freebsd-dev/sys/kern/kern_synch.c

666 lines
17 KiB
C
Raw Normal View History

1994-05-24 10:09:53 +00:00
/*-
* Copyright (c) 1982, 1986, 1990, 1991, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* 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 the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS 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.
*
* @(#)kern_synch.c 8.9 (Berkeley) 5/19/95
1999-08-28 01:08:13 +00:00
* $FreeBSD$
1994-05-24 10:09:53 +00:00
*/
#include "opt_ddb.h"
#include "opt_ktrace.h"
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/condvar.h>
1994-05-24 10:09:53 +00:00
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/proc.h>
1994-05-24 10:09:53 +00:00
#include <sys/resourcevar.h>
#include <sys/sched.h>
#include <sys/signalvar.h>
Overhaul of the SMP code. Several portions of the SMP kernel support have been made machine independent and various other adjustments have been made to support Alpha SMP. - It splits the per-process portions of hardclock() and statclock() off into hardclock_process() and statclock_process() respectively. hardclock() and statclock() call the *_process() functions for the current process so that UP systems will run as before. For SMP systems, it is simply necessary to ensure that all other processors execute the *_process() functions when the main clock functions are triggered on one CPU by an interrupt. For the alpha 4100, clock interrupts are delievered in a staggered broadcast fashion, so we simply call hardclock/statclock on the boot CPU and call the *_process() functions on the secondaries. For x86, we call statclock and hardclock as usual and then call forward_hardclock/statclock in the MD code to send an IPI to cause the AP's to execute forwared_hardclock/statclock which then call the *_process() functions. - forward_signal() and forward_roundrobin() have been reworked to be MI and to involve less hackery. Now the cpu doing the forward sets any flags, etc. and sends a very simple IPI_AST to the other cpu(s). AST IPIs now just basically return so that they can execute ast() and don't bother with setting the astpending or needresched flags themselves. This also removes the loop in forward_signal() as sched_lock closes the race condition that the loop worked around. - need_resched(), resched_wanted() and clear_resched() have been changed to take a process to act on rather than assuming curproc so that they can be used to implement forward_roundrobin() as described above. - Various other SMP variables have been moved to a MI subr_smp.c and a new header sys/smp.h declares MI SMP variables and API's. The IPI API's from machine/ipl.h have moved to machine/smp.h which is included by sys/smp.h. - The globaldata_register() and globaldata_find() functions as well as the SLIST of globaldata structures has become MI and moved into subr_smp.c. Also, the globaldata list is only available if SMP support is compiled in. Reviewed by: jake, peter Looked over by: eivind
2001-04-27 19:28:25 +00:00
#include <sys/smp.h>
#include <sys/sx.h>
#include <sys/sysctl.h>
#include <sys/sysproto.h>
#include <sys/vmmeter.h>
#ifdef DDB
#include <ddb/ddb.h>
#endif
1994-05-24 10:09:53 +00:00
#ifdef KTRACE
#include <sys/uio.h>
1994-05-24 10:09:53 +00:00
#include <sys/ktrace.h>
#endif
#include <machine/cpu.h>
2002-03-19 21:25:46 +00:00
static void sched_setup(void *dummy);
SYSINIT(sched_setup, SI_SUB_KICK_SCHEDULER, SI_ORDER_FIRST, sched_setup, NULL)
int hogticks;
int lbolt;
1994-05-24 10:09:53 +00:00
static struct callout loadav_callout;
static struct callout lbolt_callout;
struct loadavg averunnable =
{ {0, 0, 0}, FSCALE }; /* load average, of runnable procs */
/*
* Constants for averages over 1, 5, and 15 minutes
* when sampling at 5 second intervals.
*/
static fixpt_t cexp[3] = {
0.9200444146293232 * FSCALE, /* exp(-1/12) */
0.9834714538216174 * FSCALE, /* exp(-1/60) */
0.9944598480048967 * FSCALE, /* exp(-1/180) */
};
/* kernel uses `FSCALE', userland (SHOULD) use kern.fscale */
static int fscale __unused = FSCALE;
SYSCTL_INT(_kern, OID_AUTO, fscale, CTLFLAG_RD, 0, FSCALE, "");
2002-03-19 21:25:46 +00:00
static void endtsleep(void *);
static void loadav(void *arg);
static void lboltcb(void *arg);
1994-05-24 10:09:53 +00:00
/*
* We're only looking at 7 bits of the address; everything is
* aligned to 4, lots of things are aligned to greater powers
* of 2. Shift right by 8, i.e. drop the bottom 256 worth.
