d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
9539436c36
freebsd-nq/lib/libkse/thread/thr_create.c

250 lines
7.7 KiB
C
Raw Normal View History

Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/*
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
* Copyright (c) 1995-1998 John Birrell <jb@cimlogic.com.au>
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by John Birrell.
* 4. Neither the name of the author nor the names of any co-contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY JOHN BIRRELL AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
* $Id$
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
*/
#include <errno.h>
#include <stdlib.h>
#include <string.h>
#include <fcntl.h>
#include <unistd.h>
#include <sys/time.h>
#ifdef _THREAD_SAFE
#include <machine/reg.h>
#include <pthread.h>
#include "pthread_private.h"
Change the FILE locking to be by FILE, not by the underlying fd as it was. Add a FILE_WAIT state and queue threads waiting for a FILE lock. Start using the sys/queue.h macros instead of the way that MIT pthreads did it. Add a thread name to the private thread structure and a non-POSIX function to set this. This helps (me at least) when sending a SIGINFO to a threaded process to get a /tmp/uthread.dump to see what the <expletive deleted> threads are doing this time. It is nice to be able to recognise (yes, I spell that with an 's' too) which threads are which.
1998-04-11 07:47:22 +00:00
#include "libc_private.h"
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
int
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
pthread_create(pthread_t * thread, const pthread_attr_t * attr,
void *(*start_routine) (void *), void *arg)
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
{
Move the cleanup code that frees memory allocated for a dead thread from the thread kernel into a garbage collector thread which is started when the fisrt thread is created (other than the initial thread). This removes the window of opportunity where a context switch will cause a thread that has locked the malloc spinlock, to enter the thread kernel, find there is a dead thread and try to free memory, therefore trying to lock the malloc spinlock against itself. The garbage collector thread acts just like any other thread, so instead of having a spinlock to control accesses to the dead thread list, it uses a mutex and a condition variable so that it can happily wait to be signalled when a thread exists.
1998-09-30 06:36:56 +00:00
int f_gc = 0;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
int i;
int ret = 0;
int status;
Move the cleanup code that frees memory allocated for a dead thread from the thread kernel into a garbage collector thread which is started when the fisrt thread is created (other than the initial thread). This removes the window of opportunity where a context switch will cause a thread that has locked the malloc spinlock, to enter the thread kernel, find there is a dead thread and try to free memory, therefore trying to lock the malloc spinlock against itself. The garbage collector thread acts just like any other thread, so instead of having a spinlock to control accesses to the dead thread list, it uses a mutex and a condition variable so that it can happily wait to be signalled when a thread exists.
1998-09-30 06:36:56 +00:00
pthread_t gc_thread;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
pthread_t new_thread;
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
pthread_attr_t pattr;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
void *stack;
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
/*
* Locking functions in libc are required when there are
* threads other than the initial thread.
*/
__isthreaded = 1;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Allocate memory for the thread structure: */
if ((new_thread = (pthread_t) malloc(sizeof(struct pthread))) == NULL) {
/* Insufficient memory to create a thread: */
ret = EAGAIN;
} else {
/* Check if default thread attributes are required: */
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
if (attr == NULL || *attr == NULL) {
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Use the default thread attributes: */
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
pattr = &pthread_attr_default;
} else {
pattr = *attr;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
}
/* Check if a stack was specified in the thread attributes: */
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
if ((stack = pattr->stackaddr_attr) != NULL) {
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
}
/* Allocate memory for the stack: */
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
else if ((stack = (void *) malloc(pattr->stacksize_attr)) == NULL) {
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Insufficient memory to create a thread: */
ret = EAGAIN;
free(new_thread);
}
/* Check for errors: */
if (ret != 0) {
} else {
/* Initialise the thread structure: */
memset(new_thread, 0, sizeof(struct pthread));
new_thread->slice_usec = -1;
new_thread->sig_saved = 0;
new_thread->stack = stack;
new_thread->start_routine = start_routine;
new_thread->arg = arg;
Move the magic field initialisation to a place when it is more magic.
1998-04-04 07:27:29 +00:00
/*
* Write a magic value to the thread structure
* to help identify valid ones:
*/
new_thread->magic = PTHREAD_MAGIC;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Initialise the thread for signals: */
new_thread->sigmask = _thread_run->sigmask;
/* Initialise the jump buffer: */
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
setjmp(new_thread->saved_jmp_buf);
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/*
* Set up new stack frame so that it looks like it
* returned from a longjmp() to the beginning of
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
* _thread_start().
