[deliverable/linux.git] / drivers / block / brd.c

/*
 * Ram backed block device driver.
 *
 * Copyright (C) 2007 Nick Piggin
 * Copyright (C) 2007 Novell Inc.
 *
 * Parts derived from drivers/block/rd.c, and drivers/block/loop.c, copyright
 * of their respective owners.
 */

#include <linux/init.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/major.h>
#include <linux/blkdev.h>
#include <linux/bio.h>
#include <linux/highmem.h>
#include <linux/mutex.h>
#include <linux/radix-tree.h>
#include <linux/fs.h>
#include <linux/slab.h>

#include <asm/uaccess.h>

#define SECTOR_SHIFT		9
#define PAGE_SECTORS_SHIFT	(PAGE_SHIFT - SECTOR_SHIFT)
#define PAGE_SECTORS		(1 << PAGE_SECTORS_SHIFT)

/*
 * Each block ramdisk device has a radix_tree brd_pages of pages that stores
 * the pages containing the block device's contents. A brd page's ->index is
 * its offset in PAGE_SIZE units. This is similar to, but in no way connected
 * with, the kernel's pagecache or buffer cache (which sit above our block
 * device).
 */
struct brd_device {
	int		brd_number;

	struct request_queue	*brd_queue;
	struct gendisk		*brd_disk;
	struct list_head	brd_list;

	/*
	 * Backing store of pages and lock to protect it. This is the contents
	 * of the block device.
	 */
	spinlock_t		brd_lock;
	struct radix_tree_root	brd_pages;
};

/*
 * Look up and return a brd's page for a given sector.
 */
static DEFINE_MUTEX(brd_mutex);
static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector)
{
	pgoff_t idx;
	struct page *page;

	/*
	 * The page lifetime is protected by the fact that we have opened the
	 * device node -- brd pages will never be deleted under us, so we
	 * don't need any further locking or refcounting.
	 *
	 * This is strictly true for the radix-tree nodes as well (ie. we
	 * don't actually need the rcu_read_lock()), however that is not a
	 * documented feature of the radix-tree API so it is better to be
	 * safe here (we don't have total exclusion from radix tree updates
	 * here, only deletes).
	 */
	rcu_read_lock();
	idx = sector >> PAGE_SECTORS_SHIFT; /* sector to page index */
	page = radix_tree_lookup(&brd->brd_pages, idx);
	rcu_read_unlock();

	BUG_ON(page && page->index != idx);

	return page;
}

/*
 * Look up and return a brd's page for a given sector.
 * If one does not exist, allocate an empty page, and insert that. Then
 * return it.
 */
static struct page *brd_insert_page(struct brd_device *brd, sector_t sector)
{
	pgoff_t idx;
	struct page *page;
	gfp_t gfp_flags;

	page = brd_lookup_page(brd, sector);
	if (page)
		return page;

	/*
	 * Must use NOIO because we don't want to recurse back into the
	 * block or filesystem layers from page reclaim.
	 *
	 * Cannot support XIP and highmem, because our ->direct_access
	 * routine for XIP must return memory that is always addressable.
	 * If XIP was reworked to use pfns and kmap throughout, this
	 * restriction might be able to be lifted.
	 */
	gfp_flags = GFP_NOIO | __GFP_ZERO;
#ifndef CONFIG_BLK_DEV_XIP
	gfp_flags |= __GFP_HIGHMEM;
#endif
	page = alloc_page(gfp_flags);
	if (!page)
		return NULL;

	if (radix_tree_preload(GFP_NOIO)) {
		__free_page(page);
		return NULL;
	}

	spin_lock(&brd->brd_lock);
	idx = sector >> PAGE_SECTORS_SHIFT;
	page->index = idx;
	if (radix_tree_insert(&brd->brd_pages, idx, page)) {
		__free_page(page);
		page = radix_tree_lookup(&brd->brd_pages, idx);
		BUG_ON(!page);
		BUG_ON(page->index != idx);
	}
	spin_unlock(&brd->brd_lock);

	radix_tree_preload_end();

	return page;
}

static void brd_free_page(struct brd_device *brd, sector_t sector)
{
	struct page *page;
	pgoff_t idx;

	spin_lock(&brd->brd_lock);
	idx = sector >> PAGE_SECTORS_SHIFT;
	page = radix_tree_delete(&brd->brd_pages, idx);
	spin_unlock(&brd->brd_lock);
	if (page)
		__free_page(page);
}

static void brd_zero_page(struct brd_device *brd, sector_t sector)
{
	struct page *page;

	page = brd_lookup_page(brd, sector);
	if (page)
		clear_highpage(page);
}

/*
 * Free all backing store pages and radix tree. This must only be called when
 * there are no other users of the device.
 */
#define FREE_BATCH 16
static void brd_free_pages(struct brd_device *brd)
{
	unsigned long pos = 0;
	struct page *pages[FREE_BATCH];
	int nr_pages;

	do {
		int i;

		nr_pages = radix_tree_gang_lookup(&brd->brd_pages,
				(void **)pages, pos, FREE_BATCH);

		for (i = 0; i < nr_pages; i++) {
			void *ret;

			BUG_ON(pages[i]->index < pos);
			pos = pages[i]->index;
			ret = radix_tree_delete(&brd->brd_pages, pos);
			BUG_ON(!ret || ret != pages[i]);
			__free_page(pages[i]);
		}

		pos++;

		/*
		 * This assumes radix_tree_gang_lookup always returns as
		 * many pages as possible. If the radix-tree code changes,
		 * so will this have to.
		 */
	} while (nr_pages == FREE_BATCH);
}

