Commit Graph

6531 Commits

Author SHA1 Message Date
Rusty Russell
178baeba6c gossipd: get gossip_min_interval from lightningd.
Default is 5 x gossip interval == 5 minutes.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-20 06:55:00 +00:00
Rusty Russell
4cfd0524eb gossipd: simplify duplicate node_announcement check.
Write helpers to split it into non-timestamp, non-signature parts,
and simply compare those.  We extract a helper to do channel_update, too.

This is more generic than our previous approach, and simpler.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-20 06:55:00 +00:00
Rusty Russell
5ddd7866e4 gossipd: make create_node_announcement const-correct.
sig is only non-const so we can override if NULL, but talz helps
us here.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-20 06:55:00 +00:00
Rusty Russell
9578fb4035 lightningd: fix crash in cancel_channel
This path is not atomic, so we can't assume channel stays around.  It
could be the peer closes, it could be we get a parallel
fund_channel_cancel.  test_funding_cancel_race caused this crash:

    FATAL SIGNAL 6 (version 5f0a18e)
    backtrace: common/daemon.c:45 (send_backtrace) 0x55c7c373a429
    backtrace: common/daemon.c:53 (crashdump) 0x55c7c373a479
    backtrace: (null):0 ((null)) 0x7f88ee6ddf5f
    backtrace: (null):0 ((null)) 0x7f88ee6dded7
    backtrace: (null):0 ((null)) 0x7f88ee6bf534
    backtrace: ccan/ccan/tal/tal.c:93 (call_error) 0x55c7c379427c
    backtrace: ccan/ccan/tal/tal.c:165 (check_bounds) 0x55c7c3794444
    backtrace: ccan/ccan/tal/tal.c:174 (to_tal_hdr) 0x55c7c3794483
    backtrace: ccan/ccan/tal/tal.c:186 (to_tal_hdr_or_null) 0x55c7c3794504
    backtrace: ccan/ccan/tal/tal.c:421 (tal_alloc_) 0x55c7c3794c10
    backtrace: ccan/ccan/tal/tal.c:466 (tal_alloc_arr_) 0x55c7c3794ded
    backtrace: ccan/ccan/tal/str/str.c:91 (tal_vfmt_) 0x55c7c3793560
    backtrace: common/wire_error.c:22 (towire_errorfmtv) 0x55c7c3747f7b
    backtrace: common/wire_error.c:39 (towire_errorfmt) 0x55c7c37480a1
    backtrace: lightningd/channel_control.c:635 (process_check_funding_broadcast) 0x55c7c37015bb
    backtrace: lightningd/bitcoind.c:558 (process_gettxout) 0x55c7c36f8e75
    backtrace: lightningd/bitcoind.c:227 (bcli_finished) 0x55c7c36f8090
    backtrace: ccan/ccan/io/poll.c:244 (destroy_conn) 0x55c7c37869fe
    backtrace: ccan/ccan/io/poll.c:250 (destroy_conn_close_fd) 0x55c7c3786a1e
    backtrace: ccan/ccan/tal/tal.c:235 (notify) 0x55c7c3794629
    backtrace: ccan/ccan/tal/tal.c:397 (del_tree) 0x55c7c3794b18
    backtrace: ccan/ccan/tal/tal.c:481 (tal_free) 0x55c7c3794ea4
    backtrace: ccan/ccan/io/io.c:450 (io_close) 0x55c7c378521d
    backtrace: ccan/ccan/io/poll.c:449 (io_loop) 0x55c7c3787139
    backtrace: lightningd/io_loop_with_timers.c:24 (io_loop_with_timers) 0x55c7c370b26d
    backtrace: lightningd/lightningd.c:837 (main) 0x55c7c3711661
    backtrace: (null):0 ((null)) 0x7f88ee6c0b6a
    backtrace: (null):0 ((null)) 0x55c7c36f70b9
    backtrace: (null):0 ((null)) 0xffffffffffffffff

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-19 18:41:12 +08:00
Rusty Russell
794bebcc81 lightningd: add find_channel_by_id() helper, use it.
This is overkill for the moment, but we'll need it in the next patch.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-19 18:41:12 +08:00
trueptolemy
2d1a153975 plugins: Return command_param_failed() if param() fail 2019-09-19 01:07:11 +00:00
Rusty Russell
fe66b53fb9 jsonrpc: add more fields to listfunds.
This avoids having to correlate with listpeers for the most pertinent
information.

