dev_blackhole_fd was a hack, and doesn't work well now we are async
(it worked for sync comms in per-peer daemons, but now we could sneak
through a read before we get to the next write).
So, make explicit flags and use them. This is much easier now we
have all peer comms in one place.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
As connectd handles more packets itself, or diverts them to/from gossipd,
it's the only place we can implement the dev_disconnect logic.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
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>
It handles all the cases of retransmission, and in the normal case
retransmits shutdown and immediately returns for us to run closingd.
This is actually far simpler and reduces code duplication.
[ Includes fixup to stop warn_unused_result from Christian ]
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Fixed: Protocol: We could get stuck on signature exchange if we needed to retransmit the final revoke_and_ack.
This allows us to ensure a packet is read by the other end, but we
don't read anything else from them or write anything to them.
Using '+' is similar, but because it closes the connection, the peer
might notice before receiving the packet (such as if it does a write).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We should actually be including this (as it may define _GNU_SOURCE
etc) before any system headers. But where we include <assert.h> we
often didn't, because check-includes would complain that the headers
included it too.
Weaken that check, and include config.h in C files before assert.h.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This avoids overwriting the ones in git, and generally makes things neater.
We have convenience headers wire/peer_wire.h and wire/onion_wire.h to
avoid most #ifdefs: simply include those.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
I have seen some strange flakiness (under VALGRIND), which I have
traced down to dev-disconnect "+" not working as expected. In
particular, the message is not sent out before closing the fd.
This seems to fix it on Linux, though it's so intermittant that it's hard
to be completely sure.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
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>
We currently assume the daemon gives up; gossipd won't, and we want to
test it there too.
This reveals a bug (returning io_close() is bad if the call is to
duplex()), and breaks a test which now continues after dropping a
packet..
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Useful if we want to drop & suppress, for example. We change '=' to mean
do nothing to the packet.
We use this to clean up the test_reconnect_sender_add test.
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>
