uvm_km.c

基于组件方式开发操作系统的OSKIT源代码

/*	$NetBSD: uvm_km.c,v 1.41 2000/11/27 04:36:40 nisimura Exp $	*/

/* 
 * Copyright (c) 1997 Charles D. Cranor and Washington University.
 * Copyright (c) 1991, 1993, The Regents of the University of California.  
 *
 * All rights reserved.
 *
 * This code is derived from software contributed to Berkeley by
 * The Mach Operating System project at Carnegie-Mellon University.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by Charles D. Cranor,
 *      Washington University, the University of California, Berkeley and 
 *      its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	@(#)vm_kern.c   8.3 (Berkeley) 1/12/94
 * from: Id: uvm_km.c,v 1.1.2.14 1998/02/06 05:19:27 chs Exp
 *
 *
 * Copyright (c) 1987, 1990 Carnegie-Mellon University.
 * All rights reserved.
 * 
 * Permission to use, copy, modify and distribute this software and
 * its documentation is hereby granted, provided that both the copyright
 * notice and this permission notice appear in all copies of the
 * software, derivative works or modified versions, and any portions
 * thereof, and that both notices appear in supporting documentation.
 * 
 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" 
 * CONDITION.  CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND 
 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
 * 
 * Carnegie Mellon requests users of this software to return to
 *
 *  Software Distribution Coordinator  or  [e-mail]
 *  School of Computer Science
 *  Carnegie Mellon University
 *  Pittsburgh PA 15213-3890
 *
 * any improvements or extensions that they make and grant Carnegie the
 * rights to redistribute these changes.
 */

#include "opt_uvmhist.h"

/*
 * uvm_km.c: handle kernel memory allocation and management
 */

/*
 * overview of kernel memory management:
 *
 * the kernel virtual address space is mapped by "kernel_map."   kernel_map
 * starts at VM_MIN_KERNEL_ADDRESS and goes to VM_MAX_KERNEL_ADDRESS.
 * note that VM_MIN_KERNEL_ADDRESS is equal to vm_map_min(kernel_map).
 *
 * the kernel_map has several "submaps."   submaps can only appear in 
 * the kernel_map (user processes can't use them).   submaps "take over"
 * the management of a sub-range of the kernel's address space.  submaps
 * are typically allocated at boot time and are never released.   kernel
 * virtual address space that is mapped by a submap is locked by the 
 * submap's lock -- not the kernel_map's lock.
 *
 * thus, the useful feature of submaps is that they allow us to break
 * up the locking and protection of the kernel address space into smaller
 * chunks.
 *
 * the vm system has several standard kernel submaps, including:
 *   kmem_map => contains only wired kernel memory for the kernel
 *		malloc.   *** access to kmem_map must be protected
 *		by splimp() because we are allowed to call malloc()
 *		at interrupt time ***
 *   mb_map => memory for large mbufs,  *** protected by splimp ***
 *   pager_map => used to map "buf" structures into kernel space
 *   exec_map => used during exec to handle exec args
 *   etc...
 *
 * the kernel allocates its private memory out of special uvm_objects whose
 * reference count is set to UVM_OBJ_KERN (thus indicating that the objects
 * are "special" and never die).   all kernel objects should be thought of
 * as large, fixed-sized, sparsely populated uvm_objects.   each kernel 
 * object is equal to the size of kernel virtual address space (i.e. the
 * value "VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS").
 *
 * most kernel private memory lives in kernel_object.   the only exception
 * to this is for memory that belongs to submaps that must be protected
 * by splimp().    each of these submaps has their own private kernel 
 * object (e.g. kmem_object, mb_object).
 *
 * note that just because a kernel object spans the entire kernel virutal
 * address space doesn't mean that it has to be mapped into the entire space.
 * large chunks of a kernel object's space go unused either because 
 * that area of kernel VM is unmapped, or there is some other type of 
 * object mapped into that range (e.g. a vnode).    for submap's kernel
 * objects, the only part of the object that can ever be populated is the
 * offsets that are managed by the submap.
 *
 * note that the "offset" in a kernel object is always the kernel virtual
 * address minus the VM_MIN_KERNEL_ADDRESS (aka vm_map_min(kernel_map)).
 * example:
 *   suppose VM_MIN_KERNEL_ADDRESS is 0xf8000000 and the kernel does a
 *   uvm_km_alloc(kernel_map, PAGE_SIZE) [allocate 1 wired down page in the
 *   kernel map].    if uvm_km_alloc returns virtual address 0xf8235000,
 *   then that means that the page at offset 0x235000 in kernel_object is
 *   mapped at 0xf8235000.   
 *
 * note that the offsets in kmem_object and mb_object also follow this
 * rule.   this means that the offsets for kmem_object must fall in the
 * range of [vm_map_min(kmem_object) - vm_map_min(kernel_map)] to
 * [vm_map_max(kmem_object) - vm_map_min(kernel_map)], so the offsets
 * in those objects will typically not start at zero.
 *
 * kernel object have one other special property: when the kernel virtual
 * memory mapping them is unmapped, the backing memory in the object is
 * freed right away.   this is done with the uvm_km_pgremove() function.
 * this has to be done because there is no backing store for kernel pages
 * and no need to save them after they are no longer referenced.
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>

