0c6c569b69
(trunk libT) minor cleanup: improve the function names torrentConstructor() and torrentDestructor().
...
These functions actually allocate/free the memory where these objects reside, so it's clearer to name them torrentNew() and torrentDelete().
2011-02-23 06:18:09 +00:00
bf8e4594e0
(trunk libT) minor cleanup: improve the function names peerConstructor() and peerDestructor().
...
These functions actually allocate/free the memory where these peer objects reside, so it's clearer to name them peerNew() and peerDelete().
2011-02-23 06:14:06 +00:00
f2b4a59eda
(trunk libT) #4051 "Use composition for the tr_history fields in tr_peer" -- fixed.
...
If we use composition on these objects we can save a handful of pointers per peer. This isn't a big deal, but it's an easy/safe change to do.
2011-02-23 06:01:16 +00:00
2b9db3c242
(trunk libT) #4048 "use bitsets instead of bitfield in tr_completion" -- done.
...
Excuse the sprawl. Much of this didn't fit into self-contained commits.
2011-02-23 03:54:04 +00:00
094b726316
(trunk libT) remove unused code: tr_bitfieldIsEmpty()
2011-02-21 15:43:41 +00:00
32237dbdb0
(trunk libT) remove unused functions: tr_bitsetDifference() tr_bitfieldDifference()
2011-02-21 15:36:07 +00:00
6ab918a202
(trunk libT) code cleanup: moving bitset functions to their own .c file.
2011-02-21 01:40:19 +00:00
25ad6bbad2
(trunk libT) remove dead code in peer-common.h: the tr_addreq_t enum
2011-02-21 01:15:13 +00:00
1ccdb1a797
(trunk libT) web.c: free the unprocessed tasks before exiting the libcurl thread.
...
This is rare but can happen during shutdown if there are unresponsive trackers. Cleaning up the tasks improves the S/N ratio in valgrind a bit.
2011-02-21 01:13:28 +00:00
e77a627d3c
(trunk libT) mark the two "progress" args to tr_torrentSetLocation() as volatile.
2011-02-19 20:10:05 +00:00
05ff024dd8
(trunk libT) #3372 "What happened with closing idle peers?" -- fixed.
2011-02-19 12:32:41 +00:00
7a8ab0ddbe
(trunk libT) closeBadPeers() part 2 of 2: this function was serving two purposes, which wasn't clear from its name. refactor slightly to make the purpose and name match.
2011-02-19 12:30:18 +00:00
1592953439
(trunk libT) closeBadPeers() part 1 of 2: don't bother sorting these peers by activity -- we're closing -all- of them, so sorting them is pointless.
2011-02-19 12:24:00 +00:00
352981ee03
Remove obsolete UNUSED marker.
2011-02-18 21:55:42 +00:00
e2584b04ff
(trunk libT) #4047 "timeChecked for the last piece is not saved to a resume file" -- fixed.
...
A garden variety off-by-one error. good patch by jusid and revision by me.
2011-02-18 19:31:18 +00:00
e7fc697c90
Stub out UTP_Create in non-uTP builds.
2011-02-18 16:01:52 +00:00
80d06d4c34
(trunk libT) add a bool to tr_peer_stat and the RPC peer list to denote whether or not a peer's connected via µTP
2011-02-18 04:07:43 +00:00
b930026c40
(trunk libT) explicitly use the libevent2 header files, instead of the deprecated backwards-compatibility one
2011-02-18 01:25:32 +00:00
cba90ccace
(trunk libT) add stubs for UTP_Drained() and UTP_Write() to remove a few #ifdefs from peer-io.c
2011-02-18 01:18:51 +00:00
8e87147c61
(trunk libT) add a stub for UTP_Close() s.t. peer-mgr.c doesn't need the #ifdefs from r11960
2011-02-18 01:06:42 +00:00
cd04051de8
add configure script switch to enable/disable utp
2011-02-18 00:45:44 +00:00
40463b9d0a
Reinstate a variable removed by r11810.
2011-02-18 00:43:54 +00:00
828c9888b0
Don't attempt uTP connexions to peers learned from PEX that didn't signal support.
