Jeff Becker
6b2c311800
cover special case for 1 hop path builders
2021-04-19 07:02:44 -04:00
Jeff Becker
2e8f47a7fa
various cleanups
2021-04-19 07:02:44 -04:00
Jeff Becker
b36f87a4d2
stop crashing when we send dns reply for snode on service nodes
2021-04-19 07:02:44 -04:00
Jeff Becker
b01e20b4cb
add forgotten file again
2021-04-19 07:02:44 -04:00
Jeff Becker
2fa24b5eae
add untracked file
2021-04-19 07:02:44 -04:00
Jeff Becker
5eda4addc2
correctly discard invalid data on transfer traffic message
2021-04-19 07:02:43 -04:00
Jeff Becker
25e338d621
srv records
...
* add srv records in RCs if we have any
* add mechanism to add SRV records for plainquic exposed ports
* resign and republish rc or introset on srv record changes
2021-04-19 07:02:43 -04:00
Jeff Becker
a2285730b8
reduce log level
2021-04-19 07:00:27 -04:00
Jeff Becker
de1b0c9d14
add forgotten header
2021-04-19 07:00:27 -04:00
Jeff Becker
94ecf02a62
add option to run lokinet traffic only with no exit and drop all non lokinet traffic
...
triggered by: exit-node=null
2021-04-19 07:00:27 -04:00
Jeff Becker
2ed0ab1ca1
liblokinet and lnproxy updates
...
* add lokinet_status function to get the current status of the context, aka are we ready to send or not.
* catch some exceptions in lnproxy
2021-04-19 07:00:27 -04:00
Jeff Becker
e005b34169
fix up some retarded bullshit
...
* llarp::service::NameIsValid was not checking that the tld was .loki, add this check.
* make link layer initial connection timeout 5s not the session activity timeout which happens to be 60 god damn seconds.
2021-04-19 07:00:26 -04:00
Jeff Becker
0046de3e7a
service node logic change: simplify transit hop behavior to send protocol frames directly from a path transfer message
2021-04-19 06:59:07 -04:00
Jeff Becker
3f9e51e391
more liblokinet api functions
...
* get/set netid
* set log level
2021-04-19 06:59:07 -04:00
Jeff Becker
cf7603f20e
temp commit, closeReset -> close
2021-04-19 06:59:07 -04:00
Jeff Becker
48ce77ed1f
XXX: single threaded mode
2021-04-19 06:59:07 -04:00
Jeff Becker
bac9341ae0
fix: if we have not used a session yet we only check the intro for expiration
2021-04-19 06:59:07 -04:00
Jeff Becker
2188b443e9
more liblokinet api stuff
...
* add lokinet_wait_for_ready to wait until we have published introsets and such
* optionally expose ports in lnproxy
* dont save profiles by default
2021-04-19 06:59:07 -04:00
Jeff Becker
95cd275cdd
liblokinet additions:
...
* add lokinet_add_bootstrap_rc function for adding an rc from memory
* prevent stack overflow on error closing connection in quic
* add in memory nodedb
* refactor how convotags are set as active
* add initial stubs for endpoint statistics
* refactor time stuff to be a bit cleaner
* update lnproxy script with more arguments
2021-04-19 06:59:07 -04:00
Jason Rhinelander
354f0f63e5
Reduce max number of per-connection active streams to 32
...
This is the max number of simultaneous open connections to the same port
on the remote. 100 was fairly arbitrary and seems a bit high.
2021-04-19 06:59:07 -04:00
Jason Rhinelander
5e912600f8
Fix connection close handling
...
Replace stream_reset (which typically isn't called) with a stream_close
handler (which is already called whether or not it was a reset). Most
importantly, the server side needs to extend the max bidi streams
counter during stream_close (otherwise we run out when we hit the
limit and new connections just stall).
2021-04-19 06:59:07 -04:00
Jeff Becker
8bc60a59ac
fix up liblokinet api to be more friendly to ffi
...
add lnproxy exmaple of use of liblokinet
2021-04-19 06:59:06 -04:00
Jeff Becker
853cc52efb
wire up liblokinet
2021-04-19 06:59:06 -04:00
Jason Rhinelander
2a809c7a30
Add trace debugging of inbound packet type
2021-04-19 06:59:06 -04:00
Jason Rhinelander
35ead9dc6b
Remove superfluous block (FIXME: squash this with earlier "FIXME")
2021-04-19 06:59:06 -04:00
Jason Rhinelander
a2045cd966
Fix debug message
2021-04-19 06:59:06 -04:00
Jason Rhinelander
80d3738531
Increase null_iv to 8 bytes to avoid ngtcp2 stack corruption
2021-04-19 06:59:06 -04:00
Jeff Becker
309a899c9f
mark loopback convotags as forever
...
make null handler mark discard data as activity for convotag
2021-04-19 06:59:06 -04:00
Jeff Becker
f310778f81
dont regenerate convotag on loopback when we have one already
2021-04-19 06:59:06 -04:00
Jeff Becker
abf8d2fb17
fix loopback convotag
2021-04-19 06:59:06 -04:00
Jeff Becker
5b05d22bad
refactors
...
