2012-12-08 01:50:32 +04:00
|
|
|
/*
|
|
|
|
* Copyright © 2012 Intel Corporation
|
|
|
|
*
|
2015-06-12 01:39:40 +03:00
|
|
|
* Permission is hereby granted, free of charge, to any person obtaining
|
|
|
|
* a copy of this software and associated documentation files (the
|
|
|
|
* "Software"), to deal in the Software without restriction, including
|
|
|
|
* without limitation the rights to use, copy, modify, merge, publish,
|
|
|
|
* distribute, sublicense, and/or sell copies of the Software, and to
|
|
|
|
* permit persons to whom the Software is furnished to do so, subject to
|
|
|
|
* the following conditions:
|
2012-12-08 01:50:32 +04:00
|
|
|
*
|
2015-06-12 01:39:40 +03:00
|
|
|
* The above copyright notice and this permission notice (including the
|
|
|
|
* next paragraph) shall be included in all copies or substantial
|
|
|
|
* portions of the Software.
|
|
|
|
*
|
|
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
|
|
|
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
|
|
|
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
|
|
|
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
|
|
|
|
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
|
|
|
|
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
|
|
|
|
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
|
|
* SOFTWARE.
|
2012-12-08 01:50:32 +04:00
|
|
|
*/
|
|
|
|
|
2013-09-11 10:08:04 +04:00
|
|
|
#include "config.h"
|
2014-11-20 04:18:33 +03:00
|
|
|
|
2012-12-08 01:50:32 +04:00
|
|
|
#include <unistd.h>
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <sys/wait.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <assert.h>
|
|
|
|
#include <errno.h>
|
2012-12-18 19:30:16 +04:00
|
|
|
#include <signal.h>
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
#include <getopt.h>
|
CI: Never unload llvmpipe DSO whilst testing
This commit is truly horrible.
We want to run ASan with leak checking enabled in CI so we can catch
memory leaks before they're introduced. This works well with Pixman, and
with NIR-only drivers like iris or Panfrost.
But when we run under llvmpipe - which we do under CI - we start failing
because:
- Mesa pulls in llvmpipe via dlopen
- llvmpipe pulls in LLVM itself via DT_NEEDED
- initialising LLVM's global type/etc systems performs thread-local
allocations
- llvmpipe can't free those allocations since the application might
also be using LLVM
- Weston stops using GL and destroys all GL objects, leading to Mesa
unloading llvmpipe like it should
- with everything disappearing from the process's vmap, ASan can no
longer keep track of still-reachable pointers
- tests fail because LLVM is 'leaking'
Usually, an alternative is to LD_PRELOAD a shim which overrides
dlclose() to be a no-op. This is not usable here, because when
$LD_PRELOAD is not empty and ASan is not first in it, ASan immediately
errors out. Prepending ASan doesn't work, because we run our tests
through Meson (which also invokes Ninja), leading to LSan exploding
over CPython and Ninja, which is not what we're interested in.
It would be possible to inject _both_ ASan and a dlclose-does-nothing
shim DSO into the LD_PRELOAD environment for every test, but that seems
even worse, especially as Meson strongly discourages globbing for random
files in the root.
So, here we are, doing what we can: finding where swrast_dri.so (aka
llvmpipe) lives, stashing that in an environment variable, and
deliberately leaking a dlopen handle which we never close to ensure that
neither llvmpipe nor LLVM leave our process's address space before we
exit.
Signed-off-by: Daniel Stone <daniels@collabora.com>
2022-06-25 06:59:52 +03:00
|
|
|
#include <dlfcn.h>
|
2014-11-20 04:18:33 +03:00
|
|
|
|
2021-03-03 15:21:01 +03:00
|
|
|
#include "test-config.h"
|
2012-12-08 01:50:32 +04:00
|
|
|
#include "weston-test-runner.h"
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
#include "weston-testsuite-data.h"
|
|
|
|
#include "shared/string-helpers.h"
|
2012-12-08 01:50:32 +04:00
|
|
|
|
CI: Never unload llvmpipe DSO whilst testing
This commit is truly horrible.
We want to run ASan with leak checking enabled in CI so we can catch
memory leaks before they're introduced. This works well with Pixman, and
with NIR-only drivers like iris or Panfrost.
But when we run under llvmpipe - which we do under CI - we start failing
because:
- Mesa pulls in llvmpipe via dlopen
- llvmpipe pulls in LLVM itself via DT_NEEDED
- initialising LLVM's global type/etc systems performs thread-local
allocations
- llvmpipe can't free those allocations since the application might
also be using LLVM
- Weston stops using GL and destroys all GL objects, leading to Mesa
unloading llvmpipe like it should
- with everything disappearing from the process's vmap, ASan can no
longer keep track of still-reachable pointers
- tests fail because LLVM is 'leaking'
Usually, an alternative is to LD_PRELOAD a shim which overrides
dlclose() to be a no-op. This is not usable here, because when
$LD_PRELOAD is not empty and ASan is not first in it, ASan immediately
errors out. Prepending ASan doesn't work, because we run our tests
through Meson (which also invokes Ninja), leading to LSan exploding
over CPython and Ninja, which is not what we're interested in.
It would be possible to inject _both_ ASan and a dlclose-does-nothing
shim DSO into the LD_PRELOAD environment for every test, but that seems
even worse, especially as Meson strongly discourages globbing for random
files in the root.