*/
#define TABLESIZE 128
static TAILQ_HEAD(slpquehead, thread) slpque[TABLESIZE];
#define LOOKUP(x) (((intptr_t)(x) >> 8) & (TABLESIZE - 1))
1994-05-24 10:09:53 +00:00
void
sleepinit(void)
{
int i;
hogticks = (hz / 10) * 2; /* Default only. */
for (i = 0; i < TABLESIZE; i++)
TAILQ_INIT(&slpque[i]);
}
1994-05-24 10:09:53 +00:00
/*
* General sleep call. Suspends the current process until a wakeup is
* performed on the specified identifier. The process will then be made
* runnable with the specified priority. Sleeps at most timo/hz seconds
* (0 means no timeout). If pri includes PCATCH flag, signals are checked
* before and after sleeping, else signals are not checked. Returns 0 if
* awakened, EWOULDBLOCK if the timeout expires. If PCATCH is set and a
* signal needs to be delivered, ERESTART is returned if the current system
* call should be restarted if possible, and EINTR is returned if the system
* call should be interrupted by the signal (return EINTR).
*
* The mutex argument is exited before the caller is suspended, and
* entered before msleep returns. If priority includes the PDROP
* flag the mutex is not entered before returning.
1994-05-24 10:09:53 +00:00
*/
1994-05-24 10:09:53 +00:00
int
msleep(ident, mtx, priority, wmesg, timo)
1994-05-24 10:09:53 +00:00
void *ident;
struct mtx *mtx;
1994-05-24 10:09:53 +00:00
int priority, timo;
1997-11-21 11:37:03 +00:00
const char *wmesg;
1994-05-24 10:09:53 +00:00
{
struct thread *td = curthread;
struct proc *p = td->td_proc;
int sig, catch = priority & PCATCH;
int rval = 0;
WITNESS_SAVE_DECL(mtx);
1994-05-24 10:09:53 +00:00
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(1, 0);
1994-05-24 10:09:53 +00:00
#endif
Rework the witness code to work with sx locks as well as mutexes. - Introduce lock classes and lock objects. Each lock class specifies a name and set of flags (or properties) shared by all locks of a given type. Currently there are three lock classes: spin mutexes, sleep mutexes, and sx locks. A lock object specifies properties of an additional lock along with a lock name and all of the extra stuff needed to make witness work with a given lock. This abstract lock stuff is defined in sys/lock.h. The lockmgr constants, types, and prototypes have been moved to sys/lockmgr.h. For temporary backwards compatability, sys/lock.h includes sys/lockmgr.h. - Replace proc->p_spinlocks with a per-CPU list, PCPU(spinlocks), of spin locks held. By making this per-cpu, we do not have to jump through magic hoops to deal with sched_lock changing ownership during context switches. - Replace proc->p_heldmtx, formerly a list of held sleep mutexes, with proc->p_sleeplocks, which is a list of held sleep locks including sleep mutexes and sx locks. - Add helper macros for logging lock events via the KTR_LOCK KTR logging level so that the log messages are consistent. - Add some new flags that can be passed to mtx_init(): - MTX_NOWITNESS - specifies that this lock should be ignored by witness. This is used for the mutex that blocks a sx lock for example. - MTX_QUIET - this is not new, but you can pass this to mtx_init() now and no events will be logged for this lock, so that one doesn't have to change all the individual mtx_lock/unlock() operations. - All lock objects maintain an initialized flag. Use this flag to export a mtx_initialized() macro that can be safely called from drivers. Also, we on longer walk the all_mtx list if MUTEX_DEBUG is defined as witness performs the corresponding checks using the initialized flag. - The lock order reversal messages have been improved to output slightly more accurate file and line numbers.
2001-03-28 09:03:24 +00:00
WITNESS_SLEEP(0, &mtx->mtx_object);
KASSERT(timo != 0 || mtx_owned(&Giant) || mtx != NULL,
("sleeping without a mutex"));
/*
* If we are capable of async syscalls and there isn't already
* another one ready to return, start a new thread
* and queue it as ready to run. Note that there is danger here
* because we need to make sure that we don't sleep allocating
* the thread (recursion here might be bad).
* Hence the TDF_INMSLEEP flag.
