fail */
subd_send_msg(peer->ld->gossip,
take(towire_gossipctl_fail_peer(peer,
peer->unique_id)));
tal_free(peer);
/* Continue with a new peer. */
return false;
case OPENINGD:
/* Kill off openingd, forget old peer. */
peer->owner->peer = NULL;
tal_free(peer->owner);
tal_free(peer);
/* A fresh start. */
return false;
case GETTING_SIG_FROM_HSM:
/* BOLT #2:
*
* On disconnection, the funder MUST remember the channel for
* reconnection if it has broadcast the funding transaction,
* otherwise it MUST NOT.
*/
/* Free peer, which will discard HSM response. */
tal_free(peer);
/* Start afresh */
return false;
case GETTING_HSMFD:
/* Simply substitute old fd for new one. */
assert(peer->fd != -1);
close(peer->fd);
peer->fd = fd;
return true;
case CHANNELD_AWAITING_LOCKIN:
/* Kill off current channeld, if any. */
if (peer->owner) {
peer->owner->peer = NULL;
peer->owner = tal_free(peer->owner);
}
assert(peer->fd == -1);
peer->fd = fd;
/* Start a new one: first get fresh hsm fd. */
peer_set_condition(peer, CHANNELD_AWAITING_LOCKIN,
GETTING_HSMFD);
/* Get fd from hsm. */
subd_req(peer, peer->ld->hsm,
take(towire_hsmctl_hsmfd_channeld(peer,
peer->unique_id)),
-1, 1, peer_start_channeld_hsmfd, peer);
return true;
case CHANNELD_NORMAL:
/* Kill off current channeld, if any */
if (peer->owner) {
peer->owner->peer = NULL;
peer->owner = tal_free(peer->owner);
}
assert(peer->fd == -1);
peer->fd = fd;
/* Start a new one: first get fresh hsm fd. */
peer_set_condition(peer, CHANNELD_NORMAL, GETTING_HSMFD);
/* FIXME: Needs to reload state! */
/* Get fd from hsm. */
subd_req(peer, peer->ld->hsm,
take(towire_hsmctl_hsmfd_channeld(peer,
peer->unique_id)),
-1, 1, peer_start_channeld_hsmfd, peer);
return true;
case SHUTDOWND_SENT:
case SHUTDOWND_RCVD:
case CLOSINGD_SIGEXCHANGE:
case ONCHAIND_CHEATED:
case ONCHAIND_THEIR_UNILATERAL:
case ONCHAIND_OUR_UNILATERAL:
case ONCHAIND_MUTUAL:
; /* FIXME: Implement! */
}
abort();
}
void add_peer(struct lightningd *ld, u64 unique_id,
int fd, const struct pubkey *id,
const struct crypto_state *cs)
{
struct peer *peer;
const char *netname, *idname;
u8 *msg;
/* It's a reconnect? */
if (peer_reconnected(ld, id, fd, cs))
return;
/* Fresh peer. */
peer = tal(ld, struct peer);
peer->ld = ld;
peer->unique_id = unique_id;
peer->owner = NULL;
peer->scid = NULL;
peer->id = tal_dup(peer, struct pubkey, id);
peer->fd = fd;
peer->cs = tal_dup(peer, struct crypto_state, cs);
peer->funding_txid = NULL;
peer->seed = NULL;
peer->balance = NULL;
peer->state = UNINITIALIZED;
peer->channel_info = NULL;
idname = type_to_string(peer, struct pubkey, id);
/* Max 128k per peer. */
peer->log_book = new_log_book(peer, 128*1024,
get_log_level(ld->dstate.log_book));
peer->log = new_log(peer, peer->log_book, "peer %s:", idname);
/* FIXME: Don't assume protocol here! */
if (!netaddr_from_fd(peer->fd, SOCK_STREAM, IPPROTO_TCP,
&peer->netaddr)) {
log_unusual(ld->log, "Failed to get netaddr for outgoing: %s",
strerror(errno));
tal_free(peer);
return;
}
netname = netaddr_name(idname, &peer->netaddr);
log_info(peer->log, "Connected from %s", netname);
tal_free(idname);
list_add_tail(&ld->peers, &peer->list);
tal_add_destructor(peer, destroy_peer);
/* Let gossip handle it from here. */
peer->owner = peer->ld->gossip;
tal_steal(peer->owner, peer);
peer_set_condition(peer, UNINITIALIZED, GOSSIPD);
msg = towire_gossipctl_new_peer(peer, peer->unique_id, peer->cs);
subd_send_msg(peer->ld->gossip, take(msg));
subd_send_fd(peer->ld->gossip, peer->fd);
/* Peer struct longer owns fd. */
peer->fd = -1;
peer->cs = tal_free(peer->cs);
}
struct peer *peer_by_unique_id(struct lightningd *ld, u64 unique_id)
{
struct peer *p;
list_for_each(&ld->peers, p, list)
if (p->unique_id == unique_id)
return p;
return NULL;
}
struct peer *peer_by_id(struct lightningd *ld, const struct pubkey *id)
{
struct peer *p;
list_for_each(&ld->peers, p, list)
if (pubkey_eq(p->id, id))
return p;
return NULL;
}
/* When a per-peer subdaemon exits, see if we need to do anything. */
static void peer_owner_finished(struct subd *subd, int status)
{
/* If peer has moved on, do nothing. */
if (subd->peer->owner != subd) {
log_debug(subd->ld->log, "Subdaemon %s died (%i), peer moved",
subd->name, status);
return;
}
subd->peer->owner = NULL;
peer_fail(subd->peer, "Owning subdaemon %s died (%i)",
subd->name, status);
}
static int make_listen_fd(struct lightningd *ld,
int domain, void *addr, socklen_t len)
{
int fd = socket(domain, SOCK_STREAM, 0);
if (fd < 0) {
log_debug(ld->log, "Failed to create %u socket: %s",
domain, strerror(errno));
return -1;
}
if (addr) {
int on = 1;
/* Re-use, please.. */
if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &on, sizeof(on)))
log_unusual(ld->log, "Failed setting socket reuse: %s",
strerror(errno));
if (bind(fd, addr, len) != 0) {
log_unusual(ld->log, "Failed to bind on %u socket: %s",
domain, strerror(errno));
goto fail;
}
}
if (listen(fd, 5) != 0) {
log_unusual(ld->log, "Failed to listen on %u socket: %s",
domain, strerror(errno));
goto fail;
}
return fd;
fail:
close_noerr(fd);
return -1;
}
void setup_listeners(struct lightningd *ld)
{
struct sockaddr_in addr;
struct sockaddr_in6 addr6;
socklen_t len;
int fd1, fd2;
if (!ld->dstate.portnum) {
log_debug(ld->log, "Zero portnum, not listening for incoming");
return;
}
memset(&addr, 0, sizeof(addr));
addr.sin_family = AF_INET;
addr.sin_addr.s_addr = INADDR_ANY;
addr.sin_port = htons(ld->dstate.portnum);
memset(&addr6, 0, sizeof(addr6));
addr6.sin6_family = AF_INET6;
addr6.sin6_addr = in6addr_any;
addr6.sin6_port = htons(ld->dstate.portnum);
/* IPv6, since on Linux that (usually) binds to IPv4 too. */
fd1 = make_listen_fd(ld, AF_INET6, &addr6, sizeof(addr6));
if (fd1 >= 0) {
struct sockaddr_in6 in6;
len = sizeof(in6);
if (getsockname(fd1, (void *)&in6, &len) != 0) {
log_unusual(ld->log, "Failed get IPv6 sockname: %s",
strerror(errno));
close_noerr(fd1);
fd1 = -1;
} else {
addr.sin_port = in6.sin6_port;
assert(ld->dstate.portnum == ntohs(addr.sin_port));
log_debug(ld->log, "Creating IPv6 listener on port %u",
ld->dstate.portnum);
io_new_listener(ld, fd1, connection_in, ld);
}
}
/* Just in case, aim for the same port... */
fd2 = make_listen_fd(ld, AF_INET, &addr, sizeof(addr));
if (fd2 >= 0) {
len = sizeof(addr);
if (getsockname(fd2, (void *)&addr, &len) != 0) {
log_unusual(ld->log, "Failed get IPv4 sockname: %s",
