Azat Khuzhin [Thu, 10 Sep 2015 09:39:20 +0000 (12:39 +0300)]
test: fix bufferevent/bufferevent_pair_release_lock for freebsd
On FreeBSD with kqueue there is a call to evthread_debug_lock_mark_unlocked()
during event_base_free(), that will fail with an assert because of unmatched
"held_by", fix this by reseting lock callbacks to NULL before
event_base_free().
Trace:
bufferevent/bufferevent_pair_release_lock: [warn] Trying to disable lock functions after they have been set up will probaby not work.
[warn] Trying to disable lock functions after they have been set up will probaby not work.
FAIL libevent/test/regress_bufferevent.c:259: lock: lock error[err] libevent/evthread.c:277: Assertion lock->held_by == me failed in evthread_debug_lock_mark_unlocked
[New Thread 802006400 (LWP 100070/regress)]
Program received signal SIGABRT, Aborted.
[Switching to Thread 802006400 (LWP 100070/regress)]
0x000000080167d6ca in thr_kill () from /lib/libc.so.7
(gdb) bt
#0 0x000000080167d6ca in thr_kill () from /lib/libc.so.7
#1 0x0000000801752149 in abort () from /lib/libc.so.7
#2 0x00000000004dff44 in event_exit (errcode=-559030611) at libevent/log.c:105
#3 0x00000000004e053c in event_errx (eval=-559030611, fmt=0x5182cc "%s:%d: Assertion %s failed in %s") at libevent/log.c:162
#4 0x00000000004d9954 in evthread_debug_lock_mark_unlocked (mode=0, lock=0x802017060) at libevent/evthread.c:277
#5 0x00000000004d909a in debug_lock_unlock (mode=0, lock_=0x802017060) at libevent/evthread.c:290
#6 0x00000000004e132c in evsig_dealloc_ (base=0x80201e300) at libevent/signal.c:434
#7 0x00000000004e36c1 in kq_dealloc (base=0x80201e300) at libevent/kqueue.c:435
#8 0x00000000004c9a44 in event_base_free_ (base=0x80201e300, run_finalizers=1) at libevent/event.c:855
#9 0x00000000004c931a in event_base_free (base=0x0) at libevent/event.c:887
#10 0x0000000000452657 in lock_unlock_free_thread_cbs () at libevent/test/regress_bufferevent.c:279
#11 0x0000000000452621 in free_lock_unlock_profiler (data=0x8020170a0) at libevent/test/regress_bufferevent.c:317
#12 0x000000000044bc8f in test_bufferevent_pair_release_lock (arg=0x8020170a0) at libevent/test/regress_bufferevent.c:334
#13 0x00000000004b2288 in testcase_run_bare_ (testcase=0x737660) at libevent/test/tinytest.c:105
#14 0x00000000004b1e72 in testcase_run_one (group=0x738c90, testcase=0x737660) at libevent/test/tinytest.c:252
#15 0x00000000004b2930 in tinytest_main (c=3, v=0x7fffffffead0, groups=0x738c20) at libevent/test/tinytest.c:434
#16 0x00000000004982fe in main (argc=3, argv=0x7fffffffead0) at libevent/test/regress_main.c:459
(gdb) f 4
#4 0x00000000004d9954 in evthread_debug_lock_mark_unlocked (mode=0, lock=0x802017060) at libevent/evthread.c:277
277 EVUTIL_ASSERT(lock->held_by == me);
Current language: auto; currently minimal
(gdb) p lock
$1 = (struct debug_lock *) 0x802017060
(gdb) p lock->held_by
$2 = 0
(gdb) p me
$3 = 34393318400
Azat Khuzhin [Thu, 10 Sep 2015 09:17:39 +0000 (12:17 +0300)]
Merge branch 'cloudabi-fixes-pr-278'
Rebased version of #278
* cloudabi-fixes-pr-278:
Assume that ke_udata is an integer type on CloudABI.
Add missing include of <netinet/in.h>.
Include <sys/ioctl.h>, <sys/resource.h> and <sys/wait.h> optionally.
Ed Schouten [Wed, 26 Aug 2015 08:06:09 +0000 (10:06 +0200)]
Include <sys/ioctl.h>, <sys/resource.h> and <sys/wait.h> optionally.
Though CloudABI implements a very large part of POSIX, it does not
provide these header files, for the reason that there is no raw device
access, no resource limiting and no access to the global process table
through wait().
It looks like these header files are not actually needed in theory.
There don't seem to be any constructs in these source files that use
these features, but I suspect they might still be required on some
systems.
Sebastian Hahn [Thu, 13 Aug 2015 13:04:14 +0000 (15:04 +0200)]
Add a prototype for event_disable_debug_mode()
Exporting this function seems rather useless, as debugging mode has to
be enabled early and calling this function doesn't allow it to get
toggled back on later.
Azat Khuzhin [Thu, 10 Sep 2015 08:18:20 +0000 (11:18 +0300)]
test/regress_be: bufferevent_enable() shouldn't call eventcb by it's own
It must enter the event loop regardless BEV_OPT_DEFER_CALLBACKS, to avoid
potential errors with subsequent connect(), you will find more info in #43,
since this is a regression for it.
