Commit Graph

26 Commits

Author SHA1 Message Date
Rick V
b4d6f89452
try extracting dns bind addr from INI on windows 2020-03-05 12:47:45 -06:00
Rick V
244af9891b
fix a crash on some win32 systems
make clang+libc2 happy
2019-08-02 03:24:11 -05:00
Rick V
d4688ed3b3
get our TAP ifindex to pass to ipv6(1) on old platforms 2019-08-02 03:23:50 -05:00
Rick V
18be2a8562
seek only the generic TAP device
(On my system, i have a commercial VPN installed that uses its own fork of TAP-Win32 as its tunnelling driver)
2019-03-20 21:43:38 -05:00
Rick V
975b4ca743
clang-format 2019-02-26 15:48:03 -06:00
Rick V
0de253065e
remove ded code
add code for proper DNSc search
2019-02-26 15:35:04 -06:00
despair
e9c8cb9977
finally got around to writing a proper event loop for windows
just to be able to read user packets from TAP

split the UDP and TUN listeners into separate event queues

added some notes in tuntap-windows (mostly gutting it since we do a lot of the stuff ourselves)
2018-12-22 12:58:13 -06:00
despair
d1cbd58a3f
back to square one
need to be able to queue up a TUN read/write at each tick
then finish dealing with it in the main event loop
which is damn near impossible to do along with unix-style asio
that waits for data to appear/disappear before signalling
2018-12-22 12:58:05 -06:00
despair
551bf0d2bf fix TUN read pkt 2018-11-26 18:49:15 -06:00
despair
65ff1ba991 reading from TUN using native async io:
writing: 
2018-11-26 11:10:18 -06:00
despair
09713adc9f wooo new warnings in clang-7 2018-11-22 20:13:31 -06:00
despair
ae616d0fa5 minor fixes
eh

why apple why

why apple why
2018-11-19 04:29:39 -06:00
despair
be7ac352ca switch to a simulation of epoll
the native async event system on windows
is _not_ very good at getting external events
(i.e. we receive data, but we don't get any indication
that this ever happened)
2018-11-19 04:29:39 -06:00
despair
e18f786ab2 refactor event loop stuff
find some place in the C code to place the worker thread procedure
until such time that michael presents the new thread pool class

fix unix

get a new event port each time and delet in the event loop after use
2018-11-19 04:29:39 -06:00
despair
2a5a0f1577 fix warnings on windows
up next: fix netbsd
2018-11-07 18:09:49 -06:00
despair
561c5228d3 fixed race condition in event loop....dns seems to work now :DDD 2018-10-30 07:30:21 -05:00
despair
5a32ccf220 scan-build fixes on win32 code paths 2018-10-30 00:16:34 -05:00
despair86
9d546cbac4 separate read and write event ports to avoid overwriting queue events 2018-10-16 12:18:29 -05:00
despair86
3c7b7f2176 overlapped io pending is NOT an error 2018-10-06 12:04:52 -05:00
despair86
94c5f46d33 try fixing win32 event loop 2018-10-06 12:04:50 -05:00
despair86
f17727d50a if you are able to keep a windows exit node up for more than a year call me 2018-10-06 12:04:49 -05:00
despair86
363900fce9 set windows TUN to point to internal lokinet resolver 2018-10-06 12:04:49 -05:00
despair86
ca0d09142a Finally put together some skeletal TUN code for NT
up next: debugging the windows client code
stretch goal: prototype hosting a full masternode on Windows Server (still _highly_ experimental when it _does_ appear)
2018-10-06 12:04:48 -05:00
despair86
faf6022288 rip ms c++ 2018-10-01 15:08:55 -05:00
despair86
8ac7d4f6dc if we didn't specify a path to save our config in, only create .lokinet on the assumption that $HOME or $APPDATA (on NT) already exist
add win32 tun glue, fix llarp timebase
(In fact, _both_ of these are guaranteed to exist on their respective platforms.)
also, tuntap is now wired up to the windows port
2018-10-01 15:08:52 -05:00
Ryan Tharp
e41ad35c9f include libtuntap 2018-06-23 05:18:31 -07:00