rgb-cln/lightningd/peer_control.c

2815 lines
79 KiB
C

#include "lightningd.h"
#include "peer_control.h"
#include "subd.h"
#include <arpa/inet.h>
#include <bitcoin/script.h>
#include <bitcoin/tx.h>
#include <ccan/fdpass/fdpass.h>
#include <ccan/io/io.h>
#include <ccan/noerr/noerr.h>
#include <ccan/str/str.h>
#include <ccan/take/take.h>
#include <ccan/tal/str/str.h>
#include <channeld/gen_channel_wire.h>
#include <closingd/gen_closing_wire.h>
#include <common/close_tx.h>
#include <common/dev_disconnect.h>
#include <common/funding_tx.h>
#include <common/initial_commit_tx.h>
#include <common/key_derive.h>
#include <common/status.h>
#include <common/timeout.h>
#include <common/wire_error.h>
#include <errno.h>
#include <fcntl.h>
#include <gossipd/gen_gossip_wire.h>
#include <hsmd/capabilities.h>
#include <hsmd/gen_hsm_wire.h>
#include <inttypes.h>
#include <lightningd/build_utxos.h>
#include <lightningd/chaintopology.h>
#include <lightningd/gen_peer_state_names.h>
#include <lightningd/hsm_control.h>
#include <lightningd/jsonrpc.h>
#include <lightningd/log.h>
#include <lightningd/netaddress.h>
#include <lightningd/options.h>
#include <lightningd/peer_htlcs.h>
#include <onchaind/gen_onchain_wire.h>
#include <onchaind/onchain_wire.h>
#include <openingd/gen_opening_wire.h>
#include <unistd.h>
#include <wally_bip32.h>
#include <wire/gen_onion_wire.h>
#include <wire/peer_wire.h>
#include <wire/wire_sync.h>
static const u8 supported_local_features[]
= {LOCALFEATURES_INITIAL_ROUTING_SYNC};
static const u8 supported_global_features[]
= {};
struct connect {
struct list_node list;
struct pubkey id;
struct command *cmd;
};
/* FIXME: Reorder */
struct funding_channel;
static void peer_offer_channel(struct lightningd *ld,
struct funding_channel *fc,
const struct wireaddr *addr,
const struct crypto_state *cs,
const u8 *gfeatures, const u8 *lfeatures,
int peer_fd, int gossip_fd);
static bool peer_start_channeld(struct peer *peer,
const struct crypto_state *cs,
int peer_fd, int gossip_fd,
const u8 *funding_signed,
bool reconnected);
static void peer_start_closingd(struct peer *peer,
struct crypto_state *cs,
int peer_fd, int gossip_fd,
bool reconnected);
static void peer_accept_channel(struct lightningd *ld,
const struct pubkey *peer_id,
const struct wireaddr *addr,
const struct crypto_state *cs,
const u8 *gfeatures, const u8 *lfeatures,
int peer_fd, int gossip_fd,
const u8 *open_msg);
static void peer_set_owner(struct peer *peer, struct subd *owner)
{
struct subd *old_owner = peer->owner;
peer->owner = owner;
if (old_owner)
subd_release_peer(old_owner, peer);
}
static void destroy_peer(struct peer *peer)
{
/* Must not have any HTLCs! */
struct htlc_out_map_iter outi;
struct htlc_out *hout;
struct htlc_in_map_iter ini;
struct htlc_in *hin;
for (hout = htlc_out_map_first(&peer->ld->htlcs_out, &outi);
hout;
hout = htlc_out_map_next(&peer->ld->htlcs_out, &outi)) {
if (hout->key.peer != peer)
continue;
fatal("Freeing peer %s has hout %s",
peer_state_name(peer->state),
htlc_state_name(hout->hstate));
}
for (hin = htlc_in_map_first(&peer->ld->htlcs_in, &ini);
hin;
hin = htlc_in_map_next(&peer->ld->htlcs_in, &ini)) {
if (hin->key.peer != peer)
continue;
fatal("Freeing peer %s has hin %s",
peer_state_name(peer->state),
htlc_state_name(hin->hstate));
}
/* Free any old owner still hanging around. */
peer_set_owner(peer, NULL);
list_del_from(&peer->ld->peers, &peer->list);
}
u8 *get_supported_global_features(const tal_t *ctx)
{
return tal_dup_arr(ctx, u8, supported_global_features,
sizeof(supported_global_features), 0);
}
u8 *get_supported_local_features(const tal_t *ctx)
{
return tal_dup_arr(ctx, u8, supported_local_features,
sizeof(supported_local_features), 0);
}
static void sign_last_tx(struct peer *peer)
{
const tal_t *tmpctx = tal_tmpctx(peer);
u8 *funding_wscript;
struct pubkey local_funding_pubkey;
struct secrets secrets;
secp256k1_ecdsa_signature sig;
derive_basepoints(peer->seed, &local_funding_pubkey, NULL, &secrets,
NULL);
funding_wscript = bitcoin_redeem_2of2(tmpctx,
&local_funding_pubkey,
&peer->channel_info->remote_fundingkey);
/* Need input amount for signing */
peer->last_tx->input[0].amount = tal_dup(peer->last_tx->input, u64,
&peer->funding_satoshi);
sign_tx_input(peer->last_tx, 0, NULL, funding_wscript,
&secrets.funding_privkey,
&local_funding_pubkey,
&sig);
peer->last_tx->input[0].witness
= bitcoin_witness_2of2(peer->last_tx->input,
peer->last_sig,
&sig,
&peer->channel_info->remote_fundingkey,
&local_funding_pubkey);
tal_free(tmpctx);
}
static void drop_to_chain(struct peer *peer)
{
sign_last_tx(peer);
/* Keep broadcasting until we say stop (can fail due to dup,
* if they beat us to the broadcast). */
broadcast_tx(peer->ld->topology, peer, peer->last_tx, NULL);
}
void peer_fail_permanent(struct peer *peer, const u8 *msg TAKES)
{
/* BOLT #1:
*
* The channel is referred to by `channel_id` unless `channel_id` is
* zero (ie. all bytes zero), in which case it refers to all
* channels. */
static const struct channel_id all_channels;
log_unusual(peer->log, "Peer permanent failure in %s: %.*s",
peer_state_name(peer->state),
(int)tal_len(msg), (char *)msg);
/* We can have multiple errors, eg. onchaind failures. */
if (!peer->error)
peer->error = towire_error(peer, &all_channels, msg);
peer_set_owner(peer, NULL);
if (taken(msg))
tal_free(msg);
if (peer_persists(peer))
drop_to_chain(peer);
else
tal_free(peer);
return;
}
void peer_fail_permanent_str(struct peer *peer, const char *str TAKES)
{
/* Don't use tal_strdup, since we need tal_len */
u8 *msg = tal_dup_arr(peer, u8, (const u8 *)str, strlen(str) + 1, 0);
if (taken(str))
tal_free(str);
peer_fail_permanent(peer, take(msg));
}
void peer_internal_error(struct peer *peer, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
log_broken(peer->log, "Peer internal error %s: ",
peer_state_name(peer->state));
logv_add(peer->log, fmt, ap);
va_end(ap);
peer_fail_permanent_str(peer, "Internal error");
}
void peer_fail_transient(struct peer *peer, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
log_info(peer->log, "Peer transient failure in %s: ",
peer_state_name(peer->state));
logv_add(peer->log, fmt, ap);
va_end(ap);
#if DEVELOPER
if (dev_disconnect_permanent(peer->ld)) {
peer_internal_error(peer, "dev_disconnect permfail");
return;
}
#endif
peer_set_owner(peer, NULL);
/* 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));
tal_free(peer);
return;
}
/* Reconnect unless we've dropped/are dropping to chain. */
if (!peer_on_chain(peer) && peer->state != CLOSINGD_COMPLETE) {
/* Don't schedule an attempt if we disabled reconnections with
* the `--no-reconnect` flag */
if (peer->ld->config.no_reconnect)
return;
u8 *msg = towire_gossipctl_reach_peer(peer, &peer->id);
subd_send_msg(peer->ld->gossip, take(msg));
}
}
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;
/* We only persist channels/peers that have reached the opening state */
if (peer_persists(peer)) {
assert(peer->channel != NULL);
/* TODO(cdecker) Selectively save updated fields to DB */
wallet_channel_save(peer->ld->wallet, peer->channel);
}
}
static void destroy_connect(struct connect *c)
{
list_del(&c->list);
}
static struct connect *new_connect(struct lightningd *ld,
const struct pubkey *id,
struct command *cmd)
{
struct connect *c = tal(cmd, struct connect);
c->id = *id;
c->cmd = cmd;
list_add(&ld->connects, &c->list);
tal_add_destructor(c, destroy_connect);
return c;
}
static void connect_succeeded(struct lightningd *ld, const struct pubkey *id)
{
struct connect *i, *next;
/* Careful! Completing command frees connect. */
list_for_each_safe(&ld->connects, i, next, list) {
struct json_result *response;
if (!pubkey_eq(&i->id, id))
continue;
response = new_json_result(i->cmd);
json_object_start(response, NULL);
json_add_pubkey(response, "id", id);
json_object_end(response);
command_success(i->cmd, response);
}
}
static void connect_failed(struct lightningd *ld, const struct pubkey *id,
const char *error)
{
struct connect *i, *next;
/* Careful! Completing command frees connect. */
list_for_each_safe(&ld->connects, i, next, list) {
if (pubkey_eq(&i->id, id))
command_fail(i->cmd, "%s", error);
}
}
static struct peer *new_peer(struct lightningd *ld,
const struct pubkey *id,
const struct wireaddr *addr,
const u8 *gfeatures, const u8 *lfeatures,
int peer_fd)
{
struct peer *peer;
/* Need to memset since storing will access all fields */
peer = talz(ld, struct peer);
peer->error = NULL;
peer->id = *id;
peer->addr = *addr;
peer->funding_txid = NULL;
peer->remote_funding_locked = false;
peer->scid = NULL;
peer->seed = NULL;
peer->our_msatoshi = NULL;
peer->state = UNINITIALIZED;
peer->channel_info = NULL;
peer->last_tx = NULL;
peer->last_sig = NULL;
peer->last_htlc_sigs = NULL;
peer->last_was_revoke = false;
peer->last_sent_commit = NULL;
peer->remote_shutdown_scriptpubkey = NULL;
peer->local_shutdown_idx = -1;
peer->next_index[LOCAL]
= peer->next_index[REMOTE] = 0;
peer->next_htlc_id = 0;
wallet_shachain_init(ld->wallet, &peer->their_shachain);
/* FIXME: db should be keyed by (peer_id, channel_id) */
/* If we have the peer in the DB, this'll populate the fields,
* failure just indicates that the peer wasn't found in the
* DB */
wallet_peer_by_nodeid(ld->wallet, id, peer);
/* peer->channel gets populated as soon as we start opening a channel */
peer->channel = NULL;
list_add_tail(&ld->peers, &peer->list);
populate_peer(ld, peer);
return peer;
}
/**
* requires_unsupported_features - Check if we support what's being asked
*
* Given the features vector that the remote connection is expecting
* from us, we check to see if we support all even bit features, i.e.,
* the required features. We do so by subtracting our own features in
* the provided positions and see if even bits remain.