So, here we are, doing what we can: finding where swrast_dri.so (aka
llvmpipe) lives, stashing that in an environment variable, and
deliberately leaking a dlopen handle which we never close to ensure that
neither llvmpipe nor LLVM leave our process's address space before we
exit.
Signed-off-by: Daniel Stone <daniels@collabora.com>
2022-06-25 06:59:52 +03:00
|
|
|
/* This is a glibc extension; if we can't use it then make it harmless. */
|
|
|
|
#ifndef RTLD_NODELETE
|
|
|
|
#define RTLD_NODELETE 0
|
|
|
|
#endif
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
/**
|
|
|
|
* \defgroup testharness Test harness
|
|
|
|
* \defgroup testharness_private Test harness private
|
|
|
|
*/
|
2014-11-20 02:06:20 +03:00
|
|
|
|
2019-11-07 14:34:59 +03:00
|
|
|
extern const struct weston_test_entry __start_test_section, __stop_test_section;
|
2012-12-08 01:50:32 +04:00
|
|
|
|
2020-01-22 14:37:20 +03:00
|
|
|
struct weston_test_run_info {
|
|
|
|
char name[512];
|
|
|
|
int fixture_nr;
|
|
|
|
};
|
|
|
|
|
|
|
|
static const struct weston_test_run_info *test_run_info_;
|
2017-01-27 19:30:26 +03:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
/** Get the test name string with counter
|
|
|
|
*
|
2020-02-26 14:04:48 +03:00
|
|
|
* \return The test name with fixture number \c -f%%d added. For an array
|
|
|
|
* driven test, e.g. defined with TEST_P(), the name has also a \c -e%%d
|
|
|
|
* suffix to indicate the array element number.
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
*
|
|
|
|
* This is only usable from code paths inside TEST(), TEST_P(), PLUGIN_TEST()
|
|
|
|
* etc. defined functions.
|
|
|
|
*
|
|
|
|
* \ingroup testharness
|
|
|
|
*/
|
2017-01-27 19:30:26 +03:00
|
|
|
const char *
|
|
|
|
get_test_name(void)
|
|
|
|
{
|
2020-01-22 14:37:20 +03:00
|
|
|
return test_run_info_->name;
|
|
|
|
}
|
|
|
|
|
|
|
|
/** Get the current fixture index
|
|
|
|
*
|
|
|
|
* Returns the current fixture index which can be used directly as an index
|
|
|
|
* into the array passed as an argument to DECLARE_FIXTURE_SETUP_WITH_ARG().
|
|
|
|
*
|
|
|
|
* This is only usable from code paths inside TEST(), TEST_P(), PLUGIN_TEST()
|
|
|
|
* etc. defined functions.
|
|
|
|
*
|
|
|
|
* \ingroup testharness
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
get_test_fixture_index(void)
|
|
|
|
{
|
|
|
|
return test_run_info_->fixture_nr - 1;
|
2017-01-27 19:30:26 +03:00
|
|
|
}
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
/** Print into test log
|
|
|
|
*
|
|
|
|
* This is exactly like printf() except the output goes to the test log,
|
|
|
|
* which is at stderr.
|
|
|
|
*
|
|
|
|
* \param fmt printf format string
|
|
|
|
*
|
|
|
|
* \ingroup testharness
|
|
|
|
*/
|
2019-11-06 16:59:33 +03:00
|
|
|
void
|
|
|
|
testlog(const char *fmt, ...)
|
|
|
|
{
|
|
|
|
va_list argp;
|
|
|
|
|
|
|
|
va_start(argp, fmt);
|
|
|
|
vfprintf(stderr, fmt, argp);
|
|
|
|
va_end(argp);
|
|
|
|
}
|
|
|
|
|
2021-02-15 15:20:57 +03:00
|
|
|
static const void *
|
|
|
|
fixture_setup_array_get_arg(const struct fixture_setup_array *fsa, int findex)
|
|
|
|
{
|
|
|
|
const char *array_data = fsa->array;
|
|
|
|
|
|
|
|
if (array_data)
|
|
|
|
return array_data + findex * fsa->element_size;
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2021-02-15 15:43:14 +03:00
|
|
|
static const char *
|
|
|
|
fixture_setup_array_get_name(const struct fixture_setup_array *fsa, int findex)
|
|
|
|
{
|
|
|
|
const char *element_data = fixture_setup_array_get_arg(fsa, findex);
|
|
|
|
const struct fixture_metadata *meta;
|
|
|
|
|
|
|
|
if (!element_data)
|
|
|
|
return "";
|
|
|
|
|
|
|
|
meta = (const void *)(element_data + fsa->meta_offset);
|
|
|
|
|
|
|
|
return meta->name;
|
|
|
|
}
|
|
|
|
|
2019-11-07 14:34:59 +03:00
|
|
|
static const struct weston_test_entry *
|
2012-12-08 01:50:32 +04:00
|
|
|
find_test(const char *name)
|
|
|
|
{
|
2019-11-07 14:34:59 +03:00
|
|
|
const struct weston_test_entry *t;
|
2012-12-08 01:50:32 +04:00
|
|
|
|
|
|
|
for (t = &__start_test_section; t < &__stop_test_section; t++)
|
|
|
|
if (strcmp(t->name, name) == 0)
|
|
|
|
return t;
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
static enum test_result_code
|
2023-06-28 12:49:48 +03:00
|
|
|
run_test(struct wet_testsuite_data *suite_data, int fixture_nr,
|
|
|
|
const struct weston_test_entry *t, void *data,
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
int iteration)
|
2012-12-08 01:50:32 +04:00
|
|
|
{
|
2020-01-22 14:37:20 +03:00
|
|
|
struct weston_test_run_info info;
|
2017-01-27 19:30:26 +03:00
|
|
|
|
|
|
|
if (data) {
|
2020-02-26 14:04:48 +03:00
|
|
|
snprintf(info.name, sizeof(info.name), "%s-f%02d-e%02d",
|
|
|
|
t->name, fixture_nr, iteration);
|
2017-01-27 19:30:26 +03:00
|
|
|
} else {
|
2020-02-26 14:04:48 +03:00
|
|
|
snprintf(info.name, sizeof(info.name), "%s-f%02d",
|
|
|
|
t->name, fixture_nr);
|
2017-01-27 19:30:26 +03:00
|
|
|
}
|
2020-01-22 14:37:20 +03:00
|
|
|
info.fixture_nr = fixture_nr;
|
2017-01-27 19:30:26 +03:00
|
|
|
|
2020-01-22 14:37:20 +03:00
|
|
|
test_run_info_ = &info;
|
2023-06-28 12:49:48 +03:00
|
|
|
t->run(suite_data, data);
|
2020-01-22 14:37:20 +03:00
|
|
|
test_run_info_ = NULL;
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* XXX: We should return t->run(data); but that requires changing
|
|
|
|
* the function signature and stop using assert() in tests.
