rust-lightning
3 years agoEncapsulate feature flag checking and manipulation
Jeffrey Czyz [Fri, 10 Apr 2020 00:08:48 +0000 (17:08 -0700)]
Encapsulate feature flag checking and manipulation

Each feature is represented by two bits within Features' flags field.
Working with these flags requires bitwise operations, which can be error
prone. Rather than directly checking and manipulating bits, encapsulate
the bits within each feature trait and provide mechanisms for doing so.

This removes the need to comment on which features correspond to bitwise
expressions since the expressions use feature trait identifiers instead.

With this approach, byte literals and expressions can be evaluated at
compile time still. However, for these cases, knowing which byte within
the flags that a feature corresponds to still must be determined by the
implementor.

Remove the special case where initial_routing_sync has no even bit. Now,
it (bit 2) is considered known by the implementation.

3 years agoSet initial_routing_sync in InitFeatures
Jeffrey Czyz [Thu, 23 Apr 2020 16:47:15 +0000 (09:47 -0700)]
Set initial_routing_sync in InitFeatures

The initial_routing_sync feature is set by peer_handler whenever a full
sync of the network graph is desired. It is not explicitly set when
creating features with InitFeatures::supported().

An upcoming refactor will change supported() to known(), which will
return all features known by the implementation. Thus, the
initial_routing_sync flag will need to be set by default. This commit
makes the behavior change ahead of the refactor.

4 years agoUnset upfront_shutdown_script using bit clearing
Jeffrey Czyz [Fri, 10 Apr 2020 18:36:47 +0000 (11:36 -0700)]
Unset upfront_shutdown_script using bit clearing

The test_upfront_shutdown_script functional test clears this feature
flag. However, the method used to clear the flag is implemented by bit
toggling. Thus, if the flag is not set the method would actually set it.
Implement the method using bit clearing instead.

4 years agoMerge pull request #608 from TheBlueMatt/2020-04-0.0.11
Matt Corallo [Sat, 25 Apr 2020 16:42:15 +0000 (16:42 +0000)]
Merge pull request #608 from TheBlueMatt/2020-04-0.0.11

0.0.11

4 years agoMerge pull request #598 from TheBlueMatt/2020-04-559-cleanups
Matt Corallo [Sat, 25 Apr 2020 16:42:03 +0000 (16:42 +0000)]
Merge pull request #598 from TheBlueMatt/2020-04-559-cleanups

Clean up ChannelKeys API

4 years agoBump versions to 0.0.11/net-tokio 0.0.3 2020-04-0.0.11
Matt Corallo [Sat, 25 Apr 2020 02:10:00 +0000 (22:10 -0400)]
Bump versions to 0.0.11/net-tokio 0.0.3

4 years agoRemove trivial assertions in LocalCommitmentTransaction fns 2020-04-559-cleanups
Matt Corallo [Fri, 24 Apr 2020 19:31:33 +0000 (15:31 -0400)]
Remove trivial assertions in LocalCommitmentTransaction fns

We don't need to assert that transaction structure is what we
expect when the transaction is created by a function twenty lines
up in the same file.

4 years agoRemove TODOs from documentation in keysinterface
Matt Corallo [Mon, 20 Apr 2020 05:07:18 +0000 (01:07 -0400)]
Remove TODOs from documentation in keysinterface

We should never be exposing our own TODOs to the world.

4 years agoDon't modify LocalCommitmemntTransaction after construction
Matt Corallo [Thu, 23 Apr 2020 19:43:21 +0000 (15:43 -0400)]
Don't modify LocalCommitmemntTransaction after construction

Instead of adding signatures to LocalCommitmentTransactions, we
instead leave them unsigned and use them to construct signed
Transactions when we want them. This cleans up the guts of
LocalCommitmentTransaction enough that we can, and do, expose its
state to the world, allowing external signers to have a basic
awareness of what they're signing.

4 years agoBatch-sign local HTLC txn with a well-doc'd API, returning sigs
Matt Corallo [Mon, 20 Apr 2020 02:59:53 +0000 (22:59 -0400)]
Batch-sign local HTLC txn with a well-doc'd API, returning sigs

1107ab06c33bd360bdee7ee64f4b690e753003f6 introduced an API to have a
ChannelKeys implementer sign HTLC transactions by calling into the
LocalCommitmentTransaction object, which would then store the tx.

This API was incredibly awkward, both because it required an
external signer trust our own internal interfaces, but also because
it didn't allow for any inspection of what was about to be signed.

Further, it signed the HTLC transactions one-by-one in a somewhat
inefficient way, and there isn't a clear way to resolve this (as
the which-HTLC parameter has to refer to something in between the
HTLC's arbitrary index, and its index in the commitment tx, which
has "holes" for the non-HTLC outputs and skips some HTLCs).

We replace it with a new function in ChannelKeys which allows us
to sign all HTLCs in a given commitment transaction (which allows
for a bit more effeciency on the signers' part, as well as
sidesteps the which-HTLC issue). This may also simplify the signer
implementation as we will always want to sign all HTLCs spending a
given commitment transaction at once anyway.

