freebsd-skq/sys/kern/subr_rlist.c

292 lines
7.3 KiB
C
Raw Normal View History

1993-06-12 14:58:17 +00:00
/*
* Copyright (c) 1992 William F. Jolitz, TeleMuse
* 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 software is a component of "386BSD" developed by
William F. Jolitz, TeleMuse.
* 4. Neither the name of the developer nor the name "386BSD"
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS A COMPONENT OF 386BSD DEVELOPED BY WILLIAM F. JOLITZ
* AND IS INTENDED FOR RESEARCH AND EDUCATIONAL PURPOSES ONLY. THIS
* SOFTWARE SHOULD NOT BE CONSIDERED TO BE A COMMERCIAL PRODUCT.
* THE DEVELOPER URGES THAT USERS WHO REQUIRE A COMMERCIAL PRODUCT
* NOT MAKE USE THIS WORK.
*
* FOR USERS WHO WISH TO UNDERSTAND THE 386BSD SYSTEM DEVELOPED
* BY WILLIAM F. JOLITZ, WE RECOMMEND THE USER STUDY WRITTEN
* REFERENCES SUCH AS THE "PORTING UNIX TO THE 386" SERIES
* (BEGINNING JANUARY 1991 "DR. DOBBS JOURNAL", USA AND BEGINNING
* JUNE 1991 "UNIX MAGAZIN", GERMANY) BY WILLIAM F. JOLITZ AND
* LYNNE GREER JOLITZ, AS WELL AS OTHER BOOKS ON UNIX AND THE
* ON-LINE 386BSD USER MANUAL BEFORE USE. A BOOK DISCUSSING THE INTERNALS
* OF 386BSD ENTITLED "386BSD FROM THE INSIDE OUT" WILL BE AVAILABLE LATE 1992.
*
* THIS SOFTWARE IS PROVIDED BY THE DEVELOPER ``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 DEVELOPER 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.
*
* $Id: subr_rlist.c,v 1.6 1994/08/13 03:50:24 wollman Exp $
1993-06-12 14:58:17 +00:00
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/cdefs.h>
#include <sys/malloc.h>
#include <sys/rlist.h>
#include <sys/proc.h>
#include <vm/vm.h>
#include <vm/vm_map.h>
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
extern vm_map_t kernel_map;
1993-06-12 14:58:17 +00:00
/*
* Resource lists.
*/
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
#define RLIST_MIN 128
static int rlist_count=0;
static struct rlist *rlfree;
int rlist_active;
static struct rlist *
rlist_malloc()
{
struct rlist *rl;
int i;
while( rlist_count < RLIST_MIN) {
extern vm_map_t kmem_map;
int s = splhigh();
rl = (struct rlist *)kmem_malloc(kmem_map, NBPG, 0);
splx(s);
if( !rl)
break;
for(i=0;i<(NBPG/(sizeof *rl));i++) {
rl->rl_next = rlfree;
rlfree = rl;
rlist_count++;
rl++;
}
}
if( (rl = rlfree) == 0 )
panic("Cannot get an rlist entry");
--rlist_count;
rlfree = rl->rl_next;
return rl;
}
inline static void
rlist_mfree( struct rlist *rl)
{
rl->rl_next = rlfree;
rlfree = rl;
++rlist_count;
}
1993-06-12 14:58:17 +00:00
/*
* Add space to a resource list. Used to either
* initialize a list or return free space to it.
*/
void
1993-06-12 14:58:17 +00:00
rlist_free (rlp, start, end)
register struct rlist **rlp;
unsigned start, end;
{
1993-06-12 14:58:17 +00:00
struct rlist *head;
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
register struct rlist *olp = 0;
int s;
s = splhigh();
while( rlist_active)
tsleep((caddr_t)&rlist_active, PSWP, "rlistf", 0);
rlist_active = 1;
splx(s);
1993-06-12 14:58:17 +00:00
head = *rlp;
loop:
/* if nothing here, insert (tail of list) */
if (*rlp == 0) {
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
*rlp = rlist_malloc();
1993-06-12 14:58:17 +00:00
(*rlp)->rl_start = start;
(*rlp)->rl_end = end;
(*rlp)->rl_next = 0;
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
rlist_active = 0;
wakeup((caddr_t)&rlist_active);
1993-06-12 14:58:17 +00:00
return;
}
/* if new region overlaps something currently present, panic */
if (start >= (*rlp)->rl_start && start <= (*rlp)->rl_end) {
printf("Frag %d:%d, ent %d:%d ", start, end,
(*rlp)->rl_start, (*rlp)->rl_end);
panic("overlapping front rlist_free: freed twice?");
}
if (end >= (*rlp)->rl_start && end <= (*rlp)->rl_end) {
printf("Frag %d:%d, ent %d:%d ", start, end,
(*rlp)->rl_start, (*rlp)->rl_end);
panic("overlapping tail rlist_free: freed twice?");
}
/* are we adjacent to this element? (in front) */
if (end+1 == (*rlp)->rl_start) {
/* coalesce */
(*rlp)->rl_start = start;
goto scan;
}
/* are we before this element? */
if (end < (*rlp)->rl_start) {
register struct rlist *nlp;
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
nlp = rlist_malloc();
1993-06-12 14:58:17 +00:00
nlp->rl_start = start;
nlp->rl_end = end;
nlp->rl_next = *rlp;
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
/*
* If the new element is in front of the list,
* adjust *rlp, else don't.