*/
if (p->p_flag & P_KSES) {
/*
* Just don't bother if we are exiting
* and not the exiting thread or thread was marked as
* interrupted.
*/
if (catch &&
(((p->p_flag & P_WEXIT) && (p->p_singlethread != td)) ||
(td->td_flags & TDF_INTERRUPT))) {
td->td_flags &= ~TDF_INTERRUPT;
return (EINTR);
}
}
mtx_lock_spin(&sched_lock);
if (cold ) {
1994-05-24 10:09:53 +00:00
/*
* During autoconfiguration, just give interrupts
* a chance, then just return.
* Don't run any other procs or panic below,
1994-05-24 10:09:53 +00:00
* in case this is the idle process and already asleep.
*/
if (mtx != NULL && priority & PDROP)
Change the preemption code for software interrupt thread schedules and mutex releases to not require flags for the cases when preemption is not allowed: The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent switching to a higher priority thread on mutex releease and swi schedule, respectively when that switch is not safe. Now that the critical section API maintains a per-thread nesting count, the kernel can easily check whether or not it should switch without relying on flags from the programmer. This fixes a few bugs in that all current callers of swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from fast interrupt handlers and the swi_sched of softclock needed this flag. Note that to ensure that swi_sched()'s in clock and fast interrupt handlers do not switch, these handlers have to be explicitly wrapped in critical_enter/exit pairs. Presently, just wrapping the handlers is sufficient, but in the future with the fully preemptive kernel, the interrupt must be EOI'd before critical_exit() is called. (critical_exit() can switch due to a deferred preemption in a fully preemptive kernel.) I've tested the changes to the interrupt code on i386 and alpha. I have not tested ia64, but the interrupt code is almost identical to the alpha code, so I expect it will work fine. PowerPC and ARM do not yet have interrupt code in the tree so they shouldn't be broken. Sparc64 is broken, but that's been ok'd by jake and tmm who will be fixing the interrupt code for sparc64 shortly. Reviewed by: peter Tested on: i386, alpha
2002-01-05 08:47:13 +00:00
mtx_unlock(mtx);
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_unlock_spin(&sched_lock);
1994-05-24 10:09:53 +00:00
return (0);
}
Change the preemption code for software interrupt thread schedules and mutex releases to not require flags for the cases when preemption is not allowed: The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent switching to a higher priority thread on mutex releease and swi schedule, respectively when that switch is not safe. Now that the critical section API maintains a per-thread nesting count, the kernel can easily check whether or not it should switch without relying on flags from the programmer. This fixes a few bugs in that all current callers of swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from fast interrupt handlers and the swi_sched of softclock needed this flag. Note that to ensure that swi_sched()'s in clock and fast interrupt handlers do not switch, these handlers have to be explicitly wrapped in critical_enter/exit pairs. Presently, just wrapping the handlers is sufficient, but in the future with the fully preemptive kernel, the interrupt must be EOI'd before critical_exit() is called. (critical_exit() can switch due to a deferred preemption in a fully preemptive kernel.) I've tested the changes to the interrupt code on i386 and alpha. I have not tested ia64, but the interrupt code is almost identical to the alpha code, so I expect it will work fine. PowerPC and ARM do not yet have interrupt code in the tree so they shouldn't be broken. Sparc64 is broken, but that's been ok'd by jake and tmm who will be fixing the interrupt code for sparc64 shortly. Reviewed by: peter Tested on: i386, alpha
2002-01-05 08:47:13 +00:00
DROP_GIANT();
if (mtx != NULL) {
mtx_assert(mtx, MA_OWNED | MA_NOTRECURSED);
Rework the witness code to work with sx locks as well as mutexes. - Introduce lock classes and lock objects. Each lock class specifies a name and set of flags (or properties) shared by all locks of a given type. Currently there are three lock classes: spin mutexes, sleep mutexes, and sx locks. A lock object specifies properties of an additional lock along with a lock name and all of the extra stuff needed to make witness work with a given lock. This abstract lock stuff is defined in sys/lock.h. The lockmgr constants, types, and prototypes have been moved to sys/lockmgr.h. For temporary backwards compatability, sys/lock.h includes sys/lockmgr.h. - Replace proc->p_spinlocks with a per-CPU list, PCPU(spinlocks), of spin locks held. By making this per-cpu, we do not have to jump through magic hoops to deal with sched_lock changing ownership during context switches. - Replace proc->p_heldmtx, formerly a list of held sleep mutexes, with proc->p_sleeplocks, which is a list of held sleep locks including sleep mutexes and sx locks. - Add helper macros for logging lock events via the KTR_LOCK KTR logging level so that the log messages are consistent. - Add some new flags that can be passed to mtx_init(): - MTX_NOWITNESS - specifies that this lock should be ignored by witness. This is used for the mutex that blocks a sx lock for example. - MTX_QUIET - this is not new, but you can pass this to mtx_init() now and no events will be logged for this lock, so that one doesn't have to change all the individual mtx_lock/unlock() operations. - All lock objects maintain an initialized flag. Use this flag to export a mtx_initialized() macro that can be safely called from drivers. Also, we on longer walk the all_mtx list if MUTEX_DEBUG is defined as witness performs the corresponding checks using the initialized flag. - The lock order reversal messages have been improved to output slightly more accurate file and line numbers.
