mir/bind
2
0
Fork 0
mirror of https://gitlab.isc.org/isc-projects/bind9 synced 2025-08-22 10:10:06 +00:00
Code Issues Releases Wiki Activity
bind/lib/isc/thread.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

116 lines
3.1 KiB
C
Raw Normal View History

forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
/*
conditionally declare stacksize
2018-02-16 08:37:33 +11:00
* Copyright (C) Internet Systems Consortium, Inc. ("ISC")
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
*
* SPDX-License-Identifier: MPL-2.0
Update the copyright information in all files in the repository This commit converts the license handling to adhere to the REUSE specification. It specifically: 1. Adds used licnses to LICENSES/ directory 2. Add "isc" template for adding the copyright boilerplate 3. Changes all source files to include copyright and SPDX license header, this includes all the C sources, documentation, zone files, configuration files. There are notes in the doc/dev/copyrights file on how to add correct headers to the new files. 4. Handle the rest that can't be modified via .reuse/dep5 file. The binary (or otherwise unmodifiable) files could have license places next to them in <foo>.license file, but this would lead to cluttered repository and most of the files handled in the .reuse/dep5 file are system test files.
2021-06-03 08:37:05 +02:00
*
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, you can obtain one at https://mozilla.org/MPL/2.0/.
Update license headers to not include years in copyright in all applicable files
2018-02-23 09:53:12 +01:00
*
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
* See the COPYRIGHT file distributed with this work for additional
* information regarding copyright ownership.
*/
1851. [doc] Doxygen comment markup. [RT #11398]
2005-04-27 04:57:32 +00:00
/*! \file */
add RCS id string
2000-06-22 22:00:42 +00:00
3661. [bug] Address lock order reversal deadlock with inline zones. [RT #34856]
2013-10-08 11:43:08 +11:00
#if defined(HAVE_SCHED_H)
#include <sched.h>
#endif /* if defined(HAVE_SCHED_H) */
isc_thread_setaffinity()
2018-10-03 23:20:22 +02:00
#if defined(HAVE_CPUSET_H)
#include <sys/cpuset.h>
#include <sys/param.h>
#endif /* if defined(HAVE_CPUSET_H) */
Fix a typo in lib/isc/unix/thread.c
2018-11-07 19:31:06 +00:00
#if defined(HAVE_SYS_PROCSET_H)
isc_thread_setaffinity()
2018-10-03 23:20:22 +02:00
#include <sys/processor.h>
#include <sys/procset.h>
#include <sys/types.h>
#endif /* if defined(HAVE_SYS_PROCSET_H) */
Make isc_thread_join() assert internally on failure Previously isc_thread_join() would return ISC_R_UNEXPECTED on a failure to create new thread. All such occurences were caught and wrapped into assert function at higher level. The function was simplified to assert directly in the isc_thread_join() function and all caller level assertions were removed.
2019-07-18 17:47:40 +02:00
#include <isc/strerr.h>
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
#include <isc/thread.h>
add isc_thread_setconcurrency()
2000-06-22 01:48:24 +00:00
#include <isc/util.h>
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
Add isc_trampoline API to have simple accounting around threads The current isc_hp API uses internal tid_v variable that gets incremented for each new thread using hazard pointers. This tid_v variable is then used as a index to global shared table with hazard pointers state. Since the tid_v is only incremented and never decremented the table could overflow very quickly if we create set of threads for short period of time, they finish the work and cease to exist. Then we create identical set of threads and so on and so on. This is not a problem for a normal `named` operation as the set of threads is stable, but the problematic place are the unit tests where we test network manager or other APIs (task, timer) that create threads. This commits adds a thin wrapper around any function called from isc_thread_create() that adds unique-but-reusable small digit thread id that can be used as index to f.e. hazard pointer tables. The trampoline wrapper ensures that the thread ids will be reused, so the highest thread_id number doesn't grow indefinitely when threads are created and destroyed and then created again. This fixes the hazard pointer table overflow on machines with many cores. [GL #2396]
2021-02-16 15:57:39 +01:00
#include "trampoline_p.h"
Make the #define for thread stack size private by moving it to thread.c, rather than thread.h.
2000-04-19 20:48:24 +00:00
#ifndef THREAD_MINSTACKSIZE
3451. [port] Increase per thread stack size from 64K to 1M. [RT #32230]
2013-01-07 09:52:16 +11:00
#define THREAD_MINSTACKSIZE (1024U * 1024)
Make the #define for thread stack size private by moving it to thread.c, rather than thread.h.