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
*/
#if defined(__FreeBSD__)
Add FreeBSD/Alpha code to initialise a jmpbuf for a created thread. Change a bunch of __alpha references to __alpha__.
1998-03-09 04:46:26 +00:00
#if defined(__alpha__)
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
new_thread->saved_jmp_buf[0]._jb[2] = (long) _thread_start;
new_thread->saved_jmp_buf[0]._jb[4 + R_RA] = 0;
new_thread->saved_jmp_buf[0]._jb[4 + R_T12] = (long) _thread_start;
Add FreeBSD/Alpha code to initialise a jmpbuf for a created thread. Change a bunch of __alpha references to __alpha__.
1998-03-09 04:46:26 +00:00
#else
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
new_thread->saved_jmp_buf[0]._jb[0] = (long) _thread_start;
Add FreeBSD/Alpha code to initialise a jmpbuf for a created thread. Change a bunch of __alpha references to __alpha__.
1998-03-09 04:46:26 +00:00
#endif
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
#elif defined(__NetBSD__)
Add FreeBSD/Alpha code to initialise a jmpbuf for a created thread. Change a bunch of __alpha references to __alpha__.
1998-03-09 04:46:26 +00:00
#if defined(__alpha__)
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
new_thread->saved_jmp_buf[2] = (long) _thread_start;
new_thread->saved_jmp_buf[4 + R_RA] = 0;
new_thread->saved_jmp_buf[4 + R_T12] = (long) _thread_start;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
#else
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
new_thread->saved_jmp_buf[0] = (long) _thread_start;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
#endif
#else
#error "Don't recognize this operating system!"
#endif
/* The stack starts high and builds down: */
#if defined(__FreeBSD__)
Add FreeBSD/Alpha code to initialise a jmpbuf for a created thread. Change a bunch of __alpha references to __alpha__.
1998-03-09 04:46:26 +00:00
#if defined(__alpha__)
new_thread->saved_jmp_buf[0]._jb[4 + R_SP] = (long) new_thread->stack + pattr->stacksize_attr - sizeof(double);
#else
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
new_thread->saved_jmp_buf[0]._jb[2] = (int) (new_thread->stack + pattr->stacksize_attr - sizeof(double));
Add FreeBSD/Alpha code to initialise a jmpbuf for a created thread. Change a bunch of __alpha references to __alpha__.
1998-03-09 04:46:26 +00:00
#endif
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
#elif defined(__NetBSD__)
Add FreeBSD/Alpha code to initialise a jmpbuf for a created thread. Change a bunch of __alpha references to __alpha__.
1998-03-09 04:46:26 +00:00
#if defined(__alpha__)
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
new_thread->saved_jmp_buf[4 + R_SP] = (long) new_thread->stack + pattr->stacksize_attr - sizeof(double);
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
#else
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
new_thread->saved_jmp_buf[2] = (long) new_thread->stack + pattr->stacksize_attr - sizeof(double);
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
#endif
#else
#error "Don't recognize this operating system!"
#endif
/* Copy the thread attributes: */
Submitted by: John Birrell <cimaxp1!jb@werple.net.au> Here are the diffs for libc_r to get it one step closer to P1003.1c These make most of the thread/mutex/condvar structures opaque to the user. There are three functions which have been renamed with _np suffixes because they are extensions to P1003.1c (I did them for JAVA, which needs to suspend/resume threads and also start threads suspended). I've created a new header (pthread_np.h) for the non-POSIX stuff. The egrep tags stuff in /usr/src/lib/libc_r/Makefile that I uncommented doesn't work. I think its best to delete it. I don't think libc_r needs tags anyway, 'cause most of the source is in libc which does have tags. also: Here's the first batch of man pages for the thread functions. The diff to /usr/src/lib/libc_r/Makefile removes some stuff that was inherited from /usr/src/lib/libc/Makefile that should only be done with libc. also: I should have sent this diff with the pthread(3) man page. It allows people to type make -DWANT_LIBC_R world to get libc_r built with the rest of the world. I put this in the pthread(3) man page. The default is still not to build libc_r. also: The diff attached adds a pthread(3) man page to /usr/src/share/man/man3. The idea is that without libc_r installed, this man page will give people enough info to know that they have to build libc_r.