2001-03-28 09:03:24 +00:00
WITNESS_SAVE(&mtx->mtx_object, mtx);
Change the preemption code for software interrupt thread schedules and mutex releases to not require flags for the cases when preemption is not allowed: The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent switching to a higher priority thread on mutex releease and swi schedule, respectively when that switch is not safe. Now that the critical section API maintains a per-thread nesting count, the kernel can easily check whether or not it should switch without relying on flags from the programmer. This fixes a few bugs in that all current callers of swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from fast interrupt handlers and the swi_sched of softclock needed this flag. Note that to ensure that swi_sched()'s in clock and fast interrupt handlers do not switch, these handlers have to be explicitly wrapped in critical_enter/exit pairs. Presently, just wrapping the handlers is sufficient, but in the future with the fully preemptive kernel, the interrupt must be EOI'd before critical_exit() is called. (critical_exit() can switch due to a deferred preemption in a fully preemptive kernel.) I've tested the changes to the interrupt code on i386 and alpha. I have not tested ia64, but the interrupt code is almost identical to the alpha code, so I expect it will work fine. PowerPC and ARM do not yet have interrupt code in the tree so they shouldn't be broken. Sparc64 is broken, but that's been ok'd by jake and tmm who will be fixing the interrupt code for sparc64 shortly. Reviewed by: peter Tested on: i386, alpha
2002-01-05 08:47:13 +00:00
mtx_unlock(mtx);
if (priority & PDROP)
mtx = NULL;
}
KASSERT(p != NULL, ("msleep1"));
KASSERT(ident != NULL && TD_IS_RUNNING(td), ("msleep"));
CTR5(KTR_PROC, "msleep: thread %p (pid %d, %s) on %s (%p)",
td, p->p_pid, p->p_comm, wmesg, ident);
td->td_wchan = ident;
td->td_wmesg = wmesg;
TAILQ_INSERT_TAIL(&slpque[LOOKUP(ident)], td, td_slpq);
TD_SET_ON_SLEEPQ(td);
1994-05-24 10:09:53 +00:00
if (timo)
callout_reset(&td->td_slpcallout, timo, endtsleep, td);
1994-05-24 10:09:53 +00:00
/*
* We put ourselves on the sleep queue and start our timeout
* before calling thread_suspend_check, as we could stop there, and
* a wakeup or a SIGCONT (or both) could occur while we were stopped.
1994-05-24 10:09:53 +00:00
* without resuming us, thus we must be ready for sleep
* when cursig is called. If the wakeup happens while we're
* stopped, td->td_wchan will be 0 upon return from cursig.