|
|
|
|
* https://gitlab.freedesktop.org/wayland/weston/issues/311
|
|
|
|
*/
|
|
|
|
return RESULT_OK;
|
2012-12-08 01:50:32 +04:00
|
|
|
}
|
|
|
|
|
2013-09-11 10:08:30 +04:00
|
|
|
static void
|
|
|
|
list_tests(void)
|
|
|
|
{
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
const struct fixture_setup_array *fsa;
|
2019-11-07 14:34:59 +03:00
|
|
|
const struct weston_test_entry *t;
|
2021-02-15 17:14:44 +03:00
|
|
|
int i;
|
2013-09-11 10:08:30 +04:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
fsa = fixture_setup_array_get_();
|
|
|
|
|
2021-02-15 17:14:44 +03:00
|
|
|
if (fsa->n_elements > 1) {
|
|
|
|
printf("Fixture setups:\n");
|
|
|
|
for (i = 0; i < fsa->n_elements; i++) {
|
|
|
|
printf("%2d: %s\n", i + 1,
|
|
|
|
fixture_setup_array_get_name(fsa, i));
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
printf("One fixture setup.\n");
|
|
|
|
}
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
|
2021-02-15 17:14:44 +03:00
|
|
|
printf("Test names:\n");
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
for (t = &__start_test_section; t < &__stop_test_section; t++) {
|
|
|
|
printf(" %s\n", t->name);
|
|
|
|
if (t->n_elements > 1)
|
|
|
|
printf(" with array of %d cases\n", t->n_elements);
|
|
|
|
}
|
2013-09-11 10:08:30 +04:00
|
|
|
}
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
struct weston_test_harness {
|
|
|
|
int32_t fixt_ind;
|
|
|
|
char *chosen_testname;
|
|
|
|
int32_t case_ind;
|
2013-09-13 06:01:20 +04:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
struct wet_testsuite_data data;
|
|
|
|
};
|
2013-09-13 06:01:20 +04:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
typedef void (*weston_test_cb)(struct wet_testsuite_data *suite_data,
|
|
|
|
const struct weston_test_entry *t,
|
|
|
|
const void *test_data,
|
|
|
|
int iteration);
|
|
|
|
|
|
|
|
static void
|
|
|
|
for_each_test_case(struct wet_testsuite_data *data, weston_test_cb cb)
|
|
|
|
{
|
|
|
|
unsigned i;
|
|
|
|
|
|
|
|
for (i = 0; i < data->tests_count; i++) {
|
|
|
|
const struct weston_test_entry *t = &data->tests[i];
|
|
|
|
const void *current_test_data = t->table_data;
|
|
|
|
int elem;
|
|
|
|
int elem_end;
|
|
|
|
|
|
|
|
if (data->case_index == -1) {
|
|
|
|
elem = 0;
|
|
|
|
elem_end = t->n_elements;
|
|
|
|
} else {
|
|
|
|
elem = data->case_index;
|
|
|
|
elem_end = elem + 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (; elem < elem_end; elem++) {
|
|
|
|
current_test_data = (char *)t->table_data +
|
|
|
|
elem * t->element_size;
|
|
|
|
cb(data, t, current_test_data, elem);
|
|
|
|
}
|
2019-10-31 18:20:17 +03:00
|
|
|
}
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
}
|
2013-09-13 06:01:20 +04:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
static const char *
|
|
|
|
result_to_str(enum test_result_code ret)
|
|
|
|
{
|
|
|
|
static const char *names[] = {
|
|
|
|
[RESULT_FAIL] = "fail",
|
|
|
|
[RESULT_HARD_ERROR] = "hard error",
|
|
|
|
[RESULT_OK] = "ok",
|
2021-03-03 15:21:01 +03:00
|
|
|
#if WESTON_TEST_SKIP_IS_FAILURE
|
|
|
|
[RESULT_SKIP] = "skip error",
|
|
|
|
#else
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
[RESULT_SKIP] = "skip",
|
2021-03-03 15:21:01 +03:00
|
|
|
#endif
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
};
|
|
|
|
|
|
|
|
assert(ret >= 0 && ret < ARRAY_LENGTH(names));
|
|
|
|
return names[ret];
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
run_case(struct wet_testsuite_data *suite_data,
|
|
|
|
const struct weston_test_entry *t,
|
|
|
|
const void *test_data,
|
|
|
|
int iteration)
|
|
|
|
{
|
|
|
|
enum test_result_code ret;
|
|
|
|
const char *fail = "";
|
|
|
|
const char *skip = "";
|
|
|
|
int fixture_nr = suite_data->fixture_iteration + 1;
|
|
|
|
int iteration_nr = iteration + 1;
|
|
|
|
|
2021-02-15 15:43:14 +03:00
|
|
|
testlog("*** Run %s %s/%d\n",
|
|
|
|
suite_data->fixture_name, t->name, iteration_nr);
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
|
|
|
|
if (suite_data->type == TEST_TYPE_PLUGIN) {
|
2023-06-28 12:49:48 +03:00
|
|
|
ret = run_test(suite_data, fixture_nr, t, suite_data->compositor,
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
iteration);
|
|
|
|
} else {
|
2023-06-28 12:49:48 +03:00
|
|
|
ret = run_test(suite_data, fixture_nr, t, (void *)test_data,
|
|
|
|
iteration);
|