We also de-mut the LocalCommitmentTransaction passed to the
ChanKeys, instead opting to make LocalCommitmentTransaction const
and avoid storing any new HTLC-related data in it.

4 years agoReturn Result<Signature> instead of modifying args in ChannelKeys
Matt Corallo [Sun, 19 Apr 2020 21:26:41 +0000 (17:26 -0400)]
Return Result<Signature> instead of modifying args in ChannelKeys

This cleans up sign_local_commitment somewhat by returning a
Result<Signaure, ()> over the local commitment transaction instead
of modifying the struct which was passed in.

This is the first step in making LocalCommitmentTransaction a
completely pub struct, using it just to communicate enough
information to the user to allow them to construct a signaure
instead of having it contain a bunch of logic.

This should make it much easier to implement a custom ChannelKeys
by disconnecting the local commitment transaction signing from our
own datastructures.

4 years agoTrack signing of local txn in channelmonitor and refuse updates
Matt Corallo [Sun, 19 Apr 2020 18:15:56 +0000 (14:15 -0400)]
Track signing of local txn in channelmonitor and refuse updates

In e46e183084ed858f41aa304acd78503aea1a96ed we began tracking
whether a local commitment transaction had been signed and
broadcast in OnchainTxHandler, refusing to update the local
commitment transaction state in the ChannelMonitor on that basis.

This is fine, except that it doesn't make a lot of sense to store
the full local transaction state in OnchainTxHandler - we should be
providing it the unsigned local transaction at the time we wish to
broadcast and no more (just like we do all other transaction data).

4 years agoDry-up InputMaterial::Funding
Antoine Riard [Tue, 21 Apr 2020 01:19:00 +0000 (21:19 -0400)]
Dry-up InputMaterial::Funding

As channel_value last usage was for computing feerate but as this
one is static per-commitment and will always-be following specification,
we remove it.

4 years agoMerge pull request #579 from ariard/2020-04-sanitize-cltv-delay
Matt Corallo [Fri, 24 Apr 2020 22:50:45 +0000 (22:50 +0000)]
Merge pull request #579 from ariard/2020-04-sanitize-cltv-delay

Sanititze and document incoming HTLC cltv_expiry handling

4 years agoSanitize outgoing HTLC cltv_value
Antoine Riard [Fri, 10 Apr 2020 02:04:30 +0000 (22:04 -0400)]
Sanitize outgoing HTLC cltv_value

4 years agoDocument exactly our CLTV sanitization policy for final incoming HTLCs
Antoine Riard [Fri, 10 Apr 2020 01:51:29 +0000 (21:51 -0400)]
Document exactly our CLTV sanitization policy for final incoming HTLCs

We want to avoid a third-party channel closure, where a random node
by sending us a payment expiring at current height, would trigger our
onchain logic to close the channel due to a near-expiration.

4 years agoMerge pull request #607 from TheBlueMatt/2020-04-more-cleanups
Matt Corallo [Fri, 24 Apr 2020 22:17:53 +0000 (22:17 +0000)]
Merge pull request #607 from TheBlueMatt/2020-04-more-cleanups

CI and rustc warning cleanup

4 years agoInclude color when building fuzz targets in CI 2020-04-more-cleanups
Matt Corallo [Fri, 24 Apr 2020 21:35:18 +0000 (17:35 -0400)]
Include color when building fuzz targets in CI

4 years agoDisable timeout for full_stack_target runs in CI
Matt Corallo [Fri, 24 Apr 2020 20:56:20 +0000 (16:56 -0400)]
Disable timeout for full_stack_target runs in CI

It seems we've recently been seeing sporadic long-running
full_stack_target cases when running honggfuzz in CI. These
shouldn't be killed (as its possible they hit an error or a deadlock,
especially since the longest-running tests probably have the most
coverage).

4 years agoSilence fuzz build unused import warnings
Matt Corallo [Fri, 24 Apr 2020 20:24:25 +0000 (16:24 -0400)]
Silence fuzz build unused import warnings

4 years agoAddress new rustc warnings.
Matt Corallo [Fri, 24 Apr 2020 18:40:01 +0000 (14:40 -0400)]
Address new rustc warnings.

4 years agoFix step naming in fuzz tests in GitHub Actions
Matt Corallo [Fri, 24 Apr 2020 18:22:31 +0000 (14:22 -0400)]
Fix step naming in fuzz tests in GitHub Actions

4 years agoRun Github Actions on PRs, too
Matt Corallo [Fri, 24 Apr 2020 20:22:44 +0000 (16:22 -0400)]
Run Github Actions on PRs, too

4 years agoMerge pull request #593 from TheBlueMatt/2020-04-par-fuzz-check
Matt Corallo [Fri, 24 Apr 2020 20:52:19 +0000 (20:52 +0000)]
Merge pull request #593 from TheBlueMatt/2020-04-par-fuzz-check

Thread fuzz test cases

4 years agoDocument exactly our CLTV sanitization policy for incoming HTLCs
Antoine Riard [Fri, 10 Apr 2020 01:22:16 +0000 (21:22 -0400)]
Document exactly our CLTV sanitization policy for incoming HTLCs

4 years agoThread fuzz test cases 2020-04-par-fuzz-check
Matt Corallo [Fri, 21 Feb 2020 01:11:40 +0000 (20:11 -0500)]
Thread fuzz test cases

Adds threading (with logger capture) to fuzz targets so that we can
more effeciently test a failing fuzz corpus on new code changes.

