[linux-block.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/initrd.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 <linux/backing-dev.h>

#include <linux/uaccess.h>

#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 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.
	 */
	gfp_flags = GFP_NOIO | __GFP_ZERO | __GFP_HIGHMEM;
	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;
}

/*
 * 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++;

		/*
		 * It takes 3.4 seconds to remove 80GiB ramdisk.
		 * So, we need cond_resched to avoid stalling the CPU.
		 */
		cond_resched();

		/*
		 * 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;
}

/*
 * 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, unsigned int op,
			sector_t sector)
{
	void *mem;
	int err = 0;

	if (op_is_write(op)) {
		err = copy_to_brd_setup(brd, sector, len);
		if (err)
			goto out;
	}

	mem = kmap_atomic(page);
	if (!op_is_write(op)) {
		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 blk_qc_t brd_make_request(struct request_queue *q, struct bio *bio)
{
	struct brd_device *brd = bio->bi_disk->private_data;
	struct bio_vec bvec;
	sector_t sector;
	struct bvec_iter iter;

	sector = bio->bi_iter.bi_sector;
	if (bio_end_sector(bio) > get_capacity(bio->bi_disk))
		goto io_error;

	bio_for_each_segment(bvec, bio, iter) {
		unsigned int len = bvec.bv_len;
		int err;

		err = brd_do_bvec(brd, bvec.bv_page, len, bvec.bv_offset,
				  bio_op(bio), sector);
		if (err)
			goto io_error;
		sector += len >> SECTOR_SHIFT;
	}

	bio_endio(bio);
	return BLK_QC_T_NONE;
io_error:
	bio_io_error(bio);
	return BLK_QC_T_NONE;
}

static int brd_rw_page(struct block_device *bdev, sector_t sector,
		       struct page *page, unsigned int op)
{
	struct brd_device *brd = bdev->bd_disk->private_data;
	int err;

	if (PageTransHuge(page))
		return -ENOTSUPP;
	err = brd_do_bvec(brd, page, PAGE_SIZE, 0, op, sector);
	page_endio(page, op_is_write(op), err);
	return err;
}

static const struct block_device_operations brd_fops = {
	.owner =		THIS_MODULE,
	.rw_page =		brd_rw_page,
};

/*
 * And now the modules code and kernel interface.
 */
static int rd_nr = CONFIG_BLK_DEV_RAM_COUNT;
module_param(rd_nr, int, 0444);
MODULE_PARM_DESC(rd_nr, "Maximum number of brd devices");

unsigned long rd_size = CONFIG_BLK_DEV_RAM_SIZE;
module_param(rd_size, ulong, 0444);
MODULE_PARM_DESC(rd_size, "Size of each RAM disk in kbytes.");

static int max_part = 1;
module_param(max_part, int, 0444);
MODULE_PARM_DESC(max_part, "Num Minors to reserve between devices");

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);

	/* This is so fdisk will align partitions on 4k, because of
	 * direct_access API needing 4k alignment, returning a PFN
	 * (This is only a problem on very small devices <= 4M,
	 *  otherwise fdisk will align on 1M. Regardless this call
	 *  is harmless)
	 */
	blk_queue_physical_block_size(brd->brd_queue, PAGE_SIZE);
	disk = brd->brd_disk = alloc_disk(max_part);
	if (!disk)
		goto out_free_queue;
	disk->major		= RAMDISK_MAJOR;
	disk->first_minor	= i * max_part;
	disk->fops		= &brd_fops;
	disk->private_data	= brd;
	disk->flags		= GENHD_FL_EXT_DEVT;
	sprintf(disk->disk_name, "ram%d", i);
	set_capacity(disk, rd_size * 2);
	brd->brd_queue->backing_dev_info->capabilities |= BDI_CAP_SYNCHRONOUS_IO;

	/* Tell the block layer that this is not a rotational device */
	blk_queue_flag_set(QUEUE_FLAG_NONROT, brd->brd_queue);
	blk_queue_flag_clear(QUEUE_FLAG_ADD_RANDOM, brd->brd_queue);

	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, bool *new)
{
	struct brd_device *brd;