1996-08-20 08:22:01 +00:00
memcpy(&new_thread->attr, pattr, sizeof(struct pthread_attr));
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/*
* Check if this thread is to inherit the scheduling
* attributes from its parent:
*/
if (new_thread->attr.flags & PTHREAD_INHERIT_SCHED) {
/* Copy the scheduling attributes: */
[ The author's description... ] o Runnable threads are now maintained in priority queues. The implementation requires two things: 1.) The priority queues must be protected during insertion and removal of threads. Since the kernel scheduler must modify the priority queues, a spinlock for protection cannot be used. The functions _thread_kern_sched_defer() and _thread_kern_sched_undefer() were added to {un}defer kernel scheduler activation. 2.) A thread (active) priority change can be performed only when the thread is removed from the priority queue. The implementation uses a threads active priority when inserting it into the queue. A by-product is that thread switches are much faster. A separate queue is used for waiting and/or blocked threads, and it is searched at most 2 times in the kernel scheduler when there are active threads. It should be possible to reduce this to once by combining polling of threads waiting on I/O with the loop that looks for timed out threads and the minimum timeout value. o Functions to defer kernel scheduler activation were added. These are _thread_kern_sched_defer() and _thread_kern_sched_undefer() and may be called recursively. These routines do not block the scheduling signal, but latch its occurrence. The signal handler will not call the kernel scheduler when the running thread has deferred scheduling, but it will be called when running thread undefers scheduling. o Added support for _POSIX_THREAD_PRIORITY_SCHEDULING. All the POSIX routines required by this should now be implemented. One note, SCHED_OTHER, SCHED_FIFO, and SCHED_RR are required to be defined by including pthread.h. These defines are currently in sched.h. I modified pthread.h to include sched.h but don't know if this is the proper thing to do. o Added support for priority protection and inheritence mutexes. This allows definition of _POSIX_THREAD_PRIO_PROTECT and _POSIX_THREAD_PRIO_INHERIT. o Added additional error checks required by POSIX for mutexes and condition variables. o Provided a wrapper for sigpending which is marked as a hidden syscall. o Added a non-portable function as a debugging aid to allow an application to monitor thread context switches. An application can install a routine that gets called everytime a thread (explicitly created by the application) gets context switched. The routine gets passed the pthread IDs of the threads that are being switched in and out. Submitted by: Dan Eischen <eischen@vigrid.com> Changes by me: o Added a PS_SPINBLOCK state to deal with the priority inversion problem most often (I think) seen by threads calling malloc/free/realloc. o Dispatch signals to the running thread directly rather than at a context switch to avoid the situation where the switch never occurs.
1999-03-23 05:07:56 +00:00
new_thread->base_priority = _thread_run->base_priority;
new_thread->attr.prio = _thread_run->base_priority;
new_thread->attr.sched_policy = _thread_run->attr.sched_policy;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
} else {
/*
* Use just the thread priority, leaving the
* other scheduling attributes as their
* default values:
*/
[ The author's description... ] o Runnable threads are now maintained in priority queues. The implementation requires two things: 1.) The priority queues must be protected during insertion and removal of threads. Since the kernel scheduler must modify the priority queues, a spinlock for protection cannot be used. The functions _thread_kern_sched_defer() and _thread_kern_sched_undefer() were added to {un}defer kernel scheduler activation. 2.) A thread (active) priority change can be performed only when the thread is removed from the priority queue. The implementation uses a threads active priority when inserting it into the queue. A by-product is that thread switches are much faster. A separate queue is used for waiting and/or blocked threads, and it is searched at most 2 times in the kernel scheduler when there are active threads. It should be possible to reduce this to once by combining polling of threads waiting on I/O with the loop that looks for timed out threads and the minimum timeout value. o Functions to defer kernel scheduler activation were added. These are _thread_kern_sched_defer() and _thread_kern_sched_undefer() and may be called recursively. These routines do not block the scheduling signal, but latch its occurrence. The signal handler will not call the kernel scheduler when the running thread has deferred scheduling, but it will be called when running thread undefers scheduling. o Added support for _POSIX_THREAD_PRIORITY_SCHEDULING. All the POSIX routines required by this should now be implemented. One note, SCHED_OTHER, SCHED_FIFO, and SCHED_RR are required to be defined by including pthread.h. These defines are currently in sched.h. I modified pthread.h to include sched.h but don't know if this is the proper thing to do. o Added support for priority protection and inheritence mutexes. This allows definition of _POSIX_THREAD_PRIO_PROTECT and _POSIX_THREAD_PRIO_INHERIT. o Added additional error checks required by POSIX for mutexes and condition variables. o Provided a wrapper for sigpending which is marked as a hidden syscall. o Added a non-portable function as a debugging aid to allow an application to monitor thread context switches. An application can install a routine that gets called everytime a thread (explicitly created by the application) gets context switched. The routine gets passed the pthread IDs of the threads that are being switched in and out. Submitted by: Dan Eischen <eischen@vigrid.com> Changes by me: o Added a PS_SPINBLOCK state to deal with the priority inversion problem most often (I think) seen by threads calling malloc/free/realloc. o Dispatch signals to the running thread directly rather than at a context switch to avoid the situation where the switch never occurs.