2000-04-19 20:48:24 +00:00
#endif /* ifndef THREAD_MINSTACKSIZE */
Make isc_thread_create() assert internally on failure Previously isc_thread_create() would return ISC_R_UNEXPECTED on a failure to create new thread. All such occurences were caught and wrapped into assert function at higher level. The function was simplified to assert directly in the isc_thread_create() function and all caller level assertions were removed.
2019-07-18 17:24:05 +02:00
void
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
isc_thread_create(isc_threadfunc_t func, isc_threadarg_t arg,
isc_thread_t *thread) {
pthread_attr_t attr;
Add isc_trampoline API to have simple accounting around threads The current isc_hp API uses internal tid_v variable that gets incremented for each new thread using hazard pointers. This tid_v variable is then used as a index to global shared table with hazard pointers state. Since the tid_v is only incremented and never decremented the table could overflow very quickly if we create set of threads for short period of time, they finish the work and cease to exist. Then we create identical set of threads and so on and so on. This is not a problem for a normal `named` operation as the set of threads is stable, but the problematic place are the unit tests where we test network manager or other APIs (task, timer) that create threads. This commits adds a thin wrapper around any function called from isc_thread_create() that adds unique-but-reusable small digit thread id that can be used as index to f.e. hazard pointer tables. The trampoline wrapper ensures that the thread ids will be reused, so the highest thread_id number doesn't grow indefinitely when threads are created and destroyed and then created again. This fixes the hazard pointer table overflow on machines with many cores. [GL #2396]
2021-02-16 15:57:39 +01:00
isc__trampoline_t *trampoline_arg;
trampoline_arg = isc__trampoline_get(func, arg);
conditionally declare stacksize
2018-02-16 08:37:33 +11:00
#if defined(HAVE_PTHREAD_ATTR_GETSTACKSIZE) && \
defined(HAVE_PTHREAD_ATTR_SETSTACKSIZE)
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
size_t stacksize;
conditionally declare stacksize
2018-02-16 08:37:33 +11:00
#endif /* if defined(HAVE_PTHREAD_ATTR_GETSTACKSIZE) && \
* defined(HAVE_PTHREAD_ATTR_SETSTACKSIZE) */
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
int ret;
pthread_attr_init(&attr);
1496. [port] test for pthread_attr_setstacksize().
2003-07-25 05:08:51 +00:00
#if defined(HAVE_PTHREAD_ATTR_GETSTACKSIZE) && \
defined(HAVE_PTHREAD_ATTR_SETSTACKSIZE)
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
ret = pthread_attr_getstacksize(&attr, &stacksize);
Fix error reporting for POSIX Threads functions Commit 3608abc8fa6a33046e1d34a0789cf7c9547f09ad inadvertently carried over a mistake in logging pthread_cond_init() errors to the ERRNO_CHECK() preprocessor macro: instead of passing the value returned by a given pthread_*() function to strerror_r(), ERRNO_CHECK() passes the errno variable to strerror_r(). This causes bogus error reports because POSIX Threads API functions do not set the errno variable. Fix by passing the value returned by a given pthread_*() function instead of the errno variable to strerror_r(). Since this change makes the name of the affected macro (ERRNO_CHECK()) confusing, rename the latter to PTHREADS_RUNTIME_CHECK(). Also log the integer error value returned by a given pthread_*() function verbatim to rule out any further confusion in runtime error reporting.
2022-09-09 20:25:47 +02:00
PTHREADS_RUNTIME_CHECK(pthread_attr_getstacksize, ret);
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
Make the #define for thread stack size private by moving it to thread.c, rather than thread.h.
2000-04-19 20:48:24 +00:00
if (stacksize < THREAD_MINSTACKSIZE) {
ret = pthread_attr_setstacksize(&attr, THREAD_MINSTACKSIZE);
Fix error reporting for POSIX Threads functions Commit 3608abc8fa6a33046e1d34a0789cf7c9547f09ad inadvertently carried over a mistake in logging pthread_cond_init() errors to the ERRNO_CHECK() preprocessor macro: instead of passing the value returned by a given pthread_*() function to strerror_r(), ERRNO_CHECK() passes the errno variable to strerror_r(). This causes bogus error reports because POSIX Threads API functions do not set the errno variable. Fix by passing the value returned by a given pthread_*() function instead of the errno variable to strerror_r(). Since this change makes the name of the affected macro (ERRNO_CHECK()) confusing, rename the latter to PTHREADS_RUNTIME_CHECK(). Also log the integer error value returned by a given pthread_*() function verbatim to rule out any further confusion in runtime error reporting.