This API predates plugins, otherwise we'd have listutxos and listpeers
and this would simply combine them appropriately.  Still, it exists so
there's little reason not to make it more friendly.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-19 01:03:09 +00:00
Christian Decker
ac1c894255 db: Get old sqlite3_trace based db_changes working again
Seems I accidentally broke the fallback mechanism while working on the DB
abstraction.
2019-09-18 18:35:11 +02:00
trueptolemy
c5f4c9cefb hsmd: Fix a memory leak in status.c 2019-09-17 21:06:12 +02:00
trueptolemy
c737fa6b91 pay: Fix logic of the intereface find_worst_channel 2019-09-17 21:06:12 +02:00
trueptolemy
bdbfbf7e55 fundchannel: Remove the link of lightningd/json.h 2019-09-17 21:05:51 +02:00
trueptolemy
b4c6a5c0f0 plugins/.gitignore: Add fundchannel entry 2019-09-17 21:05:51 +02:00
darosior
697b501132 doc: place lightningd manpages first in index.rst 2019-09-16 13:16:08 +02:00
lisa neigut
aef5b30b81 wire: update how experimental csv patches are managed
We used to append new patches to a single file. This caused
some problems and is a lot harder to cleanup later.

This patch moves the experimental patches to their own, individual
patch files, that are named for the current BOLTVERSION, which they're
taken from.

Also moves the current patchfile over to a 'gossipqueries' one,
as it already exists.
2019-09-16 05:09:15 +00:00
Jacob Rapoport
6360ffb2ac fix macOS build & update docs 2019-09-16 05:08:42 +00:00
darosior
48fde4f0a5 wireaddr: correct the onion case in wireaddr_from_hostname
And, this time, add tests !
2019-09-16 05:05:24 +00:00
darosior
fad133ed2e wireaddr: pass an array to wireaddr_from_hostname
'wireaddr_from_hostname' now expects a tal array.
2019-09-16 05:05:24 +00:00
trueptolemy
ece0ec8bb9 CHANGELOG: Add the entry about pay can exclude error nodes 2019-09-16 12:22:06 +08:00
trueptolemy
d4104c15aa pytest: Test pay can exclude error node 2019-09-16 12:22:06 +08:00
trueptolemy
a3a0f651e3 pytest: Let fail_htlcs.py plugin generates TEMPORARY_NODE_FAILURE error 2019-09-16 12:22:06 +08:00
trueptolemy
22e8d242fc pay: Exclude nodes if the failcode of sendpay has the NODE bit set 2019-09-16 12:22:06 +08:00
trueptolemy
36852c36e6 CHANGELOG: Add the entry about getroute supports excluding nodes 2019-09-16 12:22:06 +08:00
trueptolemy
a795f5e600 doc: Update the document of getrouteabout excluding nodes 2019-09-16 12:22:06 +08:00
trueptolemy
059a6e0e0d pytest: Test excluding nodes in getroute 2019-09-16 12:22:06 +08:00
trueptolemy
5361a5d059 JSON-API: getroute now also support exclude nodes 2019-09-16 12:22:06 +08:00
trueptolemy
090a43fd3d gossip: Add the struct exclude_entry and enum exclude_entry_type 2019-09-16 12:22:06 +08:00
darosior
90fa2ae3fd plugins: load fundchannel by default
Reported by @darwin
2019-09-16 11:59:32 +08:00
darosior
04a780c715 README: Remove '-testnet' bitcoind's argument 2019-09-13 23:18:17 +00:00
Rusty Russell
a988ded3fa lightningd: fix case where config file overrides lightning-dir.
The simplest case is to explicitly load it when we see it's been
set.

This involves neatening the default config setup, to remove it from
opt_parse_from_config() and into the caller.  It also seems we don't
need to call it anymore before parsing early options: none of them
need ld->config set.

Closes: #3030
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 19:08:24 -05:00
Rusty Russell
f4e7ed179f devtools/mkencoded: tool to encode short_channel_id / other for testing
In particular, it does zlib.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
8ef996d56e devtools/mkgossip: tool to create gossip msgs for testing.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
a46e880f1d gossipd: in DEVELOPER mode, catch missing free_chan()
For memory-usage reasons, struct chan doesn't use a tal destructor, in
favor of us calling free_chan in the right places.

In DEVELOPER mode, we should check that is the case.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
5592cfb174 devtools/dump-gossipstore: add flag to show deleted entries.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
a3273d4c84 developer: IFDEV() macro.
There are some more #if DEVELOPER one-liners coming, this makes them
clear, but still lets them stand out.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
91072f56b0 developer: add 'dev-gossip-set-time' call to manipulate gossipd's time.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
768d293149 gossipd: don't get upset if we can't add channel_update.
In particular, the timestamp might be wrong once we start checking that.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
2577ad87d5 gossipd: use gossip_time_now() everywhere.
We've been slack, but it's going to be important for testing
ratelimiting.  And it currently has a minor memory leak.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
9aca38b279 pytest: increase non-developer timeout.
We overrode this with travis, but it's not a travis issue.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 05:11:56 +00:00
Rusty Russell
f554d6376f wireaddr: handle case where non-IPv6 and non-IPv4 address is returned.
Thanks clang!  Here's the error:
ommon/wireaddr.c:359:14: error: variable 'addr' is used uninitialized whenever