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

	brd = brd_alloc(i);
	if (brd) {
		brd->brd_disk->queue = brd->brd_queue;
		add_disk(brd->brd_disk);
		list_add_tail(&brd->brd_list, &brd_devices);
	}
	*new = true;
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;
	bool new;

	mutex_lock(&brd_devices_mutex);
	brd = brd_init_one(MINOR(dev) / max_part, &new);
	kobj = brd ? get_disk_and_module(brd->brd_disk) : NULL;
	mutex_unlock(&brd_devices_mutex);

	if (new)
		*part = 0;

	return kobj;
}

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

	/*
	 * brd module now has a feature to instantiate underlying device
	 * structure on-demand, provided that there is an access dev node.
	 *
	 * (1) if rd_nr is specified, create that many upfront. else
	 *     it defaults to CONFIG_BLK_DEV_RAM_COUNT
	 * (2) User can further extend brd devices by create dev node themselves
	 *     and have kernel automatically instantiate actual device
	 *     on-demand. Example:
	 *		mknod /path/devnod_name b 1 X	# 1 is the rd major
	 *		fdisk -l /path/devnod_name
	 *	If (X / max_part) was not already created it will be created
	 *	dynamically.
	 */

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

	if (unlikely(!max_part))
		max_part = 1;

	for (i = 0; i < rd_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) {
		/*
		 * associate with queue just before adding disk for
		 * avoiding to mess up failure path
		 */
		brd->brd_disk->queue = brd->brd_queue;
		add_disk(brd->brd_disk);
	}

	blk_register_region(MKDEV(RAMDISK_MAJOR, 0), 1UL << MINORBITS,
				  THIS_MODULE, brd_probe, NULL, NULL);

	pr_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");

	pr_info("brd: module NOT loaded !!!\n");
	return -ENOMEM;
}

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

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

	blk_unregister_region(MKDEV(RAMDISK_MAJOR, 0), 1UL << MINORBITS);
	unregister_blkdev(RAMDISK_MAJOR, "ramdisk");

