Ian Gulliver
|
3235211bc2
|
More sensible cleanup ordering.
|
2016-02-25 16:48:51 -08:00 |
|
Ian Gulliver
|
3155381678
|
Shut down if we run out of inputs or outputs.
|
2016-02-25 16:33:58 -08:00 |
|
Ian Gulliver
|
a9dc7c25db
|
Switch back to write(), since we have non-sockets.
|
2016-02-25 16:33:47 -08:00 |
|
Ian Gulliver
|
5e604f0596
|
In/out reference counting.
|
2016-02-25 16:17:25 -08:00 |
|
Ian Gulliver
|
aa940137c1
|
Move proto into its own dir.
|
2016-02-25 15:22:57 -08:00 |
|
Ian Gulliver
|
3816be8b3d
|
Fix leak of second+ receive/send object.
|
2016-02-25 15:17:45 -08:00 |
|
Ian Gulliver
|
4f14845c75
|
Rename --dump to --stdout. Add --stdin.
|
2016-02-25 15:12:16 -08:00 |
|
Ian Gulliver
|
46fd0c3c8b
|
Tear down resolver before others.
|
2016-02-25 11:57:23 -08:00 |
|
Ian Gulliver
|
6f082ecf32
|
Use send(MSG_NOSIGNAL) instead of write() in one other place.
|
2016-02-25 11:56:51 -08:00 |
|
Ian Gulliver
|
b4171c971a
|
Avoid double call of send_del() on write failure and from epoll loop.
|
2016-02-24 21:30:39 -08:00 |
|
Ian Gulliver
|
a408ddba2d
|
Switch from shutdown(SHUT_RD) to EPOLLIN to detect send <-> send misconfigurations.
|
2016-02-24 21:21:14 -08:00 |
|
Ian Gulliver
|
4a83b453bb
|
Ask for SIGPIPE not to be sent, instead of ignoring it.
|
2016-02-24 20:28:48 -08:00 |
|
Ian Gulliver
|
3e2c426e95
|
Fix --connect-send. Half-close connections so we can detect two read sockets talking to each other.
|
2016-02-24 20:25:31 -08:00 |
|
Ian Gulliver
|
1635534243
|
Switch from sketchy dup() juggling to callbacks to retry outgoing connection.
|
2016-02-24 20:15:09 -08:00 |
|
Ian Gulliver
|
63f5313b96
|
Set EPOLL_CLOEXEC
|
2016-02-23 22:24:03 -08:00 |
|
Ian Gulliver
|
025d14c6a6
|
Add framing to proto, since it isn't built in to protobuf.
|
2016-02-23 21:30:29 -08:00 |
|
Ian Gulliver
|
c70886e55c
|
Proto parsing working.
|
2016-02-23 19:46:42 -08:00 |
|
Ian Gulliver
|
4f48153d4f
|
More proto parsing.
|
2016-02-23 16:43:36 -08:00 |
|
Ian Gulliver
|
a33808cb33
|
Fix stats counting overflow.
|
2016-02-23 16:30:30 -08:00 |
|
Ian Gulliver
|
819518e402
|
Proto serialization support.
|
2016-02-23 16:13:11 -08:00 |
|
Ian Gulliver
|
a98eeec91e
|
Proto protocol structure.
|
2016-02-23 15:19:56 -08:00 |
|
Ian Gulliver
|
606a2029ee
|
Clean up json parsing more, add a magic value just in case.
|
2016-02-23 14:27:54 -08:00 |
|
Ian Gulliver
|
6d7c236cb9
|
Clean up json parsing.
|
2016-02-23 13:56:54 -08:00 |
|
Ian Gulliver
|
d2e521e90c
|
json input support, some shutdown fixes. json needs cleanup.
|
2016-02-23 13:49:23 -08:00 |
|
Ian Gulliver
|
07d273236d
|
Fix endless loop inside incoming handler.