/*
 * copy_to_brd_setup must be called before copy_to_brd. It may sleep.
 */
static int copy_to_brd_setup(struct brd_device *brd, sector_t sector, size_t n)
{
	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	size_t copy;

	copy = min_t(size_t, n, PAGE_SIZE - offset);
	if (!brd_insert_page(brd, sector))
		return -ENOSPC;
	if (copy < n) {
		sector += copy >> SECTOR_SHIFT;
		if (!brd_insert_page(brd, sector))
			return -ENOSPC;
	}
	return 0;
}

static void discard_from_brd(struct brd_device *brd,
			sector_t sector, size_t n)
{
	while (n >= PAGE_SIZE) {
		/*
		 * Don't want to actually discard pages here because
		 * re-allocating the pages can result in writeback
		 * deadlocks under heavy load.
		 */
		if (0)
			brd_free_page(brd, sector);
		else
			brd_zero_page(brd, sector);
		sector += PAGE_SIZE >> SECTOR_SHIFT;
		n -= PAGE_SIZE;
	}
}

/*
 * Copy n bytes from src to the brd starting at sector. Does not sleep.
 */
static void copy_to_brd(struct brd_device *brd, const void *src,
			sector_t sector, size_t n)
{
	struct page *page;
	void *dst;
	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	size_t copy;

	copy = min_t(size_t, n, PAGE_SIZE - offset);
	page = brd_lookup_page(brd, sector);
	BUG_ON(!page);

	dst = kmap_atomic(page);
	memcpy(dst + offset, src, copy);
	kunmap_atomic(dst);

	if (copy < n) {
		src += copy;
		sector += copy >> SECTOR_SHIFT;
		copy = n - copy;
		page = brd_lookup_page(brd, sector);
		BUG_ON(!page);

		dst = kmap_atomic(page);
		memcpy(dst, src, copy);
		kunmap_atomic(dst);
	}
}

/*
 * Copy n bytes to dst from the brd starting at sector. Does not sleep.
 */
static void copy_from_brd(void *dst, struct brd_device *brd,
			sector_t sector, size_t n)
{
	struct page *page;
	void *src;
	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	size_t copy;

	copy = min_t(size_t, n, PAGE_SIZE - offset);
	page = brd_lookup_page(brd, sector);
	if (page) {
		src = kmap_atomic(page);
		memcpy(dst, src + offset, copy);
		kunmap_atomic(src);
	} else
		memset(dst, 0, copy);

	if (copy < n) {
		dst += copy;
		sector += copy >> SECTOR_SHIFT;
		copy = n - copy;
		page = brd_lookup_page(brd, sector);
		if (page) {
			src = kmap_atomic(page);
			memcpy(dst, src, copy);
			kunmap_atomic(src);
		} else
			memset(dst, 0, copy);
	}
}

/*
 * Process a single bvec of a bio.
 */
static int brd_do_bvec(struct brd_device *brd, struct page *page,
			unsigned int len, unsigned int off, int rw,
			sector_t sector)
{
	void *mem;
	int err = 0;

	if (rw != READ) {
		err = copy_to_brd_setup(brd, sector, len);
		if (err)
			goto out;
	}

	mem = kmap_atomic(page);
	if (rw == READ) {
		copy_from_brd(mem + off, brd, sector, len);
		flush_dcache_page(page);
	} else {
		flush_dcache_page(page);
		copy_to_brd(brd, mem + off, sector, len);
	}
	kunmap_atomic(mem);

out:
	return err;
}

static void brd_make_request(struct request_queue *q, struct bio *bio)
{
	struct block_device *bdev = bio->bi_bdev;
	struct brd_device *brd = bdev->bd_disk->private_data;
	int rw;
	struct bio_vec bvec;
	sector_t sector;
	struct bvec_iter iter;
	int err = -EIO;

	sector = bio->bi_iter.bi_sector;
	if (bio_end_sector(bio) > get_capacity(bdev->bd_disk))
		goto out;

	if (unlikely(bio->bi_rw & REQ_DISCARD)) {
		err = 0;
		discard_from_brd(brd, sector, bio->bi_iter.bi_size);
		goto out;
	}

	rw = bio_rw(bio);
	if (rw == READA)
		rw = READ;

	bio_for_each_segment(bvec, bio, iter) {
		unsigned int len = bvec.bv_len;
		err = brd_do_bvec(brd, bvec.bv_page, len,
					bvec.bv_offset, rw, sector);
		if (err)
			break;
		sector += len >> SECTOR_SHIFT;
	}

out:
	bio_endio(bio, err);
}

static int brd_rw_page(struct block_device *bdev, sector_t sector,
		       struct page *page, int rw)
{
	struct brd_device *brd = bdev->bd_disk->private_data;
	int err = brd_do_bvec(brd, page, PAGE_CACHE_SIZE, 0, rw, sector);
	page_endio(page, rw & WRITE, err);
	return err;
}

#ifdef CONFIG_BLK_DEV_XIP
static int brd_direct_access(struct block_device *bdev, sector_t sector,
			void **kaddr, unsigned long *pfn)
{
	struct brd_device *brd = bdev->bd_disk->private_data;
	struct page *page;

	if (!brd)
		return -ENODEV;
	if (sector & (PAGE_SECTORS-1))
		return -EINVAL;
	if (sector + PAGE_SECTORS > get_capacity(bdev->bd_disk))
		return -ERANGE;
	page = brd_insert_page(brd, sector);
	if (!page)
		return -ENOSPC;
	*kaddr = page_address(page);
	*pfn = page_to_pfn(page);

	return 0;
}
#endif

static int brd_ioctl(struct block_device *bdev, fmode_t mode,
			unsigned int cmd, unsigned long arg)
{
	int error;
	struct brd_device *brd = bdev->bd_disk->private_data;

	if (cmd != BLKFLSBUF)
		return -ENOTTY;

