kmx git

Commit	Date	Message
fe008ed6	2010-09-09T14:59:27	Make all versioning changes for 2.0.7-rc, and add ChangeLog
d98511c0	2010-09-09T14:36:45	Make event_base_virtual_del() notify the base if needed
2756a10c	2010-09-09T13:43:31	Add a missing time.h include to test/regress_thread.c
fb36f9a7	2010-09-09T13:00:54	Fix an uninitialized-variable warning on windows
71b68014	2010-09-08T20:33:21	Don't decrement virutal event count twice in connect_complete.
3ec65d69	2010-09-08T19:55:13	Fix a few Windows compile warnings.
25b6a74b	2010-09-08T14:53:57	Merge branch 'tests'
6f821710	2010-09-08T14:52:37	Merge remote branch 'github/win_notify'
de412948	2010-09-08T14:52:24	Add a missing header for regress_thread.c
e7dc501e	2010-09-08T14:40:51	Implement EVBASE_NEED_NOTIFY on win32
911e0db8	2010-09-08T20:29:39	Fix a compile warning in regress_thread.c
3658b169	2010-09-08T14:12:12	Merge remote branch 'chrisd/iocp-fixes4' Conflicts: test/regress_thread.c
9580e282	2010-09-08T14:00:45	Merge branch 'th_notify_fd_reinit'
ce85280b	2010-09-08T13:29:06	Improve testing of when thread-notification occurs
4632b78e	2010-09-08T13:22:55	Minimize calls to base_notify implementation functions, thereby avoiding needless syscalls The trick here is that if we already told the base to wake up, and it hasn't woken up yet, we don't need to tell it to wake up again. This should help lots with inherently multithreaded code like IOCP.
c7a06bfa	2010-09-08T13:02:58	Avoid needlessly calling evthread_notify_base() when the loop is not running Also make sure that we always hold the base lock when calling evthread_notify_base.
57d3413c	2010-09-08T11:39:24	Merge remote branch 'github/globals'
17a14f1a	2010-09-01T11:04:57	Only process up to MAX_DEFERRED deferred_cbs at a time. If threads queue callbacks while event_process_deferred_callbacks is running, the loop may spin long enough to significantly skew timers. A unit test stressing this behavior is also in this commit.
2447fe88	2010-08-28T04:07:48	Add event_config_set_num_cpus_hint for tuning thread pools, etc.
499452f4	2010-08-28T02:44:11	IOCP-related unit test tweaks
76f7e7ae	2010-08-17T05:02:00	Some IOCP bufferevent tweaks. - Increment reference count of bufferevents before initiating overlapped operations to prevent the destructor from being called while operations are pending. The only portable way of canceling overlapped ops is to close the socket. - Translate error codes to WSA* codes. - Better handling of errors. - Add an interface to add and del "virtual" events. Because IOCP bufferevents don't register any events with the base, the event loop has no way of knowing they exist. This causes the loop to terminate prematurely. event_base_{add,del}_virtual increment/decrement base's event count so the loop runs while there are any enabled IOCP bufferevents.
d844242f	2010-08-28T02:08:27	Stop IOCP when freeing the event_base.
03afa209	2010-08-16T01:23:57	IOCP-related evbuffer fixes. - Prevent evbuffer_{add,prepend}_buffer from moving read-pinned chains. - Fix evbuffer_drain to handle read-pinned chains better. - Raise the limit on WSABUFs from two to MAX_WSABUFS for overlapped reads.
1115366e	2010-09-07T10:28:15	Fix a few memory leaks in the tests
42090072	2010-09-06T15:47:07	Move the "function to getsockname() on a listener" to regress_testutils This reverts commit fab50488fcb741884ccdfa7b83643eac3e5c9cbf. The function was, on reflection, not important enough to break the feature freeze, since it's trivial to build on your own.
c51826ff	2010-09-06T15:40:13	Merge remote branch 'github/sysqueue_include_order'
b0f284cb	2010-09-06T10:10:17	Fix event_del(0) instance in bench.c
749128b2	2010-09-04T22:02:32	Merge remote branch 'github/win32_posix_underscore'
150599e6	2010-09-04T22:00:38	Merge remote branch 'github/cdecl'
b5dd8064	2010-09-03T22:00:25	Fix uninitialized port var in http_delete_test. Last one, I hope.
3b3fb743	2010-09-03T21:03:25	Fix uninitialized variables in http_bad_request_test. (oops)
a97320ac	2010-09-03T18:48:31	Allow more than one copy of regression tests to run at once Mostly this was a matter of just removing all the hardwired ports in the test code. The http/connection_retry test is still a little screwy, though.
19521436	2010-09-03T16:42:16	Expose a function to add a nameserver by sockaddr
fab50488	2010-09-03T16:41:16	Expose a function to getsockname() on a listener's fd.
7ea8e89d	2010-09-03T15:12:35	Prefer autoreconf -ivf to manual autogen.sh Suggested by Ralph Castain
47882773	2010-09-02T13:21:17	Fix pointer-to-__cdecl-function syntax
e50c0fcc	2010-09-02T11:10:50	Use the _func() replacements for open, fstat, etc in evutil.c on win32 Remember that in a fit of ANSI C compliance, Microsoft decided to screw portability by renaming basically all the functions in unistd.h to get prefixed with an understore. For some reason, mingw didn't seem to mind, but at least some people's compilers did: see bug 3044490.
