Files
pallectrum/electrum/lnutil.py

1766 lines
73 KiB
Python
Raw Normal View History

2018-10-25 19:34:31 +02:00
# Copyright (C) 2018 The Electrum developers
# Distributed under the MIT software license, see the accompanying
# file LICENCE or http://www.opensource.org/licenses/mit-license.php
from enum import IntFlag, IntEnum
import enum
import json
from collections import namedtuple, defaultdict
from typing import NamedTuple, List, Tuple, Mapping, Optional, TYPE_CHECKING, Union, Dict, Set, Sequence
import re
lnutil.LnFeatures: limit max feature bit to 10_000 closes https://github.com/spesmilo/electrum/issues/8403 > In Python 3.10 that worked fine, however in Python 3.11 large integer check https://github.com/python/cpython/issues/95778, so now this throws an error. Apparently this change was deemed a security fix and was backported to all supported branches of CPython (going back to 3.7). i.e. it affects ~all versions of python (if sufficiently updated with bugfix patches), not just 3.11 > Some offending node aliases: > ``` > ergvein-fiatchannels > test-mainnet > arakis > ``` The features bits set by some of these nodes: ``` (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 45, 32973, 52973) (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 39, 45, 55, 32973, 52973) ``` > P.S. I see there are a lot of nodes with 253 bytes in their feature vectors. Any idea why that could happen? Note that the valid [merged-into-spec features](https://github.com/lightning/bolts/blob/50b2df24a27879e8329712c275db78876fd022fe/09-features.md) currently only go as high as ~51. However the spec does not specify how to choose feature bits for experimental stuff, so I guess some people are using values in the 50k range. The only limit imposed by the spec on the length of the features bitvector is an implicit one due to the max message size: every msg must be smaller than 65KB, and the features bitvector needs to fit inside the init message, hence it can be up to ~524K bits. (note that the features are not stored in a sparse representation in the init message and in gossip messages, so if many nodes set such high feature bits, that would noticably impact the size of the gossip). ----- Anyway, our current implementation of LnFeatures is subclassing IntFlag, and it looks like it does not work well for such large integers. I've managed to make IntFlags reasonably in python 3.11 by overriding __str__ and __repr__ (note that in cpython it is apparently only the base2<->base10 conversions that are slow, power-of-2 conversions are fast, so we can e.g. use `hex()`). However in python 3.10 and older, enum.py itself seems really slow for bigints, e.g. enum._decompose in python 3.10. Try e.g. this script, which is instant in py3.11 but takes minutes in py3.10: ```py from enum import IntFlag class c(IntFlag): known_flag_1 = 1 << 0 known_flag_2 = 1 << 1 known_flag_3 = 1 << 2 if hasattr(IntFlag, "_numeric_repr_"): # python 3.11+ _numeric_repr_ = hex def __repr__(self): return f"<{self._name_}: {hex(self._value_)}>" def __str__(self): return hex(self._value_) a = c(2**70000-1) q1 = repr(a) q2 = str(a) ``` AFAICT we have two options: either we rewrite LnFeatures so that it does not use IntFlag (and enum.py), or, for the short term as workaround, we could just reject very large feature bits. For now, I've opted to the latter, rejecting feature bits over 10k. (note that another option is bumping the min required python to 3.11, in which case with the overrides added in this commit the performance looks perfectly fine)
2023-05-08 19:37:33 +00:00
import sys
import electrum_ecc as ecc
from electrum_ecc import CURVE_ORDER, ecdsa_sig64_from_der_sig, ECPubkey, string_to_number
import attr
from .util import bfh, inv_dict, UserFacingException
from .util import list_enabled_bits
from .util import ShortID as ShortChannelID
from .util import format_short_id as format_short_channel_id
from .crypto import sha256, pw_decode_with_version_and_mac
2019-10-23 17:09:41 +02:00
from .transaction import (Transaction, PartialTransaction, PartialTxInput, TxOutpoint,
PartialTxOutput, opcodes, TxOutput)
from . import bitcoin, crypto, transaction
from . import descriptor
from .bitcoin import (redeem_script_to_address, address_to_script,
construct_witness, construct_script)
from . import segwit_addr
from .i18n import _
2018-10-04 16:22:15 +02:00
from .lnaddr import lndecode
from .bip32 import BIP32Node, BIP32_PRIME
from .transaction import BCDataStream, OPPushDataGeneric
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
from .logging import get_logger
if TYPE_CHECKING:
from .lnchannel import Channel, AbstractChannel
from .lnrouter import LNPaymentRoute
from .lnonion import OnionRoutingFailure
from .simple_config import SimpleConfig
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
_logger = get_logger(__name__)
# defined in BOLT-03:
HTLC_TIMEOUT_WEIGHT = 663
HTLC_TIMEOUT_WEIGHT_ANCHORS = 666
HTLC_SUCCESS_WEIGHT = 703
HTLC_SUCCESS_WEIGHT_ANCHORS = 706
COMMITMENT_TX_WEIGHT = 724
COMMITMENT_TX_WEIGHT_ANCHORS = 1124
HTLC_OUTPUT_WEIGHT = 172
FIXED_ANCHOR_SAT = 330
LN_MAX_FUNDING_SAT_LEGACY = pow(2, 24) - 1
DUST_LIMIT_MAX = 1000
from .json_db import StoredObject, stored_in, stored_as
def channel_id_from_funding_tx(funding_txid: str, funding_index: int) -> Tuple[bytes, bytes]:
funding_txid_bytes = bytes.fromhex(funding_txid)[::-1]
i = int.from_bytes(funding_txid_bytes, 'big') ^ funding_index
return i.to_bytes(32, 'big'), funding_txid_bytes
hex_to_bytes = lambda v: v if isinstance(v, bytes) else bytes.fromhex(v) if v is not None else None
bytes_to_hex = lambda v: repr(v.hex()) if v is not None else None
json_to_keypair = lambda v: v if isinstance(v, OnlyPubkeyKeypair) else Keypair(**v) if len(v)==2 else OnlyPubkeyKeypair(**v)
def serialize_htlc_key(scid: bytes, htlc_id: int) -> str:
return scid.hex() + ':%d'%htlc_id
def deserialize_htlc_key(htlc_key: str) -> Tuple[bytes, int]:
scid, htlc_id = htlc_key.split(':')
return bytes.fromhex(scid), int(htlc_id)
@attr.s
class OnlyPubkeyKeypair(StoredObject):
pubkey = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
@attr.s
class Keypair(OnlyPubkeyKeypair):
privkey = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
@attr.s
class ChannelConfig(StoredObject):
# shared channel config fields
payment_basepoint = attr.ib(type=OnlyPubkeyKeypair, converter=json_to_keypair)
multisig_key = attr.ib(type=OnlyPubkeyKeypair, converter=json_to_keypair)
htlc_basepoint = attr.ib(type=OnlyPubkeyKeypair, converter=json_to_keypair)
delayed_basepoint = attr.ib(type=OnlyPubkeyKeypair, converter=json_to_keypair)
revocation_basepoint = attr.ib(type=OnlyPubkeyKeypair, converter=json_to_keypair)
2020-06-10 20:13:01 +02:00
to_self_delay = attr.ib(type=int) # applies to OTHER ctx
dust_limit_sat = attr.ib(type=int) # applies to SAME ctx
max_htlc_value_in_flight_msat = attr.ib(type=int) # max val of INCOMING htlcs
max_accepted_htlcs = attr.ib(type=int) # max num of INCOMING htlcs
initial_msat = attr.ib(type=int)
2020-06-10 20:13:01 +02:00
reserve_sat = attr.ib(type=int) # applies to OTHER ctx
htlc_minimum_msat = attr.ib(type=int) # smallest value for INCOMING htlc
upfront_shutdown_script = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
announcement_node_sig = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
announcement_bitcoin_sig = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
def validate_params(self, *, funding_sat: int, config: 'SimpleConfig', peer_features: 'LnFeatures') -> None:
conf_name = type(self).__name__
for key in (
self.payment_basepoint,
self.multisig_key,
self.htlc_basepoint,
self.delayed_basepoint,
self.revocation_basepoint
):
if not (len(key.pubkey) == 33 and ecc.ECPubkey.is_pubkey_bytes(key.pubkey)):
raise Exception(f"{conf_name}. invalid pubkey in channel config")
if funding_sat < MIN_FUNDING_SAT:
raise Exception(f"funding_sat too low: {funding_sat} sat < {MIN_FUNDING_SAT}")
if not peer_features.supports(LnFeatures.OPTION_SUPPORT_LARGE_CHANNEL_OPT):
# MUST set funding_satoshis to less than 2^24 satoshi
if funding_sat > LN_MAX_FUNDING_SAT_LEGACY:
raise Exception(f"funding_sat too high: {funding_sat} sat > {LN_MAX_FUNDING_SAT_LEGACY} (legacy limit)")
if funding_sat > config.LIGHTNING_MAX_FUNDING_SAT:
raise Exception(f"funding_sat too high: {funding_sat} sat > {config.LIGHTNING_MAX_FUNDING_SAT} (config setting)")
# MUST set push_msat to equal or less than 1000 * funding_satoshis
if not (0 <= self.initial_msat <= 1000 * funding_sat):
raise Exception(f"{conf_name}. insane initial_msat={self.initial_msat}. (funding_sat={funding_sat})")
if self.reserve_sat < self.dust_limit_sat:
raise Exception(f"{conf_name}. MUST set channel_reserve_satoshis greater than or equal to dust_limit_satoshis")
if self.dust_limit_sat < bitcoin.DUST_LIMIT_UNKNOWN_SEGWIT:
raise Exception(f"{conf_name}. dust limit too low: {self.dust_limit_sat} sat")
if self.dust_limit_sat > DUST_LIMIT_MAX:
raise Exception(f"{conf_name}. dust limit too high: {self.dust_limit_sat} sat")
if self.reserve_sat > funding_sat // 100:
raise Exception(f"{conf_name}. reserve too high: {self.reserve_sat}, funding_sat: {funding_sat}")
if self.htlc_minimum_msat > 1_000:
raise Exception(f"{conf_name}. htlc_minimum_msat too high: {self.htlc_minimum_msat} msat")
HTLC_MINIMUM_MSAT_MIN = 0 # should be at least 1 really, but apparently some nodes are sending zero...
if self.htlc_minimum_msat < HTLC_MINIMUM_MSAT_MIN:
raise Exception(f"{conf_name}. htlc_minimum_msat too low: {self.htlc_minimum_msat} msat < {HTLC_MINIMUM_MSAT_MIN}")
if self.max_accepted_htlcs < 5:
raise Exception(f"{conf_name}. max_accepted_htlcs too low: {self.max_accepted_htlcs}")
if self.max_accepted_htlcs > 483:
raise Exception(f"{conf_name}. max_accepted_htlcs too high: {self.max_accepted_htlcs}")
if self.to_self_delay > MAXIMUM_REMOTE_TO_SELF_DELAY_ACCEPTED:
raise Exception(f"{conf_name}. to_self_delay too high: {self.to_self_delay} > {MAXIMUM_REMOTE_TO_SELF_DELAY_ACCEPTED}")
if self.max_htlc_value_in_flight_msat < min(1000 * funding_sat, 100_000_000):
raise Exception(f"{conf_name}. max_htlc_value_in_flight_msat is too small: {self.max_htlc_value_in_flight_msat}")
@classmethod
def cross_validate_params(
cls,
*,
local_config: 'LocalConfig',
remote_config: 'RemoteConfig',
funding_sat: int,
is_local_initiator: bool, # whether we are the funder
initial_feerate_per_kw: int,
config: 'SimpleConfig',
peer_features: 'LnFeatures',
has_anchors: bool,
) -> None:
# first we validate the configs separately
local_config.validate_params(funding_sat=funding_sat, config=config, peer_features=peer_features)
remote_config.validate_params(funding_sat=funding_sat, config=config, peer_features=peer_features)
# now do tests that need access to both configs
if is_local_initiator:
funder, fundee = LOCAL, REMOTE
funder_config, fundee_config = local_config, remote_config
else:
funder, fundee = REMOTE, LOCAL
funder_config, fundee_config = remote_config, local_config
# if channel_reserve_satoshis is less than dust_limit_satoshis within the open_channel message:
# MUST reject the channel.
if remote_config.reserve_sat < local_config.dust_limit_sat:
raise Exception("violated constraint: remote_config.reserve_sat < local_config.dust_limit_sat")
# if channel_reserve_satoshis from the open_channel message is less than dust_limit_satoshis:
# MUST reject the channel.
if local_config.reserve_sat < remote_config.dust_limit_sat:
raise Exception("violated constraint: local_config.reserve_sat < remote_config.dust_limit_sat")
# The receiving node MUST fail the channel if:
# the funder's amount for the initial commitment transaction is not
# sufficient for full fee payment.
if funder_config.initial_msat < calc_fees_for_commitment_tx(
num_htlcs=0,
feerate=initial_feerate_per_kw,
is_local_initiator=is_local_initiator,
has_anchors=has_anchors,
)[funder]:
raise Exception(
"the funder's amount for the initial commitment transaction "
"is not sufficient for full fee payment")
# The receiving node MUST fail the channel if:
# both to_local and to_remote amounts for the initial commitment transaction are
# less than or equal to channel_reserve_satoshis (see BOLT 3).
if (max(local_config.initial_msat, remote_config.initial_msat)
<= 1000 * max(local_config.reserve_sat, remote_config.reserve_sat)):
raise Exception(
"both to_local and to_remote amounts for the initial commitment "
"transaction are less than or equal to channel_reserve_satoshis")
from .simple_config import FEERATE_PER_KW_MIN_RELAY_LIGHTNING
if initial_feerate_per_kw < FEERATE_PER_KW_MIN_RELAY_LIGHTNING:
raise Exception(f"feerate lower than min relay fee. {initial_feerate_per_kw} sat/kw.")