1999-03-23 05:07:56 +00:00
new_thread->base_priority = new_thread->attr.prio;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
}
[ The author's description... ] o Runnable threads are now maintained in priority queues. The implementation requires two things: 1.) The priority queues must be protected during insertion and removal of threads. Since the kernel scheduler must modify the priority queues, a spinlock for protection cannot be used. The functions _thread_kern_sched_defer() and _thread_kern_sched_undefer() were added to {un}defer kernel scheduler activation. 2.) A thread (active) priority change can be performed only when the thread is removed from the priority queue. The implementation uses a threads active priority when inserting it into the queue. A by-product is that thread switches are much faster. A separate queue is used for waiting and/or blocked threads, and it is searched at most 2 times in the kernel scheduler when there are active threads. It should be possible to reduce this to once by combining polling of threads waiting on I/O with the loop that looks for timed out threads and the minimum timeout value. o Functions to defer kernel scheduler activation were added. These are _thread_kern_sched_defer() and _thread_kern_sched_undefer() and may be called recursively. These routines do not block the scheduling signal, but latch its occurrence. The signal handler will not call the kernel scheduler when the running thread has deferred scheduling, but it will be called when running thread undefers scheduling. o Added support for _POSIX_THREAD_PRIORITY_SCHEDULING. All the POSIX routines required by this should now be implemented. One note, SCHED_OTHER, SCHED_FIFO, and SCHED_RR are required to be defined by including pthread.h. These defines are currently in sched.h. I modified pthread.h to include sched.h but don't know if this is the proper thing to do. o Added support for priority protection and inheritence mutexes. This allows definition of _POSIX_THREAD_PRIO_PROTECT and _POSIX_THREAD_PRIO_INHERIT. o Added additional error checks required by POSIX for mutexes and condition variables. o Provided a wrapper for sigpending which is marked as a hidden syscall. o Added a non-portable function as a debugging aid to allow an application to monitor thread context switches. An application can install a routine that gets called everytime a thread (explicitly created by the application) gets context switched. The routine gets passed the pthread IDs of the threads that are being switched in and out. Submitted by: Dan Eischen <eischen@vigrid.com> Changes by me: o Added a PS_SPINBLOCK state to deal with the priority inversion problem most often (I think) seen by threads calling malloc/free/realloc. o Dispatch signals to the running thread directly rather than at a context switch to avoid the situation where the switch never occurs.
1999-03-23 05:07:56 +00:00
new_thread->active_priority = new_thread->base_priority;
new_thread->inherited_priority = 0;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Initialise the join queue for the new thread: */
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
TAILQ_INIT(&(new_thread->join_queue));
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
[ The author's description... ] o Runnable threads are now maintained in priority queues. The implementation requires two things: 1.) The priority queues must be protected during insertion and removal of threads. Since the kernel scheduler must modify the priority queues, a spinlock for protection cannot be used. The functions _thread_kern_sched_defer() and _thread_kern_sched_undefer() were added to {un}defer kernel scheduler activation. 2.) A thread (active) priority change can be performed only when the thread is removed from the priority queue. The implementation uses a threads active priority when inserting it into the queue. A by-product is that thread switches are much faster. A separate queue is used for waiting and/or blocked threads, and it is searched at most 2 times in the kernel scheduler when there are active threads. It should be possible to reduce this to once by combining polling of threads waiting on I/O with the loop that looks for timed out threads and the minimum timeout value. o Functions to defer kernel scheduler activation were added. These are _thread_kern_sched_defer() and _thread_kern_sched_undefer() and may be called recursively. These routines do not block the scheduling signal, but latch its occurrence. The signal handler will not call the kernel scheduler when the running thread has deferred scheduling, but it will be called when running thread undefers scheduling. o Added support for _POSIX_THREAD_PRIORITY_SCHEDULING. All the POSIX routines required by this should now be implemented. One note, SCHED_OTHER, SCHED_FIFO, and SCHED_RR are required to be defined by including pthread.h. These defines are currently in sched.h. I modified pthread.h to include sched.h but don't know if this is the proper thing to do. o Added support for priority protection and inheritence mutexes. This allows definition of _POSIX_THREAD_PRIO_PROTECT and _POSIX_THREAD_PRIO_INHERIT. o Added additional error checks required by POSIX for mutexes and condition variables. o Provided a wrapper for sigpending which is marked as a hidden syscall. o Added a non-portable function as a debugging aid to allow an application to monitor thread context switches. An application can install a routine that gets called everytime a thread (explicitly created by the application) gets context switched. The routine gets passed the pthread IDs of the threads that are being switched in and out. Submitted by: Dan Eischen <eischen@vigrid.com> Changes by me: o Added a PS_SPINBLOCK state to deal with the priority inversion problem most often (I think) seen by threads calling malloc/free/realloc. o Dispatch signals to the running thread directly rather than at a context switch to avoid the situation where the switch never occurs.
