freebsd-nq/sys/vm/sg_pager.c

/*-
 * Copyright (c) 2009 Advanced Computing Technologies LLC
 * Written by: John H. Baldwin <jhb@FreeBSD.org>
 * 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.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
 */

#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");

/*
 * This pager manages OBJT_SG objects.  These objects are backed by
 * a scatter/gather list of physical address ranges.
 */

#include <sys/param.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/sglist.h>
#include <vm/vm.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_pager.h>
#include <vm/uma.h>

static void sg_pager_init(void);
static vm_object_t sg_pager_alloc(void *, vm_ooffset_t, vm_prot_t,
    vm_ooffset_t, struct ucred *);
static void sg_pager_dealloc(vm_object_t);
static int sg_pager_getpages(vm_object_t, vm_page_t *, int, int);
static void sg_pager_putpages(vm_object_t, vm_page_t *, int, 
		boolean_t, int *);
static boolean_t sg_pager_haspage(vm_object_t, vm_pindex_t, int *,
		int *);

static uma_zone_t fakepg_zone;

static vm_page_t sg_pager_getfake(vm_paddr_t, vm_memattr_t);
static void sg_pager_putfake(vm_page_t);

struct pagerops sgpagerops = {
	.pgo_init =	sg_pager_init,
	.pgo_alloc =	sg_pager_alloc,
	.pgo_dealloc =	sg_pager_dealloc,
	.pgo_getpages =	sg_pager_getpages,
	.pgo_putpages =	sg_pager_putpages,
	.pgo_haspage =	sg_pager_haspage,
};

static void
sg_pager_init(void)
{

	fakepg_zone = uma_zcreate("SG fakepg", sizeof(struct vm_page),
	    NULL, NULL, NULL, NULL, UMA_ALIGN_PTR,
	    UMA_ZONE_NOFREE|UMA_ZONE_VM); 
}

static vm_object_t
sg_pager_alloc(void *handle, vm_ooffset_t size, vm_prot_t prot,
    vm_ooffset_t foff, struct ucred *cred)
{
	struct sglist *sg;
	vm_object_t object;
	vm_pindex_t npages, pindex;
	int i;

	/*
	 * Offset should be page aligned.
	 */
	if (foff & PAGE_MASK)
		return (NULL);

	/*
	 * The scatter/gather list must only include page-aligned
	 * ranges.
	 */
	npages = 0;
	sg = handle;
	for (i = 0; i < sg->sg_nseg; i++) {
		if ((sg->sg_segs[i].ss_paddr % PAGE_SIZE) != 0 ||
		    (sg->sg_segs[i].ss_len % PAGE_SIZE) != 0)
			return (NULL);
		npages += sg->sg_segs[i].ss_len / PAGE_SIZE;
	}

	/*
	 * The scatter/gather list has a fixed size.  Refuse requests
	 * to map beyond that.
	 */
	size = round_page(size);
	pindex = OFF_TO_IDX(foff + size);
	if (pindex > npages)
		return (NULL);

	/*
	 * Allocate a new object and associate it with the
	 * scatter/gather list.  It is ok for our purposes to have
	 * multiple VM objects associated with the same scatter/gather
	 * list because scatter/gather lists are static.  This is also
	 * simpler than ensuring a unique object per scatter/gather
	 * list.
	 */
	object = vm_object_allocate(OBJT_SG, npages);
	object->handle = sglist_hold(sg);
	TAILQ_INIT(&object->un_pager.sgp.sgp_pglist);
	return (object);
}

static void
sg_pager_dealloc(vm_object_t object)
{
	struct sglist *sg;
	vm_page_t m;

