This is a bit messier than I'd like, but we want to clearly remove all
dev code (not just have it uncalled), so we remove fields and functions
altogether rather than stub them out. This means we put #ifdefs in callers
in some places, but at least it's explicit.
We still run tests, but only a subset, and we run with NO_VALGRIND under
Travis to avoid increasing test times too much.
See-also: #176
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
test_routing_gossip (__main__.LightningDTests) ... lightningd: Outstanding taken pointers: lightningd/peer_control.c:2352:towire_errorfmt(ld, ((void *)0), "Can't resolve your address")
This caused by the other end closing due to the next bug.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
There is a race we see sometimes under valgrind on Travis which shows
gossipd receiving the node_announce from master before it reads the
channel_announce from channeld, and thus fails. The simplest solution
is to send the channel_announce and channel_update to master as well,
so it can ensure it sends them to gossipd in order
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
There are now only two kinds of subdaemons: global ones (hsmd, gossipd) and
per-peer ones. We can handle many callbacks internally now.
We can have a handler to set a new peer owner, and automatically do
the cleanup of the old one if necessary, since we now know which ones
are per-peer.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We currently rely on a zero exit status. That's the only difference between
onchain finished handling and other per-peer daemons, so instead we should
have an explicit "done" message. This is both clearer, and allows us to
unify.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Now the flow is much simpler from a lightningd POV:
1. If we want to connect to a peer, just send gossipd `gossipctl_reach_peer`.
2. Every new peer, gossipd hands up to lightningd, with global/local features
and the peer fd and a gossip fd using `gossip_peer_connected`
3. If lightningd doesn't want it, it just hands the peerfd and global/local
features back to gossipd using `gossipctl_handle_peer`
4. If a peer sends a non-gossip msg (eg `open_channel`) the gossipd sends
it up using `gossip_peer_nongossip`.
5. If lightningd wants to fund a channel, it simply calls `release_channel`.
Notes:
* There's no more "unique_id": we use the peer id.
* For the moment, we don't ask gossipd when we're told to list peers, so
connected peers without a channel don't appear in the JSON getpeers API.
* We add a `gossipctl_peer_addrhint` for the moment, so you can connect to
a specific ip/port, but using other sources is a TODO.
* We now (correctly) only give up on reaching a peer after we exchange init
messages, which changes the test_disconnect case.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We're going to make the ip/port optional, so they should go at the end.
In addition, using ip:port is nicer, for gethostbyaddr().
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We have to do a dance when we get a reconnect in openingd, because we
don't normally expect to free both owner and peer. It's a layering
violation: freeing a peer should clean up the owner's pointer to it,
to avoid a double free, and we can eliminate this dance.
The free order is now different, and the test_reconnect_openingd was
overprecise.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This fixes the only case where the master currently has to write directly
to the peer: re-sending an error. We make gossipd do it, by adding
a new gossipctl_fail_peer message.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We pull them from the database on-demand, where we're storing them
anyway. No need to keep them in memory as well.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
No idea why we were iterating over the list of stubs and then passing
in the index instead of a pointer to the stub directly.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
This wires in the loading of `struct htlc_stub`s on-demand when
starting `onchaind` so that we don't need to keep them in memory.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
peer_fail_permanent() frees peer->owner, but for bad_peer() we're
being called by the sd->badpeercb(), which then goes on to
io_close(conn) which is a child of sd.
We need to detach the two for this case, so neither tries to free the
other.
This leads to a corner case when the subd exits after the peer is gone:
subd->peer is NULL, so we have to handle that too.
Fixes: #282
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We have a race where we start onchaind, but state is unchanged, so checks
like peer_control.c's:
peer_ready = (peer->owner && peer->state == CHANNELD_AWAITING_LOCKIN);
if (!peer_ready) {
log_unusual(peer->log,
"Funding tx confirmed, but peer state %s %s",
peer_state_name(peer->state),
peer->owner ? peer->owner->name : "unowned");
} else {
subd_send_msg(peer->owner,
take(towire_channel_funding_locked(peer,
peer->scid)));
}
Can send to the wrong daemon.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We re-use the value for reasonable_depth given by the master, and we
tell it when our timeout transactions reach that depth.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
When we see an offered HTLC onchain, we need to use the preimage if we
know it. So we dump all the known HTLC preimages at startup, and send
new ones as we discover them.
This doesn't cover preimages we know because we're the final
recipient; that can happen if an HTLC hasn't been irrevocably
committed yet. We'll do that in a followup patch.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
If the HSM is slow it might happen that the timestamp has changed the
second time we come around, so we generate the timestamp externally
and pass it in so we're sure it won't change between calls.
Reported-by: Rusty Russell
Signed-off-by: Christian Decker <decker.christian@gmail.com>
This change is really to allow us to have a --dev-fail-on-subdaemon-fail option
so we can handle failures from subdaemons generically.
It also neatens handling so we can have an explicit callback for "peer
did something wrong" (which matters if we want to close the channel in
that case).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is the step where we broadcast the transaction to the network and
a nice place to extract the change from the transaction.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
So far we were always using the deadline in the announcements, that's
obviously not good, so this introduces the parameter as per spec.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
We weren't killing it. Eventually it would die, and peer_owner_finished()
would access subd->peer->owner, but that peer was freed already.
Closes: #261
Reported-by: Christian Decker <decker.christian@gmail.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
To reproduce the next bug, I had to ensure that one node keeps thinking it's
disconnected, then the other node reconnects, then the first node realizes
it's disconnected.
This code does that, adding a '0' dev-disconnect modifier. That means
we fork off a process which (due to pipebuf) will accept a little
data, but when the dev_disconnect file is truncated (a hacky, but
effective, signalling mechanism) will exit, as if the socket finally
realized it's not connected any more.
The python tests hang waiting for the daemon to terminate if you leave
the blackhole around; to give a clue as to what's happening in this
case I moved the log dump to before killing the daemon.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
In this case, we unset the old subd->peer, then freed subd.
peer_owner_finished dereferenced subd->peer->owner, and boom:
test_disconnect_funder (__main__.LightningDTests) ... Fatal signal 11. Log dumped in crash.log
------------------------------- Valgrind errors --------------------------------
Valgrind error file: valgrind-errors.2882
==2882== Invalid read of size 8
==2882== at 0x413F74: peer_owner_finished (peer_control.c:679)
==2882== by 0x41EA2C: destroy_subd (subd.c:381)
==2882== by 0x459700: notify (tal.c:240)
==2882== by 0x459BB1: del_tree (tal.c:400)
==2882== by 0x459FC0: tal_free (tal.c:509)
==2882== by 0x413796: peer_reconnected (peer_control.c:493)
==2882== by 0x413A6A: add_peer (peer_control.c:592)
==2882== by 0x40ED1F: handshake_succeeded (new_connection.c:186)
==2882== by 0x41E3DD: sd_msg_reply (subd.c:262)
==2882== by 0x41E6BB: sd_msg_read (subd.c:318)
==2882== by 0x41E4E6: read_fds (subd.c:283)
==2882== by 0x44DEB4: next_plan (io.c:59)
==2882== Address 0x838 is not stack'd, malloc'd or (recently) free'd
==2882==
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The logic of dispatching the announcement_signatures message was
distributed over several places and daemons. This aims to simplify it
by moving it all into `channeld`, making peer_control only report
announcement depth to `channeld`, which then takes care of the
rest. We also do not reuse the funding_locked tx watcher since it is
easier to just fire off a new watcher with the specific purpose of
waiting for the announcement_depth.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
