|
066775e3
|
2012-02-28T13:09:51
|
|
Fix evmap to work with HT-backed IO maps again.
|
|
|
ca47fa0b
|
2012-02-28T12:53:46
|
|
evutil_accept4() should take ev_socklen_t
|
|
|
d7d9f755
|
2012-02-27T18:10:45
|
|
test-fdleak: Avoid perror, since it does not capture windows socket errors
|
|
|
510839e7
|
2012-02-27T18:08:27
|
|
remove debugging prints
|
|
|
d9be5f49
|
2012-02-27T18:06:28
|
|
test-fdleak: Call event callbacks event_cb.
|
|
|
077c7e94
|
2012-02-27T18:05:49
|
|
Remove fixed port from test-fdleak, so two instances can run in parallel
|
|
|
c9ba53a3
|
2012-02-27T17:57:01
|
|
Remove an unsafe programming practice from test-fdleak; add XX comments
It's a bad idea to have a read callback read a fixed amount; Instead,
we should drain it, or have some way to know that more data is
arrived. Not unsafe in this case, but let's not encourage people to
learn bad habits.
This commit also notes some other stuff to fix.
|
|
4293a809
|
2012-02-25T07:47:49
|
|
Try and fix compilation and running of test-fdleak on Windows.
|
|
|
1c4288f4
|
2012-02-25T07:36:49
|
|
Allow compilation on machines that do not have setrlimit.
|
|
|
f7af1941
|
2012-02-25T07:21:20
|
|
Simply test-fdleak by running server & client sockets in the same event loop.
|
|
|
2ef92786
|
2012-02-23T21:40:02
|
|
Add a new test: test-fdleak which tests for fd leaks by creating many sockets.
This test opens a server socket, and forks a child which connects to that
server socket many times. It sets a low number for the max open file limit
to catch any file descriptor leaks.
It would not work on Windows since it uses fork() to be able to create both the
server and the clients.
|
|
|
74d32dd4
|
2012-02-21T20:39:17
|
|
If accept4 is absent, fall back to accept.
|
|
|
a6492cb7
|
2012-02-21T20:38:36
|
|
Avoid leaking fd in evutil_socket() by Linux.
Found by Amarin Phaosawasdi
|
|
|
a3cec909
|
2012-02-20T14:07:08
|
|
Remove the last vestiges of _EVENT_USE_EVENTLIST
|
|
|
8dbcf8d6
|
2012-02-20T13:56:13
|
|
Merge branch '21_eventlist_v3_squashed'
|
|
|
cad57535
|
2012-02-11T21:05:50
|
|
Add doxygen for event_base_dump_events
|
|
|
c89b4e63
|
2012-02-11T21:01:53
|
|
Refactor the functions that run over every event.
Now there are appropriate "for each event", "for each fd", and "for
each signal" helpers that they can use; this makes the code a bit
simpler, and far less duplicated.
This lets me turn back on the functions I disabled when removing
eventlist.
Additionally, check more lists for circularity in
event_base_assert_ok().
Add typedefs for the callback types.
Name fewer things "ctx".
Adds an implementation of Floyd's tortoise-and-hare algorithm to check
for circularity in TAILQs and LISTs, to avoid the abuse of flags that
event_base_assert_ok() was doing before.
Suggested by Dave Hart.
|
|
|
65bc91c2
|
2012-02-17T10:42:56
|
|
Fix an "unused return value" warning on write() in signal handler
|
|
|
819e98ac
|
2012-02-16T01:21:42
|
|
Start work on a 2.1 changelog and whats-new
|
|
|
a270728e
|
2012-02-15T21:27:52
|
|
Add another caveat to the TCP_DEFER_ACCEPT documentation
|
|
5880e4a1
|
2012-02-14T18:04:52
|
|
Support TCP_DEFER_ACCEPT sockopts for listeners
A listening socket can be enabled with the sockopt
TCP_DEFER_ACCEPT. This informs the kernel to not call the user-land
accept() until real data has been written to the socket.
A new flag LEV_OPT_DEFERRED_ACCEPT has been introduced to
automatically set this option. Optionally
evutil_make_tcp_listen_socket_deferred() can be called manually.
(Tweaked slightly by nickm.)
|
|
|
a220a081
|
2012-02-15T21:07:44
|
|
Merge remote-tracking branch 'github/21_fast_syscalls'
|
|
|
a63ed161
|
2012-02-15T20:26:52
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
bec50680
|
2012-02-15T20:12:32
|
|
Stop crashing in evdns when nameserver probes give a weird error
When a nameserver is down, we periodically try sending a "probe"
message to that nameserver to see if it has come back up. If a
nameserver comes up, we cancel any pending probe messages.