2013-09-13 06:01:20 +04:00
|
|
|
}
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
switch (ret) {
|
|
|
|
case RESULT_OK:
|
|
|
|
suite_data->passed++;
|
|
|
|
break;
|
|
|
|
case RESULT_FAIL:
|
|
|
|
case RESULT_HARD_ERROR:
|
|
|
|
suite_data->failed++;
|
|
|
|
fail = "not ";
|
2013-09-13 06:01:20 +04:00
|
|
|
break;
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
case RESULT_SKIP:
|
|
|
|
suite_data->skipped++;
|
|
|
|
skip = " # SKIP";
|
2021-03-03 15:21:01 +03:00
|
|
|
#if WESTON_TEST_SKIP_IS_FAILURE
|
|
|
|
fail = "not ";
|
|
|
|
#endif
|
2013-09-13 06:01:20 +04:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2021-02-15 15:43:14 +03:00
|
|
|
testlog("*** Result %s %s/%d: %s\n",
|
|
|
|
suite_data->fixture_name, t->name, iteration_nr,
|
|
|
|
result_to_str(ret));
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
|
|
|
|
suite_data->counter++;
|
2021-02-15 15:43:14 +03:00
|
|
|
printf("%sok %d %s %s/%d%s\n", fail, suite_data->counter,
|
|
|
|
suite_data->fixture_name, t->name, iteration_nr, skip);
|
2013-09-13 06:01:20 +04:00
|
|
|
}
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
/* This function might run in a new thread */
|
|
|
|
static void
|
|
|
|
testsuite_run(struct wet_testsuite_data *data)
|
|
|
|
{
|
|
|
|
for_each_test_case(data, run_case);
|
|
|
|
}
|
2013-09-13 06:01:20 +04:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
static void
|
|
|
|
count_case(struct wet_testsuite_data *suite_data,
|
|
|
|
const struct weston_test_entry *t,
|
|
|
|
const void *test_data,
|
|
|
|
int iteration)
|
|
|
|
{
|
|
|
|
suite_data->total++;
|
2013-09-13 06:01:20 +04:00
|
|
|
}
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
static void
|
|
|
|
tap_plan(struct wet_testsuite_data *data, int count_fixtures)
|
2012-12-08 01:50:32 +04:00
|
|
|
{
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
data->total = 0;
|
|
|
|
for_each_test_case(data, count_case);
|
|
|
|
|
|
|
|
printf("1..%d\n", data->total * count_fixtures);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
skip_case(struct wet_testsuite_data *suite_data,
|
|
|
|
const struct weston_test_entry *t,
|
|
|
|
const void *test_data,
|
|
|
|
int iteration)
|
|
|
|
{
|
2021-03-03 15:21:01 +03:00
|
|
|
const char *skip_error = "";
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
int iteration_nr = iteration + 1;
|
|
|
|
|
2021-03-03 15:21:01 +03:00
|
|
|
#if WESTON_TEST_SKIP_IS_FAILURE
|
|
|
|
skip_error = "not ";
|
|
|
|
#endif
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
suite_data->counter++;
|
2021-03-03 15:21:01 +03:00
|
|
|
printf("%sok %d %s %s/%d # SKIP fixture\n",
|
|
|
|
skip_error, suite_data->counter,
|
2021-02-15 15:43:14 +03:00
|
|
|
suite_data->fixture_name, t->name, iteration_nr);
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
tap_skip_fixture(struct wet_testsuite_data *data)
|
|
|
|
{
|
|
|
|
for_each_test_case(data, skip_case);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
help(const char *exe)
|
|
|
|
{
|
|
|
|
printf(
|
2021-02-17 14:25:01 +03:00
|
|
|
"Usage: %s [options] [testname [number]]\n"
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
"\n"
|
|
|
|
"This is a Weston test suite executable that runs some tests.\n"
|
|
|
|
"Options:\n"
|
2021-02-17 14:25:01 +03:00
|
|
|
" -f, --fixture N Run only fixture number N. 0 runs all (default).\n"
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
" -h, --help Print this help and exit with success.\n"
|
|
|
|
" -l, --list List all tests in this executable and exit with success.\n"
|
|
|
|
"testname: Optional; name of the test to execute instead of all tests.\n"
|
2021-02-17 14:25:01 +03:00
|
|
|
"number: Optional; for a multi-case test, run the given case only.\n"
|
|
|
|
"Both fixture and case numbering starts from 1.\n",
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
exe);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
parse_command_line(struct weston_test_harness *harness, int argc, char **argv)
|
|
|
|
{
|
|
|
|
int c;
|
|
|
|
static const struct option opts[] = {
|
|
|
|
{ "fixture", required_argument, NULL, 'f' },
|
|
|
|
{ "help", no_argument, NULL, 'h' },
|
|
|
|
{ "list", no_argument, NULL, 'l' },
|
|
|
|
{ 0, 0, NULL, 0 }
|
|
|
|
};
|
|
|
|
|
|
|
|
while ((c = getopt_long(argc, argv, "f:hl", opts, NULL)) != -1) {
