[deliverable/linux.git] / fs / logfs / dev_bdev.c

/*
 * fs/logfs/dev_bdev.c	- Device access methods for block devices
 *
 * As should be obvious for Linux kernel code, license is GPLv2
 *
 * Copyright (c) 2005-2008 Joern Engel <joern@logfs.org>
 */
#include "logfs.h"
#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/buffer_head.h>
#include <linux/gfp.h>
#include <linux/prefetch.h>

#define PAGE_OFS(ofs) ((ofs) & (PAGE_SIZE-1))

static int sync_request(struct page *page, struct block_device *bdev, int rw)
{
	struct bio bio;
	struct bio_vec bio_vec;

	bio_init(&bio);
	bio.bi_max_vecs = 1;
	bio.bi_io_vec = &bio_vec;
	bio_vec.bv_page = page;
	bio_vec.bv_len = PAGE_SIZE;
	bio_vec.bv_offset = 0;
	bio.bi_vcnt = 1;
	bio.bi_bdev = bdev;
	bio.bi_iter.bi_sector = page->index * (PAGE_SIZE >> 9);
	bio.bi_iter.bi_size = PAGE_SIZE;

	return submit_bio_wait(rw, &bio);
}

static int bdev_readpage(void *_sb, struct page *page)
{
	struct super_block *sb = _sb;
	struct block_device *bdev = logfs_super(sb)->s_bdev;
	int err;

	err = sync_request(page, bdev, READ);
	if (err) {
		ClearPageUptodate(page);
		SetPageError(page);
	} else {
		SetPageUptodate(page);
		ClearPageError(page);
	}
	unlock_page(page);
	return err;
}

static DECLARE_WAIT_QUEUE_HEAD(wq);

static void writeseg_end_io(struct bio *bio)
{
	struct bio_vec *bvec;
	int i;
	struct super_block *sb = bio->bi_private;
	struct logfs_super *super = logfs_super(sb);

	BUG_ON(bio->bi_error); /* FIXME: Retry io or write elsewhere */

	bio_for_each_segment_all(bvec, bio, i) {
		end_page_writeback(bvec->bv_page);
		put_page(bvec->bv_page);
	}
	bio_put(bio);
	if (atomic_dec_and_test(&super->s_pending_writes))
		wake_up(&wq);
}

static int __bdev_writeseg(struct super_block *sb, u64 ofs, pgoff_t index,
		size_t nr_pages)
{
	struct logfs_super *super = logfs_super(sb);
	struct address_space *mapping = super->s_mapping_inode->i_mapping;
	struct bio *bio;
	struct page *page;
	unsigned int max_pages;
	int i;

	max_pages = min_t(size_t, nr_pages, BIO_MAX_PAGES);

	bio = bio_alloc(GFP_NOFS, max_pages);
	BUG_ON(!bio);

	for (i = 0; i < nr_pages; i++) {
		if (i >= max_pages) {
			/* Block layer cannot split bios :( */
			bio->bi_vcnt = i;
			bio->bi_iter.bi_size = i * PAGE_SIZE;
			bio->bi_bdev = super->s_bdev;
			bio->bi_iter.bi_sector = ofs >> 9;
			bio->bi_private = sb;
			bio->bi_end_io = writeseg_end_io;
			atomic_inc(&super->s_pending_writes);
			submit_bio(WRITE, bio);

			ofs += i * PAGE_SIZE;
			index += i;
			nr_pages -= i;
			i = 0;

			bio = bio_alloc(GFP_NOFS, max_pages);
			BUG_ON(!bio);
		}
		page = find_lock_page(mapping, index + i);
		BUG_ON(!page);
		bio->bi_io_vec[i].bv_page = page;
		bio->bi_io_vec[i].bv_len = PAGE_SIZE;
		bio->bi_io_vec[i].bv_offset = 0;

		BUG_ON(PageWriteback(page));
		set_page_writeback(page);
		unlock_page(page);
	}
	bio->bi_vcnt = nr_pages;
	bio->bi_iter.bi_size = nr_pages * PAGE_SIZE;
	bio->bi_bdev = super->s_bdev;
	bio->bi_iter.bi_sector = ofs >> 9;
	bio->bi_private = sb;
	bio->bi_end_io = writeseg_end_io;
	atomic_inc(&super->s_pending_writes);
	submit_bio(WRITE, bio);
	return 0;
}

static void bdev_writeseg(struct super_block *sb, u64 ofs, size_t len)
{
	struct logfs_super *super = logfs_super(sb);
	int head;

	BUG_ON(super->s_flags & LOGFS_SB_FLAG_RO);

	if (len == 0) {
		/* This can happen when the object fit perfectly into a
		 * segment, the segment gets written per sync and subsequently
		 * closed.
		 */
		return;
	}
	head = ofs & (PAGE_SIZE - 1);
	if (head) {
		ofs -= head;
		len += head;
	}
	len = PAGE_ALIGN(len);
	__bdev_writeseg(sb, ofs, ofs >> PAGE_SHIFT, len >> PAGE_SHIFT);
}


static void erase_end_io(struct bio *bio)
{ 
	struct super_block *sb = bio->bi_private; 
	struct logfs_super *super = logfs_super(sb); 