	/*
	 * ram device BLKFLSBUF has special semantics, we want to actually
	 * release and destroy the ramdisk data.
	 */
	mutex_lock(&brd_mutex);
	mutex_lock(&bdev->bd_mutex);
	error = -EBUSY;
	if (bdev->bd_openers <= 1) {
		/*
		 * Kill the cache first, so it isn't written back to the
		 * device.
		 *
		 * Another thread might instantiate more buffercache here,
		 * but there is not much we can do to close that race.
		 */
		kill_bdev(bdev);
		brd_free_pages(brd);
		error = 0;
	}
	mutex_unlock(&bdev->bd_mutex);
	mutex_unlock(&brd_mutex);

	return error;
}

static const struct block_device_operations brd_fops = {
	.owner =		THIS_MODULE,
	.rw_page =		brd_rw_page,
	.ioctl =		brd_ioctl,
#ifdef CONFIG_BLK_DEV_XIP
	.direct_access =	brd_direct_access,
#endif
};

/*
 * And now the modules code and kernel interface.
 */
static int rd_nr;
int rd_size = CONFIG_BLK_DEV_RAM_SIZE;
static int max_part;
static int part_shift;
static int part_show = 0;
module_param(rd_nr, int, S_IRUGO);
MODULE_PARM_DESC(rd_nr, "Maximum number of brd devices");
module_param(rd_size, int, S_IRUGO);
MODULE_PARM_DESC(rd_size, "Size of each RAM disk in kbytes.");
module_param(max_part, int, S_IRUGO);
MODULE_PARM_DESC(max_part, "Maximum number of partitions per RAM disk");
module_param(part_show, int, S_IRUGO);
MODULE_PARM_DESC(part_show, "Control RAM disk visibility in /proc/partitions");
MODULE_LICENSE("GPL");
MODULE_ALIAS_BLOCKDEV_MAJOR(RAMDISK_MAJOR);
MODULE_ALIAS("rd");

#ifndef MODULE
/* Legacy boot options - nonmodular */
static int __init ramdisk_size(char *str)
{
	rd_size = simple_strtol(str, NULL, 0);
	return 1;
}
__setup("ramdisk_size=", ramdisk_size);
#endif

/*
 * The device scheme is derived from loop.c. Keep them in synch where possible
 * (should share code eventually).
 */
static LIST_HEAD(brd_devices);
static DEFINE_MUTEX(brd_devices_mutex);

static struct brd_device *brd_alloc(int i)
{
	struct brd_device *brd;
	struct gendisk *disk;

	brd = kzalloc(sizeof(*brd), GFP_KERNEL);
	if (!brd)
		goto out;
	brd->brd_number		= i;
	spin_lock_init(&brd->brd_lock);
	INIT_RADIX_TREE(&brd->brd_pages, GFP_ATOMIC);

	brd->brd_queue = blk_alloc_queue(GFP_KERNEL);
	if (!brd->brd_queue)
		goto out_free_dev;
	blk_queue_make_request(brd->brd_queue, brd_make_request);
	blk_queue_max_hw_sectors(brd->brd_queue, 1024);
	blk_queue_bounce_limit(brd->brd_queue, BLK_BOUNCE_ANY);

	brd->brd_queue->limits.discard_granularity = PAGE_SIZE;
	brd->brd_queue->limits.max_discard_sectors = UINT_MAX;
	brd->brd_queue->limits.discard_zeroes_data = 1;
	queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, brd->brd_queue);

	disk = brd->brd_disk = alloc_disk(1 << part_shift);
	if (!disk)
		goto out_free_queue;
	disk->major		= RAMDISK_MAJOR;
	disk->first_minor	= i << part_shift;
	disk->fops		= &brd_fops;
	disk->private_data	= brd;
	disk->queue		= brd->brd_queue;
	if (!part_show)
		disk->flags |= GENHD_FL_SUPPRESS_PARTITION_INFO;
	sprintf(disk->disk_name, "ram%d", i);
	set_capacity(disk, rd_size * 2);

	return brd;

out_free_queue:
	blk_cleanup_queue(brd->brd_queue);
out_free_dev:
	kfree(brd);
out:
	return NULL;
}

static void brd_free(struct brd_device *brd)
{
	put_disk(brd->brd_disk);
	blk_cleanup_queue(brd->brd_queue);
	brd_free_pages(brd);
	kfree(brd);
}

static struct brd_device *brd_init_one(int i)
{
	struct brd_device *brd;

	list_for_each_entry(brd, &brd_devices, brd_list) {
		if (brd->brd_number == i)
			goto out;
	}

	brd = brd_alloc(i);
	if (brd) {
		add_disk(brd->brd_disk);
		list_add_tail(&brd->brd_list, &brd_devices);
	}
out:
	return brd;
}

static void brd_del_one(struct brd_device *brd)
{
	list_del(&brd->brd_list);
	del_gendisk(brd->brd_disk);
	brd_free(brd);
}

static struct kobject *brd_probe(dev_t dev, int *part, void *data)
{
	struct brd_device *brd;
	struct kobject *kobj;

	mutex_lock(&brd_devices_mutex);
	brd = brd_init_one(MINOR(dev) >> part_shift);
	kobj = brd ? get_disk(brd->brd_disk) : NULL;
	mutex_unlock(&brd_devices_mutex);

	*part = 0;
	return kobj;
}

static int __init brd_init(void)
{
	int i, nr;
	unsigned long range;
	struct brd_device *brd, *next;