strerror(errno));
close_noerr(fd2);
fd2 = -1;
} else {
assert(ld->dstate.portnum == ntohs(addr.sin_port));
log_debug(ld->log, "Creating IPv4 listener on port %u",
ld->dstate.portnum);
io_new_listener(ld, fd2, connection_in, ld);
}
}
if (fd1 < 0 && fd2 < 0)
fatal("Could not bind to a network address on port %u",
ld->dstate.portnum);
}
static void connect_failed(struct lightningd_state *dstate,
struct connection *c)
{
tal_free(c);
}
static void json_connect(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
struct lightningd *ld = ld_from_dstate(cmd->dstate);
struct connection *c;
jsmntok_t *host, *porttok, *idtok;
const tal_t *tmpctx = tal_tmpctx(cmd);
struct pubkey id;
char *name, *port;
if (!json_get_params(buffer, params,
"host", &host,
"port", &porttok,
"id", &idtok,
NULL)) {
command_fail(cmd, "Need host, port and id to connect");
return;
}
if (!pubkey_from_hexstr(buffer + idtok->start,
idtok->end - idtok->start, &id)) {
command_fail(cmd, "id %.*s not valid",
idtok->end - idtok->start,
buffer + idtok->start);
return;
}
c = new_connection(cmd, ld, cmd, &id);
name = tal_strndup(tmpctx,
buffer + host->start, host->end - host->start);
port = tal_strndup(tmpctx,
buffer + porttok->start,
porttok->end - porttok->start);
if (!dns_resolve_and_connect(cmd->dstate, name, port,
connection_out, connect_failed, c)) {
command_fail(cmd, "DNS failed");
return;
}
tal_free(tmpctx);
}
static const struct json_command connect_command = {
"connect",
json_connect,
"Connect to a {host} at {port} expecting node {id}",
"Returns the {id} on success (once channel established)"
};
AUTODATA(json_command, &connect_command);
struct log_info {
enum log_level level;
struct json_result *response;
};
/* FIXME: Share this with jsonrpc.c's code! */
static void log_to_json(unsigned int skipped,
struct timerel diff,
enum log_level level,
const char *prefix,
const char *log,
struct log_info *info)
{
if (level < info->level)
return;
if (level != LOG_IO)
json_add_string(info->response, NULL, log);
}
static void json_getpeers(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
struct lightningd *ld = ld_from_dstate(cmd->dstate);
struct peer *p;
struct json_result *response = new_json_result(cmd);
jsmntok_t *leveltok;
struct log_info info;
json_get_params(buffer, params, "?level", &leveltok, NULL);
if (!leveltok)
;
else if (json_tok_streq(buffer, leveltok, "debug"))
info.level = LOG_DBG;
else if (json_tok_streq(buffer, leveltok, "info"))
info.level = LOG_INFORM;
else if (json_tok_streq(buffer, leveltok, "unusual"))
info.level = LOG_UNUSUAL;
else if (json_tok_streq(buffer, leveltok, "broken"))
info.level = LOG_BROKEN;
else {
command_fail(cmd, "Invalid level param");
return;
}
json_object_start(response, NULL);
json_array_start(response, "peers");
list_for_each(&ld->peers, p, list) {
json_object_start(response, NULL);
json_add_u64(response, "unique_id", p->unique_id);
json_add_string(response, "state", peer_state_name(p->state));
json_add_string(response, "netaddr",
netaddr_name(response, &p->netaddr));
if (p->id)
json_add_pubkey(response, "peerid", p->id);
if (p->owner)
json_add_string(response, "owner", p->owner->name);
if (p->scid)
json_add_short_channel_id(response, "channel", p->scid);
if (p->balance) {
json_add_u64(response, "msatoshi_to_us",
p->balance[LOCAL]);
json_add_u64(response, "msatoshi_to_them",
p->balance[REMOTE]);
}
if (leveltok) {
info.response = response;
json_array_start(response, "log");
log_each_line(p->log_book, log_to_json, &info);
json_array_end(response);
}
json_object_end(response);
}
json_array_end(response);
json_object_end(response);
command_success(cmd, response);
}
static const struct json_command getpeers_command = {
"getpeers",
json_getpeers,
"List the current peers, if {level} is set, include {log}s",
"Returns a 'peers' array"
};
AUTODATA(json_command, &getpeers_command);
struct peer *peer_from_json(struct lightningd *ld,
const char *buffer,
jsmntok_t *peeridtok)
{
struct pubkey peerid;
if (!pubkey_from_hexstr(buffer + peeridtok->start,
peeridtok->end - peeridtok->start, &peerid))
return NULL;
return peer_by_id(ld, &peerid);
}
struct funding_channel {
struct peer *peer;
struct command *cmd;
/* Details we sent to openingd to create funding. */
const struct utxo **utxomap;
u64 change;
u32 change_keyindex;
/* Funding tx once we're ready to sign and send. */
struct bitcoin_tx *funding_tx;
};
static void fail_fundchannel_command(struct funding_channel *fc)
{
/* FIXME: More details? */
command_fail(fc->cmd, "Peer died");
}
static void funding_broadcast_failed(struct peer *peer,
int exitstatus, const char *err)
{
log_unusual(peer->log, "Funding broadcast exited with %i: %s",
exitstatus, err);
/* FIXME: send PKT_ERR to peer if this happens. */
tal_free(peer);
}
static enum watch_result funding_announce_cb(struct peer *peer,
unsigned int depth,
const struct sha256_double *txid,
void *unused)
{
if (depth < ANNOUNCE_MIN_DEPTH) {
return KEEP_WATCHING;
}
if (peer->state != CHANNELD_NORMAL || !peer->owner) {
return KEEP_WATCHING;
}
subd_send_msg(peer->owner,
take(towire_channel_funding_announce_depth(peer)));
return DELETE_WATCH;
}
static enum watch_result funding_lockin_cb(struct peer *peer,
unsigned int depth,
const struct sha256_double *txid,
void *unused)
{
const char *txidstr = type_to_string(peer, struct sha256_double, txid);
struct txlocator *loc;
log_debug(peer->log, "Funding tx %s depth %u of %u",
txidstr, depth, peer->minimum_depth);
tal_free(txidstr);
if (depth < peer->minimum_depth)
return KEEP_WATCHING;
/* In theory, it could have been buried before we got back
* from accepting openingd or disconnected: just wait for next one. */
if (!peer->owner || peer->state != CHANNELD_AWAITING_LOCKIN) {
log_unusual(peer->log,
"Funding tx confirmed, but peer state %s %s",
peer_state_name(peer->state),
peer->owner ? peer->owner->name : "unowned");
return KEEP_WATCHING;
}
loc = locate_tx(peer, peer->ld->topology, txid);
peer->scid = tal(peer, struct short_channel_id);
peer->scid->blocknum = loc->blkheight;
peer->scid->txnum = loc->index;
peer->scid->outnum = peer->funding_outnum;
tal_free(loc);
subd_send_msg(peer->owner,
take(towire_channel_funding_locked(peer, peer->scid)));
/* BOLT #7:
*
* If sent, `announcement_signatures` messages MUST NOT be sent until
* `funding_locked` has been sent, and the funding transaction is has
* at least 6 confirmations.
