asterisk/tests/test_sched.c

487 lines
13 KiB
C
Raw Normal View History

/*
* Asterisk -- An open source telephony toolkit.
*
* Copyright (C) 2009, Digium, Inc.
*
* Russell Bryant <russell@digium.com>
*
* See http://www.asterisk.org for more information about
* the Asterisk project. Please do not directly contact
* any of the maintainers of this project for assistance;
* the project provides a web site, mailing lists and IRC
* channels for your use.
*
* This program is free software, distributed under the terms of
* the GNU General Public License Version 2. See the LICENSE file
* at the top of the source tree.
*/
/*! \file
*
* \brief ast_sched performance test module
*
* \author Russell Bryant <russell@digium.com>
*/
/*** MODULEINFO
<depend>TEST_FRAMEWORK</depend>
<support_level>core</support_level>
***/
#include "asterisk.h"
#include <inttypes.h>
#include "asterisk/module.h"
#include "asterisk/utils.h"
#include "asterisk/sched.h"
#include "asterisk/test.h"
#include "asterisk/cli.h"
#include "asterisk/astobj2.h"
static int sched_cb(const void *data)
{
return 0;
}
static int order_check;
static int order_check_failed;
static void sched_order_check(struct ast_test *test, int order)
{
++order_check;
if (order_check != order) {
ast_test_status_update(test, "Unexpected execution order: expected:%d got:%d\n",
order, order_check);
order_check_failed = 1;
}
}
static int sched_order_1_cb(const void *data)
{
sched_order_check((void *) data, 1);
return 0;
}
static int sched_order_2_cb(const void *data)
{
sched_order_check((void *) data, 2);
return 0;
}
static int sched_order_3_cb(const void *data)
{
sched_order_check((void *) data, 3);
return 0;
}
static int sched_order_4_cb(const void *data)
{
sched_order_check((void *) data, 4);
return 0;
}
static int sched_order_5_cb(const void *data)
{
sched_order_check((void *) data, 5);
return 0;
}
static int sched_order_6_cb(const void *data)
{
sched_order_check((void *) data, 6);
return 0;
}
static int sched_order_7_cb(const void *data)
{
sched_order_check((void *) data, 7);
return 0;
}
static int sched_order_8_cb(const void *data)
{
sched_order_check((void *) data, 8);
return 0;
}
AST_TEST_DEFINE(sched_test_order)
{
struct ast_sched_context *con;
enum ast_test_result_state res = AST_TEST_FAIL;
int id1, id2, id3, wait;
switch (cmd) {
case TEST_INIT:
info->name = "sched_test_order";
info->category = "/main/sched/";
info->summary = "Test ordering of events in the scheduler API";
info->description =
"This test ensures that events are properly ordered by the "
"time they are scheduled to execute in the scheduler API.";
return AST_TEST_NOT_RUN;
case TEST_EXECUTE:
break;
}
if (!(con = ast_sched_context_create())) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test,
"Test failed - could not create scheduler context\n");
return AST_TEST_FAIL;
}
/* Add 3 scheduler entries, and then remove them, ensuring that the result
* of ast_sched_wait() looks appropriate at each step along the way. */
if ((wait = ast_sched_wait(con)) != -1) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test,
"ast_sched_wait() should have returned -1, returned '%d'\n",
wait);
goto return_cleanup;
}
if ((id1 = ast_sched_add(con, 100000, sched_cb, NULL)) == -1) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test, "Failed to add scheduler entry\n");
goto return_cleanup;
}
if ((wait = ast_sched_wait(con)) > 100000) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test,
"ast_sched_wait() should have returned <= 100000, returned '%d'\n",
wait);
goto return_cleanup;
}
if ((id2 = ast_sched_add(con, 10000, sched_cb, NULL)) == -1) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test, "Failed to add scheduler entry\n");
goto return_cleanup;
}
if ((wait = ast_sched_wait(con)) > 10000) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test,
"ast_sched_wait() should have returned <= 10000, returned '%d'\n",
