Age | Commit message (Collapse) | Author | |
---|---|---|---|
2018-02-04 | Align with go library layout | Mathias Hall-Andersen | |
2018-02-02 | Fixed tests | Mathias Hall-Andersen | |
2018-02-02 | Clear cryptographic state when interface down | Mathias Hall-Andersen | |
Attempts to clear the cryptographic state for every peer when the device goes down. | |||
2018-02-02 | Rework of entire locking system | Mathias Hall-Andersen | |
Locking on the Device instance is now much more fined-grained, seperating out the fields into "resources" st. most common interactions only require a small number. | |||
2018-01-26 | Added initial version of peer teardown | Mathias Hall-Andersen | |
There is a double lock issue with device.Close which has yet to be resolved. | |||
2018-01-13 | Work on timer teardown + bug fixes | Mathias Hall-Andersen | |
Added waitgroups to peer struct for routine start / stop synchronisation | |||
2017-12-29 | Peer timer teardown | Mathias Hall-Andersen | |
2017-11-30 | Refactor timers.go | Mathias Hall-Andersen | |
2017-11-18 | Moved endpoint into interface and simplified peer | Mathias Hall-Andersen | |
2017-10-27 | Fixed message header length in conn_linux | Mathias Hall-Andersen | |
2017-10-16 | Initial implementation of source caching | Mathias Hall-Andersen | |
Yet untested. | |||
2017-10-08 | Added new UDPBind interface | Mathias Hall-Andersen | |
2017-09-20 | Added last_minute_handshake_guard | Mathias Hall-Andersen | |
- Added last_minute_handshake_guard and reverted keypair changes. - Added comment explaining the state of Go in releation to handling cryptographic state in memory. - Decreased logging level of netsh test | |||
2017-08-14 | Improved cookie/mac computation code | Mathias Hall-Andersen | |
2017-08-07 | Number of fixes in response to code review | Mathias Hall-Andersen | |
This version cannot complete a handshake. The program will panic upon receiving any message on the UDP socket. | |||
2017-07-27 | Improved timer code | Mathias Hall-Andersen | |
2017-07-18 | Added last_handshake_time fields to UAPI | Mathias Hall-Andersen | |
2017-07-13 | Terminate on interface deletion | Mathias Hall-Andersen | |
Program now terminates when the interface is removed Increases the number of os threads (relevant for Go <1.5, not tested) More consistent commenting Improved logging (additional peer information) | |||
2017-07-08 | Improved timer state machine | Mathias Hall-Andersen | |
2017-07-08 | Added source verification | Mathias Hall-Andersen | |
2017-07-06 | Fixed broken test | Mathias Hall-Andersen | |
2017-07-06 | Initial working full exchange | Mathias Hall-Andersen | |
The implementation is now capable of connecting to another wireguard instance, complete a handshake and exchange transport messages. | |||
2017-07-01 | Handshake negotiation functioning | Mathias Hall-Andersen | |
2017-06-30 | Completed initial version of outbound flow | Mathias Hall-Andersen | |
2017-06-28 | Work on UAPI | Mathias Hall-Andersen | |
Cross-platform API (get operation) Handshake initiation creation process Outbound packet flow Fixes from code-review | |||
2017-06-27 | Implemented MAC1/2 calculation | Mathias Hall-Andersen | |
2017-06-26 | Begin implementation of outbound work queue | Mathias Hall-Andersen | |
2017-06-26 | Begin work on outbound packet flow | Mathias Hall-Andersen | |
2017-06-24 | Restructuring of noise impl. | Mathias Hall-Andersen | |
2017-06-04 | Beginning work on TUN interface | Mathias Hall-Andersen | |
And outbound routing I am not entirely convinced the use of net.IP is a good idea, since the internal representation of net.IP is a byte slice and all constructor functions in "net" return 16 byte slices (padded for IPv4), while the use in this project uses 4 byte slices. Which may be confusing. | |||
2017-06-01 | Inital implementation of trie | Mathias Hall-Andersen | |
2017-05-30 | Beginning work on UAPI and routing table | Mathias Hall-Andersen | |