#include <uvm/uvm.h>

/*
 * global data structures
 */

vm_map_t kernel_map = NULL;

struct vmi_list vmi_list;
simple_lock_data_t vmi_list_slock;

/*
 * local data structues
 */

#ifdef OSKIT
extern struct vmspace		vmspace0;
#define kernel_map_store	(vmspace0.vm_map)
#else
static struct vm_map		kernel_map_store;
#endif
static struct uvm_object	kmem_object_store;
static struct uvm_object	mb_object_store;

/*
 * All pager operations here are NULL, but the object must have
 * a pager ops vector associated with it; various places assume
 * it to be so.
 */
static struct uvm_pagerops	km_pager;

/*
 * uvm_km_init: init kernel maps and objects to reflect reality (i.e.
 * KVM already allocated for text, data, bss, and static data structures).
 *
 * => KVM is defined by VM_MIN_KERNEL_ADDRESS/VM_MAX_KERNEL_ADDRESS.
 *    we assume that [min -> start] has already been allocated and that
 *    "end" is the end.
 */

void
uvm_km_init(start, end)
	vaddr_t start, end;
{
	vaddr_t base = VM_MIN_KERNEL_ADDRESS;

	/*
	 * first, initialize the interrupt-safe map list.
	 */
	LIST_INIT(&vmi_list);
	simple_lock_init(&vmi_list_slock);

	/*
	 * next, init kernel memory objects.
	 */

	/* kernel_object: for pageable anonymous kernel memory */
	uao_init();
	uvm.kernel_object = uao_create(VM_MAX_KERNEL_ADDRESS -
				 VM_MIN_KERNEL_ADDRESS, UAO_FLAG_KERNOBJ);

	/*
	 * kmem_object: for use by the kernel malloc().  Memory is always
	 * wired, and this object (and the kmem_map) can be accessed at
	 * interrupt time.
	 */
	simple_lock_init(&kmem_object_store.vmobjlock);
	kmem_object_store.pgops = &km_pager;
	TAILQ_INIT(&kmem_object_store.memq);
	kmem_object_store.uo_npages = 0;
	/* we are special.  we never die */
	kmem_object_store.uo_refs = UVM_OBJ_KERN_INTRSAFE; 
	uvmexp.kmem_object = &kmem_object_store;

	/*
	 * mb_object: for mbuf cluster pages on platforms which use the
	 * mb_map.  Memory is always wired, and this object (and the mb_map)
	 * can be accessed at interrupt time.
	 */
	simple_lock_init(&mb_object_store.vmobjlock);
	mb_object_store.pgops = &km_pager;
	TAILQ_INIT(&mb_object_store.memq);
	mb_object_store.uo_npages = 0;
	/* we are special.  we never die */
	mb_object_store.uo_refs = UVM_OBJ_KERN_INTRSAFE; 
	uvmexp.mb_object = &mb_object_store;