1994-05-24 10:09:53 +00:00
*/
if (catch) {
CTR3(KTR_PROC, "msleep caught: thread %p (pid %d, %s)", td,
p->p_pid, p->p_comm);
td->td_flags |= TDF_SINTR;
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_unlock_spin(&sched_lock);
PROC_LOCK(p);
sig = cursig(td);
if (sig == 0 && thread_suspend_check(1))
sig = SIGSTOP;
mtx_lock_spin(&sched_lock);
Change the preemption code for software interrupt thread schedules and mutex releases to not require flags for the cases when preemption is not allowed: The purpose of the MTX_NOSWITCH and SWI_NOSWITCH flags is to prevent switching to a higher priority thread on mutex releease and swi schedule, respectively when that switch is not safe. Now that the critical section API maintains a per-thread nesting count, the kernel can easily check whether or not it should switch without relying on flags from the programmer. This fixes a few bugs in that all current callers of swi_sched() used SWI_NOSWITCH, when in fact, only the ones called from fast interrupt handlers and the swi_sched of softclock needed this flag. Note that to ensure that swi_sched()'s in clock and fast interrupt handlers do not switch, these handlers have to be explicitly wrapped in critical_enter/exit pairs. Presently, just wrapping the handlers is sufficient, but in the future with the fully preemptive kernel, the interrupt must be EOI'd before critical_exit() is called. (critical_exit() can switch due to a deferred preemption in a fully preemptive kernel.) I've tested the changes to the interrupt code on i386 and alpha. I have not tested ia64, but the interrupt code is almost identical to the alpha code, so I expect it will work fine. PowerPC and ARM do not yet have interrupt code in the tree so they shouldn't be broken. Sparc64 is broken, but that's been ok'd by jake and tmm who will be fixing the interrupt code for sparc64 shortly. Reviewed by: peter Tested on: i386, alpha
2002-01-05 08:47:13 +00:00
PROC_UNLOCK(p);
if (sig != 0) {
if (TD_ON_SLEEPQ(td))
unsleep(td);
} else if (!TD_ON_SLEEPQ(td))
1994-05-24 10:09:53 +00:00
catch = 0;
} else
1994-05-24 10:09:53 +00:00
sig = 0;
/*
* Let the scheduler know we're about to voluntarily go to sleep.
*/
sched_sleep(td, priority & PRIMASK);
if (TD_ON_SLEEPQ(td)) {
p->p_stats->p_ru.ru_nvcsw++;
TD_SET_SLEEPING(td);
mi_switch();
}
/*
* We're awake from voluntary sleep.
*/
CTR3(KTR_PROC, "msleep resume: thread %p (pid %d, %s)", td, p->p_pid,
p->p_comm);
KASSERT(TD_IS_RUNNING(td), ("running but not TDS_RUNNING"));
td->td_flags &= ~TDF_SINTR;
if (td->td_flags & TDF_TIMEOUT) {
td->td_flags &= ~TDF_TIMEOUT;
if (sig == 0)
rval = EWOULDBLOCK;
} else if (td->td_flags & TDF_TIMOFAIL) {
td->td_flags &= ~TDF_TIMOFAIL;
} else if (timo && callout_stop(&td->td_slpcallout) == 0) {
/*
* This isn't supposed to be pretty. If we are here, then
* the endtsleep() callout is currently executing on another
* CPU and is either spinning on the sched_lock or will be
* soon. If we don't synchronize here, there is a chance
* that this process may msleep() again before the callout
* has a chance to run and the callout may end up waking up
* the wrong msleep(). Yuck.
*/
TD_SET_SLEEPING(td);
p->p_stats->p_ru.ru_nivcsw++;
mi_switch();
td->td_flags &= ~TDF_TIMOFAIL;
}
if ((td->td_flags & TDF_INTERRUPT) && (priority & PCATCH) &&
(rval == 0)) {
td->td_flags &= ~TDF_INTERRUPT;
rval = EINTR;
}
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_unlock_spin(&sched_lock);
if (rval == 0 && catch) {
PROC_LOCK(p);
/* XXX: shouldn't we always be calling cursig() */
if (sig != 0 || (sig = cursig(td))) {
if (SIGISMEMBER(p->p_sigacts->ps_sigintr, sig))
rval = EINTR;
else
rval = ERESTART;
}
PROC_UNLOCK(p);
1994-05-24 10:09:53 +00:00
}
#ifdef KTRACE
if (KTRPOINT(td, KTR_CSW))
ktrcsw(0, 0);
1994-05-24 10:09:53 +00:00
#endif
PICKUP_GIANT();
if (mtx != NULL) {
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_lock(mtx);
Rework the witness code to work with sx locks as well as mutexes. - Introduce lock classes and lock objects. Each lock class specifies a name and set of flags (or properties) shared by all locks of a given type. Currently there are three lock classes: spin mutexes, sleep mutexes, and sx locks. A lock object specifies properties of an additional lock along with a lock name and all of the extra stuff needed to make witness work with a given lock. This abstract lock stuff is defined in sys/lock.h. The lockmgr constants, types, and prototypes have been moved to sys/lockmgr.h. For temporary backwards compatability, sys/lock.h includes sys/lockmgr.h. - Replace proc->p_spinlocks with a per-CPU list, PCPU(spinlocks), of spin locks held. By making this per-cpu, we do not have to jump through magic hoops to deal with sched_lock changing ownership during context switches. - Replace proc->p_heldmtx, formerly a list of held sleep mutexes, with proc->p_sleeplocks, which is a list of held sleep locks including sleep mutexes and sx locks. - Add helper macros for logging lock events via the KTR_LOCK KTR logging level so that the log messages are consistent. - Add some new flags that can be passed to mtx_init(): - MTX_NOWITNESS - specifies that this lock should be ignored by witness. This is used for the mutex that blocks a sx lock for example. - MTX_QUIET - this is not new, but you can pass this to mtx_init() now and no events will be logged for this lock, so that one doesn't have to change all the individual mtx_lock/unlock() operations. - All lock objects maintain an initialized flag. Use this flag to export a mtx_initialized() macro that can be safely called from drivers. Also, we on longer walk the all_mtx list if MUTEX_DEBUG is defined as witness performs the corresponding checks using the initialized flag. - The lock order reversal messages have been improved to output slightly more accurate file and line numbers.