2022-09-09 20:25:47 +02:00
PTHREADS_RUNTIME_CHECK(pthread_attr_setstacksize, ret);
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
}
- if pthreads are not found, build a nonthreaded version - check for pthread_attr_getstacksize
2000-12-22 20:50:05 +00:00
#endif /* if defined(HAVE_PTHREAD_ATTR_GETSTACKSIZE) && \
1496. [port] test for pthread_attr_setstacksize().
2003-07-25 05:08:51 +00:00
* defined(HAVE_PTHREAD_ATTR_SETSTACKSIZE) */
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
Add isc_trampoline API to have simple accounting around threads The current isc_hp API uses internal tid_v variable that gets incremented for each new thread using hazard pointers. This tid_v variable is then used as a index to global shared table with hazard pointers state. Since the tid_v is only incremented and never decremented the table could overflow very quickly if we create set of threads for short period of time, they finish the work and cease to exist. Then we create identical set of threads and so on and so on. This is not a problem for a normal `named` operation as the set of threads is stable, but the problematic place are the unit tests where we test network manager or other APIs (task, timer) that create threads. This commits adds a thin wrapper around any function called from isc_thread_create() that adds unique-but-reusable small digit thread id that can be used as index to f.e. hazard pointer tables. The trampoline wrapper ensures that the thread ids will be reused, so the highest thread_id number doesn't grow indefinitely when threads are created and destroyed and then created again. This fixes the hazard pointer table overflow on machines with many cores. [GL #2396]
2021-02-16 15:57:39 +01:00
ret = pthread_create(thread, &attr, isc__trampoline_run,
trampoline_arg);
Fix error reporting for POSIX Threads functions Commit 3608abc8fa6a33046e1d34a0789cf7c9547f09ad inadvertently carried over a mistake in logging pthread_cond_init() errors to the ERRNO_CHECK() preprocessor macro: instead of passing the value returned by a given pthread_*() function to strerror_r(), ERRNO_CHECK() passes the errno variable to strerror_r(). This causes bogus error reports because POSIX Threads API functions do not set the errno variable. Fix by passing the value returned by a given pthread_*() function instead of the errno variable to strerror_r(). Since this change makes the name of the affected macro (ERRNO_CHECK()) confusing, rename the latter to PTHREADS_RUNTIME_CHECK(). Also log the integer error value returned by a given pthread_*() function verbatim to rule out any further confusion in runtime error reporting.
2022-09-09 20:25:47 +02:00
PTHREADS_RUNTIME_CHECK(pthread_create, ret);
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
pthread_attr_destroy(&attr);
Make isc_thread_create() assert internally on failure Previously isc_thread_create() would return ISC_R_UNEXPECTED on a failure to create new thread. All such occurences were caught and wrapped into assert function at higher level. The function was simplified to assert directly in the isc_thread_create() function and all caller level assertions were removed.
2019-07-18 17:24:05 +02:00
return;
forgot to add this, thanks Bob. :)
2000-04-19 20:34:34 +00:00
}
add isc_thread_setconcurrency()
2000-06-22 01:48:24 +00:00
Make isc_thread_join() assert internally on failure Previously isc_thread_join() would return ISC_R_UNEXPECTED on a failure to create new thread. All such occurences were caught and wrapped into assert function at higher level. The function was simplified to assert directly in the isc_thread_join() function and all caller level assertions were removed.
2019-07-18 17:47:40 +02:00
void
isc_thread_join(isc_thread_t thread, isc_threadresult_t *result) {
int ret = pthread_join(thread, result);
Add an ERRNO_CHECK() preprocessor macro In a number of situations in pthreads-related code, a common sequence of steps is taken: if the value returned by a library function is not 0, pass errno to strerror_r(), log the string returned by the latter, and immediately abort execution. Add an ERRNO_CHECK() preprocessor macro which takes those exact steps and use it wherever (conveniently) possible. Notes: 1. The "log the return value of strerror_r() and abort" pattern is used in a number of other places that this commit does not touch; only "!= 0" checks followed by isc_error_fatal() calls with non-customized error messages are replaced here. 2. This change temporarily breaks file name & line number reporting for isc__mutex_init() errors, to prevent breaking the build. This issue will be rectified in a subsequent change.