	/*
	 * Free up our fake pages.
	 */
	while ((m = TAILQ_FIRST(&object->un_pager.sgp.sgp_pglist)) != 0) {
		TAILQ_REMOVE(&object->un_pager.sgp.sgp_pglist, m, pageq);
		sg_pager_putfake(m);
	}
	
	sg = object->handle;
	sglist_free(sg);
}

static int
sg_pager_getpages(vm_object_t object, vm_page_t *m, int count, int reqpage)
{
	struct sglist *sg;
	vm_page_t m_paddr, page;
	vm_pindex_t offset;
	vm_paddr_t paddr;
	vm_memattr_t memattr;
	size_t space;
	int i;

	VM_OBJECT_LOCK_ASSERT(object, MA_OWNED);
	sg = object->handle;
	memattr = object->memattr;
	VM_OBJECT_UNLOCK(object);
	offset = m[reqpage]->pindex;

	/*
	 * Lookup the physical address of the requested page.  An initial
	 * value of '1' instead of '0' is used so we can assert that the
	 * page is found since '0' can be a valid page-aligned physical
	 * address.
	 */
	space = 0;
	paddr = 1;
	for (i = 0; i < sg->sg_nseg; i++) {
		if (space + sg->sg_segs[i].ss_len <= (offset * PAGE_SIZE)) {
			space += sg->sg_segs[i].ss_len;
			continue;
		}
		paddr = sg->sg_segs[i].ss_paddr + offset * PAGE_SIZE - space;
		break;
	}
	KASSERT(paddr != 1, ("invalid SG page index"));

	/* If "paddr" is a real page, perform a sanity check on "memattr". */
	if ((m_paddr = vm_phys_paddr_to_vm_page(paddr)) != NULL &&
	    pmap_page_get_memattr(m_paddr) != memattr) {
		memattr = pmap_page_get_memattr(m_paddr);
		printf(
	    "WARNING: A device driver has set \"memattr\" inconsistently.\n");
	}

	/* Return a fake page for the requested page. */
	KASSERT(!(m[reqpage]->flags & PG_FICTITIOUS),
	    ("backing page for SG is fake"));

	/* Construct a new fake page. */
	printf("SG: getting fake page for paddr %lx\n", paddr);
	page = sg_pager_getfake(paddr, memattr);
	VM_OBJECT_LOCK(object);
	TAILQ_INSERT_TAIL(&object->un_pager.sgp.sgp_pglist, page, pageq);

	/* Free the original pages and insert this fake page into the object. */
	vm_page_lock_queues();
	for (i = 0; i < count; i++) {
		printf("SG: freeing VM page %p\n", m[i]);
		vm_page_free(m[i]);
	}
	vm_page_unlock_queues();
	printf("SG: Inserting new fake page\n");
	vm_page_insert(page, object, offset);
	m[reqpage] = page;

	return (VM_PAGER_OK);
}

static void
sg_pager_putpages(vm_object_t object, vm_page_t *m, int count,
    boolean_t sync, int *rtvals)
{

	panic("sg_pager_putpage called");
}

static boolean_t
sg_pager_haspage(vm_object_t object, vm_pindex_t pindex, int *before,
    int *after)
{

	if (before != NULL)
		*before = 0;
	if (after != NULL)
		*after = 0;
	return (TRUE);
}

/*
 * Create a fictitious page with the specified physical address and memory
 * attribute.  The memory attribute is the only the machine-dependent aspect
 * of a fictitious page that must be initialized.
 */
static vm_page_t
sg_pager_getfake(vm_paddr_t paddr, vm_memattr_t memattr)
{
	vm_page_t m;

	m = uma_zalloc(fakepg_zone, M_WAITOK | M_ZERO);
	m->phys_addr = paddr;
	/* Fictitious pages don't use "segind". */
	m->flags = PG_FICTITIOUS;
	/* Fictitious pages don't use "order" or "pool". */
	m->oflags = VPO_BUSY;
	m->wire_count = 1;
	pmap_page_set_memattr(m, memattr);
	return (m);
}

static void
sg_pager_putfake(vm_page_t m)
{

	if (!(m->flags & PG_FICTITIOUS))
		panic("sg_pager_putfake: bad page");
	uma_zfree(fakepg_zone, m);
}