|
|
|
|
switch (c) {
|
|
|
|
case 'f':
|
|
|
|
if (!safe_strtoint(optarg, &harness->fixt_ind)) {
|
|
|
|
fprintf(stderr,
|
|
|
|
"Error: '%s' does not look like a number (command line).\n",
|
|
|
|
optarg);
|
|
|
|
exit(RESULT_HARD_ERROR);
|
|
|
|
}
|
|
|
|
harness->fixt_ind--; /* convert base-1 to base 0 */
|
|
|
|
break;
|
|
|
|
case 'h':
|
|
|
|
help(argv[0]);
|
|
|
|
exit(RESULT_OK);
|
|
|
|
case 'l':
|
2013-09-11 10:08:30 +04:00
|
|
|
list_tests();
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
exit(RESULT_OK);
|
|
|
|
case 0:
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
exit(RESULT_HARD_ERROR);
|
2013-09-11 10:08:30 +04:00
|
|
|
}
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
if (optind < argc)
|
|
|
|
harness->chosen_testname = argv[optind++];
|
2013-09-11 10:08:30 +04:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
if (optind < argc) {
|
|
|
|
if (!safe_strtoint(argv[optind], &harness->case_ind)) {
|
|
|
|
fprintf(stderr,
|
|
|
|
"Error: '%s' does not look like a number (command line).\n",
|
|
|
|
argv[optind]);
|
|
|
|
exit(RESULT_HARD_ERROR);
|
2014-11-20 02:06:20 +03:00
|
|
|
}
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
harness->case_ind--; /* convert base-1 to base 0 */
|
|
|
|
optind++;
|
|
|
|
}
|
2014-11-20 02:06:20 +03:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
if (optind < argc) {
|
|
|
|
fprintf(stderr, "Unexpected extra arguments given (command line).\n\n");
|
|
|
|
help(argv[0]);
|
|
|
|
exit(RESULT_HARD_ERROR);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct weston_test_harness *
|
|
|
|
weston_test_harness_create(int argc, char **argv)
|
|
|
|
{
|
|
|
|
const struct fixture_setup_array *fsa;
|
|
|
|
struct weston_test_harness *harness;
|
|
|
|
|
|
|
|
harness = zalloc(sizeof(*harness));
|
|
|
|
assert(harness);
|
|
|
|
|
|
|
|
harness->fixt_ind = -1;
|
|
|
|
harness->case_ind = -1;
|
|
|
|
parse_command_line(harness, argc, argv);
|
|
|
|
|
|
|
|
fsa = fixture_setup_array_get_();
|
|
|
|
if (harness->fixt_ind < -1 || harness->fixt_ind >= fsa->n_elements) {
|
|
|
|
fprintf(stderr,
|
|
|
|
"Error: fixture index %d (command line) is invalid for this program.\n",
|
|
|
|
harness->fixt_ind + 1);
|
|
|
|
exit(RESULT_HARD_ERROR);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (harness->chosen_testname) {
|
|
|
|
const struct weston_test_entry *t;
|
|
|
|
|
|
|
|
t = find_test(harness->chosen_testname);
|
|
|
|
if (!t) {
|
|
|
|
fprintf(stderr,
|
|
|
|
"Error: test '%s' not found (command line).\n",
|
|
|
|
harness->chosen_testname);
|
|
|
|
exit(RESULT_HARD_ERROR);
|
2012-12-08 01:50:32 +04:00
|
|
|
}
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
if (harness->case_ind < -1 ||
|
|
|
|
harness->case_ind >= t->n_elements) {
|
|
|
|
fprintf(stderr,
|
|
|
|
"Error: case index %d (command line) is invalid for this test.\n",
|
|
|
|
harness->case_ind + 1);
|
|
|
|
exit(RESULT_HARD_ERROR);
|
2012-12-08 01:50:32 +04:00
|
|
|
}
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
|
|
|
|
harness->data.tests = t;
|
|
|
|
harness->data.tests_count = 1;
|
|
|
|
harness->data.case_index = harness->case_ind;
|
|
|
|
} else {
|
|
|
|
harness->data.tests = &__start_test_section;
|
|
|
|
harness->data.tests_count =
|
|
|
|
&__stop_test_section - &__start_test_section;
|
|
|
|
harness->data.case_index = -1;
|
2012-12-08 01:50:32 +04:00
|
|
|
}
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
harness->data.run = testsuite_run;
|
|
|
|
|
|
|
|
return harness;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
weston_test_harness_destroy(struct weston_test_harness *harness)
|
|
|
|
{
|
|
|
|
free(harness);
|
|
|
|
}
|
|
|
|
|
|
|
|
static enum test_result_code
|
|
|
|
counts_to_result(const struct wet_testsuite_data *data)
|
|
|
|
{
|
2021-03-03 15:21:01 +03:00
|
|
|
#if WESTON_TEST_SKIP_IS_FAILURE
|
|
|
|
if (data->skipped > 0)
|
|
|
|
return RESULT_FAIL;
|
|
|
|
#endif
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
/* RESULT_SKIP is reserved for fixture setup itself skipping everything */
|
|
|
|
if (data->total == data->passed + data->skipped)
|
|
|
|
return RESULT_OK;
|
|
|
|
return RESULT_FAIL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/** Execute all tests as client tests
|
|
|
|
*
|
|
|
|
* \param harness The test harness context.