*
* @bitmap: the features bitmap the peer is asking for
* @supportmap: what do we support
* @smlen: how long is our supportmap
*/
static bool requires_unsupported_features(const u8 *bitmap,
const u8 *supportmap,
size_t smlen)
{
size_t len = tal_count(bitmap);
u8 support;
for (size_t i=0; i<len; i++) {
/* Find matching bitmap byte in supportmap, 0x00 if none */
if (len > smlen) {
support = 0x00;
} else {
support = supportmap[smlen-1];
}
/* Cancel out supported bits, check for even bits */
if ((~support & bitmap[i]) & 0x55)
return true;
}
return false;
}
bool unsupported_features(const u8 *gfeatures, const u8 *lfeatures)
{
return requires_unsupported_features(gfeatures,
supported_global_features,
sizeof(supported_global_features))
|| requires_unsupported_features(lfeatures,
supported_local_features,
sizeof(supported_local_features));
}
/* Extract channel_id from various packets, return true if possible. */
static bool extract_channel_id(const u8 *in_pkt, struct channel_id *channel_id)
{
u64 ignored_u64;
u32 ignored_u32;
u16 ignored_u16;
u8 ignored_u8;
struct pubkey ignored_pubkey;
struct sha256_double ignored_shadouble;
if (fromwire_channel_reestablish(in_pkt, NULL, channel_id,
&ignored_u64, &ignored_u64))
return true;
if (fromwire_open_channel(in_pkt, NULL, &ignored_shadouble,
channel_id, &ignored_u64,
&ignored_u64, &ignored_u64,
&ignored_u64, &ignored_u64,
&ignored_u64, &ignored_u32,
&ignored_u16, &ignored_u16,
&ignored_pubkey, &ignored_pubkey,
&ignored_pubkey, &ignored_pubkey,
&ignored_pubkey, &ignored_pubkey,
&ignored_u8))
return true;
return false;
}
/**
* peer_channel_new -- Instantiate a new channel for the given peer and save it
*
* We are about to open a channel with the peer, either due to a
* nongossip message from remote, or because we initiated an
* open. This creates the `struct wallet_channel` for the peer and
* stores it in the database.
*
* @w: the wallet to store the information in
* @peer: the peer we are opening a channel to
*
* This currently overwrites peer->channel, so can only be used if we
* allow a single channel per peer.
*/
static struct wallet_channel *peer_channel_new(struct wallet *w,
struct peer *peer)
{
struct wallet_channel *wc = tal(peer, struct wallet_channel);
wc->peer = peer;
wallet_peer_by_nodeid(w, &peer->id, peer);
wc->id = 0;
wallet_channel_save(w, wc);
return wc;
}
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->config.locktime_max;
*max_minimum_depth = ld->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->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 lightning_openingd, for consistency. */
ours->channel_reserve_satoshis = 0;
};
/* Gossipd tells us a peer has connected */
void peer_connected(struct lightningd *ld, const u8 *msg,
int peer_fd, int gossip_fd)
{
struct pubkey id;
struct crypto_state cs;
u8 *gfeatures, *lfeatures;
u8 *error;
struct peer *peer;
struct wireaddr addr;
if (!fromwire_gossip_peer_connected(msg, msg, NULL,
&id, &addr, &cs,
&gfeatures, &lfeatures))
fatal("Gossip gave bad GOSSIP_PEER_CONNECTED message %s",
tal_hex(msg, msg));
if (unsupported_features(gfeatures, lfeatures)) {
log_unusual(ld->log, "peer %s offers unsupported features %s/%s",
type_to_string(msg, struct pubkey, &id),
tal_hex(msg, gfeatures),
tal_hex(msg, lfeatures));
error = towire_errorfmt(msg, NULL,
"We only support globalfeatures %s"
" and localfeatures %s",
tal_hexstr(msg,
supported_global_features,
sizeof(supported_global_features)),
tal_hexstr(msg,
supported_local_features,
sizeof(supported_local_features)));
goto send_error;
}
/* Now, do we already know this peer? */
peer = peer_by_id(ld, &id);
if (peer) {
log_debug(peer->log, "Peer has reconnected, state %s",
peer_state_name(peer->state));
/* FIXME: We can have errors for multiple channels. */
if (peer->error) {
error = peer->error;
goto send_error;
}
#if DEVELOPER
if (dev_disconnect_permanent(ld)) {
peer_internal_error(peer, "dev_disconnect permfail");
error = peer->error;
goto send_error;
}
#endif
switch (peer->state) {
/* This can't happen. */
case UNINITIALIZED:
abort();
/* Reconnect: discard old one. */
case OPENINGD:
peer = tal_free(peer);
goto return_to_gossipd;
case ONCHAIND_CHEATED:
case ONCHAIND_THEIR_UNILATERAL:
case ONCHAIND_OUR_UNILATERAL:
case FUNDING_SPEND_SEEN:
case ONCHAIND_MUTUAL:
/* If they try to reestablish channel, we'll send
* error then */
goto return_to_gossipd;
case CHANNELD_AWAITING_LOCKIN:
case CHANNELD_NORMAL:
case CHANNELD_SHUTTING_DOWN:
/* Stop any existing daemon, without triggering error
* on this peer. */
peer_set_owner(peer, NULL);
peer->addr = addr;
peer_start_channeld(peer, &cs, peer_fd, gossip_fd, NULL,
true);
return;
case CLOSINGD_SIGEXCHANGE:
case CLOSINGD_COMPLETE:
/* Stop any existing daemon, without triggering error
* on this peer. */
peer_set_owner(peer, NULL);
peer->addr = addr;
peer_start_closingd(peer, &cs, peer_fd, gossip_fd,
true);
return;
}
abort();
}
return_to_gossipd:
/* Otherwise, we hand back to gossipd, to continue. */
msg = towire_gossipctl_handle_peer(msg, &id, &addr, &cs,
gfeatures, lfeatures, NULL);
subd_send_msg(ld->gossip, take(msg));
subd_send_fd(ld->gossip, peer_fd);
close(gossip_fd);
/* If we were waiting for connection, we succeeded. */
connect_succeeded(ld, &id);
return;
send_error:
/* Hand back to gossipd, with an error packet. */
connect_failed(ld, &id, sanitize_error(msg, error, NULL));
msg = towire_gossipctl_handle_peer(msg, &id, &addr, &cs,
gfeatures, lfeatures, error);
subd_send_msg(ld->gossip, take(msg));
subd_send_fd(ld->gossip, peer_fd);
close(gossip_fd);
}
void peer_sent_nongossip(struct lightningd *ld,
const struct pubkey *id,
const struct wireaddr *addr,
const struct crypto_state *cs,
const u8 *gfeatures,
const u8 *lfeatures,
int peer_fd, int gossip_fd,
const u8 *in_msg)
{
struct channel_id *channel_id, extracted_channel_id;
struct peer *peer;
u8 *error, *msg;
if (!extract_channel_id(in_msg, &extracted_channel_id))
channel_id = NULL;
else
channel_id = &extracted_channel_id;
/* FIXME: match state too; we can have multiple onchain
* (ie. dead) channels for the same peer. */
peer = peer_by_id(ld, id);
if (peer) {
error = towire_errorfmt(ld, channel_id,
"Unexpected message %i in state %s",
fromwire_peektype(in_msg),
peer_state_name(peer->state));
goto send_error;
}
/* Open request? */
if (fromwire_peektype(in_msg) == WIRE_OPEN_CHANNEL) {
peer_accept_channel(ld, id, addr, cs, gfeatures, lfeatures,
peer_fd, gossip_fd, in_msg);
return;
}
/* Weird request. */
error = towire_errorfmt(ld, channel_id,
"Unexpected message %i for unknown peer",
fromwire_peektype(in_msg));
send_error:
/* Hand back to gossipd, with an error packet. */
connect_failed(ld, id, sanitize_error(error, error, NULL));
msg = towire_gossipctl_handle_peer(error, id, addr, cs,
gfeatures, lfeatures, error);
subd_send_msg(ld->gossip, take(msg));
subd_send_fd(ld->gossip, peer_fd);
close(gossip_fd);
tal_free(error);
}
/* We copy per-peer entries above --log-level into the main log. */
static void copy_to_parent_log(const char *prefix,
enum log_level level,
bool continued,
const char *str,
struct peer *peer)
{
const char *idstr = type_to_string(peer, struct pubkey, &peer->id);
if (continued)
log_add(peer->ld->log, "peer %s: ... %s", idstr, str);
else
log_(peer->ld->log, level, "peer %s: %s", idstr, str);
tal_free(idstr);
}
void populate_peer(struct lightningd *ld, struct peer *peer)
{
const char *idname;
struct pubkey *id = &peer->id;
idname = type_to_string(peer, struct pubkey, id);
peer->ld = ld;
/* Max 128k per peer. */
peer->log_book = new_log_book(peer, 128*1024,
get_log_level(ld->log_book));
peer->log = new_log(peer, peer->log_book, "peer %s:", idname);
set_log_outfn(peer->log_book, copy_to_parent_log, peer);
tal_free(idname);
tal_add_destructor(peer, destroy_peer);
}
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;
}
static void json_connect(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
jsmntok_t *hosttok, *idtok;
struct pubkey id;
const char *name, *port, *colon;
struct wireaddr addr;
u8 *msg;
if (!json_get_params(buffer, params,
"id", &idtok,
"?host", &hosttok,
NULL)) {
command_fail(cmd, "Need id to connect");
return;
}
if (!json_tok_pubkey(buffer, idtok, &id)) {
command_fail(cmd, "id %.*s not valid",
idtok->end - idtok->start,
buffer + idtok->start);
return;
}
if (hosttok) {
colon = memchr(buffer + hosttok->start, ':',
hosttok->end - hosttok->start);
if (colon) {
name = tal_strndup(cmd, buffer + hosttok->start,
colon - (buffer + hosttok->start));
port = tal_strndup(cmd, colon + 1,
(buffer + hosttok->end) - colon - 1);
} else {
name = tal_strndup(cmd, buffer + hosttok->start,
hosttok->end - hosttok->start);
port = tal_strdup(cmd, stringify(DEFAULT_PORT));
}
addr.port = atoi(port);
if (!parse_wireaddr(name, &addr, addr.port) || !addr.port)
command_fail(cmd, "host %s:%s not valid", name, port);
/* Tell it about the address. */
