]> git.bitcoin.ninja Git - rust-lightning/log
rust-lightning
13 months agoWrap long onion_message fuzz strings
Jeffrey Czyz [Wed, 25 Oct 2023 21:37:26 +0000 (16:37 -0500)]
Wrap long onion_message fuzz strings

Some editors like vim slow to a crawl when scrolling over long strings
when syntax highlighting is turned on. Limit the length in fuzz strings
to avoid this.

13 months agoMerge pull request #2678 from TheBlueMatt/2023-10-0.0.118 v0.0.118
Matt Corallo [Tue, 24 Oct 2023 01:26:30 +0000 (01:26 +0000)]
Merge pull request #2678 from TheBlueMatt/2023-10-0.0.118

Cut 0.0.118

13 months agoBump crate versions to lightning 0.0.118, invoice 0.26 2023-10-0.0.118
Matt Corallo [Fri, 20 Oct 2023 23:44:49 +0000 (23:44 +0000)]
Bump crate versions to lightning 0.0.118, invoice 0.26

13 months ago0.0.118 release notes
Matt Corallo [Fri, 20 Oct 2023 20:46:30 +0000 (20:46 +0000)]
0.0.118 release notes

13 months agoMerge pull request #2679 from TheBlueMatt/2023-10-116-bindings-1
Matt Corallo [Mon, 23 Oct 2023 22:58:15 +0000 (22:58 +0000)]
Merge pull request #2679 from TheBlueMatt/2023-10-116-bindings-1

Small bindings tweaks for 0.0.118

13 months agoFix CI on rustc 1.50 and below 2023-10-116-bindings-1
Matt Corallo [Mon, 23 Oct 2023 19:50:12 +0000 (19:50 +0000)]
Fix CI on rustc 1.50 and below

rustc doesn't allow `--features` with `-p`, so we simply skip the
steps that rely on it.

13 months agoUse a tuple, not a struct, for `PendingOnionMessage` in bindings
Matt Corallo [Mon, 23 Oct 2023 18:55:17 +0000 (18:55 +0000)]
Use a tuple, not a struct, for `PendingOnionMessage` in bindings

Bindings aren't currently able to handle a struct with a generic
which is actually exposed - we map all structs concretely to a
single type, whereas having fluctuating types on a struct requires
mapping the inner field to a trait first.

Since this isn't super practical, we make `PendingOnionMessage` a
tuple in bindings, rather than a struct.

13 months agoMerge pull request #2676 from TheBlueMatt/2023-10-various-followups
Matt Corallo [Mon, 23 Oct 2023 19:16:29 +0000 (19:16 +0000)]
Merge pull request #2676 from TheBlueMatt/2023-10-various-followups

Various Followups to 2039 and 2674

13 months agoAdd relevant no-export tags to functions returning builders 2023-10-various-followups
Matt Corallo [Mon, 23 Oct 2023 16:49:49 +0000 (16:49 +0000)]
Add relevant no-export tags to functions returning builders

Because we can't map move semantics in most languages, we also
can't map our current builders. Thus, we have to mark them
no-export.

13 months agoDrop an unnecessary no-export on ParsedOnionMessageContents
Matt Corallo [Mon, 23 Oct 2023 16:50:05 +0000 (16:50 +0000)]
Drop an unnecessary no-export on ParsedOnionMessageContents

13 months agoDo not compile the `Simple*` type aliases in `c_bindings` at all
Matt Corallo [Sat, 21 Oct 2023 02:42:48 +0000 (02:42 +0000)]
Do not compile the `Simple*` type aliases in `c_bindings` at all

Because the bindings changes now require further changes to our
type definitions, avoiding building the `Simple*` type aliases
entirely makes the patchset there simpler.

13 months agoFix (and test) the `c_bindings` build flag
Matt Corallo [Sat, 21 Oct 2023 01:08:38 +0000 (01:08 +0000)]
Fix (and test) the `c_bindings` build flag

Rather than only building with the `c_bindings` flag in certain
crates, we go ahead and test all crates with the flag in CI here.

13 months agoRemove some additional excess words in `ConfirmationTarget` docs
Matt Corallo [Fri, 20 Oct 2023 18:19:57 +0000 (18:19 +0000)]
Remove some additional excess words in `ConfirmationTarget` docs

13 months agoRemove a redundant sentence in `ConfirmationTarget` docs
Matt Corallo [Fri, 20 Oct 2023 18:11:56 +0000 (18:11 +0000)]
Remove a redundant sentence in `ConfirmationTarget` docs

... and correct direction which causes force-closure in another
sentence.

13 months agoDrop unused `use` import.
Matt Corallo [Fri, 20 Oct 2023 18:09:39 +0000 (18:09 +0000)]
Drop unused `use` import.