	/*
	 * brd module now has a feature to instantiate underlying device
	 * structure on-demand, provided that there is an access dev node.
	 * However, this will not work well with user space tool that doesn't
	 * know about such "feature".  In order to not break any existing
	 * tool, we do the following:
	 *
	 * (1) if rd_nr is specified, create that many upfront, and this
	 *     also becomes a hard limit.
	 * (2) if rd_nr is not specified, create CONFIG_BLK_DEV_RAM_COUNT
	 *     (default 16) rd device on module load, user can further
	 *     extend brd device by create dev node themselves and have
	 *     kernel automatically instantiate actual device on-demand.
	 */

	part_shift = 0;
	if (max_part > 0) {
		part_shift = fls(max_part);

		/*
		 * Adjust max_part according to part_shift as it is exported
		 * to user space so that user can decide correct minor number
		 * if [s]he want to create more devices.
		 *
		 * Note that -1 is required because partition 0 is reserved
		 * for the whole disk.
		 */
		max_part = (1UL << part_shift) - 1;
	}

	if ((1UL << part_shift) > DISK_MAX_PARTS)
		return -EINVAL;

	if (rd_nr > 1UL << (MINORBITS - part_shift))
		return -EINVAL;

	if (rd_nr) {
		nr = rd_nr;
		range = rd_nr << part_shift;
	} else {
		nr = CONFIG_BLK_DEV_RAM_COUNT;
		range = 1UL << MINORBITS;
	}

	if (register_blkdev(RAMDISK_MAJOR, "ramdisk"))
		return -EIO;

	for (i = 0; i < nr; i++) {
		brd = brd_alloc(i);
		if (!brd)
			goto out_free;
		list_add_tail(&brd->brd_list, &brd_devices);
	}

	/* point of no return */

	list_for_each_entry(brd, &brd_devices, brd_list)
		add_disk(brd->brd_disk);

	blk_register_region(MKDEV(RAMDISK_MAJOR, 0), range,
				  THIS_MODULE, brd_probe, NULL, NULL);

	printk(KERN_INFO "brd: module loaded\n");
	return 0;

out_free:
	list_for_each_entry_safe(brd, next, &brd_devices, brd_list) {
		list_del(&brd->brd_list);
		brd_free(brd);
	}
	unregister_blkdev(RAMDISK_MAJOR, "ramdisk");

	return -ENOMEM;
}

static void __exit brd_exit(void)
{
	unsigned long range;
	struct brd_device *brd, *next;

	range = rd_nr ? rd_nr << part_shift : 1UL << MINORBITS;

	list_for_each_entry_safe(brd, next, &brd_devices, brd_list)
		brd_del_one(brd);

	blk_unregister_region(MKDEV(RAMDISK_MAJOR, 0), range);
	unregister_blkdev(RAMDISK_MAJOR, "ramdisk");
}