4 years agoMerge pull request #587 from TheBlueMatt/2020-04-mpp-timeout
Matt Corallo [Fri, 24 Apr 2020 19:07:22 +0000 (19:07 +0000)]
Merge pull request #587 from TheBlueMatt/2020-04-mpp-timeout

Time out HTLCs before they expire

4 years agoAdd test for timing out HTLCs which are in the holding cell 2020-04-mpp-timeout
Matt Corallo [Mon, 20 Apr 2020 20:55:31 +0000 (16:55 -0400)]
Add test for timing out HTLCs which are in the holding cell

4 years agoTime out AwatingRemoteRAA outgoing HTLCs when we reach cltv_expiry
Antoine Riard [Sat, 8 Dec 2018 03:09:58 +0000 (22:09 -0500)]
Time out AwatingRemoteRAA outgoing HTLCs when we reach cltv_expiry

In case of committing out-of-time outgoing HTLCs, we force
ourselves to close the channel to avoid remote peer claims on a
non-backed HTLC

4 years agoAdd test for partial-send MPP due to monitor update failure
Matt Corallo [Tue, 14 Apr 2020 01:04:03 +0000 (21:04 -0400)]
Add test for partial-send MPP due to monitor update failure

Relatively simple test that, after a monitor update fails, we get
the right return value and continue with the bits of the MPP that
did not send after the monitor updating is restored.

4 years agoAdd a test for timeout'ing HTLCs which claim to be a part of an MPP
Matt Corallo [Thu, 19 Mar 2020 04:34:15 +0000 (00:34 -0400)]
Add a test for timeout'ing HTLCs which claim to be a part of an MPP

This is a key test for our automatic HTLC time-out logic, as it
ensures we don't allow an HTLC which indicates we should wait for
additional HTLCs before responding to cause us to force-close a
channel due to HTLC near-timeout.

4 years agoExpand expect_payment_failed!() to take error codes and use it more
Matt Corallo [Mon, 20 Apr 2020 19:46:35 +0000 (15:46 -0400)]
Expand expect_payment_failed!() to take error codes and use it more

expect_payment_failed!() was introduced after many of the tests
which could use it were written, so we take this opportunity to
switch them over now, increasing test coverage slightly by always
checking the payment hash expected.

4 years agoTime out incoming HTLCs when we reach cltv_expiry (+ test)
Matt Corallo [Thu, 9 Jan 2020 19:09:25 +0000 (14:09 -0500)]
Time out incoming HTLCs when we reach cltv_expiry (+ test)

We only do this for incoming HTLCs directly as we rely on channel
closure and HTLC-Timeout broadcast to fail any HTLCs which we
relayed onwards where our next-hop doesn't update_fail in time.

4 years agoAdd a constant and docs for when we should fail an HTLC.
Matt Corallo [Fri, 24 Apr 2020 00:39:01 +0000 (20:39 -0400)]
Add a constant and docs for when we should fail an HTLC.

4 years agoMerge pull request #601 from D4nte/ci-in-github-action
Matt Corallo [Fri, 24 Apr 2020 18:14:55 +0000 (18:14 +0000)]
Merge pull request #601 from D4nte/ci-in-github-action

Add GitHub Action to build the project

4 years agoAdd a pop of colour
Franck Royer [Fri, 24 Apr 2020 05:06:15 +0000 (15:06 +1000)]
Add a pop of colour

4 years agoAdd CI using GitHub Action
Franck Royer [Tue, 21 Apr 2020 10:24:39 +0000 (20:24 +1000)]
Add CI using GitHub Action

Same setup than Travis except for removing
`rm -f target/debug/lightning-*` as I do not believe
such file would exist on a fresh run.

I have not setup caching at this stage. The library is
small so I don't think it'd be that necessary/helpful.

I'd recommend to let both CI run for a bit to compare
performance and stability. The CI setup is straightforward
so I do not foresee any issue with GitHub actions.

Once happy, Travis file can be removed and branch
protection checks can be updated to block on the GitHub
actions.

You can also check the [Coverage report](https://codecov.io/gh/D4nte/rust-lightning/tree/752a58bc0441a49a0513f2cad979ad9e2621312a/lightning/src/chain) to ensure it is as expected.