13 months agoAvoid a redundant allocation in `InvoiceError` handling in one case
Matt Corallo [Fri, 20 Oct 2023 17:38:19 +0000 (17:38 +0000)]
Avoid a redundant allocation in `InvoiceError` handling in one case

... by passing an owned `String`, rather than taking an `&str` and
`to_owned()`ing it.

13 months agoUse `Default::default()` for scoring params in tests
Matt Corallo [Fri, 20 Oct 2023 17:34:12 +0000 (17:34 +0000)]
Use `Default::default()` for scoring params in tests

In 26c1639ab69d6780c97a118f09e42cb42304088a we switched to using
`Default::default()` to initialize `()` for scoring parameters in
tests. A number of `()`s slipped back in recently, which we replace
here.

13 months agoMerge pull request #2667 from wpaulino/random-htlc-holder-sigs-non-anchors
Matt Corallo [Fri, 20 Oct 2023 22:55:08 +0000 (22:55 +0000)]
Merge pull request #2667 from wpaulino/random-htlc-holder-sigs-non-anchors

Use sign_holder_htlc_transaction to sign non-anchors holder HTLCs

13 months agoMerge pull request #2674 from wpaulino/consider-anchor-outputs-value-balances
Matt Corallo [Fri, 20 Oct 2023 22:54:08 +0000 (22:54 +0000)]
Merge pull request #2674 from wpaulino/consider-anchor-outputs-value-balances

Consider anchor outputs value throughout balance checks and computations

13 months agoMove HTLCDescriptor to sign module
Wilmer Paulino [Mon, 16 Oct 2023 19:21:52 +0000 (12:21 -0700)]
Move HTLCDescriptor to sign module

Now that `HTLCDescriptor` is no longer specific to anchors, it doesn't
make sense for it to live in the `bump_transaction` module anymore.

13 months agoDon't sign holder HTLCs along with holder commitments
Wilmer Paulino [Fri, 13 Oct 2023 21:09:37 +0000 (14:09 -0700)]
Don't sign holder HTLCs along with holder commitments

`sign_holder_commitment_and_htlcs` never really made sense. Unlike
`sign_counterparty_commitment`, the signatures for holder HTLC
transactions may be required much later than the commitment
transaction's. While it was nice for us to only reach the signer once to
obtain all holder signatures, it's not really ideal anymore as we want
our signatures to be random and not reused.

We no longer return all holder HTLC signatures and instead defer to
obtaining them via `EcdsaChannelSigner::sign_holder_htlc_transaction`.

13 months agoRemove caching of holder HTLC signatures
Wilmer Paulino [Fri, 13 Oct 2023 20:58:59 +0000 (13:58 -0700)]
Remove caching of holder HTLC signatures

Since we want our holder HTLC signatures to be randomly generated and
not reused, our existing caches are useless now, so we opt to remove
them.

13 months agoUse sign_holder_htlc_transaction to sign non-anchors holder HTLCs
Wilmer Paulino [Fri, 13 Oct 2023 20:52:23 +0000 (13:52 -0700)]
Use sign_holder_htlc_transaction to sign non-anchors holder HTLCs

We want to ensure we use fresh random signatures to prevent certain
classes of transaction replacement attacks at the bitcoin P2P layer.
This was already covered for commitment transactions and zero fee holder
HTLC transactions, but was missing for holder HTLC transactions on
non-anchors channels.

We can easily do this by reusing the existing
`EcdsaChannelSigner::sign_holder_htlc_transaction` method and
circumventing the existing `holder_htlc_sigs/prev_holder_htlc_sigs`
caches, which will be removed in a later commit anyway.

13 months agoApply a default max fee rather than none when paying for BOLT12
Matt Corallo [Fri, 20 Oct 2023 17:31:42 +0000 (17:31 +0000)]
Apply a default max fee rather than none when paying for BOLT12

If the user declines to specify a `max_total_routing_fee_msat` in
the new BOLT12 payment methods, rather than defaulting to no limit
on the fee we pay at all, we should default to our "usual default",
ie the one calculated in
`RouteParameters::from_payment_params_and_value`.

We do this here, as well as documenting the behavior on the payment
methods.

13 months agoOnly account for fee spike buffer multiple on non-anchor channels 2023-10-2674-fuzz-test
Wilmer Paulino [Thu, 19 Oct 2023 16:29:21 +0000 (09:29 -0700)]
Only account for fee spike buffer multiple on non-anchor channels

Anchor outputs channels are no longer susceptible to fee spikes as they
now mostly target the dynamic minimum mempool fee and can contribute the
remainder of fees when closing.

13 months agoConsider anchor outputs value on channel open
Wilmer Paulino [Thu, 19 Oct 2023 16:27:57 +0000 (09:27 -0700)]
Consider anchor outputs value on channel open

We should make sure the funding amount of a channel can cover all its
associated costs, including the value of anchor outputs, to make sure
that it is actually usable once "opened".