wait);
goto return_cleanup;
}
if ((id3 = ast_sched_add(con, 1000, sched_cb, NULL)) == -1) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test, "Failed to add scheduler entry\n");
goto return_cleanup;
}
if ((wait = ast_sched_wait(con)) > 1000) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test,
"ast_sched_wait() should have returned <= 1000, returned '%d'\n",
wait);
goto return_cleanup;
}
if (ast_sched_del(con, id3) == -1) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test, "Failed to remove scheduler entry\n");
goto return_cleanup;
}
if ((wait = ast_sched_wait(con)) <= 1000) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test,
"ast_sched_wait() should have returned > 1000, returned '%d'\n",
wait);
goto return_cleanup;
}
if (ast_sched_del(con, id2) == -1) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test, "Failed to remove scheduler entry\n");
goto return_cleanup;
}
if ((wait = ast_sched_wait(con)) <= 10000) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test,
"ast_sched_wait() should have returned > 10000, returned '%d'\n",
wait);
goto return_cleanup;
}
if (ast_sched_del(con, id1) == -1) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test, "Failed to remove scheduler entry\n");
goto return_cleanup;
}
if ((wait = ast_sched_wait(con)) != -1) {
Various updates to the unit test API. 1) It occurred to me that the difference in usage between the error ast_str and the ast_test_update_status() usage has turned out to be a bit ambiguous in practice. In a lot of cases, the same message was being sent to both. In other cases, it was only sent to one or the other. My opinion now is that in every case, I think it makes sense to do both; we should output it to the CLI as well as save it off for logging purposes. This change results in most of the changes in this diff, since it required changes to all existing unit tests. It also allowed for some simplifications of unit test API implementation code. 2) Update ast_test_status_update() to include the file, function, and line number for the code providing the update. 3) There are some formatting tweaks here and there. Hopefully they aren't too distracting for code review purposes. Reviewboard's diff viewer seems to do a pretty good job of pointing out when something is a whitespace change. 4) I moved the md5_test and sha1_test into the test_utils module. It seemed like a better approach since these tests are so tiny. 5) I changed the number of nodes used in heap_test_2 from 1 million to 100 thousand. The only reason for this was to reduce the time it took for this test to run. 6) Remove an unused function prototype that was at the bottom of utils.h. 7) Simplify test_insert() using the LIST_INSERT_SORTALPHA() macro. The one minor difference in behavior is that it no longer checks for a test registered with the same name. 8) Expand the code in test_alloc() to provide specific error messages for each failure case, to clearly inform developers if they forget to set the name, summary, description, etc. 9) Tweak the output of the "test show registered" CLI command. I swapped the name and category to have the category first. It seemed more natural since that is the sort key. 10) Don't output the status ast_str in the "test show results" CLI command. This is going to tend to be pretty verbose, so just leave that for the detailed test logs (test generate results). Review: https://reviewboard.asterisk.org/r/493/ git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@245864 65c4cc65-6c06-0410-ace0-fbb531ad65f3
2010-02-09 23:32:14 +00:00
ast_test_status_update(test,
"ast_sched_wait() should have returned -1, returned '%d'\n",
wait);
goto return_cleanup;
}
/*
* Schedule immediate and delayed entries to check the order
* that they get executed. They must get executed at the
* time they expire in the order they were added.