Cancelling a probe message while handling the probe's response would
result in a access-after-free or a double-free, so when we notice that
we're about to call a nameserver up because of having received a probe
from it, we need to check whether current response is the response
from the probe.
There was a case where we didn't to that, though: when the resolver
gave us an unusual error response to our request that it resolve
google.com. This is pretty rare, but apparently it can happen with
some weird cacheing nameservers -- the one on the mikrotik router, for
example. Without this patch, we would crash with a NULL pointer
derefernce.
Thanks to Hannes Sowa for finding this issue and helping me track it
down.
|
|
|
a37cbfda
|
2012-02-14T15:44:03
|
|
Merge remote-tracking branch 'origin/patches-2.0'
Conflicts:
configure.in
|
|
|
2d67b638
|
2012-02-14T15:37:58
|
|
Changed OPENSSL_LDFLAGS to OPENSSL_LIBADD
|
|
|
92781968
|
2012-02-14T15:01:02
|
|
Added OPENSSL_LDFLAGS env variable which is appended to SSL checks.
If openssl is not installed system-wide or not compiled as a shared library,
some systems require various link flags (e.g., -ld).
|
|
|
2ed44302
|
2012-02-14T11:48:55
|
|
If open(O_CLOEXEC) fails, fall back to fcntl(CLOEXEC)
This is needed for folks who build with recent Linux kernel headers
but run with older kernels.
|
|
|
f088d0c5
|
2012-02-13T18:34:00
|
|
Fix up usage of reserved identifiers in ht-internal.h
|
|
|
4900c5f9
|
2012-02-13T18:04:05
|
|
Port changes from Tor for ht-internal.h
|
|
|
46e5bb7b
|
2012-02-13T17:59:14
|
|
Invert the sense and the name of HT_CACHE_HASH_VALUES
This is meant to make it easier to merge ht-internal.h back upstream
into Tor.
|
|
|
6c81be74
|
2012-02-13T17:49:17
|
|
Synchronize with upstream tinytest
|
|
|
a9dc063c
|
2012-02-13T17:30:46
|
|
Merge in a punctuation fix from upstream tinytest
|
|
|
bbea8d6b
|
2012-02-13T12:19:08
|
|
Move libevent 1.x headers to include/, to put all public headers in one place.
|
|
|
27931976
|
2012-02-11T21:20:47
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
03dce42d
|
2012-02-11T21:17:18
|
|
Tweak the evutil_open_closeonexec patch to work on windows, old unixes.
Windows doesn't have a mode_t as far as I can tell.
Some unixes, iirc, don't like three-argument open without O_CREAT.
|
|
|
d2b5f722
|
2012-02-11T17:23:17
|
|
Make uses of open() close-on-exec safe by introducing evutil_open_closeonexec.
In a multi-process/threaded environment, opening fds internally
without the close-on-exec flag could leak fds to child processes.
|
|
|
a6503944
|
2012-02-11T12:04:15
|
|
Add a bufferevent_getcb() to find a bufferevent's current callbacks
|
|
|
bf2c5a77
|
2012-02-10T23:56:49
|
|
Also make win32select.c conditional for IDE users
|
|
|
87a7cded
|
2012-02-10T23:52:27
|
|
Merge branch 'ifdef' of git://github.com/rosslagerwall/libevent
|
|
|
de69eb7b
|
2012-02-10T23:51:11
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
0c483170
|
2012-02-10T22:20:05
|
|
Add an empty section to the changelog for 2.0.18-stable
|
|
|
c5b6e464
|
2012-02-10T22:18:19
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
5a807b7a
|
2012-02-10T22:17:59
|
|
Correct a name in the credits. oops
|
|
|
8fd7a243
|
2012-02-10T22:16:46
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
2d7bf0f7
|
2012-02-10T22:16:14
|
|
Bump version to 2.0.17-stable-dev
|
|
|
2d915460
|
2012-02-10T18:50:17
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
83e58ccd
|
2012-02-10T18:39:31
|
|
Update the credits in the readme
|
|
|
0cdbd642
|
2012-02-10T18:36:10
|
|
Finalize the changelog
|
|
|
2b0a2c45
|
2012-02-10T18:04:04
|
|
Fix compilation on osx
|
|
|
539466e5
|
2012-02-10T17:33:50
|
|
Merge remote-tracking branch 'origin/patches-2.0'
Conflicts:
Makefile.am
WIN32-Code/event2/event-config.h
configure.in
|
|
|
e49e2891
|
2012-02-10T17:29:53
|
|
Update copyright notices to 2012
|
|
|
8152b4c2
|
2012-02-10T16:49:05
|
|
Bump version to 2.0.17-stable
|
|
|
bac906c7
|
2012-02-10T16:39:46
|
|
Prefer epoll_create1 on Linuxen that have it
|
|
|
33fca629