4 years agoMerge pull request #597 from TheBlueMatt/2020-04-more-chanmon-cleanups
Matt Corallo [Thu, 23 Apr 2020 18:15:47 +0000 (18:15 +0000)]
Merge pull request #597 from TheBlueMatt/2020-04-more-chanmon-cleanups

Few more ChannelMonitor Cleanups

4 years agoDe-Option<> current_local_signed_commitment_tx in ChannelMonitor 2020-04-more-chanmon-cleanups
Matt Corallo [Sun, 19 Apr 2020 03:13:18 +0000 (23:13 -0400)]
De-Option<> current_local_signed_commitment_tx in ChannelMonitor

Since we now are always initialised with an initial local commitment
transaction available now, we might as well take advantage of it and
stop using an Option<> where we don't need to.

4 years agoDelay creating outbound ChannelMonitor until funding_signed rcpt
Matt Corallo [Sat, 18 Apr 2020 20:35:01 +0000 (16:35 -0400)]
Delay creating outbound ChannelMonitor until funding_signed rcpt

Previously, we created the initial ChannelMonitor on outbound
channels when we generated the funding_created message. This was
somewhat unnecessary as, at that time, we hadn't yet received
clearance to broadcast our initial funding transaction, and thus
there should never be any use for a ChannelMonitor. It also
complicated ChannelMonitor a bit as, at this point, we didn't have
an initial local commitment transaction.

By moving the creation of the initial ChannelMonitor to when we
receive our counterparty's funding_signed, we can ensure that any
ChannelMonitor will always have both a latest remote commitment tx
and a latest local commitment tx for broadcast.

This also fixes a strange API where we would close a channel
unceremoniously on peer-disconnection if we hadn't yet received the
funding_signed, but we'd already have a ChannelMonitor for that
channel. While it isn't strictly a bug (some potential DoS issues
aside), it is strange that these two definitions of a channel being
open were not in sync.

4 years agoAdd HTLC/extra data in LocalCommitmentTransaction from construction
Matt Corallo [Sat, 18 Apr 2020 04:10:24 +0000 (00:10 -0400)]
Add HTLC/extra data in LocalCommitmentTransaction from construction

1107ab06c33bd360bdee7ee64f4b690e753003f6 introduced some additional
metadata, including per-HTLC data in LocalCommitmentTransaction. To
keep diff reasonable it did so in ChannelMonitor after the
LocalCommitmentTransaction had been constructed and passed over the
wall, but there's little reason to do so - we should just be
constructing them with the data from the start, filled in by Channel.

This cleans up some internal interfaces a bit, slightly reduces
some data duplication and moves us one step forward to exposing
the guts of LocalCommitmentTransaction publicly in a sensible way.

4 years agoDrop redundant parameters in sign_local_commitment_tx
Matt Corallo [Sat, 18 Apr 2020 01:26:38 +0000 (21:26 -0400)]
Drop redundant parameters in sign_local_commitment_tx

The ChanKeys is created with knowledge of the Channel's value and
funding redeemscript up-front, so we should not be providing it
when making signing requests.

4 years agoAvoid a few useless clone() calls in onchaintx.rs
Matt Corallo [Sun, 19 Apr 2020 05:16:20 +0000 (01:16 -0400)]
Avoid a few useless clone() calls in onchaintx.rs

3d640da5c343111f538f006996c13c9a98e0d9e6 looped over a new HashMap
new_claims, clone()ing entries out of it right before droppng the
whole thing. This is an obvious candidate for drain(..).

4 years agoClarify documentation on deserialization reorg requirements
Matt Corallo [Tue, 21 Apr 2020 20:18:49 +0000 (16:18 -0400)]
Clarify documentation on deserialization reorg requirements

4 years agoUse more generic name as it will be use in GitHub Action too
Franck Royer [Tue, 21 Apr 2020 22:00:43 +0000 (08:00 +1000)]
Use more generic name as it will be use in GitHub Action too

4 years agoDrop ChannelMonitor::write_for_watchtower
Matt Corallo [Sun, 19 Apr 2020 01:36:33 +0000 (21:36 -0400)]
Drop ChannelMonitor::write_for_watchtower

Not only was watchtower mode never implemented, but the bits that
we had were removed some time ago. It doesn't seem likely we'll
move forward with a "watchtower-mode" ChannelMonitor, instead
we'll likely have some other, separate struct for this.

4 years agoMerge pull request #602 from murtyjones/test-byte-utils
Matt Corallo [Wed, 22 Apr 2020 20:45:04 +0000 (20:45 +0000)]
Merge pull request #602 from murtyjones/test-byte-utils

Add tests to byte_utils.rs

4 years agoMerge pull request #603 from moneyball/patch-3
Matt Corallo [Wed, 22 Apr 2020 20:43:45 +0000 (20:43 +0000)]
Merge pull request #603 from moneyball/patch-3

fixed typo

4 years agofixed typo
Steve Lee [Wed, 22 Apr 2020 17:52:32 +0000 (10:52 -0700)]
fixed typo

4 years agoUpdate lightning/src/util/byte_utils.rs
Marty Jones [Wed, 22 Apr 2020 00:29:46 +0000 (20:29 -0400)]
Update lightning/src/util/byte_utils.rs