Azat Khuzhin [Wed, 9 Sep 2015 15:08:02 +0000 (18:08 +0300)]
Merge branch 'http-server-EOF-at-read-v3'
This patchset fixes an issue with detecting EOF in http server and add some
regression tests to make sure that we will not break anything.
* http-server-EOF-at-read-v3:
http: fix detecting EOF without write
test/regress_http: cover write during read
test/regress_http: verify that closecb will be called without multiple write
Azat Khuzhin [Tue, 8 Sep 2015 13:40:55 +0000 (16:40 +0300)]
http: fix detecting EOF without write
Before this patch http server don't knows when client disconnected until it
will try to write to it, IOW to detect is client still alive you need to write
something to client socket, however it is not convenient since it requires to
store all clients somewhere and poll them periodically, and I don't see any
regressions if we will leave EV_READ always (like libevhtp do), since we
already reset read callback in evhttp_write_buffer() (see
http/write_during_read).
Also since we don't disable EV_READ anymore we don't need some enable EV_READ,
so we will reduce number of epoll_ctl() calls.
Azat Khuzhin [Wed, 9 Sep 2015 14:45:44 +0000 (17:45 +0300)]
test/regress_http: cover write during read
This is the regression for evhttp_write_buffer() where we reset readcb to avoid
illegal state:
http/write_during_read: [err] evhttp_read_cb: illegal connection state 7
Azat Khuzhin [Wed, 2 Sep 2015 08:55:14 +0000 (11:55 +0300)]
test/regress: fix bufferevent_pair_release_lock with EVENT_DEBUG_MODE
Before this patch you will see next error:
$ EVENT_DEBUG_MODE= regress --no-fork bufferevent/bufferevent_pair_release_lock
bufferevent/bufferevent_pair_release_lock: [err] evthread initialization must be called BEFORE anything else!
Azat Khuzhin [Wed, 2 Sep 2015 16:34:15 +0000 (19:34 +0300)]
Merge branch 'be-openssl-drop-fd_is_set-v4'
This must fix some issues with bufferevent openssl layer + some cleanups.
* be-openssl-drop-fd_is_set-v4:
be_openssl: don't use *_auto() in do_handshake() we can't have fd == -1 there
be_openssl: don't call set_open_callbacks() if fd == -1
be_openssl: introduce be_openssl_auto_fd() helper
be_openssl: introduce set_open_callbacks_auto()
be_openssl: get rid off hackish "fd_is_set", to fix some corner cases
Azat Khuzhin [Wed, 2 Sep 2015 09:40:26 +0000 (12:40 +0300)]
be_openssl: introduce set_open_callbacks_auto()
This will split cases when we need to extract fd (cases when we have fd==-1
passed to set_open_callbacks()), and cases when we mustn't have to do this --
SET_FD via be_openssl_ctrl().
Azat Khuzhin [Sat, 22 Aug 2015 18:38:18 +0000 (21:38 +0300)]
be_openssl: get rid off hackish "fd_is_set", to fix some corner cases
This patch is a cleanup and a bug fix, it drops ```fd_is_set``` flag, and
replace it with some checks to event_initialized(), and now we will not call
event_assign() on already added event, plus we will delete event when we really
have to (this patch fixes the case when server is down, IOW before this patch
we will not call event_del() because ```fd_is_set``` was reset to 0) and this
will fix some issues with retries in http layer for ssl.
Azat Khuzhin [Wed, 2 Sep 2015 16:18:08 +0000 (19:18 +0300)]
Merge branch 'test-regress-be-openssl-v2'
This patchset is a bunch of regression tests for bufferevent openssl layer,
some of them already show some bugs, that bugs will be fixed in the next
patches.
* test-regress-be-openssl-v2:
test/regress_ssl: check events fd/pending after timeout triggered
test/regress_ssl: cover case when server didn't up (failed with timeout)
test/regress_ssl: covert that we can't change fd with underlying
test/regress_ssl: cover that events (read/write) at finish not pending
test/regress_ssl: cover fd manipulations
test/regress_ssl: convert open_ssl_bufevs() to mask
test/regress_ssl: convert client/server to mask too
test/regress_ssl: cover "allow_dirty_shutdown"
test/regress_ssl: convert regress_bufferevent_openssl() to bitmask
Azat Khuzhin [Wed, 2 Sep 2015 14:49:41 +0000 (17:49 +0300)]
test/regress_ssl: check events fd/pending after timeout triggered
In this case client can't connect to server, and this bring to the front some
bugs with assigning on already added events (because of ```fd_is_set``` stuff),
for more info see #258, since this is the reproducible for it.
Ed Schouten [Tue, 25 Aug 2015 13:24:39 +0000 (15:24 +0200)]
Don't use BSD u_* types.
These types are not part of POSIX. As we only use them in a small number
of places, we'd better replace them by C standard types. This makes a
larger part of the code build for CloudABI.
