Charles Kerr
|
9fffb2be35
|
(trunk libT) Yippie, plug a small memory leak reported by valgrind!
|
2010-02-10 05:27:14 +00:00 |
|
Charles Kerr
|
a95aae5591
|
(trunk) as promised, remove the announce logging code. now safe for general-purpose nightly build testing again.
|
2010-02-10 04:50:21 +00:00 |
|
Charles Kerr
|
acc8e64102
|
(trunk) another test build that logs tracker announces to a file in /tmp, so you probably shouldn't use it unless I've asked you to. After hudson-t is done building a debug dmg of this, I'll commit r10155 to remove the logging code
|
2010-02-10 04:35:59 +00:00 |
|
Charles Kerr
|
2cbfe53717
|
(trunk libT) #2887 "dangling pointer in 1.81" -- fixed in trunk for 1.90 / 1.8x
|
2010-02-10 03:26:19 +00:00 |
|
Charles Kerr
|
a8a5a6cae5
|
(trunk) revert that last commit, which was only intended for about a dozen people. Unless you're one of them (and you don't know, then you're not) DON'T USE r10145. r10145 -- **ONLY** r10145 -- stored user's tracker announces to a logfile in /tmp.
|
2010-02-09 02:46:40 +00:00 |
|
Charles Kerr
|
8e51476258
|
pulling my hair out.... gaaaah
|
2010-02-09 01:47:16 +00:00 |
|
Charles Kerr
|
8b3fb8a5e8
|
(trunk libT) #2854 "`tracker did not respond' errors" -- lengthen the time we wait before timing out. This was shortened not too many releases ago to prevent a logjam from dead tpb tracker announces; however, we now have better ways of doing that and the short deadline may be contributing to the "did not respond" errors.
|
2010-02-05 04:54:08 +00:00 |
|
Charles Kerr
|
d118db0f37
|
(trunk libT) remove dead announce fields like supportscrypto and requirescrypto. these don't appear to be supported by any of the major clients so why waste the announce bandwidth...
|
2010-02-03 06:32:14 +00:00 |
|
Charles Kerr
|
5014319622
|
(trunk libT) experimental commit for faster tracker announces
|
2010-02-03 03:38:13 +00:00 |
|
Charles Kerr
|
931558aea3
|
(trunk libT) fix DNS timing issue reported by gn0s1s in irc
|
2010-01-28 18:25:18 +00:00 |
|
Charles Kerr
|
dd41b5c257
|
(trunk libT) #2819 "announce URLs containing raw IP addresses won't announce/scrape in 1.82" -- possible fix
|
2010-01-28 17:24:23 +00:00 |
|
Charles Kerr
|
a334b422c3
|
(trunk) remove unnecessary #includes
|
2010-01-28 13:33:40 +00:00 |
|
Charles Kerr
|
a888c4d425
|
(trunk libT) web.c: add a cache for unresolvable hostnames
|
2010-01-27 06:40:12 +00:00 |
|
Charles Kerr
|
3ace0afaae
|
(trunk libT) #2792 "since 1.80 update downloads failing" -- try to handle non-resolving tracker addresses without blocking in curl
|
2010-01-26 07:22:50 +00:00 |
|
Charles Kerr
|
c2adab97b2
|
(trunk libT) fix typo from prev commit
|
2010-01-25 04:26:08 +00:00 |
|
Charles Kerr
|
4164b17977
|
(trunk libT) extend the lifespan of the cached DNS entries
|
2010-01-25 04:24:48 +00:00 |
|
Charles Kerr
|
1d1b09b75b
|
(trunk libT) #2804 "segmentation fault add-by-url when url scheme is ftp" -- fixed in trunk for 1.83
|
2010-01-25 04:00:08 +00:00 |
|
Charles Kerr
|
ae966e0725
|
(trunk libT) increase the announce timeout from 30 seconds to 45 seconds.
