a8d7a31327
Integrate arp into NetconEthernetTap.
11 years ago
c1a53a2653
ARP cache and responder agent code for use in netcon and iOS.
11 years ago
0d386f1c31
Add a bit of useful testing instrumentation to SqliteNetworkController.
11 years ago
85b90f122a
Final std::map<> from Switch, and add some smallish default values for hash size.
11 years ago
3dba016a93
Almost done... very few std::map<>s remaining in any spot that matters.
11 years ago
db0369e9b8
Remove way-overkill multimap from Switch.
11 years ago
0ab3e49be9
Starting in on Switch... kill map in defrag queue, which will probably improve performance pretty decently under high load with lots of peers.
11 years ago
f116c4b9c0
... and another ...
11 years ago
307e44f7c8
Two for one! (std::map removal)
11 years ago
d1341578d8
... and another one!
11 years ago
7b8ce16057
Another std::map<> dies.
11 years ago
3a959a7763
Swap out std::map<> for Hashtable<> for main peer database in Topology. (ongoing std::map-ectomy)
11 years ago
cfd101c9b8
Add entries() to go with keys() for future use.
11 years ago
da9a720c3f
Hash table bug fix, and add copy constructor and assignment operator for principle of least surprise.
11 years ago
b11ffc9635
Integrate Hashtable into Multicaster, where @mwarning found heaviest std::map() overhead.
11 years ago
3947807b1f
A simple and fast Hashtable, tested but not yet integrated with anything.
11 years ago
5fcef91278
Ignore old root-topology if present -- fix for upgrading on Linux and possibly elsewhere.
11 years ago
facb009a1d
Add security notice to auto-update info in -h output, and fix a missing paren.
11 years ago
8d09c37140
Remove a bit of redundant logic, and also announce MULTICAST_LIKEs to controllers (for future use).
11 years ago
f6ced547be
Dead code removal.
11 years ago
eea8d58afa
docs,cleanup
11 years ago
21e6850722
Cancel NAT-t attempts if peer is no longer "alive"
11 years ago
5986d83738
Kill more kittens.
11 years ago
4564dd95ff
Revert... no luck with any of that.
11 years ago
d2bfdfa6e7
Play with NAT-t tweaks some more.
11 years ago
b69afa010e
Disable type punning on ARM by ifdef.
11 years ago
708aac1ea7
Remove some left over debug code, and fix attempt to send to self if we are an active bridge.
11 years ago
17bfd4d55e
Add TRACE for NAT-t debugging.
11 years ago
b31071463c
Try another NAT traversal improvement.
11 years ago
dda376c9eb
Nuke some abandoned code.
11 years ago
821f1f366e
Fix to NAT escalation sequence.
11 years ago
e99eda4a4a
Fix IP scoping bug, and disable remotely reported surface push... not helping. :(
11 years ago
fadb291962
Fix infinite loop typo.
11 years ago
f0003ea922
Push remote surface as reported by peers along with known interface direct paths to assist with (some) NAT traversal. (trying this, may back out if not effective)
11 years ago
e30ba3e138
Eliminate some aggressive port scanning NAT-t behavior that has proven ineffective.
11 years ago
7a15d8a7e3
Fix leaving of networks to actually call Network::destroy().
11 years ago
d647a587a1
(1) Fix updating of network revision counter on member change.
...
(2) Go back to timestamp as certificate revision number. This is simpler
and more robust than using the network revision number for this and
forcing network revision fast-forward, which could cause some peers
to fall off the horizon when you don't want them to.
11 years ago
b3516c599b
Add a rate limiting circuit breaker to the network controller to prevent flooding attacks and race conditions.
11 years ago
3ba54c7e35
Eliminate some poorly thought out optimizations from the netconf/controller interaction,
...
and go ahead and bump version to 1.0.4.
For a while in 1.0.3 -dev I was trying to optimize out repeated network controller
requests by using a ratcheting mechanism. If the client received a network config
that was indeed different from the one it had, it would respond by instantlly
requesting it again.
Not sure what I was thinking. It's fundamentally unsafe to respond to a message
with another message of the same type -- it risks a race condition. In this case
that's exactly what could happen.
It just isn't worth the added complexity to avoid a tiny, tiny amount of network
overhead, so I've taken this whole path out.
A few extra bytes every two minutes isn't worth fretting about, but as I recall
the reason for this optimization was to save CPU on the controller. This can be
achieved by just caching responses in memory *there* and serving those same
responses back out if they haven't changed.
I think I developed that 'ratcheting' stuff before I went full time on this. It's
hard to develop stuff like this without hours of sustained focus.
11 years ago
8c18c60c00
sp.
11 years ago
88949a750f
Workaround for uclibc missing map::operator==()
11 years ago
547b1c6157
Add additional TRACE output in pushDirectPaths.
11 years ago
0b354803f3
Clean up some YAGNI issues with implementation of GitHub issue #180 , and make best path choice aware of path rank.
11 years ago
4bf3bcbd55
Fixes to PUSH_DIRECT_PATHS.
11 years ago
3f0eca72f7
ZT_TRACE build fix.
11 years ago
fe20f0d7cd
Put back legacy code to listen for LAN announcements to support same network location with pre-1.0.4 clients.
11 years ago
a297e4a5bf
Add build def ZT_NO_TYPE_PUNNING, which when defined disables type punning code that might cause unaligned access errors on architectures that care (e.g. Android/ARM)
11 years ago
412389ec75
Implement ERROR_UNWATNED_MULTICAST
11 years ago
cf6f30963c
Kill a potential source of type punning BUS errors on Android, and besides that hack probably did not improve performance at all given the short lenghts of things compared with secureEq()
11 years ago
3f567a07ca
Save a little bit of RAM by getting rid of overkill CMWC4096 non-crypto PRNG and replacing it with a simple non-crypto PRNG that just uses Salsa20.
11 years ago
Adam Ierymenko