1999-03-23 05:07:56 +00:00
/* Initialize the mutex queue: */
TAILQ_INIT(&new_thread->mutexq);
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Initialise hooks in the thread structure: */
new_thread->specific_data = NULL;
new_thread->cleanup = NULL;
new_thread->flags = 0;
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
new_thread->poll_data.nfds = 0;
new_thread->poll_data.fds = NULL;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
/*
* Defer signals to protect the scheduling queues
* from access by the signal handler:
*/
_thread_kern_sig_defer();
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
Move the cleanup code that frees memory allocated for a dead thread from the thread kernel into a garbage collector thread which is started when the fisrt thread is created (other than the initial thread). This removes the window of opportunity where a context switch will cause a thread that has locked the malloc spinlock, to enter the thread kernel, find there is a dead thread and try to free memory, therefore trying to lock the malloc spinlock against itself. The garbage collector thread acts just like any other thread, so instead of having a spinlock to control accesses to the dead thread list, it uses a mutex and a condition variable so that it can happily wait to be signalled when a thread exists.
1998-09-30 06:36:56 +00:00
/*
* Check if the garbage collector thread
* needs to be started.
*/
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
f_gc = (TAILQ_FIRST(&_thread_list) == _thread_initial);
Move the cleanup code that frees memory allocated for a dead thread from the thread kernel into a garbage collector thread which is started when the fisrt thread is created (other than the initial thread). This removes the window of opportunity where a context switch will cause a thread that has locked the malloc spinlock, to enter the thread kernel, find there is a dead thread and try to free memory, therefore trying to lock the malloc spinlock against itself. The garbage collector thread acts just like any other thread, so instead of having a spinlock to control accesses to the dead thread list, it uses a mutex and a condition variable so that it can happily wait to be signalled when a thread exists.
1998-09-30 06:36:56 +00:00
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Add the thread to the linked list of all threads: */
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
TAILQ_INSERT_HEAD(&_thread_list, new_thread, tle);
[ The author's description... ] o Runnable threads are now maintained in priority queues. The implementation requires two things: 1.) The priority queues must be protected during insertion and removal of threads. Since the kernel scheduler must modify the priority queues, a spinlock for protection cannot be used. The functions _thread_kern_sched_defer() and _thread_kern_sched_undefer() were added to {un}defer kernel scheduler activation. 2.) A thread (active) priority change can be performed only when the thread is removed from the priority queue. The implementation uses a threads active priority when inserting it into the queue. A by-product is that thread switches are much faster. A separate queue is used for waiting and/or blocked threads, and it is searched at most 2 times in the kernel scheduler when there are active threads. It should be possible to reduce this to once by combining polling of threads waiting on I/O with the loop that looks for timed out threads and the minimum timeout value. o Functions to defer kernel scheduler activation were added. These are _thread_kern_sched_defer() and _thread_kern_sched_undefer() and may be called recursively. These routines do not block the scheduling signal, but latch its occurrence. The signal handler will not call the kernel scheduler when the running thread has deferred scheduling, but it will be called when running thread undefers scheduling. o Added support for _POSIX_THREAD_PRIORITY_SCHEDULING. All the POSIX routines required by this should now be implemented. One note, SCHED_OTHER, SCHED_FIFO, and SCHED_RR are required to be defined by including pthread.h. These defines are currently in sched.h. I modified pthread.h to include sched.h but don't know if this is the proper thing to do. o Added support for priority protection and inheritence mutexes. This allows definition of _POSIX_THREAD_PRIO_PROTECT and _POSIX_THREAD_PRIO_INHERIT. o Added additional error checks required by POSIX for mutexes and condition variables. o Provided a wrapper for sigpending which is marked as a hidden syscall. o Added a non-portable function as a debugging aid to allow an application to monitor thread context switches. An application can install a routine that gets called everytime a thread (explicitly created by the application) gets context switched. The routine gets passed the pthread IDs of the threads that are being switched in and out. Submitted by: Dan Eischen <eischen@vigrid.com> Changes by me: o Added a PS_SPINBLOCK state to deal with the priority inversion problem most often (I think) seen by threads calling malloc/free/realloc. o Dispatch signals to the running thread directly rather than at a context switch to avoid the situation where the switch never occurs.
