Trying to put all the disconnect logic into the same path was a dumb
idea. If you asked to reconnect but passed in an 'unsaved' channel, we
would not call the 'reconnect' code.
Instead, we make a differentiation between "unsaved" channels
(ones that we haven't received commitment tx for) and handle the
disconnect for these separate from where we want to do a reconnect.
We generally hang things off our JSON response (this pattern predates
tmpctx!) but sometimes it gets reported as a memleak. I'd prefer not
to mark JSON responses as "notleak", since they can be allocated for
a while), so use tmpctx here.
```
E ValueError:
E Node errors:
E Global errors:
E - Node /tmp/ltests-spnausnb/test_htlc_out_timeout_1/lightning-1/ has memory leaks: [
E {
E "backtrace": [
E "ccan/ccan/tal/tal.c:442 (tal_alloc_)",
E "ccan/ccan/tal/tal.c:471 (tal_alloc_arr_)",
E "wallet/wallet.c:1775 (wallet_state_change_get)",
E "lightningd/peer_control.c:922 (json_add_channel)",
E "lightningd/peer_control.c:1424 (json_add_peer)",
E "lightningd/peer_control.c:1454 (json_listpeers)",
E "lightningd/jsonrpc.c:643 (command_exec)",
E "lightningd/jsonrpc.c:767 (rpc_command_hook_final)",
E "lightningd/plugin_hook.c:275 (plugin_hook_call_)",
E "lightningd/jsonrpc.c:855 (plugin_hook_call_rpc_command)",
E "lightningd/jsonrpc.c:942 (parse_request)",
E "lightningd/jsonrpc.c:1033 (read_json)",
E "ccan/ccan/io/io.c:59 (next_plan)",
E "ccan/ccan/io/io.c:435 (io_do_always)",
E "ccan/ccan/io/poll.c:300 (handle_always)",
E "ccan/ccan/io/poll.c:377 (io_loop)",
E "lightningd/io_loop_with_timers.c:24 (io_loop_with_timers)",
E "lightningd/lightningd.c:1097 (main)"
E ],
E "label": "wallet/wallet.c:1775:struct state_change_entry[]",
E "parents": [
E "common/json_stream.c:29:struct json_stream",
E "ccan/ccan/io/io.c:91:struct io_conn",
E "lightningd/lightningd.c:116:struct lightningd"
E ],
E "value": "0x55c6b02150b8"
E }
E ]
```
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This means remembering the connection direction. We also use the address to try
to reconnect, which we shouldn't bother with if they connect to us.
For peers from the database, we currently always save the addr: we shouldn't really
do this if they connected to us, since it's not useful for reconnecting (we don't
show the addr in JSON reply to listpeers unless we're connected, so it's only an
internal issue). This is left for future work.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This matters: if we connected, the address is probably usable for future connections.
But if they connected, the port is probably not (but the IP address may be).
Changelog-Added: JSON-RPC: `connect` returns "direction" ("in": they iniatated, or "out": we initiated)
Changelog-Added: plugins: `peer_connected` hook and `connect` notifications have "direction" field.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
You can now activate dual-funded channels using the
`--experimental-dual-fund` flag
Changelog-Changed: Config: `--experimental-dual-fund` runtime flag will enable dual-funded protocol on this node
Otherwise, we might find an address other than the one given and
the user might think that address worked.
Fixes: #4185
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Added: JSON-RPC: `connect` returns `address` it actually connected to
Changelog-Added: lightningd: experimental-shutdown-wrong-funding to allow remote nodes to close incorrectly opened channels.
Changelog-Added: JSON-RPC: close has a new `wrong_funding` option to try to close out unused channels where we messed up the funding tx.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Looks like #4394 treated a symptom but not the root cause. We were
actually sending the message framed with the WIRE_CUSTOMMSG_OUT and
the length prefix over the encrypted connection to the peer. It just
happened to be a valid custommsg...
This fixes the issue, and this time I made sure we actually send the
raw message over the wire. However for backward compatibility we
needed to imitate the faulty behavior which is 90% of this patch :-)
Changelog-Fixed: plugin: `dev-sendcustommsg` included the type and length prefix when sending a message.
Changelog-Added: JSON-RPC: `listpeers` now includes 'last_feerate', 'next_feerate', 'initial_feerate' and 'next_fee_step' for channels in state DUALOPEND_AWAITING_LOCKIN
fixup! listpeers: include feerate info for RBF-candidate channels
We move over to the new "warning" paradigm, instead of using
an "rbf_fail" message.
Every failure is either a warning or an error; on warnings we
hang up and reconnect later, effectively resetting the state.
Users have no idea what they would pay for unilateral closes.
At least this gives them a clue!
Reported-by: @az0re on IRC.
Changelog-Added: JSON-RPC: `listpeers` now shows latest feerate and unilaral close fee.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We were always prefixing the `message` field with the internal type
prefix 0x0407, followed by the length prefix. Neither is needed since
the type being constant is of no interest to the plugin and the length
being implicit due to the JSON-encoding.