f0056d04	2010-09-02T12:06:58	Declare signal handler function as "__cdecl" on Windows. I swear, they must have half a dozen different calling conventions. (goes to check) Holy crud. They actually do. There's __cdecl, __stdcall, __fastcall, "thiscall", "naked" and the obsolete "__pascal", "__fortran", and "__syscall". And don't forget WINAPI and __far. Anyways, this should fix 3044488 if I got it right.
ca9048f1	2010-09-02T11:36:44	Move evkeyvalq into a separate header for evhttp_parse_query users The evhttp_parse_query API is a bit misdesigned; all the other evkeyvalq stuff is abstract and lets you get away with having a header stub, but evhttp_parse_query seems to require that you instantiate an empty evkeyvalq of your own.
d3ceca80	2010-09-02T11:27:57	Declare evkeyvalq and event_list even if event_struct.h comes before sys/queue.h Fixes bug 3036645 reported by Mihai Draghicioiu
495ed667	2010-09-01T16:36:30	Close th_notify_fds and open a new pair on reinit After a fork, you want subthreads to wake up the event_base in the child process, not to have the child process and the main process fight over who wakes up whom. Related to a problem found by Nicholas Marriott while debugging 3048812.
5de2bcb7	2010-09-01T16:03:39	On windows, make lock/thread function tables static This requires us to have a separate implementation of the lock macros that indirects to a set of functions. Fortunately, this isn't too hard to do. This may be a fix for bug 3042969, where our openssl dll and our libevent dll each got their own version of the thread stuff.
5218d2a8	2010-09-01T15:56:22	Make defer-internal.h use lock macros, not direct cess to lock fns
1fdec20f	2010-09-01T15:01:39	Stop using global arrays to implement the EVUTIL_ctype functions These apparently made libtool sad on win32, and the function call overhead here should be negligable anyway.
acc4aca4	2010-08-30T11:35:06	Fix a bug in our win32 condition implementation The do ... while loop in our wait code could spin while waiting because the event object wasn't reset until there were no longer any events waiting to be woken up. We also want to reset the event object if the count of events to wake up reaches zero. Found by Chris Davis. Fixes bug 3053358.
d61b2f33	2010-08-26T15:09:58	Fix an issue with forking and signal socketpairs in select/poll backends Nicholas Marriott identifies an issue where a signal socketpair doesn't get recreated if the event backend doesn't set event_reinit. See bug 3048812 There may be a similar issue with respect to th_notify_fd
041989fb	2010-08-17T14:44:12	Rename regress_pthread.c to regress_thread.c
f1074b77	2010-08-23T12:01:45	Detect events with no ev_base; warn instead of crashing
743f8665	2010-08-23T11:48:46	Honor NDEBUG; build without warnings with NDEBUG; make NDEBUG always-off in unit test code
970e6ad2	2010-08-19T14:00:06	Avoid deadlock when activating signals. Fixes bug 3048812. Based on patch by Nicholas Marriott.
6123d129	2010-08-19T09:38:44	Fix kquue correctness test on x84_64 Apparently, in our configure.in check for a working kqueue, we were leaving some fields unset that seemed to irritate 64-bit kqueue a lot. Found by Christopher Layne
4c32b9de	2010-08-18T12:02:25	Fix logic error in win32 TRY_LOCK that caused problems with rate-limiting
d0b88433	2010-08-18T10:49:12	Set close-on-exec bit for filedescriptors created by dns subsystem. Based on patch for 1.4 by Ralf Schmitt.
da6e7cd4	2010-08-17T13:26:03	Correctly detect failure to delete bufferevent read-timeout event Gilad Benjamini noted that we check the error code for deleting a write-timeout event twice, and the read timeout not at all. This shouldn't be a bit problem, since it's really hard for a delete to fail on a timeout-only event, but it's worth fixing. Fixes bug 3046787
5fb10958	2010-08-16T22:55:45	Add a unit test for conditions
d74ae381	2010-08-16T12:38:22	Make the regress_pthread.c tests work on windows with current test APIs
e0972c21	2010-08-17T13:18:18	Use conditions instead of current_event_lock to fix a deadlock. Avi Bab correctly noted as bug 3044479 the fact that any thread blocking on current_event_lock will do so while holding th_base_lock, making it impossible for the currently running event's callback to call any other functions that require th_base_lock. This patch switches the current_event_lock code to instead use a condition variable that we wait on if we're trying to mess with a currently-executing event, and that we signal when we're done executing a callback if anybody is waiting on it.