|
2012-02-10T16:33:25
|
|
Save some syscalls when constructing a socket for a bufferevent
|
|
|
713e570a
|
2012-02-10T16:32:50
|
|
Save some syscalls when creating evdns sockets
|
|
|
af6c9d8e
|
2012-02-10T16:32:32
|
|
Save syscalls when constructing listener sockets for evhttp
|
|
|
a35f396f
|
2012-02-10T16:25:53
|
|
Use pipes for telling signals to main thread when possible
|
|
|
ca76cd93
|
2012-02-10T16:15:10
|
|
Use a wrapper function to create the notification pipe/socketpair/eventfd
|
|
|
4970329a
|
2012-02-10T16:14:09
|
|
The LEV_OPT_CLOSE_ON_EXEC flag now applies to accepted listener sockets too
|
|
|
7e9e2893
|
2012-02-10T16:13:30
|
|
Minimize syscalls during socket creation in listener.c
|
|
|
a1c042bf
|
2012-02-10T15:55:15
|
|
Infrastructure for using faster/fewer syscalls when creating sockets
Linux provides some features that allow avoiding extra calls to
fcntl when creating new nonblocking/close-on-exec sockets, so
we can add wrapper functions to emulate those when they are not
available.
Additionally, even when we are emulating those functions, we can
take a fast path that cuts our fcntl calls in half: we don't need to
look up the previous value of a file's flags when we have just
created it.
|
|
|
e465623a
|
2012-02-10T12:05:39
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
5d7bfa15
|
2012-02-10T11:24:51
|
|
In the kqueue backend, do not report EBADF as an EV_READ
We were doing this because of (correct) reports that NetBSD gives an
EBADF when you try to add the write side of a pipe for which the
read side has been closed. But on most kqueue platforms, that
doesn't happen, and on *all* kqueue platforms, reporting a
nonexistent fd (which we usually have if we have seen EBADF) as
readable tends to give programs a case of the vapors.
Nicholas Marriott wrote the original patch here; I did the comment
fixes.
|
|
|
df19ba6a
|
2012-02-09T15:16:02
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
19715a60
|
2012-02-09T15:07:19
|
|
Avoid crash when freeing event_iocp and using event_set_mem_functions
There was a calloc that needed to be an mm_calloc.
Reported by "fffvvvzz" on sourceforge. Ticket 3486114
|
|
|
76d4c929
|
2012-02-08T18:46:00
|
|
Put #ifdef around some files to support alternate build systems.
|
|
|
6f6cebe3
|
2012-02-06T21:18:25
|
|
Bring the changelog up to date
|
|
|
c00416f7
|
2012-02-06T12:39:09
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
5b4b8126
|
2012-02-06T12:24:49
|
|
Loop on filtering SSL reads until we are blocked or exhausted.
This is not a perfect fix, but it's much much better than the
current buggy behavior, which could lead to filtering SSL
connections that just stopped reading.
Based on ideas by Maseeb Abdul Qadir and Mark Ellzey.
|
|
|
4dee4cc7
|
2012-02-02T11:45:23
|
|
Replace more C99/C++ comments with oldschool /* */ comments
|
|
|
7c46d4aa
|
2012-02-02T11:41:37
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
d84d9170
|
2011-11-21T09:41:55
|
|
Use C-style comments in C source files (for compatibility with compilers such as xlc on AIX).
|
|
|
032aac15
|
2012-02-01T14:57:44
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
d6094b16
|
2012-02-01T14:56:43
|
|
evdns: fix a bug in circular-queue implementation
found by Wang Qin
|
|
e4a56ed5
|
2012-01-31T18:58:59
|
|
Fix a compile warning in event_reinit
Introduced in 272033efe575a9dc7ec6f123a96afba5c69ff1c6
|
|
|
604569bf
|
2012-01-27T16:35:04
|
|
Remove the eventqueue list and the ev_next pointers.
Those pointers were once used to maintain a complete list of
inserted IO and signal events. But such a list is now available by
walking over ev_io_map and ev_signal_map! So all they did was
require extra pointer operations to maintain, and extra 8-16 bytes
of storage in each struct event.
To be cowardly and keep the option of going back to having this
redundancy, I'm wrapping the removed code in a set of ifdefs.
This is a first cut; it needs cleanups and stress-testing!! In
particular, it just plain disables a couple of functions that could
probably be saved.