*/
#define DELAYED_SAME_EXPIRE 300 /* ms */
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, DELAYED_SAME_EXPIRE, sched_order_1_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, 0, sched_order_1_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, DELAYED_SAME_EXPIRE, sched_order_2_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, 0, sched_order_2_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, DELAYED_SAME_EXPIRE, sched_order_3_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, 0, sched_order_3_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, DELAYED_SAME_EXPIRE, sched_order_4_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, 0, sched_order_4_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, DELAYED_SAME_EXPIRE, sched_order_5_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, 0, sched_order_5_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, DELAYED_SAME_EXPIRE, sched_order_6_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, 0, sched_order_6_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, DELAYED_SAME_EXPIRE, sched_order_7_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, 0, sched_order_7_cb, test), res, return_cleanup);
ast_test_validate_cleanup(test, -1 < ast_sched_add(con, DELAYED_SAME_EXPIRE, sched_order_8_cb, test), res, return_cleanup);
/* Check order of scheduled immediate entries. */
order_check = 0;
order_check_failed = 0;
usleep(50 * 1000);/* Ensure that all the immediate entries are ready to expire */
ast_test_validate_cleanup(test, 7 == ast_sched_runq(con), res, return_cleanup);
ast_test_validate_cleanup(test, !order_check_failed, res, return_cleanup);
/* Check order of scheduled entries expiring at the same time. */
order_check = 0;
order_check_failed = 0;
usleep((DELAYED_SAME_EXPIRE + 50) * 1000);/* Ensure that all the delayed entries are ready to expire */
ast_test_validate_cleanup(test, 8 == ast_sched_runq(con), res, return_cleanup);
ast_test_validate_cleanup(test, !order_check_failed, res, return_cleanup);
if ((wait = ast_sched_wait(con)) != -1) {
ast_test_status_update(test,
"ast_sched_wait() should have returned -1, returned '%d'\n",
wait);
goto return_cleanup;
}
res = AST_TEST_PASS;
return_cleanup:
ast_sched_context_destroy(con);
return res;
}
static char *handle_cli_sched_bench(struct ast_cli_entry *e, int cmd, struct ast_cli_args *a)
{
struct ast_sched_context *con;
struct timeval start;
unsigned int num, i;
int *sched_ids = NULL;
switch (cmd) {
case CLI_INIT:
e->command = "sched benchmark";
e->usage = ""
"Usage: sched benchmark <num>\n"
"";
return NULL;
case CLI_GENERATE:
return NULL;
}
if (a->argc != e->args + 1) {
return CLI_SHOWUSAGE;
}
if (sscanf(a->argv[e->args], "%u", &num) != 1) {
return CLI_SHOWUSAGE;
}
if (!(con = ast_sched_context_create())) {
ast_cli(a->fd, "Test failed - could not create scheduler context\n");
return CLI_FAILURE;
}
if (!(sched_ids = ast_malloc(sizeof(*sched_ids) * num))) {
ast_cli(a->fd, "Test failed - memory allocation failure\n");
goto return_cleanup;
}
ast_cli(a->fd, "Testing ast_sched_add() performance - timing how long it takes "
"to add %u entries at random time intervals from 0 to 60 seconds\n", num);
start = ast_tvnow();
for (i = 0; i < num; i++) {
long when = labs(ast_random()) % 60000;
if ((sched_ids[i] = ast_sched_add(con, when, sched_cb, NULL)) == -1) {
ast_cli(a->fd, "Test failed - sched_add returned -1\n");
goto return_cleanup;
}
}
ast_cli(a->fd, "Test complete - %" PRIi64 " us\n", ast_tvdiff_us(ast_tvnow(), start));
ast_cli(a->fd, "Testing ast_sched_del() performance - timing how long it takes "
"to delete %u entries with random time intervals from 0 to 60 seconds\n", num);
start = ast_tvnow();
for (i = 0; i < num; i++) {