d4977b52	2010-08-17T13:15:34	Add a condition variable backend, with implementations for pthreads and win32 The interface from the user's POV is similar to the locking implementation: either provide a structure full of function pointers, or just call evthread_use__threads() and everything will be okay. The internal interface is meant to vaguely resemble pthread_cond_, which Windows people will better recognize as ConditionVariable.
39201728	2010-08-13T17:19:18	Make definition of WIN32_LEAN_AND_MEAN in event.h conditional. (Inclusion order might trigger a warning for redefinition on Windows.)
60433a0a	2010-08-13T17:08:59	Clean up syntax on TAILQ_ENTRY() usage Though the C standards allow it, it's apparently possible to get MSVC upset by saying "struct { int field; } (declarator);" instead of "struct {int field; } declarator;", so let's just not do that. Bugfix for 3044492 (commit msg by nickm)
a4af9be1	2010-08-13T11:41:37	Make include/event2/event-config.h not included in source dist As a generated file, it shouldn't get included in our source distribution. Apparently there is an automake incant for this: nobase_ even stacks with nodist_ .
4022b287	2010-08-13T11:34:39	Change include order in Makefile.nmake If there is an event-config.h in include/event2 (either because we screwed up packaging like in 2.0.6-rc or because we previously tried building with mingw and we didn't make distclean in the middle), we want MSVC to find the one one in WIN32-Code/include/event2 first. Found by Gilad Benjamini.
4991669b	2010-08-10T21:05:52	Update the whatsnew-2.0.txt document
33200e72	2010-08-10T15:03:14	Document that DNS_NO_SEARCH is an obsolete alias for DNS_QUERY_NO_SEARCH
3808168d	2010-08-10T15:02:50	Completely remove the (mostly-removed) obsolete thread functions.
9b5c19ac	2010-08-10T11:46:00	Merge branch 'more_pkgconfig'
ebcb1f03	2010-08-10T11:11:25	Add pkgconfig files for libevent_{openssl,pthreads}
0bffe43a	2010-08-09T12:08:40	Fix a nasty dangling-event bug when using rate-limiting groups When we freed a bufferevent that was in a rate-limiting group and blocked on IO, the process of freeing it caused it to get removed from the group. But removing the bufferevent from the group made its limits get removed, which could make it get un-suspended and in turn cause its events to get re-added. Since we would then immediately _free_ the events, this would result in dangling pointers. Fixes bug 3041007.
ff481a8e	2010-08-06T23:22:01	Increment vesion to 2.0.6-rc-dev
556d0f40	2010-08-06T21:19:44	Increment versions to 2.0.6-rc
f6582640	2010-08-06T21:11:01	Changlog and new acknowledgements for 2.0.6-rc
ec347b92	2010-07-07T16:45:03	Move event-config.h to include/event2 This change means that all required include files are in event2, and all files not in event2/* are optional.
5fa30d2b	2010-08-06T17:13:27	Fix evdns build with -DUNICODE It turns out that GetProcAddress always takes its second argument as a C string, regardless of whether unicode is on or not.
28f31a4f	2010-08-06T16:36:23	Fix unit tests with -DUSE_DEBUG enabled If you were to enable USE_DEBUG and slog through all 700+ MB of debugging output, you'd find that one of the unit tests failed, since it tested the debug logging code, but the string it expected and the string it logged differed by a tab vs 2 spaces.
ea1ea3d6	2010-08-06T13:34:51	Make test-changelist count cpu usage right on win32
9b602096	2010-08-06T13:03:17	Use AF_INET socketpair to test sendfile on Solaris
57b30cd7	2010-08-06T13:01:32	Turn our socketpair() replacement into its own function This patch splits the formerly windows-only case of evutil_socketpair() into an (internal-use-only) function named evutil_ersatz_socketpair(), and makes it build and work right on non-Windows hosts. We need this for convenience to test sendfile on solaris, where socketpair can't give you an AF_INET pair, and sendfile() won't work on AF_UNIX.
de1f5d6d	2010-08-06T09:33:55	Remember to initialize timeout events for bufferevent_async Fixes segfaults in unit tests.
170ffd2b	2010-08-05T15:57:30	Increase the tolerance in our unit tests for sloppy clocks. (Apparently openbsd in virtualbox just doesn't keep very accurate time.)
e996b3d4	2010-08-05T15:51:16	Make tests quieter on local dns resolver failure
4558ce23	2010-08-04T17:31:28	Merge branch 'openssl_ratelim'
b2c6202d	2010-08-04T15:52:32	Fix an assertion bug in test-ratelim If the rate limit was low enough, then the echo_conns wouldn't finish inside the 300 msec we allowed for them to close. Instead, count the number of connections we have, and keep waiting until they are all closed.
42f6b624	2010-08-04T15:51:14	Add option to test-ratelim to test min_share
6ae53d67	2010-08-04T15:44:08	Add an interface to expose min_share in ratelimiting groups