13 months agoConsider anchor outputs value on inbound HTLCs
Wilmer Paulino [Thu, 19 Oct 2023 16:27:30 +0000 (09:27 -0700)]
Consider anchor outputs value on inbound HTLCs

This could lead us to accept HTLCs that would put the sender below
their reserve, which must never happen.

13 months agoConsider anchor outputs value in get_available_balances
Wilmer Paulino [Thu, 19 Oct 2023 16:25:23 +0000 (09:25 -0700)]
Consider anchor outputs value in get_available_balances

This could lead us to sending/forwarding HTLCs that would put us below
our reserve, forcing our counterparty to close the channel on us due to
an invalid update.

13 months agoRun chanmon_consistency_test with anchor outputs channels
Wilmer Paulino [Thu, 19 Oct 2023 16:22:50 +0000 (09:22 -0700)]
Run chanmon_consistency_test with anchor outputs channels

13 months agoProvide missing derivation parameters to OnchainTxHandler
Wilmer Paulino [Fri, 13 Oct 2023 20:47:45 +0000 (13:47 -0700)]
Provide missing derivation parameters to OnchainTxHandler

`OnchainTxHandler` will need to construct `HTLCDescriptor`s for holder
HTLCs, but it did not have access to all of the derivation parameters
that need to be provided.

13 months agoSupport signing non-anchors HTLCs with HTLCDescriptor
Wilmer Paulino [Fri, 13 Oct 2023 20:49:50 +0000 (13:49 -0700)]
Support signing non-anchors HTLCs with HTLCDescriptor

We plan to use `EcdsaChannelSigner::sign_holder_htlc_transaction` to
also sign holder HTLC transactions on non-anchor outputs channels.
`HTLCDescriptor` was only used in an anchor outputs context, so a few
things needed changing, mostly to handle the different scripts and
feerate.

13 months agoMerge pull request #2660 from benthecarman/flexible-fee-rate
Matt Corallo [Fri, 20 Oct 2023 17:37:17 +0000 (17:37 +0000)]
Merge pull request #2660 from benthecarman/flexible-fee-rate

More flexible fee rate estimates

13 months agoMore flexible fee rate estimates
benthecarman [Thu, 12 Oct 2023 20:43:30 +0000 (15:43 -0500)]
More flexible fee rate estimates

13 months agoMerge pull request #2039 from jkczyz/2023-02-offer-flow
Matt Corallo [Fri, 20 Oct 2023 16:40:17 +0000 (16:40 +0000)]
Merge pull request #2039 from jkczyz/2023-02-offer-flow

BOLT 12 Offers message flow

13 months agoMerge pull request #2670 from yanganto/socket-addr-to-string
Matt Corallo [Fri, 20 Oct 2023 15:57:43 +0000 (15:57 +0000)]
Merge pull request #2670 from yanganto/socket-addr-to-string

Impl `Display` for SocketAddress

13 months agoFix PaymentConstraints::max_cltv_expiry docs
Jeffrey Czyz [Fri, 20 Oct 2023 01:32:00 +0000 (20:32 -0500)]
Fix PaymentConstraints::max_cltv_expiry docs

13 months agoFix build warnings
Jeffrey Czyz [Thu, 19 Oct 2023 23:02:58 +0000 (18:02 -0500)]
Fix build warnings

13 months agoExpand request_refund_payment docs for limitations
Jeffrey Czyz [Thu, 19 Oct 2023 21:36:02 +0000 (16:36 -0500)]
Expand request_refund_payment docs for limitations

13 months agoOnion message routing to immediate peers.
Jeffrey Czyz [Thu, 19 Oct 2023 20:50:19 +0000 (15:50 -0500)]
Onion message routing to immediate peers.

DefaultMessageRouter always fails. Update it so that it can route to a
directly connected peer. This is needed for an Offers minimum viable
product.

13 months agoAdd privacy section to pay_for_offer docs
Jeffrey Czyz [Thu, 19 Oct 2023 14:45:30 +0000 (09:45 -0500)]
Add privacy section to pay_for_offer docs

13 months agoOrganize create_refund and pay_for_offer docs
Jeffrey Czyz [Thu, 19 Oct 2023 14:37:47 +0000 (09:37 -0500)]
Organize create_refund and pay_for_offer docs

13 months agoDocument InvoiceRequestFailed in ChannelManager
Jeffrey Czyz [Thu, 19 Oct 2023 14:16:08 +0000 (09:16 -0500)]
Document InvoiceRequestFailed in ChannelManager

13 months agoRevert "Config-guard Event::InvoiceRequestFailed"
Jeffrey Czyz [Thu, 12 Oct 2023 21:10:51 +0000 (16:10 -0500)]
Revert "Config-guard Event::InvoiceRequestFailed"

This reverts commit c7219e46831751c52026932294c33fce24121d84.

