Several fields used in tracking on-chain HTLC outputs were
named `input_idx` despite referring to the output index in the
commitment transaction. Here they are all renamed
`commitment_tx_output_idx` for clarity.
For direct channels, the channel liquidity is known with certainty. Use
this knowledge in ProbabilisticScorer by either penalizing with the
per-hop penalty or u64::max_value depending on the amount.
Scorers could benefit from having the channel's EffectiveCapacity rather
than a u64 msat value. For instance, ProbabilisticScorer can give a more
accurate penalty when given the ExactLiquidity variant. Pass a struct
wrapping the effective capacity, the proposed amount, and any in-flight
HTLC value.
For route hints, the aggregate next hops path penalty and CLTV delta
should be computed after considering each hop rather than before.
Otherwise, these aggregate values will include values from the current
hop, too.
When scoring route hints, the amount passed to the scorer should include
any fees needed for subsequent hops. This worked correctly for single-
hop hints since there are no further hops, but not for multi-hint hops
(except the final one).
Using EffectiveCapacity in scoring gives more accurate success
probabilities when the maximum HTLC value is less than the channel
capacity. Change EffectiveCapacity to prefer the channel's capacity
over its maximum HTLC limit, but still use the latter for route finding.
This update also includes a minor refactor. The return type of
`pending_monitor_events` has been changed to a `Vec` tuple with the
`OutPoint` type. This associates a `Vec` of `MonitorEvent`s with a
funding outpoint.
We've also renamed `source/sink_channel_id` to `prev/next_channel_id` in
the favour of clarity.
As the `counterparty_node_id` is now required to be passed back to the
`ChannelManager` to accept or reject an inbound channel request, the
documentation is updated to reflect that.
Instead of including a `Secp256k1` context per
`PeerChannelEncryptor`, which is relatively expensive memory-wise
and nontrivial CPU-wise to construct, we should keep one for all
peers and simply reuse it.
This is relatively trivial so we do so in this commit.
Since its trivial to do so, we also take this opportunity to
randomize the new PeerManager context.
Because we handle messages (which can take some time, persisting
things to disk or validating cryptographic signatures) with the
top-level read lock, but require the top-level write lock to
connect new peers or handle disconnection, we are particularly
sensitive to writer starvation issues.
Rust's libstd RwLock does not provide any fairness guarantees,
using whatever the OS provides as-is. On Linux, pthreads defaults
to starving writers, which Rust's RwLock exposes to us (without
any configurability).
Here we work around that issue by blocking readers if there are
pending writers, optimizing for readable code over
perfectly-optimized blocking.
This avoids any extra calls to `read_event` after a write fails to
flush the write buffer fully, as is required by the PeerManager
API (though it isn't critical).
Only one instance of PeerManager::process_events can run at a time,
and each run always finishes all available work before returning.
Thus, having several threads blocked on the process_events lock
doesn't accomplish anything but blocking more threads.
Here we limit the number of blocked calls on process_events to two
- one processing events and one blocked at the top which will
process all available events after the first completes.
Because the peers write lock "blocks the world", and happens after
each read event, always taking the write lock has pretty severe
impacts on parallelism. Instead, here, we only take the global
write lock if we have to disconnect a peer.
Unlike very ancient versions of lightning-net-tokio, this does not
rely on a single global process_events future, but instead has one
per connection. This could still cause significant contention, so
we'll ensure only two process_events calls can exist at once in
the next few commits.
Users are required to only ever call `read_event` serially
per-peer, thus we actually don't need any locks while we're
processing messages - we can only be processing messages in one
thread per-peer.
That said, we do need to ensure that another thread doesn't
disconnect the peer we're processing messages for, as that could
result in a peer_disconencted call while we're processing a
message for the same peer - somewhat nonsensical.
This significantly improves parallelism especially during gossip
processing as it avoids waiting on the entire set of individual
peer locks to forward a gossip message while several other threads
are validating gossip messages with their individual peer locks
held.