@stored_as('local_config')
@attr.s
class LocalConfig(ChannelConfig):
channel_seed = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex) # type: Optional[bytes]
funding_locked_received = attr.ib(type=bool)
current_commitment_signature = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
current_htlc_signatures = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
per_commitment_secret_seed = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
2020-03-31 12:43:43 +02:00
@classmethod
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
def from_seed(cls, **kwargs):
channel_seed = kwargs['channel_seed']
node = BIP32Node.from_rootseed(channel_seed, xtype='standard')
2020-03-31 12:43:43 +02:00
keypair_generator = lambda family: generate_keypair(node, family)
kwargs['per_commitment_secret_seed'] = keypair_generator(LnKeyFamily.REVOCATION_ROOT).privkey
kwargs['multisig_key'] = keypair_generator(LnKeyFamily.MULTISIG)
kwargs['htlc_basepoint'] = keypair_generator(LnKeyFamily.HTLC_BASE)
kwargs['delayed_basepoint'] = keypair_generator(LnKeyFamily.DELAY_BASE)
kwargs['revocation_basepoint'] = keypair_generator(LnKeyFamily.REVOCATION_BASE)
static_remotekey = kwargs.pop('static_remotekey')
static_payment_key = kwargs.pop('static_payment_key')
if static_payment_key:
# We derive the payment_basepoint from a static secret (derived from
# the wallet seed) and a public nonce that is revealed
# when the funding transaction is spent. This way we can restore the
# payment_basepoint, needed for sweeping in the event of a force close.
kwargs['payment_basepoint'] = derive_payment_basepoint(
static_payment_secret=static_payment_key.privkey,
funding_pubkey=kwargs['multisig_key'].pubkey
)
elif static_remotekey: # we automatically sweep to a wallet address
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
kwargs['payment_basepoint'] = OnlyPubkeyKeypair(static_remotekey)
else:
# we expect all our channels to use option_static_remotekey, so ending up here likely indicates an issue...
kwargs['payment_basepoint'] = keypair_generator(LnKeyFamily.PAYMENT_BASE)
2020-03-31 12:43:43 +02:00
return LocalConfig(**kwargs)
def validate_params(self, *, funding_sat: int, config: 'SimpleConfig', peer_features: 'LnFeatures') -> None:
conf_name = type(self).__name__
# run base checks regardless whether LOCAL/REMOTE config
super().validate_params(funding_sat=funding_sat, config=config, peer_features=peer_features)
# run some stricter checks on LOCAL config (make sure we ourselves do the sane thing,
# even if we are lenient with REMOTE for compatibility reasons)
HTLC_MINIMUM_MSAT_MIN = 1
if self.htlc_minimum_msat < HTLC_MINIMUM_MSAT_MIN:
raise Exception(f"{conf_name}. htlc_minimum_msat too low: {self.htlc_minimum_msat} msat < {HTLC_MINIMUM_MSAT_MIN}")
@stored_as('remote_config')
@attr.s
class RemoteConfig(ChannelConfig):
next_per_commitment_point = attr.ib(type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
current_per_commitment_point = attr.ib(default=None, type=bytes, converter=hex_to_bytes, repr=bytes_to_hex)
@stored_in('fee_updates')
@attr.s
class FeeUpdate(StoredObject):
rate = attr.ib(type=int) # in sat/kw
ctn_local = attr.ib(default=None, type=int)
ctn_remote = attr.ib(default=None, type=int)
2019-07-23 19:23:39 +02:00
@stored_as('constraints')
@attr.s
class ChannelConstraints(StoredObject):
flags = attr.ib(type=int, converter=int)
capacity = attr.ib(type=int) # in sat
is_initiator = attr.ib(type=bool) # note: sometimes also called "funder"
funding_txn_minimum_depth = attr.ib(type=int)
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
CHANNEL_BACKUP_VERSION_LATEST = 1
KNOWN_CHANNEL_BACKUP_VERSIONS = (0, 1,)
assert CHANNEL_BACKUP_VERSION_LATEST in KNOWN_CHANNEL_BACKUP_VERSIONS
@attr.s
class ChannelBackupStorage(StoredObject):
funding_txid = attr.ib(type=str)
funding_index = attr.ib(type=int, converter=int)
funding_address = attr.ib(type=str)
is_initiator = attr.ib(type=bool)
def funding_outpoint(self):
return Outpoint(self.funding_txid, self.funding_index)
def channel_id(self):
chan_id, _ = channel_id_from_funding_tx(self.funding_txid, self.funding_index)
return chan_id
@stored_in('onchain_channel_backups')
@attr.s
class OnchainChannelBackupStorage(ChannelBackupStorage):
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
node_id_prefix = attr.ib(type=bytes, converter=hex_to_bytes) # remote node pubkey
@stored_in('imported_channel_backups')
@attr.s
class ImportedChannelBackupStorage(ChannelBackupStorage):
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
node_id = attr.ib(type=bytes, converter=hex_to_bytes) # remote node pubkey
privkey = attr.ib(type=bytes, converter=hex_to_bytes) # local node privkey
host = attr.ib(type=str)
port = attr.ib(type=int, converter=int)
channel_seed = attr.ib(type=bytes, converter=hex_to_bytes)
local_delay = attr.ib(type=int, converter=int)
remote_delay = attr.ib(type=int, converter=int)
remote_payment_pubkey = attr.ib(type=bytes, converter=hex_to_bytes)
remote_revocation_pubkey = attr.ib(type=bytes, converter=hex_to_bytes)
local_payment_pubkey = attr.ib(type=bytes, converter=hex_to_bytes) # type: Optional[bytes]
def to_bytes(self) -> bytes:
vds = BCDataStream()
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
vds.write_uint16(CHANNEL_BACKUP_VERSION_LATEST)
vds.write_boolean(self.is_initiator)
vds.write_bytes(self.privkey, 32)
vds.write_bytes(self.channel_seed, 32)
vds.write_bytes(self.node_id, 33)
vds.write_bytes(bfh(self.funding_txid), 32)
vds.write_uint16(self.funding_index)
vds.write_string(self.funding_address)
vds.write_bytes(self.remote_payment_pubkey, 33)
vds.write_bytes(self.remote_revocation_pubkey, 33)
vds.write_uint16(self.local_delay)
vds.write_uint16(self.remote_delay)
vds.write_string(self.host)
vds.write_uint16(self.port)
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
vds.write_bytes(self.local_payment_pubkey, 33)
return bytes(vds.input)
@staticmethod
def from_bytes(s: bytes) -> "ImportedChannelBackupStorage":
vds = BCDataStream()
vds.write(s)
version = vds.read_uint16()
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
if version not in KNOWN_CHANNEL_BACKUP_VERSIONS:
raise Exception(f"unknown version for channel backup: {version}")
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
is_initiator = vds.read_boolean()
privkey = vds.read_bytes(32)
channel_seed = vds.read_bytes(32)
node_id = vds.read_bytes(33)
funding_txid = vds.read_bytes(32).hex()
funding_index = vds.read_uint16()
funding_address = vds.read_string()
remote_payment_pubkey = vds.read_bytes(33)
remote_revocation_pubkey = vds.read_bytes(33)
local_delay = vds.read_uint16()
remote_delay = vds.read_uint16()
host = vds.read_string()
port = vds.read_uint16()
if version >= 1:
local_payment_pubkey = vds.read_bytes(33)
else:
local_payment_pubkey = None
return ImportedChannelBackupStorage(
fix sweeping chan after local force-close using cb scenario: - user opens a lightning channel and exports an "imported channel backup" - user closes channel via local-force-close - local ctx is published, to_local output has user's funds and they are CSV-locked for days - user restores wallet file from seed and imports channel backup - new wallet file should be able to sweep coins from to_local output (after CSV expires) This was not working previously, as the local_payment_basepoint was not included in the imported channel backups, and the code was interpreting the lack of this as the channel not having option_static_remotekey enabled. This resulted in lnutil.extract_ctn_from_tx using an incorrect funder_payment_basepoint, and lnsweep not recognising the ctx due to the garbage ctn value. The imported channel backup serialisation format is slightly changed to include the previously missing field, and its version number is bumped (0->1). We allow importing both version 0 and version 1 backups, however v0 backups cannot handle the above described scenario (they can only be used to request a remote-force-close). Note that we were/are setting the missing local_payment_basepoint to the pubkey of one of the wallet change addresses, which is bruteforceable if necessary, but I think it is not worth the complexity to add this bruteforce logic. Also note that the bruteforcing could only be done after the local-force-close was broadcast. Ideally people with existing channels and already exported v0 backups should re-export v1 backups... Not sure how to handle this. closes https://github.com/spesmilo/electrum/issues/8516
2023-07-14 14:21:50 +00:00
is_initiator=is_initiator,
privkey=privkey,
channel_seed=channel_seed,
node_id=node_id,
funding_txid=funding_txid,
funding_index=funding_index,
funding_address=funding_address,
remote_payment_pubkey=remote_payment_pubkey,
remote_revocation_pubkey=remote_revocation_pubkey,
local_delay=local_delay,
remote_delay=remote_delay,
host=host,
port=port,
local_payment_pubkey=local_payment_pubkey,
)
@staticmethod
def from_encrypted_str(data: str, *, password: str) -> "ImportedChannelBackupStorage":
if not data.startswith('channel_backup:'):
raise ValueError("missing or invalid magic bytes")
encrypted = data[15:]
decrypted = pw_decode_with_version_and_mac(encrypted, password)
return ImportedChannelBackupStorage.from_bytes(decrypted)
class ScriptHtlc(NamedTuple):
redeem_script: bytes
htlc: 'UpdateAddHtlc'
2019-10-23 17:09:41 +02:00
# FIXME duplicate of TxOutpoint in transaction.py??
@stored_as('funding_outpoint')
@attr.s
class Outpoint(StoredObject):
txid = attr.ib(type=str)
output_index = attr.ib(type=int)
2018-09-12 16:17:10 +02:00
def to_str(self):
return "{}:{}".format(self.txid, self.output_index)
class HtlcLog(NamedTuple):
success: bool
amount_msat: int # amount for receiver (e.g. from invoice)
route: Optional['LNPaymentRoute'] = None
preimage: Optional[bytes] = None
error_bytes: Optional[bytes] = None
failure_msg: Optional['OnionRoutingFailure'] = None
sender_idx: Optional[int] = None
trampoline_fee_level: Optional[int] = None
2020-03-10 16:44:23 +01:00
def formatted_tuple(self):
route = self.route
route_str = '%d'%len(route)
short_channel_id = None
if not self.success:
sender_idx = self.sender_idx
failure_msg = self.failure_msg
if sender_idx is not None:
try:
short_channel_id = route[sender_idx + 1].short_channel_id
except IndexError:
# payment destination reported error
short_channel_id = _("Destination node")
message = failure_msg.code_name()
2020-03-10 16:44:23 +01:00
else:
short_channel_id = route[-1].short_channel_id
message = _('Success')
chan_str = str(short_channel_id) if short_channel_id else _("Unknown")
2020-03-10 16:44:23 +01:00
return route_str, chan_str, message
2018-10-15 11:47:42 +02:00
class LightningError(Exception): pass
class UnableToDeriveSecret(LightningError): pass
2018-10-23 20:32:18 +02:00
class RemoteMisbehaving(LightningError): pass
2018-10-18 22:56:40 +02:00
2018-11-05 20:29:44 +01:00
class NotFoundChanAnnouncementForUpdate(Exception): pass
class InvalidGossipMsg(Exception):
"""e.g. signature check failed"""
2018-11-05 20:29:44 +01:00
2019-11-18 05:12:13 +01:00
class PaymentFailure(UserFacingException): pass
2021-02-19 09:27:28 +01:00
class NoPathFound(PaymentFailure):
def __str__(self):
return _('No path found')
2018-10-18 22:56:40 +02:00
class LNProtocolError(Exception):
"""Raised in peer methods to trigger an error message."""
class LNProtocolWarning(Exception):
"""Raised in peer methods to trigger a warning message."""