msg = towire_gossipctl_peer_addrhint(cmd, &id, &addr);
subd_send_msg(cmd->ld->gossip, take(msg));
}
/* Now tell it to try reaching it. */
msg = towire_gossipctl_reach_peer(cmd, &id);
subd_send_msg(cmd->ld->gossip, take(msg));
/* Leave this here for gossip_peer_connected */
new_connect(cmd->ld, &id, cmd);
}
static const struct json_command connect_command = {
"connect",
json_connect,
"Connect to {id} at {host} (which can end in ':port' if not default)",
"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);
}
struct getpeers_args {
struct command *cmd;
/* If non-NULL, they want logs too */
enum log_level *ll;
};
static void gossipd_getpeers_complete(struct subd *gossip, const u8 *msg,
const int *fds,
struct getpeers_args *gpa)
{
/* This is a little sneaky... */
struct pubkey *ids;
struct wireaddr *addrs;
struct json_result *response = new_json_result(gpa->cmd);
struct peer *p;
if (!fromwire_gossip_getpeers_reply(msg, msg, NULL, &ids, &addrs)) {
command_fail(gpa->cmd, "Bad response from gossipd");
return;
}
/* First the peers not just gossiping. */
json_object_start(response, NULL);
json_array_start(response, "peers");
list_for_each(&gpa->cmd->ld->peers, p, list) {
json_object_start(response, NULL);
json_add_string(response, "state", peer_state_name(p->state));
json_array_start(response, "netaddr");
if (p->addr.type != ADDR_TYPE_PADDING)
json_add_string(response, NULL,
type_to_string(response, struct wireaddr,
&p->addr));
json_array_end(response);
json_add_pubkey(response, "peerid", &p->id);
json_add_bool(response, "connected", p->owner != NULL);
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->our_msatoshi) {
json_add_u64(response, "msatoshi_to_us",
*p->our_msatoshi);
json_add_u64(response, "msatoshi_total",
p->funding_satoshi * 1000);
}
if (gpa->ll) {
struct log_info info;
info.level = *gpa->ll;
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);
}
for (size_t i = 0; i < tal_count(ids); i++) {
/* Don't report peers in both, which can happen if they're
* reconnecting */
if (peer_by_id(gpa->cmd->ld, ids + i))
continue;
json_object_start(response, NULL);
/* Fake state. */
json_add_string(response, "state", "GOSSIPING");
json_add_pubkey(response, "peerid", ids+i);
json_array_start(response, "netaddr");
if (addrs[i].type != ADDR_TYPE_PADDING)
json_add_string(response, NULL,
type_to_string(response, struct wireaddr,
addrs + i));
json_array_end(response);
json_add_bool(response, "connected", "true");
json_add_string(response, "owner", gossip->name);
json_object_end(response);
}
json_array_end(response);
json_object_end(response);
command_success(gpa->cmd, response);
}
static void json_getpeers(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
jsmntok_t *leveltok;
struct getpeers_args *gpa = tal(cmd, struct getpeers_args);
gpa->cmd = cmd;
json_get_params(buffer, params, "?level", &leveltok, NULL);
if (leveltok) {
gpa->ll = tal(gpa, enum log_level);
if (json_tok_streq(buffer, leveltok, "debug"))
*gpa->ll = LOG_DBG;
else if (json_tok_streq(buffer, leveltok, "info"))
*gpa->ll = LOG_INFORM;
else if (json_tok_streq(buffer, leveltok, "unusual"))
*gpa->ll = LOG_UNUSUAL;
else if (json_tok_streq(buffer, leveltok, "broken"))
*gpa->ll = LOG_BROKEN;
else {
command_fail(cmd, "Invalid level param");
return;
}
} else
gpa->ll = NULL;
/* Get peers from gossipd. */
subd_req(cmd, cmd->ld->gossip,
take(towire_gossip_getpeers_request(cmd)),
-1, 0, gossipd_getpeers_complete, gpa);
}
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 (!json_tok_pubkey(buffer, peeridtok, &peerid))
return NULL;
return peer_by_id(ld, &peerid);
}
struct funding_channel {
struct command *cmd; /* Which also owns us. */
/* Peer we're trying to reach. */
struct pubkey peerid;
/* Details of how to make funding. */
const struct utxo **utxomap;
u64 change;
u32 change_keyindex;
u64 funding_satoshi, push_msat;
/* Peer, once we have one. */
struct peer *peer;
/* 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)
{
peer_internal_error(peer, "Funding broadcast exited with %i: %s",
exitstatus, err);
}
static enum watch_result funding_announce_cb(struct peer *peer,
const struct bitcoin_tx *tx,
unsigned int depth,
void *unused)
{
if (depth < ANNOUNCE_MIN_DEPTH) {
return KEEP_WATCHING;
}
if (!peer->owner || !streq(peer->owner->name, "lightning_channeld")) {
log_debug(peer->ld->log,
"Funding tx announce ready, but peer is not owned by channeld");
return KEEP_WATCHING;
}
subd_send_msg(peer->owner,
take(towire_channel_funding_announce_depth(peer)));
return DELETE_WATCH;
}
/* We dump all the known preimages when onchaind starts up. */
static void onchaind_tell_fulfill(struct peer *peer)
{
struct htlc_in_map_iter ini;
struct htlc_in *hin;
u8 *msg;
for (hin = htlc_in_map_first(&peer->ld->htlcs_in, &ini);
hin;
hin = htlc_in_map_next(&peer->ld->htlcs_in, &ini)) {
if (hin->key.peer != peer)
continue;
/* BOLT #5:
*
* If the node receives (or already knows) a payment preimage
* for an unresolved HTLC output it was offered for which it
* has committed to an outgoing HTLC, it MUST *resolve* the
* output by spending it. Otherwise, if the other node is not
* irrevocably committed to the HTLC, it MUST NOT *resolve*
* the output by spending it.
*/
/* We only set preimage once it's irrevocably committed, and
* we spend even if we don't have an outgoing HTLC (eg. local
* payyment complete) */
if (!hin->preimage)
continue;
msg = towire_onchain_known_preimage(peer, hin->preimage);
subd_send_msg(peer->owner, take(msg));
}
}
static void handle_onchain_init_reply(struct peer *peer, const u8 *msg)
{
u8 state;
if (!fromwire_onchain_init_reply(msg, NULL, &state)) {
peer_internal_error(peer, "Invalid onchain_init_reply");
return;
}
if (!peer_state_on_chain(state)) {
peer_internal_error(peer,
"Invalid onchain_init_reply state %u (%s)",
state, peer_state_name(state));
return;
}
peer_set_condition(peer, FUNDING_SPEND_SEEN, state);
/* Tell it about any preimages we know. */
onchaind_tell_fulfill(peer);
}
static enum watch_result onchain_tx_watched(struct peer *peer,
const struct bitcoin_tx *tx,
unsigned int depth,
void *unused)
{
u8 *msg;
struct sha256_double txid;
if (depth == 0) {
log_unusual(peer->log, "Chain reorganization!");
peer_set_owner(peer, NULL);
/* FIXME!
topology_rescan(peer->ld->topology, peer->funding_txid);
*/
/* We will most likely be freed, so this is a noop */
return KEEP_WATCHING;
}
bitcoin_txid(tx, &txid);
msg = towire_onchain_depth(peer, &txid, depth);
subd_send_msg(peer->owner, take(msg));
return KEEP_WATCHING;
}
static void watch_tx_and_outputs(struct peer *peer,
const struct bitcoin_tx *tx);
static enum watch_result onchain_txo_watched(struct peer *peer,
const struct bitcoin_tx *tx,
size_t input_num,
const struct block *block,
void *unused)
{
u8 *msg;
watch_tx_and_outputs(peer, tx);
msg = towire_onchain_spent(peer, tx, input_num, block->height);
subd_send_msg(peer->owner, take(msg));
/* We don't need to keep watching: If this output is double-spent
* (reorg), we'll get a zero depth cb to onchain_tx_watched, and
* restart onchaind. */
return DELETE_WATCH;
}
/* To avoid races, we watch the tx and all outputs. */
static void watch_tx_and_outputs(struct peer *peer,
const struct bitcoin_tx *tx)
{
struct sha256_double txid;
struct txwatch *txw;
bitcoin_txid(tx, &txid);
/* Make txwatch a parent of txo watches, so we can unwatch together. */
txw = watch_tx(peer->owner, peer->ld->topology, peer, tx,
onchain_tx_watched, NULL);
for (size_t i = 0; i < tal_count(tx->output); i++)
watch_txo(txw, peer->ld->topology, peer, &txid, i,
onchain_txo_watched, NULL);
}
static void handle_onchain_broadcast_tx(struct peer *peer, const u8 *msg)
{
struct bitcoin_tx *tx;
tx = tal(msg, struct bitcoin_tx);
if (!fromwire_onchain_broadcast_tx(msg, NULL, tx)) {
peer_internal_error(peer, "Invalid onchain_broadcast_tx");
return;
}
/* We don't really care if it fails, we'll respond via watch. */
broadcast_tx(peer->ld->topology, peer, tx, NULL);
}
static void handle_onchain_unwatch_tx(struct peer *peer, const u8 *msg)
{
/* FIXME: unwatch tx and children here. */
}
static void handle_extracted_preimage(struct peer *peer, const u8 *msg)
{
struct preimage preimage;
if (!fromwire_onchain_extracted_preimage(msg, NULL, &preimage)) {
peer_internal_error(peer, "Invalid extracted_preimage");
return;
}
onchain_fulfilled_htlc(peer, &preimage);
}
static void handle_missing_htlc_output(struct peer *peer, const u8 *msg)
{
struct htlc_stub htlc;
if (!fromwire_onchain_missing_htlc_output(msg, NULL, &htlc)) {
peer_internal_error(peer, "Invalid missing_htlc_output");
return;
}
/* BOLT #5:
*
* For any committed HTLC which does not have an output in this
* commitment transaction, the node MUST fail the corresponding
* incoming HTLC (if any) once the commitment transaction has reached
* reasonable depth, and MAY fail it sooner if no valid commitment
* transaction contains an output corresponding to the HTLC.
