d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
0c32d97ab5
freebsd-skq/sys/vm/vm_zeroidle.c

155 lines
3.6 KiB
C
Raw Normal View History

Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
/*-
* Copyright (c) 1994 John Dyson
* Copyright (c) 2001 Matt Dillon
*
* All rights reserved. Terms for use and redistribution
* are covered by the BSD Copyright as found in /usr/src/COPYRIGHT.
*
* from: @(#)vm_machdep.c 7.3 (Berkeley) 5/13/91
* Utah $Hdr: vm_machdep.c 1.16.1.1 89/06/23$
*/
Use __FBSDID().
2003-06-11 23:50:51 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
#include <sys/param.h>
#include <sys/systm.h>
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
#include <sys/kernel.h>
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
#include <sys/proc.h>
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
#include <sys/resourcevar.h>
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
#include <sys/vmmeter.h>
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
#include <sys/lock.h>
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
#include <sys/mutex.h>
- Create a new scheduler api that is defined in sys/sched.h - Begin moving scheduler specific functionality into sched_4bsd.c - Replace direct manipulation of scheduler data with hooks provided by the new api. - Remove KSE specific state modifications and single runq assumptions from kern_switch.c Reviewed by: -arch
2002-10-12 05:32:24 +00:00
#include <sys/sched.h>
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
#include <sys/sysctl.h>
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
#include <sys/kthread.h>
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
#include <vm/vm.h>
#include <vm/vm_page.h>
SYSCTL_DECL(_vm_stats_misc);
static int cnt_prezero;
SYSCTL_INT(_vm_stats_misc, OID_AUTO,
cnt_prezero, CTLFLAG_RD, &cnt_prezero, 0, "");
Add a special page zero entry point intended to be called via the single threaded VM pagezero kthread outside of Giant. For some platforms, this is really easy since it can just use the direct mapped region. For others, IPI sending is involved or there are other issues, so grab Giant when needed. We still have preemption issues to deal with, but Alan Cox has an interesting suggestion on how to minimize the problem on x86. Use Luigi's hack for preserving the (lack of) priority. Turn the idle zeroing back on since it can now actually do something useful outside of Giant in many cases.
2002-07-08 04:24:26 +00:00
static int idlezero_enable = 1;
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
SYSCTL_INT(_vm, OID_AUTO, idlezero_enable, CTLFLAG_RW, &idlezero_enable, 0, "");
TUNABLE_INT("vm.idlezero_enable", &idlezero_enable);
static int idlezero_maxrun = 16;
SYSCTL_INT(_vm, OID_AUTO, idlezero_maxrun, CTLFLAG_RW, &idlezero_maxrun, 0, "");
TUNABLE_INT("vm.idlezero_maxrun", &idlezero_maxrun);
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
/*
* Implement the pre-zeroed page mechanism.
*/
#define ZIDLE_LO(v) ((v) * 2 / 3)
#define ZIDLE_HI(v) ((v) * 4 / 5)
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
static int zero_state;
static int
vm_page_zero_check(void)
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
{
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
if (!idlezero_enable)
return 0;
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
/*
* Attempt to maintain approximately 1/2 of our free pages in a
* PG_ZERO'd state. Add some hysteresis to (attempt to) avoid
* generally zeroing a page when the system is near steady-state.
* Otherwise we might get 'flutter' during disk I/O / IPC or
* fast sleeps. We also do not want to be continuously zeroing
* pages because doing so may flush our L1 and L2 caches too much.
*/
if (zero_state && vm_page_zero_count >= ZIDLE_LO(cnt.v_free_count))
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
return 0;
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
if (vm_page_zero_count >= ZIDLE_HI(cnt.v_free_count))
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
return 0;
return 1;
}
static int
vm_page_zero_idle(void)
{
static int free_rover;
vm_page_t m;
vm_page_queue_free_mtx is a spin mutex, not a normal sleep mutex. I do not know why this didn't panic my box, but I have most certainly been using it: peter@overcee[3:14pm]~src/sys/i386/i386-110> sysctl -a | grep zero vm.stats.misc.zero_page_count: 2235 vm.stats.misc.cnt_prezero: 638951 vm.idlezero_enable: 1 vm.idlezero_maxrun: 16 Submitted by: Tor.Egge@cvsup.no.freebsd.org Approved by: Tor's patches are never wrong. :-)
2002-07-08 23:12:37 +00:00
mtx_lock_spin(&vm_page_queue_free_mtx);
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
zero_state = 0;
m = vm_pageq_find(PQ_FREE, free_rover, FALSE);
if (m != NULL && (m->flags & PG_ZERO) == 0) {
o Use vm_pageq_remove_nowakeup() and vm_pageq_enqueue() in vm_page_zero_idle() instead of partially duplicated implementations. In particular, this change guarantees that the number of free pages in the free queue(s) matches the global free page count when Giant is released. Submitted by: peter (via his p4 "pmap" branch)
2002-07-16 19:39:40 +00:00
vm_pageq_remove_nowakeup(m);
vm_page_queue_free_mtx is a spin mutex, not a normal sleep mutex. I do not know why this didn't panic my box, but I have most certainly been using it: peter@overcee[3:14pm]~src/sys/i386/i386-110> sysctl -a | grep zero vm.stats.misc.zero_page_count: 2235 vm.stats.misc.cnt_prezero: 638951 vm.idlezero_enable: 1 vm.idlezero_maxrun: 16 Submitted by: Tor.Egge@cvsup.no.freebsd.org Approved by: Tor's patches are never wrong. :-)
2002-07-08 23:12:37 +00:00
mtx_unlock_spin(&vm_page_queue_free_mtx);
Add a special page zero entry point intended to be called via the single threaded VM pagezero kthread outside of Giant. For some platforms, this is really easy since it can just use the direct mapped region. For others, IPI sending is involved or there are other issues, so grab Giant when needed. We still have preemption issues to deal with, but Alan Cox has an interesting suggestion on how to minimize the problem on x86. Use Luigi's hack for preserving the (lack of) priority. Turn the idle zeroing back on since it can now actually do something useful outside of Giant in many cases.