009f3005	2010-08-04T14:54:38	Fix rate-limit calculation on openssl bufferevents. When you're doing rate limiting on an openssl connection, you nearly always want to limit the number of bytes sent and received over the wire, not the number of bytes read or written over the secure transport.
13b912e4	2010-08-01T21:39:48	Fixed compilation of sample/le-proxy.c on win32
42a8c711	2010-07-31T17:10:04	Build more cleanly with NetBSDs that dislike toupper(char) To be fair, when char can be signed, if toupper doesn't take negative characters, toupper(char) is a very bad idea. So let's just use the nice safe EVUTIL_TOUPPER instead. (It explicitly only upcases ASCII, but we only use it for identifiers that we know to be ASCII anyway).
840a72fb	2010-07-29T20:32:40	Fix badly-behaved subtest of dns/bufferevent_connect_hostname The bufferevent_connect_hostname test was specifying AF_INET, but the gethostbyname test we were using to see what error to expect was using PF_UNSPEC, leading to possible divergence of results.
9e725f72	2010-07-29T18:12:05	Improve error message for failed epoll to make debugging easier.
cc2379d2	2010-07-26T14:48:32	Constify a couple of arguments to evdns_server_request_add_*_reply
e1c1167c	2010-07-22T14:38:08	Replace (unused,always 0) is_tcp argument to evdns_add_server_port*() with flags Since we weren't using it for anything, and we always failed if it was set, we're allowed to change the future semantics of setting it.
7e87a599	2010-07-21T14:33:42	Stop asserting when asked for a (unsupported) TCP dns port. Just return NULL.
c9913174	2010-07-19T15:31:19	Change bufferevent_openssl::do_write so it doesn't call SSL_write with a 0 length buffer I was running into a problem when using bufferevent_openssl with a very simple echo server. My server simply bufferevent_read_buffer 'd data into an evbuffer and then passed that evbuffer straight to bufferevent_write_buffer. The problem was every now and again the write would fail for no apparent reason. I tracked it down to SSL_write being called with the amount of data to send being 0. This patch alters do_write in bufferevent_openssl so that it skips io_vecs with 0 length.
7c926916	2010-07-19T15:14:10	Make test/test.sh call test-changelist Eventually test-changelist should expand to try more cases, maybe query the status of the actual changelist somehow, and integrate it with the rest of the unit tests. Also, add test-changelist to gitignore.
cb927a51	2010-07-19T15:03:43	Fix whitespace.
b00d4c04	2010-07-19T14:58:08	Remove unused variables in test/test-changelist.c
cf249e7d	2010-07-19T13:44:56	Possible fix to 100% cpu usage with epoll and openssl I'm running a fairly simple bit of test code using libevent2 with epoll and openssl bufferevents and I've run into a 100% cpu usage problem. Looking into it 100% usage was caused by epoll_wait constantly returning write events on the openssl socket when it shouldn't really have been looking for write events at all (N_ACTIVE_CALLBACKS() was returning 0 also). Looking a bit deeper eventbuffer_openssl socket seems to be requesting that the EV_WRITE event be removed when it should, but the event isn't actually being removed from epoll. Continuing to follow this I think I've found a bug in event_changelist_del. For evpoll event_del calls event_changelist_del which caches the change which is then actioned later when evpoll_dispatch is called. In event_changlist_del there is a check so that if the currently changed action is an add then the cached action is changed to a no-op rather than a delete (which makes sense). The problem arises if there are more than two add or delete operations between calls to dispatch, in this case it's possible that the delete is turned into a no-op when it shouldn't have been. For example starting with the event on, a delete followed by an add and then another delete results in a no-op when it should have been a delete (I added a fair bit of debug output that seems to confirm this behaviour). I've applied a small change that checks the original old_event stored with the change and only converts the delete to a no-op if the event isn't on in old_event. This seems to have fixed my problem.
928b7d49	2010-07-16T12:38:23	Merge branch 'pncalls_fix'
65abdc20	2010-07-16T09:11:09	Fix wrong sie calculation of iovec buffers when exact=1 The old code had a bug where the 'exact' flag to 1 in _evbuffer_read_setup_vecs would never actually make the iov_len field of the last iovec get truncated. This patch fixes that.
7c2dea16	2010-07-13T11:06:08	Pass flags to fcntl(F_SETFL) and fcntl(F_SETFD) as int, not long Everybody but Linux documents this as taking an int, and Linux is very tolerant of getting an int instead. If it weren't, everybody doing fcntl(fd,F_SETFL,O_NONBLOCK) would break, since the glibc headers define O_NONBLOCK as an int literal.