*/
onchain_failed_our_htlc(peer, &htlc, "missing in commitment tx");
}
static void handle_onchain_htlc_timeout(struct peer *peer, const u8 *msg)
{
struct htlc_stub htlc;
if (!fromwire_onchain_htlc_timeout(msg, NULL, &htlc)) {
peer_internal_error(peer, "Invalid onchain_htlc_timeout");
return;
}
/* BOLT #5:
*
* If the HTLC output has *timed out* and not been *resolved*, the node
* MUST *resolve* the output and MUST fail the corresponding incoming
* HTLC (if any) once the resolving transaction has reached reasonable
* depth.
*/
onchain_failed_our_htlc(peer, &htlc, "timed out");
}
/* If peer is NULL, free them all (for shutdown) */
void free_htlcs(struct lightningd *ld, const struct peer *peer)
{
struct htlc_out_map_iter outi;
struct htlc_out *hout;
struct htlc_in_map_iter ini;
struct htlc_in *hin;
bool deleted;
/* FIXME: Implement check_htlcs to ensure no dangling hout->in ptrs! */
do {
deleted = false;
for (hout = htlc_out_map_first(&ld->htlcs_out, &outi);
hout;
hout = htlc_out_map_next(&ld->htlcs_out, &outi)) {
if (peer && hout->key.peer != peer)
continue;
tal_free(hout);
deleted = true;
}
for (hin = htlc_in_map_first(&ld->htlcs_in, &ini);
hin;
hin = htlc_in_map_next(&ld->htlcs_in, &ini)) {
if (peer && hin->key.peer != peer)
continue;
tal_free(hin);
deleted = true;
}
/* Can skip over elements due to iterating while deleting. */
} while (deleted);
}
static void handle_irrevocably_resolved(struct peer *peer, const u8 *msg)
{
/* FIXME: Implement check_htlcs to ensure no dangling hout->in ptrs! */
free_htlcs(peer->ld, peer);
/* FIXME: Remove peer from db. */
log_info(peer->log, "onchaind complete, forgetting peer");
/* This will also free onchaind. */
tal_free(peer);
}
static unsigned int onchain_msg(struct subd *sd, const u8 *msg, const int *fds)
{
enum onchain_wire_type t = fromwire_peektype(msg);
switch (t) {
case WIRE_ONCHAIN_INIT_REPLY:
handle_onchain_init_reply(sd->peer, msg);
break;
case WIRE_ONCHAIN_BROADCAST_TX:
handle_onchain_broadcast_tx(sd->peer, msg);
break;
case WIRE_ONCHAIN_UNWATCH_TX:
handle_onchain_unwatch_tx(sd->peer, msg);
break;
case WIRE_ONCHAIN_EXTRACTED_PREIMAGE:
handle_extracted_preimage(sd->peer, msg);
break;
case WIRE_ONCHAIN_MISSING_HTLC_OUTPUT:
handle_missing_htlc_output(sd->peer, msg);
break;
case WIRE_ONCHAIN_HTLC_TIMEOUT:
handle_onchain_htlc_timeout(sd->peer, msg);
break;
case WIRE_ONCHAIN_ALL_IRREVOCABLY_RESOLVED:
handle_irrevocably_resolved(sd->peer, msg);
break;
/* We send these, not receive them */
case WIRE_ONCHAIN_INIT:
case WIRE_ONCHAIN_SPENT:
case WIRE_ONCHAIN_DEPTH:
case WIRE_ONCHAIN_HTLC:
case WIRE_ONCHAIN_KNOWN_PREIMAGE:
break;
}
return 0;
}
static u8 *p2wpkh_for_keyidx(const tal_t *ctx, struct lightningd *ld, u64 keyidx)
{
struct pubkey shutdownkey;
if (!bip32_pubkey(ld->wallet->bip32_base, &shutdownkey, keyidx))
return NULL;
return scriptpubkey_p2wpkh(ctx, &shutdownkey);
}
/* If we want to know if this HTLC is missing, return depth. */
static bool tell_if_missing(const struct peer *peer, struct htlc_stub *stub,
bool *tell_immediate)
{
struct htlc_out *hout;
/* Keep valgrind happy. */
*tell_immediate = false;
/* Is it a current HTLC? */
hout = find_htlc_out_by_ripemd(peer, &stub->ripemd);
if (!hout)
return false;
/* BOLT #5:
*
* For any committed HTLC which does not have an output in this
* commitment transaction, the node MUST fail the corresponding
* incoming HTLC (if any) once the commitment transaction has reached
* reasonable depth, and MAY fail it sooner if no valid commitment
* transaction contains an output corresponding to the HTLC.
*/
if (hout->hstate >= RCVD_ADD_REVOCATION
&& hout->hstate < SENT_REMOVE_REVOCATION)
*tell_immediate = true;
log_debug(peer->log, "We want to know if htlc %"PRIu64" is missing (%s)",
hout->key.id, *tell_immediate ? "immediate" : "later");
return true;
}
/* With a reorg, this can get called multiple times; each time we'll kill
* onchaind (like any other owner), and restart */
static enum watch_result funding_spent(struct peer *peer,
const struct bitcoin_tx *tx,
size_t input_num,
const struct block *block,
void *unused)
{
u8 *msg, *scriptpubkey;
struct sha256_double our_last_txid;
s64 keyindex;
struct pubkey ourkey;
struct htlc_stub *stubs;
peer_fail_permanent_str(peer, "Funding transaction spent");
/* We could come from almost any state. */
peer_set_condition(peer, peer->state, FUNDING_SPEND_SEEN);
peer_set_owner(peer, new_peer_subd(peer->ld,
"lightning_onchaind", peer,
onchain_wire_type_name,
onchain_msg,
NULL));
if (!peer->owner) {
log_broken(peer->log, "Could not subdaemon onchain: %s",
strerror(errno));
return KEEP_WATCHING;
}
stubs = wallet_htlc_stubs(peer, peer->ld->wallet, peer->channel);
if (!stubs) {
log_broken(peer->log, "Could not load htlc_stubs");
return KEEP_WATCHING;
}
/* We re-use this key to send other outputs to. */
if (peer->local_shutdown_idx >= 0)
keyindex = peer->local_shutdown_idx;
else {
/* FIXME: Save to db */
keyindex = wallet_get_newindex(peer->ld);
if (keyindex < 0) {
log_broken(peer->log, "Could not get keyindex");
return KEEP_WATCHING;
}
}
scriptpubkey = p2wpkh_for_keyidx(peer, peer->ld, keyindex);
if (!scriptpubkey) {
peer_internal_error(peer,
"Can't get shutdown script %"PRIu64,
keyindex);
return DELETE_WATCH;
}
if (!bip32_pubkey(peer->ld->wallet->bip32_base, &ourkey, keyindex)) {
peer_internal_error(peer,
"Can't get shutdown key %"PRIu64,
keyindex);
return DELETE_WATCH;
}
/* This could be a mutual close, but it doesn't matter. */
bitcoin_txid(peer->last_tx, &our_last_txid);
msg = towire_onchain_init(peer, peer->seed, &peer->their_shachain.chain,
peer->funding_satoshi,
&peer->channel_info->old_remote_per_commit,
&peer->channel_info->remote_per_commit,
peer->our_config.to_self_delay,
peer->channel_info->their_config.to_self_delay,
get_feerate(peer->ld->topology,
FEERATE_NORMAL),
peer->our_config.dust_limit_satoshis,
&peer->channel_info->theirbase.revocation,
&our_last_txid,
scriptpubkey,
peer->remote_shutdown_scriptpubkey,
&ourkey,
peer->funder,
&peer->channel_info->theirbase.payment,
&peer->channel_info->theirbase.htlc,
&peer->channel_info->theirbase.delayed_payment,
tx,
block->height,
/* FIXME: config for 'reasonable depth' */
3,
peer->last_htlc_sigs,
tal_count(stubs));
subd_send_msg(peer->owner, take(msg));
/* FIXME: Don't queue all at once, use an empty cb... */
for (size_t i = 0; i < tal_count(stubs); i++) {
bool tell_immediate;
bool tell = tell_if_missing(peer, &stubs[i], &tell_immediate);
msg = towire_onchain_htlc(peer, &stubs[i],
tell, tell_immediate);
subd_send_msg(peer->owner, take(msg));
}
watch_tx_and_outputs(peer, tx);
/* We keep watching until peer finally deleted, for reorgs. */
return KEEP_WATCHING;
}
static enum watch_result funding_lockin_cb(struct peer *peer,
const struct bitcoin_tx *tx,
unsigned int depth,
void *unused)
{
struct sha256_double txid;
const char *txidstr;
struct txlocator *loc;
bool peer_ready;
bitcoin_txid(tx, &txid);
txidstr = type_to_string(peer, struct sha256_double, &txid);
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;
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);
/* In theory, it could have been buried before we got back
* from accepting openingd or disconnected: just wait for next one. */
peer_ready = (peer->owner && peer->state == CHANNELD_AWAITING_LOCKIN);
if (!peer_ready) {
log_unusual(peer->log,
"Funding tx confirmed, but peer state %s %s",
peer_state_name(peer->state),
peer->owner ? peer->owner->name : "unowned");
} else {
subd_send_msg(peer->owner,
take(towire_channel_funding_locked(peer,
peer->scid)));
}
/* BOLT #7:
*
* If the `open_channel` message had the `announce_channel` bit set,
* then both nodes must send the `announcement_signatures` message,
* otherwise they MUST NOT.