*/
if (depth >= ANNOUNCE_MIN_DEPTH) {
subd_send_msg(peer->owner,
take(towire_channel_funding_announce_depth(peer)));
} else {
watch_txid(peer, peer->ld->topology, peer, txid,
funding_announce_cb, NULL);
}
return DELETE_WATCH;
}
/* FIXME: Reshuffle. */
static void peer_start_channeld(struct peer *peer, enum peer_state oldstate);
static bool opening_got_hsm_funding_sig(struct subd *hsm, const u8 *resp,
const int *fds,
struct funding_channel *fc)
{
secp256k1_ecdsa_signature *sigs;
struct bitcoin_tx *tx = fc->funding_tx;
size_t i;
if (!fromwire_hsmctl_sign_funding_reply(fc, resp, NULL, &sigs))
fatal("HSM gave bad sign_funding_reply %s",
tal_hex(fc, resp));
if (tal_count(sigs) != tal_count(tx->input))
fatal("HSM gave %zu sigs, needed %zu",
tal_count(sigs), tal_count(tx->input));
/* Create input parts from signatures. */
for (i = 0; i < tal_count(tx->input); i++) {
struct pubkey key;
if (!bip32_pubkey(fc->peer->ld->bip32_base,
&key, fc->utxomap[i]->keyindex))
fatal("Cannot generate BIP32 key for UTXO %u",
fc->utxomap[i]->keyindex);
/* P2SH inputs have same witness. */
tx->input[i].witness
= bitcoin_witness_p2wpkh(tx, &sigs[i], &key);
}
/* Send it out and watch for confirms. */
broadcast_tx(hsm->ld->topology, fc->peer, tx, funding_broadcast_failed);
watch_tx(fc->peer, fc->peer->ld->topology, fc->peer, tx,
funding_lockin_cb, NULL);
/* We could defer until after funding locked, but makes testing
* harder. */
tal_del_destructor(fc, fail_fundchannel_command);
command_success(fc->cmd, null_response(fc->cmd));
/* Start normal channel daemon. */
peer_start_channeld(fc->peer, GETTING_SIG_FROM_HSM);
tal_free(fc);
return true;
}
struct decoding_htlc {
struct peer *peer;
u64 id;
u32 amount_msat;
u32 cltv_expiry;
struct sha256 payment_hash;
u8 onion[TOTAL_PACKET_SIZE];
struct secret shared_secret;
};
static void fail_htlc(struct peer *peer, struct htlc_end *hend, const u8 *msg)
{
u8 *reply = wrap_onionreply(hend, hend->shared_secret, msg);
subd_send_msg(peer->owner,
take(towire_channel_fail_htlc(peer, hend->htlc_id, reply)));
if (taken(msg))
tal_free(msg);
}
static void fail_local_htlc(struct peer *peer, struct htlc_end *hend, const u8 *msg)
{
u8 *reply;
enum onion_type failcode = fromwire_peektype(msg);
log_broken(peer->log, "failed htlc %"PRIu64" code 0x%04x (%s)",
hend->htlc_id, failcode, onion_type_name(failcode));
reply = create_onionreply(hend, hend->shared_secret, msg);
fail_htlc(peer, hend, reply);
}
static u8 *make_failmsg(const tal_t *ctx, const struct htlc_end *hend,
enum onion_type failcode)
{
struct sha256 *onion_sha = NULL;
u8 *channel_update = NULL;
if (failcode & BADONION) {
/* FIXME: need htlc_end->sha? */
}
if (failcode & UPDATE) {
/* FIXME: Ask gossip daemon for channel_update. */
}
switch (failcode) {
case WIRE_INVALID_REALM:
return towire_invalid_realm(ctx);
case WIRE_TEMPORARY_NODE_FAILURE:
return towire_temporary_node_failure(ctx);
case WIRE_PERMANENT_NODE_FAILURE:
return towire_permanent_node_failure(ctx);
case WIRE_REQUIRED_NODE_FEATURE_MISSING:
return towire_required_node_feature_missing(ctx);
case WIRE_INVALID_ONION_VERSION:
return towire_invalid_onion_version(ctx, onion_sha);
case WIRE_INVALID_ONION_HMAC:
return towire_invalid_onion_hmac(ctx, onion_sha);
case WIRE_INVALID_ONION_KEY:
return towire_invalid_onion_key(ctx, onion_sha);
case WIRE_TEMPORARY_CHANNEL_FAILURE:
return towire_temporary_channel_failure(ctx, channel_update);
case WIRE_PERMANENT_CHANNEL_FAILURE:
return towire_permanent_channel_failure(ctx);
case WIRE_REQUIRED_CHANNEL_FEATURE_MISSING:
return towire_required_channel_feature_missing(ctx);
case WIRE_UNKNOWN_NEXT_PEER:
return towire_unknown_next_peer(ctx);
case WIRE_AMOUNT_BELOW_MINIMUM:
return towire_amount_below_minimum(ctx, hend->msatoshis, channel_update);
case WIRE_FEE_INSUFFICIENT:
return towire_fee_insufficient(ctx, hend->msatoshis, channel_update);
case WIRE_INCORRECT_CLTV_EXPIRY:
/* FIXME: ctlv! */
return towire_incorrect_cltv_expiry(ctx, 0, channel_update);
case WIRE_EXPIRY_TOO_SOON:
return towire_expiry_too_soon(ctx, channel_update);
case WIRE_UNKNOWN_PAYMENT_HASH:
return towire_unknown_payment_hash(ctx);
case WIRE_INCORRECT_PAYMENT_AMOUNT:
return towire_incorrect_payment_amount(ctx);
case WIRE_FINAL_EXPIRY_TOO_SOON:
return towire_final_expiry_too_soon(ctx);
case WIRE_FINAL_INCORRECT_CLTV_EXPIRY:
/* FIXME: ctlv! */
return towire_final_incorrect_cltv_expiry(ctx, 0);
case WIRE_FINAL_INCORRECT_HTLC_AMOUNT:
return towire_final_incorrect_htlc_amount(ctx, hend->msatoshis);
}
abort();
}
/* BOLT #4:
*
* * `amt_to_forward` - The amount in milli-satoshi to forward to the next
* (outgoing) hop specified within the routing information.
*
* This value MUST factor in the computed fee for this particular hop. When
* processing an incoming Sphinx packet along with the HTLC message it's
* encapsulated within, if the following inequality doesn't hold, then the
* HTLC should be rejected as it indicates a prior node in the path has
* deviated from the specified paramters:
*
* incoming_htlc_amt - fee >= amt_to_forward
*
* Where `fee` is calculated according to the receving node's advertised fee
* schema as described in [BOLT 7](https://github.com/lightningnetwork/lightning-rfc/blob/master/07-routing-gossip.md#htlc-fees), or 0 if this node is the
* final hop.
*/
static bool check_amount(struct htlc_end *hend,
u64 amt_to_forward, u64 amt_in_htlc, u64 fee)
{
if (amt_in_htlc - fee >= amt_to_forward)
return true;
log_debug(hend->peer->ld->log, "HTLC %"PRIu64" incorrect amount:"
" %"PRIu64" in, %"PRIu64" out, fee reqd %"PRIu64,
hend->htlc_id, amt_in_htlc, amt_to_forward, fee);
return false;
}
/* BOLT #4:
*
* * `outgoing_cltv_value` - The CLTV value that the _outgoing_ HTLC carrying
* the packet should have.