fe008ed6

2010-09-09T14:59:27

Make all versioning changes for 2.0.7-rc, and add ChangeLog

d98511c0

2010-09-09T14:36:45

Make event_base_virtual_del() notify the base if needed

2756a10c

2010-09-09T13:43:31

Add a missing time.h include to test/regress_thread.c

fb36f9a7

2010-09-09T13:00:54

Fix an uninitialized-variable warning on windows

71b68014

2010-09-08T20:33:21

Don't decrement virutal event count twice in connect_complete.

3ec65d69

2010-09-08T19:55:13

Fix a few Windows compile warnings.

25b6a74b

2010-09-08T14:53:57

Merge branch 'tests'

6f821710

2010-09-08T14:52:37

Merge remote branch 'github/win_notify'

de412948

2010-09-08T14:52:24

Add a missing header for regress_thread.c

e7dc501e

2010-09-08T14:40:51

Implement EVBASE_NEED_NOTIFY on win32

911e0db8

2010-09-08T20:29:39

Fix a compile warning in regress_thread.c

3658b169

2010-09-08T14:12:12

Merge remote branch 'chrisd/iocp-fixes4' Conflicts: test/regress_thread.c

9580e282

2010-09-08T14:00:45

Merge branch 'th_notify_fd_reinit'

ce85280b

2010-09-08T13:29:06

Improve testing of when thread-notification occurs

4632b78e

2010-09-08T13:22:55

Minimize calls to base_notify implementation functions, thereby avoiding needless syscalls The trick here is that if we already told the base to wake up, and it hasn't woken up yet, we don't need to tell it to wake up again. This should help lots with inherently multithreaded code like IOCP.

c7a06bfa

2010-09-08T13:02:58

Avoid needlessly calling evthread_notify_base() when the loop is not running Also make sure that we always hold the base lock when calling evthread_notify_base.

57d3413c

2010-09-08T11:39:24

Merge remote branch 'github/globals'

17a14f1a

2010-09-01T11:04:57

Only process up to MAX_DEFERRED deferred_cbs at a time. If threads queue callbacks while event_process_deferred_callbacks is running, the loop may spin long enough to significantly skew timers. A unit test stressing this behavior is also in this commit.

2447fe88

2010-08-28T04:07:48

Add event_config_set_num_cpus_hint for tuning thread pools, etc.

499452f4

2010-08-28T02:44:11

IOCP-related unit test tweaks

76f7e7ae

2010-08-17T05:02:00

Some IOCP bufferevent tweaks. - Increment reference count of bufferevents before initiating overlapped operations to prevent the destructor from being called while operations are pending. The only portable way of canceling overlapped ops is to close the socket. - Translate error codes to WSA* codes. - Better handling of errors. - Add an interface to add and del "virtual" events. Because IOCP bufferevents don't register any events with the base, the event loop has no way of knowing they exist. This causes the loop to terminate prematurely. event_base_{add,del}_virtual increment/decrement base's event count so the loop runs while there are any enabled IOCP bufferevents.

d844242f

2010-08-28T02:08:27

Stop IOCP when freeing the event_base.

03afa209

2010-08-16T01:23:57

IOCP-related evbuffer fixes. - Prevent evbuffer_{add,prepend}_buffer from moving read-pinned chains. - Fix evbuffer_drain to handle read-pinned chains better. - Raise the limit on WSABUFs from two to MAX_WSABUFS for overlapped reads.

1115366e

2010-09-07T10:28:15

Fix a few memory leaks in the tests

42090072

2010-09-06T15:47:07

Move the "function to getsockname() on a listener" to regress_testutils This reverts commit fab50488fcb741884ccdfa7b83643eac3e5c9cbf. The function was, on reflection, not important enough to break the feature freeze, since it's trivial to build on your own.

c51826ff

2010-09-06T15:40:13

Merge remote branch 'github/sysqueue_include_order'

b0f284cb

2010-09-06T10:10:17

Fix event_del(0) instance in bench.c

749128b2

2010-09-04T22:02:32

Merge remote branch 'github/win32_posix_underscore'

150599e6

2010-09-04T22:00:38

Merge remote branch 'github/cdecl'

b5dd8064

2010-09-03T22:00:25

Fix uninitialized port var in http_delete_test. Last one, I hope.

3b3fb743

2010-09-03T21:03:25

Fix uninitialized variables in http_bad_request_test. (oops)

a97320ac

2010-09-03T18:48:31

Allow more than one copy of regression tests to run at once Mostly this was a matter of just removing all the hardwired ports in the test code. The http/connection_retry test is still a little screwy, though.

19521436

2010-09-03T16:42:16

Expose a function to add a nameserver by sockaddr

fab50488

2010-09-03T16:41:16

Expose a function to getsockname() on a listener's fd.

