freebsd-nq/sys/kern/subr_rlist.c

312 lines
8.1 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:
1995-05-30 08:16:23 +00:00
* This software is a component of "386BSD" developed by
1993-06-12 14:58:17 +00:00
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.
*
1995-05-30 08:16:23 +00:00
* 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
1993-06-12 14:58:17 +00:00
* NOT MAKE USE THIS WORK.
*
* FOR USERS WHO WISH TO UNDERSTAND THE 386BSD SYSTEM DEVELOPED
1995-05-30 08:16:23 +00:00
* 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
1993-06-12 14:58:17 +00:00
* 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.
*
*/
/*
* Changes Copyright (C) 1995, David Greenman & John Dyson; This software may
* be used, modified, copied, distributed, and sold, in both source and
* binary form provided that the above copyright and these terms are
* retained. Under no circumstances is the author responsible for the proper
* functioning of this software, nor does the author assume any responsibility
* for damages incurred with its use.
*
* $Id: subr_rlist.c,v 1.18 1996/10/12 21:35:25 dyson Exp $
1993-06-12 14:58:17 +00:00
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/rlist.h>
#include <sys/proc.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/vm_kern.h>
#include <vm/vm_extern.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
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;
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
static struct rlist *rlfree;
static struct rlist *rlist_malloc __P((void));
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
static struct rlist *
rlist_malloc()
{
struct rlist *rl;
int i;
while( rlist_count < RLIST_MIN) {
int s = splhigh();
rl = (struct rlist *)kmem_alloc(kernel_map, PAGE_SIZE);
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
splx(s);
if( !rl)
break;
1995-05-30 08:16:23 +00:00
for(i=0;i<(PAGE_SIZE/(sizeof *rl));i++) {
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
rl->rl_next = rlfree;
rlfree = rl;
rlist_count++;
rl++;
}
}
if( (rl = rlfree) == 0 )
panic("Cannot get an rlist entry");
1995-05-30 08:16:23 +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_count;
rlfree = rl->rl_next;
return rl;
}
inline static void
rlist_mfree( struct rlist *rl)
{
rl->rl_next = rlfree;
rlfree = rl;
++rlist_count;
}
1995-05-30 08:16:23 +00:00
void
rlist_free(rlh, start, end)
struct rlisthdr *rlh;
u_int start, end;
{
struct rlist **rlp = &rlh->rlh_list;
struct rlist *prev_rlp = NULL, *cur_rlp = *rlp, *next_rlp = NULL;
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;
s = splhigh();
while (rlh->rlh_lock & RLH_LOCKED) {
rlh->rlh_lock |= RLH_DESIRED;
tsleep(rlh, PSWP, "rlistf", 0);
}
rlh->rlh_lock |= RLH_LOCKED;
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
splx(s);
1993-06-12 14:58:17 +00:00
/*
* Traverse the list looking for an entry after the one we want
* to insert.
*/
while (cur_rlp != NULL) {
if (start < cur_rlp->rl_start)
break;
#ifdef DIAGNOSTIC
if (prev_rlp) {
if (prev_rlp->rl_end + 1 == cur_rlp->rl_start)
panic("rlist_free: missed coalesce opportunity");
if (prev_rlp->rl_end == cur_rlp->rl_start)
panic("rlist_free: entries overlap");
if (prev_rlp->rl_end > cur_rlp->rl_start)
panic("entries out of order");
}
#endif
prev_rlp = cur_rlp;
cur_rlp = cur_rlp->rl_next;
1993-06-12 14:58:17 +00:00
}
if (cur_rlp != NULL) {
1993-06-12 14:58:17 +00:00
if (end >= cur_rlp->rl_start)
panic("rlist_free: free end overlaps already freed area");
1993-06-12 14:58:17 +00:00
if (prev_rlp) {
if (start <= prev_rlp->rl_end)
panic("rlist_free: free start overlaps already freed area");
/*
* Attempt to append
*/
if (prev_rlp->rl_end + 1 == start) {
prev_rlp->rl_end = end;
/*
* Attempt to prepend and coalesce
*/
if (end + 1 == cur_rlp->rl_start) {
prev_rlp->rl_end = cur_rlp->rl_end;
prev_rlp->rl_next = cur_rlp->rl_next;
rlist_mfree(cur_rlp);
}
goto done;
}
}
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
/*
* Attempt to prepend
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 (end + 1 == cur_rlp->rl_start) {
cur_rlp->rl_start = start;
goto done;
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
}
1993-06-12 14:58:17 +00:00
}
/*
* Reached the end of the list without finding a larger entry.
* Append to last entry if there is one and it's adjacent.
*/
if (prev_rlp) {
if (start <= prev_rlp->rl_end)
panic("rlist_free: free start overlaps already freed area at list tail");
/*
* Attempt to append
*/
if (prev_rlp->rl_end + 1 == start) {
prev_rlp->rl_end = end;
goto done;
}
1993-06-12 14:58:17 +00:00
}
/*
* Could neither append nor prepend; allocate a new entry.
*/
next_rlp = cur_rlp;
cur_rlp = rlist_malloc();
cur_rlp->rl_start = start;
cur_rlp->rl_end = end;
cur_rlp->rl_next = next_rlp;
if (prev_rlp) {
prev_rlp->rl_next = cur_rlp;
} else {
/*
* No previous - this entry is the new list head.
*/
*rlp = cur_rlp;
1993-06-12 14:58:17 +00:00
}
done:
rlh->rlh_lock &= ~RLH_LOCKED;
if (rlh->rlh_lock & RLH_DESIRED) {
wakeup(rlh);
rlh->rlh_lock &= ~RLH_DESIRED;
}
return;
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 (rlh, size, loc)
struct rlisthdr *rlh;
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
unsigned size, *loc;
{
struct rlist **rlp = &rlh->rlh_list;
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
s = splhigh();
while (rlh->rlh_lock & RLH_LOCKED) {
rlh->rlh_lock |= RLH_DESIRED;
tsleep(rlh, PSWP, "rlistf", 0);
}
rlh->rlh_lock |= RLH_LOCKED;
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
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
}
rlh->rlh_lock &= ~RLH_LOCKED;
if (rlh->rlh_lock & RLH_DESIRED) {
wakeup(rlh);
rlh->rlh_lock &= ~RLH_DESIRED;
}
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
}
rlh->rlh_lock &= ~RLH_LOCKED;
if (rlh->rlh_lock & RLH_DESIRED) {
wakeup(rlh);
rlh->rlh_lock &= ~RLH_DESIRED;
}
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
rlist_destroy (rlh)
struct rlisthdr *rlh;
{
struct rlist **rlp = &rlh->rlh_list;
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
}
}