13 months agoUse ChannelManager as OffersMessageHandler
Jeffrey Czyz [Tue, 6 Jun 2023 17:00:54 +0000 (12:00 -0500)]
Use ChannelManager as OffersMessageHandler

13 months agoOffersMessageHandler impl for ChannelManager
Jeffrey Czyz [Fri, 1 Sep 2023 19:04:27 +0000 (14:04 -0500)]
OffersMessageHandler impl for ChannelManager

Define the BOLT 12 message flow in ChannelManager's
OffersMessageHandler implementation.
- An invoice_request message results in responding with an invoice
  message if it can be verified that the request is for a valid offer.
- An invoice is paid if it can be verified to have originated from a
  sent invoice_request or a refund.
- An invoice_error is sent in some failure cases.
- Initial messages enqueued for sending are released to OnionMessenger

13 months agoCheck offer expiry when building invoice in no-std
Jeffrey Czyz [Thu, 19 Oct 2023 22:49:13 +0000 (17:49 -0500)]
Check offer expiry when building invoice in no-std

Building an invoice will fail if the underlying offer or refund has
already expired. The check was skipped in no-std since there is no
system clock. However, the invoice creation time can be used instead.
This prevents responding to an invoice request if the offer has already
expired.

13 months agoGrammar fix in docs
Jeffrey Czyz [Fri, 1 Sep 2023 14:40:34 +0000 (09:40 -0500)]
Grammar fix in docs

13 months agoBOLT12 invoice_feature methods for ChannelManager
Jeffrey Czyz [Fri, 1 Sep 2023 14:38:44 +0000 (09:38 -0500)]
BOLT12 invoice_feature methods for ChannelManager

13 months agoQualify BOLT11 ChannelManager invoice_features
Jeffrey Czyz [Fri, 1 Sep 2023 14:22:43 +0000 (09:22 -0500)]
Qualify BOLT11 ChannelManager invoice_features

13 months agoUtility for creating and sending Bolt12Invoices
Jeffrey Czyz [Fri, 15 Sep 2023 18:40:41 +0000 (13:40 -0500)]
Utility for creating and sending Bolt12Invoices

Add a utility to ChannelManager for creating a Bolt12Invoice for a
Refund such that the ChannelManager can recognize the PaymentHash and
reconstruct the PaymentPreimage from the PaymentSecret, the latter of
which is contained in a BlindedPath within the invoice.

13 months agoUtility for paying for an Offer
Jeffrey Czyz [Thu, 14 Sep 2023 18:32:51 +0000 (13:32 -0500)]
Utility for paying for an Offer

Add a utility to ChannelManager for sending an InvoiceRequest for an
Offer such that derived keys are used for the payer id. This allows for
stateless verification of any Invoice messages before it is paid.

Also tracks future payments using the given PaymentId such that the
corresponding Invoice is paid only once.

13 months agoAbsolute expiry or timer tick payment expiration
Jeffrey Czyz [Thu, 19 Oct 2023 19:38:16 +0000 (14:38 -0500)]
Absolute expiry or timer tick payment expiration

Pending outbound payments use an absolute expiry to determine when they
are considered stale and should be fail. In `no-std`, this may result in
long timeouts as the highest seen block time is used. Instead, allow for
expiration based on timer ticks. This will be use in an upcoming commit
for invoice request expiration.

13 months agoStore OffersMessages for later sending
Jeffrey Czyz [Thu, 14 Sep 2023 19:50:56 +0000 (14:50 -0500)]
Store OffersMessages for later sending

Upcoming commits will add utilities for sending an InvoiceRequest for an
Offer and an Invoice for a Refund. These messages need to be enqueued so
that they can be released in ChannelManager's implementation of
OffersMessageHandler to OnionMessenger for sending.

These messages do not need to be serialized as they must be resent upon
restart.

13 months agoExpand docs on failing expired outbound payments
Jeffrey Czyz [Thu, 19 Oct 2023 13:45:25 +0000 (08:45 -0500)]
Expand docs on failing expired outbound payments

13 months agoimpl Display for SocketAddress
Antonio Yang [Wed, 18 Oct 2023 10:36:03 +0000 (18:36 +0800)]
impl Display for SocketAddress

13 months agoMerge pull request #2636 from slanesuke/impl-ToSocketAddrs-for-Hostname
Elias Rohrer [Fri, 20 Oct 2023 07:52:29 +0000 (09:52 +0200)]
Merge pull request #2636 from slanesuke/impl-ToSocketAddrs-for-Hostname