2001-03-28 09:03:24 +00:00
WITNESS_RESTORE(&mtx->mtx_object, mtx);
}
return (rval);
1994-05-24 10:09:53 +00:00
}
/*
* Implement timeout for msleep()
*
1994-05-24 10:09:53 +00:00
* If process hasn't been awakened (wchan non-zero),
* set timeout flag and undo the sleep. If proc
* is stopped, just unsleep so it will remain stopped.
2000-12-01 04:55:52 +00:00
* MP-safe, called without the Giant mutex.
1994-05-24 10:09:53 +00:00
*/
1997-11-22 08:35:46 +00:00
static void
1994-05-24 10:09:53 +00:00
endtsleep(arg)
void *arg;
{
register struct thread *td = arg;
1994-05-24 10:09:53 +00:00
CTR3(KTR_PROC, "endtsleep: thread %p (pid %d, %s)",
td, td->td_proc->p_pid, td->td_proc->p_comm);
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_lock_spin(&sched_lock);
/*
* This is the other half of the synchronization with msleep()
* described above. If the TDS_TIMEOUT flag is set, we lost the
* race and just need to put the process back on the runqueue.
*/
if (TD_ON_SLEEPQ(td)) {
TAILQ_REMOVE(&slpque[LOOKUP(td->td_wchan)], td, td_slpq);
TD_CLR_ON_SLEEPQ(td);
td->td_flags |= TDF_TIMEOUT;
} else {
td->td_flags |= TDF_TIMOFAIL;
}
TD_CLR_SLEEPING(td);
setrunnable(td);
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_unlock_spin(&sched_lock);
1994-05-24 10:09:53 +00:00
}
/*
* Abort a thread, as if an interrupt had occured. Only abort
* interruptable waits (unfortunatly it isn't only safe to abort others).
* This is about identical to cv_abort().
* Think about merging them?
* Also, whatever the signal code does...
*/
void
abortsleep(struct thread *td)
{
mtx_assert(&sched_lock, MA_OWNED);
/*
* If the TDF_TIMEOUT flag is set, just leave. A
* timeout is scheduled anyhow.
*/
if ((td->td_flags & (TDF_TIMEOUT | TDF_SINTR)) == TDF_SINTR) {
if (TD_ON_SLEEPQ(td)) {
unsleep(td);
TD_CLR_SLEEPING(td);
setrunnable(td);
}
}
}
1994-05-24 10:09:53 +00:00
/*
* Remove a process from its wait queue
*/
void
unsleep(struct thread *td)
1994-05-24 10:09:53 +00:00
{
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_lock_spin(&sched_lock);
if (TD_ON_SLEEPQ(td)) {
TAILQ_REMOVE(&slpque[LOOKUP(td->td_wchan)], td, td_slpq);
TD_CLR_ON_SLEEPQ(td);
1994-05-24 10:09:53 +00:00
}
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_unlock_spin(&sched_lock);
1994-05-24 10:09:53 +00:00
}
/*
2001-07-06 01:16:43 +00:00
* Make all processes sleeping on the specified identifier runnable.