*
* cltv-expiry - cltv-expiry-delta = outgoing_cltv_value
*
* Inclusion of this field allows a node to both authenticate the information
* specified by the original sender and the paramaters of the HTLC forwarded,
* and ensure the original sender is using the current `cltv-expiry-delta` value.
* If there is no next hop, `cltv-expiry-delta` is zero.
* If the values don't correspond, then the HTLC should be failed+rejected as
* this indicates the incoming node has tampered with the intended HTLC
* values, or the origin has an obsolete `cltv-expiry-delta` value.
* The node MUST be consistent in responding to an unexpected
* `outgoing_cltv_value` whether it is the final hop or not, to avoid
* leaking that information.
*/
static bool check_ctlv(struct htlc_end *hend,
u32 ctlv_expiry, u32 outgoing_cltv_value, u32 delta)
{
if (ctlv_expiry - delta == outgoing_cltv_value)
return true;
log_debug(hend->peer->ld->log, "HTLC %"PRIu64" incorrect CLTV:"
" %u in, %u out, delta reqd %u",
hend->htlc_id, ctlv_expiry, outgoing_cltv_value, delta);
return false;
}
static void fulfill_htlc(struct htlc_end *hend, const struct preimage *preimage)
{
u8 *msg;
hend->peer->balance[LOCAL] += hend->msatoshis;
hend->peer->balance[REMOTE] -= hend->msatoshis;
/* FIXME: fail the peer if it doesn't tell us that htlc fulfill is
* committed before deadline.
*/
msg = towire_channel_fulfill_htlc(hend->peer, hend->htlc_id, preimage);
subd_send_msg(hend->peer->owner, take(msg));
}
static void handle_localpay(struct htlc_end *hend,
u32 cltv_expiry,
const struct sha256 *payment_hash,
u64 amt_to_forward,
u32 outgoing_cltv_value)
{
u8 *err;
struct invoice *invoice;
/* BOLT #4:
*
* If the `amt_to_forward` does not match the `incoming_htlc_amt` of
* the HTLC at the final hop:
*
* 1. type: 19 (`final_incorrect_htlc_amount`)
* 2. data:
* * [4:incoming-htlc-amt]
*/
if (!check_amount(hend, amt_to_forward, hend->msatoshis, 0)) {
err = towire_final_incorrect_htlc_amount(hend, hend->msatoshis);
goto fail;
}
/* BOLT #4:
*
* If the `outgoing_cltv_value` does not match the `ctlv-expiry` of
* the HTLC at the final hop:
*
* 1. type: 18 (`final_incorrect_cltv_expiry`)
* 2. data:
* * [4:cltv-expiry]
*/
if (!check_ctlv(hend, cltv_expiry, outgoing_cltv_value, 0)) {
err = towire_final_incorrect_cltv_expiry(hend, cltv_expiry);
goto fail;
}
invoice = find_unpaid(hend->peer->ld->dstate.invoices, payment_hash);
if (!invoice) {
err = towire_unknown_payment_hash(hend);
goto fail;
}
/* BOLT #4:
*
* If the amount paid is less than the amount expected, the final node
* MUST fail the HTLC. If the amount paid is more than twice the
* amount expected, the final node SHOULD fail the HTLC. This allows
* the sender to reduce information leakage by altering the amount,
* without allowing accidental gross overpayment:
*
* 1. type: PERM|16 (`incorrect_payment_amount`)
*/
if (hend->msatoshis < invoice->msatoshi) {
err = towire_incorrect_payment_amount(hend);
goto fail;
} else if (hend->msatoshis > invoice->msatoshi * 2) {
err = towire_incorrect_payment_amount(hend);
goto fail;
}
/* BOLT #4:
*
* If the `cltv-expiry` is too low, the final node MUST fail the HTLC:
*/
if (get_block_height(hend->peer->ld->topology)
+ hend->peer->ld->dstate.config.deadline_blocks >= cltv_expiry) {
log_debug(hend->peer->log,
"Expiry cltv %u too close to current %u + deadline %u",
cltv_expiry,
get_block_height(hend->peer->ld->topology),
hend->peer->ld->dstate.config.deadline_blocks);
err = towire_final_expiry_too_soon(hend);
goto fail;
}
connect_htlc_end(&hend->peer->ld->htlc_ends, hend);
log_info(hend->peer->ld->log, "Resolving invoice '%s' with HTLC %"PRIu64,
invoice->label, hend->htlc_id);
fulfill_htlc(hend, &invoice->r);
resolve_invoice(&hend->peer->ld->dstate, invoice);
return;
fail:
fail_local_htlc(hend->peer, hend, take(err));
tal_free(hend);
}
/*
* A catchall in case outgoing peer disconnects before getting fwd.
*
* We could queue this and wait for it to come back, but this is simple.
*/
static void hend_subd_died(struct htlc_end *hend)
{
/* FIXME: Ask gossip daemon for channel_update. */
u8 *channel_update = NULL;
u8 *failmsg = towire_temporary_channel_failure(hend->other_end,
channel_update);
u8 *msg = towire_channel_fail_htlc(hend->other_end,
hend->other_end->htlc_id,
failmsg);
log_debug(hend->other_end->peer->owner->log,
"Failing HTLC %"PRIu64" due to peer death",
hend->other_end->htlc_id);
subd_send_msg(hend->other_end->peer->owner, take(msg));
tal_free(failmsg);
}
static bool rcvd_htlc_reply(struct subd *subd, const u8 *msg, const int *fds,
struct htlc_end *hend)
{
u16 failure_code;
u8 *failurestr;
if (!fromwire_channel_offer_htlc_reply(msg, msg, NULL,
&hend->htlc_id,
&failure_code,
&failurestr)) {
log_broken(subd->log, "Bad channel_offer_htlc_reply");
tal_free(hend);
return false;
}
if (failure_code) {
log_debug(hend->other_end->peer->owner->log,
"HTLC failed from other daemon: %s (%.*s)",
onion_type_name(failure_code),
(int)tal_len(failurestr), (char *)failurestr);
msg = make_failmsg(msg, hend->other_end, failure_code);
subd_send_msg(hend->other_end->peer->owner, take(msg));
tal_free(hend);
return true;
}
tal_del_destructor(hend, hend_subd_died);
/* Add it to lookup table. */
connect_htlc_end(&hend->peer->ld->htlc_ends, hend);
return true;
}
static void forward_htlc(struct htlc_end *hend,
u32 cltv_expiry,
const struct sha256 *payment_hash,
u64 amt_to_forward,
u32 outgoing_cltv_value,
const struct pubkey *next_hop,
const u8 next_onion[TOTAL_PACKET_SIZE])
{
u8 *err, *msg;
u64 fee;
struct lightningd *ld = hend->peer->ld;
struct peer *next = peer_by_pubkey(ld, next_hop);
if (!next) {
err = towire_unknown_next_peer(hend);
goto fail;
}
if (!peer_can_add_htlc(next)) {
log_info(next->log, "Attempt to forward HTLC but not ready");
err = towire_unknown_next_peer(hend);
goto fail;
}
/* BOLT #7:
*
* The node creating `channel_update` SHOULD accept HTLCs which pay a
* fee equal or greater than:
*
* fee-base-msat + htlc-amount-msat * fee-proportional-millionths / 1000000
*/
if (mul_overflows_u64(amt_to_forward,
ld->dstate.config.fee_per_satoshi)) {
