GitHub/libevent
1
0
Fork 0
mirror of https://github.com/libevent/libevent.git synced 2025-01-09 00:56:20 +08:00
Code Issues Packages Projects Releases Wiki Activity
libevent/test/Makefile.am

82 lines
2.5 KiB
Makefile
Raw Normal View History

Do not inhibit automake dependencies generation It fixes make parallel builds.
2010-04-12 12:50:17 +02:00
AUTOMAKE_OPTIONS = foreign
automake-ify test dir, too svn:r44
2003-03-08 05:24:26 +00:00
Correct an AM_CFLAGS to an AM_CPPFLAGS in test/Makefile.am
2011-02-13 02:27:42 -05:00
AM_CPPFLAGS = -I$(top_srcdir) -I$(top_srcdir)/compat -I$(top_srcdir)/include -I../include -DTINYTEST_LOCAL
automake-ify test dir, too svn:r44
2003-03-08 05:24:26 +00:00
Remove redundant stuff from EXTRA_DIST To a first approximation, sources that are mentioned anywhere in an automake file don't need to get mentioned in EXTRA_DIST.
2010-03-12 13:22:47 -05:00
EXTRA_DIST = regress.rpc regress.gen.h regress.gen.c test.sh
including the tagging code that is required by event_rpcgen.py; test the new functionality. svn:r172
2005-08-22 01:34:34 +00:00
Use AM_CONDITIONAL, not AC_SUBST, to tell automake to build regress For some reason, this makes mingw much happier about building regress.exe by default. No, I'm afraid I don't know why.
2011-04-23 02:35:17 -04:00
noinst_PROGRAMS = test-init test-eof test-weof test-time \
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.
2010-07-19 13:44:56 +12:00
bench bench_cascade bench_http bench_httpclient test-ratelim \
test-changelist
Use AM_CONDITIONAL, not AC_SUBST, to tell automake to build regress For some reason, this makes mingw much happier about building regress.exe by default. No, I'm afraid I don't know why.
2011-04-23 02:35:17 -04:00
if BUILD_REGRESS
noinst_PROGRAMS += regress
endif
Add configure options to disable installation, regression tests The main reason for disabling installation is if you're building libevent as a subpackage for embedding: you want to have your main package's "make all" build libevent, but you don't want your main package's "make install" to install libevent.
2011-02-13 02:05:04 -05:00
EXTRA_PROGRAMS = regress
Bump to the latest version of tinytest This lets us do without libevent-specific code in tinytest.c, and lets us add a feature to skip individual tests from the command line.
2010-09-09 16:13:09 -04:00
noinst_HEADERS = tinytest.h tinytest_macros.h regress.h tinytest_local.h
automake-ify test dir, too svn:r44
2003-03-08 05:24:26 +00:00
Support the standard 'make check' target in place of 'make verify' Based on patch 2816088 from Zack Weinberg
2010-03-12 13:09:28 -05:00
TESTS = $(top_srcdir)/test/test.sh
Add configure options to disable installation, regression tests The main reason for disabling installation is if you're building libevent as a subpackage for embedding: you want to have your main package's "make all" build libevent, but you don't want your main package's "make install" to install libevent.
2011-02-13 02:05:04 -05:00
BUILT_SOURCES =
if BUILD_REGRESS
BUILT_SOURCES += regress.gen.c regress.gen.h
endif
including the tagging code that is required by event_rpcgen.py; test the new functionality. svn:r172
2005-08-22 01:34:34 +00:00
test_init_SOURCES = test-init.c
split libevent into two extra libraries libevent_core and libevent_extra svn:r469
2007-11-03 23:45:38 +00:00
test_init_LDADD = ../libevent_core.la
including the tagging code that is required by event_rpcgen.py; test the new functionality. svn:r172
2005-08-22 01:34:34 +00:00
test_eof_SOURCES = test-eof.c
split libevent into two extra libraries libevent_core and libevent_extra svn:r469
2007-11-03 23:45:38 +00:00
test_eof_LDADD = ../libevent_core.la
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.
2010-07-19 13:44:56 +12:00
test_changelist_SOURCES = test-changelist.c
test_changelist_LDADD = ../libevent_core.la
including the tagging code that is required by event_rpcgen.py; test the new functionality. svn:r172
2005-08-22 01:34:34 +00:00
test_weof_SOURCES = test-weof.c
split libevent into two extra libraries libevent_core and libevent_extra svn:r469
2007-11-03 23:45:38 +00:00
test_weof_LDADD = ../libevent_core.la
including the tagging code that is required by event_rpcgen.py; test the new functionality. svn:r172
2005-08-22 01:34:34 +00:00
test_time_SOURCES = test-time.c
split libevent into two extra libraries libevent_core and libevent_extra svn:r469
2007-11-03 23:45:38 +00:00
test_time_LDADD = ../libevent_core.la
Testing code for bufferevent rate-limiting. This is not part of the regression tests, since running it necessarily takes a while. There is a new test-ratelim test; run it with '-h' for an argument to see its options.