7ea8e89d

2010-09-03T15:12:35

Prefer autoreconf -ivf to manual autogen.sh Suggested by Ralph Castain

47882773

2010-09-02T13:21:17

Fix pointer-to-__cdecl-function syntax

e50c0fcc

2010-09-02T11:10:50

Use the _func() replacements for open, fstat, etc in evutil.c on win32 Remember that in a fit of ANSI C compliance, Microsoft decided to screw portability by renaming basically all the functions in unistd.h to get prefixed with an understore. For some reason, mingw didn't seem to mind, but at least some people's compilers did: see bug 3044490.

f0056d04

2010-09-02T12:06:58

Declare signal handler function as "__cdecl" on Windows. I swear, they must have half a dozen different calling conventions. (goes to check) Holy crud. They actually do. There's __cdecl, __stdcall, __fastcall, "thiscall", "naked" and the obsolete "__pascal", "__fortran", and "__syscall". And don't forget WINAPI and __far. Anyways, this should fix 3044488 if I got it right.

ca9048f1

2010-09-02T11:36:44

Move evkeyvalq into a separate header for evhttp_parse_query users The evhttp_parse_query API is a bit misdesigned; all the other evkeyvalq stuff is abstract and lets you get away with having a header stub, but evhttp_parse_query seems to require that you instantiate an empty evkeyvalq of your own.

d3ceca80

2010-09-02T11:27:57

Declare evkeyvalq and event_list even if event_struct.h comes before sys/queue.h Fixes bug 3036645 reported by Mihai Draghicioiu

495ed667

2010-09-01T16:36:30

Close th_notify_fds and open a new pair on reinit After a fork, you want subthreads to wake up the event_base in the child process, not to have the child process and the main process fight over who wakes up whom. Related to a problem found by Nicholas Marriott while debugging 3048812.

5de2bcb7

2010-09-01T16:03:39

On windows, make lock/thread function tables static This requires us to have a separate implementation of the lock macros that indirects to a set of functions. Fortunately, this isn't too hard to do. This may be a fix for bug 3042969, where our openssl dll and our libevent dll each got their own version of the thread stuff.

5218d2a8

2010-09-01T15:56:22

Make defer-internal.h use lock macros, not direct cess to lock fns

1fdec20f

2010-09-01T15:01:39

Stop using global arrays to implement the EVUTIL_ctype functions These apparently made libtool sad on win32, and the function call overhead here should be negligable anyway.

acc4aca4

2010-08-30T11:35:06

Fix a bug in our win32 condition implementation The do ... while loop in our wait code could spin while waiting because the event object wasn't reset until there were no longer any events waiting to be woken up. We also want to reset the event object if the count of events to wake up reaches zero. Found by Chris Davis. Fixes bug 3053358.

d61b2f33

2010-08-26T15:09:58

Fix an issue with forking and signal socketpairs in select/poll backends Nicholas Marriott identifies an issue where a signal socketpair doesn't get recreated if the event backend doesn't set event_reinit. See bug 3048812 There may be a similar issue with respect to th_notify_fd

041989fb

2010-08-17T14:44:12

Rename regress_pthread.c to regress_thread.c

f1074b77

2010-08-23T12:01:45

Detect events with no ev_base; warn instead of crashing

743f8665

2010-08-23T11:48:46

Honor NDEBUG; build without warnings with NDEBUG; make NDEBUG always-off in unit test code

970e6ad2

2010-08-19T14:00:06

Avoid deadlock when activating signals. Fixes bug 3048812. Based on patch by Nicholas Marriott.

6123d129

2010-08-19T09:38:44

Fix kquue correctness test on x84_64 Apparently, in our configure.in check for a working kqueue, we were leaving some fields unset that seemed to irritate 64-bit kqueue a lot. Found by Christopher Layne

4c32b9de

2010-08-18T12:02:25

Fix logic error in win32 TRY_LOCK that caused problems with rate-limiting

d0b88433

2010-08-18T10:49:12

Set close-on-exec bit for filedescriptors created by dns subsystem. Based on patch for 1.4 by Ralf Schmitt.

da6e7cd4

2010-08-17T13:26:03

Correctly detect failure to delete bufferevent read-timeout event Gilad Benjamini noted that we check the error code for deleting a write-timeout event twice, and the read timeout not at all. This shouldn't be a bit problem, since it's really hard for a delete to fail on a timeout-only event, but it's worth fixing. Fixes bug 3046787

5fb10958

2010-08-16T22:55:45

Add a unit test for conditions

d74ae381

2010-08-16T12:38:22

Make the regress_pthread.c tests work on windows with current test APIs

e0972c21

2010-08-17T13:18:18

Use conditions instead of current_event_lock to fix a deadlock. Avi Bab correctly noted as bug 3044479 the fact that any thread blocking on current_event_lock will do so while holding th_base_lock, making it impossible for the currently running event's callback to call any other functions that require th_base_lock. This patch switches the current_event_lock code to instead use a condition variable that we wait on if we're trying to mess with a currently-executing event, and that we signal when we're done executing a callback if anybody is waiting on it.

d4977b52

2010-08-17T13:15:34

Add a condition variable backend, with implementations for pthreads and win32 The interface from the user's POV is similar to the locking implementation: either provide a structure full of function pointers, or just call evthread_use_*_threads() and everything will be okay. The internal interface is meant to vaguely resemble pthread_cond_*, which Windows people will better recognize as *ConditionVariable*.