1999-03-23 05:07:56 +00:00
if (pattr->suspend == PTHREAD_CREATE_SUSPENDED) {
new_thread->state = PS_SUSPENDED;
PTHREAD_WAITQ_INSERT(new_thread);
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
}
else {
[ The author's description... ] o Runnable threads are now maintained in priority queues. The implementation requires two things: 1.) The priority queues must be protected during insertion and removal of threads. Since the kernel scheduler must modify the priority queues, a spinlock for protection cannot be used. The functions _thread_kern_sched_defer() and _thread_kern_sched_undefer() were added to {un}defer kernel scheduler activation. 2.) A thread (active) priority change can be performed only when the thread is removed from the priority queue. The implementation uses a threads active priority when inserting it into the queue. A by-product is that thread switches are much faster. A separate queue is used for waiting and/or blocked threads, and it is searched at most 2 times in the kernel scheduler when there are active threads. It should be possible to reduce this to once by combining polling of threads waiting on I/O with the loop that looks for timed out threads and the minimum timeout value. o Functions to defer kernel scheduler activation were added. These are _thread_kern_sched_defer() and _thread_kern_sched_undefer() and may be called recursively. These routines do not block the scheduling signal, but latch its occurrence. The signal handler will not call the kernel scheduler when the running thread has deferred scheduling, but it will be called when running thread undefers scheduling. o Added support for _POSIX_THREAD_PRIORITY_SCHEDULING. All the POSIX routines required by this should now be implemented. One note, SCHED_OTHER, SCHED_FIFO, and SCHED_RR are required to be defined by including pthread.h. These defines are currently in sched.h. I modified pthread.h to include sched.h but don't know if this is the proper thing to do. o Added support for priority protection and inheritence mutexes. This allows definition of _POSIX_THREAD_PRIO_PROTECT and _POSIX_THREAD_PRIO_INHERIT. o Added additional error checks required by POSIX for mutexes and condition variables. o Provided a wrapper for sigpending which is marked as a hidden syscall. o Added a non-portable function as a debugging aid to allow an application to monitor thread context switches. An application can install a routine that gets called everytime a thread (explicitly created by the application) gets context switched. The routine gets passed the pthread IDs of the threads that are being switched in and out. Submitted by: Dan Eischen <eischen@vigrid.com> Changes by me: o Added a PS_SPINBLOCK state to deal with the priority inversion problem most often (I think) seen by threads calling malloc/free/realloc. o Dispatch signals to the running thread directly rather than at a context switch to avoid the situation where the switch never occurs.
1999-03-23 05:07:56 +00:00
new_thread->state = PS_RUNNING;
PTHREAD_PRIOQ_INSERT_TAIL(new_thread);
}
/*
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
* Undefer and handle pending signals, yielding
* if necessary.
[ The author's description... ] o Runnable threads are now maintained in priority queues. The implementation requires two things: 1.) The priority queues must be protected during insertion and removal of threads. Since the kernel scheduler must modify the priority queues, a spinlock for protection cannot be used. The functions _thread_kern_sched_defer() and _thread_kern_sched_undefer() were added to {un}defer kernel scheduler activation. 2.) A thread (active) priority change can be performed only when the thread is removed from the priority queue. The implementation uses a threads active priority when inserting it into the queue. A by-product is that thread switches are much faster. A separate queue is used for waiting and/or blocked threads, and it is searched at most 2 times in the kernel scheduler when there are active threads. It should be possible to reduce this to once by combining polling of threads waiting on I/O with the loop that looks for timed out threads and the minimum timeout value. o Functions to defer kernel scheduler activation were added. These are _thread_kern_sched_defer() and _thread_kern_sched_undefer() and may be called recursively. These routines do not block the scheduling signal, but latch its occurrence. The signal handler will not call the kernel scheduler when the running thread has deferred scheduling, but it will be called when running thread undefers scheduling. o Added support for _POSIX_THREAD_PRIORITY_SCHEDULING. All the POSIX routines required by this should now be implemented. One note, SCHED_OTHER, SCHED_FIFO, and SCHED_RR are required to be defined by including pthread.h. These defines are currently in sched.h. I modified pthread.h to include sched.h but don't know if this is the proper thing to do. o Added support for priority protection and inheritence mutexes. This allows definition of _POSIX_THREAD_PRIO_PROTECT and _POSIX_THREAD_PRIO_INHERIT. o Added additional error checks required by POSIX for mutexes and condition variables. o Provided a wrapper for sigpending which is marked as a hidden syscall. o Added a non-portable function as a debugging aid to allow an application to monitor thread context switches. An application can install a routine that gets called everytime a thread (explicitly created by the application) gets context switched. The routine gets passed the pthread IDs of the threads that are being switched in and out. Submitted by: Dan Eischen <eischen@vigrid.com> Changes by me: o Added a PS_SPINBLOCK state to deal with the priority inversion problem most often (I think) seen by threads calling malloc/free/realloc. o Dispatch signals to the running thread directly rather than at a context switch to avoid the situation where the switch never occurs.
