Since we only care about the latest commits, we can simply associate a
state with each HTLC, rather than using queues of HTLCs associated
with each commitment transaction.
This works far better in the context of a database.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
From doing a code walkthrough with Christian Decker; unnecessary const in
bitcoin/tx.c, an erroneous FIXME, a missing comment, and an unused struct.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is the command an actual user would use: it figures out the fee
and route, and pays it if it can.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
If a block triggers two peers to close, we ran io_break() on both of them; the
second overrode the first and we didn't end up freeing that one.
Rather than chase such bugs in future, simply iterate to see if any
peers need freeing.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Note that the base fee is in millisatoshi, the proportional fee is
in microsatoshi per satoshi. ie. 1,000,000 means charge 1 satoshi for
every satoshi carried.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Most HTLCs we offer are triggered by an incoming HTLC from a different
peer. Save this "source" htlc, so we can fail/fulfill it when we
fail/fulfill this one.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
No more copies!
I tried changing the cstate->side[].htlcs to htlc_map rather than a
simple pointer array, but we rely on those array indices heavily for
permutation mapping, and it turned into a major rewrite (especially
for the steal case).
Eventually, we're going to want to reconstruct the commit info for
older commit txs rather than keeping all the permutation and
per-commit-info HTLC information in memory, so we can do the work
then.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It's a more logical name, and a more logical place. We change
"funding" to "channel" in the remaining exposed symbols, too.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is the more normal case; find by ID. The low-level json commands are
really just for testing.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The protocol still supports both, but we now only support blocks.
It's hard to do risk management with timeouts in seconds, given block
variance. This is also signficantly simpler, as HTLC timeouts are
always fired in response to blocks, not wall-clock times.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We need to know when changes are fully committed by both sides:
1) For their HTLC_ADDs, this is when we can fulfill/fail/route.
2) For their HTLC_FAILs, this is when we can fail incoming.
For HTLC_FULFULL we don't need to wait: as soon as we know the preimage
we can propogate it.
For the moment, we simply log and assert; acting on it comes later.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We've been stuffing these into sha256s, but they're actually nonces.
Create a new structure for that for clarity.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
And make the add/fail/fulfill arg a pointer to a union htlc_staging
directly, removing struct htlc_progress.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
There's no real reason to avoid commands for the next commit; this has
the benefit that we can remove the infrastructure to queue commands.
The only exceptions are the commit command and the opening phase.
We still only allow one commit at a time, but that's mainly run off a
timer which can try again later. For the JSONRPC API used for
testing, we can simply fail the commit if one is in progress.
For opening we add an explicit peer_open_complete() call in place of
using the command infrastructure.
Commands are now outside the state machine altogether: we simply have
it return the new state instead of the command status. The JSONRPC
functions can also now run commands directly.
This removes the idea of "peercond" as well: you can simply examine
the states to determine whether an input is valid. There are
fine-grained helpers for this now, too.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We're about to allow changes while we're waiting for a commit ack.
This means we can't have a single "unacked changes" queue; when we
receive the revocation reply, we need to apply the unacked changes
known at the time we sent the commit, not any we've created since
then.
Note that we still only have a single staged_commit; we never have two
outstanding commits, since for simplicity we will still block
following update_commit pending the reply to the current one.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We already removed the on-chain states, now we remove the "clearing" state
(which wasn't fully implemented anyway).
This turns into two smaller state machines: one for clearing, which
still allows HTLCs to be failed and fulfilled, and one for mutual
close negotiation which only allows close_signature messages.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
As per lightning-rfc commit 8ee09e749990a11fa53bea03d5961cfde4be4616,
we remove the acks from the protocol now they're no longer needed (and
all the infrastructure).
We also place the commit number in the commit_info where it logically
belongs, removing it from the peer struct.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
From BOLT#2 (rev 8ee09e749990a11fa53bea03d5961cfde4be4616):
Thus each node (conceptually) tracks:
...
3. Two *unacked changesets*: one for the local commitment (their proposals) and one for the remote (our proposals)
4. Two *acked changesets*: one for the local commitment (our proposals, acknowledged) and one for the remote (their proposals, acknowledged).
(Note that an implementation MAY optimize this internally, for
example, pre-applying the changesets in some cases).
In our case, we apply the unacked changes immediately into
staging_cstate, and save them in an unacked_changes array. That array
gets applied to staging_cstate as soon as it's acked (we only allow
one outstanding update_commit, so we only need one array).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
While useful for testing, it doesn't make sense to have an explicit commit
command; we should commit whenever there are outstanding changes.
We have a 10ms timer to allow limited batching, however.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Once we see an on-chain tx, we ignore the state machine and handle it
as per the onchain.md draft. This specifies a *resolution* for each
output, and we're done when they're irrevocable.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It's not quite true: if we offer the anchor, we have a commitinfo
without their signature yet. So make it a pointer again. Since we
always allocate struct commit_info with talz, it starts as a NULL
pointer.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It's primitive, but we re-broadcast any txs not included in the main
chain every time the tip moves. We only track transactions we are
watching, but that turns out to cover every transaction we generate
anyway.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
You can't re-enter the state machine from a callback, so this allows you
to queue an input for when it returns.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We just use a p2sh to a single address for the moment, but that's simply for
non-segwit wallets; we'll pay to whatever the other side specifies.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rather than p2sh of a 2of2, it's now a version 0 witness program.
This means that the commit transaction input and mutual close
transaction input are both different.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
There isn't a single blockhash; we may be on multiple forks. But the one
caller which cares is commit_tx_depth(), which wants to know if the tx is
spendable yet. So that uses get_last_mediantime().
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We now keep a list of commitment transaction states for "us" and
"them", as well as a "struct channel_state" for staged changes.
We manipulate these structures as we send out packets, receive
packets, or receive acknowledgement of packets. In particular, we
update the other nodes' staging_cstate as we send out our requests,
and update our own staging_cstate are we receive acks. When we
receive a request, we update both (as we immediately send out our
ack).
The RPC output is changed; rather than expose the complexity, we
expose our last committed state: what would happen if we have to drop
to the blockchain now.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rather than creating packets then queueing them, call out to functions
which do both. This moves us towards doing more work in those functions
where we send out a request, which is sometimes clearer.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
And divide fees as specified there.
We still use fixed values rather than floating, and we don't send or
handle update_fee messages.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We don't actually implement closing when we have HTLCs (we should
allow it, as that's what the clearing phase is for), since soon we'll
rewrite HTLC to match the async HTLC protocol of BOLT #2.
Note that this folds the close paths, using a simple check if we have
a close transaction. That's a slight state layer violation, but
reduces code duplication.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This means we send the first two revocation hashes; this is important
once we move to a commit model as we need to send (unsolicited) the
signature for the *next* commit tx so we need its commit hash.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This encapsulates proposals more cleanly, and is important when we change
the protocol to have more than one outstanding at a time.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell
sstone