4 years agoadd tests
Marty Jones [Wed, 22 Apr 2020 00:27:35 +0000 (20:27 -0400)]
add tests

4 years agoMerge pull request #594 from TheBlueMatt/2020-04-cleanups
Matt Corallo [Mon, 20 Apr 2020 21:54:35 +0000 (21:54 +0000)]
Merge pull request #594 from TheBlueMatt/2020-04-cleanups

Trivial Cleanups

4 years agoMerge pull request #596 from D4nte/586-update-htlc-error-code-content
Matt Corallo [Mon, 20 Apr 2020 05:14:18 +0000 (05:14 +0000)]
Merge pull request #596 from D4nte/586-update-htlc-error-code-content

Include `height` to `incorrect_or_unknown_payment_details` failure

4 years agoTest that `height` is included for incorrect payment details
Franck Royer [Sun, 19 Apr 2020 21:30:16 +0000 (07:30 +1000)]
Test that `height` is included for incorrect payment details

Ensure that the best know blockchain height is included in the
data of `incorrect_or_unknown_payment_details` message failure.

4 years agoInclude `height` to `incorrect_or_unknown_payment_details` failure
Franck Royer [Sun, 19 Apr 2020 12:14:14 +0000 (22:14 +1000)]
Include `height` to `incorrect_or_unknown_payment_details` failure

`incorrect_or_unknown_payment_details` failure message,
`0x4000 (PERM) | 15`, should include the following data:
- [u64:htlc_msat]
- [u32:height]
This patches ensure that the height is included in all
the occurrences of this failure message.

4 years agoDe-Option<> some fields in ChannelMonitor which are set at init 2020-04-cleanups
Matt Corallo [Sun, 19 Apr 2020 01:33:54 +0000 (21:33 -0400)]
De-Option<> some fields in ChannelMonitor which are set at init

After we moved the ChannelMonitor creation later during Channel
init, we never went back and cleaned up ChannelMonitor to remove
a number of now-useless Option<>s, so we do that now.

4 years agoDrop std::error::Error impl for DecodeError
Matt Corallo [Sat, 18 Apr 2020 01:14:54 +0000 (21:14 -0400)]
Drop std::error::Error impl for DecodeError

It appears to be effectively-deprecated in Rust now, and didn't
really appear to serve a lot of purpose anyway.

4 years agoConcretize some types in fuzz, addressing new rustc warnings
Matt Corallo [Sat, 18 Apr 2020 01:09:08 +0000 (21:09 -0400)]
Concretize some types in fuzz, addressing new rustc warnings

4 years agoFix new rustc warnings for unnecessary parenthesis
Matt Corallo [Sat, 18 Apr 2020 01:06:40 +0000 (21:06 -0400)]
Fix new rustc warnings for unnecessary parenthesis

4 years agoDrop uneccessary indirection in map-updating in 1107ab06c3
Matt Corallo [Fri, 17 Apr 2020 23:31:24 +0000 (19:31 -0400)]
Drop uneccessary indirection in map-updating in 1107ab06c3

1107ab06c33bd360bdee7ee64f4b690e753003f6 added a Vec of future
updates to apply during a loop, fixing a borrow checker issue that
didn't exist in the merged version of the patch. This simply reverts
that small part of the change.

4 years agoMerge pull request #558 from TheBlueMatt/2020-03-555-nits
Matt Corallo [Sun, 19 Apr 2020 00:19:21 +0000 (00:19 +0000)]
Merge pull request #558 from TheBlueMatt/2020-03-555-nits

Flatten ChannelMonitor substructs that don't add clarity

4 years agoFlatten ChannelMonitor substructs that don't add clarity 2020-03-555-nits
Matt Corallo [Sat, 21 Mar 2020 22:29:17 +0000 (18:29 -0400)]
Flatten ChannelMonitor substructs that don't add clarity

The new OnchainDetection struct (which is the remnants of the old
KeyStorage enum, which was removed in 1dbda4faedc33506e63176e6a456)
doesn't really add any clarity to ChannelMonitor, so best to just
drop it and move its members into ChannelMonitor directly.

4 years agoDrop uneccessary clone() introduced in 16fba9fd664522ac8d24111547b4
Matt Corallo [Sat, 21 Mar 2020 22:21:52 +0000 (18:21 -0400)]
Drop uneccessary clone() introduced in 16fba9fd664522ac8d24111547b4

4 years agoMerge pull request #559 from ariard/2020-03-move-local-commitment
Matt Corallo [Sat, 18 Apr 2020 00:05:11 +0000 (00:05 +0000)]
Merge pull request #559 from ariard/2020-03-move-local-commitment

Split parsing and transaction management for local transactions between Chanmon/Onchain

4 years agoAdd test_update_err_monitor_lockdown
Antoine Riard [Wed, 8 Apr 2020 01:07:55 +0000 (21:07 -0400)]
Add test_update_err_monitor_lockdown

This test tries the new lockdown logic in case of a signed-and-broadcast
local commitment transaction while a concurrent ChannelMonitorUpdate for
a next _local_ commitment is submitted from offchain. Update is rejected
as expected with a ChannelMonitorUpdateErr.