2022-07-13 13:19:32 +02:00
Fix error reporting for POSIX Threads functions Commit 3608abc8fa6a33046e1d34a0789cf7c9547f09ad inadvertently carried over a mistake in logging pthread_cond_init() errors to the ERRNO_CHECK() preprocessor macro: instead of passing the value returned by a given pthread_*() function to strerror_r(), ERRNO_CHECK() passes the errno variable to strerror_r(). This causes bogus error reports because POSIX Threads API functions do not set the errno variable. Fix by passing the value returned by a given pthread_*() function instead of the errno variable to strerror_r(). Since this change makes the name of the affected macro (ERRNO_CHECK()) confusing, rename the latter to PTHREADS_RUNTIME_CHECK(). Also log the integer error value returned by a given pthread_*() function verbatim to rule out any further confusion in runtime error reporting.
2022-09-09 20:25:47 +02:00
PTHREADS_RUNTIME_CHECK(pthread_join, ret);
Make isc_thread_join() assert internally on failure Previously isc_thread_join() would return ISC_R_UNEXPECTED on a failure to create new thread. All such occurences were caught and wrapped into assert function at higher level. The function was simplified to assert directly in the isc_thread_join() function and all caller level assertions were removed.
2019-07-18 17:47:40 +02:00
}
[master] give threads unique names to assist debugging 4602. [func] Threads are now set to human-readable names to assist debugging, when supported by the OS. [RT #43234]
2017-04-21 13:58:22 -07:00
void
isc_thread_setname(isc_thread_t thread, const char *name) {
Enable system extensions macros using AC_USE_SYSTEM_EXTENSIONS and remove handcrafted code that enabled __EXTENSIONS__ and _GNU_SOURCE
2018-08-28 21:27:48 +02:00
#if defined(HAVE_PTHREAD_SETNAME_NP) && !defined(__APPLE__)
[master] give threads unique names to assist debugging 4602. [func] Threads are now set to human-readable names to assist debugging, when supported by the OS. [RT #43234]
2017-04-21 13:58:22 -07:00
/*
* macOS has pthread_setname_np but only works on the
* current thread so it's not used here
*/
Fix the build on NetBSD due to the differences in pthread implementation. Signed-off-by: Ondřej Surý <ondrej@sury.org>
2018-11-22 15:15:11 +01:00
#if defined(__NetBSD__)
(void)pthread_setname_np(thread, name, NULL);
#else /* if defined(__NetBSD__) */
[master] give threads unique names to assist debugging 4602. [func] Threads are now set to human-readable names to assist debugging, when supported by the OS. [RT #43234]
2017-04-21 13:58:22 -07:00
(void)pthread_setname_np(thread, name);
Fix the build on NetBSD due to the differences in pthread implementation. Signed-off-by: Ondřej Surý <ondrej@sury.org>
2018-11-22 15:15:11 +01:00
#endif /* if defined(__NetBSD__) */
[master] give threads unique names to assist debugging 4602. [func] Threads are now set to human-readable names to assist debugging, when supported by the OS. [RT #43234]
2017-04-21 13:58:22 -07:00
#elif defined(HAVE_PTHREAD_SET_NAME_NP)
(void)pthread_set_name_np(thread, name);
#else /* if defined(HAVE_PTHREAD_SETNAME_NP) && !defined(__APPLE__) */
UNUSED(thread);
UNUSED(name);
#endif /* if defined(HAVE_PTHREAD_SETNAME_NP) && !defined(__APPLE__) */
}
3661. [bug] Address lock order reversal deadlock with inline zones. [RT #34856]
2013-10-08 11:43:08 +11:00
void
isc_thread_yield(void) {
#if defined(HAVE_SCHED_YIELD)
sched_yield();
#elif defined(HAVE_PTHREAD_YIELD)
pthread_yield();
#elif defined(HAVE_PTHREAD_YIELD_NP)
pthread_yield_np();
#endif /* if defined(HAVE_SCHED_YIELD) */
}
Reference in New Issue Copy Permalink