2009-12-23 07:54:13 -05:00
test_ratelim_SOURCES = test-ratelim.c
Fix test-ratelim compilation on Linux. I'd forgotten to include time.h, and to link against libm.
2009-12-30 13:50:52 -05:00
test_ratelim_LDADD = ../libevent_core.la -lm
Refactor unit tests using my spiffy new "tinytest" framework. The big win here is that we can get process-level isolation. This has been tested to work okay on at least Linux and Win32. Only the tests in regress.c have been converted wrapped in the new wrapper functions; the others are still on the old system. svn:r1073
2009-01-29 23:19:57 +00:00
Move evbuffer tests into their own file. svn:r1087
2009-02-01 01:07:12 +00:00
regress_SOURCES = regress.c regress_buffer.c regress_http.c regress_dns.c \
move dns utility functions into a separate file so that we can use them for http testing
2010-01-14 16:53:25 -08:00
regress_testutils.c regress_testutils.h \
make it work with older versions of automake; from Scott Lamb svn:r858
2008-06-22 16:00:48 +00:00
regress_rpc.c regress.gen.c regress.gen.h regress_et.c \
More evconnlistener unit tests: bump its coverage from 71% to 83%. svn:r1392
2009-07-30 17:01:38 +00:00
regress_bufferevent.c regress_listener.c \
Add a randomized test for heap correctness. svn:r1229
2009-04-23 00:33:37 +00:00
regress_util.c tinytest.c regress_main.c regress_minheap.c \
Rename regress_pthread.c to regress_thread.c
2010-08-17 14:44:12 -04:00
$(regress_thread_SOURCES) $(regress_zlib_SOURCES)
New functions to provide sane threading callbacks with pthreads and win32 threading implementations. svn:r1031
2009-01-21 07:51:25 +00:00
if PTHREADS
Rename regress_pthread.c to regress_thread.c
2010-08-17 14:44:12 -04:00
regress_thread_SOURCES = regress_thread.c
Make threading functions global, like the mm_ functions. Use the libevent_pthread.la library in regress_pthread. svn:r1121
2009-02-12 22:19:54 +00:00
PTHREAD_LIBS += ../libevent_pthreads.la
Provide OpenSSL style support for multiple threads accessing the same event_base svn:r684
2008-03-02 21:18:33 +00:00
endif
Make the regress_pthread.c tests work on windows with current test APIs
2010-08-16 12:38:22 -04:00
if BUILD_WIN32
Rename regress_pthread.c to regress_thread.c
2010-08-17 14:44:12 -04:00
regress_thread_SOURCES = regress_thread.c
Make the regress_pthread.c tests work on windows with current test APIs
2010-08-16 12:38:22 -04:00
endif
provide example bufferevent filters doing compression and decompression as additional regression test svn:r751
2008-04-30 04:31:10 +00:00
if ZLIB_REGRESS
make it work with older versions of automake; from Scott Lamb svn:r858
2008-06-22 16:00:48 +00:00
regress_zlib_SOURCES = regress_zlib.c
provide example bufferevent filters doing compression and decompression as additional regression test svn:r751
2008-04-30 04:31:10 +00:00
endif
Oops: actually commit changes to build and use regress_iocp svn:r1255
2009-04-30 19:05:43 +00:00
if BUILD_WIN32
regress_SOURCES += regress_iocp.c
endif
Bufferevent support for openssl. This code adds a new Bufferevent type that is only compiled when the openssl library is present. It supports using an SSL object and an event alert mechanism, which can either be an fd or an underlying bufferevent. There is still more work to do: the unit tests are incomplete, and we need to support flush and shutdown much better. Sometimes events are generated needlessly: this will hose performance. There's a new encrypting proxy in sample/le-proxy.c. This code has only been tested on OSX, and nowhere else. svn:r1382
2009-07-28 04:03:57 +00:00
Use $(), not @@, for LIBEVENT_GC_SECTIONS
2011-02-22 16:13:49 -05:00
regress_LDADD = $(LIBEVENT_GC_SECTIONS) ../libevent.la $(PTHREAD_LIBS) $(ZLIB_LIBS)
Make sure TINYTEST_LOCAL is defined when building tinytest.c
2011-05-26 00:03:38 +02:00
regress_CPPFLAGS = $(AM_CPPFLAGS) $(PTHREAD_CFLAGS) $(ZLIB_CFLAGS)
minor compilation and regresson fixes; from Frank Denis svn:r952