Reported-by: Ilya Evdokimov
Changelog-Fixed: plugin: The `custommsg` hook no longer includes the internal type prefix and length prefix in its `payload`
Changelog-Deprecated: plugin: The `message` field on the `custommsg` hook is deprecated in favor of the `payload` field, which skips the internal prefix.
Since we turned many errors into warnings, we want our tests to fail
when they happen unexpectedly. We make WARNING clear in the strings
we print, too, to help out.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
No more sending "all-channel" errors; in particular, gossipd now only
sends warnings (which make us hang up), not errors, and peer_connected
rejections are warnings (and disconnect), not errors.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Changed: Plugins: `peer_connected` rejections now send a warning, not an error, to the peer.
This is in line with the warnings draft, where all-zeroes in a
channel_id is no longer special (i.e. it will be ignored).
But gossipd would send these if it got upset with us, so it's best
practice to ignore them for now anyway.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Added: Protocol: we treat error messages from peer which refer to "all channels" as warnings, not errors.
rearranges the`peer_connected_hook_payload` definition to the location
where this is used in the file.
Fixes certain blanklines and linebreaks to make the code look nicer.
payload is owned by the peer, which is freed in this case, then we
free payload (again).
==1404== Invalid read of size 8
==1404== at 0x1F39E8: to_tal_hdr (tal.c:174)
==1404== by 0x1F43A4: tal_free (tal.c:479)
==1404== by 0x14B3D1: peer_connected_hook_cb (peer_control.c:1087)
==1404== by 0x15D6E9: plugin_hook_call_ (plugin_hook.c:288)
==1404== by 0x14B40E: plugin_hook_call_peer_connected (peer_control.c:1090)
==1404== by 0x14B5B8: peer_connected (peer_control.c:1135)
==1404== by 0x122FCF: connectd_msg (connect_control.c:310)
==1404== by 0x160291: sd_msg_read (subd.c:480)
==1404== by 0x15FBE7: read_fds (subd.c:308)
==1404== by 0x1E37D1: next_plan (io.c:59)
==1404== by 0x1E434E: do_plan (io.c:407)
==1404== by 0x1E438C: io_ready (io.c:417)
==1404== Address 0x2fcd2268 is 24 bytes inside a block of size 336 free'd
==1404== at 0x4C32D3B: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==1404== by 0x1F416E: del_tree (tal.c:421)
==1404== by 0x1F40F2: del_tree (tal.c:412)
==1404== by 0x1F442C: tal_free (tal.c:486)
==1404== by 0x148816: delete_peer (peer_control.c:120)
==1404== by 0x148899: maybe_delete_peer (peer_control.c:136)
==1404== by 0x13A970: destroy_uncommitted_channel (opening_common.c:29)
==1404== by 0x1F3BB1: notify (tal.c:240)
==1404== by 0x1F40A0: del_tree (tal.c:402)
==1404== by 0x1F442C: tal_free (tal.c:486)
==1404== by 0x13D3E9: peer_start_openingd (opening_control.c:911)
==1404== by 0x14B3C2: peer_connected_hook_cb (peer_control.c:1086)
==1404== Block was alloc'd at
==1404== at 0x4C31B0F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==1404== by 0x1F3C1B: allocate (tal.c:250)
==1404== by 0x1F41B4: tal_alloc_ (tal.c:428)
==1404== by 0x14B454: peer_connected (peer_control.c:1105)
==1404== by 0x122FCF: connectd_msg (connect_control.c:310)
==1404== by 0x160291: sd_msg_read (subd.c:480)
==1404== by 0x15FBE7: read_fds (subd.c:308)
==1404== by 0x1E37D1: next_plan (io.c:59)
==1404== by 0x1E434E: do_plan (io.c:407)
==1404== by 0x1E438C: io_ready (io.c:417)
==1404== by 0x1E6552: io_loop (poll.c:445)
==1404== by 0x12E2AD: io_loop_with_timers (io_loop_with_timers.c:24)
Fixes: #4329
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Prior to this, all reconnect logic lived in channeld. If you
disconnected before we finished building a funding transaction, that was
no big deal. Now, however, we're waiting for the funding to lock in in
dualopend, instead of handing straight to channeld to wait.
So we need a way to restart dualopend.
v2 channel opens are going to happen over in dualopend. In order
to make sure that these don't end up in the wrong place/to keep track of
the difference between "waiting for sigs" and "have merely initiatlized
a channel", we add two new states to the channel state machine.
A channel that 'originates' in dualopend will only ever arrive at
channeld in the state CHANNELD_NORMAL.
Since this all stays in dualopend/dual_open_control, we can hold
onto the openchannel_signed command to wait for a response here locally.
Previously we were splitting across the channeld/openingd boundary.
Our new "decode" command will also handle bolt11. We make a few cleanups:
1. Avoid type_to_string() in JSON, instead use format functions directly.
2. Don't need to escape description now that JSON core does that for us.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
niftynei
Rusty Russell
Christian Decker
Michael Schmoock