39201728

2010-08-13T17:19:18

Make definition of WIN32_LEAN_AND_MEAN in event.h conditional. (Inclusion order might trigger a warning for redefinition on Windows.)

60433a0a

2010-08-13T17:08:59

Clean up syntax on TAILQ_ENTRY() usage Though the C standards allow it, it's apparently possible to get MSVC upset by saying "struct { int field; } (declarator);" instead of "struct {int field; } declarator;", so let's just not do that. Bugfix for 3044492 (commit msg by nickm)

a4af9be1

2010-08-13T11:41:37

Make include/event2/event-config.h not included in source dist As a generated file, it shouldn't get included in our source distribution. Apparently there is an automake incant for this: nobase_ even stacks with nodist_ .

4022b287

2010-08-13T11:34:39

Change include order in Makefile.nmake If there is an event-config.h in include/event2 (either because we screwed up packaging like in 2.0.6-rc or because we previously tried building with mingw and we didn't make distclean in the middle), we want MSVC to find the one one in WIN32-Code/include/event2 first. Found by Gilad Benjamini.

4991669b

2010-08-10T21:05:52

Update the whatsnew-2.0.txt document

33200e72

2010-08-10T15:03:14

Document that DNS_NO_SEARCH is an obsolete alias for DNS_QUERY_NO_SEARCH

3808168d

2010-08-10T15:02:50

Completely remove the (mostly-removed) obsolete thread functions.

9b5c19ac

2010-08-10T11:46:00

Merge branch 'more_pkgconfig'

ebcb1f03

2010-08-10T11:11:25

Add pkgconfig files for libevent_{openssl,pthreads}

0bffe43a

2010-08-09T12:08:40

Fix a nasty dangling-event bug when using rate-limiting groups When we freed a bufferevent that was in a rate-limiting group and blocked on IO, the process of freeing it caused it to get removed from the group. But removing the bufferevent from the group made its limits get removed, which could make it get un-suspended and in turn cause its events to get re-added. Since we would then immediately _free_ the events, this would result in dangling pointers. Fixes bug 3041007.

ff481a8e

2010-08-06T23:22:01

Increment vesion to 2.0.6-rc-dev

556d0f40

2010-08-06T21:19:44

Increment versions to 2.0.6-rc

f6582640

2010-08-06T21:11:01

Changlog and new acknowledgements for 2.0.6-rc

ec347b92

2010-07-07T16:45:03

Move event-config.h to include/event2 This change means that all required include files are in event2, and all files not in event2/* are optional.

5fa30d2b

2010-08-06T17:13:27

Fix evdns build with -DUNICODE It turns out that GetProcAddress always takes its second argument as a C string, regardless of whether unicode is on or not.

28f31a4f

2010-08-06T16:36:23

Fix unit tests with -DUSE_DEBUG enabled If you were to enable USE_DEBUG and slog through all 700+ MB of debugging output, you'd find that one of the unit tests failed, since it tested the debug logging code, but the string it expected and the string it logged differed by a tab vs 2 spaces.

ea1ea3d6

2010-08-06T13:34:51

Make test-changelist count cpu usage right on win32

9b602096

2010-08-06T13:03:17

Use AF_INET socketpair to test sendfile on Solaris

57b30cd7

2010-08-06T13:01:32

Turn our socketpair() replacement into its own function This patch splits the formerly windows-only case of evutil_socketpair() into an (internal-use-only) function named evutil_ersatz_socketpair(), and makes it build and work right on non-Windows hosts. We need this for convenience to test sendfile on solaris, where socketpair can't give you an AF_INET pair, and sendfile() won't work on AF_UNIX.

de1f5d6d

2010-08-06T09:33:55

Remember to initialize timeout events for bufferevent_async Fixes segfaults in unit tests.

170ffd2b

2010-08-05T15:57:30

Increase the tolerance in our unit tests for sloppy clocks. (Apparently openbsd in virtualbox just doesn't keep very accurate time.)

e996b3d4

2010-08-05T15:51:16

Make tests quieter on local dns resolver failure

4558ce23

2010-08-04T17:31:28

Merge branch 'openssl_ratelim'

b2c6202d

2010-08-04T15:52:32

Fix an assertion bug in test-ratelim If the rate limit was low enough, then the echo_conns wouldn't finish inside the 300 msec we allowed for them to close. Instead, count the number of connections we have, and keep waiting until they are all closed.