1999-03-23 05:07:56 +00:00
*/
In the words of the author: o The polling mechanism for I/O readiness was changed from select() to poll(). In additon, a wrapped version of poll() is now provided. o The wrapped select routine now converts each fd_set to a poll array so that the thread scheduler doesn't have to perform a bitwise search for selected fds each time file descriptors are polled for I/O readiness. o The thread scheduler was modified to use a new queue (_workq) for threads that need work. Threads waiting for I/O readiness and spinblocks are added to the work queue in addition to the waiting queue. This reduces the time spent forming/searching the array of file descriptors being polled. o The waiting queue (_waitingq) is now maintained in order of thread wakeup time. This allows the thread scheduler to find the nearest wakeup time by looking at the first thread in the queue instead of searching the entire queue. o Removed file descriptor locking for select/poll routines. An application should not rely on the threads library for providing this locking; if necessary, the application should use mutexes to protect selecting/polling of file descriptors. o Retrieve and use the kernel clock rate/resolution at startup instead of hardcoding the clock resolution to 10 msec (tested with kernel running at 1000 HZ). o All queues have been changed to use queue.h macros. These include the queues of all threads, dead threads, and threads waiting for file descriptor locks. o Added reinitialization of the GC mutex and condition variable after a fork. Also prevented reallocation of the ready queue after a fork. o Prevented the wrapped close routine from closing the thread kernel pipes. o Initialized file descriptor table for stdio entries at thread init. o Provided additional flags to indicate to what queues threads belong. o Moved TAILQ initialization for statically allocated mutex and condition variables to after the spinlock. o Added dispatching of signals to pthread_kill. Removing the dispatching of signals from thread activation broke sigsuspend when pthread_kill was used to send a signal to a thread. o Temporarily set the state of a thread to PS_SUSPENDED when it is first created and placed in the list of threads so that it will not be accidentally scheduled before becoming a member of one of the scheduling queues. o Change the signal handler to queue signals to the thread kernel pipe if the scheduling queues are protected. When scheduling queues are unprotected, signals are then dequeued and handled. o Ensured that all installed signal handlers block the scheduling signal and that the scheduling signal handler blocks all other signals. This ensures that the signal handler is only interruptible for and by non-scheduling signals. An atomic lock is used to decide which instance of the signal handler will handle pending signals. o Removed _lock_thread_list and _unlock_thread_list as they are no longer used to protect the thread list. o Added missing RCS IDs to modified files. o Added checks for appropriate queue membership and activity when adding, removing, and searching the scheduling queues. These checks add very little overhead and are enabled when compiled with _PTHREADS_INVARIANTS defined. Suggested and implemented by Tor Egge with some modification by me. o Close a race condition in uthread_close. (Tor Egge) o Protect the scheduling queues while modifying them in pthread_cond_signal and _thread_fd_unlock. (Tor Egge) o Ensure that when a thread gets a mutex, the mutex is on that threads list of owned mutexes. (Tor Egge) o Set the kernel-in-scheduler flag in _thread_kern_sched_state and _thread_kern_sched_state_unlock to prevent a scheduling signal from calling the scheduler again. (Tor Egge) o Don't use TAILQ_FOREACH macro while searching the waiting queue for threads in a sigwait state, because a change of state destroys the TAILQ link. It is actually safe to do so, though, because once a sigwaiting thread is found, the loop ends and the function returns. (Tor Egge) o When dispatching signals to threads, make the thread inherit the signal deferral flag of the currently running thread. (Tor Egge) Submitted by: Daniel Eischen <eischen@vigrid.com> and Tor Egge <Tor.Egge@fast.no>
1999-06-20 08:28:48 +00:00
_thread_kern_sig_undefer();