# TODO make some of these values configurable?
REDEEM_AFTER_DOUBLE_SPENT_DELAY = 30
2020-04-03 18:54:02 +02:00
CHANNEL_OPENING_TIMEOUT = 24*60*60
# Small capacity channels are problematic for many reasons. As the onchain fees start to become
# significant compared to the capacity, things start to break down. e.g. the counterparty
# force-closing the channel costs much of the funds in the channel.
# Closing a channel uses ~200 vbytes onchain, feerates could spike to 100 sat/vbyte or even higher;
# that in itself is already 20_000 sats. This mining fee is reserved and cannot be used for payments.
# The value below is chosen arbitrarily to be one order of magnitude higher than that.
MIN_FUNDING_SAT = 200_000
##### CLTV-expiry-delta-related values
# see https://github.com/lightningnetwork/lightning-rfc/blob/master/02-peer-protocol.md#cltv_expiry_delta-selection
# the minimum cltv_expiry accepted for newly received HTLCs
# note: when changing, consider Blockchain.is_tip_stale()
MIN_FINAL_CLTV_DELTA_ACCEPTED = 144
# set it a tiny bit higher for invoices as blocks could get mined
# during forward path of payment
MIN_FINAL_CLTV_DELTA_FOR_INVOICE = MIN_FINAL_CLTV_DELTA_ACCEPTED + 3
# the deadline for offered HTLCs:
# the deadline after which the channel has to be failed and timed out on-chain
NBLOCK_DEADLINE_DELTA_AFTER_EXPIRY_FOR_OFFERED_HTLCS = 1
# the deadline for received HTLCs this node has fulfilled:
# the deadline after which the channel has to be failed and the HTLC fulfilled on-chain before its cltv_expiry
NBLOCK_DEADLINE_DELTA_BEFORE_EXPIRY_FOR_RECEIVED_HTLCS = 72
NBLOCK_CLTV_DELTA_TOO_FAR_INTO_FUTURE = 28 * 144
MAXIMUM_REMOTE_TO_SELF_DELAY_ACCEPTED = 2016
class RevocationStore:
# closely based on code in lightningnetwork/lnd
START_INDEX = 2 ** 48 - 1
def __init__(self, storage):
if len(storage) == 0:
storage['index'] = self.START_INDEX
storage['buckets'] = {}
self.storage = storage
self.buckets = storage['buckets']
def add_next_entry(self, hsh):
index = self.storage['index']
new_element = ShachainElement(index=index, secret=hsh)
bucket = count_trailing_zeros(index)
for i in range(0, bucket):
this_bucket = self.buckets[i]
e = shachain_derive(new_element, this_bucket.index)
if e != this_bucket:
raise Exception("hash is not derivable: {} {} {}".format(e.secret.hex(), this_bucket.secret.hex(), this_bucket.index))
self.buckets[bucket] = new_element
self.storage['index'] = index - 1
def retrieve_secret(self, index: int) -> bytes:
assert index <= self.START_INDEX, index
for i in range(0, 49):
bucket = self.buckets.get(i)
if bucket is None:
raise UnableToDeriveSecret()
try:
element = shachain_derive(bucket, index)
except UnableToDeriveSecret:
continue
return element.secret
raise UnableToDeriveSecret()
def __eq__(self, o):
return type(o) is RevocationStore and self.serialize() == o.serialize()
def __hash__(self):
return hash(json.dumps(self.serialize(), sort_keys=True))
def count_trailing_zeros(index):
""" BOLT-03 (where_to_put_secret) """
try:
return list(reversed(bin(index)[2:])).index("1")
except ValueError:
return 48
def shachain_derive(element, to_index):
def get_prefix(index, pos):
mask = (1 << 64) - 1 - ((1 << pos) - 1)
return index & mask
from_index = element.index
zeros = count_trailing_zeros(from_index)
if from_index != get_prefix(to_index, zeros):
raise UnableToDeriveSecret("prefixes are different; index not derivable")
return ShachainElement(
get_per_commitment_secret_from_seed(element.secret, to_index, zeros),
to_index)
class ShachainElement(NamedTuple):
secret: bytes
index: int
def __str__(self):
return "ShachainElement(" + self.secret.hex() + "," + str(self.index) + ")"
@stored_in('buckets', tuple)
def read(*x):
return ShachainElement(bfh(x[0]), int(x[1]))
def get_per_commitment_secret_from_seed(seed: bytes, i: int, bits: int = 48) -> bytes:
"""Generate per commitment secret."""
per_commitment_secret = bytearray(seed)
for bitindex in range(bits - 1, -1, -1):
mask = 1 << bitindex
if i & mask:
per_commitment_secret[bitindex // 8] ^= 1 << (bitindex % 8)
per_commitment_secret = bytearray(sha256(per_commitment_secret))
bajts = bytes(per_commitment_secret)
return bajts
def secret_to_pubkey(secret: int) -> bytes:
assert type(secret) is int
return ecc.ECPrivkey.from_secret_scalar(secret).get_public_key_bytes(compressed=True)
2018-10-15 11:16:35 +02:00
def derive_pubkey(basepoint: bytes, per_commitment_point: bytes) -> bytes:
2020-02-06 20:59:57 +01:00
p = ecc.ECPubkey(basepoint) + ecc.GENERATOR * ecc.string_to_number(sha256(per_commitment_point + basepoint))
return p.get_public_key_bytes()
def derive_privkey(secret: int, per_commitment_point: bytes) -> int:
assert type(secret) is int
basepoint_bytes = secret_to_pubkey(secret)
basepoint = secret + ecc.string_to_number(sha256(per_commitment_point + basepoint_bytes))
basepoint %= CURVE_ORDER
return basepoint
def derive_blinded_pubkey(basepoint: bytes, per_commitment_point: bytes) -> bytes:
k1 = ecc.ECPubkey(basepoint) * ecc.string_to_number(sha256(basepoint + per_commitment_point))
k2 = ecc.ECPubkey(per_commitment_point) * ecc.string_to_number(sha256(per_commitment_point + basepoint))
return (k1 + k2).get_public_key_bytes()
def derive_blinded_privkey(basepoint_secret: bytes, per_commitment_secret: bytes) -> bytes:
basepoint = ecc.ECPrivkey(basepoint_secret).get_public_key_bytes(compressed=True)
per_commitment_point = ecc.ECPrivkey(per_commitment_secret).get_public_key_bytes(compressed=True)
k1 = ecc.string_to_number(basepoint_secret) * ecc.string_to_number(sha256(basepoint + per_commitment_point))
k2 = ecc.string_to_number(per_commitment_secret) * ecc.string_to_number(sha256(per_commitment_point + basepoint))
sum = (k1 + k2) % ecc.CURVE_ORDER
return int.to_bytes(sum, length=32, byteorder='big', signed=False)
def derive_payment_basepoint(static_payment_secret: bytes, funding_pubkey: bytes) -> Keypair:
assert isinstance(static_payment_secret, bytes)
assert isinstance(funding_pubkey, bytes)
payment_basepoint = ecc.ECPrivkey(sha256(static_payment_secret + funding_pubkey))
return Keypair(
pubkey=payment_basepoint.get_public_key_bytes(),
privkey=payment_basepoint.get_secret_bytes()
)
def make_htlc_tx_output(
amount_msat,
local_feerate,
revocationpubkey,
local_delayedpubkey,
success,
to_self_delay,
has_anchors: bool
) -> Tuple[bytes, PartialTxOutput]:
assert type(amount_msat) is int
assert type(local_feerate) is int
script = make_commitment_output_to_local_witness_script(
revocation_pubkey=revocationpubkey,
to_self_delay=to_self_delay,
delayed_pubkey=local_delayedpubkey,
)
p2wsh = bitcoin.redeem_script_to_address('p2wsh', script)
weight = effective_htlc_tx_weight(success=success, has_anchors=has_anchors)
fee = local_feerate * weight
fee = fee // 1000 * 1000
final_amount_sat = (amount_msat - fee) // 1000
assert final_amount_sat > 0, final_amount_sat
2019-10-23 17:09:41 +02:00
output = PartialTxOutput.from_address_and_value(p2wsh, final_amount_sat)
return script, output
def make_htlc_tx_witness(remotehtlcsig: bytes, localhtlcsig: bytes,
payment_preimage: bytes, witness_script: bytes) -> bytes:
assert type(remotehtlcsig) is bytes
assert type(localhtlcsig) is bytes
assert type(payment_preimage) is bytes
assert type(witness_script) is bytes
return construct_witness([0, remotehtlcsig, localhtlcsig, payment_preimage, witness_script])
def make_htlc_tx_inputs(htlc_output_txid: str, htlc_output_index: int,
amount_msat: int, witness_script: bytes) -> List[PartialTxInput]:
assert type(htlc_output_txid) is str
assert type(htlc_output_index) is int
assert type(amount_msat) is int
assert type(witness_script) is bytes
2019-10-23 17:09:41 +02:00
txin = PartialTxInput(prevout=TxOutpoint(txid=bfh(htlc_output_txid), out_idx=htlc_output_index),
nsequence=0)
txin.witness_script = witness_script
2019-10-23 17:09:41 +02:00
txin.script_sig = b''
txin._trusted_value_sats = amount_msat // 1000
c_inputs = [txin]
return c_inputs
def make_htlc_tx(*, cltv_abs: int, inputs: List[PartialTxInput], output: PartialTxOutput) -> PartialTransaction:
assert type(cltv_abs) is int
c_outputs = [output]
tx = PartialTransaction.from_io(inputs, c_outputs, locktime=cltv_abs, version=2)
return tx
def make_offered_htlc(
*,
revocation_pubkey: bytes,
remote_htlcpubkey: bytes,
local_htlcpubkey: bytes,
payment_hash: bytes,
has_anchors: bool,
) -> bytes:
assert type(revocation_pubkey) is bytes
assert type(remote_htlcpubkey) is bytes
assert type(local_htlcpubkey) is bytes
assert type(payment_hash) is bytes
script_opcodes = [
opcodes.OP_DUP,
opcodes.OP_HASH160,
bitcoin.hash_160(revocation_pubkey),
opcodes.OP_EQUAL,
opcodes.OP_IF,
opcodes.OP_CHECKSIG,
opcodes.OP_ELSE,
remote_htlcpubkey,
opcodes.OP_SWAP,
opcodes.OP_SIZE,
32,
opcodes.OP_EQUAL,
opcodes.OP_NOTIF,
opcodes.OP_DROP,
2,
opcodes.OP_SWAP,
local_htlcpubkey,
2,
opcodes.OP_CHECKMULTISIG,
opcodes.OP_ELSE,
opcodes.OP_HASH160,
crypto.ripemd(payment_hash),
opcodes.OP_EQUALVERIFY,
opcodes.OP_CHECKSIG,
opcodes.OP_ENDIF,
]
if has_anchors:
script_opcodes.extend([1, opcodes.OP_CHECKSEQUENCEVERIFY, opcodes.OP_DROP])
script_opcodes.append(opcodes.OP_ENDIF)
script = construct_script(script_opcodes)
return script
def make_received_htlc(
*,
revocation_pubkey: bytes,
remote_htlcpubkey: bytes,
local_htlcpubkey: bytes,
payment_hash: bytes,
cltv_abs: int,
has_anchors: bool,
) -> bytes:
for i in [revocation_pubkey, remote_htlcpubkey, local_htlcpubkey, payment_hash]:
assert type(i) is bytes
assert type(cltv_abs) is int
script_opcodes = [
opcodes.OP_DUP,
opcodes.OP_HASH160,
bitcoin.hash_160(revocation_pubkey),
opcodes.OP_EQUAL,
opcodes.OP_IF,
opcodes.OP_CHECKSIG,
opcodes.OP_ELSE,
remote_htlcpubkey,
opcodes.OP_SWAP,
opcodes.OP_SIZE,
32,
opcodes.OP_EQUAL,
opcodes.OP_IF,
opcodes.OP_HASH160,
crypto.ripemd(payment_hash),
opcodes.OP_EQUALVERIFY,
2,
opcodes.OP_SWAP,
local_htlcpubkey,
2,
opcodes.OP_CHECKMULTISIG,
opcodes.OP_ELSE,
opcodes.OP_DROP,
cltv_abs,
opcodes.OP_CHECKLOCKTIMEVERIFY,
opcodes.OP_DROP,
opcodes.OP_CHECKSIG,
opcodes.OP_ENDIF,
]
if has_anchors:
script_opcodes.extend([1, opcodes.OP_CHECKSEQUENCEVERIFY, opcodes.OP_DROP])
script_opcodes.append(opcodes.OP_ENDIF)
script = construct_script(script_opcodes)
return script
WITNESS_TEMPLATE_OFFERED_HTLC = [
opcodes.OP_DUP,
opcodes.OP_HASH160,
OPPushDataGeneric(None),
opcodes.OP_EQUAL,
opcodes.OP_IF,
opcodes.OP_CHECKSIG,
opcodes.OP_ELSE,
OPPushDataGeneric(None),
opcodes.OP_SWAP,
opcodes.OP_SIZE,
OPPushDataGeneric(lambda x: x==1),
opcodes.OP_EQUAL,
opcodes.OP_NOTIF,
opcodes.OP_DROP,
opcodes.OP_2,
opcodes.OP_SWAP,
OPPushDataGeneric(None),
opcodes.OP_2,
opcodes.OP_CHECKMULTISIG,
opcodes.OP_ELSE,
opcodes.OP_HASH160,
OPPushDataGeneric(None),
opcodes.OP_EQUALVERIFY,
opcodes.OP_CHECKSIG,
opcodes.OP_ENDIF,
opcodes.OP_ENDIF,
]
WITNESS_TEMPLATE_RECEIVED_HTLC = [
opcodes.OP_DUP,
opcodes.OP_HASH160,
OPPushDataGeneric(None),
opcodes.OP_EQUAL,
opcodes.OP_IF,
opcodes.OP_CHECKSIG,
opcodes.OP_ELSE,
OPPushDataGeneric(None),
opcodes.OP_SWAP,
opcodes.OP_SIZE,
OPPushDataGeneric(lambda x: x==1),
opcodes.OP_EQUAL,
opcodes.OP_IF,
opcodes.OP_HASH160,
OPPushDataGeneric(None),
opcodes.OP_EQUALVERIFY,
opcodes.OP_2,
opcodes.OP_SWAP,
OPPushDataGeneric(None),
opcodes.OP_2,
opcodes.OP_CHECKMULTISIG,
opcodes.OP_ELSE,
opcodes.OP_DROP,
OPPushDataGeneric(None),
opcodes.OP_CHECKLOCKTIMEVERIFY,
opcodes.OP_DROP,
opcodes.OP_CHECKSIG,
opcodes.OP_ENDIF,
opcodes.OP_ENDIF,
]
def make_htlc_output_witness_script(
*,
is_received_htlc: bool,
remote_revocation_pubkey: bytes,
remote_htlc_pubkey: bytes,
local_htlc_pubkey: bytes,
payment_hash: bytes,
cltv_abs: Optional[int],
has_anchors: bool,
) -> bytes:
if is_received_htlc:
return make_received_htlc(
revocation_pubkey=remote_revocation_pubkey,
remote_htlcpubkey=remote_htlc_pubkey,
local_htlcpubkey=local_htlc_pubkey,
payment_hash=payment_hash,
cltv_abs=cltv_abs,
has_anchors=has_anchors,
)
else:
return make_offered_htlc(
revocation_pubkey=remote_revocation_pubkey,
remote_htlcpubkey=remote_htlc_pubkey,
local_htlcpubkey=local_htlc_pubkey,
payment_hash=payment_hash,
has_anchors=has_anchors,
)
def get_ordered_channel_configs(chan: 'AbstractChannel', for_us: bool) -> Tuple[Union[LocalConfig, RemoteConfig],
Union[LocalConfig, RemoteConfig]]:
conf = chan.config[LOCAL] if for_us else chan.config[REMOTE]
other_conf = chan.config[LOCAL] if not for_us else chan.config[REMOTE]
return conf, other_conf
def possible_output_idxs_of_htlc_in_ctx(*, chan: 'Channel', pcp: bytes, subject: 'HTLCOwner',
htlc_direction: 'Direction', ctx: Transaction,
htlc: 'UpdateAddHtlc') -> Set[int]:
amount_msat, cltv_abs, payment_hash = htlc.amount_msat, htlc.cltv_abs, htlc.payment_hash
for_us = subject == LOCAL
conf, other_conf = get_ordered_channel_configs(chan=chan, for_us=for_us)
other_revocation_pubkey = derive_blinded_pubkey(other_conf.revocation_basepoint.pubkey, pcp)
other_htlc_pubkey = derive_pubkey(other_conf.htlc_basepoint.pubkey, pcp)
htlc_pubkey = derive_pubkey(conf.htlc_basepoint.pubkey, pcp)
witness_script = make_htlc_output_witness_script(
is_received_htlc=htlc_direction == RECEIVED,
remote_revocation_pubkey=other_revocation_pubkey,
remote_htlc_pubkey=other_htlc_pubkey,
local_htlc_pubkey=htlc_pubkey,
payment_hash=payment_hash,
cltv_abs=cltv_abs,
has_anchors=chan.has_anchors(),
)
htlc_address = redeem_script_to_address('p2wsh', witness_script)
candidates = ctx.get_output_idxs_from_address(htlc_address)
return {output_idx for output_idx in candidates
if ctx.outputs()[output_idx].value == htlc.amount_msat // 1000}
def map_htlcs_to_ctx_output_idxs(*, chan: 'Channel', ctx: Transaction, pcp: bytes,
subject: 'HTLCOwner', ctn: int) -> Dict[Tuple['Direction', 'UpdateAddHtlc'], Tuple[int, int]]:
"""Returns a dict from (htlc_dir, htlc) to (ctx_output_idx, htlc_relative_idx)"""
htlc_to_ctx_output_idx_map = {} # type: Dict[Tuple[Direction, UpdateAddHtlc], int]
unclaimed_ctx_output_idxs = set(range(len(ctx.outputs())))
offered_htlcs = chan.included_htlcs(subject, SENT, ctn=ctn)
offered_htlcs.sort(key=lambda htlc: htlc.cltv_abs)
received_htlcs = chan.included_htlcs(subject, RECEIVED, ctn=ctn)
received_htlcs.sort(key=lambda htlc: htlc.cltv_abs)
for direction, htlcs in zip([SENT, RECEIVED], [offered_htlcs, received_htlcs]):
for htlc in htlcs:
cands = sorted(possible_output_idxs_of_htlc_in_ctx(chan=chan,
pcp=pcp,
subject=subject,
htlc_direction=direction,
ctx=ctx,
htlc=htlc))
for ctx_output_idx in cands:
if ctx_output_idx in unclaimed_ctx_output_idxs:
unclaimed_ctx_output_idxs.discard(ctx_output_idx)
htlc_to_ctx_output_idx_map[(direction, htlc)] = ctx_output_idx
break
# calc htlc_relative_idx
inverse_map = {ctx_output_idx: (direction, htlc)
for ((direction, htlc), ctx_output_idx) in htlc_to_ctx_output_idx_map.items()}
return {inverse_map[ctx_output_idx]: (ctx_output_idx, htlc_relative_idx)
for htlc_relative_idx, ctx_output_idx in enumerate(sorted(inverse_map))}
def make_htlc_tx_with_open_channel(*, chan: 'Channel', pcp: bytes, subject: 'HTLCOwner', ctn: int,
htlc_direction: 'Direction', commit: Transaction, ctx_output_idx: int,
2019-10-23 17:09:41 +02:00
htlc: 'UpdateAddHtlc', name: str = None) -> Tuple[bytes, PartialTransaction]:
amount_msat, cltv_abs, payment_hash = htlc.amount_msat, htlc.cltv_abs, htlc.payment_hash
for_us = subject == LOCAL
conf, other_conf = get_ordered_channel_configs(chan=chan, for_us=for_us)
delayedpubkey = derive_pubkey(conf.delayed_basepoint.pubkey, pcp)
other_revocation_pubkey = derive_blinded_pubkey(other_conf.revocation_basepoint.pubkey, pcp)
other_htlc_pubkey = derive_pubkey(other_conf.htlc_basepoint.pubkey, pcp)
htlc_pubkey = derive_pubkey(conf.htlc_basepoint.pubkey, pcp)
# HTLC-success for the HTLC spending from a received HTLC output
# if we do not receive, and the commitment tx is not for us, they receive, so it is also an HTLC-success
is_htlc_success = htlc_direction == RECEIVED
witness_script_of_htlc_tx_output, htlc_tx_output = make_htlc_tx_output(
amount_msat=amount_msat,
local_feerate=chan.get_feerate(subject, ctn=ctn),
revocationpubkey=other_revocation_pubkey,
local_delayedpubkey=delayedpubkey,
success=is_htlc_success,
to_self_delay=other_conf.to_self_delay,
has_anchors=chan.has_anchors(),
)
witness_script_in = make_htlc_output_witness_script(
is_received_htlc=is_htlc_success,
remote_revocation_pubkey=other_revocation_pubkey,
remote_htlc_pubkey=other_htlc_pubkey,
local_htlc_pubkey=htlc_pubkey,
payment_hash=payment_hash,
cltv_abs=cltv_abs,
has_anchors=chan.has_anchors(),
)
htlc_tx_inputs = make_htlc_tx_inputs(
commit.txid(), ctx_output_idx,
amount_msat=amount_msat,
witness_script=witness_script_in)
if chan.has_anchors():
htlc_tx_inputs[0].nsequence = 1
if is_htlc_success:
cltv_abs = 0
htlc_tx = make_htlc_tx(cltv_abs=cltv_abs, inputs=htlc_tx_inputs, output=htlc_tx_output)
return witness_script_of_htlc_tx_output, htlc_tx
def make_funding_input(local_funding_pubkey: bytes, remote_funding_pubkey: bytes,
2019-10-23 17:09:41 +02:00
funding_pos: int, funding_txid: str, funding_sat: int) -> PartialTxInput:
pubkeys = sorted([local_funding_pubkey.hex(), remote_funding_pubkey.hex()])
# commitment tx input
2019-10-23 17:09:41 +02:00
prevout = TxOutpoint(txid=bfh(funding_txid), out_idx=funding_pos)
c_input = PartialTxInput(prevout=prevout)
ppubkeys = [descriptor.PubkeyProvider.parse(pk) for pk in pubkeys]
multi = descriptor.MultisigDescriptor(pubkeys=ppubkeys, thresh=2, is_sorted=True)
c_input.script_descriptor = descriptor.WSHDescriptor(subdescriptor=multi)
2019-10-23 17:09:41 +02:00
c_input._trusted_value_sats = funding_sat
return c_input
class HTLCOwner(IntEnum):
LOCAL = 1
REMOTE = -LOCAL
def inverted(self) -> 'HTLCOwner':
return -self
def __neg__(self) -> 'HTLCOwner':
return HTLCOwner(super().__neg__())
class Direction(IntEnum):
2020-03-27 19:06:30 +01:00
SENT = -1 # in the context of HTLCs: "offered" HTLCs
RECEIVED = 1 # in the context of HTLCs: "received" HTLCs
SENT = Direction.SENT
RECEIVED = Direction.RECEIVED
LOCAL = HTLCOwner.LOCAL
REMOTE = HTLCOwner.REMOTE
def make_commitment_outputs(
*,
fees_per_participant: Mapping[HTLCOwner, int],
local_amount_msat: int,
remote_amount_msat: int,
local_script: bytes,
remote_script: bytes,
htlcs: List[ScriptHtlc],
dust_limit_sat: int,
has_anchors: bool,
local_anchor_script: Optional[str],
remote_anchor_script: Optional[str]
) -> Tuple[List[PartialTxOutput], List[PartialTxOutput]]:
# determine HTLC outputs and trim below dust to know if anchors need to be included
htlc_outputs = []
for script, htlc in htlcs:
addr = bitcoin.redeem_script_to_address('p2wsh', script)
if htlc.amount_msat // 1000 > dust_limit_sat:
htlc_outputs.append(
PartialTxOutput(
scriptpubkey=address_to_script(addr),
value=htlc.amount_msat // 1000
))