*/
if (!(peer->channel_flags & CHANNEL_FLAGS_ANNOUNCE_CHANNEL))
return DELETE_WATCH;
/* Tell channeld that we have reached the announce_depth and
* that it may send the announcement_signatures upon receiving
* funding_locked, or right now if it already received it
* before. If we are at the right depth, call the callback
* directly, otherwise schedule a callback */
if (depth >= ANNOUNCE_MIN_DEPTH)
funding_announce_cb(peer, tx, depth, NULL);
else
watch_txid(peer, peer->ld->topology, peer, &txid,
funding_announce_cb, NULL);
return DELETE_WATCH;
}
static void opening_got_hsm_funding_sig(struct funding_channel *fc,
int peer_fd, int gossip_fd,
const u8 *resp,
const struct crypto_state *cs)
{
secp256k1_ecdsa_signature *sigs;
struct bitcoin_tx *tx = fc->funding_tx;
u8 *linear;
u64 change_satoshi;
struct json_result *response = new_json_result(fc->cmd);
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->wallet->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(fc->peer->ld->topology, fc->peer, tx, funding_broadcast_failed);
watch_tx(fc->peer, fc->peer->ld->topology, fc->peer, tx,
funding_lockin_cb, NULL);
/* Extract the change output and add it to the DB */
wallet_extract_owned_outputs(fc->peer->ld->wallet, tx, &change_satoshi);
/* FIXME: Remove arg from cb? */
watch_txo(fc->peer, fc->peer->ld->topology, fc->peer,
fc->peer->funding_txid, fc->peer->funding_outnum,
funding_spent, NULL);
/* We could defer until after funding locked, but makes testing
* harder. */
tal_del_destructor(fc, fail_fundchannel_command);
json_object_start(response, NULL);
linear = linearize_tx(response, tx);
json_add_hex(response, "tx", linear, tal_len(linear));
json_object_end(response);
command_success(fc->cmd, response);
/* Start normal channel daemon. */
peer_start_channeld(fc->peer, cs, peer_fd, gossip_fd, NULL, false);
peer_set_condition(fc->peer, OPENINGD, CHANNELD_AWAITING_LOCKIN);
wallet_confirm_utxos(fc->peer->ld->wallet, fc->utxomap);
tal_free(fc);
}
/* 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. This is typically called twice: once with the dummy
* signature to get it signed and a second time to build the full
* packet with the signature. The timestamp is handed in since that is
* the only thing that may change between the dummy creation and the
* call with a signature.*/
static u8 *create_node_announcement(const tal_t *ctx, struct lightningd *ld,
secp256k1_ecdsa_signature *sig,
u32 timestamp)
{
u8 *features = NULL;
u8 *addresses = tal_arr(ctx, u8, 0);
u8 *announcement;
size_t i;
if (!sig) {
sig = tal(ctx, secp256k1_ecdsa_signature);
memset(sig, 0, sizeof(*sig));
}
for (i = 0; i < tal_count(ld->wireaddrs); i++)
towire_wireaddr(&addresses, ld->wireaddrs+i);
announcement =
towire_node_announcement(ctx, sig, features, timestamp,
&ld->id, ld->rgb, (u8 *)ld->alias,
addresses);
return announcement;
}
/* 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_announce(struct peer *peer, const u8 *msg)
{
struct lightningd *ld = peer->ld;
tal_t *tmpctx = tal_tmpctx(peer);
secp256k1_ecdsa_signature sig;
u8 *cannounce, *cupdate;
u8 *announcement, *wrappedmsg;
u32 timestamp = time_now().ts.tv_sec;
if (!fromwire_channel_announce(msg, msg, NULL, &cannounce, &cupdate)) {
peer_internal_error(peer, "bad fromwire_channel_announced %s",
tal_hex(peer, msg));
return;
}
msg = towire_hsmctl_node_announcement_sig_req(
tmpctx, create_node_announcement(tmpctx, ld, NULL, timestamp));
if (!wire_sync_write(ld->hsm_fd, take(msg)))
fatal("Could not write to HSM: %s", strerror(errno));
msg = hsm_sync_read(tmpctx, ld);
if (!fromwire_hsmctl_node_announcement_sig_reply(msg, NULL, &sig))
fatal("HSM returned an invalid node_announcement sig");
/* 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. */
announcement = create_node_announcement(tmpctx, ld, &sig, timestamp);
/* We have to send channel_announce before channel_update and
* node_announcement */
wrappedmsg = towire_gossip_forwarded_msg(tmpctx, cannounce);
subd_send_msg(ld->gossip, take(wrappedmsg));
wrappedmsg = towire_gossip_forwarded_msg(tmpctx, cupdate);
subd_send_msg(ld->gossip, take(wrappedmsg));
wrappedmsg = towire_gossip_forwarded_msg(tmpctx, announcement);
subd_send_msg(ld->gossip, take(wrappedmsg));
tal_free(tmpctx);
}
static void peer_got_funding_locked(struct peer *peer, const u8 *msg)
{
struct pubkey next_per_commitment_point;
if (!fromwire_channel_got_funding_locked(msg, NULL,
&next_per_commitment_point)) {
peer_internal_error(peer, "bad channel_got_funding_locked %s",
tal_hex(peer, msg));
return;
}
if (peer->remote_funding_locked) {
peer_internal_error(peer, "channel_got_funding_locked twice");
return;
}
update_per_commit_point(peer, &next_per_commitment_point);
log_debug(peer->log, "Got funding_locked");
peer->remote_funding_locked = true;
}
static void peer_got_shutdown(struct peer *peer, const u8 *msg)
{
u8 *scriptpubkey;
if (!fromwire_channel_got_shutdown(peer, msg, NULL, &scriptpubkey)) {
peer_internal_error(peer, "bad channel_got_shutdown %s",
tal_hex(peer, msg));
return;
}
/* FIXME: Add to spec that we must allow repeated shutdown! */
peer->remote_shutdown_scriptpubkey
= tal_free(peer->remote_shutdown_scriptpubkey);
peer->remote_shutdown_scriptpubkey = scriptpubkey;
/* BOLT #2:
*
* A sending node MUST set `scriptpubkey` to one of the following forms:
*
* 1. `OP_DUP` `OP_HASH160` `20` 20-bytes `OP_EQUALVERIFY` `OP_CHECKSIG`
* (pay to pubkey hash), OR
* 2. `OP_HASH160` `20` 20-bytes `OP_EQUAL` (pay to script hash), OR
* 3. `OP_0` `20` 20-bytes (version 0 pay to witness pubkey), OR
* 4. `OP_0` `32` 32-bytes (version 0 pay to witness script hash)
*
* A receiving node SHOULD fail the connection if the `scriptpubkey`
* is not one of those forms. */
if (!is_p2pkh(scriptpubkey, NULL) && !is_p2sh(scriptpubkey, NULL)
&& !is_p2wpkh(scriptpubkey, NULL) && !is_p2wsh(scriptpubkey, NULL)) {
char *str = tal_fmt(peer, "Bad shutdown scriptpubkey %s",
tal_hex(peer, scriptpubkey));
peer_fail_permanent_str(peer, take(str));
return;
}
if (peer->local_shutdown_idx == -1) {
u8 *scriptpubkey;
peer->local_shutdown_idx = wallet_get_newindex(peer->ld);
if (peer->local_shutdown_idx == -1) {
peer_internal_error(peer,
"Can't get local shutdown index");
return;
}
peer_set_condition(peer, CHANNELD_NORMAL, CHANNELD_SHUTTING_DOWN);
/* BOLT #2:
*
* A sending node MUST set `scriptpubkey` to one of the
* following forms:
*
* ...3. `OP_0` `20` 20-bytes (version 0 pay to witness pubkey),
*/
scriptpubkey = p2wpkh_for_keyidx(msg, peer->ld,
peer->local_shutdown_idx);
if (!scriptpubkey) {
peer_internal_error(peer,
"Can't get shutdown script %"PRIu64,
peer->local_shutdown_idx);
return;
}
/* BOLT #2:
*
* A receiving node MUST reply to a `shutdown` message with a
* `shutdown` once there are no outstanding updates on the
* peer, unless it has already sent a `shutdown`.
