Commit Graph

6824 Commits

Author SHA1 Message Date
Jason Rhinelander
e5aa96ed7c Upgrade to clang-format 11
The formatting changes here are not too large and also nice: in
particular clang-format-11 appears to do better handling of `{...}`
constructor braces/arguments.
2021-03-05 13:36:31 -04:00
Jeff
76ee6734db
Merge pull request #1557 from jagerman/uvw-overhaul
libuv -> uvw overhaul
2021-03-04 17:12:10 -05:00
Jeff
57a5f8cb15
Merge pull request #1567 from jagerman/loki-core-docker-deps
Add oxen-core deps to docker images
2021-03-04 17:04:26 -05:00
Jeff Becker
27685b1b6e
fix docker lines and dont install recommended packages 2021-03-04 17:03:02 -05:00
Jason Rhinelander
78a43e7163 Add oxen-core deps to docker images 2021-03-04 17:21:45 -04:00
Jason Rhinelander
c00de1e4ec Add wakeup call into transit_hop UpstreamHandler 2021-03-04 16:51:18 -04:00
Jason Rhinelander
89b582bf60 Modernize [[maybe_unused]] attribute 2021-03-04 16:51:18 -04:00
Jason Rhinelander
4b5387a078 Remove unused ShouldFlushNow method 2021-03-04 16:51:18 -04:00
Jason Rhinelander
f9dc308f75 Tick event loop on data queuing 2021-03-04 16:51:18 -04:00
Jason Rhinelander
895acc45ff EventLoop: add public wakeup() method, and call it from call()
call(), when invoked from the logic thread, wasn't triggering a wakeup
which stalled some traffic (such as client-to-snode packets).

Fix it by triggering a wakeup on `call()`, and expose it because this is
a useful thing to be able to do.
2021-03-04 16:51:18 -04:00
Jason Rhinelander
f8d5e106a0 Uncurse SockAddr's sockaddr* operator
The `const sockaddr*` conversion was returning a pointer to the
sockaddr_in6, but with its family set to AF_INET, so the receiving code
would have treated interpreting the result as a sockaddr_in pointer --
but it isn't, and trying to get the IPv4 from the mis-reinterpreted
sockaddr_in6 fails; this then prevented uvw UDP binding (which uses the
sockaddr* operator) from working.

This changes it to always leave the family value at the proper value and
using the ipv6_is_mapped_ipv4 check to determine when we have (and
should return) a pointer to the sockaddr_in rather than the
sockaddr_in6.

Also fixes toString() to compare the full prefix instead of just bytes
10 and 11.
2021-03-04 16:51:18 -04:00
Jason Rhinelander
62fc1bbc48 Make llarp_buffer_t constructible from an rvalue reference 2021-03-04 16:51:18 -04:00
Jason Rhinelander
f8486604c4 Export UDP file descriptor (for android) 2021-03-04 16:51:18 -04:00
Jason Rhinelander
d995766436 Android & hive fixes 2021-03-04 16:51:18 -04:00
Jason Rhinelander
673d065ed8 Remove HasNextPacket() (and fix win32)
We don't use it anymore (instead we just ReadNextPacket until it returns
an empty result).
2021-03-04 16:51:18 -04:00
Jason Rhinelander
6e05dc693f Remove unused arguments (and ants) 2021-03-04 16:51:18 -04:00
Jason Rhinelander
2c827870c9 Simplify code of many f's
We don't need to go through the contention killer here, that was mainly
for (long ago) debugging reasons.
2021-03-04 16:51:18 -04:00
Jason Rhinelander
36edabd9b0 Rename EventLoop::udp to EventLoop::make_udp
Makes it consistent with the make_waker, etc.

Also removes some debugging.
2021-03-04 16:51:18 -04:00
Jason Rhinelander
afd856bca0 Remove unused junk (also fix win32 compilation fail) 2021-03-04 16:51:18 -04:00
Jason Rhinelander
ced035d401 Switch uvw to a fork with libuv backwards compat support 2021-03-04 16:51:18 -04:00
Jason Rhinelander
096f37fe4e Fix linux compilation & narrowing conversion warning 2021-03-04 16:51:18 -04:00
Jason Rhinelander
e198bfd3e1 Remove unused variable 2021-03-04 16:51:18 -04:00
Jason Rhinelander
c4559d158e Make format 2021-03-04 16:51:18 -04:00
Jason Rhinelander
ccc7b5c9e9 Merge Logic functions into EventLoop
loop->call(...) is similar to the old logic->Call(...), but is smart
about the current thread: if called from within the event loop it simply
runs the argument directly, otherwise it queues it.