module_init(brd_init);
module_exit(brd_exit);
Commit	Line	Data
9db5579b NP	1	/*
	2	* Ram backed block device driver.
	3	*
	4	* Copyright (C) 2007 Nick Piggin
	5	* Copyright (C) 2007 Novell Inc.
	6	*
	7	* Parts derived from drivers/block/rd.c, and drivers/block/loop.c, copyright
	8	* of their respective owners.
	9	*/
	10
	11	#include <linux/init.h>
	12	#include <linux/module.h>
	13	#include <linux/moduleparam.h>
	14	#include <linux/major.h>
	15	#include <linux/blkdev.h>
	16	#include <linux/bio.h>
	17	#include <linux/highmem.h>
2a48fc0a	18	#include <linux/mutex.h>
9db5579b	19	#include <linux/radix-tree.h>
ff01bb48	20	#include <linux/fs.h>
5a0e3ad6	21	#include <linux/slab.h>
9db5579b NP	22
	23	#include <asm/uaccess.h>
	24
	25	#define SECTOR_SHIFT 9
	26	#define PAGE_SECTORS_SHIFT (PAGE_SHIFT - SECTOR_SHIFT)
	27	#define PAGE_SECTORS (1 << PAGE_SECTORS_SHIFT)
	28
	29	/*
	30	* Each block ramdisk device has a radix_tree brd_pages of pages that stores
	31	* the pages containing the block device's contents. A brd page's ->index is
	32	* its offset in PAGE_SIZE units. This is similar to, but in no way connected
	33	* with, the kernel's pagecache or buffer cache (which sit above our block
	34	* device).
	35	*/
	36	struct brd_device {
	37	int brd_number;
9db5579b NP	38
	39	struct request_queue *brd_queue;
	40	struct gendisk *brd_disk;
	41	struct list_head brd_list;
	42
	43	/*
	44	* Backing store of pages and lock to protect it. This is the contents
	45	* of the block device.
	46	*/
	47	spinlock_t brd_lock;
	48	struct radix_tree_root brd_pages;
	49	};
	50
	51	/*
	52	* Look up and return a brd's page for a given sector.
	53	*/
2a48fc0a	54	static DEFINE_MUTEX(brd_mutex);
9db5579b NP	55	static struct page brd_lookup_page(struct brd_device brd, sector_t sector)
	56	{
	57	pgoff_t idx;
	58	struct page *page;
	59
	60	/*
	61	* The page lifetime is protected by the fact that we have opened the
	62	* device node -- brd pages will never be deleted under us, so we
	63	* don't need any further locking or refcounting.
	64	*
	65	* This is strictly true for the radix-tree nodes as well (ie. we
	66	* don't actually need the rcu_read_lock()), however that is not a
	67	* documented feature of the radix-tree API so it is better to be
	68	* safe here (we don't have total exclusion from radix tree updates
	69	* here, only deletes).
	70	*/
	71	rcu_read_lock();
	72	idx = sector >> PAGE_SECTORS_SHIFT; /* sector to page index */
	73	page = radix_tree_lookup(&brd->brd_pages, idx);
	74	rcu_read_unlock();
	75
	76	BUG_ON(page && page->index != idx);
	77
	78	return page;
	79	}
	80
	81	/*
	82	* Look up and return a brd's page for a given sector.
	83	* If one does not exist, allocate an empty page, and insert that. Then
	84	* return it.
	85	*/
	86	static struct page brd_insert_page(struct brd_device brd, sector_t sector)
	87	{
	88	pgoff_t idx;
	89	struct page *page;
75acb9cd	90	gfp_t gfp_flags;
9db5579b NP	91
	92	page = brd_lookup_page(brd, sector);
	93	if (page)
	94	return page;
	95
	96	/*
	97	* Must use NOIO because we don't want to recurse back into the
	98	* block or filesystem layers from page reclaim.
75acb9cd NP	99	*
	100	* Cannot support XIP and highmem, because our ->direct_access
	101	* routine for XIP must return memory that is always addressable.
	102	* If XIP was reworked to use pfns and kmap throughout, this
	103	* restriction might be able to be lifted.
9db5579b	104	*/
75acb9cd NP	105	gfp_flags = GFP_NOIO \| __GFP_ZERO;
	106	#ifndef CONFIG_BLK_DEV_XIP
	107	gfp_flags \|= __GFP_HIGHMEM;
	108	#endif
26defe34	109	page = alloc_page(gfp_flags);
9db5579b NP	110	if (!page)
	111	return NULL;
	112
	113	if (radix_tree_preload(GFP_NOIO)) {
	114	__free_page(page);
	115	return NULL;
	116	}
	117
	118	spin_lock(&brd->brd_lock);
	119	idx = sector >> PAGE_SECTORS_SHIFT;
dfd20b2b	120	page->index = idx;
9db5579b NP	121	if (radix_tree_insert(&brd->brd_pages, idx, page)) {
	122	__free_page(page);
	123	page = radix_tree_lookup(&brd->brd_pages, idx);
	124	BUG_ON(!page);
	125	BUG_ON(page->index != idx);
dfd20b2b	126	}
9db5579b NP	127	spin_unlock(&brd->brd_lock);
	128
	129	radix_tree_preload_end();
	130
	131	return page;
	132	}
	133
b7c33571 NP	134	static void brd_free_page(struct brd_device *brd, sector_t sector)
	135	{
	136	struct page *page;
	137	pgoff_t idx;
	138
	139	spin_lock(&brd->brd_lock);
	140	idx = sector >> PAGE_SECTORS_SHIFT;
	141	page = radix_tree_delete(&brd->brd_pages, idx);
	142	spin_unlock(&brd->brd_lock);
	143	if (page)
	144	__free_page(page);
	145	}
	146
	147	static void brd_zero_page(struct brd_device *brd, sector_t sector)
	148	{
	149	struct page *page;
	150
	151	page = brd_lookup_page(brd, sector);
	152	if (page)
	153	clear_highpage(page);
	154	}
	155
9db5579b NP	156	/*
	157	* Free all backing store pages and radix tree. This must only be called when
	158	* there are no other users of the device.
	159	*/
	160	#define FREE_BATCH 16
	161	static void brd_free_pages(struct brd_device *brd)