|
2010-01-24 23:32:49 +00:00 |
|
Charles Kerr
|
7e5e4db797
|
(trunk libT) #2783 "blocklist is empty" -- fix this 1.82 regression. apparently update.transmissionbt.com doesn't like it when you make the ":80" in the Host: header explicit instead of implicit... so probably other web servers have the same problem too.
|
2010-01-24 19:22:22 +00:00 |
|
Charles Kerr
|
d97a04b001
|
(trunk libT) when adding the Host: header by hand, make include the host's port
|
2010-01-23 07:51:51 +00:00 |
|
Charles Kerr
|
2be24ec9a8
|
(trunk libT) #2783 "starting with r9992, IP blocklist is empty" -- when replacing a URL's hostname with an IP address before passing the URL to curl, manually set the Host: header to the original hostname. (Xref: #2781)
|
2010-01-22 07:27:19 +00:00 |
|
Charles Kerr
|
0f6214d9ae
|
(trunk libT) if we're calling evdns_init(), we probably ought to call evdns_shutdown() on exit...
|
2010-01-22 03:39:21 +00:00 |
|
Charles Kerr
|
1f44272678
|
(trunk libT) #2781 "1.80 freezes on startup" -- possible fix. implement nonblocking dns before we hand the URLs over to libcurl
|
2010-01-22 02:40:11 +00:00 |
|
Charles Kerr
|
e2feb075b9
|
(trunk libT) reinstate proxies in the http announce
|
2010-01-10 20:15:56 +00:00 |
|
Charles Kerr
|
2ac553361a
|
(trunk libT) #2416 "crash in event_queue_insert" -- I think this is happening if we queue up two changes, one to listen for read/write, and then a second one to delete the polling, in kqueue's event mechanism, at libcurl's request. Let's try disabling kqueue in libevent and use poll/select instead.
|
2010-01-10 19:52:01 +00:00 |
|
Mitchell Livingston
|
14da64ec06
|
assorted small cleanup and log removal
|
2010-01-10 16:26:30 +00:00 |
|
Mitchell Livingston
|
5af45426d4
|
random code optimization: when escaping http, use an if with a lot less comparisons rather than a switch
|
2010-01-10 14:56:04 +00:00 |
|
Mitchell Livingston
|
8f6e1dedf5
|
revert r9505, which appears to cause a crash on launch
|
2010-01-10 14:36:53 +00:00 |
|
Charles Kerr
|
5a60e4549d
|
(trunk libT) #2416 "crash in event_queue_insert" -- another guess, based on the idea that we might queue two actions (one to insert, one to delete) on the same event before libevent's dispatcher is invoked
|
2010-01-10 08:53:17 +00:00 |
|
Charles Kerr
|
69cb1362e3
|
(trunk libT) #2416 "crash in event_queue_insert" -- see if we can get the program to crash in a different way.
|
2010-01-09 07:27:45 +00:00 |
|
Charles Kerr
|
0bb355d8df
|
(trunk libT) #2416 "crash in event_queue_insert" -- continue throwing the kitchen sink at this ticket. (1) disable proxies to see if that makes any difference (2) rearrange the sequencing in sock_cb() to unconditionally delete the active event before doing anything else (3) use a libevent timer instead of CURLOPT_TIMEOUT to workaround curl bug http://tinyurl.com/ycm5d3e
|
2010-01-08 21:45:56 +00:00 |
|
Charles Kerr
|
5c26afdb52
|
happy new year!
|
2010-01-04 21:00:47 +00:00 |
|
Charles Kerr
|
339173d1b4
|
(trunk libT) make sure to restart curl's periodic timer
|
2010-01-04 20:06:39 +00:00 |
|
Charles Kerr
|
8ae3911477
|
(trunk libT) #2416 "crash in event_queue_insert" -- I am really quite sick of this bug. It's not reproducible at all under Linux and appears to be a bug in libevent's kqueue implementation. In this commit, we leak the event object on OS X.