Ed Schouten [Tue, 25 Aug 2015 13:32:25 +0000 (15:32 +0200)]
Remove BSD-ism: TIMEVAL_TO_TIMESPEC().
Systems like CloudABI implement kqueue() but do not provide the
BSD-specific TIMEVAL_TO_TIMESPEC() macro. Change the code to perform
this manually, as it is not hard to do this conversion.
Azat Khuzhin [Wed, 19 Aug 2015 19:52:22 +0000 (22:52 +0300)]
be: fix sockaddr_in6 type definition for win32
bufferevent-internal.h:217:23: error: field 'in6' has incomplete type
struct sockaddr_in6 in6;
^
CI: https://ci.appveyor.com/project/nmathewson/libevent/build/2.1.5.108
Azat Khuzhin [Tue, 18 Aug 2015 18:19:11 +0000 (21:19 +0300)]
Merge branch 'reuse-conn_address-on-retry-v11'
There is regression tests and also this code worked/tested during crawling a
huge number of pages (billions).
* reuse-conn_address-on-retry-v11:
be_sock: bufferevent_socket_set_conn_address(): assert instead of silent no-op
http: reuse connected address only with EVHTTP_CON_REUSE_CONNECTED_ADDR
be_sock: sanity check in bufferevent_socket_set_conn_address()
be: replace sockaddr_storage with sockaddr_in6 for conn_address
be: we don't need to use getpeername() we we have conn_address
be: replace conn_address by full struct instead of pointer
test/http: cover retrying with saved conn_address by shutting down dns server
http: use IP address that we got before (if any) during retrying
bufferevent: move conn_address out from http into bufferevent
be: make @sa const for bufferevent_socket_connect()
util: make @sa const for evutil_socket_connect_()
Azat Khuzhin [Tue, 2 Dec 2014 11:24:19 +0000 (14:24 +0300)]
be: replace sockaddr_storage with sockaddr_in6 for conn_address
We need only ipv6/ipv4 for connect in bufferevent, and since
sockaddr_storage is pretty big (128 bytes) it will be better to use
sockaddr_in6 here (it will fit ipv4 too).
Azat Khuzhin [Wed, 12 Nov 2014 17:23:46 +0000 (20:23 +0300)]
http: use IP address that we got before (if any) during retrying
Before this patch every time we are retrying our request we resolve
domain, but we could optimize this (since UDP is slow) by using cached
conn_address value, so do this.
Azat Khuzhin [Sat, 15 Nov 2014 18:46:11 +0000 (21:46 +0300)]
bufferevent: move conn_address out from http into bufferevent
In http the only case when when we could store it is when we already
connected, *but* if we are doing request using domain name, then we need
to do request to nameserver to get IP address, and this is handled by
bufferevent.
So when we have IP address (from nameserver) and don't have connection
to this IP address, we could already cache it to avoid extra DNS
requests (since UDP is slow), and we can't do this from http layer, only
from bufferevent.
Mark Ellzey [Fri, 15 May 2015 09:58:14 +0000 (02:58 -0700)]
Debug mode option to error on evthread init AFTER other event calls.
- A handy event_enable_debug_mode() feature which will error and abort the
application if any thread-aware libevent functions are called BEFORE the
evthread API has been initialized (manually, or through
evthread_use_windows_threads() / evthread_use_pthreads()
- This is done by setting the global debug variable
'event_debug_created_threadable_ctx_' whenever the following functions
are called:
evthreadimpl_lock_alloc_()
evthreadimpl_cond_alloc_()
event_base_new_with_config() <- this checks to see if the thread
callbacks are enabled first, so we
have to manually set the variable.
Mark Ellzey [Mon, 11 May 2015 16:47:40 +0000 (12:47 -0400)]
Fix the link for appveyor OpenSSL installer (WIN32)
* change and move the openssl self installer off to a site we own
- the old link was 404, probably due to being replaced with a newer
non-vuln version. But since we are only using this installer to
auto-build with on appveyor (not as a release), then having a file
we don't change and own seems to be a better solution.
* reduce verbosity
Mark Ellzey [Mon, 11 May 2015 16:06:01 +0000 (12:06 -0400)]
Fix garbage value in socketpair util function, stdint?
* Fixed an issue with evutil_ersatz_socketpair_, listen_addr could all
be compared against with agarbage values. So just memset it before
using it anywhere.
* Nick might punch me in the face, but if we have stdint.h; (as in
EVENT__HAVE_STDINT_H is defined), might as well use those instead of
the manual [U]INT[X}_MAX/MIN muck in there now.
Mark Ellzey [Tue, 5 May 2015 17:37:00 +0000 (10:37 -0700)]
Call underlying bev ctrl SET_FD on filtered bufferevents
If a bufferevent_filter is set on an underlying bufferevent which has
ctrl functions, bufferevent_filter needs to handle this.
For now I have added just BEV_CTRL_SET_FD, since this is needed for
bufferevent_sock to assign file descriptors to the proper
bufferevent_read/write callbacks.
A good example of the problem can be found in issue #237
https://github.com/libevent/libevent/issues/237