	162	{
	163	unsigned long pos = 0;
	164	struct page *pages[FREE_BATCH];
	165	int nr_pages;
	166
	167	do {
	168	int i;
	169
	170	nr_pages = radix_tree_gang_lookup(&brd->brd_pages,
	171	(void **)pages, pos, FREE_BATCH);
	172
	173	for (i = 0; i < nr_pages; i++) {
	174	void *ret;
	175
	176	BUG_ON(pages[i]->index < pos);
	177	pos = pages[i]->index;
	178	ret = radix_tree_delete(&brd->brd_pages, pos);
	179	BUG_ON(!ret \|\| ret != pages[i]);
	180	__free_page(pages[i]);
	181	}
	182
	183	pos++;
	184
	185	/*
	186	* This assumes radix_tree_gang_lookup always returns as
	187	* many pages as possible. If the radix-tree code changes,
	188	* so will this have to.
	189	*/
	190	} while (nr_pages == FREE_BATCH);
	191	}
	192
	193	/*
	194	* copy_to_brd_setup must be called before copy_to_brd. It may sleep.
	195	*/
	196	static int copy_to_brd_setup(struct brd_device *brd, sector_t sector, size_t n)
	197	{
	198	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	199	size_t copy;
	200
	201	copy = min_t(size_t, n, PAGE_SIZE - offset);
	202	if (!brd_insert_page(brd, sector))
96f8d8e0	203	return -ENOSPC;
9db5579b NP	204	if (copy < n) {
	205	sector += copy >> SECTOR_SHIFT;
	206	if (!brd_insert_page(brd, sector))
96f8d8e0	207	return -ENOSPC;
9db5579b NP	208	}
	209	return 0;
	210	}
	211
b7c33571 NP	212	static void discard_from_brd(struct brd_device *brd,
	213	sector_t sector, size_t n)
	214	{
	215	while (n >= PAGE_SIZE) {
	216	/*
	217	* Don't want to actually discard pages here because
	218	* re-allocating the pages can result in writeback
	219	* deadlocks under heavy load.
	220	*/
	221	if (0)
	222	brd_free_page(brd, sector);
	223	else
	224	brd_zero_page(brd, sector);
	225	sector += PAGE_SIZE >> SECTOR_SHIFT;
	226	n -= PAGE_SIZE;
	227	}
	228	}
	229
9db5579b NP	230	/*
	231	* Copy n bytes from src to the brd starting at sector. Does not sleep.
	232	*/
	233	static void copy_to_brd(struct brd_device brd, const void src,
	234	sector_t sector, size_t n)
	235	{
	236	struct page *page;
	237	void *dst;
	238	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	239	size_t copy;
	240
	241	copy = min_t(size_t, n, PAGE_SIZE - offset);
	242	page = brd_lookup_page(brd, sector);
	243	BUG_ON(!page);
	244
cfd8005c	245	dst = kmap_atomic(page);
9db5579b	246	memcpy(dst + offset, src, copy);
cfd8005c	247	kunmap_atomic(dst);
9db5579b NP	248
	249	if (copy < n) {
	250	src += copy;
	251	sector += copy >> SECTOR_SHIFT;
	252	copy = n - copy;
	253	page = brd_lookup_page(brd, sector);
	254	BUG_ON(!page);
	255
cfd8005c	256	dst = kmap_atomic(page);
9db5579b	257	memcpy(dst, src, copy);
cfd8005c	258	kunmap_atomic(dst);
9db5579b NP	259	}
	260	}
	261
	262	/*
	263	* Copy n bytes to dst from the brd starting at sector. Does not sleep.
	264	*/
	265	static void copy_from_brd(void dst, struct brd_device brd,
	266	sector_t sector, size_t n)
	267	{
	268	struct page *page;
	269	void *src;
	270	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	271	size_t copy;
	272
	273	copy = min_t(size_t, n, PAGE_SIZE - offset);
	274	page = brd_lookup_page(brd, sector);
	275	if (page) {
cfd8005c	276	src = kmap_atomic(page);
9db5579b	277	memcpy(dst, src + offset, copy);
cfd8005c	278	kunmap_atomic(src);
9db5579b NP	279	} else
	280	memset(dst, 0, copy);
	281
	282	if (copy < n) {
	283	dst += copy;
	284	sector += copy >> SECTOR_SHIFT;
	285	copy = n - copy;
	286	page = brd_lookup_page(brd, sector);
	287	if (page) {
cfd8005c	288	src = kmap_atomic(page);
9db5579b	289	memcpy(dst, src, copy);
cfd8005c	290	kunmap_atomic(src);
9db5579b NP	291	} else
	292	memset(dst, 0, copy);
	293	}
	294	}
	295
	296	/*
	297	* Process a single bvec of a bio.
	298	*/
	299	static int brd_do_bvec(struct brd_device brd, struct page page,
	300	unsigned int len, unsigned int off, int rw,
	301	sector_t sector)
	302	{
	303	void *mem;
	304	int err = 0;
	305
	306	if (rw != READ) {
	307	err = copy_to_brd_setup(brd, sector, len);
	308	if (err)
	309	goto out;
	310	}
	311
cfd8005c	312	mem = kmap_atomic(page);
9db5579b NP	313	if (rw == READ) {
	314	copy_from_brd(mem + off, brd, sector, len);
	315	flush_dcache_page(page);
c2572f2b NP	316	} else {
c2572f2b NP	317	flush_dcache_page(page);
9db5579b	318	copy_to_brd(brd, mem + off, sector, len);
c2572f2b	319	}
cfd8005c	320	kunmap_atomic(mem);
9db5579b NP	321
	322	out:
	323	return err;
	324	}
	325
5a7bbad2	326	static void brd_make_request(struct request_queue q, struct bio bio)
9db5579b NP	327	{
	328	struct block_device *bdev = bio->bi_bdev;
	329	struct brd_device *brd = bdev->bd_disk->private_data;
	330	int rw;
7988613b	331	struct bio_vec bvec;
9db5579b	332	sector_t sector;
7988613b	333	struct bvec_iter iter;
9db5579b NP	334	int err = -EIO;
9db5579b NP	335
4f024f37	336	sector = bio->bi_iter.bi_sector;
f73a1c7d	337	if (bio_end_sector(bio) > get_capacity(bdev->bd_disk))
9db5579b NP	338	goto out;
9db5579b NP	339
7b6d91da	340	if (unlikely(bio->bi_rw & REQ_DISCARD)) {
b7c33571	341	err = 0;
