This is in preparation for gossipd feeding us the latest channel_updates,
rather than having lightningd and channeld query gossipd when it wants
to send an onion error with an update included.
This means gossipd will start telling us the updates, so we need the
channels loaded first.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
lightningd would race with the subd destructor to do the waitpid(),
resulting in UNUSUAL log messages, but also us missing if a plugin
was killed via a signal.
We can also get rid of the gratuitous waitpid() in test_subdaemons.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Get rid of the 'movement_idx', since we replay events now.
Since we're removing a field from the 'coin_movement' event emission, we
bump the version type.
Changelog-Updated: `coin_movements` events have been revamped and are now on version 2.
The only thing that needs ld->wallet after this is destroy_invoices_waiter (off jsonrpc)
Could not find any other destructors (destroy_*) that need wallet or db access after this.
Any db access would now segfault.
The idea is to have different default ports for different networks.
Current default port is `9735` for everything. Let's use it for
the mainnet and reuse the difference added to the default port
from `rpc_port` values in `bitcoin/chainstate.c`.
Testnet would be `19735` (adding rpc_port - 8332 = `10000`).
Signet would be `39735` (adding rpc_port - 8332 = `30000`).
Regtest would be `19846` (adding rpc_port - 8332 = `10111`).
With Vincenzo's kind pair-programming help over tmate.
Two other commits were squashed into this one so that bisecting
never ends up in half-baked state:
1. chainparams: Fix regtest default rpc_port
bitcoind -help says this:
-rpcport=<port>
Listen for JSON-RPC connections on <port> (default: 8332, testnet:
18332, signet: 38332, regtest: 18443)
2. test_gossip: Default port for regtest
hex: 2607 is now .... (could be 4d86 but Elements uses another port)
dec: 9735 is now any port (could be 19846 ^^ but now is for any port)
The lines which were binding to default port were removed as the
default port is different on each network.
NOTE: Remember not to modify gossip_store tests which loads everything raw
including the checksums.
Changelog-Changed: If the port is unspecified, the default port is chosen according to used network similarly to Bitcoin Core.
And turn "" includes into full-path (which makes it easier to put
config.h first, and finds some cases check-includes.sh missed
previously).
config.h sets _GNU_SOURCE which really needs to be done before any
'#includes': we mainly got away with it with glibc, but other platforms
like Alpine may have stricter requirements.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Setting SIGCHLD back to default (i.e. ignored) makes waitpid hang on an
old SIGCHLD that was still in the pipe?
This happens running test_important_plugin with developer=1:
(or with dev=0 and build-in plugins subscribed to "shutdown")
0 0x00007ff8336b6437 in __GI___waitpid (pid=-1, stat_loc=0x0, options=1) at ../sysdeps/unix/sysv/linux/waitpid.c:30
1 0x000055fb468f733a in sigchld_rfd_in (conn=0x55fb47c7cfc8, ld=0x55fb47bdce58) at lightningd/lightningd.c:785
2 0x000055fb469bcc6b in next_plan (conn=0x55fb47c7cfc8, plan=0x55fb47c7cfe8) at ccan/ccan/io/io.c:59
3 0x000055fb469bd80b in do_plan (conn=0x55fb47c7cfc8, plan=0x55fb47c7cfe8, idle_on_epipe=false) at ccan/ccan/io/io.c:407
4 0x000055fb469bd849 in io_ready (conn=0x55fb47c7cfc8, pollflags=1) at ccan/ccan/io/io.c:417
5 0x000055fb469bfa26 in io_loop (timers=0x55fb47c41198, expired=0x7ffdf4be9028) at ccan/ccan/io/poll.c:453
6 0x000055fb468f1be9 in io_loop_with_timers (ld=0x55fb47bdce58) at lightningd/io_loop_with_timers.c:21
7 0x000055fb46924817 in shutdown_plugins (ld=0x55fb47bdce58) at lightningd/plugin.c:2114
8 0x000055fb468f7c92 in main (argc=22, argv=0x7ffdf4be9228) at lightningd/lightningd.c:1195
since PR #3867 utxos are unreserved by height, destroy_utxos and
related functions are not used anymore so clean them up also
However free(ld->jsonrpc) still needs to happen before free(ld) because its
destructors need list_head pointers from ld
because:
- shutdown_subdaemons can trigger db write, comments in that function say so at least
- resurrecting the main event loop with subdaemons still running is counter productive
in shutting down activity (such as htlc's, hook_calls etc.)