Similarly most of the other event loop calls are also now thread-aware:
for example, `call_later(...)` can queue the job directly when called if
in the event loop rather than having to double-queue through the even
loop (once to call, then inside the call to initiate the time).
2021-03-04 16:51:18 -04:00
Jason Rhinelander
5b555ee5aa Replace libuv with uvw & related refactoring
- removes all the llarp_ev_* functions, replacing with methods/classes/functions in the llarp
  namespace.
- banish ev/ev.h to the void
- Passes various things by const lvalue ref, especially shared_ptr's that don't need to be copied
  (to avoid an atomic refcount increment/decrement).
- Add a llarp::UDPHandle abstract class for UDP handling
- Removes the UDP tick handler; code that needs tick can just do a separate handler on the event
  loop outside the UDP socket.
- Adds an "OwnedBuffer" which owns its own memory but is implicitly convertible to a llarp_buffer_t.
  This is mostly needed to take over ownership of buffers from uvw without copying them as,
  currently, uvw does its own allocation (pending some open upstream issues/PRs).
- Logic:
  - add `make_caller`/`call_forever`/`call_every` utility functions to abstract Call wrapping and
    dependent timed tasks.
  - Add inLogicThread() so that code can tell its inside the logic thread (typically for
    debugging assertions).
  - get rid of janky integer returns and dealing with cancellations on call_later: the other methods
    added here and the event loop code remove the need for them.
- Event loop:
  - redo everything with uvw instead of libuv
  - rename EventLoopWakeup::Wakeup to EventLoopWakeup::Trigger to better reflect what it does.
  - add EventLoopRepeater for repeated events, and replace the code that reschedules itself every
    time it is called with a repeater.
  - Split up `EventLoop::run()` into a non-virtual base method and abstract `run_loop()` methods;
    the base method does a couple extra setup/teardown things that don't need to be in the derived class.
  - udp_listen is replaced with ev->udp(...) which returns a new UDPHandle object rather that
    needing gross C-style-but-not-actually-C-compatible structs.
  - Remove unused register_poll_fd_(un)readable
  - Use shared_ptr for EventLoopWakeup rather than returning a raw pointer; uvw lets us not have to
    worry about having the event loop class maintain ownership of it.
  - Add factory EventLoop::create() function to create a default (uvw-based) event loop (previously
    this was one of the llarp_ev_blahblah unnamespaced functions).
  - ev_libuv: this is mostly rewritten; all of the glue code/structs, in particular, are gone as
    they are no longer needed with uvw.
- DNS:
  - Rename DnsHandler to DnsInterceptor to better describe what it does (this is the code that
    intercepts all DNS to the tun IP range for Android).
- endpoint:
  - remove unused "isolated network" code
  - remove distinct (but actually always the same) variables for router/endpoint logic objects
- llarp_buffer_t
  - make constructors type-safe against being called with points to non-size-1 values
- tun packet reading:
  - read all available packets off the device/file descriptor; previously we were reading one packet
    at a time then returning to the event loop to poll again.
  - ReadNextPacket() now returns a 0-size packet if the read would block (so that we can implement
    the previous point).
  - ReadNextPacket() now throws on I/O error
- Miscellaneous code cleanups/simplifications
2021-03-04 16:51:18 -04:00
Jason Rhinelander
c71d3527bd C++17 simplification 2021-03-04 16:51:18 -04:00
Jason Rhinelander
8aa1b702cd Disallow evil compilation
-fpermissive allows nasty things that are (apparently) needed on
Windows, so only enable it there.

vla are evil, so don't just warn but error on it.

remove duplicate -Wno-unknown-pragmas
2021-03-04 16:51:18 -04:00
Jason Rhinelander
ffe9bbc984 Tweak clang-tidy rules
Turn off modernize-use-trailing-return because it's *really* annoying
(and clangd picks it up when using it for editing).