4f024f37	342	discard_from_brd(brd, sector, bio->bi_iter.bi_size);
b7c33571 NP	343	goto out;
	344	}
	345
9db5579b NP	346	rw = bio_rw(bio);
	347	if (rw == READA)
	348	rw = READ;
	349
7988613b KO	350	bio_for_each_segment(bvec, bio, iter) {
	351	unsigned int len = bvec.bv_len;
	352	err = brd_do_bvec(brd, bvec.bv_page, len,
	353	bvec.bv_offset, rw, sector);
9db5579b NP	354	if (err)
	355	break;
	356	sector += len >> SECTOR_SHIFT;
	357	}
	358
	359	out:
	360	bio_endio(bio, err);
9db5579b NP	361	}
9db5579b NP	362
a72132c3 MW	363	static int brd_rw_page(struct block_device *bdev, sector_t sector,
	364	struct page *page, int rw)
	365	{
	366	struct brd_device *brd = bdev->bd_disk->private_data;
	367	int err = brd_do_bvec(brd, page, PAGE_CACHE_SIZE, 0, rw, sector);
	368	page_endio(page, rw & WRITE, err);
	369	return err;
	370	}
	371
75acb9cd	372	#ifdef CONFIG_BLK_DEV_XIP
b7c33571	373	static int brd_direct_access(struct block_device *bdev, sector_t sector,
30afcb4b	374	void *kaddr, unsigned long pfn)
75acb9cd NP	375	{
	376	struct brd_device *brd = bdev->bd_disk->private_data;
	377	struct page *page;
	378
	379	if (!brd)
	380	return -ENODEV;
	381	if (sector & (PAGE_SECTORS-1))
	382	return -EINVAL;
	383	if (sector + PAGE_SECTORS > get_capacity(bdev->bd_disk))
	384	return -ERANGE;
	385	page = brd_insert_page(brd, sector);
	386	if (!page)
96f8d8e0	387	return -ENOSPC;
30afcb4b JH	388	*kaddr = page_address(page);
30afcb4b JH	389	*pfn = page_to_pfn(page);
75acb9cd NP	390
	391	return 0;
	392	}
	393	#endif
	394
2b9ecd03	395	static int brd_ioctl(struct block_device *bdev, fmode_t mode,
9db5579b NP	396	unsigned int cmd, unsigned long arg)
	397	{
	398	int error;
9db5579b NP	399	struct brd_device *brd = bdev->bd_disk->private_data;
	400
	401	if (cmd != BLKFLSBUF)
	402	return -ENOTTY;
	403
	404	/*
	405	* ram device BLKFLSBUF has special semantics, we want to actually
	406	* release and destroy the ramdisk data.
	407	*/
2a48fc0a	408	mutex_lock(&brd_mutex);
9db5579b NP	409	mutex_lock(&bdev->bd_mutex);
	410	error = -EBUSY;
	411	if (bdev->bd_openers <= 1) {
	412	/*
ff01bb48 AV	413	* Kill the cache first, so it isn't written back to the
ff01bb48 AV	414	* device.
9db5579b NP	415	*
	416	* Another thread might instantiate more buffercache here,
	417	* but there is not much we can do to close that race.
	418	*/
ff01bb48	419	kill_bdev(bdev);
9db5579b NP	420	brd_free_pages(brd);
	421	error = 0;
	422	}
	423	mutex_unlock(&bdev->bd_mutex);
2a48fc0a	424	mutex_unlock(&brd_mutex);
9db5579b NP	425
	426	return error;
	427	}
	428
83d5cde4	429	static const struct block_device_operations brd_fops = {
75acb9cd	430	.owner = THIS_MODULE,
a72132c3	431	.rw_page = brd_rw_page,
8a6cfeb6	432	.ioctl = brd_ioctl,
75acb9cd NP	433	#ifdef CONFIG_BLK_DEV_XIP
	434	.direct_access = brd_direct_access,
	435	#endif
9db5579b NP	436	};
	437
	438	/*
	439	* And now the modules code and kernel interface.
	440	*/
	441	static int rd_nr;
	442	int rd_size = CONFIG_BLK_DEV_RAM_SIZE;
d7853d1f LV	443	static int max_part;
d7853d1f LV	444	static int part_shift;
aeac3181	445	static int part_show = 0;
8892cbaf	446	module_param(rd_nr, int, S_IRUGO);
9db5579b	447	MODULE_PARM_DESC(rd_nr, "Maximum number of brd devices");
8892cbaf	448	module_param(rd_size, int, S_IRUGO);
9db5579b	449	MODULE_PARM_DESC(rd_size, "Size of each RAM disk in kbytes.");
8892cbaf	450	module_param(max_part, int, S_IRUGO);
d7853d1f	451	MODULE_PARM_DESC(max_part, "Maximum number of partitions per RAM disk");
aeac3181 DM	452	module_param(part_show, int, S_IRUGO);
aeac3181 DM	453	MODULE_PARM_DESC(part_show, "Control RAM disk visibility in /proc/partitions");
9db5579b NP	454	MODULE_LICENSE("GPL");
9db5579b NP	455	MODULE_ALIAS_BLOCKDEV_MAJOR(RAMDISK_MAJOR);
efedf51c	456	MODULE_ALIAS("rd");
9db5579b NP	457
	458	#ifndef MODULE
	459	/* Legacy boot options - nonmodular */
	460	static int __init ramdisk_size(char *str)
	461	{
	462	rd_size = simple_strtol(str, NULL, 0);
	463	return 1;
	464	}
1adbee50	465	__setup("ramdisk_size=", ramdisk_size);
9db5579b NP	466	#endif
	467
	468	/*
	469	* The device scheme is derived from loop.c. Keep them in synch where possible
	470	* (should share code eventually).
	471	*/
	472	static LIST_HEAD(brd_devices);
	473	static DEFINE_MUTEX(brd_devices_mutex);
	474
	475	static struct brd_device *brd_alloc(int i)
	476	{
	477	struct brd_device *brd;
	478	struct gendisk *disk;
	479
	480	brd = kzalloc(sizeof(*brd), GFP_KERNEL);
	481	if (!brd)
	482	goto out;
	483	brd->brd_number = i;
	484	spin_lock_init(&brd->brd_lock);
	485	INIT_RADIX_TREE(&brd->brd_pages, GFP_ATOMIC);
	486
	487	brd->brd_queue = blk_alloc_queue(GFP_KERNEL);
	488	if (!brd->brd_queue)
	489	goto out_free_dev;
	490	blk_queue_make_request(brd->brd_queue, brd_make_request);
086fa5ff	491	blk_queue_max_hw_sectors(brd->brd_queue, 1024);
9db5579b NP	492	blk_queue_bounce_limit(brd->brd_queue, BLK_BOUNCE_ANY);
9db5579b NP	493
b7c33571 NP	494	brd->brd_queue->limits.discard_granularity = PAGE_SIZE;