4 years agoMonitor should panic on receiving buggy update sequences
Antoine Riard [Mon, 6 Apr 2020 22:54:45 +0000 (18:54 -0400)]
Monitor should panic on receiving buggy update sequences

Channel shouldn't send a ChannelForceClosed update followed by
a LatestLocalCommitmentTxInfo as it would be a programming error
leading to risk of money loss. Force-closing the channel will
broadcast the local commitment transaction, if the revocation
secret for this one is released after its broadcast, it would
allow remote party to claim outputs on this transaction using
the revocation path.

4 years agoMake acces and signature of local commitment transaction unique
Antoine Riard [Mon, 6 Apr 2020 22:32:57 +0000 (18:32 -0400)]
Make acces and signature of local commitment transaction unique

Local commitment transaction broadcast can be triggered by a)
a Channel force-close or b) reaching some block height implying
a onchain HTLC-timeout. If one of this condition is fulfilled,
commitment is signed and from then any state update would be
rejected.

ChannelMonitor init at Channel creation need to be refactored
before to make get_fully_signed_local_tx infaillible to avoid
choking in the test framework.

4 years agoRemove temporary anti-duplicata logic
Antoine Riard [Mon, 9 Mar 2020 23:56:42 +0000 (19:56 -0400)]
Remove temporary anti-duplicata logic

4 years agoMove HTLC tx generation in OnchainTxHandler
Antoine Riard [Mon, 9 Mar 2020 22:15:35 +0000 (18:15 -0400)]
Move HTLC tx generation in OnchainTxHandler

HTLC Transaction can't be bumped without sighash changes
so their gneeration is one-time for nwo. We move them in
OnchainTxHandler for simplifying ChannelMonitor and to prepare
storage of keys material behind one external signer interface.

Some tests break due to change in transaction broadcaster order.
Number of transactions may vary because of temporary anti-duplicata
tweak can't dissociate between 2- broadcast from different
origins (ChannelMonitor, ChannelManager) and 2-broadcast from same
component.

4 years agoAdd OnchainTxHandler::get_fully_signed_htlc
Antoine Riard [Mon, 23 Mar 2020 05:30:48 +0000 (01:30 -0400)]
Add OnchainTxHandler::get_fully_signed_htlc

In case of channel force-closure, access to local commitment
transactions and its dependent HTLCs is needed. Instead of using
broadcast_by_local_state which registers outpoint to claim and
outputs to watch which are going to be discarded in this case,
we simply ask OnchainTxHandler to build and sign HTLC transactions
through new API.

4 years agoCache csv_local inside OnchainTxHandler
Antoine Riard [Mon, 23 Mar 2020 05:36:37 +0000 (01:36 -0400)]
Cache csv_local inside OnchainTxHandler

csv_local is csv_delay encumbering local revokable_redeemscript
for to_local an htlc output on local commitment/HTLC transactions.

4 years agoCache per-HTLC data in OnchainTxHandler::HTLCTxCache
Antoine Riard [Sat, 21 Mar 2020 22:52:00 +0000 (18:52 -0400)]
Cache per-HTLC data in OnchainTxHandler::HTLCTxCache

Splitting further parsing from transaction generation, we cache
transaction elements needed for local HTLC transaction inside
OnchainTxHandler. Duplicated data will be removed from ChannelMonitor
in future commits.

4 years agoCache HTLC transaction material inside OnchainTxHandler
Antoine Riard [Tue, 7 Apr 2020 22:46:14 +0000 (18:46 -0400)]
Cache HTLC transaction material inside OnchainTxHandler

Splitting further parsing from transaction generation, we cache
transaction elements needed for local HTLC transaction inside
OnchainTxHandler. Duplicated data will be removed from ChannelMonitor
in future commits.

4 years agoRemove signing htlc transaction from ChannelMonitor
Antoine Riard [Sat, 21 Mar 2020 19:39:19 +0000 (15:39 -0400)]
Remove signing htlc transaction from ChannelMonitor

Extend external signer interface to sign HTLC transactions on its
behalf without seckey passing. This move will allow us to remove
key access access from ChannelMonitor hot memory in further work.

HTLC transactions should stay half-signed by remote until
we need to broadcast them for timing-out/claiming HTLCs onchain.

4 years agoRemove per_input_material introduce in commit 0011713
Antoine Riard [Sat, 21 Mar 2020 19:48:11 +0000 (15:48 -0400)]
Remove per_input_material introduce in commit 0011713

Caching of input material for HTLC transaction was introducted
prevently but since then API (InputMaterial) has changed
between ChannelMonitor and OnchainTxHandler

4 years agoAccess signed local commitment through OnchainTxHandler
Antoine Riard [Sat, 21 Mar 2020 02:41:12 +0000 (22:41 -0400)]
Access signed local commitment through OnchainTxHandler

Implementing dynamic fee bumping implied to cache transaction material
including its witness, to generate a bumped version if needed.