42f6b624

2010-08-04T15:51:14

Add option to test-ratelim to test min_share

6ae53d67

2010-08-04T15:44:08

Add an interface to expose min_share in ratelimiting groups

009f3005

2010-08-04T14:54:38

Fix rate-limit calculation on openssl bufferevents. When you're doing rate limiting on an openssl connection, you nearly always want to limit the number of bytes sent and received over the wire, not the number of bytes read or written over the secure transport.

13b912e4

2010-08-01T21:39:48

Fixed compilation of sample/le-proxy.c on win32

42a8c711

2010-07-31T17:10:04

Build more cleanly with NetBSDs that dislike toupper(char) To be fair, when char can be signed, if toupper doesn't take negative characters, toupper(char) is a very bad idea. So let's just use the nice safe EVUTIL_TOUPPER instead. (It explicitly only upcases ASCII, but we only use it for identifiers that we know to be ASCII anyway).

840a72fb

2010-07-29T20:32:40

Fix badly-behaved subtest of dns/bufferevent_connect_hostname The bufferevent_connect_hostname test was specifying AF_INET, but the gethostbyname test we were using to see what error to expect was using PF_UNSPEC, leading to possible divergence of results.

9e725f72

2010-07-29T18:12:05

Improve error message for failed epoll to make debugging easier.

cc2379d2

2010-07-26T14:48:32

Constify a couple of arguments to evdns_server_request_add_*_reply

e1c1167c

2010-07-22T14:38:08

Replace (unused,always 0) is_tcp argument to evdns_add_server_port*() with flags Since we weren't using it for anything, and we always failed if it was set, we're allowed to change the future semantics of setting it.

7e87a599

2010-07-21T14:33:42

Stop asserting when asked for a (unsupported) TCP dns port. Just return NULL.

c9913174

2010-07-19T15:31:19

Change bufferevent_openssl::do_write so it doesn't call SSL_write with a 0 length buffer I was running into a problem when using bufferevent_openssl with a very simple echo server. My server simply bufferevent_read_buffer 'd data into an evbuffer and then passed that evbuffer straight to bufferevent_write_buffer. The problem was every now and again the write would fail for no apparent reason. I tracked it down to SSL_write being called with the amount of data to send being 0. This patch alters do_write in bufferevent_openssl so that it skips io_vecs with 0 length.

7c926916

2010-07-19T15:14:10

Make test/test.sh call test-changelist Eventually test-changelist should expand to try more cases, maybe query the status of the actual changelist somehow, and integrate it with the rest of the unit tests. Also, add test-changelist to gitignore.

cb927a51

2010-07-19T15:03:43

Fix whitespace.

b00d4c04

2010-07-19T14:58:08

Remove unused variables in test/test-changelist.c

cf249e7d

2010-07-19T13:44:56

Possible fix to 100% cpu usage with epoll and openssl I'm running a fairly simple bit of test code using libevent2 with epoll and openssl bufferevents and I've run into a 100% cpu usage problem. Looking into it 100% usage was caused by epoll_wait constantly returning write events on the openssl socket when it shouldn't really have been looking for write events at all (N_ACTIVE_CALLBACKS() was returning 0 also). Looking a bit deeper eventbuffer_openssl socket seems to be requesting that the EV_WRITE event be removed when it should, but the event isn't actually being removed from epoll. Continuing to follow this I think I've found a bug in event_changelist_del. For evpoll event_del calls event_changelist_del which caches the change which is then actioned later when evpoll_dispatch is called. In event_changlist_del there is a check so that if the currently changed action is an add then the cached action is changed to a no-op rather than a delete (which makes sense). The problem arises if there are more than two add or delete operations between calls to dispatch, in this case it's possible that the delete is turned into a no-op when it shouldn't have been. For example starting with the event on, a delete followed by an add and then another delete results in a no-op when it should have been a delete (I added a fair bit of debug output that seems to confirm this behaviour). I've applied a small change that checks the original old_event stored with the change and only converts the delete to a no-op if the event isn't on in old_event. This seems to have fixed my problem.

928b7d49

2010-07-16T12:38:23

Merge branch 'pncalls_fix'

65abdc20

2010-07-16T09:11:09

Fix wrong sie calculation of iovec buffers when exact=1 The old code had a bug where the 'exact' flag to 1 in _evbuffer_read_setup_vecs would never actually make the iov_len field of the last iovec get truncated. This patch fixes that.

7c2dea16

2010-07-13T11:06:08

Pass flags to fcntl(F_SETFL) and fcntl(F_SETFD) as int, not long Everybody but Linux documents this as taking an int, and Linux is very tolerant of getting an int instead. If it weren't, everybody doing fcntl(fd,F_SETFL,O_NONBLOCK) would break, since the glibc headers define O_NONBLOCK as an int literal.

kc3-lang/libevent

Log