[ The author's description... ] o Runnable threads are now maintained in priority queues. The implementation requires two things: 1.) The priority queues must be protected during insertion and removal of threads. Since the kernel scheduler must modify the priority queues, a spinlock for protection cannot be used. The functions _thread_kern_sched_defer() and _thread_kern_sched_undefer() were added to {un}defer kernel scheduler activation. 2.) A thread (active) priority change can be performed only when the thread is removed from the priority queue. The implementation uses a threads active priority when inserting it into the queue. A by-product is that thread switches are much faster. A separate queue is used for waiting and/or blocked threads, and it is searched at most 2 times in the kernel scheduler when there are active threads. It should be possible to reduce this to once by combining polling of threads waiting on I/O with the loop that looks for timed out threads and the minimum timeout value. o Functions to defer kernel scheduler activation were added. These are _thread_kern_sched_defer() and _thread_kern_sched_undefer() and may be called recursively. These routines do not block the scheduling signal, but latch its occurrence. The signal handler will not call the kernel scheduler when the running thread has deferred scheduling, but it will be called when running thread undefers scheduling. o Added support for _POSIX_THREAD_PRIORITY_SCHEDULING. All the POSIX routines required by this should now be implemented. One note, SCHED_OTHER, SCHED_FIFO, and SCHED_RR are required to be defined by including pthread.h. These defines are currently in sched.h. I modified pthread.h to include sched.h but don't know if this is the proper thing to do. o Added support for priority protection and inheritence mutexes. This allows definition of _POSIX_THREAD_PRIO_PROTECT and _POSIX_THREAD_PRIO_INHERIT. o Added additional error checks required by POSIX for mutexes and condition variables. o Provided a wrapper for sigpending which is marked as a hidden syscall. o Added a non-portable function as a debugging aid to allow an application to monitor thread context switches. An application can install a routine that gets called everytime a thread (explicitly created by the application) gets context switched. The routine gets passed the pthread IDs of the threads that are being switched in and out. Submitted by: Dan Eischen <eischen@vigrid.com> Changes by me: o Added a PS_SPINBLOCK state to deal with the priority inversion problem most often (I think) seen by threads calling malloc/free/realloc. o Dispatch signals to the running thread directly rather than at a context switch to avoid the situation where the switch never occurs.
1999-03-23 05:07:56 +00:00
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Return a pointer to the thread structure: */
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
(*thread) = new_thread;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
Change signal model to match POSIX (i.e. one set of signal handlers for the process, not a separate set for each thread). By default, the process now only has signal handlers installed for SIGVTALRM, SIGINFO and SIGCHLD. The thread kernel signal handler is installed for other signals on demand. This means that SIG_IGN and SIG_DFL processing is now left to the kernel, not the thread kernel. Change the signal dispatch to no longer use a signal thread, and call the signal handler using the stack of the thread that has the signal pending. Change the atomic lock method to use test-and-set asm code with a yield if blocked. This introduces separate locks for each type of object instead of blocking signals to prevent a context switch. It was this blocking of signals that caused the performance degradation the people have noted. This is a *big* change!
1998-04-29 09:59:34 +00:00
/* Schedule the new user thread: */
_thread_kern_sched(NULL);
Move the cleanup code that frees memory allocated for a dead thread from the thread kernel into a garbage collector thread which is started when the fisrt thread is created (other than the initial thread). This removes the window of opportunity where a context switch will cause a thread that has locked the malloc spinlock, to enter the thread kernel, find there is a dead thread and try to free memory, therefore trying to lock the malloc spinlock against itself. The garbage collector thread acts just like any other thread, so instead of having a spinlock to control accesses to the dead thread list, it uses a mutex and a condition variable so that it can happily wait to be signalled when a thread exists.
1998-09-30 06:36:56 +00:00
/*
* Start a garbage collector thread
* if necessary.
*/
if (f_gc && pthread_create(&gc_thread,NULL,
_thread_gc,NULL) != 0)
PANIC("Can't create gc thread");
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
}
}
/* Return the status: */
return (ret);
}
void
_thread_start(void)
{
Fix some bugs in pthread scheduler: make pthread_yield() more reliable, threads always (I hope) preempted at least every 0.1 sec, as intended. PR: bin/7744 Submitted by: "Richard Seaman, Jr." <dick@tar.com>
1998-10-09 19:01:30 +00:00
/* We just left the scheduler via longjmp: */
_thread_kern_in_sched = 0;
Reviewed by: julian Submitted by: john birrel One version of the pthreads library another will follow with differnt actions under some cases.. not QUITE complete
1996-01-22 00:23:58 +00:00
/* Run the current thread's start routine with argument: */
pthread_exit(_thread_run->start_routine(_thread_run->arg));
/* This point should never be reached. */
PANIC("Thread has resumed after exit");
}
#endif
Reference in New Issue Copy Permalink