|
2016-02-23 12:31:48 -08:00 |
|
Ian Gulliver
|
49edb43232
|
Close stdout and stderr before exit, to make valgrind output easier reading.
|
2016-02-23 12:11:07 -08:00 |
|
Ian Gulliver
|
ab6b8cd5e9
|
Track and clean up receive objects.
|
2016-02-23 12:08:53 -08:00 |
|
Ian Gulliver
|
77ee1422c1
|
Sprinkle CLOEXEC around.
|
2016-02-23 12:00:17 -08:00 |
|
Ian Gulliver
|
f676cc531a
|
Fix fd leak.
|
2016-02-23 11:57:40 -08:00 |
|
Ian Gulliver
|
289dcdbb54
|
Change resolution to use a single resolve thread, and to clean up after itself.
|
2016-02-23 11:53:17 -08:00 |
|
Ian Gulliver
|
c6a2150a7d
|
Initial pass at async resolving. This is ugly and leaks memory and I hate it, so dedicated thread incoming.
|
2016-02-23 10:46:40 -08:00 |
|
Ian Gulliver
|
ebdb3f9dc0
|
Pass source ID through to JSON output.
|
2016-02-22 22:00:49 -08:00 |
|
Ian Gulliver
|
bccdc212f2
|
More info in JSON header.
|
2016-02-22 21:55:45 -08:00 |
|
Ian Gulliver
|
bc9219b4a8
|
Split server out.
|
2016-02-22 21:53:25 -08:00 |
|
Ian Gulliver
|
c6e1eb9d42
|
Switch stats to use CLOCK_MONOTONIC_COARSE
|
2016-02-22 16:59:42 -08:00 |
|
Ian Gulliver
|
2ce2ed5bf5
|
Split out packet. Delete common.
|
2016-02-22 16:58:13 -08:00 |
|
Ian Gulliver
|
b5ab0cb0a6
|
Split out peer
|
2016-02-22 16:49:43 -08:00 |
|
Ian Gulliver
|
b5219d9c1a
|
Split out buf
|
2016-02-22 16:41:34 -08:00 |
|
Ian Gulliver
|
f36642b823
|
Split out hex
|
2016-02-22 16:36:27 -08:00 |
|
Ian Gulliver
|
21536eee33
|
Move retry delay calculation into wakeup, since it's only used there.
|
2016-02-22 16:33:00 -08:00 |
|
Ian Gulliver
|
d6a629fa7e
|
Switch to an internal implementation of uuid generation, so we can use the randomness pool.
|
2016-02-22 16:27:44 -08:00 |
|
Ian Gulliver
|
d03590439f
|
Pool randomness.
|
2016-02-22 14:57:27 -08:00 |
|
Ian Gulliver
|
16685b8d05
|
Split out rand
|
2016-02-22 14:45:18 -08:00 |
|
Ian Gulliver
|
f255170967
|
Move wakeup into the main thread to remove complexity. Looks like CLOCK_MONOTONIC_COARSE doesn't need a syscall.
|
2016-02-22 14:37:00 -08:00 |
|
Ian Gulliver
|
70cfaa94aa
|
Human time for max backoff
|
2016-02-21 16:57:42 -08:00 |
|
Ian Gulliver
|
a48c4a24f5
|
Avoid a spinloop waiting for zero ms.
|
2016-02-21 16:55:31 -08:00 |
|
Ian Gulliver
|
c5a87e89a4
|
Restore airspy_adsb
|
2016-02-21 16:52:36 -08:00 |
|
Ian Gulliver
|
8f36ad0db8
|
Wakeup/delay fixes, and actually add retries for incoming and outgoing.
|
2016-02-21 16:47:27 -08:00 |
|
Ian Gulliver
|
835a144d6e
|
Genericize the rand code.
|
2016-02-21 15:56:07 -08:00 |
|
Ian Gulliver
|
6384e0f20d
|
Wakeup actually working (probably), add delay calculator.
|
2016-02-21 15:51:38 -08:00 |
|