Impl ToSocketAddrs for SocketAddress

13 months agoMerge pull request #2666 from tnull/2023-10-observable-update
Matt Corallo [Thu, 19 Oct 2023 21:49:02 +0000 (21:49 +0000)]
Merge pull request #2666 from tnull/2023-10-observable-update

13 months agoMerge pull request #2661 from TheBlueMatt/2023-10-dup-claim-chan-hang
Matt Corallo [Thu, 19 Oct 2023 17:53:46 +0000 (17:53 +0000)]
Merge pull request #2661 from TheBlueMatt/2023-10-dup-claim-chan-hang

Immediately unblock channels on duplicate claims

13 months agoImmediately unblock channels on duplicate claims 2023-10-dup-claim-chan-hang
Matt Corallo [Wed, 11 Oct 2023 14:01:28 +0000 (14:01 +0000)]
Immediately unblock channels on duplicate claims

When `MonitorUpdateCompletionAction`s were added, we didn't
consider the case of a duplicate claim during normal HTLC
processing (as the handling only had an `if let` rather than a
`match`, which made the branch easy to miss). This can lead to a
channel freezing indefinitely if an HTLC is claimed (without a
`commitment_signed`), the peer disconnects, and then the HTLC is
claimed again, leading to a never-completing
`MonitorUpdateCompletionAction`.

The fix is simple - if we get back an
`UpdateFulfillCommitFetch::DuplicateClaim` when claiming from the
inbound edge, immediately unlock the outbound edge channel with a
new `MonitorUpdateCompletionAction::FreeOtherChannelImmediately`.

Here we implement this fix by actually generating the new variant
when a claim is duplicative.

13 months agoAdd an immediately-freeing `MonitorUpdateCompletionAction`.
Matt Corallo [Wed, 11 Oct 2023 13:56:00 +0000 (13:56 +0000)]
Add an immediately-freeing `MonitorUpdateCompletionAction`.

When `MonitorUpdateCompletionAction`s were added, we didn't
consider the case of a duplicate claim during normal HTLC
processing (as the handling only had an `if let` rather than a
`match`, which made the branch easy to miss). This can lead to a
channel freezing indefinitely if an HTLC is claimed (without a
`commitment_signed`), the peer disconnects, and then the HTLC is
claimed again, leading to a never-completing
`MonitorUpdateCompletionAction`.

The fix is simple - if we get back an
`UpdateFulfillCommitFetch::DuplicateClaim` when claiming from the
inbound edge, immediately unlock the outbound edge channel with a
new `MonitorUpdateCompletionAction::FreeOtherChannelImmediately`.

Here we add the new variant, which we start generating in the next
commit.

13 months agoIndicate to `claim_funds_internal` that we're replaying on startup
Matt Corallo [Thu, 12 Oct 2023 22:26:07 +0000 (22:26 +0000)]
Indicate to `claim_funds_internal` that we're replaying on startup

While we'd previously avoided this, this is sadly now required in
the next commit.

13 months agoLog when we prepare to block a channel's next `revoke_and_ack`
Matt Corallo [Wed, 11 Oct 2023 01:39:26 +0000 (01:39 +0000)]
Log when we prepare to block a channel's next `revoke_and_ack`

This may help in debugging blocking actions in the future.

13 months agoDon't apply PathFailure::ChannelUpdateMessage
Elias Rohrer [Thu, 19 Oct 2023 15:00:50 +0000 (17:00 +0200)]
Don't apply PathFailure::ChannelUpdateMessage

If we receive a channel update from an intermediary via a failure onion
we shouldn't apply them in a persisted and network-observable way to our
network graph, as this might introduce a privacy leak. Here, we
therefore avoid applying such updates to our network graph.

13 months agoAllow to verify channel updates without applying them
Elias Rohrer [Fri, 13 Oct 2023 09:30:15 +0000 (11:30 +0200)]
Allow to verify channel updates without applying them

We introduce a new `NetworkGraph::verify_channel_update` method that
allows to check whether an update would be applied by `update_channel`.

13 months agoMerge pull request #2578 from jkczyz/2023-09-offer-utilities
Matt Corallo [Thu, 19 Oct 2023 02:41:48 +0000 (02:41 +0000)]
Merge pull request #2578 from jkczyz/2023-09-offer-utilities

BOLT 12 Offers utilities

13 months agoImpl ToSocketAddrs for SocketAddress
slanesuke [Sat, 30 Sep 2023 23:05:00 +0000 (17:05 -0600)]
Impl ToSocketAddrs for SocketAddress

13 months agoClean up peel_onion name, parameters, and docs
Jeffrey Czyz [Wed, 18 Oct 2023 23:24:07 +0000 (18:24 -0500)]
Clean up peel_onion name, parameters, and docs

For consistency with other functions and doc cleanliness.