      'if' condition is false [-Werror,-Wsometimes-uninitialized]
                } else if (addrinfo->ai_family == AF_INET6) {
                           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
common/wireaddr.c:364:25: note: uninitialized use occurs here
                tal_arr_expand(addrs, addr);
                                      ^~~~
./common/utils.h:27:16: note: expanded from macro 'tal_arr_expand'
                (*(p))[n] = (s);                                        \
                             ^
common/wireaddr.c:359:10: note: remove the 'if' if its condition is always true
                } else if (addrinfo->ai_family == AF_INET6) {
                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
common/wireaddr.c:354:3: note: variable 'addr' is declared here
                struct wireaddr addr;
                ^

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2019-09-12 01:30:47 +00:00
darosior
5cded7863f connectd: correct a segfault in add_seed_addrs
lightning_connectd: FATAL SIGNAL 6 (version v0.6.1-1964-g226e2ae-modded)
0x5610a4a41b5d send_backtrace
	common/daemon.c:40
0x5610a4a41c03 crashdump
	common/daemon.c:53
0x7ff6bf71e83f ???
	???:0
0x7ff6bf71e7bb ???
	???:0
0x7ff6bf709534 ???
	???:0
0x5610a4a7a169 call_error
	ccan/ccan/tal/tal.c:93
0x5610a4a7a331 check_bounds
	ccan/ccan/tal/tal.c:165
0x5610a4a7a38f to_tal_hdr
	ccan/ccan/tal/tal.c:176
0x5610a4a7a3f1 to_tal_hdr_or_null
	ccan/ccan/tal/tal.c:186
0x5610a4a7b2d9 tal_bytelen
	ccan/ccan/tal/tal.c:632
0x5610a4a3a238 add_seed_addrs
	connectd/connectd.c:1282
0x5610a4a3a85c try_connect_peer
	connectd/connectd.c:1374
0x5610a4a3aaa2 connect_to_peer
	connectd/connectd.c:1419
2019-09-12 01:30:47 +00:00
lisa neigut
904a1384ef tests: de-flake lightningd_still_loading 2019-09-11 23:56:27 +00:00
lisa neigut
fe6c7f8f80 gossip queries: patch up valgrind errors in tests
These were giving me valgrind errors locally; fixed now.
2019-09-11 23:56:27 +00:00
lisa neigut
a35677fa0f fundchannel: use plugin; delete interior impl
Switch over to using the fundchannel plugin.
2019-09-11 23:56:27 +00:00
lisa neigut
548672ea0a fundchannel: populate max_funding from chainparams
Since we pull the chainparams out to get the funding_addr placeholder,
we can also get the max_funding now!
2019-09-11 23:56:27 +00:00
lisa neigut
c261309f18 test: fund wallet before testing fundchannel
After switching to a plugin, we verify that we can fund a channel
before we check to contact a peer. We'll need to have a funded wallet
to pass the check in this test that verifies that 'fundchannel' cannot
be called for a peer after fundchannel_start is.
2019-09-11 23:56:27 +00:00
lisa neigut
42ef64941a fundchannel: have address work with any network
`txprepare` verifies that you're sending to an address on the
right network, so this builds a placeholder address for the
right network.
2019-09-11 23:56:27 +00:00
lisa neigut
7d069239aa fundchannel: only cap to 'max' if keyword 'all' passed in
in order to preserve current behavior, we cap at max if specified 'all';
otherwise we fail since the amount requested is larger than the
channel max capacity
2019-09-11 23:56:27 +00:00
lisa neigut
d9843c1957 fundchannel: handle 'all' for satoshi amount
Previously, we'd fail on 'all'. `fundchannel_start` needs an amount
in order to start a funding transaction.

The way that we approach this is to first call `txprepare` with a
placeholder address and the 'all' amount; this will return the maximum
amount available. We then clamp this to the max_funding (currently
hardcoded, in the future we'd want to consult our/the peer's features) and
then use the amount on the output in the prepared transaction as the
funding amount. We then pass this amount to fundchannel_start,
after we've started it successfully we cancel the held placeholder
transaction and prepare a second transaction for the exact amount,
using the funding address that fundchannel_start passed back.
2019-09-11 23:56:27 +00:00
lisa neigut
fdcacbc7f0 wallet: check that funding_tx is affordabe if we trim change
In some cases, the funding transaction is affordable only
if we leave out the change amount.
2019-09-11 23:56:27 +00:00
lisa neigut
38e404af51 wallet: add flag to specify whether or not to include change output
Will allow for more intelligent change policy
2019-09-11 23:56:27 +00:00