	pr_info("brd: module unloaded\n");
}

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>
287f3ca5	12	#include <linux/initrd.h>
9db5579b NP	13	#include <linux/module.h>
	14	#include <linux/moduleparam.h>
	15	#include <linux/major.h>
	16	#include <linux/blkdev.h>
	17	#include <linux/bio.h>
	18	#include <linux/highmem.h>
2a48fc0a	19	#include <linux/mutex.h>
9db5579b	20	#include <linux/radix-tree.h>
ff01bb48	21	#include <linux/fs.h>
5a0e3ad6	22	#include <linux/slab.h>
23c47d2a	23	#include <linux/backing-dev.h>
9db5579b	24
7c0f6ba6	25	#include <linux/uaccess.h>
9db5579b	26
9db5579b NP	27	#define PAGE_SECTORS_SHIFT (PAGE_SHIFT - SECTOR_SHIFT)
	28	#define PAGE_SECTORS (1 << PAGE_SECTORS_SHIFT)
	29
	30	/*
	31	* Each block ramdisk device has a radix_tree brd_pages of pages that stores
	32	* the pages containing the block device's contents. A brd page's ->index is
	33	* its offset in PAGE_SIZE units. This is similar to, but in no way connected
	34	* with, the kernel's pagecache or buffer cache (which sit above our block
	35	* device).
	36	*/
	37	struct brd_device {
	38	int brd_number;
9db5579b NP	39
	40	struct request_queue *brd_queue;
	41	struct gendisk *brd_disk;
	42	struct list_head brd_list;
	43
	44	/*
	45	* Backing store of pages and lock to protect it. This is the contents
	46	* of the block device.
	47	*/
	48	spinlock_t brd_lock;
	49	struct radix_tree_root brd_pages;
	50	};
	51
	52	/*
	53	* Look up and return a brd's page for a given sector.
	54	*/
	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.
	99	*/
f6b50160	100	gfp_flags = GFP_NOIO \| __GFP_ZERO \| __GFP_HIGHMEM;
26defe34	101	page = alloc_page(gfp_flags);
9db5579b NP	102	if (!page)
	103	return NULL;
	104
	105	if (radix_tree_preload(GFP_NOIO)) {
	106	__free_page(page);
	107	return NULL;
	108	}
	109
	110	spin_lock(&brd->brd_lock);
	111	idx = sector >> PAGE_SECTORS_SHIFT;
dfd20b2b	112	page->index = idx;
9db5579b NP	113	if (radix_tree_insert(&brd->brd_pages, idx, page)) {
	114	__free_page(page);
	115	page = radix_tree_lookup(&brd->brd_pages, idx);
	116	BUG_ON(!page);
	117	BUG_ON(page->index != idx);
dfd20b2b	118	}
9db5579b NP	119	spin_unlock(&brd->brd_lock);
	120
	121	radix_tree_preload_end();
	122
	123	return page;
	124	}
	125
	126	/*
	127	* Free all backing store pages and radix tree. This must only be called when
	128	* there are no other users of the device.
	129	*/
	130	#define FREE_BATCH 16
	131	static void brd_free_pages(struct brd_device *brd)
	132	{
	133	unsigned long pos = 0;
	134	struct page *pages[FREE_BATCH];
	135	int nr_pages;
	136
	137	do {
	138	int i;
	139
	140	nr_pages = radix_tree_gang_lookup(&brd->brd_pages,
	141	(void **)pages, pos, FREE_BATCH);
	142
	143	for (i = 0; i < nr_pages; i++) {
	144	void *ret;
	145
	146	BUG_ON(pages[i]->index < pos);
	147	pos = pages[i]->index;
	148	ret = radix_tree_delete(&brd->brd_pages, pos);
	149	BUG_ON(!ret \|\| ret != pages[i]);
	150	__free_page(pages[i]);
	151	}
	152
	153	pos++;
	154
936b33f7 MP	155	/*
	156	* It takes 3.4 seconds to remove 80GiB ramdisk.
	157	* So, we need cond_resched to avoid stalling the CPU.
	158	*/
	159	cond_resched();
	160
9db5579b NP	161	/*
	162	* This assumes radix_tree_gang_lookup always returns as
	163	* many pages as possible. If the radix-tree code changes,
	164	* so will this have to.
	165	*/
	166	} while (nr_pages == FREE_BATCH);
	167	}
	168
	169	/*
	170	* copy_to_brd_setup must be called before copy_to_brd. It may sleep.
	171	*/
	172	static int copy_to_brd_setup(struct brd_device *brd, sector_t sector, size_t n)
	173	{
	174	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	175	size_t copy;
	176
	177	copy = min_t(size_t, n, PAGE_SIZE - offset);
	178	if (!brd_insert_page(brd, sector))
96f8d8e0	179	return -ENOSPC;
9db5579b NP	180	if (copy < n) {
	181	sector += copy >> SECTOR_SHIFT;