13 months agoInclude a one-hop blinded path in Offer and Refund
Jeffrey Czyz [Fri, 3 Mar 2023 15:38:45 +0000 (09:38 -0600)]
Include a one-hop blinded path in Offer and Refund

While this doesn't add much privacy over not including any blinded
paths, it allows us to exercise code for receiving on blinded paths.

13 months agoStateless offer and refund builder utilities
Jeffrey Czyz [Tue, 14 Feb 2023 03:54:37 +0000 (21:54 -0600)]
Stateless offer and refund builder utilities

Add utility functions to ChannelManager for creating OfferBuilder,
and RefundBuilder such that derived keys are used for the signing
pubkey and payer id, respectively. This allows for stateless
verification of any InvoiceRequest and Invoice messages.

Later, blinded paths can be included in the returned builders.

Also tracks future payments using the given PaymentId such that the
corresponding Invoice is paid only once.

13 months agoAwait for invoices using an absolute expiry
Jeffrey Czyz [Tue, 17 Oct 2023 14:59:39 +0000 (09:59 -0500)]
Await for invoices using an absolute expiry

PendingOutboundPayment::AwaitingInvoice counts the number of timer ticks
that have passed awaiting a Bolt12Invoice for an InvoiceRequest. When a
constant INVOICE_REQUEST_TIMEOUT_TICKS has passed, the payment is
forgotten. However, this mechanism is insufficient for the Refund
scenario, where the Refund's expiration should be used instead.

Change AwaitingInvoice to store an absolute expiry instead. When
removing stale payments, pass the `SystemTime` in `std` and the highest
block time minus two hours in `no-std`.

13 months agoRemove outdated docs
Jeffrey Czyz [Tue, 17 Oct 2023 15:47:48 +0000 (10:47 -0500)]
Remove outdated docs

13 months agoAllow sending onion messages to 1-hop blinded path
Jeffrey Czyz [Mon, 11 Sep 2023 22:40:43 +0000 (17:40 -0500)]
Allow sending onion messages to 1-hop blinded path

This allows for specifying the introduction node as the message
recipient.

13 months agoEnqueue onion messages in handlers
Jeffrey Czyz [Thu, 14 Sep 2023 02:19:50 +0000 (21:19 -0500)]
Enqueue onion messages in handlers

When constructing onion messages to send initially (opposed to replying
to one from a handler), the user must construct an OnionMessagePath first
before calling OnionMessener::send_onion_message. Additionally, having a
reference to OnionMessener isn't always desirable. For instance, in an
upcoming commit, ChannelManager will implement OffersMessageHandler,
which OnionMessenger needs a reference to. If ChannelManager had a
reference to OnionMessenger, too, there would be a dependency cycle.

Instead, modify OffersMessageHandler and CustomOnionMessageHandler's
interfaces to include a method for releasing pending onion messages.
That way, ChannelManager may, for instance, construct and enqueue an
InvoiceRequest for sending without needing a reference to
OnionMessenger.

Additionally, OnionMessenger has responsibility for path finding just as
it does when replying to messages from a handler. It performs this when
extracting messages from the handlers before returning the next message
to send to a peer.

13 months agoGeneralize CustomOnionMessageContents trait
Jeffrey Czyz [Tue, 19 Sep 2023 17:35:13 +0000 (12:35 -0500)]
Generalize CustomOnionMessageContents trait

Rename CustomOnionMessageContents to OnionMessageContents and use it as
a trait bound on messages passed to OnionMessenger methods. This allows
using the trait in an upcoming commit as a bound on the contents of
PendingOnionMessage.

Also, make ParsedOnionMessageContent implement OnionMessageContents so
that Payload can be bounded by OnionMessageContents directly, but used
when either reading a ParsedOnionMessageContent or writing a specific
type of OnionMessageContents (e.g., OffersMessage).

13 months agoRename OnionMessageContents
Jeffrey Czyz [Tue, 19 Sep 2023 15:59:53 +0000 (10:59 -0500)]
Rename OnionMessageContents

In preparation for needing the name OnionMessageContents for a trait to
bound methods, rename it to ParsedOnionMessageContents. In the next
commit, it's use will be limited to reading only, and the new trait will
be a bound on method parameters instead.

13 months agoGeneralize respond_with_onion_message
Jeffrey Czyz [Wed, 13 Sep 2023 19:13:05 +0000 (14:13 -0500)]
Generalize respond_with_onion_message

OnionMessenger can send onion message responses from its handlers using
respond_with_onion_message, which finds a path to the destination and
enqueues the response for sending. Generalize this as it can be used not
only for responses but for initial sends as well.