* add path sequence numbers on routing messages
* reduce log level in debug mode
* wire up loopback style sending to ourself
2021-04-19 06:59:06 -04:00
Jeff Becker
f86a2daf83
fixes
...
* Add service::Endpoint::HasOutboundConvo
* dont mark outbound convos as inbound
* order quic packets
2021-04-19 06:59:06 -04:00
Jeff Becker
59c9e997f2
build paths faster and limit path builds at edge router instead of via a time counter for all edges
2021-04-19 06:59:05 -04:00
Jason Rhinelander
ff2e79ce38
Remove superfluous log statement
...
The send_packet() we call here already prints the buffer.
2021-04-19 06:59:05 -04:00
Jason Rhinelander
312222e881
Properly handle ngtcp2 returning UINT64_MAX
2021-04-19 06:59:05 -04:00
Jason Rhinelander
2ea84a61f9
Replace timer->again() with stop/start()
...
repeat()/again() sets a repeating timer, which we don't want; stop and
start it instead so that it fires just once.
2021-04-19 06:59:05 -04:00
Jason Rhinelander
68e96a4e5d
Remove obsolete fixme
...
Remote addr and ecn are handled.
2021-04-19 06:59:05 -04:00
Jason Rhinelander
7982581cfd
Revisit/reduce quic logging
...
Demote many things to Trace.
2021-04-19 06:59:05 -04:00
Jason Rhinelander
3c630b260a
Don't install stream forwarding here
...
Stream forward on the client-side TCP connection gets set up within in
initial_client_data_handler (which also handles reading the initial
stream version byte).
2021-04-19 06:59:05 -04:00
Jason Rhinelander
f534103971
Simplify/fix expiry timer logic
...
Also fix a potential transmission delay because `again()` wasn't being
called when the expiry is already passed (i.e. meaning we should run
immediately).
2021-04-19 06:59:05 -04:00
Jason Rhinelander
ac34835c12
Fix/refactor stream closing
...
Make stream closing with expiring connections work better. Fixes an
issue where the stream's uv_async could outlive the stream and/or
connection and segfault.
2021-04-19 06:59:05 -04:00
Jason Rhinelander
60c813d306
Schedule retransmit after blocked send
...
In the standalone plainquic code we triggered a retransmit when the
socket became writeable again, but that doesn't work here, so just
schedule it right away to let ngtcp2 worry about retrying.
2021-04-19 06:59:05 -04:00
Jason Rhinelander
71beae5923
Stop retransmit timer when destroying a connection
...
Otherwise we segfault because it outlives the Connection.
2021-04-19 06:59:05 -04:00
Jason Rhinelander
027243cbcc
Remove obsolete fixme
2021-04-19 06:59:04 -04:00
Jason Rhinelander
233cb86191
Sever tunnel links earlier
...
When we get an error on the tcp connection immediately sever the link to
the quic tunnel so that it doesn't keep trying to forward data to it.
2021-04-19 06:59:04 -04:00
Jeff Becker
73572b317b
stuff
2021-04-19 06:59:04 -04:00
Jason Rhinelander
4cea33a139
Fix Server's missing opt-out of ring buffer
...
For now we still steal buffers from uvw.
In the future I'd like to change that, but it's still uvw
work-in-progress to support custom data allocators, and so for now we
still steal data buffers from uvw.
2021-04-19 06:59:04 -04:00
Jason Rhinelander
99954f7501
Add some more useful debug logging
2021-04-19 06:59:04 -04:00
Jason Rhinelander
b8be889291
Add missing client TCP accept/forwarding handlers
...
Somehow the TCP client connection accept and stream forwarding got
dropped in the quic refactor.
2021-04-19 06:59:04 -04:00
Jason Rhinelander
738f16366b
Try outbound convos if no inbound convos work (FIXME)
...
FIXME - eliminate this block of code that used to be an `else` before
PR'ing.
2021-04-19 06:59:04 -04:00