1994-05-24 10:09:53 +00:00
*/
void
2001-07-06 01:16:43 +00:00
wakeup(ident)
1994-05-24 10:09:53 +00:00
register void *ident;
{
register struct slpquehead *qp;
register struct thread *td;
struct thread *ntd;
struct proc *p;
1994-05-24 10:09:53 +00:00
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_lock_spin(&sched_lock);
1994-05-24 10:09:53 +00:00
qp = &slpque[LOOKUP(ident)];
restart:
for (td = TAILQ_FIRST(qp); td != NULL; td = ntd) {
ntd = TAILQ_NEXT(td, td_slpq);
if (td->td_wchan == ident) {
unsleep(td);
TD_CLR_SLEEPING(td);
setrunnable(td);
p = td->td_proc;
CTR3(KTR_PROC,"wakeup: thread %p (pid %d, %s)",
td, p->p_pid, p->p_comm);
goto restart;
}
}
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_unlock_spin(&sched_lock);
}
/*
* Make a process sleeping on the specified identifier runnable.
2000-05-07 05:09:45 +00:00
* May wake more than one process if a target process is currently
2001-07-06 01:16:43 +00:00
* swapped out.
*/
void
2001-07-06 01:16:43 +00:00
wakeup_one(ident)
register void *ident;
{
register struct slpquehead *qp;
register struct thread *td;
register struct proc *p;
struct thread *ntd;
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_lock_spin(&sched_lock);
qp = &slpque[LOOKUP(ident)];
for (td = TAILQ_FIRST(qp); td != NULL; td = ntd) {
ntd = TAILQ_NEXT(td, td_slpq);
if (td->td_wchan == ident) {
unsleep(td);
TD_CLR_SLEEPING(td);
setrunnable(td);
p = td->td_proc;
CTR3(KTR_PROC,"wakeup1: thread %p (pid %d, %s)",
td, p->p_pid, p->p_comm);
break;
}
1994-05-24 10:09:53 +00:00
}
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_unlock_spin(&sched_lock);
1994-05-24 10:09:53 +00:00
}
/*
* The machine independent parts of mi_switch().
*/
void
mi_switch(void)
1994-05-24 10:09:53 +00:00
{
struct bintime new_switchtime;
struct thread *td = curthread; /* XXX */
struct proc *p = td->td_proc; /* XXX */
struct kse *ke = td->td_kse;
u_int sched_nest;
1994-05-24 10:09:53 +00:00
mtx_assert(&sched_lock, MA_OWNED | MA_NOTRECURSED);
KASSERT(!TD_ON_RUNQ(td), ("mi_switch: called by old code"));
#ifdef INVARIANTS
if (!TD_ON_LOCK(td) &&
!TD_ON_RUNQ(td) &&
!TD_IS_RUNNING(td))
mtx_assert(&Giant, MA_NOTOWNED);
#endif
KASSERT(td->td_critnest == 1,
("mi_switch: switch in a critical section"));
1994-05-24 10:09:53 +00:00
/*
* Compute the amount of time during which the current
* process was running, and add that to its total so far.
*/
binuptime(&new_switchtime);
bintime_add(&p->p_runtime, &new_switchtime);
bintime_sub(&p->p_runtime, PCPU_PTR(switchtime));
1994-05-24 10:09:53 +00:00
#ifdef DDB
/*
* Don't perform context switches from the debugger.
*/
if (db_active) {
mtx_unlock_spin(&sched_lock);
db_error("Context switches not allowed in the debugger.");
}
#endif
1994-05-24 10:09:53 +00:00
/*
* Check if the process exceeds its cpu resource allocation. If
* over max, arrange to kill the process in ast().
1994-05-24 10:09:53 +00:00
*/
if (p->p_cpulimit != RLIM_INFINITY &&
p->p_runtime.sec > p->p_cpulimit) {
p->p_sflag |= PS_XCPU;
ke->ke_flags |= KEF_ASTPENDING;
}
1994-05-24 10:09:53 +00:00
/*
* Finish up stats for outgoing thread.
1994-05-24 10:09:53 +00:00
*/
cnt.v_swtch++;
PCPU_SET(switchtime, new_switchtime);
CTR3(KTR_PROC, "mi_switch: old thread %p (pid %d, %s)", td, p->p_pid,
p->p_comm);
sched_nest = sched_lock.mtx_recurse;
sched_switchout(td);
cpu_switch(); /* SHAZAM!!*/
sched_lock.mtx_recurse = sched_nest;
sched_lock.mtx_lock = (uintptr_t)td;
sched_switchin(td);
/*
* Start setting up stats etc. for the incoming thread.
* Similar code in fork_exit() is returned to by cpu_switch()
* in the case of a new thread/process.