	/*
	 * init the map and reserve allready allocated kernel space 
	 * before installing.
	 */

	uvm_map_setup(&kernel_map_store, base, end, VM_MAP_PAGEABLE);
	kernel_map_store.pmap = pmap_kernel();
	if (uvm_map(&kernel_map_store, &base, start - base, NULL,
	    UVM_UNKNOWN_OFFSET, 0, UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL,
	    UVM_INH_NONE, UVM_ADV_RANDOM,UVM_FLAG_FIXED)) != KERN_SUCCESS)
		panic("uvm_km_init: could not reserve space for kernel");
	
	/*
	 * install!
	 */

	kernel_map = &kernel_map_store;
}

/*
 * uvm_km_suballoc: allocate a submap in the kernel map.   once a submap
 * is allocated all references to that area of VM must go through it.  this
 * allows the locking of VAs in kernel_map to be broken up into regions.
 *
 * => if `fixed' is true, *min specifies where the region described
 *      by the submap must start
 * => if submap is non NULL we use that as the submap, otherwise we
 *	alloc a new map
 */
struct vm_map *
uvm_km_suballoc(map, min, max, size, flags, fixed, submap)
	struct vm_map *map;
	vaddr_t *min, *max;		/* OUT, OUT */
	vsize_t size;
	int flags;
	boolean_t fixed;
	struct vm_map *submap;
{
	int mapflags = UVM_FLAG_NOMERGE | (fixed ? UVM_FLAG_FIXED : 0);

	size = round_page(size);	/* round up to pagesize */

	/*
	 * first allocate a blank spot in the parent map
	 */

	if (uvm_map(map, min, size, NULL, UVM_UNKNOWN_OFFSET, 0,
	    UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_NONE,
	    UVM_ADV_RANDOM, mapflags)) != KERN_SUCCESS) {
	       panic("uvm_km_suballoc: unable to allocate space in parent map");
	}

	/*
	 * set VM bounds (min is filled in by uvm_map)
	 */

	*max = *min + size;

	/*
	 * add references to pmap and create or init the submap
	 */

	pmap_reference(vm_map_pmap(map));
	if (submap == NULL) {
		submap = uvm_map_create(vm_map_pmap(map), *min, *max, flags);
		if (submap == NULL)
			panic("uvm_km_suballoc: unable to create submap");
	} else {
		uvm_map_setup(submap, *min, *max, flags);
		submap->pmap = vm_map_pmap(map);
	}

	/*
	 * now let uvm_map_submap plug in it...
	 */

	if (uvm_map_submap(map, *min, *max, submap) != KERN_SUCCESS)
		panic("uvm_km_suballoc: submap allocation failed");

	return(submap);
}

/*
 * uvm_km_pgremove: remove pages from a kernel uvm_object.
 *
 * => when you unmap a part of anonymous kernel memory you want to toss
 *    the pages right away.    (this gets called from uvm_unmap_...).
 */

#define UKM_HASH_PENALTY 4      /* a guess */

void
uvm_km_pgremove(uobj, start, end)
	struct uvm_object *uobj;
	vaddr_t start, end;
{
	boolean_t by_list;
	struct vm_page *pp, *ppnext;
	vaddr_t curoff;
	UVMHIST_FUNC("uvm_km_pgremove"); UVMHIST_CALLED(maphist);

	KASSERT(uobj->pgops == &aobj_pager);
	simple_lock(&uobj->vmobjlock);

	/* choose cheapest traversal */
	by_list = (uobj->uo_npages <=
	     ((end - start) >> PAGE_SHIFT) * UKM_HASH_PENALTY);
 
	if (by_list)
		goto loop_by_list;