/* FIXME: Add channel update */
err = towire_fee_insufficient(hend, hend->msatoshis, NULL);
goto fail;
}
fee = ld->dstate.config.fee_base
+ amt_to_forward * ld->dstate.config.fee_per_satoshi / 1000000;
if (!check_amount(hend, amt_to_forward, hend->msatoshis, fee)) {
/* FIXME: Add channel update */
err = towire_fee_insufficient(hend, hend->msatoshis, NULL);
goto fail;
}
if (!check_ctlv(hend, cltv_expiry, outgoing_cltv_value,
ld->dstate.config.deadline_blocks)) {
/* FIXME: Add channel update */
err = towire_incorrect_cltv_expiry(hend, cltv_expiry, NULL);
goto fail;
}
/* BOLT #4:
*
* If the ctlv-expiry is too near, we tell them the the current channel
* setting for the outgoing channel:
* 1. type: UPDATE|14 (`expiry_too_soon`)
* 2. data:
* * [2:len]
* * [len:channel_update]
*/
if (get_block_height(next->ld->topology)
+ next->ld->dstate.config.deadline_blocks >= outgoing_cltv_value) {
log_debug(hend->peer->log,
"Expiry cltv %u too close to current %u + deadline %u",
outgoing_cltv_value,
get_block_height(next->ld->topology),
next->ld->dstate.config.deadline_blocks);
/* FIXME: Add channel update */
err = towire_expiry_too_soon(hend, NULL);
goto fail;
}
/* Make sure daemon owns it, in case it fails. */
hend->other_end = tal(next->owner, struct htlc_end);
hend->other_end->which_end = HTLC_DST;
hend->other_end->peer = next;
hend->other_end->other_end = hend;
hend->other_end->pay_command = NULL;
hend->other_end->msatoshis = amt_to_forward;
tal_add_destructor(hend->other_end, hend_subd_died);
msg = towire_channel_offer_htlc(next, amt_to_forward,
outgoing_cltv_value,
payment_hash, next_onion);
subd_req(next->owner, next->owner, take(msg), -1, 0,
rcvd_htlc_reply, hend->other_end);
return;
fail:
fail_local_htlc(hend->peer, hend, take(err));
tal_free(hend);
}
/* We received a resolver reply, which gives us the node_ids of the
* channel we want to forward over */
static bool channel_resolve_reply(struct subd *gossip, const u8 *msg,
const int *fds, struct htlc_end *hend)
{
struct pubkey *nodes, *peer_id;
if (!fromwire_gossip_resolve_channel_reply(msg, msg, NULL, &nodes)) {
log_broken(gossip->log,
"bad fromwire_gossip_resolve_channel_reply %s",
tal_hex(msg, msg));
return false;
}
if (tal_count(nodes) == 0) {
fail_htlc(hend->peer, hend,
take(towire_unknown_next_peer(hend)));
tal_free(hend);
return true;
} else if (tal_count(nodes) != 2) {
log_broken(gossip->log,
"fromwire_gossip_resolve_channel_reply has %zu nodes",
tal_count(nodes));
return false;
}
/* Get the other peer matching the id that is not us */
if (pubkey_cmp(&nodes[0], &gossip->ld->dstate.id) == 0) {
peer_id = &nodes[1];
} else {
peer_id = &nodes[0];
}
forward_htlc(hend, hend->cltv_expiry, &hend->payment_hash,
hend->amt_to_forward, hend->outgoing_cltv_value, peer_id,
hend->next_onion);
/* FIXME(cdecker) Cleanup things we stuffed into hend before (maybe?) */
return true;
}
static int peer_accepted_htlc(struct peer *peer, const u8 *msg)
{
bool forward;
struct htlc_end *hend;
u8 *req;
hend = tal(msg, struct htlc_end);
hend->shared_secret = tal(hend, struct secret);
if (!fromwire_channel_accepted_htlc(msg, NULL,
&hend->htlc_id, &hend->msatoshis,
&hend->cltv_expiry, &hend->payment_hash,
hend->next_onion, &forward,
&hend->amt_to_forward,
&hend->outgoing_cltv_value,
&hend->next_channel,
hend->shared_secret)) {
log_broken(peer->log, "bad fromwire_channel_accepted_htlc %s",
tal_hex(peer, msg));
return -1;
}
tal_steal(peer, hend);
hend->which_end = HTLC_SRC;
hend->peer = peer;
hend->other_end = NULL;
hend->pay_command = NULL;
if (forward) {
req = towire_gossip_resolve_channel_request(msg, &hend->next_channel);
log_broken(peer->log, "Asking gossip to resolve channel %d/%d/%d", hend->next_channel.blocknum, hend->next_channel.txnum, hend->next_channel.outnum);
subd_req(hend, peer->ld->gossip, req, -1, 0, channel_resolve_reply, hend);
/* FIXME(cdecker) Stuff all this info into hend */
} else
handle_localpay(hend, hend->cltv_expiry, &hend->payment_hash,
hend->amt_to_forward, hend->outgoing_cltv_value);
return 0;
}
static int peer_fulfilled_htlc(struct peer *peer, const u8 *msg)
{
u64 id;
struct preimage preimage;
struct htlc_end *hend;
if (!fromwire_channel_fulfilled_htlc(msg, NULL, &id, &preimage)) {
log_broken(peer->log, "bad fromwire_channel_fulfilled_htlc %s",
tal_hex(peer, msg));
return -1;
}
hend = find_htlc_end(&peer->ld->htlc_ends, peer, id, HTLC_DST);
if (!hend) {
log_broken(peer->log,
"channel_fulfilled_htlc unknown htlc %"PRIu64,
id);
return -1;
}
/* They fulfilled our HTLC. Credit them, forward as required. */
peer->balance[REMOTE] += hend->msatoshis;
peer->balance[LOCAL] -= hend->msatoshis;
if (hend->other_end)
fulfill_htlc(hend->other_end, &preimage);
else
payment_succeeded(peer->ld, hend, &preimage);
tal_free(hend);
return 0;
}
static int peer_failed_htlc(struct peer *peer, const u8 *msg)
{
u64 id;
u8 *reason;
struct htlc_end *hend;
enum onion_type failcode;
struct onionreply *reply;
if (!fromwire_channel_failed_htlc(msg, msg, NULL, &id, &reason)) {
log_broken(peer->log, "bad fromwire_channel_failed_htlc %s",
tal_hex(peer, msg));
return -1;
}
hend = find_htlc_end(&peer->ld->htlc_ends, peer, id, HTLC_DST);
if (!hend) {
log_broken(peer->log,
"channel_failed_htlc unknown htlc %"PRIu64,
id);
return -1;
}
if (hend->other_end) {
fail_htlc(hend->other_end->peer, hend->other_end,
reason);
} else {
size_t numhops = tal_count(hend->path_secrets);
struct secret *shared_secrets = tal_arr(hend, struct secret, numhops);
for (size_t i=0; i<numhops; i++) {
shared_secrets[i] = hend->path_secrets[i];
}
reply = unwrap_onionreply(msg, shared_secrets, numhops, reason);
if (!reply) {
log_info(peer->log, "htlc %"PRIu64" failed with bad reply (%s)",
id, tal_hex(msg, msg));
failcode = WIRE_PERMANENT_NODE_FAILURE;
} else {
failcode = fromwire_peektype(reply->msg);
log_info(peer->log, "htlc %"PRIu64" failed with code 0x%04x (%s)",
id, failcode, onion_type_name(failcode));
}
payment_failed(peer->ld, hend, NULL, failcode);
}
tal_free(hend);
return 0;
}
/* FIXME: Encrypt! */