2002-07-08 04:24:26 +00:00
pmap_zero_page_idle(m);
vm_page_queue_free_mtx is a spin mutex, not a normal sleep mutex. I do not know why this didn't panic my box, but I have most certainly been using it: peter@overcee[3:14pm]~src/sys/i386/i386-110> sysctl -a | grep zero vm.stats.misc.zero_page_count: 2235 vm.stats.misc.cnt_prezero: 638951 vm.idlezero_enable: 1 vm.idlezero_maxrun: 16 Submitted by: Tor.Egge@cvsup.no.freebsd.org Approved by: Tor's patches are never wrong. :-)
2002-07-08 23:12:37 +00:00
mtx_lock_spin(&vm_page_queue_free_mtx);
o Remove the acquisition and release of Giant from the idle priority thread that pre-zeroes free pages. o Remove GIANT_REQUIRED from some low-level page queue functions. (Instead assertions on the page queue lock are being added to the higher-level functions, like vm_page_wire(), etc.) In collaboration with: peter
2002-07-18 17:40:07 +00:00
m->flags |= PG_ZERO;
o Use vm_pageq_remove_nowakeup() and vm_pageq_enqueue() in vm_page_zero_idle() instead of partially duplicated implementations. In particular, this change guarantees that the number of free pages in the free queue(s) matches the global free page count when Giant is released. Submitted by: peter (via his p4 "pmap" branch)
2002-07-16 19:39:40 +00:00
vm_pageq_enqueue(PQ_FREE + m->pc, m);
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
++vm_page_zero_count;
++cnt_prezero;
if (vm_page_zero_count >= ZIDLE_HI(cnt.v_free_count))
zero_state = 1;
}
free_rover = (free_rover + PQ_PRIME2) & PQ_L2_MASK;
vm_page_queue_free_mtx is a spin mutex, not a normal sleep mutex. I do not know why this didn't panic my box, but I have most certainly been using it: peter@overcee[3:14pm]~src/sys/i386/i386-110> sysctl -a | grep zero vm.stats.misc.zero_page_count: 2235 vm.stats.misc.cnt_prezero: 638951 vm.idlezero_enable: 1 vm.idlezero_maxrun: 16 Submitted by: Tor.Egge@cvsup.no.freebsd.org Approved by: Tor's patches are never wrong. :-)
2002-07-08 23:12:37 +00:00
mtx_unlock_spin(&vm_page_queue_free_mtx);
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
return 1;
}
/* Called by vm_page_free to hint that a new page is available */
void
vm_page_zero_idle_wakeup(void)
{
Add a special page zero entry point intended to be called via the single threaded VM pagezero kthread outside of Giant. For some platforms, this is really easy since it can just use the direct mapped region. For others, IPI sending is involved or there are other issues, so grab Giant when needed. We still have preemption issues to deal with, but Alan Cox has an interesting suggestion on how to minimize the problem on x86. Use Luigi's hack for preserving the (lack of) priority. Turn the idle zeroing back on since it can now actually do something useful outside of Giant in many cases.
2002-07-08 04:24:26 +00:00
if (idlezero_enable && vm_page_zero_check())
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
wakeup(&zero_state);
}
static void
vm_pagezero(void)
{
Set P_NOLOAD on the pagezero kthread so that it doesn't artificially skew the loadav. This is not real load. If you have a nice process running in the background, pagezero may sit in the run queue for ages and add one to the loadav, and thereby affecting other scheduling decisions.