Turn off modernize-use-nodiscard as well because it's already annoying.

Add bugprone-* as well because it includes a lot of useful checks.
2021-03-04 16:51:18 -04:00
Jason Rhinelander
dfb62d06ce Shut up noisy catch tests 2021-03-04 16:51:18 -04:00
Jason Rhinelander
8b00eacabf Fix signed/unsigned char comparison bug
These functions would not working properly if given a high bit char
value.
2021-03-04 16:51:18 -04:00
Jason Rhinelander
00df8b85da Bump required oxenmq version
Current lokinet dev relies on something from oxenmq dev branch which is
in the not-yet-released next version; without this lokinet tries using
the system 1.2.3 install, which then fails the build.
2021-03-04 16:51:18 -04:00
Jason Rhinelander
e3aff6c36c Add more junk & clangd symlinks to .gitignore
Ignore any root directory starting with build so that you can have many
build directories (build, build2, build-win32, ...) that get ignored.

Add .cache which clangd creates

Ignore compile_commands.json (and generate it by default) which helps
clangd, but needs to be symlinked to a compile_commands.json in a build
dir so isn't suitable for inclusion in the repo.
2021-03-04 16:51:18 -04:00
Jeff
db515e48ec
Merge pull request #1561 from majestrate/catch2_migrate-2021-03-03
totally remove gtest/gmock from unit tests.
2021-03-04 15:50:36 -05:00
Jeff
abf057d6e3
Merge pull request #1536 from majestrate/memleak-fix-2021-02-15
plug super slow memory leak
2021-03-04 15:50:11 -05:00
Jeff
72c9b0f1b4
Merge pull request #1535 from majestrate/no-iwp-multiack-2021-02-13
only do single ack packets
2021-03-04 15:49:12 -05:00
Jeff Becker
4cf0f1ee35
decaying hashset unit test fixes 2021-03-04 15:47:09 -05:00
Jeff Becker
5fd15fa044
generate keys in unit test 2021-03-04 15:46:12 -05:00
Jeff Becker
6c8e5256d6
catchAll -> testAll 2021-03-04 15:45:52 -05:00
Jeff Becker
60bfdacc23
get rid of the rest of gtest/gmock 2021-03-04 15:41:53 -05:00
lyyn
ece91e87fc
Migrate tests from gtest to catch2 2021-03-04 15:41:42 -05:00
Jeff Becker
ec0f45eeb7
plug memory leak in outbound message queue.
sometimes we use rxid in our pathid so we need to clear those queues too.
if we don't it'll leak from never clearing those queues.
2021-03-04 15:29:09 -05:00
Jeff Becker
7dbd25f271
don't leave old multi ack codepath in 2021-03-04 15:25:11 -05:00
Jeff Becker
8a4417cb1a
reduce test message count so it can pass 2021-03-04 15:21:51 -05:00
Jeff Becker
c9ff917e0d
revert priority queue ordering change 2021-03-04 15:21:51 -05:00
Jeff Becker
ad3c23ba2b
simplify call 2021-03-04 15:21:51 -05:00
Jeff Becker
5cffc3b0f8
consolidate rx message handling in iwp
* add toggle flag for using multi ack
* check replay filter before processing message
2021-03-04 15:21:51 -05:00
Jeff Becker
e6ac7e721d
use a std::map instead of a std::unordered_map for holding message buffers by their sequentially increasing message id.
when retransmissions happen, they will be executed in a deterministic order by their delivery id.
this reduces the jitter from delayed message delivery on link layer as with an unordered map the order of iteration is "random",
so when we iterate all messages for retransmission we do it lowest id first (the messages queued first).
2021-03-04 15:21:51 -05:00
Jeff Becker
74f707ee01
only do single ack packets 2021-03-04 15:21:51 -05:00
Jeff
4992629f20
Merge pull request #1563 from majestrate/dont-multiplex-from-outbound-context-2021-03-03
use lowest latency path for send context
2021-03-04 15:16:40 -05:00
Jeff
d5baeaa425
Merge pull request #1564 from majestrate/exit-backwards-comapt-issue-2021-03-03
unbreak ipv4 exits
2021-03-04 14:30:53 -05:00