# BOLT-03: "Base commitment transaction fees are extracted from the funder's amount;
# if that amount is insufficient, the entire amount of the funder's output is used."
non_htlc_outputs = []
to_local_amt_msat = local_amount_msat - fees_per_participant[LOCAL]
to_remote_amt_msat = remote_amount_msat - fees_per_participant[REMOTE]
anchor_outputs = []
# if no anchor scripts are set, we ignore anchor outputs, useful when this
# function is used to determine outputs for a collaborative close
if has_anchors and local_anchor_script and remote_anchor_script:
local_pays_anchors = bool(fees_per_participant[LOCAL])
# we always allocate for two anchor outputs even if they are not added
if local_pays_anchors:
to_local_amt_msat -= 2 * FIXED_ANCHOR_SAT * 1000
else:
to_remote_amt_msat -= 2 * FIXED_ANCHOR_SAT * 1000
# include anchors for outputs that materialize, include both if there are HTLCs present
if to_local_amt_msat // 1000 >= dust_limit_sat or htlc_outputs:
anchor_outputs.append(PartialTxOutput(scriptpubkey=local_anchor_script, value=FIXED_ANCHOR_SAT))
if to_remote_amt_msat // 1000 >= dust_limit_sat or htlc_outputs:
anchor_outputs.append(PartialTxOutput(scriptpubkey=remote_anchor_script, value=FIXED_ANCHOR_SAT))
# if funder cannot afford feerate, their output might go negative, so take max(0, x) here
to_local_amt_msat = max(0, to_local_amt_msat)
to_remote_amt_msat = max(0, to_remote_amt_msat)
non_htlc_outputs.append(PartialTxOutput(scriptpubkey=local_script, value=to_local_amt_msat // 1000))
non_htlc_outputs.append(PartialTxOutput(scriptpubkey=remote_script, value=to_remote_amt_msat // 1000))
c_outputs_filtered = list(filter(lambda x: x.value >= dust_limit_sat, non_htlc_outputs + htlc_outputs))
c_outputs = c_outputs_filtered + anchor_outputs
return htlc_outputs, c_outputs
def effective_htlc_tx_weight(success: bool, has_anchors: bool):
# for anchors-zero-fee-htlc we set an effective weight of zero
# we only trim htlcs below dust, as in the anchors commitment format,
# the fees for the hltc transaction don't need to be subtracted from
# the htlc output, but fees are taken from extra attached inputs
if has_anchors:
return HTLC_SUCCESS_WEIGHT_ANCHORS if success else HTLC_TIMEOUT_WEIGHT_ANCHORS
else:
return HTLC_SUCCESS_WEIGHT if success else HTLC_TIMEOUT_WEIGHT
def offered_htlc_trim_threshold_sat(*, dust_limit_sat: int, feerate: int, has_anchors: bool) -> int:
# offered htlcs strictly below this amount will be trimmed (from ctx).
# feerate is in sat/kw
# returns value in sat
weight = effective_htlc_tx_weight(success=False, has_anchors=has_anchors)
return dust_limit_sat + weight * feerate // 1000
def received_htlc_trim_threshold_sat(*, dust_limit_sat: int, feerate: int, has_anchors: bool) -> int:
# received htlcs strictly below this amount will be trimmed (from ctx).
# feerate is in sat/kw
# returns value in sat
weight = effective_htlc_tx_weight(success=True, has_anchors=has_anchors)
return dust_limit_sat + weight * feerate // 1000
def fee_for_htlc_output(*, feerate: int) -> int:
# feerate is in sat/kw
# returns fee in msat
return feerate * HTLC_OUTPUT_WEIGHT
def calc_fees_for_commitment_tx(*, num_htlcs: int, feerate: int,
is_local_initiator: bool, round_to_sat: bool = True, has_anchors: bool) -> Dict['HTLCOwner', int]:
# feerate is in sat/kw
# returns fees in msats
# note: BOLT-02 specifies that msat fees need to be rounded down to sat.
# However, the rounding needs to happen for the total fees, so if the return value
# is to be used as part of additional fee calculation then rounding should be done after that.
if has_anchors:
commitment_tx_weight = COMMITMENT_TX_WEIGHT_ANCHORS
else:
commitment_tx_weight = COMMITMENT_TX_WEIGHT
overall_weight = commitment_tx_weight + num_htlcs * HTLC_OUTPUT_WEIGHT
fee = feerate * overall_weight
if round_to_sat:
fee = fee // 1000 * 1000
return {
LOCAL: fee if is_local_initiator else 0,
REMOTE: fee if not is_local_initiator else 0,
}
2020-03-30 03:49:50 +02:00
def make_commitment(
*,
ctn: int,
local_funding_pubkey: bytes,
remote_funding_pubkey: bytes,
remote_payment_pubkey: bytes,
funder_payment_basepoint: bytes,
fundee_payment_basepoint: bytes,
revocation_pubkey: bytes,
delayed_pubkey: bytes,
to_self_delay: int,
funding_txid: str,
funding_pos: int,
funding_sat: int,
local_amount: int,
remote_amount: int,
dust_limit_sat: int,
fees_per_participant: Mapping[HTLCOwner, int],
htlcs: List[ScriptHtlc],
has_anchors: bool
2020-03-30 03:49:50 +02:00
) -> PartialTransaction:
c_input = make_funding_input(local_funding_pubkey, remote_funding_pubkey,
funding_pos, funding_txid, funding_sat)
obs = get_obscured_ctn(ctn, funder_payment_basepoint, fundee_payment_basepoint)
locktime = (0x20 << 24) + (obs & 0xffffff)
sequence = (0x80 << 24) + (obs >> 24)
2019-10-23 17:09:41 +02:00
c_input.nsequence = sequence
c_inputs = [c_input]
# commitment tx outputs
local_address = make_commitment_output_to_local_address(revocation_pubkey, to_self_delay, delayed_pubkey)
remote_address = make_commitment_output_to_remote_address(remote_payment_pubkey, has_anchors)
local_anchor_address = None
remote_anchor_address = None
if has_anchors:
local_anchor_address = make_commitment_output_to_anchor_address(local_funding_pubkey)
remote_anchor_address = make_commitment_output_to_anchor_address(remote_funding_pubkey)
2020-03-30 03:49:50 +02:00
# note: it is assumed that the given 'htlcs' are all non-dust (dust htlcs already trimmed)
# BOLT-03: "Transaction Input and Output Ordering
# Lexicographic ordering: see BIP69. In the case of identical HTLC outputs,
# the outputs are ordered in increasing cltv_expiry order."
# so we sort by cltv_expiry now; and the later BIP69-sort is assumed to be *stable*
htlcs = list(htlcs)
htlcs.sort(key=lambda x: x.htlc.cltv_abs)
2019-10-23 17:09:41 +02:00
htlc_outputs, c_outputs_filtered = make_commitment_outputs(
fees_per_participant=fees_per_participant,
local_amount_msat=local_amount,
remote_amount_msat=remote_amount,
local_script=address_to_script(local_address),
remote_script=address_to_script(remote_address),
htlcs=htlcs,
dust_limit_sat=dust_limit_sat,
has_anchors=has_anchors,
local_anchor_script=address_to_script(local_anchor_address) if local_anchor_address else None,
remote_anchor_script=address_to_script(remote_anchor_address) if remote_anchor_address else None
)
assert sum(x.value for x in c_outputs_filtered) <= funding_sat, (c_outputs_filtered, funding_sat)
# create commitment tx
2019-10-23 17:09:41 +02:00
tx = PartialTransaction.from_io(c_inputs, c_outputs_filtered, locktime=locktime, version=2)
return tx
def make_commitment_output_to_local_witness_script(
revocation_pubkey: bytes, to_self_delay: int, delayed_pubkey: bytes,
) -> bytes:
assert type(revocation_pubkey) is bytes
assert type(to_self_delay) is int
assert type(delayed_pubkey) is bytes
script = construct_script([
opcodes.OP_IF,
revocation_pubkey,
opcodes.OP_ELSE,
to_self_delay,
opcodes.OP_CHECKSEQUENCEVERIFY,
opcodes.OP_DROP,
delayed_pubkey,
opcodes.OP_ENDIF,
opcodes.OP_CHECKSIG,
])
return script
def make_commitment_output_to_local_address(
revocation_pubkey: bytes, to_self_delay: int, delayed_pubkey: bytes) -> str:
local_script = make_commitment_output_to_local_witness_script(revocation_pubkey, to_self_delay, delayed_pubkey)
return bitcoin.redeem_script_to_address('p2wsh', local_script)
def make_commitment_output_to_remote_witness_script(remote_payment_pubkey: bytes) -> bytes:
assert isinstance(remote_payment_pubkey, bytes)
script = construct_script([
remote_payment_pubkey,
opcodes.OP_CHECKSIGVERIFY,
opcodes.OP_1,
opcodes.OP_CHECKSEQUENCEVERIFY,
])
return script
def make_commitment_output_to_remote_address(remote_payment_pubkey: bytes, has_anchors: bool) -> str:
if has_anchors:
remote_script = make_commitment_output_to_remote_witness_script(remote_payment_pubkey)
2024-11-20 12:05:05 +01:00
return bitcoin.redeem_script_to_address('p2wsh', remote_script)
else:
return bitcoin.pubkey_to_address('p2wpkh', remote_payment_pubkey.hex())
def make_commitment_output_to_anchor_witness_script(funding_pubkey: bytes) -> bytes:
assert isinstance(funding_pubkey, bytes)
script = construct_script([
funding_pubkey,
opcodes.OP_CHECKSIG,
opcodes.OP_IFDUP,
opcodes.OP_NOTIF,
opcodes.OP_16,
opcodes.OP_CHECKSEQUENCEVERIFY,
opcodes.OP_ENDIF,
])
return script
def make_commitment_output_to_anchor_address(funding_pubkey: bytes) -> str:
script = make_commitment_output_to_anchor_witness_script(funding_pubkey)
2024-11-20 12:05:05 +01:00
return bitcoin.redeem_script_to_address('p2wsh', script)
2019-10-23 17:09:41 +02:00
def sign_and_get_sig_string(tx: PartialTransaction, local_config, remote_config):
tx.sign({local_config.multisig_key.pubkey: local_config.multisig_key.privkey})
sig = tx.inputs()[0].sigs_ecdsa[local_config.multisig_key.pubkey]
sig_64 = ecdsa_sig64_from_der_sig(sig[:-1])
return sig_64
def funding_output_script(local_config: 'LocalConfig', remote_config: 'RemoteConfig') -> bytes:
return funding_output_script_from_keys(local_config.multisig_key.pubkey, remote_config.multisig_key.pubkey)
def funding_output_script_from_keys(pubkey1: bytes, pubkey2: bytes) -> bytes:
pubkeys = sorted([pubkey1.hex(), pubkey2.hex()])
return transaction.multisig_script(pubkeys, 2)
def get_obscured_ctn(ctn: int, funder: bytes, fundee: bytes) -> int:
mask = int.from_bytes(sha256(funder + fundee)[-6:], 'big')
return ctn ^ mask
2019-10-23 17:09:41 +02:00
def extract_ctn_from_tx(tx: Transaction, txin_index: int, funder_payment_basepoint: bytes,
fundee_payment_basepoint: bytes) -> int:
tx.deserialize()
locktime = tx.locktime
2019-10-23 17:09:41 +02:00
sequence = tx.inputs()[txin_index].nsequence
obs = ((sequence & 0xffffff) << 24) + (locktime & 0xffffff)
return get_obscured_ctn(obs, funder_payment_basepoint, fundee_payment_basepoint)
def extract_ctn_from_tx_and_chan(tx: Transaction, chan: 'AbstractChannel') -> int:
funder_conf = chan.config[LOCAL] if chan.is_initiator() else chan.config[REMOTE]
fundee_conf = chan.config[LOCAL] if not chan.is_initiator() else chan.config[REMOTE]
2018-09-12 16:17:10 +02:00
return extract_ctn_from_tx(tx, txin_index=0,
funder_payment_basepoint=funder_conf.payment_basepoint.pubkey,
fundee_payment_basepoint=fundee_conf.payment_basepoint.pubkey)
2018-09-12 16:17:10 +02:00
2018-07-26 21:08:25 +02:00
class LnFeatureContexts(enum.Flag):
INIT = enum.auto()
NODE_ANN = enum.auto()
CHAN_ANN_AS_IS = enum.auto()
CHAN_ANN_ALWAYS_ODD = enum.auto()
CHAN_ANN_ALWAYS_EVEN = enum.auto()
INVOICE = enum.auto()
LNFC = LnFeatureContexts
_ln_feature_direct_dependencies = defaultdict(set) # type: Dict[LnFeatures, Set[LnFeatures]]
_ln_feature_contexts = {} # type: Dict[LnFeatures, LnFeatureContexts]
class LnFeatures(IntFlag):
2018-10-09 20:10:26 +02:00
OPTION_DATA_LOSS_PROTECT_REQ = 1 << 0
OPTION_DATA_LOSS_PROTECT_OPT = 1 << 1
_ln_feature_contexts[OPTION_DATA_LOSS_PROTECT_OPT] = (LNFC.INIT | LnFeatureContexts.NODE_ANN)
_ln_feature_contexts[OPTION_DATA_LOSS_PROTECT_REQ] = (LNFC.INIT | LnFeatureContexts.NODE_ANN)
2018-10-09 20:10:26 +02:00
INITIAL_ROUTING_SYNC = 1 << 3
_ln_feature_contexts[INITIAL_ROUTING_SYNC] = LNFC.INIT