*/
if( olp) {
olp->rl_next = nlp;
} else {
*rlp = nlp;
}
rlist_active = 0;
wakeup((caddr_t)&rlist_active);
1993-06-12 14:58:17 +00:00
return;
}
/* are we adjacent to this element? (at tail) */
if ((*rlp)->rl_end + 1 == start) {
/* coalesce */
(*rlp)->rl_end = end;
goto scan;
}
/* are we after this element */
if (start > (*rlp)->rl_end) {
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
olp = *rlp;
1993-06-12 14:58:17 +00:00
rlp = &((*rlp)->rl_next);
goto loop;
} else
panic("rlist_free: can't happen");
scan:
/* can we coalesce list now that we've filled a void? */
{
register struct rlist *lp, *lpn;
for (lp = head; lp->rl_next ;) {
lpn = lp->rl_next;
/* coalesce ? */
if (lp->rl_end + 1 == lpn->rl_start) {
lp->rl_end = lpn->rl_end;
lp->rl_next = lpn->rl_next;
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
rlist_mfree(lpn);
1993-06-12 14:58:17 +00:00
} else
lp = lp->rl_next;
}
}
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
rlist_active = 0;
wakeup((caddr_t)&rlist_active);
1993-06-12 14:58:17 +00:00
}
/*
* Obtain a region of desired size from a resource list.
* If nothing available of that size, return 0. Otherwise,
* return a value of 1 and set resource start location with
* "*loc". (Note: loc can be zero if we don't wish the value)
*/
int rlist_alloc (rlp, size, loc)
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
struct rlist **rlp;
unsigned size, *loc;
{
1993-06-12 14:58:17 +00:00
register struct rlist *lp;
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
int s;
register struct rlist *olp = 0;
1993-06-12 14:58:17 +00:00
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
s = splhigh();
while( rlist_active)
tsleep((caddr_t)&rlist_active, PSWP, "rlista", 0);
rlist_active = 1;
splx(s);
1993-06-12 14:58:17 +00:00
/* walk list, allocating first thing that's big enough (first fit) */
for (; *rlp; rlp = &((*rlp)->rl_next))
if(size <= (*rlp)->rl_end - (*rlp)->rl_start + 1) {
/* hand it to the caller */
if (loc) *loc = (*rlp)->rl_start;
(*rlp)->rl_start += size;
/* did we eat this element entirely? */
if ((*rlp)->rl_start > (*rlp)->rl_end) {
lp = (*rlp)->rl_next;
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
rlist_mfree(*rlp);
/*
* if the deleted element was in fromt
* of the list, adjust *rlp, else don't.
*/
if (olp) {
olp->rl_next = lp;
} else {
*rlp = lp;
}
1993-06-12 14:58:17 +00:00
}
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
rlist_active = 0;
wakeup((caddr_t)&rlist_active);
1993-06-12 14:58:17 +00:00
return (1);
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
} else {
olp = *rlp;
1993-06-12 14:58:17 +00:00
}
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
rlist_active = 0;
wakeup((caddr_t)&rlist_active);
1993-06-12 14:58:17 +00:00
/* nothing in list that's big enough */
return (0);
}
/*
* Finished with this resource list, reclaim all space and
* mark it as being empty.
*/
void
1993-06-12 14:58:17 +00:00
rlist_destroy (rlp)
struct rlist **rlp;
{
1993-06-12 14:58:17 +00:00
struct rlist *lp, *nlp;
lp = *rlp;
*rlp = 0;
for (; lp; lp = nlp) {
nlp = lp->rl_next;
First round of changes from John Dyson and myself to the VM system. This set improves performance and fixes the following problems (description from John Dyson): 1. Growing swap space problem in both static usage and in situations with lots of fork/execs in heavy paging situations. 2. Sparse swap space allocation (internal fragmentation.) 3. General swap_pager slowness. Additionally, the new swap_pager also provides hooks for multi-page I/O that is currently being developed (in early testing phases.) Problem #1 is a result of a problem where objects cannot be collapsed once a pager has been allocated for them. This problem has been solved by relaxing the restriction by allowing the pages contained in a shadow object's pager be copied to the parent object's pager. The copy is afforded by manipulating pointers to the disk blocks on the swap space. Since an improved swap_pager has already been developed with the data structures to support the copy operation, this new swap_pager has been introduced. Also, shadow object bypass in the collapse code has been enhanced to support checking for pages on disk. The vm_pageout daemon has also been modified to defer creation of an object's pager when the object's shadow is paging. This allows efficient immediate collapsing of a shadow into a parent object under many circumstances without the creation of an intermediate pager. Problem #2 is solved by the allocation data structures and algorithms in the new swap_pager. Additionally, a newer version of this new swap_pager is being tested that permits multiple page I/O and mitigation of the fragmentation problems associated with allocation of large contiguous blocks of swap space. Problem #3 is addressed by better algorithms and a fix of a couple of bugs in the swap_pager. Additionally, this new pager has a growth path allowing multi-page inputs from disk. Approximately 50% performance improvement can be expected under certain circumstances when using this pager in the standard single page mode. (Actually, I've seen more like twice the speed in my tests. -DLG)
1993-12-22 12:51:39 +00:00
rlist_mfree(lp);
1993-06-12 14:58:17 +00:00
}
}