2011-02-18 00:43:49 +00:00
5af8e26251
Use large kernel buffers for the UDP socket when uTP is enabled.
...
Since we're using a single UDP socket to implement multiple uTP sockets,
and since we're not always timely in servicing an incoming UDP packet,
it's important to use a large receive buffer. The send buffer is probably
less critical, we increase it nonetheless.
2011-02-18 00:43:47 +00:00
a348c5d421
Don't call tr_utpClose in toggle_utp.
...
The uTP timer needs to run even when uTP has been disabled, or else the
uTP sockets will never time out.
2011-02-18 00:43:45 +00:00
2729d3c291
Don't mark atoms as non-uTP unless we failed to connect.
2011-02-18 00:43:43 +00:00
55b050f18e
Clear utp_failed flag upon seeing an announcement for ut_holepunch.
2011-02-18 00:43:39 +00:00
07addc7d0d
Try uTP connections by default.
2011-02-18 00:43:37 +00:00
2ba3305412
When failing to connect over uTP, fall back to TCP.
2011-02-18 00:43:34 +00:00
704f984bdf
Implement atom->utp_failed and peerMgrSetUpFailed.
2011-02-18 00:43:31 +00:00
5ba2e40f79
When uTP is disabled, don't call UTP_CheckTimeouts too often.
2011-02-18 00:43:27 +00:00
25b8d08537
When uTP is enabled, open uTP connections to some peers.
...
Since we don't implement falling back to TCP yet, we're very
conservative: we only use uTP when we have good reasons to believe
the peer speaks uTP.
2011-02-18 00:43:24 +00:00
cc96b12bbc
Fix typo.
2011-02-18 00:43:21 +00:00
004d30337b
when sending PEX messages, unset the holepunch flag because we don't support holepunch.
2011-02-18 00:41:32 +00:00
f00799f80c
if we successfully finish a handshake using uTP, mark the peer as supporting uTP
2011-02-18 00:41:06 +00:00
75badcd4ab
silence compiler warning
2011-02-18 00:40:53 +00:00
71486f8c86
call UTP_RBDrained when our read buffer is empty
2011-02-18 00:40:38 +00:00
e4b4812caa
add dbgmsg statements to peer-io's uTP code
2011-02-18 00:40:22 +00:00
9626aa4506
uTP peers need tr_peerIoFlush() to work during the handshake phase.
2011-02-18 00:40:01 +00:00
f6a85cb16b
implement bandwidth limiting for uploads to uTP peers
2011-02-18 00:39:33 +00:00
3c56dbea2a
Remove more debugging messages.
2011-02-18 00:36:36 +00:00
f33a772b00
Call didWriteWrapper in utp_on_write.
2011-02-18 00:36:33 +00:00
e35d1aacaf
Don't propagate information about ut_holepunch over PEX.
...
It's not useful, says Greg.
2011-02-18 00:36:31 +00:00
c9c2ad7c57
Don't use read_enabled in get_rb_size.
...
We're no longer maintaining read_enabled, as there are simpler ways of
doing the same. Additionally, we used RBDrained incorrectly.
2011-02-18 00:36:27 +00:00
77c034122e
Revert "Maintain information about enabled/disabled I/O."
...
This reverts commit 604f8fd10db747e57d8710cc53326c0e114ad32c.
2011-02-18 00:36:24 +00:00
c9a7becf87
Remove debugging messages.
2011-02-18 00:36:21 +00:00
88f4eac81f
Implement outgoing uTP connections.
2011-02-18 00:36:19 +00:00
456dbdd935
Export uTP send_to callback.
2011-02-18 00:36:16 +00:00
11114a74c0
Fix peerIoReconnect for uTP sockets.
2011-02-18 00:36:12 +00:00
dae7db8e10
Implement pacing of reads.
...
This should cause uTP sockets to respect read bandwidth limits. I'm not so
sure about the values we return for the read buffer size -- perhaps we
should allow some slack for network latency?
2011-02-18 00:36:09 +00:00
Jordan Lee