static u8 *malformed_msg(const tal_t *ctx, enum onion_type type,
const struct sha256 *sha256_of_onion)
{
u8 *channel_update;
/* FIXME: check the reported SHA matches what we sent! */
switch (type) {
case WIRE_INVALID_ONION_VERSION:
return towire_invalid_onion_version(ctx, sha256_of_onion);
case WIRE_INVALID_ONION_HMAC:
return towire_invalid_onion_hmac(ctx, sha256_of_onion);
case WIRE_INVALID_ONION_KEY:
return towire_invalid_onion_key(ctx, sha256_of_onion);
default:
/* FIXME: Ask gossip daemon for channel_update. */
channel_update = NULL;
return towire_temporary_channel_failure(ctx, channel_update);
}
}
static int peer_failed_malformed_htlc(struct peer *peer, const u8 *msg)
{
u64 id;
struct htlc_end *hend;
struct sha256 sha256_of_onion;
u16 failcode;
if (!fromwire_channel_malformed_htlc(msg, NULL, &id,
&sha256_of_onion, &failcode)) {
log_broken(peer->log, "bad fromwire_channel_malformed_htlc %s",
tal_hex(peer, msg));
return -1;
}
hend = find_htlc_end(&peer->ld->htlc_ends, peer, id, HTLC_DST);
if (!hend) {
log_broken(peer->log,
"channel_malformed_htlc unknown htlc %"PRIu64,
id);
return -1;
}
if (hend->other_end) {
/* Not really a local failure, but since the failing
* peer could not derive its shared secret it cannot
* create a valid HMAC, so we do it on his behalf */
fail_local_htlc(hend->other_end->peer, hend->other_end,
malformed_msg(msg, failcode, &sha256_of_onion));
} else {
payment_failed(peer->ld, hend, NULL, failcode);
}
tal_free(hend);
return 0;
}
/* Create a node_announcement with the given signature. It may be NULL
* in the case we need to create a provisional announcement for the
* HSM to sign. */
static u8 *create_node_announcement(const tal_t *ctx, struct lightningd *ld,
secp256k1_ecdsa_signature *sig)
{
u32 timestamp = time_now().ts.tv_sec;
u8 rgb[3] = {0x77, 0x88, 0x99};
u8 alias[32];
u8 *features = NULL;
u8 *addresses = tal_arr(ctx, u8, 0);
u8 *announcement;
if (!sig) {
sig = tal(ctx, secp256k1_ecdsa_signature);
memset(sig, 0, sizeof(*sig));
}
if (ld->dstate.config.ipaddr.type != ADDR_TYPE_PADDING) {
towire_ipaddr(&addresses, &ld->dstate.config.ipaddr);
}
memset(alias, 0, sizeof(alias));
announcement =
towire_node_announcement(ctx, sig, timestamp, &ld->dstate.id, rgb,
alias, features, addresses);
return announcement;
}
/* We got the signature for out provisional node_announcement back
* from the HSM, create the real announcement and forward it to
* gossipd so it can take care of forwarding it. */
static bool send_node_announcement_got_sig(struct subd *hsm, const u8 *msg,
const int *fds,
struct lightningd *ld)
{
tal_t *tmpctx = tal_tmpctx(hsm);
secp256k1_ecdsa_signature sig;
u8 *announcement, *wrappedmsg;
if (!fromwire_hsmctl_node_announcement_sig_reply(msg, NULL, &sig)) {
log_debug(ld->log,
"HSM returned an invalid node_announcement sig");
return false;
}
announcement = create_node_announcement(tmpctx, ld, &sig);
wrappedmsg = towire_gossip_forwarded_msg(tmpctx, announcement);
subd_send_msg(ld->gossip, take(wrappedmsg));
tal_free(tmpctx);
return true;
}
/* We were informed by channeld that it announced the channel and sent
* an update, so we can now start sending a node_announcement. The
* first step is to build the provisional announcement and ask the HSM
* to sign it. */
static void peer_channel_announced(struct peer *peer, const u8 *msg)
{
struct lightningd *ld = peer->ld;
tal_t *tmpctx = tal_tmpctx(ld);
u8 *req;
req = towire_hsmctl_node_announcement_sig_req(
tmpctx, create_node_announcement(tmpctx, ld, NULL));
subd_req(ld, ld->hsm, take(req), -1, 0,
send_node_announcement_got_sig, ld);
tal_free(tmpctx);
}
static int channel_msg(struct subd *sd, const u8 *msg, const int *unused)
{
enum channel_wire_type t = fromwire_peektype(msg);
switch (t) {
case WIRE_CHANNEL_NORMAL_OPERATION:
peer_set_condition(sd->peer,
CHANNELD_AWAITING_LOCKIN, CHANNELD_NORMAL);
break;
case WIRE_CHANNEL_ACCEPTED_HTLC:
return peer_accepted_htlc(sd->peer, msg);
case WIRE_CHANNEL_FULFILLED_HTLC:
return peer_fulfilled_htlc(sd->peer, msg);
case WIRE_CHANNEL_FAILED_HTLC:
return peer_failed_htlc(sd->peer, msg);
case WIRE_CHANNEL_MALFORMED_HTLC:
return peer_failed_malformed_htlc(sd->peer, msg);
case WIRE_CHANNEL_ANNOUNCED:
peer_channel_announced(sd->peer, msg);
break;
/* We never see fatal ones. */
case WIRE_CHANNEL_BAD_COMMAND:
case WIRE_CHANNEL_HSM_FAILED:
case WIRE_CHANNEL_CRYPTO_FAILED:
case WIRE_CHANNEL_INTERNAL_ERROR:
case WIRE_CHANNEL_PEER_WRITE_FAILED:
case WIRE_CHANNEL_PEER_READ_FAILED:
case WIRE_CHANNEL_PEER_BAD_MESSAGE:
/* And we never get these from channeld. */
case WIRE_CHANNEL_INIT:
case WIRE_CHANNEL_FUNDING_LOCKED:
case WIRE_CHANNEL_FUNDING_ANNOUNCE_DEPTH:
case WIRE_CHANNEL_OFFER_HTLC:
case WIRE_CHANNEL_FULFILL_HTLC:
case WIRE_CHANNEL_FAIL_HTLC:
case WIRE_CHANNEL_PING:
/* Replies go to requests. */
case WIRE_CHANNEL_OFFER_HTLC_REPLY:
case WIRE_CHANNEL_PING_REPLY:
break;
}
return 0;
}
/* We've got fd from HSM for channeld */
static bool peer_start_channeld_hsmfd(struct subd *hsm, const u8 *resp,
const int *fds,
struct peer *peer)
{
u8 *initmsg;
const struct config *cfg = &peer->ld->dstate.config;
peer->owner = new_subd(peer->ld, peer->ld,
"lightningd_channel", peer,
channel_wire_type_name,
channel_msg,
peer_owner_finished,
peer->fd,
peer->gossip_client_fd, fds[0], -1);
if (!peer->owner) {
log_unusual(peer->log, "Could not subdaemon channel: %s",
strerror(errno));
peer_fail(peer, "Failed to subdaemon channel");
return true;
}
log_debug(peer->log, "Waiting for funding confirmations");
peer_set_condition(peer, GETTING_HSMFD, CHANNELD_AWAITING_LOCKIN);
initmsg = towire_channel_init(peer,
peer->funding_txid,
peer->funding_outnum,
&peer->our_config,
&peer->channel_info->their_config,
&peer->channel_info->commit_sig,
peer->cs,
&peer->channel_info->remote_fundingkey,
&peer->channel_info->theirbase.revocation,
&peer->channel_info->theirbase.payment,
&peer->channel_info->theirbase.delayed_payment,
&peer->channel_info->their_per_commit_point,
peer->funder == LOCAL,
cfg->fee_base,
cfg->fee_per_satoshi,
peer->funding_satoshi,
peer->push_msat,
peer->seed,
&peer->ld->dstate.id,