|
|
|
|
* \param setup The compositor configuration.
|
|
|
|
*
|
|
|
|
* Initializes the compositor with the given setup and executes the compositor.
|
|
|
|
* The compositor creates a new thread where all tests in the test program are
|
|
|
|
* serially executed. Once the thread finishes, the compositor returns from its
|
|
|
|
* event loop and cleans up.
|
|
|
|
*
|
|
|
|
* Returns RESULT_SKIP if the requested compositor features, e.g. GL-renderer,
|
|
|
|
* are not built.
|
|
|
|
*
|
|
|
|
* \sa DECLARE_FIXTURE_SETUP(), DECLARE_FIXTURE_SETUP_WITH_ARG()
|
|
|
|
* \ingroup testharness
|
|
|
|
*/
|
|
|
|
enum test_result_code
|
|
|
|
weston_test_harness_execute_as_client(struct weston_test_harness *harness,
|
|
|
|
const struct compositor_setup *setup)
|
|
|
|
{
|
|
|
|
struct wet_testsuite_data *data = &harness->data;
|
|
|
|
|
|
|
|
data->type = TEST_TYPE_CLIENT;
|
|
|
|
return execute_compositor(setup, data);
|
|
|
|
}
|
|
|
|
|
|
|
|
/** Execute all tests as plugin tests
|
|
|
|
*
|
|
|
|
* \param harness The test harness context.
|
|
|
|
* \param setup The compositor configuration.
|
|
|
|
*
|
|
|
|
* Initializes the compositor with the given setup and executes the compositor.
|
|
|
|
* The compositor executes all tests in the test program serially from an idle
|
|
|
|
* handler, then returns from its event loop and cleans up.
|
|
|
|
*
|
|
|
|
* Returns RESULT_SKIP if the requested compositor features, e.g. GL-renderer,
|
|
|
|
* are not built.
|
|
|
|
*
|
|
|
|
* \sa DECLARE_FIXTURE_SETUP(), DECLARE_FIXTURE_SETUP_WITH_ARG()
|
|
|
|
* \ingroup testharness
|
|
|
|
*/
|
|
|
|
enum test_result_code
|
|
|
|
weston_test_harness_execute_as_plugin(struct weston_test_harness *harness,
|
|
|
|
const struct compositor_setup *setup)
|
|
|
|
{
|
|
|
|
struct wet_testsuite_data *data = &harness->data;
|
|
|
|
|
|
|
|
data->type = TEST_TYPE_PLUGIN;
|
|
|
|
return execute_compositor(setup, data);
|
|
|
|
}
|
|
|
|
|
|
|
|
/** Execute all tests as standalone tests
|
|
|
|
*
|
|
|
|
* \param harness The test harness context.
|
|
|
|
*
|
|
|
|
* Executes all tests in the test program serially without any further setup,
|
|
|
|
* particularly without any compositor instance created.
|
|
|
|
*
|
|
|
|
* \sa DECLARE_FIXTURE_SETUP(), DECLARE_FIXTURE_SETUP_WITH_ARG()
|
|
|
|
* \ingroup testharness
|
|
|
|
*/
|
|
|
|
enum test_result_code
|
|
|
|
weston_test_harness_execute_standalone(struct weston_test_harness *harness)
|
|
|
|
{
|
|
|
|
struct wet_testsuite_data *data = &harness->data;
|
|
|
|
|
|
|
|
data->type = TEST_TYPE_STANDALONE;
|
|
|
|
data->run(data);
|
|
|
|
|
|
|
|
return RESULT_OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
/** Fixture data array getter method
|
|
|
|
*
|
|
|
|
* DECLARE_FIXTURE_SETUP_WITH_ARG() overrides this in test programs.
|
|
|
|
* The default implementation has no data and makes the tests run once.