ChannelMonitor is slowly rescoped to its parsing function with ongoing
patchset and data duplicata are removed. If signed local commitment tx
access is needed, it's done through OnchainTxHandler extended API

For test framework purpose, we use the test-only method
ChannelMonitor::unsafe_get_latest_local_commitment_txn to intentionally
generate unsafe local commitment to exerce revocation logic.

4 years agoCache current local commitment number in ChannelMonitor.
Antoine Riard [Tue, 24 Mar 2020 23:26:28 +0000 (19:26 -0400)]
Cache current local commitment number in ChannelMonitor.

By caching current local commitment number instead of deciphering
it from local commitment tx, we may remove local commitment tx
from ChannelMonitor in next commit.

4 years agoMove local commitment tx generation in OnchainTxHandler
Antoine Riard [Tue, 3 Mar 2020 23:51:50 +0000 (18:51 -0500)]
Move local commitment tx generation in OnchainTxHandler

Local Commitment Transaction can't be bumped without anchor outputs
so their generation is one-time for now. We move them in
OnchainTxHandler for simplifying ChannelMonitor and to prepare
storage of keys material behind one external signer interface.

Some tests break due to change in transaction broadcast order but
number of transactions broadcast should stay the same.

4 years agoPrevent any update of local commitment transaction once signed
Antoine Riard [Fri, 27 Mar 2020 21:53:52 +0000 (17:53 -0400)]
Prevent any update of local commitment transaction once signed

To prevent any unsafe state discrepancy between offchain and onchain,
once local commitment transaction has been signed due to an event
(either block height for HTLC-timeout or channel force-closure), don't
allow any further update of local commitment transaction view
to avoid delivery of revocation secret to counterparty for the
aformentionned signed transaction.

4 years agoCache local_commitment_tx in OnchainTxHandler
Antoine Riard [Sat, 21 Mar 2020 00:26:23 +0000 (20:26 -0400)]
Cache local_commitment_tx in OnchainTxHandler

As transaction generation and signature is headed to be moved
inside OnchainTxHandler, cache local_commitment_tx signed by remote.

If access to local commitment transaction is needed, we extend Onchain
TxHandler API to do so.

4 years agoCache funding_redeemscript in OnchainTxHandler
Antoine Riard [Sat, 21 Mar 2020 00:06:13 +0000 (20:06 -0400)]
Cache funding_redeemscript in OnchainTxHandler

As transaction generation and signature is headed to be moved
inside OnchainTxHandler, cache any usefule witness element.

4 years agoRemove duplicata for local commitment+HTLC txn
Antoine Riard [Tue, 3 Mar 2020 22:35:36 +0000 (17:35 -0500)]
Remove duplicata for local commitment+HTLC txn

Previously, we would regenerate this class of txn twice due to
block-rescan triggered by new watching outputs registered.

This commmit doesn't change behavior, it only tweaks TestBroadcaster
to ensure we modify cleanly tests anticipating next commit
refactor.

4 years agoMerge pull request #589 from jkczyz/2020-04-feature-fixes
Matt Corallo [Fri, 17 Apr 2020 20:03:06 +0000 (20:03 +0000)]
Merge pull request #589 from jkczyz/2020-04-feature-fixes

Missing feature checks and tests

4 years agoAdd missing known features in constructor
Jeffrey Czyz [Thu, 16 Apr 2020 02:29:20 +0000 (19:29 -0700)]
Add missing known features in constructor

PaymentSecret and BasicMPP were added in d39f25839ad9ff5e2a0e1dd8919e991828d1bf35
but were not accounted for in with_known_relevant_init_flags.

4 years agoRemove signing local commitment transaction from ChannelMonitor
Antoine Riard [Fri, 20 Mar 2020 22:04:01 +0000 (18:04 -0400)]
Remove signing local commitment transaction from ChannelMonitor

Extend external signer interface to sign local commitment transactions
on its behalf without seckey passing. This move will allow us to remove
key access from ChannelMonitor hot memory in further work.

Local commitment transaction should stay half-signed by remote until
we need to broadcast for a channel force-close or a HTLC to timeout onchain.

Add an unsafe test-only version of sign_local_commitment to fulfill our
test_framework needs.

4 years agoAdd missing feature sanity checks
Jeffrey Czyz [Thu, 16 Apr 2020 01:21:29 +0000 (18:21 -0700)]
Add missing feature sanity checks

PaymentSecret and BasicMPP were added in d39f25839ad9ff5e2a0e1dd8919e991828d1bf35
but the sanity test for supported features was missed.

4 years agoMerge pull request #441 from TheBlueMatt/2020-01-mpp
Matt Corallo [Wed, 15 Apr 2020 01:36:44 +0000 (01:36 +0000)]
Merge pull request #441 from TheBlueMatt/2020-01-mpp

Multi-Path Payments

4 years agoPass Route to send_payment as a reference, not move 2020-01-mpp
Matt Corallo [Tue, 7 Apr 2020 17:36:58 +0000 (13:36 -0400)]
Pass Route to send_payment as a reference, not move

ChannelManager::send_payment stopped utilizing its ownership of the
Route with MPP (which, for readability, now clone()s the individual
paths when creating HTLCSource::OutboundRoute objects). While this
isn't ideal, it likely also makes sense to ensure that the user has
access to the Route after sending to correlate individual path
failures with the paths in the route or, in the future, retry
individual paths.