2002-07-19 21:06:01 +00:00
struct thread *td;
struct proc *p;
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
struct rtprio rtp;
int pages = 0;
Add a special page zero entry point intended to be called via the single threaded VM pagezero kthread outside of Giant. For some platforms, this is really easy since it can just use the direct mapped region. For others, IPI sending is involved or there are other issues, so grab Giant when needed. We still have preemption issues to deal with, but Alan Cox has an interesting suggestion on how to minimize the problem on x86. Use Luigi's hack for preserving the (lack of) priority. Turn the idle zeroing back on since it can now actually do something useful outside of Giant in many cases.
2002-07-08 04:24:26 +00:00
int pri;
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
Set P_NOLOAD on the pagezero kthread so that it doesn't artificially skew the loadav. This is not real load. If you have a nice process running in the background, pagezero may sit in the run queue for ages and add one to the loadav, and thereby affecting other scheduling decisions.
2002-07-19 21:06:01 +00:00
td = curthread;
p = td->td_proc;
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
rtp.prio = RTP_PRIO_MAX;
rtp.type = RTP_PRIO_IDLE;
Process priority is locked by the sched_lock, not the proc lock.
2001-09-01 20:16:30 +00:00
mtx_lock_spin(&sched_lock);
In a threaded world, differnt priorirites become properties of different entities. Make it so. Reviewed by: jhb@freebsd.org (john baldwin)
2002-02-11 20:37:54 +00:00
rtp_to_pri(&rtp, td->td_ksegrp);
Add a special page zero entry point intended to be called via the single threaded VM pagezero kthread outside of Giant. For some platforms, this is really easy since it can just use the direct mapped region. For others, IPI sending is involved or there are other issues, so grab Giant when needed. We still have preemption issues to deal with, but Alan Cox has an interesting suggestion on how to minimize the problem on x86. Use Luigi's hack for preserving the (lack of) priority. Turn the idle zeroing back on since it can now actually do something useful outside of Giant in many cases.
2002-07-08 04:24:26 +00:00
pri = td->td_priority;
Process priority is locked by the sched_lock, not the proc lock.
2001-09-01 20:16:30 +00:00
mtx_unlock_spin(&sched_lock);
Set P_NOLOAD on the pagezero kthread so that it doesn't artificially skew the loadav. This is not real load. If you have a nice process running in the background, pagezero may sit in the run queue for ages and add one to the loadav, and thereby affecting other scheduling decisions.
2002-07-19 21:06:01 +00:00
PROC_LOCK(p);
p->p_flag |= P_NOLOAD;
PROC_UNLOCK(p);
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
for (;;) {
if (vm_page_zero_check()) {
pages += vm_page_zero_idle();
- Create a new scheduler api that is defined in sys/sched.h - Begin moving scheduler specific functionality into sched_4bsd.c - Replace direct manipulation of scheduler data with hooks provided by the new api. - Remove KSE specific state modifications and single runq assumptions from kern_switch.c Reviewed by: -arch
2002-10-12 05:32:24 +00:00
if (pages > idlezero_maxrun || sched_runnable()) {
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
mtx_lock_spin(&sched_lock);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_proc->p_stats->p_ru.ru_nvcsw++;
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
mi_switch();
mtx_unlock_spin(&sched_lock);
pages = 0;
}
} else {
Add a special page zero entry point intended to be called via the single threaded VM pagezero kthread outside of Giant. For some platforms, this is really easy since it can just use the direct mapped region. For others, IPI sending is involved or there are other issues, so grab Giant when needed. We still have preemption issues to deal with, but Alan Cox has an interesting suggestion on how to minimize the problem on x86. Use Luigi's hack for preserving the (lack of) priority. Turn the idle zeroing back on since it can now actually do something useful outside of Giant in many cases.
2002-07-08 04:24:26 +00:00
tsleep(&zero_state, pri, "pgzero", hz * 300);
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
pages = 0;
Move vm_page_zero_idle() from machine-dependant sections to a machine-independant source file, vm/vm_zeroidle.c. It was exactly the same for all platforms and updating them all was getting annoying.
2001-07-05 01:32:42 +00:00
}
}
}
Rename a static variable to avoid future conflicts.
2003-04-04 12:08:42 +00:00
static struct proc *pagezero_proc;
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
static struct kproc_desc pagezero_kp = {
"pagezero",
vm_pagezero,
Rename a static variable to avoid future conflicts.
2003-04-04 12:08:42 +00:00
&pagezero_proc
Implement idle zeroing of pages. I've been tinkering with this on and off since John Dyson left his work-in-progress. It is off by default for now. sysctl vm.zeroidle_enable=1 to turn it on. There are some hacks here to deal with the present lack of preemption - we yield after doing a small number of pages since we wont preempt otherwise. This is basically Matt's algorithm [with hysteresis] with an idle process to call it in a similar way it used to be called from the idle loop. I cleaned up the includes a fair bit here too.
2001-08-25 05:00:44 +00:00
};
SYSINIT(pagezero, SI_SUB_KTHREAD_VM, SI_ORDER_ANY, kproc_start, &pagezero_kp)
Reference in New Issue Copy Permalink