2018-10-09 20:10:26 +02:00
OPTION_UPFRONT_SHUTDOWN_SCRIPT_REQ = 1 << 4
OPTION_UPFRONT_SHUTDOWN_SCRIPT_OPT = 1 << 5
_ln_feature_contexts[OPTION_UPFRONT_SHUTDOWN_SCRIPT_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_UPFRONT_SHUTDOWN_SCRIPT_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
2018-10-09 20:10:26 +02:00
GOSSIP_QUERIES_REQ = 1 << 6
GOSSIP_QUERIES_OPT = 1 << 7
_ln_feature_contexts[GOSSIP_QUERIES_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[GOSSIP_QUERIES_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
VAR_ONION_REQ = 1 << 8
VAR_ONION_OPT = 1 << 9
_ln_feature_contexts[VAR_ONION_OPT] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
_ln_feature_contexts[VAR_ONION_REQ] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
GOSSIP_QUERIES_EX_REQ = 1 << 10
GOSSIP_QUERIES_EX_OPT = 1 << 11
_ln_feature_direct_dependencies[GOSSIP_QUERIES_EX_OPT] = {GOSSIP_QUERIES_OPT}
_ln_feature_contexts[GOSSIP_QUERIES_EX_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[GOSSIP_QUERIES_EX_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
2019-12-13 14:07:11 +01:00
OPTION_STATIC_REMOTEKEY_REQ = 1 << 12
OPTION_STATIC_REMOTEKEY_OPT = 1 << 13
_ln_feature_contexts[OPTION_STATIC_REMOTEKEY_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_STATIC_REMOTEKEY_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
PAYMENT_SECRET_REQ = 1 << 14
PAYMENT_SECRET_OPT = 1 << 15
_ln_feature_direct_dependencies[PAYMENT_SECRET_OPT] = {VAR_ONION_OPT}
_ln_feature_contexts[PAYMENT_SECRET_OPT] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
_ln_feature_contexts[PAYMENT_SECRET_REQ] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
BASIC_MPP_REQ = 1 << 16
BASIC_MPP_OPT = 1 << 17
_ln_feature_direct_dependencies[BASIC_MPP_OPT] = {PAYMENT_SECRET_OPT}
_ln_feature_contexts[BASIC_MPP_OPT] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
_ln_feature_contexts[BASIC_MPP_REQ] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
OPTION_SUPPORT_LARGE_CHANNEL_REQ = 1 << 18
OPTION_SUPPORT_LARGE_CHANNEL_OPT = 1 << 19
_ln_feature_contexts[OPTION_SUPPORT_LARGE_CHANNEL_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_SUPPORT_LARGE_CHANNEL_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
OPTION_ANCHOR_OUTPUTS_REQ = 1 << 20
OPTION_ANCHOR_OUTPUTS_OPT = 1 << 21
_ln_feature_direct_dependencies[OPTION_ANCHOR_OUTPUTS_OPT] = {OPTION_STATIC_REMOTEKEY_OPT}
_ln_feature_contexts[OPTION_ANCHOR_OUTPUTS_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_ANCHOR_OUTPUTS_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
OPTION_ANCHORS_ZERO_FEE_HTLC_REQ = 1 << 22
OPTION_ANCHORS_ZERO_FEE_HTLC_OPT = 1 << 23
_ln_feature_direct_dependencies[OPTION_ANCHORS_ZERO_FEE_HTLC_OPT] = {OPTION_STATIC_REMOTEKEY_OPT}
_ln_feature_contexts[OPTION_ANCHORS_ZERO_FEE_HTLC_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_ANCHORS_ZERO_FEE_HTLC_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
# Temporary number.
OPTION_TRAMPOLINE_ROUTING_REQ_ECLAIR = 1 << 148
OPTION_TRAMPOLINE_ROUTING_OPT_ECLAIR = 1 << 149
_ln_feature_contexts[OPTION_TRAMPOLINE_ROUTING_REQ_ECLAIR] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
_ln_feature_contexts[OPTION_TRAMPOLINE_ROUTING_OPT_ECLAIR] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
# We use a different bit because Phoenix cannot do end-to-end multi-trampoline routes
OPTION_TRAMPOLINE_ROUTING_REQ_ELECTRUM = 1 << 150
OPTION_TRAMPOLINE_ROUTING_OPT_ELECTRUM = 1 << 151
_ln_feature_contexts[OPTION_TRAMPOLINE_ROUTING_REQ_ELECTRUM] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
_ln_feature_contexts[OPTION_TRAMPOLINE_ROUTING_OPT_ELECTRUM] = (LNFC.INIT | LNFC.NODE_ANN | LNFC.INVOICE)
OPTION_SHUTDOWN_ANYSEGWIT_REQ = 1 << 26
OPTION_SHUTDOWN_ANYSEGWIT_OPT = 1 << 27
_ln_feature_contexts[OPTION_SHUTDOWN_ANYSEGWIT_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_SHUTDOWN_ANYSEGWIT_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
2022-01-18 14:40:16 +01:00
OPTION_CHANNEL_TYPE_REQ = 1 << 44
OPTION_CHANNEL_TYPE_OPT = 1 << 45
_ln_feature_contexts[OPTION_CHANNEL_TYPE_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_CHANNEL_TYPE_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
OPTION_SCID_ALIAS_REQ = 1 << 46
OPTION_SCID_ALIAS_OPT = 1 << 47
_ln_feature_contexts[OPTION_SCID_ALIAS_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_SCID_ALIAS_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
OPTION_ZEROCONF_REQ = 1 << 50
OPTION_ZEROCONF_OPT = 1 << 51
_ln_feature_direct_dependencies[OPTION_ZEROCONF_OPT] = {OPTION_SCID_ALIAS_OPT}
_ln_feature_contexts[OPTION_ZEROCONF_REQ] = (LNFC.INIT | LNFC.NODE_ANN)
_ln_feature_contexts[OPTION_ZEROCONF_OPT] = (LNFC.INIT | LNFC.NODE_ANN)
2020-11-20 08:51:01 +01:00
def validate_transitive_dependencies(self) -> bool:
# for all even bit set, set corresponding odd bit:
features = self # copy
flags = list_enabled_bits(features)
for flag in flags:
if flag % 2 == 0:
features |= 1 << get_ln_flag_pair_of_bit(flag)
# Check dependencies. We only check that the direct dependencies of each flag set
# are satisfied: this implies that transitive dependencies are also satisfied.
flags = list_enabled_bits(features)
for flag in flags:
for dependency in _ln_feature_direct_dependencies[1 << flag]:
if not (dependency & features):
return False
return True
def for_init_message(self) -> 'LnFeatures':
features = LnFeatures(0)
for flag in list_enabled_bits(self):
if LnFeatureContexts.INIT & _ln_feature_contexts[1 << flag]:
features |= (1 << flag)
return features
def for_node_announcement(self) -> 'LnFeatures':
features = LnFeatures(0)
for flag in list_enabled_bits(self):
if LnFeatureContexts.NODE_ANN & _ln_feature_contexts[1 << flag]:
features |= (1 << flag)
return features
def for_invoice(self) -> 'LnFeatures':
features = LnFeatures(0)
for flag in list_enabled_bits(self):
if LnFeatureContexts.INVOICE & _ln_feature_contexts[1 << flag]:
features |= (1 << flag)
return features
def for_channel_announcement(self) -> 'LnFeatures':
features = LnFeatures(0)
for flag in list_enabled_bits(self):
ctxs = _ln_feature_contexts[1 << flag]
if LnFeatureContexts.CHAN_ANN_AS_IS & ctxs:
features |= (1 << flag)
elif LnFeatureContexts.CHAN_ANN_ALWAYS_EVEN & ctxs:
if flag % 2 == 0:
features |= (1 << flag)
elif LnFeatureContexts.CHAN_ANN_ALWAYS_ODD & ctxs:
if flag % 2 == 0:
flag = get_ln_flag_pair_of_bit(flag)
features |= (1 << flag)
return features
def supports(self, feature: 'LnFeatures') -> bool:
"""Returns whether given feature is enabled.
Helper function that tries to hide the complexity of even/odd bits.
For example, instead of:
bool(myfeatures & LnFeatures.VAR_ONION_OPT or myfeatures & LnFeatures.VAR_ONION_REQ)
you can do:
myfeatures.supports(LnFeatures.VAR_ONION_OPT)
"""
if (1 << (feature.bit_length() - 1)) != feature:
raise ValueError(f"'feature' cannot be a combination of features: {feature}")
if feature.bit_length() % 2 == 0: # feature is OPT
feature_other = feature >> 1
else: # feature is REQ
feature_other = feature << 1
return (self & feature != 0) or (self & feature_other != 0)
def get_names(self) -> Sequence[str]:
r = []
for flag in list_enabled_bits(self):
feature_name = LnFeatures(1 << flag).name
r.append(feature_name or f"bit_{flag}")
return r
lnutil.LnFeatures: limit max feature bit to 10_000 closes https://github.com/spesmilo/electrum/issues/8403 > In Python 3.10 that worked fine, however in Python 3.11 large integer check https://github.com/python/cpython/issues/95778, so now this throws an error. Apparently this change was deemed a security fix and was backported to all supported branches of CPython (going back to 3.7). i.e. it affects ~all versions of python (if sufficiently updated with bugfix patches), not just 3.11 > Some offending node aliases: > ``` > ergvein-fiatchannels > test-mainnet > arakis > ``` The features bits set by some of these nodes: ``` (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 45, 32973, 52973) (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 39, 45, 55, 32973, 52973) ``` > P.S. I see there are a lot of nodes with 253 bytes in their feature vectors. Any idea why that could happen? Note that the valid [merged-into-spec features](https://github.com/lightning/bolts/blob/50b2df24a27879e8329712c275db78876fd022fe/09-features.md) currently only go as high as ~51. However the spec does not specify how to choose feature bits for experimental stuff, so I guess some people are using values in the 50k range. The only limit imposed by the spec on the length of the features bitvector is an implicit one due to the max message size: every msg must be smaller than 65KB, and the features bitvector needs to fit inside the init message, hence it can be up to ~524K bits. (note that the features are not stored in a sparse representation in the init message and in gossip messages, so if many nodes set such high feature bits, that would noticably impact the size of the gossip). ----- Anyway, our current implementation of LnFeatures is subclassing IntFlag, and it looks like it does not work well for such large integers. I've managed to make IntFlags reasonably in python 3.11 by overriding __str__ and __repr__ (note that in cpython it is apparently only the base2<->base10 conversions that are slow, power-of-2 conversions are fast, so we can e.g. use `hex()`). However in python 3.10 and older, enum.py itself seems really slow for bigints, e.g. enum._decompose in python 3.10. Try e.g. this script, which is instant in py3.11 but takes minutes in py3.10: ```py from enum import IntFlag class c(IntFlag): known_flag_1 = 1 << 0 known_flag_2 = 1 << 1 known_flag_3 = 1 << 2 if hasattr(IntFlag, "_numeric_repr_"): # python 3.11+ _numeric_repr_ = hex def __repr__(self): return f"<{self._name_}: {hex(self._value_)}>" def __str__(self): return hex(self._value_) a = c(2**70000-1) q1 = repr(a) q2 = str(a) ``` AFAICT we have two options: either we rewrite LnFeatures so that it does not use IntFlag (and enum.py), or, for the short term as workaround, we could just reject very large feature bits. For now, I've opted to the latter, rejecting feature bits over 10k. (note that another option is bumping the min required python to 3.11, in which case with the overrides added in this commit the performance looks perfectly fine)
2023-05-08 19:37:33 +00:00
if hasattr(IntFlag, "_numeric_repr_"): # python 3.11+
# performance improvement (avoid base2<->base10), see #8403
_numeric_repr_ = hex
def __repr__(self):
# performance improvement (avoid base2<->base10), see #8403
return f"<{self._name_}: {hex(self._value_)}>"
def __str__(self):
# performance improvement (avoid base2<->base10), see #8403
return hex(self._value_)
@stored_as('channel_type', _type=None)
2022-01-18 14:40:16 +01:00
class ChannelType(IntFlag):
OPTION_LEGACY_CHANNEL = 0
OPTION_STATIC_REMOTEKEY = 1 << 12
OPTION_ANCHOR_OUTPUTS = 1 << 20
OPTION_ANCHORS_ZERO_FEE_HTLC_TX = 1 << 22
OPTION_SCID_ALIAS = 1 << 46
OPTION_ZEROCONF = 1 << 50
2022-01-18 14:40:16 +01:00
def discard_unknown_and_check(self):
"""Discards unknown flags and checks flag combination."""