	/* by hash */

	for (curoff = start ; curoff < end ; curoff += PAGE_SIZE) {
		pp = uvm_pagelookup(uobj, curoff);
		if (pp == NULL)
			continue;

		UVMHIST_LOG(maphist,"  page 0x%x, busy=%d", pp,
		    pp->flags & PG_BUSY, 0, 0);

		/* now do the actual work */
		if (pp->flags & PG_BUSY) {
			/* owner must check for this when done */
			pp->flags |= PG_RELEASED;
		} else {
			/* free the swap slot... */
			uao_dropswap(uobj, curoff >> PAGE_SHIFT);

			/*
			 * ...and free the page; note it may be on the
			 * active or inactive queues.
			 */
			uvm_lock_pageq();
			uvm_pagefree(pp);
			uvm_unlock_pageq();
		}
	}
	simple_unlock(&uobj->vmobjlock);
	return;

loop_by_list:

	for (pp = TAILQ_FIRST(&uobj->memq); pp != NULL; pp = ppnext) {
		ppnext = TAILQ_NEXT(pp, listq);
		if (pp->offset < start || pp->offset >= end) {
			continue;
		}

		UVMHIST_LOG(maphist,"  page 0x%x, busy=%d", pp,
		    pp->flags & PG_BUSY, 0, 0);

		if (pp->flags & PG_BUSY) {
			/* owner must check for this when done */
			pp->flags |= PG_RELEASED;
		} else {
			/* free the swap slot... */
			uao_dropswap(uobj, pp->offset >> PAGE_SHIFT);

			/*
			 * ...and free the page; note it may be on the
			 * active or inactive queues.
			 */
			uvm_lock_pageq();
			uvm_pagefree(pp);
			uvm_unlock_pageq();
		}
	}
	simple_unlock(&uobj->vmobjlock);
}


/*
 * uvm_km_pgremove_intrsafe: like uvm_km_pgremove(), but for "intrsafe"
 *    objects
 *
 * => when you unmap a part of anonymous kernel memory you want to toss
 *    the pages right away.    (this gets called from uvm_unmap_...).
 * => none of the pages will ever be busy, and none of them will ever
 *    be on the active or inactive queues (because these objects are
 *    never allowed to "page").
 */

void
uvm_km_pgremove_intrsafe(uobj, start, end)
	struct uvm_object *uobj;
	vaddr_t start, end;
{
	boolean_t by_list;
	struct vm_page *pp, *ppnext;
	vaddr_t curoff;
	UVMHIST_FUNC("uvm_km_pgremove_intrsafe"); UVMHIST_CALLED(maphist);

	KASSERT(UVM_OBJ_IS_INTRSAFE_OBJECT(uobj));
	simple_lock(&uobj->vmobjlock);		/* lock object */

	/* choose cheapest traversal */
	by_list = (uobj->uo_npages <=
	     ((end - start) >> PAGE_SHIFT) * UKM_HASH_PENALTY);
 
	if (by_list)
		goto loop_by_list;

	/* by hash */

	for (curoff = start ; curoff < end ; curoff += PAGE_SIZE) {
		pp = uvm_pagelookup(uobj, curoff);
		if (pp == NULL) {
			continue;
		}

		UVMHIST_LOG(maphist,"  page 0x%x, busy=%d", pp,
		    pp->flags & PG_BUSY, 0, 0);
		KASSERT((pp->flags & PG_BUSY) == 0);
		KASSERT((pp->pqflags & PQ_ACTIVE) == 0);
		KASSERT((pp->pqflags & PQ_INACTIVE) == 0);
		uvm_pagefree(pp);
	}
	simple_unlock(&uobj->vmobjlock);
	return;

loop_by_list:

	for (pp = TAILQ_FIRST(&uobj->memq); pp != NULL; pp = ppnext) {
		ppnext = TAILQ_NEXT(pp, listq);
		if (pp->offset < start || pp->offset >= end) {
			continue;