There seems to be a need for an evmap_{io,signal}_foreach() or something.
|
|
|
7622d265
|
2012-01-27T15:10:28
|
|
Make test-changelist faster
|
|
|
39b3f38d
|
2012-01-27T14:39:18
|
|
Check changelist as part of checking representational integrity
|
|
|
2c4b5de1
|
2012-01-27T14:30:41
|
|
Restore fast-path event_reinit() for slower backends
We used to use the needs_reinit flag in struct eventop to indicate
whether an event backend had shared state across a fork(), and
therefore would require us to construct a new event backend. But
when we realized that the signal notification fds and the thread
notification fds would always be shared across forks, we stopped
looking at it.
This patch restores the old behavior so that poll, select, and
win32select don't need to do a linear scan over all pending
fds/signals when they do a reinit. Their life is hard enough
already.
|
|
|
272033ef
|
2012-01-27T13:54:05
|
|
Make event_reinit() more robust and maintainable
Previously, event_reinit required a bunch of really dubious hacks,
and violated a lot of abstraction barriers to mess around with lists
of internal events and "pretend" to re-add them.
The new (and fairly well commented!) implementation tries to be much
smarter, by isolating the changes as much as possible to the backend
state, and minimizing the amount of abstraction violations.
Specifically, we now use event_del() to remove events we want to
remove, rather than futzing around with queues in event_reinit().
To avoid bogus calls to evsel->del(), we temporarily replace evsel
with a null-object stub.
Also, we now push the responsibility for calling evsel->add() down
into the evmap code, so that we don't actually need to unlink and
re-link all of our events.
|
|
|
4b7baec6
|
2012-01-26T16:56:33
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
9f9e259f
|
2012-01-26T15:04:24
|
|
Better workaround for Linux 3.2 edge-triggered epoll bug
On further investigation, it appears that this problem is limited to
AF_UNIX sockets, so let's just do the test on AF_INET sockets.
|
|
|
2c6b3246
|
2012-01-24T17:18:52
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
c94a5f2a
|
2012-01-24T17:15:50
|
|
Do a memberwise comparison of threading function tables
Doing a memcmp risks comparing uninitialized padding bytes at the
end of the structure.
|
|
|
95e2455c
|
2012-01-24T16:08:00
|
|
When including an -internal.h header outside the main tree, do so early
Some of our unit tests and sample code need functions and structures
defined in an -internal.h header. But that can freak out OpenSolaris,
where stdio.h wants to define _FILE_OFFSET_BITS unless it's already
defined, and then evconfig-internal.h defines it. Regular users
should never ever use our -internal.h headers, so the solution is
to make sure that if we're going to use them ourselves, we do so
before system headers.
|
|
|
b9f7e5fd
|
2012-01-24T15:58:11
|
|
Allow more slop in deferred_cb_skew test; freebsd needs it
|
|
|
ea30a435
|
2012-01-24T15:30:51
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
3aa0e8a9
|
2012-01-24T15:30:31
|
|
Remove a couple of now-unused variables
|
|
|
438d4ff2
|
2012-01-24T15:29:39
|
|
Make event_base integrity check work on windows
|
|
|
edc6d7d7
|
2012-01-24T15:09:04
|
|
Start writing a changelog for 2.0.17-stable
|
|
|
b77b43fc
|
2012-01-24T14:59:37
|
|
Use test_timeval_diff_eq more consistently
|
|
|
67a1763b
|
2012-01-24T14:33:10
|
|
Make regression tests run over 3x faster.
This was mainly a matter of reducing timeouts and delays, paying
special attention to every test that took longer than a second to
finish.
We could do better here; IMO anything over .7 sec is probably too
long, but it's a big win as it is.
Remember, interactive computing is a big win over batch processing:
anything that makes you get up and walk away from the terminal might
as well be making you carry your punch cards over to the mainframe.
|
|
|
350a3c40
|
2012-01-24T14:34:04
|
|
New evhttp function to adjust initial retry timeout
|
|
|
ab14f7c1
|
2012-01-24T11:45:00
|
|
Reduce the timeout in the main/fork test.
There was no reason for it to be so long, except for the lack of a usleep
|
|
|
f25d9d32
|
2012-01-24T11:42:26
|
|
Add an (internal) usleep function for use by unit tests
|
|
|
18653fca
|
2012-01-24T11:17:06
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
5408ff3b
|
2012-01-24T11:16:26
|
|
Oops:remove an accidentally committed "sleep(1)" in a unit test
|
|
|
10996a1a
|
2012-01-24T11:04:56
|
|
Merge remote-tracking branch 'origin/patches-2.0'
|
|
|
dab91877
|
2012-01-24T11:04:19
|
|
Workaround in the unit tests for an apparent epoll bug in Linux 3.2
|