Thus, the easiest solution is to just take the Route by reference,
allowing the user to retain ownership.

4 years agopanic!() when serializing OnionHopDatas with value > 21m BTC
Matt Corallo [Sat, 21 Mar 2020 22:49:30 +0000 (18:49 -0400)]
panic!() when serializing OnionHopDatas with value > 21m BTC

Add documentation to the struct fields noting this to avoid missing
docs when various msg structs become public.

4 years agoTest basic AMP payments in chanmon_consistency
Matt Corallo [Fri, 10 Jan 2020 22:24:33 +0000 (17:24 -0500)]
Test basic AMP payments in chanmon_consistency

4 years agoRefactor test utils and add a simple MPP send/claim test.
Matt Corallo [Tue, 7 Jan 2020 01:30:08 +0000 (20:30 -0500)]
Refactor test utils and add a simple MPP send/claim test.

4 years agoImplement multipath sends using payment_secret.
Matt Corallo [Tue, 7 Jan 2020 01:29:33 +0000 (20:29 -0500)]
Implement multipath sends using payment_secret.

This rather dramatically changes the return type of send_payment
making it much clearer when resending is safe and allowing us to
return a list of Results since different paths may have different
return values.

4 years agoRefactor payment-claim logic to ensure MPP-claim atomicity
Matt Corallo [Mon, 30 Mar 2020 20:24:19 +0000 (16:24 -0400)]
Refactor payment-claim logic to ensure MPP-claim atomicity

Previously if we claimed an MPP where a previous-hop channel was
closed while we were waitng for the user to provide us the preimage
we'd simply skip claiming that HTLC without letting the user know.

This refactors the claim logic to first check that all the channels
are still available (which is actually all we need - we really
mostly care about updating the channel monitors, not the channels
themselves) and then claim the HTLCs in the same lock, ensuring
atomicity.

4 years agoExpand the Route object to include multiple paths.
Matt Corallo [Sat, 4 Jan 2020 00:31:40 +0000 (19:31 -0500)]
Expand the Route object to include multiple paths.

Rather big diff, but its all mechanical and doesn't introduce any
new features.

4 years agoAdd/announce features for payment_secret and basic_mpp
Matt Corallo [Wed, 15 Jan 2020 04:31:21 +0000 (23:31 -0500)]
Add/announce features for payment_secret and basic_mpp

4 years agoImpl Base AMP in the receive pipeline and expose payment_secret
Matt Corallo [Thu, 2 Jan 2020 06:23:48 +0000 (01:23 -0500)]
Impl Base AMP in the receive pipeline and expose payment_secret

Base AMP is centered around the concept of a 'payment_secret` - an
opaque 32-byte random string which is used to authenticate the
sender to the recipient as well as tie the various HTLCs which
make up one payment together. This new field gets exposed in a
number of places, though sadly only as an Option for backwards
compatibility when sending to a receiver/receiving from a sender
which does not support Base AMP.

Sadly a huge diff here, but almost all of it is changing the method
signatures for sending/receiving/failing HTLCs and the
PaymentReceived event, which all now need to expose an
Option<[u8; 32]> for the payment_secret.

It doesn't yet properly fail back pending HTLCs when the full AMP
payment is never received (which should result in accidental
channel force-closures). Further, as sending AMP payments is not
yet supported, the only test here is a simple single-path payment
with a payment_secret in it.

4 years agoRefuse to deserialize OnionHopDatas with values > 21 million
Matt Corallo [Thu, 2 Jan 2020 02:13:48 +0000 (21:13 -0500)]
Refuse to deserialize OnionHopDatas with values > 21 million

We should probably do this for all values (and define a newtype
for msat values), but this will do for now.

4 years agoSupport (de)serializing payment_data in onion TLVs and track them
Matt Corallo [Thu, 2 Jan 2020 01:20:42 +0000 (20:20 -0500)]
Support (de)serializing payment_data in onion TLVs and track them

This is the first step in Base AMP support, just tracking the
relevant data in internal datastructures.

4 years agoSplit only-receive/forward data out of PendingHTLCInfo into an enum
Matt Corallo [Wed, 1 Jan 2020 22:39:51 +0000 (17:39 -0500)]
Split only-receive/forward data out of PendingHTLCInfo into an enum

This should avoid blowing up the size of the struct when we add
additional data that is only relevant for receive.

4 years agoMerge pull request #582 from arik-so/trailing_newline_config
Matt Corallo [Sat, 11 Apr 2020 20:28:26 +0000 (20:28 +0000)]
Merge pull request #582 from arik-so/trailing_newline_config

Mandate new line at end of file in editorconfig.