if (ast_sched_del(con, sched_ids[i]) == -1) {
ast_cli(a->fd, "Test failed - sched_del returned -1\n");
goto return_cleanup;
}
}
ast_cli(a->fd, "Test complete - %" PRIi64 " us\n", ast_tvdiff_us(ast_tvnow(), start));
return_cleanup:
ast_sched_context_destroy(con);
if (sched_ids) {
ast_free(sched_ids);
}
return CLI_SUCCESS;
}
struct test_obj {
ast_mutex_t lock;
ast_cond_t cond;
int scheduledCBstarted;
int id;
};
static void test_obj_cleanup(void *data)
{
struct test_obj *obj = data;
ast_mutex_destroy(&obj->lock);
ast_cond_destroy(&obj->cond);
}
static int lockingcb(const void *data)
{
struct test_obj *obj = (struct test_obj *)data;
struct timespec delay = {3,0};
ast_mutex_lock(&obj->lock);
obj->scheduledCBstarted = 1;
ast_cond_signal(&obj->cond);
ast_mutex_unlock(&obj->lock);
ao2_ref(obj, -1);
while (nanosleep(&delay, &delay));
/* sleep to force this scheduled event to remain running long
* enough for the scheduling thread to unlock and call
* AST_SCHED_DEL_UNREF
*/
return 0;
}
AST_TEST_DEFINE(sched_test_freebird)
{
struct test_obj * obj;
struct ast_sched_context * con;
enum ast_test_result_state res = AST_TEST_FAIL;
int refs;
switch (cmd) {
case TEST_INIT:
info->name = "sched_test_freebird";
info->category = "/main/sched/";
info->summary = "Test deadlock avoidance and double-unref";
info->description =
"This tests a call to AST_SCHED_DEL_UNREF on a running event.";
return AST_TEST_NOT_RUN;
case TEST_EXECUTE:
res = AST_TEST_PASS;
break;
}
obj = ao2_alloc(sizeof(struct test_obj), test_obj_cleanup);
if (!obj) {
ast_test_status_update(test,
"ao2_alloc() did not return an object\n");
return AST_TEST_FAIL;
}
obj->scheduledCBstarted = 0;
con = ast_sched_context_create();
if (!con) {
ast_test_status_update(test,
"ast_sched_context_create() did not return a context\n");
ao2_cleanup(obj);
return AST_TEST_FAIL;
}
if (ast_sched_start_thread(con)) {
ast_test_status_update(test, "Failed to start test thread\n");
ao2_cleanup(obj);
ast_sched_context_destroy(con);
return AST_TEST_FAIL;
}
/* This double reference is to ensure that the object isn't destroyed prematurely
* in a case where it is unreffed an additional time.
*/
ao2_ref(obj, +2);
if ((obj->id = ast_sched_add(con, 0, lockingcb, obj)) == -1) {
ast_test_status_update(test, "Failed to add scheduler entry\n");
ao2_ref(obj, -3);
ast_sched_context_destroy(con);
return AST_TEST_FAIL;
}
ast_mutex_lock(&obj->lock);
while(obj->scheduledCBstarted == 0) {
/* Wait for the scheduled thread to indicate that it has started so we can
* then call the AST_SCHED_DEL_UNREF macro
*/
ast_cond_wait(&obj->cond,&obj->lock);
}
ast_mutex_unlock(&obj->lock);
ast_test_status_update(test, "Received signal, calling Scedule and UNREF\n");
ast_test_status_update(test, "ID: %d\n", obj->id);
AST_SCHED_DEL_UNREF(con, obj->id, ao2_ref(obj, -1));
refs = ao2_ref(obj, 0);
switch(refs){
case 2:
ast_test_status_update(test, "Correct number of references '2'\n");
break;
default:
ast_test_status_update(test, "Incorrect number of references '%d'\n",
refs);
res = AST_TEST_FAIL;
break;
}
/* Based on success or failure, the refcount could change
*/
while(ao2_ref(obj, -1) > 1);
ast_sched_context_destroy(con);
return res;
}
static struct ast_cli_entry cli_sched[] = {
AST_CLI_DEFINE(handle_cli_sched_bench, "Benchmark ast_sched add/del performance"),
};
static int unload_module(void)
{
AST_TEST_UNREGISTER(sched_test_order);
AST_TEST_UNREGISTER(sched_test_freebird);
ast_cli_unregister_multiple(cli_sched, ARRAY_LEN(cli_sched));
return 0;
}
static int load_module(void)
{
AST_TEST_REGISTER(sched_test_order);
AST_TEST_REGISTER(sched_test_freebird);
ast_cli_register_multiple(cli_sched, ARRAY_LEN(cli_sched));
return AST_MODULE_LOAD_SUCCESS;
}
AST_MODULE_INFO_STANDARD(ASTERISK_GPL_KEY, "ast_sched performance test module");