13 months agoImport msgs::OnionMessage
Jeffrey Czyz [Fri, 13 Oct 2023 16:12:57 +0000 (11:12 -0500)]
Import msgs::OnionMessage

13 months agoClean up onion messenger parameters and docs
Jeffrey Czyz [Thu, 5 Oct 2023 01:23:34 +0000 (15:23 -1000)]
Clean up onion messenger parameters and docs

13 months agoAvoid overloading introduction_node_id
Jeffrey Czyz [Tue, 12 Sep 2023 22:18:49 +0000 (17:18 -0500)]
Avoid overloading introduction_node_id

13 months agoRemove OnionMessageProvider
Jeffrey Czyz [Tue, 12 Sep 2023 21:57:15 +0000 (16:57 -0500)]
Remove OnionMessageProvider

OnionMessageProvider is a super-trait of OnionMessageHandler, but they
don't need to be used separately. Additionally, the former is misplaced
in the events module. Remove OnionMessageProvider and add it's only
method, next_onion_message_for_peer, into OnionMessageHandler.

13 months agoUpdate OnionMessenger docs for Offers
Jeffrey Czyz [Tue, 12 Sep 2023 21:14:16 +0000 (16:14 -0500)]
Update OnionMessenger docs for Offers

13 months agoMerge pull request #2663 from TheBlueMatt/2023-10-peer-race-send-discon
Matt Corallo [Wed, 18 Oct 2023 21:54:57 +0000 (21:54 +0000)]
Merge pull request #2663 from TheBlueMatt/2023-10-peer-race-send-discon

Fix race between outbound messages and peer disconnection

13 months agoMerge pull request #2658 from wpaulino/bogus-channel-reestablish
Matt Corallo [Wed, 18 Oct 2023 21:47:31 +0000 (21:47 +0000)]
Merge pull request #2658 from wpaulino/bogus-channel-reestablish

Send bogus ChannelReestablish for unknown channels

13 months agoMerge pull request #2599 from Evanfeenstra/peel-onion
Matt Corallo [Wed, 18 Oct 2023 21:11:52 +0000 (21:11 +0000)]
Merge pull request #2599 from Evanfeenstra/peel-onion

public static peel_onion method on OnionMessenger

13 months agoFix race between outbound messages and peer disconnection 2023-10-peer-race-send-discon
Matt Corallo [Wed, 18 Oct 2023 15:22:26 +0000 (15:22 +0000)]
Fix race between outbound messages and peer disconnection

Previously, outbound messages held in `process_events` could race
with peer disconnection, allowing a message intended for a peer
before disconnection to be sent to the same peer after
disconnection.

The fix is simple - hold the peers read lock while we fetch
pending messages from peers (as we disconnect with the write lock).

13 months agoRelease short_to_chan_info lock throughout forwarding_channel_not_found
Wilmer Paulino [Mon, 16 Oct 2023 20:29:06 +0000 (13:29 -0700)]
Release short_to_chan_info lock throughout forwarding_channel_not_found

Not doing so caused a lock order inversion between the locks
`ChannelManager::best_block` and `ChannelManager::short_to_chan_info`
after the addition of `test_trigger_lnd_force_close`.

It turns out that we were holding the `short_to_chan_info` for longer
than needed when processing HTLC forwards. We only need to acquire it to
quickly obtain channel info, and there aren't any other locks within
`forwarding_channel_not_found` that depend on it being held.

13 months agoDisconnect peer when force closing a funded channel with an error
Wilmer Paulino [Fri, 13 Oct 2023 16:28:35 +0000 (09:28 -0700)]
Disconnect peer when force closing a funded channel with an error

We do this to ensure that the counterparty will always broadcast their
latest state when we broadcast ours. Usually, they'll do this with the
`error` message alone, but if they don't receive it or ignore it, then
we'll force them to broadcast by sending them a bogus
`channel_reestablish` upon reconnecting. Note that this doesn't apply to
unfunded channels as there is no commitment transaction to broadcast.

13 months agoSend bogus ChannelReestablish for unknown channels
Wilmer Paulino [Wed, 11 Oct 2023 16:42:05 +0000 (09:42 -0700)]
Send bogus ChannelReestablish for unknown channels

Unfortunately, lnd doesn't force close on errors
(https://github.com/lightningnetwork/lnd/blob/abb1e3463f3a83bbb843d5c399869dbe930ad94f/htlcswitch/link.go#L2119).
One of the few ways to get an lnd counterparty to force close is by
replicating what they do when restoring static channel backups (SCBs).
They send an invalid `ChannelReestablish` with `0` commitment numbers
and an invalid `your_last_per_commitment_secret`.

Since we received a `ChannelReestablish` for a channel that doesn't
exist, we can assume it's likely the channel closed from our point of
view, but it remains open on the counterparty's side. By sending this
bogus `ChannelReestablish` message now as a response to theirs, we
trigger them to force close broadcasting their latest state. If the
closing transaction from our point of view remains unconfirmed, it'll
enter a race with the counterparty's to-be-broadcast latest commitment
transaction.