|
|
|
|
*
|
|
|
|
* \ingroup testharness
|
|
|
|
*/
|
|
|
|
__attribute__((weak)) const struct fixture_setup_array *
|
|
|
|
fixture_setup_array_get_(void)
|
|
|
|
{
|
|
|
|
/* A dummy fixture without a data array. */
|
|
|
|
static const struct fixture_setup_array default_fsa = {
|
|
|
|
.array = NULL,
|
|
|
|
.element_size = 0,
|
|
|
|
.n_elements = 1,
|
2021-02-15 15:43:14 +03:00
|
|
|
.meta_offset = 0,
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
};
|
|
|
|
|
|
|
|
return &default_fsa;
|
|
|
|
}
|
|
|
|
|
|
|
|
/** Fixture setup function
|
|
|
|
*
|
|
|
|
* DECLARE_FIXTURE_SETUP() and DECLARE_FIXTURE_SETUP_WITH_ARG() override
|
|
|
|
* this in test programs.
|
|
|
|
* The default implementation just calls
|
|
|
|
* weston_test_harness_execute_standalone().
|
|
|
|
*
|
|
|
|
* \ingroup testharness
|
|
|
|
*/
|
|
|
|
__attribute__((weak)) enum test_result_code
|
|
|
|
fixture_setup_run_(struct weston_test_harness *harness, const void *arg_)
|
|
|
|
{
|
|
|
|
return weston_test_harness_execute_standalone(harness);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
fixture_report(const struct wet_testsuite_data *d, enum test_result_code ret)
|
|
|
|
{
|
|
|
|
int fixture_nr = d->fixture_iteration + 1;
|
|
|
|
|
2021-02-15 15:43:14 +03:00
|
|
|
testlog("--- Fixture %d (%s) %s: passed %d, skipped %d, failed %d, total %d\n",
|
|
|
|
fixture_nr, d->fixture_name, result_to_str(ret),
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
d->passed, d->skipped, d->failed, d->total);
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
|
|
|
main(int argc, char *argv[])
|
|
|
|
{
|
|
|
|
struct weston_test_harness *harness;
|
|
|
|
enum test_result_code ret;
|
|
|
|
enum test_result_code result = RESULT_OK;
|
|
|
|
const struct fixture_setup_array *fsa;
|
CI: Never unload llvmpipe DSO whilst testing
This commit is truly horrible.
We want to run ASan with leak checking enabled in CI so we can catch
memory leaks before they're introduced. This works well with Pixman, and
with NIR-only drivers like iris or Panfrost.
But when we run under llvmpipe - which we do under CI - we start failing
because:
- Mesa pulls in llvmpipe via dlopen
- llvmpipe pulls in LLVM itself via DT_NEEDED
- initialising LLVM's global type/etc systems performs thread-local
allocations
- llvmpipe can't free those allocations since the application might
also be using LLVM
- Weston stops using GL and destroys all GL objects, leading to Mesa
unloading llvmpipe like it should
- with everything disappearing from the process's vmap, ASan can no
longer keep track of still-reachable pointers
- tests fail because LLVM is 'leaking'
Usually, an alternative is to LD_PRELOAD a shim which overrides
dlclose() to be a no-op. This is not usable here, because when
$LD_PRELOAD is not empty and ASan is not first in it, ASan immediately
errors out. Prepending ASan doesn't work, because we run our tests
through Meson (which also invokes Ninja), leading to LSan exploding
over CPython and Ninja, which is not what we're interested in.
It would be possible to inject _both_ ASan and a dlclose-does-nothing
shim DSO into the LD_PRELOAD environment for every test, but that seems
even worse, especially as Meson strongly discourages globbing for random
files in the root.
So, here we are, doing what we can: finding where swrast_dri.so (aka
llvmpipe) lives, stashing that in an environment variable, and
deliberately leaking a dlopen handle which we never close to ensure that
neither llvmpipe nor LLVM leave our process's address space before we
exit.
Signed-off-by: Daniel Stone <daniels@collabora.com>
2022-06-25 06:59:52 +03:00
|
|
|
const char *leak_dl_handle;
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
int fi;
|
|
|
|
int fi_end;
|
|
|
|
|
CI: Never unload llvmpipe DSO whilst testing
This commit is truly horrible.
We want to run ASan with leak checking enabled in CI so we can catch
memory leaks before they're introduced. This works well with Pixman, and
with NIR-only drivers like iris or Panfrost.
But when we run under llvmpipe - which we do under CI - we start failing
because:
- Mesa pulls in llvmpipe via dlopen
- llvmpipe pulls in LLVM itself via DT_NEEDED
- initialising LLVM's global type/etc systems performs thread-local
allocations
- llvmpipe can't free those allocations since the application might
also be using LLVM
- Weston stops using GL and destroys all GL objects, leading to Mesa
unloading llvmpipe like it should
- with everything disappearing from the process's vmap, ASan can no
longer keep track of still-reachable pointers
- tests fail because LLVM is 'leaking'
Usually, an alternative is to LD_PRELOAD a shim which overrides
dlclose() to be a no-op. This is not usable here, because when
$LD_PRELOAD is not empty and ASan is not first in it, ASan immediately
errors out. Prepending ASan doesn't work, because we run our tests
through Meson (which also invokes Ninja), leading to LSan exploding
over CPython and Ninja, which is not what we're interested in.
It would be possible to inject _both_ ASan and a dlclose-does-nothing
shim DSO into the LD_PRELOAD environment for every test, but that seems
even worse, especially as Meson strongly discourages globbing for random
files in the root.
So, here we are, doing what we can: finding where swrast_dri.so (aka
llvmpipe) lives, stashing that in an environment variable, and
deliberately leaking a dlopen handle which we never close to ensure that
neither llvmpipe nor LLVM leave our process's address space before we
exit.