peer->id,
time_to_msec(cfg->commit_time),
cfg->deadline_blocks);
/* We don't expect a response: we are triggered by funding_depth_cb. */
subd_send_msg(peer->owner, take(initmsg));
peer->fd = -1;
peer->cs = tal_free(peer->cs);
return true;
}
/* opening is done, start lightningd_channel for peer. */
static void peer_start_channeld(struct peer *peer, enum peer_state oldstate)
{
/* Unowned: back to being owned by main daemon. */
peer->owner = NULL;
tal_steal(peer->ld, peer);
log_debug(peer->log, "Waiting for HSM file descriptor");
/* Now we can consider balance set. */
peer->balance = tal_arr(peer, u64, NUM_SIDES);
peer->balance[peer->funder] = peer->funding_satoshi * 1000 - peer->push_msat;
peer->balance[!peer->funder] = peer->push_msat;
peer_set_condition(peer, oldstate, GETTING_HSMFD);
/* Get fd from hsm. */
subd_req(peer, peer->ld->hsm,
take(towire_hsmctl_hsmfd_channeld(peer, peer->unique_id)),
-1, 1, peer_start_channeld_hsmfd, peer);
}
static bool opening_funder_finished(struct subd *opening, const u8 *resp,
const int *fds,
struct funding_channel *fc)
{
u8 *msg;
struct channel_info *channel_info;
struct utxo *utxos;
struct sha256_double funding_txid;
struct pubkey changekey;
struct pubkey local_fundingkey;
assert(tal_count(fds) == 2);
fc->peer->fd = fds[0];
fc->peer->gossip_client_fd = fds[1];
fc->peer->cs = tal(fc->peer, struct crypto_state);
/* At this point, we care about peer */
fc->peer->channel_info = channel_info
= tal(fc->peer, struct channel_info);
if (!fromwire_opening_funder_reply(resp, NULL,
&channel_info->their_config,
&channel_info->commit_sig,
fc->peer->cs,
&channel_info->theirbase.revocation,
&channel_info->theirbase.payment,
&channel_info->theirbase.delayed_payment,
&channel_info->their_per_commit_point,
&fc->peer->minimum_depth,
&channel_info->remote_fundingkey,
&funding_txid)) {
log_broken(fc->peer->log, "bad OPENING_FUNDER_REPLY %s",
tal_hex(resp, resp));
tal_free(fc->peer);
return false;
}
/* Generate the funding tx. */
if (fc->change
&& !bip32_pubkey(fc->peer->ld->bip32_base,
&changekey, fc->change_keyindex))
fatal("Error deriving change key %u", fc->change_keyindex);
derive_basepoints(fc->peer->seed, &local_fundingkey,
NULL, NULL, NULL, NULL, 0);
fc->funding_tx = funding_tx(fc, &fc->peer->funding_outnum,
fc->utxomap, fc->peer->funding_satoshi,
&local_fundingkey,
&channel_info->remote_fundingkey,
fc->change, &changekey,
fc->peer->ld->bip32_base);
fc->peer->funding_txid = tal(fc->peer, struct sha256_double);
bitcoin_txid(fc->funding_tx, fc->peer->funding_txid);
if (!structeq(fc->peer->funding_txid, &funding_txid)) {
peer_fail(fc->peer, "Funding txid mismatch:"
" satoshi %"PRIu64" change %"PRIu64" changeidx %u"
" localkey %s remotekey %s",
fc->peer->funding_satoshi,
fc->change, fc->change_keyindex,
type_to_string(fc, struct pubkey, &local_fundingkey),
type_to_string(fc, struct pubkey,
&channel_info->remote_fundingkey));
return false;
}
/* Get HSM to sign the funding tx. */
log_debug(fc->peer->log, "Getting HSM to sign funding tx");
utxos = from_utxoptr_arr(fc, fc->utxomap);
msg = towire_hsmctl_sign_funding(fc, fc->peer->funding_satoshi,
fc->change, fc->change_keyindex,
&local_fundingkey,
&channel_info->remote_fundingkey,
utxos);
tal_free(utxos);
fc->peer->owner = NULL;
peer_set_condition(fc->peer, OPENINGD, GETTING_SIG_FROM_HSM);
subd_req(fc, fc->peer->ld->hsm, take(msg), -1, 0,
opening_got_hsm_funding_sig, fc);
/* Tell opening daemon to exit. */
return false;
}
static bool opening_fundee_finished(struct subd *opening,
const u8 *reply,
const int *fds,
struct peer *peer)
{
u8 *funding_msg_enc;
struct channel_info *channel_info;
log_debug(peer->log, "Got opening_fundee_finish_response");
assert(tal_count(fds) == 2);
peer->fd = fds[0];
peer->gossip_client_fd = fds[1];
peer->cs = tal(peer, struct crypto_state);
/* At this point, we care about peer */
peer->channel_info = channel_info = tal(peer, struct channel_info);
peer->funding_txid = tal(peer, struct sha256_double);
if (!fromwire_opening_fundee_reply(reply, reply, NULL,
&channel_info->their_config,
&channel_info->commit_sig,
peer->cs,
&channel_info->theirbase.revocation,
&channel_info->theirbase.payment,
&channel_info->theirbase.delayed_payment,
&channel_info->their_per_commit_point,
&channel_info->remote_fundingkey,
peer->funding_txid,
&peer->funding_outnum,
&peer->funding_satoshi,
&peer->push_msat,
&funding_msg_enc)) {
log_broken(peer->log, "bad OPENING_FUNDEE_REPLY %s",
tal_hex(reply, reply));
return false;
}
log_debug(peer->log, "Watching funding tx %s",
type_to_string(reply, struct sha256_double,
peer->funding_txid));
watch_txid(peer, peer->ld->topology, peer, peer->funding_txid,
funding_lockin_cb, NULL);
/* FIXME: Remove synchronous write! */
if (write(peer->fd, funding_msg_enc, tal_len(funding_msg_enc))
!= tal_len(funding_msg_enc)) {
log_broken(peer->log, "Could not write funding_signed msg");
return false;
}
/* On to normal operation! */
peer->owner = NULL;
peer_start_channeld(peer, OPENINGD);
/* Tell opening daemon to exit. */
return false;
}
static void channel_config(struct lightningd *ld,
struct channel_config *ours,
u32 *max_to_self_delay,
u32 *max_minimum_depth,
u64 *min_effective_htlc_capacity_msat)
{
/* FIXME: depend on feerate. */
*max_to_self_delay = ld->dstate.config.locktime_max;
*max_minimum_depth = ld->dstate.config.anchor_confirms_max;
/* This is 1c at $1000/BTC */
*min_effective_htlc_capacity_msat = 1000000;
/* BOLT #2:
*
* The sender SHOULD set `dust-limit-satoshis` to a sufficient
* value to allow commitment transactions to propagate through
* the Bitcoin network.
*/
ours->dust_limit_satoshis = 546;
ours->max_htlc_value_in_flight_msat = UINT64_MAX;
/* Don't care */
ours->htlc_minimum_msat = 0;
/* BOLT #2:
*
* The sender SHOULD set `to-self-delay` sufficient to ensure
* the sender can irreversibly spend a commitment transaction
* output in case of misbehavior by the receiver.
*/
ours->to_self_delay = ld->dstate.config.locktime_blocks;
/* BOLT #2:
*
* It MUST fail the channel if `max-accepted-htlcs` is greater than
* 483.