flags = list_enabled_bits(self)
known_channel_types = []
for flag in flags:
channel_type = ChannelType(1 << flag)
if channel_type.name:
known_channel_types.append(channel_type)
final_channel_type = known_channel_types[0]
for channel_type in known_channel_types[1:]:
final_channel_type |= channel_type
final_channel_type.check_combinations()
return final_channel_type
def check_combinations(self):
basic_type = self & ~(ChannelType.OPTION_SCID_ALIAS | ChannelType.OPTION_ZEROCONF)
if basic_type not in [
ChannelType.OPTION_STATIC_REMOTEKEY,
ChannelType.OPTION_ANCHOR_OUTPUTS | ChannelType.OPTION_STATIC_REMOTEKEY,
ChannelType.OPTION_ANCHORS_ZERO_FEE_HTLC_TX | ChannelType.OPTION_STATIC_REMOTEKEY
]:
2022-01-18 14:40:16 +01:00
raise ValueError("Channel type is not a valid flag combination.")
def complies_with_features(self, features: LnFeatures) -> bool:
flags = list_enabled_bits(self)
complies = True
for flag in flags:
feature = LnFeatures(1 << flag)
complies &= features.supports(feature)
return complies
def to_bytes_minimal(self):
# MUST use the smallest bitmap possible to represent the channel type.
bit_length =self.value.bit_length()
byte_length = bit_length // 8 + int(bool(bit_length % 8))
return self.to_bytes(byte_length, byteorder='big')
2022-01-18 14:37:49 +01:00
@property
def name_minimal(self):
if self.name:
return self.name.replace('OPTION_', '')
else:
return str(self)
2022-01-18 14:37:49 +01:00
2022-01-18 14:40:16 +01:00
del LNFC # name is ambiguous without context
# features that are actually implemented and understood in our codebase:
# (note: this is not what we send in e.g. init!)
# (note: specify both OPT and REQ here)
LN_FEATURES_IMPLEMENTED = (
LnFeatures(0)
| LnFeatures.OPTION_DATA_LOSS_PROTECT_OPT | LnFeatures.OPTION_DATA_LOSS_PROTECT_REQ
| LnFeatures.GOSSIP_QUERIES_OPT | LnFeatures.GOSSIP_QUERIES_REQ
| LnFeatures.OPTION_STATIC_REMOTEKEY_OPT | LnFeatures.OPTION_STATIC_REMOTEKEY_REQ
| LnFeatures.VAR_ONION_OPT | LnFeatures.VAR_ONION_REQ
2020-03-24 20:07:00 +01:00
| LnFeatures.PAYMENT_SECRET_OPT | LnFeatures.PAYMENT_SECRET_REQ
| LnFeatures.BASIC_MPP_OPT | LnFeatures.BASIC_MPP_REQ
| LnFeatures.OPTION_TRAMPOLINE_ROUTING_OPT_ELECTRUM | LnFeatures.OPTION_TRAMPOLINE_ROUTING_REQ_ELECTRUM
| LnFeatures.OPTION_SHUTDOWN_ANYSEGWIT_OPT | LnFeatures.OPTION_SHUTDOWN_ANYSEGWIT_REQ
| LnFeatures.OPTION_CHANNEL_TYPE_OPT | LnFeatures.OPTION_CHANNEL_TYPE_REQ
| LnFeatures.OPTION_SCID_ALIAS_OPT | LnFeatures.OPTION_SCID_ALIAS_REQ
)
2018-10-09 20:10:26 +02:00
def get_ln_flag_pair_of_bit(flag_bit: int) -> int:
2018-10-09 20:10:26 +02:00
"""Ln Feature flags are assigned in pairs, one even, one odd. See BOLT-09.
Return the other flag from the pair.
e.g. 6 -> 7
e.g. 7 -> 6
"""
if flag_bit % 2 == 0:
return flag_bit + 1
else:
return flag_bit - 1
class IncompatibleOrInsaneFeatures(Exception): pass
class UnknownEvenFeatureBits(IncompatibleOrInsaneFeatures): pass
class IncompatibleLightningFeatures(IncompatibleOrInsaneFeatures): pass
def ln_compare_features(our_features: 'LnFeatures', their_features: int) -> 'LnFeatures':
"""Returns negotiated features.
Raises IncompatibleLightningFeatures if incompatible.
"""
our_flags = set(list_enabled_bits(our_features))
their_flags = set(list_enabled_bits(their_features))
# check that they have our required features, and disable the optional features they don't have
for flag in our_flags:
if flag not in their_flags and get_ln_flag_pair_of_bit(flag) not in their_flags:
# they don't have this feature we wanted :(
if flag % 2 == 0: # even flags are compulsory
raise IncompatibleLightningFeatures(f"remote does not support {LnFeatures(1 << flag)!r}")
our_features ^= 1 << flag # disable flag
else:
# They too have this flag.
# For easier feature-bit-testing, if this is an even flag, we also
# set the corresponding odd flag now.
if flag % 2 == 0 and our_features & (1 << flag):
our_features |= 1 << get_ln_flag_pair_of_bit(flag)
# check that we have their required features
for flag in their_flags:
if flag not in our_flags and get_ln_flag_pair_of_bit(flag) not in our_flags:
# we don't have this feature they wanted :(
if flag % 2 == 0: # even flags are compulsory
raise IncompatibleLightningFeatures(f"remote wanted feature we don't have: {LnFeatures(1 << flag)!r}")
return our_features
lnutil.LnFeatures: limit max feature bit to 10_000 closes https://github.com/spesmilo/electrum/issues/8403 > In Python 3.10 that worked fine, however in Python 3.11 large integer check https://github.com/python/cpython/issues/95778, so now this throws an error. Apparently this change was deemed a security fix and was backported to all supported branches of CPython (going back to 3.7). i.e. it affects ~all versions of python (if sufficiently updated with bugfix patches), not just 3.11 > Some offending node aliases: > ``` > ergvein-fiatchannels > test-mainnet > arakis > ``` The features bits set by some of these nodes: ``` (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 45, 32973, 52973) (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 39, 45, 55, 32973, 52973) ``` > P.S. I see there are a lot of nodes with 253 bytes in their feature vectors. Any idea why that could happen? Note that the valid [merged-into-spec features](https://github.com/lightning/bolts/blob/50b2df24a27879e8329712c275db78876fd022fe/09-features.md) currently only go as high as ~51. However the spec does not specify how to choose feature bits for experimental stuff, so I guess some people are using values in the 50k range. The only limit imposed by the spec on the length of the features bitvector is an implicit one due to the max message size: every msg must be smaller than 65KB, and the features bitvector needs to fit inside the init message, hence it can be up to ~524K bits. (note that the features are not stored in a sparse representation in the init message and in gossip messages, so if many nodes set such high feature bits, that would noticably impact the size of the gossip). ----- Anyway, our current implementation of LnFeatures is subclassing IntFlag, and it looks like it does not work well for such large integers. I've managed to make IntFlags reasonably in python 3.11 by overriding __str__ and __repr__ (note that in cpython it is apparently only the base2<->base10 conversions that are slow, power-of-2 conversions are fast, so we can e.g. use `hex()`). However in python 3.10 and older, enum.py itself seems really slow for bigints, e.g. enum._decompose in python 3.10. Try e.g. this script, which is instant in py3.11 but takes minutes in py3.10: ```py from enum import IntFlag class c(IntFlag): known_flag_1 = 1 << 0 known_flag_2 = 1 << 1 known_flag_3 = 1 << 2 if hasattr(IntFlag, "_numeric_repr_"): # python 3.11+ _numeric_repr_ = hex def __repr__(self): return f"<{self._name_}: {hex(self._value_)}>" def __str__(self): return hex(self._value_) a = c(2**70000-1) q1 = repr(a) q2 = str(a) ``` AFAICT we have two options: either we rewrite LnFeatures so that it does not use IntFlag (and enum.py), or, for the short term as workaround, we could just reject very large feature bits. For now, I've opted to the latter, rejecting feature bits over 10k. (note that another option is bumping the min required python to 3.11, in which case with the overrides added in this commit the performance looks perfectly fine)
2023-05-08 19:37:33 +00:00
if hasattr(sys, "get_int_max_str_digits"):
# check that the user or other library has not lowered the limit (from default)
assert sys.get_int_max_str_digits() >= 4300, f"sys.get_int_max_str_digits() too low: {sys.get_int_max_str_digits()}"
def validate_features(features: int) -> LnFeatures:
"""Raises IncompatibleOrInsaneFeatures if
- a mandatory feature is listed that we don't recognize, or
- the features are inconsistent
lnutil.LnFeatures: limit max feature bit to 10_000 closes https://github.com/spesmilo/electrum/issues/8403 > In Python 3.10 that worked fine, however in Python 3.11 large integer check https://github.com/python/cpython/issues/95778, so now this throws an error. Apparently this change was deemed a security fix and was backported to all supported branches of CPython (going back to 3.7). i.e. it affects ~all versions of python (if sufficiently updated with bugfix patches), not just 3.11 > Some offending node aliases: > ``` > ergvein-fiatchannels > test-mainnet > arakis > ``` The features bits set by some of these nodes: ``` (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 45, 32973, 52973) (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 39, 45, 55, 32973, 52973) ``` > P.S. I see there are a lot of nodes with 253 bytes in their feature vectors. Any idea why that could happen? Note that the valid [merged-into-spec features](https://github.com/lightning/bolts/blob/50b2df24a27879e8329712c275db78876fd022fe/09-features.md) currently only go as high as ~51. However the spec does not specify how to choose feature bits for experimental stuff, so I guess some people are using values in the 50k range. The only limit imposed by the spec on the length of the features bitvector is an implicit one due to the max message size: every msg must be smaller than 65KB, and the features bitvector needs to fit inside the init message, hence it can be up to ~524K bits. (note that the features are not stored in a sparse representation in the init message and in gossip messages, so if many nodes set such high feature bits, that would noticably impact the size of the gossip). ----- Anyway, our current implementation of LnFeatures is subclassing IntFlag, and it looks like it does not work well for such large integers. I've managed to make IntFlags reasonably in python 3.11 by overriding __str__ and __repr__ (note that in cpython it is apparently only the base2<->base10 conversions that are slow, power-of-2 conversions are fast, so we can e.g. use `hex()`). However in python 3.10 and older, enum.py itself seems really slow for bigints, e.g. enum._decompose in python 3.10. Try e.g. this script, which is instant in py3.11 but takes minutes in py3.10: ```py from enum import IntFlag class c(IntFlag): known_flag_1 = 1 << 0 known_flag_2 = 1 << 1 known_flag_3 = 1 << 2 if hasattr(IntFlag, "_numeric_repr_"): # python 3.11+ _numeric_repr_ = hex def __repr__(self): return f"<{self._name_}: {hex(self._value_)}>" def __str__(self): return hex(self._value_) a = c(2**70000-1) q1 = repr(a) q2 = str(a) ``` AFAICT we have two options: either we rewrite LnFeatures so that it does not use IntFlag (and enum.py), or, for the short term as workaround, we could just reject very large feature bits. For now, I've opted to the latter, rejecting feature bits over 10k. (note that another option is bumping the min required python to 3.11, in which case with the overrides added in this commit the performance looks perfectly fine)
2023-05-08 19:37:33 +00:00
For convenience, returns the parsed features.