*/
subd_send_msg(peer->owner,
take(towire_channel_send_shutdown(peer,
scriptpubkey)));
}
/* TODO(cdecker) Selectively save updated fields to DB */
wallet_channel_save(peer->ld->wallet, peer->channel);
}
void peer_last_tx(struct peer *peer, struct bitcoin_tx *tx,
const secp256k1_ecdsa_signature *sig)
{
/* FIXME: save to db. */
tal_free(peer->last_sig);
peer->last_sig = tal_dup(peer, secp256k1_ecdsa_signature, sig);
tal_free(peer->last_tx);
peer->last_tx = tal_steal(peer, tx);
}
/* Is this better than the last tx we were holding? */
static bool better_closing_fee(struct peer *peer, const struct bitcoin_tx *tx)
{
u64 weight, fee, last_fee, ideal_fee, min_fee;
s64 old_diff, new_diff;
size_t i;
/* Calculate actual fee. */
fee = peer->funding_satoshi;
for (i = 0; i < tal_count(tx->output); i++)
fee -= tx->output[i].amount;
last_fee = peer->funding_satoshi;
for (i = 0; i < tal_count(peer->last_tx); i++)
last_fee -= peer->last_tx->output[i].amount;
/* Weight once we add in sigs. */
weight = measure_tx_cost(tx) + 74 * 2;
min_fee = get_feerate(peer->ld->topology, FEERATE_SLOW) * weight / 1000;
if (fee < min_fee)
return false;
ideal_fee = get_feerate(peer->ld->topology, FEERATE_NORMAL) * weight / 1000;
/* We prefer fee which is closest to our ideal. */
old_diff = imaxabs((s64)ideal_fee - (s64)last_fee);
new_diff = imaxabs((s64)ideal_fee - (s64)fee);
return (new_diff < old_diff);
}
static void peer_received_closing_signature(struct peer *peer, const u8 *msg)
{
secp256k1_ecdsa_signature sig;
struct bitcoin_tx *tx = tal(msg, struct bitcoin_tx);
if (!fromwire_closing_received_signature(msg, NULL, &sig, tx)) {
peer_internal_error(peer, "Bad closing_received_signature %s",
tal_hex(peer, msg));
return;
}
/* FIXME: Make sure signature is correct! */
if (better_closing_fee(peer, tx)) {
/* TODO(cdecker) Selectively save updated fields to DB */
wallet_channel_save(peer->ld->wallet, peer->channel);
peer_last_tx(peer, tx, &sig);
}
/* OK, you can continue now. */
subd_send_msg(peer->owner,
take(towire_closing_received_signature_reply(peer)));
}
static void peer_closing_complete(struct peer *peer, const u8 *msg)
{
if (!fromwire_closing_complete(msg, NULL)) {
peer_internal_error(peer, "Bad closing_complete %s",
tal_hex(peer, msg));
return;
}
/* Retransmission only, ignore closing. */
if (peer->state == CLOSINGD_COMPLETE)
return;
drop_to_chain(peer);
peer_set_condition(peer, CLOSINGD_SIGEXCHANGE, CLOSINGD_COMPLETE);
}
static unsigned closing_msg(struct subd *sd, const u8 *msg, const int *fds)
{
enum closing_wire_type t = fromwire_peektype(msg);
switch (t) {
case WIRE_CLOSING_RECEIVED_SIGNATURE:
peer_received_closing_signature(sd->peer, msg);
break;
case WIRE_CLOSING_COMPLETE:
peer_closing_complete(sd->peer, msg);
break;
/* We send these, not receive them */
case WIRE_CLOSING_INIT:
case WIRE_CLOSING_RECEIVED_SIGNATURE_REPLY:
break;
}
return 0;
}
static void peer_start_closingd(struct peer *peer,
struct crypto_state *cs,
int peer_fd, int gossip_fd,
bool reconnected)
{
const tal_t *tmpctx = tal_tmpctx(peer);
u8 *initmsg, *local_scriptpubkey;
u64 minfee, maxfee, startfee, feelimit;
u64 num_revocations;
if (peer->local_shutdown_idx == -1
|| !peer->remote_shutdown_scriptpubkey) {
peer_internal_error(peer,
"Can't start closing: local %s remote %s",
peer->local_shutdown_idx == -1
? "not shutdown" : "shutdown",
peer->remote_shutdown_scriptpubkey
? "shutdown" : "not shutdown");
tal_free(tmpctx);
return;
}
peer_set_owner(peer, new_peer_subd(peer->ld,
"lightning_closingd", peer,
closing_wire_type_name, closing_msg,
take(&peer_fd), take(&gossip_fd),
NULL));
if (!peer->owner) {
log_unusual(peer->log, "Could not subdaemon closing: %s",
strerror(errno));
peer_fail_transient(peer, "Failed to subdaemon closing");
tal_free(tmpctx);
return;
}
local_scriptpubkey = p2wpkh_for_keyidx(tmpctx, peer->ld,
peer->local_shutdown_idx);
if (!local_scriptpubkey) {
peer_internal_error(peer,
"Can't generate local shutdown scriptpubkey");
tal_free(tmpctx);
return;
}
/* BOLT #2:
*
* A sending node MUST set `fee_satoshis` lower than or equal
* to the base fee of the final commitment transaction as
* calculated in [BOLT
* #3](03-transactions.md#fee-calculation).
*/
feelimit = commit_tx_base_fee(peer->channel_info->feerate_per_kw[LOCAL],
0);
maxfee = commit_tx_base_fee(get_feerate(peer->ld->topology,
FEERATE_IMMEDIATE), 0);
minfee = commit_tx_base_fee(get_feerate(peer->ld->topology,
FEERATE_SLOW), 0);
startfee = commit_tx_base_fee(get_feerate(peer->ld->topology,
FEERATE_NORMAL), 0);
if (maxfee > feelimit)
maxfee = feelimit;
if (startfee > feelimit)
startfee = feelimit;
if (minfee > feelimit)
minfee = feelimit;
num_revocations
= revocations_received(&peer->their_shachain.chain);
/* BOLT #3:
*
* The amounts for each output MUST BE rounded down to whole satoshis.
*/
initmsg = towire_closing_init(tmpctx,
cs,
peer->seed,
peer->funding_txid,
peer->funding_outnum,
peer->funding_satoshi,
&peer->channel_info->remote_fundingkey,
peer->funder,
*peer->our_msatoshi / 1000,
peer->funding_satoshi
- *peer->our_msatoshi / 1000,
peer->our_config.dust_limit_satoshis,
minfee, maxfee, startfee,
local_scriptpubkey,
peer->remote_shutdown_scriptpubkey,
reconnected,
peer->next_index[LOCAL],
peer->next_index[REMOTE],
num_revocations);
/* We don't expect a response: it will give us feedback on
* signatures sent and received, then closing_complete. */
subd_send_msg(peer->owner, take(initmsg));
tal_free(tmpctx);
}
static void peer_start_closingd_after_shutdown(struct peer *peer, const u8 *msg,
const int *fds)
{
struct crypto_state cs;
/* We expect 2 fds. */
assert(tal_count(fds) == 2);
if (!fromwire_channel_shutdown_complete(msg, NULL, &cs)) {
peer_internal_error(peer, "bad shutdown_complete: %s",
tal_hex(peer, msg));
return;
}
/* This sets peer->owner, closes down channeld. */
peer_start_closingd(peer, &cs, fds[0], fds[1], false);
peer_set_condition(peer, CHANNELD_SHUTTING_DOWN, CLOSINGD_SIGEXCHANGE);
}
static unsigned channel_msg(struct subd *sd, const u8 *msg, const int *fds)
{
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_SENDING_COMMITSIG:
peer_sending_commitsig(sd->peer, msg);
break;
case WIRE_CHANNEL_GOT_COMMITSIG:
peer_got_commitsig(sd->peer, msg);
break;
case WIRE_CHANNEL_GOT_REVOKE:
peer_got_revoke(sd->peer, msg);
break;
case WIRE_CHANNEL_ANNOUNCE:
peer_channel_announce(sd->peer, msg);
break;
case WIRE_CHANNEL_GOT_FUNDING_LOCKED:
peer_got_funding_locked(sd->peer, msg);
break;
case WIRE_CHANNEL_GOT_SHUTDOWN:
peer_got_shutdown(sd->peer, msg);
break;
case WIRE_CHANNEL_SHUTDOWN_COMPLETE:
/* We expect 2 fds. */
if (!fds)
return 2;
peer_start_closingd_after_shutdown(sd->peer, msg, fds);
break;
/* 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:
case WIRE_CHANNEL_GOT_COMMITSIG_REPLY:
case WIRE_CHANNEL_GOT_REVOKE_REPLY:
case WIRE_CHANNEL_SENDING_COMMITSIG_REPLY:
case WIRE_CHANNEL_SEND_SHUTDOWN:
case WIRE_CHANNEL_DEV_REENABLE_COMMIT:
case WIRE_CHANNEL_FEERATES:
/* Replies go to requests. */
case WIRE_CHANNEL_OFFER_HTLC_REPLY:
case WIRE_CHANNEL_PING_REPLY:
case WIRE_CHANNEL_DEV_REENABLE_COMMIT_REPLY:
break;
}
return 0;
}
static bool peer_start_channeld(struct peer *peer,
const struct crypto_state *cs,
int peer_fd, int gossip_fd,
const u8 *funding_signed,
bool reconnected)
{
const tal_t *tmpctx = tal_tmpctx(peer);
u8 *msg, *initmsg;
int hsmfd;
const struct config *cfg = &peer->ld->config;
struct added_htlc *htlcs;
enum htlc_state *htlc_states;
struct fulfilled_htlc *fulfilled_htlcs;
enum side *fulfilled_sides;
struct failed_htlc *failed_htlcs;
enum side *failed_sides;
struct short_channel_id funding_channel_id;
const u8 *shutdown_scriptpubkey;
u64 num_revocations;
/* Now we can consider balance set. */
if (!reconnected) {
assert(!peer->our_msatoshi);
peer->our_msatoshi = tal(peer, u64);
if (peer->funder == LOCAL)
*peer->our_msatoshi
= peer->funding_satoshi * 1000 - peer->push_msat;
else
*peer->our_msatoshi = peer->push_msat;
} else
assert(peer->our_msatoshi);
msg = towire_hsmctl_client_hsmfd(tmpctx, &peer->id, HSM_CAP_SIGN_GOSSIP | HSM_CAP_ECDH);
if (!wire_sync_write(peer->ld->hsm_fd, take(msg)))
fatal("Could not write to HSM: %s", strerror(errno));
msg = hsm_sync_read(tmpctx, peer->ld);
if (!fromwire_hsmctl_client_hsmfd_reply(msg, NULL))
fatal("Bad reply from HSM: %s", tal_hex(tmpctx, msg));