13 months agopublic static peel_onion method on OnionMessenger
Evan Feenstra [Tue, 26 Sep 2023 18:29:16 +0000 (11:29 -0700)]
public static peel_onion method on OnionMessenger

13 months agoMerge pull request #2662 from jkczyz/2023-10-chain-hash
valentinewallace [Tue, 17 Oct 2023 15:05:45 +0000 (11:05 -0400)]
Merge pull request #2662 from jkczyz/2023-10-chain-hash

Use `ChainHash` instead of `BlockHash` as applicable

13 months agoMerge pull request #2665 from TheBlueMatt/2023-10-scanable-test-logs
valentinewallace [Mon, 16 Oct 2023 19:59:32 +0000 (15:59 -0400)]
Merge pull request #2665 from TheBlueMatt/2023-10-scanable-test-logs

Make test log lines somewhat more eye-scannable

13 months agoUse ChainHash instead of BlockHash as applicable
Jeffrey Czyz [Fri, 13 Oct 2023 22:01:19 +0000 (17:01 -0500)]
Use ChainHash instead of BlockHash as applicable

ChainHash is more appropriate for places where an arbitrary BlockHash is
not desirable. This type was introduced in later versions of the bitcoin
crate, thus BlockHash was used instead.

Using ChainHash also makes it easier to check if ChannelManager is
compatible with an Offer.

13 months agoMerge pull request #2664 from TheBlueMatt/2023-10-ci-disk-usage
Matt Corallo [Mon, 16 Oct 2023 14:56:34 +0000 (14:56 +0000)]
Merge pull request #2664 from TheBlueMatt/2023-10-ci-disk-usage

Reduce disk usage in CI

13 months agoMerge pull request #2625 from tnull/2023-09-moar-tests-n-fixes
Matt Corallo [Sun, 15 Oct 2023 20:18:56 +0000 (20:18 +0000)]
Merge pull request #2625 from tnull/2023-09-moar-tests-n-fixes

Improve test coverage of #2575 router fixes

13 months agoMake test log lines somewhat more eye-scannable 2023-10-scanable-test-logs
Matt Corallo [Thu, 12 Oct 2023 21:15:49 +0000 (21:15 +0000)]
Make test log lines somewhat more eye-scannable

When running tests, our log output should be reasonably readable
by developers, but currently it repeats the module twice (via the
module and file name), and then starts the log line at a variable
location.

Instead, we only print the module and then align the start of the
log lines so that the output is much more scannable.

13 months agoReduce disk usage in CI 2023-10-ci-disk-usage
Matt Corallo [Sat, 14 Oct 2023 18:41:34 +0000 (18:41 +0000)]
Reduce disk usage in CI

Recently github appears to have reduced the available free disk
space in actions runs, causing CI to fail with out of space errors.
Here we simply run `cargo clean` a few times in CI to reduce our
disk usage somewhat.

13 months agoMerge pull request #2655 from TheBlueMatt/2023-10-no-test-net
Matt Corallo [Sat, 14 Oct 2023 02:10:31 +0000 (02:10 +0000)]
Merge pull request #2655 from TheBlueMatt/2023-10-no-test-net

Replace `lightning-block-sync` test that depended on `foo.com`

13 months agoMerge pull request #2639 from vladimirfomene/fix-final-incorrect-cltv
valentinewallace [Fri, 13 Oct 2023 21:12:11 +0000 (17:12 -0400)]
Merge pull request #2639 from vladimirfomene/fix-final-incorrect-cltv

Fix final incorrect cltv

13 months agofix: use the update_add_htlc's cltv_expiry for comparison
Vladimir Fomene [Mon, 2 Oct 2023 19:05:42 +0000 (22:05 +0300)]
fix: use the update_add_htlc's cltv_expiry for comparison

As noted in BOLT 4, we should be using the update_add_htlc's cltv_expiry,
not the CLTV expiry set by the sender in the onion for this comparison.
See here: https://github.com/lightning/bolts/blob/4dcc377209509b13cf89a4b91fde7d478f5b46d8/04-onion-routing.md?plain=1#L334

13 months agoReplace `lightning-block-sync` test that depended on `foo.com` 2023-10-no-test-net
Matt Corallo [Mon, 9 Oct 2023 03:24:54 +0000 (03:24 +0000)]
Replace `lightning-block-sync` test that depended on `foo.com`

Our tests should generally not rely on internet access, and should
not rely on the behavior of any given remote server. However, one
of the `endpoint_tests` in `lightning-block-sync::http` relied on
`foo.com` resolving to a single socket address, which both might
change in the future and makes our tests fail without internet.