Signed-off-by: Daniel Stone <daniels@collabora.com>
2022-06-25 06:59:52 +03:00
|
|
|
/* This is horrific, but it gives us working leak checking. If we
|
|
|
|
* actually unload llvmpipe, then we also unload LLVM, and some global
|
|
|
|
* setup it's done - which llvmpipe can't tear down because the actual
|
|
|
|
* client might be using LLVM instead.
|
|
|
|
*
|
|
|
|
* Turns out if llvmpipe is always live, then the pointers are always
|
|
|
|
* reachable, so LeakSanitizer just tells us about our own code rather
|
|
|
|
* than LLVM's, so ...
|
|
|
|
*/
|
|
|
|
leak_dl_handle = getenv("WESTON_CI_LEAK_DL_HANDLE");
|
|
|
|
if (leak_dl_handle)
|
|
|
|
(void) dlopen(leak_dl_handle, RTLD_LAZY | RTLD_GLOBAL | RTLD_NODELETE);
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
harness = weston_test_harness_create(argc, argv);
|
|
|
|
|
|
|
|
fsa = fixture_setup_array_get_();
|
|
|
|
|
|
|
|
if (harness->fixt_ind == -1) {
|
|
|
|
fi = 0;
|
|
|
|
fi_end = fsa->n_elements;
|
|
|
|
} else {
|
|
|
|
fi = harness->fixt_ind;
|
|
|
|
fi_end = fi + 1;
|
|
|
|
}
|
|
|
|
|
2023-01-10 00:38:06 +03:00
|
|
|
printf("TAP version 13\n");
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
tap_plan(&harness->data, fi_end - fi);
|
|
|
|
testlog("Iterating through %d fixtures.\n", fi_end - fi);
|
|
|
|
|
|
|
|
for (; fi < fi_end; fi++) {
|
2021-02-15 15:20:57 +03:00
|
|
|
const void *arg = fixture_setup_array_get_arg(fsa, fi);
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
|
|
|
|
harness->data.fixture_iteration = fi;
|
2021-02-15 15:43:14 +03:00
|
|
|
harness->data.fixture_name = fixture_setup_array_get_name(fsa,
|
|
|
|
fi);
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
harness->data.passed = 0;
|
|
|
|
harness->data.skipped = 0;
|
|
|
|
harness->data.failed = 0;
|
|
|
|
|
2021-02-15 15:43:14 +03:00
|
|
|
testlog("--- Fixture %d (%s)...\n",
|
|
|
|
fi + 1, harness->data.fixture_name);
|
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
ret = fixture_setup_run_(harness, arg);
|
|
|
|
fixture_report(&harness->data, ret);
|
|
|
|
|
|
|
|
if (ret == RESULT_SKIP) {
|
|
|
|
tap_skip_fixture(&harness->data);
|
2021-03-03 15:21:01 +03:00
|
|
|
#if WESTON_TEST_SKIP_IS_FAILURE
|
|
|
|
ret = RESULT_FAIL;
|
|
|
|
#else
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
continue;
|
2021-03-03 15:21:01 +03:00
|
|
|
#endif
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
if (ret != RESULT_OK && result != RESULT_HARD_ERROR)
|
|
|
|
result = ret;
|
|
|
|
else if (counts_to_result(&harness->data) != RESULT_OK)
|
|
|
|
result = RESULT_FAIL;
|
|
|
|
}
|
2014-02-07 12:34:48 +04:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
weston_test_harness_destroy(harness);
|
2012-12-08 01:50:32 +04:00
|
|
|
|
tests: thread-based client harness
This replaces the old test harness with a new one.
The old harness relied on fork()'ing each test which makes tests independent,
but makes debugging them harder. The new harness runs client code in a thread
instead of a new process. A side-effect of not fork()'ing anymore is that any
failure will stop running a test series short. Fortunately we do not have any
tests that are expected to crash or fail.
The old harness executed 'weston' from Meson, with lots of setup as both
command line options and environment variables. The new harness executes
wet_main() instead: the test program itself calls the compositor main function
to execute the compositor in-process. Command line arguments are configured in
the test program itself, not in meson.build. Environment variables aside, you
are able to run a test by simply executing the test program, even if it is a
plugin test.
The new harness adds a new type of iteration: fixtures. For now, fixtures are
used to set up the compositor for tests that need a compositor. If necessary, a
fixture setup may include a data array of arbitrary type for executing the test
series for each element in the array. This will be most useful for running
screenshooting tests with both Pixman- and GL-renderers.
The new harness outputs TAP formatted results into stdout. Meson is not
switched to consume TAP yet though, because it would require a Meson version
requirement bump and would not have any benefits at this time. OTOH outputting
TAP is trivial and sets up a clear precedent of random test chatter belonging
to stderr.
This commit migrates only few tests to actually make use of the new features:
roles is a basic client test, subsurface-shot is a client test that
demonstrates the fixture array, and plugin-registry is a plugin test. The rest
of the tests will be migrated later.
Once all tests are migrated, we can remove the test-specific setup from
meson.build, leaving only the actual build instructions in there.
The not migrated tests and stand-alone tests suffer only a minor change: they
no longer fork() for each TEST(), otherwise they keep running as before.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2019-11-01 15:02:15 +03:00
|
|
|
return result;
|
2012-12-08 01:50:32 +04:00
|
|
|
}
|