	495	brd->brd_queue->limits.max_discard_sectors = UINT_MAX;
	496	brd->brd_queue->limits.discard_zeroes_data = 1;
	497	queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, brd->brd_queue);
	498
d7853d1f	499	disk = brd->brd_disk = alloc_disk(1 << part_shift);
9db5579b NP	500	if (!disk)
	501	goto out_free_queue;
	502	disk->major = RAMDISK_MAJOR;
d7853d1f	503	disk->first_minor = i << part_shift;
9db5579b NP	504	disk->fops = &brd_fops;
	505	disk->private_data = brd;
	506	disk->queue = brd->brd_queue;
aeac3181 DM	507	if (!part_show)
aeac3181 DM	508	disk->flags \|= GENHD_FL_SUPPRESS_PARTITION_INFO;
9db5579b NP	509	sprintf(disk->disk_name, "ram%d", i);
	510	set_capacity(disk, rd_size * 2);
	511
	512	return brd;
	513
	514	out_free_queue:
	515	blk_cleanup_queue(brd->brd_queue);
	516	out_free_dev:
	517	kfree(brd);
	518	out:
	519	return NULL;
	520	}
	521
	522	static void brd_free(struct brd_device *brd)
	523	{
	524	put_disk(brd->brd_disk);
	525	blk_cleanup_queue(brd->brd_queue);
	526	brd_free_pages(brd);
	527	kfree(brd);
	528	}
	529
	530	static struct brd_device *brd_init_one(int i)
	531	{
	532	struct brd_device *brd;
	533
	534	list_for_each_entry(brd, &brd_devices, brd_list) {
	535	if (brd->brd_number == i)
	536	goto out;
	537	}
	538
	539	brd = brd_alloc(i);
	540	if (brd) {
	541	add_disk(brd->brd_disk);
	542	list_add_tail(&brd->brd_list, &brd_devices);
	543	}
	544	out:
	545	return brd;
	546	}
	547
	548	static void brd_del_one(struct brd_device *brd)
	549	{
	550	list_del(&brd->brd_list);
	551	del_gendisk(brd->brd_disk);
	552	brd_free(brd);
	553	}
	554
	555	static struct kobject brd_probe(dev_t dev, int part, void *data)
	556	{
	557	struct brd_device *brd;
	558	struct kobject *kobj;
	559
	560	mutex_lock(&brd_devices_mutex);
af465668	561	brd = brd_init_one(MINOR(dev) >> part_shift);
a207f593	562	kobj = brd ? get_disk(brd->brd_disk) : NULL;
9db5579b NP	563	mutex_unlock(&brd_devices_mutex);
	564
	565	*part = 0;
	566	return kobj;
	567	}
	568
	569	static int __init brd_init(void)
	570	{
	571	int i, nr;
	572	unsigned long range;
	573	struct brd_device brd, next;
	574
	575	/*
	576	* brd module now has a feature to instantiate underlying device
	577	* structure on-demand, provided that there is an access dev node.
	578	* However, this will not work well with user space tool that doesn't
	579	* know about such "feature". In order to not break any existing
	580	* tool, we do the following:
	581	*
	582	* (1) if rd_nr is specified, create that many upfront, and this
	583	* also becomes a hard limit.
13868b76 NK	584	* (2) if rd_nr is not specified, create CONFIG_BLK_DEV_RAM_COUNT
	585	* (default 16) rd device on module load, user can further
	586	* extend brd device by create dev node themselves and have
	587	* kernel automatically instantiate actual device on-demand.
9db5579b	588	*/
d7853d1f LV	589
d7853d1f LV	590	part_shift = 0;
8892cbaf	591	if (max_part > 0) {
d7853d1f LV	592	part_shift = fls(max_part);
d7853d1f LV	593
8892cbaf NK	594	/*
	595	* Adjust max_part according to part_shift as it is exported
	596	* to user space so that user can decide correct minor number
	597	* if [s]he want to create more devices.
	598	*
	599	* Note that -1 is required because partition 0 is reserved
	600	* for the whole disk.
	601	*/
	602	max_part = (1UL << part_shift) - 1;
	603	}
	604
315980c8 NK	605	if ((1UL << part_shift) > DISK_MAX_PARTS)
	606	return -EINVAL;
	607
d7853d1f	608	if (rd_nr > 1UL << (MINORBITS - part_shift))
9db5579b NP	609	return -EINVAL;
	610
	611	if (rd_nr) {
	612	nr = rd_nr;
af465668	613	range = rd_nr << part_shift;
9db5579b NP	614	} else {
9db5579b NP	615	nr = CONFIG_BLK_DEV_RAM_COUNT;
af465668	616	range = 1UL << MINORBITS;
9db5579b NP	617	}
	618
	619	if (register_blkdev(RAMDISK_MAJOR, "ramdisk"))
	620	return -EIO;
	621
	622	for (i = 0; i < nr; i++) {
	623	brd = brd_alloc(i);
	624	if (!brd)
	625	goto out_free;
	626	list_add_tail(&brd->brd_list, &brd_devices);
	627	}
	628
	629	/* point of no return */
	630
	631	list_for_each_entry(brd, &brd_devices, brd_list)
	632	add_disk(brd->brd_disk);
	633
	634	blk_register_region(MKDEV(RAMDISK_MAJOR, 0), range,
	635	THIS_MODULE, brd_probe, NULL, NULL);
	636
	637	printk(KERN_INFO "brd: module loaded\n");
	638	return 0;
	639
	640	out_free:
	641	list_for_each_entry_safe(brd, next, &brd_devices, brd_list) {
	642	list_del(&brd->brd_list);
	643	brd_free(brd);
	644	}
c82f2966	645	unregister_blkdev(RAMDISK_MAJOR, "ramdisk");
9db5579b	646
9db5579b NP	647	return -ENOMEM;
	648	}
	649
	650	static void __exit brd_exit(void)
	651	{
	652	unsigned long range;
	653	struct brd_device brd, next;
	654
af465668	655	range = rd_nr ? rd_nr << part_shift : 1UL << MINORBITS;
9db5579b NP	656
	657	list_for_each_entry_safe(brd, next, &brd_devices, brd_list)
	658	brd_del_one(brd);
	659
	660	blk_unregister_region(MKDEV(RAMDISK_MAJOR, 0), range);
	661	unregister_blkdev(RAMDISK_MAJOR, "ramdisk");
	662	}
	663
	664	module_init(brd_init);
	665	module_exit(brd_exit);
	666