	182	if (!brd_insert_page(brd, sector))
96f8d8e0	183	return -ENOSPC;
9db5579b NP	184	}
	185	return 0;
	186	}
	187
	188	/*
	189	* Copy n bytes from src to the brd starting at sector. Does not sleep.
	190	*/
	191	static void copy_to_brd(struct brd_device brd, const void src,
	192	sector_t sector, size_t n)
	193	{
	194	struct page *page;
	195	void *dst;
	196	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	197	size_t copy;
	198
	199	copy = min_t(size_t, n, PAGE_SIZE - offset);
	200	page = brd_lookup_page(brd, sector);
	201	BUG_ON(!page);
	202
cfd8005c	203	dst = kmap_atomic(page);
9db5579b	204	memcpy(dst + offset, src, copy);
cfd8005c	205	kunmap_atomic(dst);
9db5579b NP	206
	207	if (copy < n) {
	208	src += copy;
	209	sector += copy >> SECTOR_SHIFT;
	210	copy = n - copy;
	211	page = brd_lookup_page(brd, sector);
	212	BUG_ON(!page);
	213
cfd8005c	214	dst = kmap_atomic(page);
9db5579b	215	memcpy(dst, src, copy);
cfd8005c	216	kunmap_atomic(dst);
9db5579b NP	217	}
	218	}
	219
	220	/*
	221	* Copy n bytes to dst from the brd starting at sector. Does not sleep.
	222	*/
	223	static void copy_from_brd(void dst, struct brd_device brd,
	224	sector_t sector, size_t n)
	225	{
	226	struct page *page;
	227	void *src;
	228	unsigned int offset = (sector & (PAGE_SECTORS-1)) << SECTOR_SHIFT;
	229	size_t copy;
	230
	231	copy = min_t(size_t, n, PAGE_SIZE - offset);
	232	page = brd_lookup_page(brd, sector);
	233	if (page) {
cfd8005c	234	src = kmap_atomic(page);
9db5579b	235	memcpy(dst, src + offset, copy);
cfd8005c	236	kunmap_atomic(src);
9db5579b NP	237	} else
	238	memset(dst, 0, copy);
	239
	240	if (copy < n) {
	241	dst += copy;
	242	sector += copy >> SECTOR_SHIFT;
	243	copy = n - copy;
	244	page = brd_lookup_page(brd, sector);
	245	if (page) {
cfd8005c	246	src = kmap_atomic(page);
9db5579b	247	memcpy(dst, src, copy);
cfd8005c	248	kunmap_atomic(src);
9db5579b NP	249	} else
	250	memset(dst, 0, copy);
	251	}
	252	}
	253
	254	/*
	255	* Process a single bvec of a bio.
	256	*/
	257	static int brd_do_bvec(struct brd_device brd, struct page page,
3f289dcb	258	unsigned int len, unsigned int off, unsigned int op,
9db5579b NP	259	sector_t sector)
	260	{
	261	void *mem;
	262	int err = 0;
	263
3f289dcb	264	if (op_is_write(op)) {
9db5579b NP	265	err = copy_to_brd_setup(brd, sector, len);
	266	if (err)
	267	goto out;
	268	}
	269
cfd8005c	270	mem = kmap_atomic(page);
3f289dcb	271	if (!op_is_write(op)) {
9db5579b NP	272	copy_from_brd(mem + off, brd, sector, len);
9db5579b NP	273	flush_dcache_page(page);
c2572f2b NP	274	} else {
c2572f2b NP	275	flush_dcache_page(page);
9db5579b	276	copy_to_brd(brd, mem + off, sector, len);
c2572f2b	277	}
cfd8005c	278	kunmap_atomic(mem);
9db5579b NP	279
	280	out:
	281	return err;
	282	}
	283
dece1635	284	static blk_qc_t brd_make_request(struct request_queue q, struct bio bio)
9db5579b	285	{
74d46992	286	struct brd_device *brd = bio->bi_disk->private_data;
7988613b	287	struct bio_vec bvec;
9db5579b	288	sector_t sector;
7988613b	289	struct bvec_iter iter;
9db5579b	290
4f024f37	291	sector = bio->bi_iter.bi_sector;
74d46992	292	if (bio_end_sector(bio) > get_capacity(bio->bi_disk))
4246a0b6	293	goto io_error;
9db5579b	294
7988613b KO	295	bio_for_each_segment(bvec, bio, iter) {
7988613b KO	296	unsigned int len = bvec.bv_len;
4246a0b6 CH	297	int err;
4246a0b6 CH	298
c11f0c0b	299	err = brd_do_bvec(brd, bvec.bv_page, len, bvec.bv_offset,
3f289dcb	300	bio_op(bio), sector);
9db5579b	301	if (err)
4246a0b6	302	goto io_error;
9db5579b NP	303	sector += len >> SECTOR_SHIFT;
	304	}
	305
4246a0b6	306	bio_endio(bio);
dece1635	307	return BLK_QC_T_NONE;
4246a0b6 CH	308	io_error:
4246a0b6 CH	309	bio_io_error(bio);
dece1635	310	return BLK_QC_T_NONE;
9db5579b NP	311	}
9db5579b NP	312
a72132c3	313	static int brd_rw_page(struct block_device *bdev, sector_t sector,