hsmfd = fdpass_recv(peer->ld->hsm_fd);
if (hsmfd < 0)
fatal("Could not read fd from HSM: %s", strerror(errno));
peer_set_owner(peer, new_peer_subd(peer->ld,
"lightning_channeld", peer,
channel_wire_type_name,
channel_msg,
take(&peer_fd),
take(&gossip_fd),
take(&hsmfd), NULL));
if (!peer->owner) {
log_unusual(peer->log, "Could not subdaemon channel: %s",
strerror(errno));
peer_fail_transient(peer, "Failed to subdaemon channel");
tal_free(tmpctx);
return true;
}
peer_htlcs(tmpctx, peer, &htlcs, &htlc_states, &fulfilled_htlcs,
&fulfilled_sides, &failed_htlcs, &failed_sides);
if (peer->scid) {
funding_channel_id = *peer->scid;
log_debug(peer->log, "Already have funding locked in");
} else {
log_debug(peer->log, "Waiting for funding confirmations");
memset(&funding_channel_id, 0, sizeof(funding_channel_id));
}
if (peer->local_shutdown_idx != -1) {
shutdown_scriptpubkey
= p2wpkh_for_keyidx(tmpctx, peer->ld,
peer->local_shutdown_idx);
} else
shutdown_scriptpubkey = NULL;
num_revocations = revocations_received(&peer->their_shachain.chain);
initmsg = towire_channel_init(tmpctx,
&get_chainparams(peer->ld)
->genesis_blockhash,
peer->funding_txid,
peer->funding_outnum,
peer->funding_satoshi,
&peer->our_config,
&peer->channel_info->their_config,
peer->channel_info->feerate_per_kw,
get_feerate(peer->ld->topology, FEERATE_NORMAL),
get_feerate(peer->ld->topology, FEERATE_IMMEDIATE) * 5,
peer->last_sig,
cs,
&peer->channel_info->remote_fundingkey,
&peer->channel_info->theirbase.revocation,
&peer->channel_info->theirbase.payment,
&peer->channel_info->theirbase.htlc,
&peer->channel_info->theirbase.delayed_payment,
&peer->channel_info->remote_per_commit,
&peer->channel_info->old_remote_per_commit,
peer->funder,
cfg->fee_base,
cfg->fee_per_satoshi,
*peer->our_msatoshi,
peer->seed,
&peer->ld->id,
&peer->id,
time_to_msec(cfg->commit_time),
cfg->cltv_expiry_delta,
peer->last_was_revoke,
peer->last_sent_commit,
peer->next_index[LOCAL],
peer->next_index[REMOTE],
num_revocations,
peer->next_htlc_id,
htlcs, htlc_states,
fulfilled_htlcs, fulfilled_sides,
failed_htlcs, failed_sides,
peer->scid != NULL,
peer->remote_funding_locked,
&funding_channel_id,
reconnected,
shutdown_scriptpubkey,
peer->remote_shutdown_scriptpubkey != NULL,
peer->channel_flags,
funding_signed);
/* We don't expect a response: we are triggered by funding_depth_cb. */
subd_send_msg(peer->owner, take(initmsg));
tal_free(tmpctx);
return true;
}
static bool peer_commit_initial(struct peer *peer)
{
peer->next_index[LOCAL] = peer->next_index[REMOTE] = 1;
return true;
}
static void 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;
struct crypto_state cs;
secp256k1_ecdsa_signature remote_commit_sig;
struct bitcoin_tx *remote_commit;
assert(tal_count(fds) == 2);
/* At this point, we care about peer */
fc->peer->channel_info = channel_info
= tal(fc->peer, struct channel_info);
remote_commit = tal(resp, struct bitcoin_tx);
/* This is a new channel_info->their_config so set its ID to 0 */
fc->peer->channel_info->their_config.id = 0;
if (!fromwire_opening_funder_reply(resp, NULL,
&channel_info->their_config,
remote_commit,
&remote_commit_sig,
&cs,
&channel_info->theirbase.revocation,
&channel_info->theirbase.payment,
&channel_info->theirbase.htlc,
&channel_info->theirbase.delayed_payment,
&channel_info->remote_per_commit,
&fc->peer->minimum_depth,
&channel_info->remote_fundingkey,
&funding_txid,
&channel_info->feerate_per_kw[REMOTE])) {
peer_internal_error(fc->peer, "bad funder_reply: %s",
tal_hex(resp, resp));
return;
}
/* Feerates begin identical. */
channel_info->feerate_per_kw[LOCAL]
= channel_info->feerate_per_kw[REMOTE];
/* old_remote_per_commit not valid yet, copy valid one. */
channel_info->old_remote_per_commit = channel_info->remote_per_commit;
/* Now, keep the initial commit as our last-tx-to-broadast. */
peer_last_tx(fc->peer, remote_commit, &remote_commit_sig);
/* Generate the funding tx. */
if (fc->change
&& !bip32_pubkey(fc->peer->ld->wallet->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);
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->wallet->bip32_base);
log_debug(fc->peer->log, "Funding tx has %zi inputs, %zu outputs:",
tal_count(fc->funding_tx->input),
tal_count(fc->funding_tx->output));
for (size_t i = 0; i < tal_count(fc->funding_tx->input); i++) {
log_debug(fc->peer->log, "%zi: %"PRIu64" satoshi (%s) %s\n",
i, fc->utxomap[i]->amount,
fc->utxomap[i]->is_p2sh ? "P2SH" : "SEGWIT",
type_to_string(ltmp, struct sha256_double,
&fc->funding_tx->input[i].txid));
}
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_internal_error(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;
}
if (!peer_commit_initial(fc->peer)) {
peer_internal_error(fc->peer, "Initial peer to db failed");
return;
}
/* 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);
/* Unowned (will free openingd). */
peer_set_owner(fc->peer, NULL);
if (!wire_sync_write(fc->peer->ld->hsm_fd, take(msg)))
fatal("Could not write to HSM: %s", strerror(errno));
msg = hsm_sync_read(fc, fc->peer->ld);
opening_got_hsm_funding_sig(fc, fds[0], fds[1], msg, &cs);
}
static void opening_fundee_finished(struct subd *opening,
const u8 *reply,
const int *fds,
struct peer *peer)
{
u8 *funding_signed;
struct channel_info *channel_info;
struct crypto_state cs;
secp256k1_ecdsa_signature remote_commit_sig;
struct bitcoin_tx *remote_commit;
log_debug(peer->log, "Got opening_fundee_finish_response");
assert(tal_count(fds) == 2);
remote_commit = tal(reply, struct bitcoin_tx);
/* At this point, we care about peer */
peer->channel_info = channel_info = tal(peer, struct channel_info);
/* This is a new channel_info->their_config, set its ID to 0 */
peer->channel_info->their_config.id = 0;
peer->funding_txid = tal(peer, struct sha256_double);
if (!fromwire_opening_fundee_reply(peer, reply, NULL,
&channel_info->their_config,
remote_commit,
&remote_commit_sig,
&cs,
&channel_info->theirbase.revocation,
&channel_info->theirbase.payment,
&channel_info->theirbase.htlc,
&channel_info->theirbase.delayed_payment,
&channel_info->remote_per_commit,
&channel_info->remote_fundingkey,
peer->funding_txid,
&peer->funding_outnum,
&peer->funding_satoshi,
&peer->push_msat,
&peer->channel_flags,
&channel_info->feerate_per_kw[REMOTE],
&funding_signed)) {
peer_internal_error(peer, "bad OPENING_FUNDEE_REPLY %s",
tal_hex(reply, reply));
return;
}
/* Feerates begin identical. */
channel_info->feerate_per_kw[LOCAL]
= channel_info->feerate_per_kw[REMOTE];
/* old_remote_per_commit not valid yet, copy valid one. */
channel_info->old_remote_per_commit = channel_info->remote_per_commit;
/* Now, keep the initial commit as our last-tx-to-broadast. */
peer_last_tx(peer, remote_commit, &remote_commit_sig);
if (!peer_commit_initial(peer))
return;
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 arg from cb? */
watch_txo(peer, peer->ld->topology, peer, peer->funding_txid,
peer->funding_outnum, funding_spent, NULL);
/* Unowned (will free openingd). */
peer_set_owner(peer, NULL);
/* On to normal operation! */
peer_start_channeld(peer, &cs, fds[0], fds[1], funding_signed, false);
peer_set_condition(peer, OPENINGD, CHANNELD_AWAITING_LOCKIN);
}
/* Peer has spontaneously exited from gossip due to open msg */
static void peer_accept_channel(struct lightningd *ld,
const struct pubkey *peer_id,
const struct wireaddr *addr,
const struct crypto_state *cs,
const u8 *gfeatures, const u8 *lfeatures,
int peer_fd, int gossip_fd,
const u8 *open_msg)
{
u32 max_to_self_delay, max_minimum_depth;
u64 min_effective_htlc_capacity_msat;
u8 *msg;
struct peer *peer;
assert(fromwire_peektype(open_msg) == WIRE_OPEN_CHANNEL);
/* We make a new peer. */
peer = new_peer(ld, peer_id, addr, gfeatures, lfeatures, peer_fd);
peer_set_condition(peer, UNINITIALIZED, OPENINGD);
peer_set_owner(peer,
new_peer_subd(ld, "lightning_openingd", peer,
opening_wire_type_name, NULL,
take(&peer_fd), take(&gossip_fd), NULL));
if (!peer->owner) {
peer_fail_transient(peer, "Failed to subdaemon opening: %s",
strerror(errno));
return;
}
/* They will open channel. */
peer->funder = REMOTE;
/* BOLT #2:
*
* The sender SHOULD set `minimum_depth` to a number of blocks it
* considers reasonable to avoid double-spending of the funding
* transaction.