"""
lnutil.LnFeatures: limit max feature bit to 10_000 closes https://github.com/spesmilo/electrum/issues/8403 > In Python 3.10 that worked fine, however in Python 3.11 large integer check https://github.com/python/cpython/issues/95778, so now this throws an error. Apparently this change was deemed a security fix and was backported to all supported branches of CPython (going back to 3.7). i.e. it affects ~all versions of python (if sufficiently updated with bugfix patches), not just 3.11 > Some offending node aliases: > ``` > ergvein-fiatchannels > test-mainnet > arakis > ``` The features bits set by some of these nodes: ``` (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 45, 32973, 52973) (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 39, 45, 55, 32973, 52973) ``` > P.S. I see there are a lot of nodes with 253 bytes in their feature vectors. Any idea why that could happen? Note that the valid [merged-into-spec features](https://github.com/lightning/bolts/blob/50b2df24a27879e8329712c275db78876fd022fe/09-features.md) currently only go as high as ~51. However the spec does not specify how to choose feature bits for experimental stuff, so I guess some people are using values in the 50k range. The only limit imposed by the spec on the length of the features bitvector is an implicit one due to the max message size: every msg must be smaller than 65KB, and the features bitvector needs to fit inside the init message, hence it can be up to ~524K bits. (note that the features are not stored in a sparse representation in the init message and in gossip messages, so if many nodes set such high feature bits, that would noticably impact the size of the gossip). ----- Anyway, our current implementation of LnFeatures is subclassing IntFlag, and it looks like it does not work well for such large integers. I've managed to make IntFlags reasonably in python 3.11 by overriding __str__ and __repr__ (note that in cpython it is apparently only the base2<->base10 conversions that are slow, power-of-2 conversions are fast, so we can e.g. use `hex()`). However in python 3.10 and older, enum.py itself seems really slow for bigints, e.g. enum._decompose in python 3.10. Try e.g. this script, which is instant in py3.11 but takes minutes in py3.10: ```py from enum import IntFlag class c(IntFlag): known_flag_1 = 1 << 0 known_flag_2 = 1 << 1 known_flag_3 = 1 << 2 if hasattr(IntFlag, "_numeric_repr_"): # python 3.11+ _numeric_repr_ = hex def __repr__(self): return f"<{self._name_}: {hex(self._value_)}>" def __str__(self): return hex(self._value_) a = c(2**70000-1) q1 = repr(a) q2 = str(a) ``` AFAICT we have two options: either we rewrite LnFeatures so that it does not use IntFlag (and enum.py), or, for the short term as workaround, we could just reject very large feature bits. For now, I've opted to the latter, rejecting feature bits over 10k. (note that another option is bumping the min required python to 3.11, in which case with the overrides added in this commit the performance looks perfectly fine)
2023-05-08 19:37:33 +00:00
if features.bit_length() > 10_000:
# This is an implementation-specific limit for how high feature bits we allow.
# Needed as LnFeatures subclasses IntFlag, and uses ints internally.
# See https://docs.python.org/3/library/stdtypes.html#integer-string-conversion-length-limitation
raise IncompatibleOrInsaneFeatures(f"features bitvector too large: {features.bit_length()=} > 10_000")
features = LnFeatures(features)
enabled_features = list_enabled_bits(features)
for fbit in enabled_features:
if (1 << fbit) & LN_FEATURES_IMPLEMENTED == 0 and fbit % 2 == 0:
raise UnknownEvenFeatureBits(fbit)
2020-11-20 08:51:01 +01:00
if not features.validate_transitive_dependencies():
raise IncompatibleOrInsaneFeatures(f"not all transitive dependencies are set. "
f"features={features}")
lnutil.LnFeatures: limit max feature bit to 10_000 closes https://github.com/spesmilo/electrum/issues/8403 > In Python 3.10 that worked fine, however in Python 3.11 large integer check https://github.com/python/cpython/issues/95778, so now this throws an error. Apparently this change was deemed a security fix and was backported to all supported branches of CPython (going back to 3.7). i.e. it affects ~all versions of python (if sufficiently updated with bugfix patches), not just 3.11 > Some offending node aliases: > ``` > ergvein-fiatchannels > test-mainnet > arakis > ``` The features bits set by some of these nodes: ``` (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 45, 32973, 52973) (1, 7, 8, 11, 13, 14, 17, 19, 23, 27, 39, 45, 55, 32973, 52973) ``` > P.S. I see there are a lot of nodes with 253 bytes in their feature vectors. Any idea why that could happen? Note that the valid [merged-into-spec features](https://github.com/lightning/bolts/blob/50b2df24a27879e8329712c275db78876fd022fe/09-features.md) currently only go as high as ~51. However the spec does not specify how to choose feature bits for experimental stuff, so I guess some people are using values in the 50k range. The only limit imposed by the spec on the length of the features bitvector is an implicit one due to the max message size: every msg must be smaller than 65KB, and the features bitvector needs to fit inside the init message, hence it can be up to ~524K bits. (note that the features are not stored in a sparse representation in the init message and in gossip messages, so if many nodes set such high feature bits, that would noticably impact the size of the gossip). ----- Anyway, our current implementation of LnFeatures is subclassing IntFlag, and it looks like it does not work well for such large integers. I've managed to make IntFlags reasonably in python 3.11 by overriding __str__ and __repr__ (note that in cpython it is apparently only the base2<->base10 conversions that are slow, power-of-2 conversions are fast, so we can e.g. use `hex()`). However in python 3.10 and older, enum.py itself seems really slow for bigints, e.g. enum._decompose in python 3.10. Try e.g. this script, which is instant in py3.11 but takes minutes in py3.10: ```py from enum import IntFlag class c(IntFlag): known_flag_1 = 1 << 0 known_flag_2 = 1 << 1 known_flag_3 = 1 << 2 if hasattr(IntFlag, "_numeric_repr_"): # python 3.11+ _numeric_repr_ = hex def __repr__(self): return f"<{self._name_}: {hex(self._value_)}>" def __str__(self): return hex(self._value_) a = c(2**70000-1) q1 = repr(a) q2 = str(a) ``` AFAICT we have two options: either we rewrite LnFeatures so that it does not use IntFlag (and enum.py), or, for the short term as workaround, we could just reject very large feature bits. For now, I've opted to the latter, rejecting feature bits over 10k. (note that another option is bumping the min required python to 3.11, in which case with the overrides added in this commit the performance looks perfectly fine)
2023-05-08 19:37:33 +00:00
return features
2020-03-24 20:07:00 +01:00
def derive_payment_secret_from_payment_preimage(payment_preimage: bytes) -> bytes:
"""Returns secret to be put into invoice.
Derivation is deterministic, based on the preimage.
Crucially the payment_hash must be derived in an independent way from this.
"""
# Note that this could be random data too, but then we would need to store it.
# We derive it identically to clightning, so that we cannot be distinguished:
# https://github.com/ElementsProject/lightning/blob/faac4b28adee5221e83787d64cd5d30b16b62097/lightningd/invoice.c#L115
modified = bytearray(payment_preimage)
modified[0] ^= 1
return sha256(bytes(modified))
def get_compressed_pubkey_from_bech32(bech32_pubkey: str) -> bytes:
decoded_bech32 = segwit_addr.bech32_decode(bech32_pubkey)
hrp = decoded_bech32.hrp
data_5bits = decoded_bech32.data
if decoded_bech32.encoding is None:
raise ValueError("Bad bech32 checksum")
if decoded_bech32.encoding != segwit_addr.Encoding.BECH32:
raise ValueError("Bad bech32 encoding: must be using vanilla BECH32")
if hrp != 'ln':
raise Exception('unexpected hrp: {}'.format(hrp))
data_8bits = segwit_addr.convertbits(data_5bits, 5, 8, False)
# pad with zeroes
COMPRESSED_PUBKEY_LENGTH = 33
data_8bits = data_8bits + ((COMPRESSED_PUBKEY_LENGTH - len(data_8bits)) * [0])
return bytes(data_8bits)
2018-08-01 18:30:35 +02:00
def make_closing_tx(local_funding_pubkey: bytes, remote_funding_pubkey: bytes,
2019-10-23 17:09:41 +02:00
funding_txid: str, funding_pos: int, funding_sat: int,
outputs: List[PartialTxOutput]) -> PartialTransaction:
c_input = make_funding_input(local_funding_pubkey, remote_funding_pubkey,
funding_pos, funding_txid, funding_sat)
2019-10-23 17:09:41 +02:00
c_input.nsequence = 0xFFFF_FFFF
tx = PartialTransaction.from_io([c_input], outputs, locktime=0, version=2)
return tx
# key derivation
# originally based on lnd/keychain/derivation.go
# notes:
# - Add a new path for each use case. Do not reuse existing paths.
# (to avoid having to carefully consider if reuse would be safe)
# - Always prefer to use hardened derivation for new paths you add.
# (to avoid having to carefully consider if unhardened would be safe)
class LnKeyFamily(IntEnum):
MULTISIG = 0 | BIP32_PRIME
REVOCATION_BASE = 1 | BIP32_PRIME
HTLC_BASE = 2 | BIP32_PRIME
PAYMENT_BASE = 3 | BIP32_PRIME
DELAY_BASE = 4 | BIP32_PRIME
REVOCATION_ROOT = 5 | BIP32_PRIME
NODE_KEY = 6
BACKUP_CIPHER = 7 | BIP32_PRIME
PAYMENT_SECRET_KEY = 8 | BIP32_PRIME
NOSTR_KEY = 9 | BIP32_PRIME
2020-02-21 12:18:28 +01:00
def generate_keypair(node: BIP32Node, key_family: LnKeyFamily) -> Keypair:
node2 = node.subkey_at_private_derivation([key_family, 0, 0])
2020-02-21 12:18:28 +01:00
k = node2.eckey.get_secret_bytes()
cK = ecc.ECPrivkey(k).get_public_key_bytes()
return Keypair(cK, k)
def generate_random_keypair() -> Keypair:
import secrets
k = secrets.token_bytes(32)
cK = ecc.ECPrivkey(k).get_public_key_bytes()
return Keypair(cK, k)
NUM_MAX_HOPS_IN_PAYMENT_PATH = 20
NUM_MAX_EDGES_IN_PAYMENT_PATH = NUM_MAX_HOPS_IN_PAYMENT_PATH
@attr.s(frozen=True)
class UpdateAddHtlc:
amount_msat = attr.ib(type=int, kw_only=True)
payment_hash = attr.ib(type=bytes, kw_only=True, converter=hex_to_bytes, repr=lambda val: val.hex())
cltv_abs = attr.ib(type=int, kw_only=True)
timestamp = attr.ib(type=int, kw_only=True)
htlc_id = attr.ib(type=int, kw_only=True, default=None)
@stored_in('adds', tuple)
def from_tuple(amount_msat, payment_hash, cltv_abs, htlc_id, timestamp) -> 'UpdateAddHtlc':
return UpdateAddHtlc(
amount_msat=amount_msat,
payment_hash=payment_hash,
cltv_abs=cltv_abs,
htlc_id=htlc_id,
timestamp=timestamp)
def to_json(self):
return (self.amount_msat, self.payment_hash, self.cltv_abs, self.htlc_id, self.timestamp)
class OnionFailureCodeMetaFlag(IntFlag):
BADONION = 0x8000
PERM = 0x4000
NODE = 0x2000
UPDATE = 0x1000
class PaymentFeeBudget(NamedTuple):
fee_msat: int
# The cltv budget covers the cost of route to get to the destination, but excluding the
# cltv-delta the destination wants for itself. (e.g. "min_final_cltv_delta" is excluded)
cltv: int # this is cltv-delta-like, no absolute heights here!
#num_htlc: int
@classmethod
def default(cls, *, invoice_amount_msat: int, config: 'SimpleConfig') -> 'PaymentFeeBudget':
millionths_orig = config.LIGHTNING_PAYMENT_FEE_MAX_MILLIONTHS
millionths = min(max(0, millionths_orig), 250_000) # clamp into [0, 25%]
cutoff_orig = config.LIGHTNING_PAYMENT_FEE_CUTOFF_MSAT
cutoff = min(max(0, cutoff_orig), 10_000_000) # clamp into [0, 10k sat]
if millionths != millionths_orig:
_logger.warning(
f"PaymentFeeBudget. found insane fee millionths in config. "
f"clamped: {millionths_orig}->{millionths}")
if cutoff != cutoff_orig:
_logger.warning(
f"PaymentFeeBudget. found insane fee cutoff in config. "
f"clamped: {cutoff_orig}->{cutoff}")
# for small payments, fees <= constant cutoff are fine
# for large payments, the max fee is percentage-based
fee_msat = invoice_amount_msat * millionths // 1_000_000
fee_msat = max(fee_msat, cutoff)
return PaymentFeeBudget(
fee_msat=fee_msat,
cltv=NBLOCK_CLTV_DELTA_TOO_FAR_INTO_FUTURE,
)