|
2009-12-29 00:00:33 +00:00 |
|
Charles Kerr
|
6151881a03
|
(trunk libT) #2416 "crash in event_queue_insert" -- instead of freeing libevent's event objects, pool and reuse them. *sigh*
|
2009-12-28 23:29:59 +00:00 |
|
Charles Kerr
|
ee58472ec2
|
(trunk libT) fix minor bug in r9651 "Reduce SO_SNDBUF and SO_RCVBUF for tracker announce/scrape messages" -- libcurl's CURLOPT_SOCKOPTFUNCTION doesn't return void; it returns an int that is nonzero if an error occurs.
|
2009-12-28 23:25:50 +00:00 |
|
Charles Kerr
|
ac202f13bb
|
(trunk libT) remove dead code -- tr_assert()
|
2009-12-28 23:11:26 +00:00 |
|
Charles Kerr
|
bc86c0b364
|
(trunk libT) #2416 "crash in event_queue_insert" -- maybe this will finally fix it...
|
2009-12-26 23:57:25 +00:00 |
|
Charles Kerr
|
8903e1b4e3
|
(trunk libT) #2416 "crash in event_queue_insert" -- another experimental commit. /as an experiment/ let's just leak that event struct and see if that makes the crash go away. Obviously leaking those objects is not a long term fix but this will help point the way towards the real fix.
|
2009-12-24 18:50:34 +00:00 |
|
Charles Kerr
|
b8d7e76750
|
(trunk libT) #2416 "crash in event_queue_insert" -- possible fix. it looks like the kqueue client is accessing the event after the callback and after event_del() is being called. *If* that's the case I don't know if it's a libevent bug or if libtransmission is making assumptions it shouldn't've... but it's easy enough to keep the memory in a pool and free it later when it's safer to do so.
|
2009-12-23 16:12:18 +00:00 |
|
Charles Kerr
|
ef650d0398
|
(trunk libT) it certainly smells like we're using freed memory in the libcurl + libevent code in web.c... let's trash the structures right before free()ing them
|
2009-12-18 17:32:16 +00:00 |
|
Charles Kerr
|
0601442f58
|
(trunk libT) #2416 "crash in event_queue_insert" -- another stab at fixing this. yay
|
2009-12-14 18:24:26 +00:00 |
|
Charles Kerr
|
87aefaf5ff
|
(trunk libT) #1699 "announce to multiple trackers at once" -- fix minor bug in reporting scrape results when an announce returns enough information that a separate scrape isn't necessary.
|
2009-12-14 17:17:05 +00:00 |
|
Charles Kerr
|
ee0de4b06c
|
(trunk libT) keep simplifying web.c... is there anything left to strip out?
|
2009-12-14 14:25:22 +00:00 |
|
Charles Kerr
|
c33aef7e2d
|
(trunk libT) more web.c simplification + handle an unhandled case CURL_POLL_NONE in the socket callback.
|
2009-12-14 12:54:30 +00:00 |
|
Charles Kerr
|
2653cd5153
|
(trunk libT) keep simplifying web.c. also, fix an improbable FMR at shutdown
|
2009-12-14 05:11:33 +00:00 |
|
Charles Kerr
|
6d9e6ececc
|
(trunk libT) remove some more obsolete debug messages and other cruft from web.c
|
2009-12-13 19:33:02 +00:00 |
|
Charles Kerr
|
485b64a5e3
|
(trunk libT) it never ends
|
2009-12-13 17:54:01 +00:00 |
|
Charles Kerr
|
a89d2217af
|
(trunk libT) #2416 "crash in event_queue_insert()"
|
2009-12-11 15:41:34 +00:00 |
|
Charles Kerr
|
5f1f2b1e85
|
(trunk libT) use curl_multi_socket_action(), even on OS X.
|
2009-12-10 19:05:21 +00:00 |