*/
CTR3(KTR_PROC, "mi_switch: new thread %p (pid %d, %s)", td, p->p_pid,
p->p_comm);
if (PCPU_GET(switchtime.sec) == 0)
binuptime(PCPU_PTR(switchtime));
PCPU_SET(switchticks, ticks);
/*
* Call the switchin function while still holding the scheduler lock
* (used by the idlezero code and the general page-zeroing code)
*/
if (td->td_switchin)
td->td_switchin();
/*
* If the last thread was exiting, finish cleaning it up.
*/
if ((td = PCPU_GET(deadthread))) {
PCPU_SET(deadthread, NULL);
thread_stash(td);
}
1994-05-24 10:09:53 +00:00
}
/*
* Change process state to be runnable,
* placing it on the run queue if it is in memory,
* and awakening the swapper if it isn't in memory.
*/
void
setrunnable(struct thread *td)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
2001-09-21 19:16:12 +00:00
mtx_assert(&sched_lock, MA_OWNED);
switch (p->p_state) {
case PRS_ZOMBIE:
2001-09-21 19:16:12 +00:00
panic("setrunnable(1)");
default:
break;
}
switch (td->td_state) {
case TDS_RUNNING:
case TDS_RUNQ:
return;
case TDS_INHIBITED:
/*
* If we are only inhibited because we are swapped out
* then arange to swap in this process. Otherwise just return.
*/
if (td->td_inhibitors != TDI_SWAPPED)
return;
case TDS_CAN_RUN:
break;
1994-05-24 10:09:53 +00:00
default:
printf("state is 0x%x", td->td_state);
panic("setrunnable(2)");
1994-05-24 10:09:53 +00:00
}
if ((p->p_sflag & PS_INMEM) == 0) {
if ((p->p_sflag & PS_SWAPPINGIN) == 0) {
p->p_sflag |= PS_SWAPINREQ;
wakeup(&proc0);
}
} else
sched_wakeup(td);
1994-05-24 10:09:53 +00:00
}
/*
* Compute a tenex style load average of a quantity on
* 1, 5 and 15 minute intervals.
* XXXKSE Needs complete rewrite when correct info is available.
* Completely Bogus.. only works with 1:1 (but compiles ok now :-)
*/
static void
loadav(void *arg)
{
int i, nrun;
struct loadavg *avg;
struct proc *p;
struct thread *td;
avg = &averunnable;
sx_slock(&allproc_lock);
nrun = 0;
FOREACH_PROC_IN_SYSTEM(p) {
FOREACH_THREAD_IN_PROC(p, td) {
switch (td->td_state) {
case TDS_RUNQ:
case TDS_RUNNING:
if ((p->p_flag & P_NOLOAD) != 0)
goto nextproc;
nrun++; /* XXXKSE */
default:
break;
}
nextproc:
continue;
}
}
sx_sunlock(&allproc_lock);
for (i = 0; i < 3; i++)
avg->ldavg[i] = (cexp[i] * avg->ldavg[i] +
nrun * FSCALE * (FSCALE - cexp[i])) >> FSHIFT;
/*
* Schedule the next update to occur after 5 seconds, but add a
* random variation to avoid synchronisation with processes that
* run at regular intervals.
*/
callout_reset(&loadav_callout, hz * 4 + (int)(random() % (hz * 2 + 1)),
loadav, NULL);
}
static void
lboltcb(void *arg)
{
wakeup(&lbolt);
callout_reset(&lbolt_callout, hz, lboltcb, NULL);
}
/* ARGSUSED */
static void
sched_setup(dummy)
void *dummy;
{
callout_init(&loadav_callout, 0);
callout_init(&lbolt_callout, 1);
/* Kick off timeout driven events by calling first time. */
loadav(NULL);
lboltcb(NULL);
}
/*
* General purpose yield system call
*/
int
yield(struct thread *td, struct yield_args *uap)
{
struct ksegrp *kg = td->td_ksegrp;
mtx_assert(&Giant, MA_NOTOWNED);
2001-09-21 19:21:18 +00:00
mtx_lock_spin(&sched_lock);
kg->kg_proc->p_stats->p_ru.ru_nvcsw++;
sched_prio(td, PRI_MAX_TIMESHARE);
mi_switch();
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
mtx_unlock_spin(&sched_lock);
2001-09-21 19:21:18 +00:00
td->td_retval[0] = 0;
return (0);
}