*/
ours->max_accepted_htlcs = 483;
/* This is filled in by lightningd_opening, for consistency. */
ours->channel_reserve_satoshis = 0;
};
/* Peer has spontaneously exited from gossip due to msg */
void peer_fundee_open(struct peer *peer, const u8 *from_peer)
{
struct lightningd *ld = peer->ld;
u32 max_to_self_delay, max_minimum_depth;
u64 min_effective_htlc_capacity_msat;
u8 *msg;
/* Note: gossipd handles unknown packets, so we don't have to worry
* about ignoring odd ones here. */
if (fromwire_peektype(from_peer) != WIRE_OPEN_CHANNEL) {
log_unusual(peer->log, "Strange message to exit gossip: %u",
fromwire_peektype(from_peer));
peer_fail(peer, "Bad message during gossiping: %s",
tal_hex(peer, from_peer));
return;
}
peer_set_condition(peer, GOSSIPD, OPENINGD);
peer->owner = new_subd(ld, ld, "lightningd_opening", peer,
opening_wire_type_name,
NULL, peer_owner_finished,
peer->fd, peer->gossip_client_fd, -1);
if (!peer->owner) {
peer_fail(peer, "Failed to subdaemon opening: %s",
strerror(errno));
return;
}
/* We handed off peer fd and gossip fd */
peer->fd = -1;
peer->gossip_client_fd = -1;
/* They will open channel. */
peer->funder = REMOTE;
/* BOLT #2:
*
* The sender SHOULD set `minimum-depth` to an amount where
* the sender considers reorganizations to be low risk.
*/
peer->minimum_depth = ld->dstate.config.anchor_confirms;
channel_config(ld, &peer->our_config,
&max_to_self_delay, &max_minimum_depth,
&min_effective_htlc_capacity_msat);
peer->seed = tal(peer, struct privkey);
derive_peer_seed(ld, peer->seed, peer->id);
msg = towire_opening_init(peer, &peer->our_config,
max_to_self_delay,
min_effective_htlc_capacity_msat,
peer->cs, peer->seed);
peer->cs = tal_free(peer->cs);
subd_send_msg(peer->owner, take(msg));
msg = towire_opening_fundee(peer, peer->minimum_depth,
7500, 150000, from_peer);
/* Careful here! Their message could push us overlength! */
if (tal_len(msg) >= 65536) {
peer_fail(peer, "Unacceptably long open_channel");
return;
}
subd_req(peer, peer->owner, take(msg), -1, 2,
opening_fundee_finished, peer);
}
/* Peer has been released from gossip. Start opening. */
static bool gossip_peer_released(struct subd *gossip,
const u8 *resp,
const int *fds,
struct funding_channel *fc)
{
struct lightningd *ld = fc->peer->ld;
u32 max_to_self_delay, max_minimum_depth;
u64 min_effective_htlc_capacity_msat;
u64 id;
u8 *msg;
struct subd *opening;
struct utxo *utxos;
u8 *bip32_base;
assert(tal_count(fds) == 2);
fc->peer->fd = fds[0];
fc->peer->cs = tal(fc, struct crypto_state);
fc->peer->gossip_client_fd = fds[1];
if (!fromwire_gossipctl_release_peer_reply(resp, NULL, &id, fc->peer->cs))
fatal("Gossup daemon gave invalid reply %s",
tal_hex(gossip, resp));
if (id != fc->peer->unique_id)
fatal("Gossup daemon release gave %"PRIu64" not %"PRIu64,
id, fc->peer->unique_id);
peer_set_condition(fc->peer, GOSSIPD, OPENINGD);
opening = new_subd(fc->peer->ld, ld,
"lightningd_opening", fc->peer,
opening_wire_type_name,
NULL, peer_owner_finished,
fc->peer->fd, fc->peer->gossip_client_fd, -1);
if (!opening) {
peer_fail(fc->peer, "Failed to subdaemon opening: %s",
strerror(errno));
return true;
}
fc->peer->owner = opening;
/* They took our fds. */
fc->peer->fd = -1;
fc->peer->gossip_client_fd = -1;
/* We will fund channel */
fc->peer->funder = LOCAL;
channel_config(ld, &fc->peer->our_config,
&max_to_self_delay, &max_minimum_depth,
&min_effective_htlc_capacity_msat);
fc->peer->seed = tal(fc->peer, struct privkey);
derive_peer_seed(ld, fc->peer->seed, fc->peer->id);
msg = towire_opening_init(fc, &fc->peer->our_config,
max_to_self_delay,
min_effective_htlc_capacity_msat,
fc->peer->cs, fc->peer->seed);
fc->peer->cs = tal_free(fc->peer->cs);
subd_send_msg(opening, take(msg));
utxos = from_utxoptr_arr(fc, fc->utxomap);
bip32_base = tal_arr(fc, u8, BIP32_SERIALIZED_LEN);
if (bip32_key_serialize(fc->peer->ld->bip32_base, BIP32_FLAG_KEY_PUBLIC,
bip32_base, tal_len(bip32_base))
!= WALLY_OK)
fatal("Can't serialize bip32 public key");
/* FIXME: Real feerate! */
msg = towire_opening_funder(fc, fc->peer->funding_satoshi,
fc->peer->push_msat,
15000, max_minimum_depth,
fc->change, fc->change_keyindex,
utxos, bip32_base);
subd_req(fc, opening, take(msg), -1, 2, opening_funder_finished, fc);
return true;
}
static void json_fund_channel(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
struct lightningd *ld = ld_from_dstate(cmd->dstate);
jsmntok_t *peertok, *satoshitok;
struct funding_channel *fc = tal(cmd, struct funding_channel);
u8 *msg;
if (!json_get_params(buffer, params,
"id", &peertok,
"satoshi", &satoshitok,
NULL)) {
command_fail(cmd, "Need peerid and satoshi");
return;
}
fc->cmd = cmd;
fc->peer = peer_from_json(ld, buffer, peertok);
if (!fc->peer) {
command_fail(cmd, "Could not find peer with that peerid");
return;
}
if (fc->peer->owner != ld->gossip) {
command_fail(cmd, "Peer not ready for connection");
return;
}
if (!json_tok_u64(buffer, satoshitok, &fc->peer->funding_satoshi)) {
command_fail(cmd, "Invalid satoshis");
return;
}
/* FIXME: Support push_msat? */
fc->peer->push_msat = 0;
/* Try to do this now, so we know if insufficient funds. */
/* FIXME: Feerate & dustlimit */
fc->utxomap = build_utxos(fc, ld, fc->peer->funding_satoshi, 15000, 600,
&fc->change, &fc->change_keyindex);
if (!fc->utxomap) {
command_fail(cmd, "Cannot afford funding transaction");
return;
}
msg = towire_gossipctl_release_peer(cmd, fc->peer->unique_id);
/* Tie this fc lifetime (and hence utxo release) to the peer */
tal_steal(fc->peer, fc);
tal_add_destructor(fc, fail_fundchannel_command);
subd_req(fc, ld->gossip, msg, -1, 2, gossip_peer_released, fc);
}
static const struct json_command fund_channel_command = {
"fundchannel",
json_fund_channel,
"Fund channel with {id} using {satoshi} satoshis",
"Returns once channel established"
};
AUTODATA(json_command, &fund_channel_command);
const char *peer_state_name(enum peer_state state)
{
size_t i;
for (i = 0; enum_peer_state_names[i].name; i++)
if (enum_peer_state_names[i].v == state)
return enum_peer_state_names[i].name;
return "unknown";
}