*/
peer->minimum_depth = ld->config.anchor_confirms;
channel_config(ld, &peer->our_config,
&max_to_self_delay, &max_minimum_depth,
&min_effective_htlc_capacity_msat);
/* Store the channel in the database in order to get a channel
* ID that is unique and which we can base the peer_seed on */
peer->channel = peer_channel_new(ld->wallet, peer);
wallet_channel_save(peer->ld->wallet, peer->channel);
peer->seed = tal(peer, struct privkey);
derive_peer_seed(ld, peer->seed, &peer->id, peer->channel->id);
msg = towire_opening_init(peer, get_chainparams(ld)->index,
&peer->our_config,
max_to_self_delay,
min_effective_htlc_capacity_msat,
cs, peer->seed);
subd_send_msg(peer->owner, take(msg));
/* BOLT #2:
*
* Given the variance in fees, and the fact that the transaction may
* be spent in the future, it's a good idea for the fee payer to keep
* a good margin, say 5x the expected fee requirement */
msg = towire_opening_fundee(peer, peer->minimum_depth,
get_feerate(ld->topology, FEERATE_SLOW),
get_feerate(ld->topology, FEERATE_IMMEDIATE)
* 5,
open_msg);
subd_req(peer, peer->owner, take(msg), -1, 2,
opening_fundee_finished, peer);
}
static void peer_offer_channel(struct lightningd *ld,
struct funding_channel *fc,
const struct wireaddr *addr,
const struct crypto_state *cs,
const u8 *gfeatures, const u8 *lfeatures,
int peer_fd, int gossip_fd)
{
u8 *msg;
u32 max_to_self_delay, max_minimum_depth;
u64 min_effective_htlc_capacity_msat;
struct utxo *utxos;
/* We make a new peer. */
fc->peer = new_peer(ld, &fc->peerid, addr,
gfeatures, lfeatures, peer_fd);
fc->peer->funding_satoshi = fc->funding_satoshi;
fc->peer->push_msat = fc->push_msat;
peer_set_condition(fc->peer, UNINITIALIZED, OPENINGD);
peer_set_owner(fc->peer,
new_peer_subd(ld,
"lightning_openingd", fc->peer,
opening_wire_type_name, NULL,
take(&peer_fd), take(&gossip_fd), NULL));
if (!fc->peer->owner) {
fc->peer = tal_free(fc->peer);
command_fail(fc->cmd,
"Failed to launch openingd: %s",
strerror(errno));
return;
}
/* FIXME: This is wrong in several ways.
*
* 1. We should set the temporary channel id *now*, so that's the
* key.
* 2. We don't need the peer or channel in db until peer_persists().
*/
/* Store the channel in the database in order to get a channel
* ID that is unique and which we can base the peer_seed on */
fc->peer->channel = peer_channel_new(ld->wallet, fc->peer);
wallet_channel_save(fc->peer->ld->wallet, fc->peer->channel);
fc->peer->seed = tal(fc->peer, struct privkey);
derive_peer_seed(ld, fc->peer->seed, &fc->peer->id,
fc->peer->channel->id);
/* 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->channel_flags = OUR_CHANNEL_FLAGS;
msg = towire_opening_init(fc,
get_chainparams(ld)->index,
&fc->peer->our_config,
max_to_self_delay,
min_effective_htlc_capacity_msat,
cs, fc->peer->seed);
subd_send_msg(fc->peer->owner, take(msg));
utxos = from_utxoptr_arr(fc, fc->utxomap);
msg = towire_opening_funder(fc, fc->peer->funding_satoshi,
fc->peer->push_msat,
get_feerate(ld->topology, FEERATE_IMMEDIATE),
max_minimum_depth,
fc->change, fc->change_keyindex,
fc->peer->channel_flags,
utxos, fc->peer->ld->wallet->bip32_base);
/* Peer now owns fc; if it dies, we fail fc. */
tal_steal(fc->peer, fc);
tal_add_destructor(fc, fail_fundchannel_command);
subd_req(fc, fc->peer->owner,
take(msg), -1, 2, opening_funder_finished, fc);
}
/* Peer has been released from gossip. Start opening. */
static void gossip_peer_released(struct subd *gossip,
const u8 *resp,
const int *fds,
struct funding_channel *fc)
{
struct lightningd *ld = gossip->ld;
struct crypto_state cs;
u8 *gfeatures, *lfeatures;
struct wireaddr addr;
/* We could have raced with peer doing something else. */
fc->peer = peer_by_id(ld, &fc->peerid);
if (!fromwire_gossipctl_release_peer_reply(fc, resp, NULL, &addr, &cs,
&gfeatures, &lfeatures)) {
if (!fromwire_gossipctl_release_peer_replyfail(resp, NULL)) {
fatal("Gossip daemon gave invalid reply %s",
tal_hex(gossip, resp));
}
if (fc->peer)
command_fail(fc->cmd, "Peer already %s",
peer_state_name(fc->peer->state));
else
command_fail(fc->cmd, "Peer not connected");
return;
}
assert(tal_count(fds) == 2);
/* We asked to release this peer, but another raced in? Corner case,
* close this is easiest. */
if (fc->peer) {
command_fail(fc->cmd, "Peer already %s",
peer_state_name(fc->peer->state));
close(fds[0]);
close(fds[1]);
return;
}
/* OK, offer peer a channel. */
peer_offer_channel(ld, fc, &addr, &cs, gfeatures, lfeatures,
fds[0], fds[1]);
}
static void json_fund_channel(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
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 id and satoshi");
return;
}
fc->cmd = cmd;
if (!pubkey_from_hexstr(buffer + peertok->start,
peertok->end - peertok->start, &fc->peerid)) {
command_fail(cmd, "Could not parse id");
return;
}
if (!json_tok_u64(buffer, satoshitok, &fc->funding_satoshi)) {
command_fail(cmd, "Invalid satoshis");
return;
}
/* FIXME: Support push_msat? */
fc->push_msat = 0;
/* Try to do this now, so we know if insufficient funds. */
/* FIXME: dustlimit */
fc->utxomap = build_utxos(fc, cmd->ld, fc->funding_satoshi,
get_feerate(cmd->ld->topology, FEERATE_NORMAL),
600, &fc->change, &fc->change_keyindex);
if (!fc->utxomap) {
command_fail(cmd, "Cannot afford funding transaction");
return;
}
msg = towire_gossipctl_release_peer(cmd, &fc->peerid);
subd_req(fc, cmd->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 {tx} once channel established"
};
AUTODATA(json_command, &fund_channel_command);
static void json_close(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
jsmntok_t *peertok;
struct peer *peer;
if (!json_get_params(buffer, params,
"id", &peertok,
NULL)) {
command_fail(cmd, "Need id");
return;
}
peer = peer_from_json(cmd->ld, buffer, peertok);
if (!peer) {
command_fail(cmd, "Could not find peer with that id");
return;
}
/* Easy case: peer can simply be forgotten. */
if (!peer_persists(peer)) {
peer_fail_permanent(peer, NULL);
command_success(cmd, null_response(cmd));
return;
}
/* Normal case. */
if (peer->state == CHANNELD_NORMAL) {
u8 *shutdown_scriptpubkey;
peer->local_shutdown_idx = wallet_get_newindex(peer->ld);
if (peer->local_shutdown_idx == -1) {
command_fail(cmd, "Failed to get new key for shutdown");
return;
}
shutdown_scriptpubkey = p2wpkh_for_keyidx(cmd, peer->ld,
peer->local_shutdown_idx);
if (!shutdown_scriptpubkey) {
command_fail(cmd, "Failed to get script for shutdown");
return;
}
peer_set_condition(peer, CHANNELD_NORMAL, CHANNELD_SHUTTING_DOWN);
if (peer->owner)
subd_send_msg(peer->owner,
take(towire_channel_send_shutdown(peer,
shutdown_scriptpubkey)));
command_success(cmd, null_response(cmd));
} else
command_fail(cmd, "Peer is in state %s",
peer_state_name(peer->state));
}
static const struct json_command close_command = {
"close",
json_close,
"Close the channel with peer {id}",
"Returns an empty result on success"
};
AUTODATA(json_command, &close_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";
}
#if DEVELOPER
static void json_sign_last_tx(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
jsmntok_t *peertok;
struct peer *peer;
struct json_result *response = new_json_result(cmd);
u8 *linear;
if (!json_get_params(buffer, params,
"id", &peertok,
NULL)) {
command_fail(cmd, "Need id");
return;
}
peer = peer_from_json(cmd->ld, buffer, peertok);
if (!peer) {
command_fail(cmd, "Could not find peer with that id");
return;
}
if (!peer->last_tx) {
command_fail(cmd, "Peer has no final transaction");
return;
}
log_debug(peer->log, "dev-sign-last-tx: signing tx with %zu outputs",
tal_count(peer->last_tx->output));
sign_last_tx(peer);
linear = linearize_tx(cmd, peer->last_tx);
json_object_start(response, NULL);
json_add_hex(response, "tx", linear, tal_len(linear));
json_object_end(response);
command_success(cmd, response);
}
static const struct json_command dev_sign_last_tx = {
"dev-sign-last-tx",
json_sign_last_tx,
"Sign and return the last commitment transaction",
"Sign last transaction with peer @id, return as @tx."
" This should never be called outside testing!"
};
AUTODATA(json_command, &dev_sign_last_tx);
static void json_dev_fail(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
jsmntok_t *peertok;
struct peer *peer;
if (!json_get_params(buffer, params,
"id", &peertok,
NULL)) {
command_fail(cmd, "Need id");
return;
}
peer = peer_from_json(cmd->ld, buffer, peertok);
if (!peer) {
command_fail(cmd, "Could not find peer with that id");
return;
}
peer_internal_error(peer, "Failing due to dev-fail command");
command_success(cmd, null_response(cmd));
}
static const struct json_command dev_fail_command = {
"dev-fail",
json_dev_fail,
"Fail with peer {id}",
"Returns {} on success"
};
AUTODATA(json_command, &dev_fail_command);
static void dev_reenable_commit_finished(struct subd *channeld,
const u8 *resp,
const int *fds,
struct command *cmd)
{
command_success(cmd, null_response(cmd));
}
static void json_dev_reenable_commit(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
jsmntok_t *peertok;
struct peer *peer;
u8 *msg;
if (!json_get_params(buffer, params,
"id", &peertok,
NULL)) {
command_fail(cmd, "Need id");
return;
}
peer = peer_from_json(cmd->ld, buffer, peertok);
if (!peer) {
command_fail(cmd, "Could not find peer with that id");
return;
}
if (!peer->owner) {
command_fail(cmd, "Peer has no owner");
return;
}
if (!streq(peer->owner->name, "lightning_channeld")) {
command_fail(cmd, "Peer owned by %s", peer->owner->name);
return;
}
msg = towire_channel_dev_reenable_commit(peer);
subd_req(peer, peer->owner, take(msg), -1, 0,
dev_reenable_commit_finished, cmd);
}
static const struct json_command dev_reenable_commit = {
"dev-reenable-commit",
json_dev_reenable_commit,
"Reenable the commit timer on peer {id}",
"Returns {} on success"
};
AUTODATA(json_command, &dev_reenable_commit);
#endif /* DEVELOPER */