3f289dcb	314	struct page *page, unsigned int op)
a72132c3 MW	315	{
a72132c3 MW	316	struct brd_device *brd = bdev->bd_disk->private_data;
98cc093c HY	317	int err;
	318
	319	if (PageTransHuge(page))
	320	return -ENOTSUPP;
3f289dcb TH	321	err = brd_do_bvec(brd, page, PAGE_SIZE, 0, op, sector);
3f289dcb TH	322	page_endio(page, op_is_write(op), err);
a72132c3 MW	323	return err;
	324	}
	325
83d5cde4	326	static const struct block_device_operations brd_fops = {
75acb9cd	327	.owner = THIS_MODULE,
a72132c3	328	.rw_page = brd_rw_page,
9db5579b NP	329	};
	330
	331	/*
	332	* And now the modules code and kernel interface.
	333	*/
937af5ec	334	static int rd_nr = CONFIG_BLK_DEV_RAM_COUNT;
5657a819	335	module_param(rd_nr, int, 0444);
9db5579b	336	MODULE_PARM_DESC(rd_nr, "Maximum number of brd devices");
937af5ec	337
366f4aea	338	unsigned long rd_size = CONFIG_BLK_DEV_RAM_SIZE;
5657a819	339	module_param(rd_size, ulong, 0444);
9db5579b	340	MODULE_PARM_DESC(rd_size, "Size of each RAM disk in kbytes.");
937af5ec BH	341
937af5ec BH	342	static int max_part = 1;
5657a819	343	module_param(max_part, int, 0444);
937af5ec BH	344	MODULE_PARM_DESC(max_part, "Num Minors to reserve between devices");
937af5ec BH	345
9db5579b NP	346	MODULE_LICENSE("GPL");
9db5579b NP	347	MODULE_ALIAS_BLOCKDEV_MAJOR(RAMDISK_MAJOR);
efedf51c	348	MODULE_ALIAS("rd");
9db5579b NP	349
	350	#ifndef MODULE
	351	/* Legacy boot options - nonmodular */
	352	static int __init ramdisk_size(char *str)
	353	{
	354	rd_size = simple_strtol(str, NULL, 0);
	355	return 1;
	356	}
1adbee50	357	__setup("ramdisk_size=", ramdisk_size);
9db5579b NP	358	#endif
	359
	360	/*
	361	* The device scheme is derived from loop.c. Keep them in synch where possible
	362	* (should share code eventually).
	363	*/
	364	static LIST_HEAD(brd_devices);
	365	static DEFINE_MUTEX(brd_devices_mutex);
	366
	367	static struct brd_device *brd_alloc(int i)
	368	{
	369	struct brd_device *brd;
	370	struct gendisk *disk;
	371
	372	brd = kzalloc(sizeof(*brd), GFP_KERNEL);
	373	if (!brd)
	374	goto out;
	375	brd->brd_number = i;
	376	spin_lock_init(&brd->brd_lock);
	377	INIT_RADIX_TREE(&brd->brd_pages, GFP_ATOMIC);
	378
	379	brd->brd_queue = blk_alloc_queue(GFP_KERNEL);
	380	if (!brd->brd_queue)
	381	goto out_free_dev;
c8fa3173	382
9db5579b	383	blk_queue_make_request(brd->brd_queue, brd_make_request);
086fa5ff	384	blk_queue_max_hw_sectors(brd->brd_queue, 1024);
9db5579b	385
c8fa3173 BH	386	/* This is so fdisk will align partitions on 4k, because of
	387	* direct_access API needing 4k alignment, returning a PFN
	388	* (This is only a problem on very small devices <= 4M,
	389	* otherwise fdisk will align on 1M. Regardless this call
	390	* is harmless)
	391	*/
	392	blk_queue_physical_block_size(brd->brd_queue, PAGE_SIZE);
937af5ec	393	disk = brd->brd_disk = alloc_disk(max_part);
9db5579b NP	394	if (!disk)
	395	goto out_free_queue;
	396	disk->major = RAMDISK_MAJOR;
937af5ec	397	disk->first_minor = i * max_part;
9db5579b NP	398	disk->fops = &brd_fops;
9db5579b NP	399	disk->private_data = brd;
937af5ec	400	disk->flags = GENHD_FL_EXT_DEVT;
9db5579b NP	401	sprintf(disk->disk_name, "ram%d", i);
9db5579b NP	402	set_capacity(disk, rd_size * 2);
153fcd5f	403	brd->brd_queue->backing_dev_info->capabilities \|= BDI_CAP_SYNCHRONOUS_IO;
9db5579b	404
316ba573	405	/* Tell the block layer that this is not a rotational device */
153fcd5f ML	406	blk_queue_flag_set(QUEUE_FLAG_NONROT, brd->brd_queue);
153fcd5f ML	407	blk_queue_flag_clear(QUEUE_FLAG_ADD_RANDOM, brd->brd_queue);
316ba573	408
9db5579b NP	409	return brd;
	410
	411	out_free_queue:
	412	blk_cleanup_queue(brd->brd_queue);
	413	out_free_dev:
	414	kfree(brd);
	415	out:
	416	return NULL;
	417	}
	418
	419	static void brd_free(struct brd_device *brd)