- custom behavior injected by plugins via hooks should be consistent, see test
in previous commmit
IDEA:
in shutdown_plugins, when starting new io_loop:
- A plugin that is still running can return a jsonrpc_request response, this triggers
response_cb, which cannot be handled because subdaemons are gone -> so any response_cb should be blocked/aborted
- jsonrpc is still there, so users (such as plugins) can make new jsonrpc_request's which
cannot be handled because subdaemons are gone -> so new rpc_request should also be blocked
- But we do want to send/receive notifications and log messages (handled in jsonrpc as jsonrpc_notification)
as these do not trigger subdaemon calls or db_write's
Log messages and notifications do not have "id" field, where jsonrpc_request *do* have an "id" field
PLAN (hypothesis):
- hack into plugin_read_json_one OR plugin_response_handle to filter-out json with
an "id" field, this should
block/abandon any jsonrpc_request responses (and new jsonrpc_requests for plugins?)
Q. Can internal (so not via plugin) jsonrpc_requests called in the main io_loop return/revive in
the shutdown io_loop?
A. No. All code under lightningd/ returning command_still_pending depends on either a subdaemon, timer or
plugin. In shutdown loop the subdaemons are dead, timer struct cleared and plugins will be taken
care of (in next commits).
fixup: we can only io_break the main io_loop once
This loads up 20MB of plugins temporarily; we seem to be getting OOM
killed under CI and I wonder if this is contributing.
Doesn't significantly reduce runtime here, but I have lots of memory.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Fixes: #4868
ChangeLog-Fixed: We now no longer self-limit the number of file descriptors (which limits the number of channels) in sufficiently modern systems, or where we can access `/proc` or `/dev/fd`. We still self-limit on old systems where we cannot find the list of open files on `/proc` or `/dev/fd`, so if you need > ~4000 channels, upgrade or mount `/proc`.
This also inadvertently fixes a latent bug: before this patch, in the
`subd` function in `lightningd/subd.c`, we would close `execfail[1]`
*before* doing an `exec`.
We use an EOF on `execfail[1]` as a signal that `exec` succeeded (the
fd is marked CLOEXEC), and otherwise use it to pump `errno` to the
parent.
The intent is that this fd should be kept open until `exec`, at which
point CLOEXEC triggers and close that fd and sends the EOF, *or* if
`exec` fails we can send the `errno` to the parent process vua that
pipe-end.
However, in the previous version, we end up closing that fd *before*
reaching `exec`, either in the loop which `dup2`s passed-in fds (by
overwriting `execfail[1]` with a `dup2`) or in the "close everything"
loop, which does not guard against `execfail[1]`, only
`dev_disconnect_fd`.
After recent header files clean-up it was not possible to
build c-lightning 7401b2682. This patch fixes it both for
Alpine Linux and OpenBSD.
Proposed-by: nathanael <nathanael@dalliard.ch>
Changelog-None
Before:
Ten builds, laptop -j5, no ccache:
```
real 0m36.686000-38.956000(38.608+/-0.65)s
user 2m32.864000-42.253000(40.7545+/-2.7)s
sys 0m16.618000-18.316000(17.8531+/-0.48)s
```
Ten builds, laptop -j5, ccache (warm):
```
real 0m8.212000-8.577000(8.39989+/-0.13)s
user 0m12.731000-13.212000(12.9751+/-0.17)s
sys 0m3.697000-3.902000(3.83722+/-0.064)s
```
After:
Ten builds, laptop -j5, no ccache: 8% faster
```
real 0m33.802000-35.773000(35.468+/-0.54)s
user 2m19.073000-27.754000(26.2542+/-2.3)s
sys 0m15.784000-17.173000(16.7165+/-0.37)s
```
Ten builds, laptop -j5, ccache (warm): 1% faster
```
real 0m8.200000-8.485000(8.30138+/-0.097)s
user 0m12.485000-13.100000(12.7344+/-0.19)s
sys 0m3.702000-3.889000(3.78787+/-0.056)s
```
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
That was quick!