2008-11-27 19:34:07 +00:00
regress_LDFLAGS = $(PTHREAD_CFLAGS)
Refactor unit tests using my spiffy new "tinytest" framework. The big win here is that we can get process-level isolation. This has been tested to work okay on at least Linux and Win32. Only the tests in regress.c have been converted wrapped in the new wrapper functions; the others are still on the old system. svn:r1073
2009-01-29 23:19:57 +00:00
Bufferevent support for openssl. This code adds a new Bufferevent type that is only compiled when the openssl library is present. It supports using an SSL object and an event alert mechanism, which can either be an fd or an underlying bufferevent. There is still more work to do: the unit tests are incomplete, and we need to support flush and shutdown much better. Sometimes events are generated needlessly: this will hose performance. There's a new encrypting proxy in sample/le-proxy.c. This code has only been tested on OSX, and nowhere else. svn:r1382
2009-07-28 04:03:57 +00:00
if OPENSSL
regress_SOURCES += regress_ssl.c
Libtool brings in the dependencies of libevent_openssl.la automatically
2011-05-04 16:37:58 +02:00
regress_LDADD += ../libevent_openssl.la
Bufferevent support for openssl. This code adds a new Bufferevent type that is only compiled when the openssl library is present. It supports using an SSL object and an event alert mechanism, which can either be an fd or an underlying bufferevent. There is still more work to do: the unit tests are incomplete, and we need to support flush and shutdown much better. Sometimes events are generated needlessly: this will hose performance. There's a new encrypting proxy in sample/le-proxy.c. This code has only been tested on OSX, and nowhere else. svn:r1382
2009-07-28 04:03:57 +00:00
endif
including the tagging code that is required by event_rpcgen.py; test the new functionality. svn:r172
2005-08-22 01:34:34 +00:00
bench_SOURCES = bench.c
Use $(), not @@, for LIBEVENT_GC_SECTIONS
2011-02-22 16:13:49 -05:00
bench_LDADD = $(LIBEVENT_GC_SECTIONS) ../libevent.la
add another benchmark that tests cascading events svn:r566
2007-11-30 02:21:33 +00:00
bench_cascade_SOURCES = bench_cascade.c
Use $(), not @@, for LIBEVENT_GC_SECTIONS
2011-02-22 16:13:49 -05:00
bench_cascade_LDADD = $(LIBEVENT_GC_SECTIONS) ../libevent.la
add a simple http server for potential benchmarking svn:r823
2008-05-15 01:54:20 +00:00
bench_http_SOURCES = bench_http.c
Use $(), not @@, for LIBEVENT_GC_SECTIONS
2011-02-22 16:13:49 -05:00
bench_http_LDADD = $(LIBEVENT_GC_SECTIONS) ../libevent.la
Add a new bench_httpclient for a trivial codecon demo. svn:r1198
2009-04-18 00:12:52 +00:00
bench_httpclient_SOURCES = bench_httpclient.c
Use $(), not @@, for LIBEVENT_GC_SECTIONS
2011-02-22 16:13:49 -05:00
bench_httpclient_LDADD = $(LIBEVENT_GC_SECTIONS) ../libevent_core.la
including the tagging code that is required by event_rpcgen.py; test the new functionality. svn:r172
2005-08-22 01:34:34 +00:00
r15764@tombo: nickm | 2007-12-31 15:46:16 -0500 Forward-port from 1.4 branch: Make generated rpc files build-depend on event_rpcgen. svn:r626
2007-12-31 20:47:12 +00:00
regress.gen.c regress.gen.h: regress.rpc $(top_srcdir)/event_rpcgen.py
r15094@catbus: nickm | 2007-09-18 10:52:11 -0400 Patch from Christopher Lane: Make regress.gen.* buildable from outside source directory. svn:r434
2007-09-18 15:11:53 +00:00
$(top_srcdir)/event_rpcgen.py $(srcdir)/regress.rpc || echo "No Python installed"
automake-ify test dir, too svn:r44
2003-03-08 05:24:26 +00:00
DISTCLEANFILES = *~
Support the standard 'make check' target in place of 'make verify' Based on patch 2816088 from Zack Weinberg
2010-03-12 13:09:28 -05:00
verify: check
automake-ify test dir, too svn:r44
2003-03-08 05:24:26 +00:00
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.
2010-07-19 13:44:56 +12:00
bench test-init test-eof test-weof test-time test-changelist: ../libevent.la
Reference in New Issue Copy Permalink