	420	{
	421	put_disk(brd->brd_disk);
	422	blk_cleanup_queue(brd->brd_queue);
	423	brd_free_pages(brd);
	424	kfree(brd);
	425	}
	426
937af5ec	427	static struct brd_device brd_init_one(int i, bool new)
9db5579b NP	428	{
	429	struct brd_device *brd;
	430
937af5ec	431	*new = false;
9db5579b NP	432	list_for_each_entry(brd, &brd_devices, brd_list) {
	433	if (brd->brd_number == i)
	434	goto out;
	435	}
	436
	437	brd = brd_alloc(i);
	438	if (brd) {
153fcd5f	439	brd->brd_disk->queue = brd->brd_queue;
9db5579b NP	440	add_disk(brd->brd_disk);
	441	list_add_tail(&brd->brd_list, &brd_devices);
	442	}
937af5ec	443	*new = true;
9db5579b NP	444	out:
	445	return brd;
	446	}
	447
	448	static void brd_del_one(struct brd_device *brd)
	449	{
	450	list_del(&brd->brd_list);
	451	del_gendisk(brd->brd_disk);
	452	brd_free(brd);
	453	}
	454
	455	static struct kobject brd_probe(dev_t dev, int part, void *data)
	456	{
	457	struct brd_device *brd;
	458	struct kobject *kobj;
937af5ec	459	bool new;
9db5579b NP	460
9db5579b NP	461	mutex_lock(&brd_devices_mutex);
937af5ec	462	brd = brd_init_one(MINOR(dev) / max_part, &new);
3079c22e	463	kobj = brd ? get_disk_and_module(brd->brd_disk) : NULL;
9db5579b NP	464	mutex_unlock(&brd_devices_mutex);
9db5579b NP	465
937af5ec BH	466	if (new)
	467	*part = 0;
	468
9db5579b NP	469	return kobj;
	470	}
	471
	472	static int __init brd_init(void)
	473	{
9db5579b	474	struct brd_device brd, next;
937af5ec	475	int i;
9db5579b NP	476
	477	/*
	478	* brd module now has a feature to instantiate underlying device
	479	* structure on-demand, provided that there is an access dev node.
9db5579b	480	*
937af5ec BH	481	* (1) if rd_nr is specified, create that many upfront. else
	482	* it defaults to CONFIG_BLK_DEV_RAM_COUNT
	483	* (2) User can further extend brd devices by create dev node themselves
	484	* and have kernel automatically instantiate actual device
	485	* on-demand. Example:
	486	* mknod /path/devnod_name b 1 X # 1 is the rd major
	487	* fdisk -l /path/devnod_name
	488	* If (X / max_part) was not already created it will be created
	489	* dynamically.
9db5579b	490	*/
d7853d1f	491
9db5579b NP	492	if (register_blkdev(RAMDISK_MAJOR, "ramdisk"))
	493	return -EIO;
	494
937af5ec BH	495	if (unlikely(!max_part))
	496	max_part = 1;
	497
	498	for (i = 0; i < rd_nr; i++) {
9db5579b NP	499	brd = brd_alloc(i);
	500	if (!brd)
	501	goto out_free;
	502	list_add_tail(&brd->brd_list, &brd_devices);
	503	}
	504
	505	/* point of no return */
	506
153fcd5f ML	507	list_for_each_entry(brd, &brd_devices, brd_list) {
	508	/*
	509	* associate with queue just before adding disk for
	510	* avoiding to mess up failure path
	511	*/
	512	brd->brd_disk->queue = brd->brd_queue;
9db5579b	513	add_disk(brd->brd_disk);
153fcd5f	514	}
9db5579b	515
937af5ec	516	blk_register_region(MKDEV(RAMDISK_MAJOR, 0), 1UL << MINORBITS,
9db5579b NP	517	THIS_MODULE, brd_probe, NULL, NULL);
9db5579b NP	518
937af5ec	519	pr_info("brd: module loaded\n");
9db5579b NP	520	return 0;
	521
	522	out_free:
	523	list_for_each_entry_safe(brd, next, &brd_devices, brd_list) {
	524	list_del(&brd->brd_list);
	525	brd_free(brd);
	526	}
c82f2966	527	unregister_blkdev(RAMDISK_MAJOR, "ramdisk");
9db5579b	528
937af5ec	529	pr_info("brd: module NOT loaded !!!\n");
9db5579b NP	530	return -ENOMEM;
	531	}
	532
	533	static void __exit brd_exit(void)
	534	{
9db5579b NP	535	struct brd_device brd, next;
9db5579b NP	536
9db5579b NP	537	list_for_each_entry_safe(brd, next, &brd_devices, brd_list)
	538	brd_del_one(brd);
	539
937af5ec	540	blk_unregister_region(MKDEV(RAMDISK_MAJOR, 0), 1UL << MINORBITS);
9db5579b	541	unregister_blkdev(RAMDISK_MAJOR, "ramdisk");
937af5ec BH	542
937af5ec BH	543	pr_info("brd: module unloaded\n");
9db5579b NP	544	}
	545
	546	module_init(brd_init);
	547	module_exit(brd_exit);
	548