We remove the 50% test, since the default is now to use quickclose.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Added: Protocol: We now perform quick-close if the peer supports it.
Based on a commit by @niftynei, but:
- Separated quickclose logic from main loop.
- I made it indep of anchor_outputs, use and option instead.
- Disable if they've specified how to negotiate.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Not necessary yet, but it will be once shutdown starts waiting for
plugins to respond: we don't want these to try to access the bcli
plugin once it's freed.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It's a legacy from when it didn't have an ld pointer to access ld's
timer structure. Now it's just confusing.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This renames all occurences of use_proxy_always to always_use_proxy
to keep it inline with config values. This was a bit confusing.
Only significant change is that the payload in the plugins init
requests also contained the old name. No plugin currently seems to make
use of this variable yet. The old name 'use_proxy_always' is added when
deprecated APIs is enabled.
Changelog-Deprecated: Plugins: Renames plugin init 'use_proxy_always' to 'always_use_proxy'
In fact, we make it compulsory, which means if you don't understand it
you'll hang up on us!
Add some logging for that in future.
Changelog-Changed: Protocol: All new invoices require a payment_secret (i.e. modern TLV format onion)
Changelog-Changed: Protocol: We can no longer connect to peers which don't support `payment_secret`.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Useful for regtest and testnet. Sure, you shouldn't use this on mainnet,
but I haven't restricted it because our users are usually pretty clever.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Fixes: #1806
Changelog-Added: config: `force_feerates` option to allow overriding feerate estimates (mainly for regtest).
Incoming HTLCs are rejected by the HTLC logic if the payload contains
an even type that `lightningd` doesn't recognize. This is to prevent
us from accidentally accepting a payment that has extra semantics
attached (for example if we get a keysend payment and don't know what
to do with the TLV field containing the message we should reject it,
otherwise the overall semantics of the message delivery fail).
You still shouldn't do this (you could get some transient failures),
but at least you have a decent chance if you reinstall over a running
daemon, instead of getting confusing internal errors if message
formats have changed.
Changelog-Added: lightningd: we now try to restart if subdaemons are upgraded underneath us.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Fixes: #4346
The main change which affects us is that 2016 blocks to forget a channel
is a fixed number in the spec; we make this clear by renaming the
(developer-only) max_funding_unconfirmed to dev_max_funding_unconfirmed
and making it compile DEVELOPER only.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We were reporting the failure immediately but still continuing with
the startup. This could happen if an important plugin ends up in a
race with another plugin (important or not) for a contended
resource (CLI option or RPC method name). We would eventually notice
that we were supposed to abort, but at that point we already processed
a couple of blocks, loaded the entire state, etc.
This just aborts early with a sane error message.
Changelog-Added: plugin: If there is a misconfiguration with important plugins we now abort early with a more descriptive error message.
Reported-by: PsySc0rpi0n
Reported-by: Ján Sáreník <@jsarenik>
We use waitpid() manually for subdaemons, so we need to step
around that (otherwise we could simply ignore them).
We could destroy subdaemons only once they've exited, but
that works badly with the sd->conn, which will be freed
when error (i.e. close) is detected, so the current code
is probably the best compromise.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
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.
This avoids duplication of both logic and error-prone values, such as
the salt. Grouping all hsm encryption logic into a public API will also
allow us to fuzz it.
Signed-off-by: Antoine Poinsot <darosior@protonmail.com>
`openchannel_signed` commands hang out across the openingd/channeld
boundary -- we don't return until we've successfully broadcast the
transaction (or timed out waiting for them to send a tx_sigs back).
The 100MB log buffer has been the biggest memory footprint for the daemon.
Keeping 10MB for emergency log dumps seems sufficient.
This has been mentioned in the last developer meeting.
Changelog-Changed: In-memory log buffer reduced from 100MB to 10MB
We're going to want this for bolt13 formation as well.
As a result of reworking the logic into "candidate selection" then
"route hint selection", we need to change the way round-robin works.
We use a simple incrementing index now.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