	BUG_ON(bio->bi_error); /* FIXME: Retry io or write elsewhere */ 
	BUG_ON(bio->bi_vcnt == 0); 
	bio_put(bio); 
	if (atomic_dec_and_test(&super->s_pending_writes))
		wake_up(&wq); 
} 

static int do_erase(struct super_block *sb, u64 ofs, pgoff_t index,
		size_t nr_pages)
{
	struct logfs_super *super = logfs_super(sb);
	struct bio *bio;
	unsigned int max_pages;
	int i;

	max_pages = min_t(size_t, nr_pages, BIO_MAX_PAGES);

	bio = bio_alloc(GFP_NOFS, max_pages);
	BUG_ON(!bio);

	for (i = 0; i < nr_pages; i++) {
		if (i >= max_pages) {
			/* Block layer cannot split bios :( */
			bio->bi_vcnt = i;
			bio->bi_iter.bi_size = i * PAGE_SIZE;
			bio->bi_bdev = super->s_bdev;
			bio->bi_iter.bi_sector = ofs >> 9;
			bio->bi_private = sb;
			bio->bi_end_io = erase_end_io;
			atomic_inc(&super->s_pending_writes);
			submit_bio(WRITE, bio);

			ofs += i * PAGE_SIZE;
			index += i;
			nr_pages -= i;
			i = 0;

			bio = bio_alloc(GFP_NOFS, max_pages);
			BUG_ON(!bio);
		}
		bio->bi_io_vec[i].bv_page = super->s_erase_page;
		bio->bi_io_vec[i].bv_len = PAGE_SIZE;
		bio->bi_io_vec[i].bv_offset = 0;
	}
	bio->bi_vcnt = nr_pages;
	bio->bi_iter.bi_size = nr_pages * PAGE_SIZE;
	bio->bi_bdev = super->s_bdev;
	bio->bi_iter.bi_sector = ofs >> 9;
	bio->bi_private = sb;
	bio->bi_end_io = erase_end_io;
	atomic_inc(&super->s_pending_writes);
	submit_bio(WRITE, bio);
	return 0;
}

static int bdev_erase(struct super_block *sb, loff_t to, size_t len,
		int ensure_write)
{
	struct logfs_super *super = logfs_super(sb);

	BUG_ON(to & (PAGE_SIZE - 1));
	BUG_ON(len & (PAGE_SIZE - 1));

	if (super->s_flags & LOGFS_SB_FLAG_RO)
		return -EROFS;

	if (ensure_write) {
		/*
		 * Object store doesn't care whether erases happen or not.
		 * But for the journal they are required.  Otherwise a scan
		 * can find an old commit entry and assume it is the current
		 * one, travelling back in time.
		 */
		do_erase(sb, to, to >> PAGE_SHIFT, len >> PAGE_SHIFT);
	}

	return 0;
}

static void bdev_sync(struct super_block *sb)
{
	struct logfs_super *super = logfs_super(sb);

	wait_event(wq, atomic_read(&super->s_pending_writes) == 0);
}

static struct page *bdev_find_first_sb(struct super_block *sb, u64 *ofs)
{
	struct logfs_super *super = logfs_super(sb);
	struct address_space *mapping = super->s_mapping_inode->i_mapping;
	filler_t *filler = bdev_readpage;

	*ofs = 0;
	return read_cache_page(mapping, 0, filler, sb);
}

static struct page *bdev_find_last_sb(struct super_block *sb, u64 *ofs)
{
	struct logfs_super *super = logfs_super(sb);
	struct address_space *mapping = super->s_mapping_inode->i_mapping;
	filler_t *filler = bdev_readpage;
	u64 pos = (super->s_bdev->bd_inode->i_size & ~0xfffULL) - 0x1000;
	pgoff_t index = pos >> PAGE_SHIFT;

	*ofs = pos;
	return read_cache_page(mapping, index, filler, sb);
}

static int bdev_write_sb(struct super_block *sb, struct page *page)
{
	struct block_device *bdev = logfs_super(sb)->s_bdev;

	/* Nothing special to do for block devices. */
	return sync_request(page, bdev, WRITE);
}

static void bdev_put_device(struct logfs_super *s)
{
	blkdev_put(s->s_bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
}

static int bdev_can_write_buf(struct super_block *sb, u64 ofs)
{
	return 0;
}

static const struct logfs_device_ops bd_devops = {
	.find_first_sb	= bdev_find_first_sb,
	.find_last_sb	= bdev_find_last_sb,
	.write_sb	= bdev_write_sb,
	.readpage	= bdev_readpage,
	.writeseg	= bdev_writeseg,
	.erase		= bdev_erase,
	.can_write_buf	= bdev_can_write_buf,
	.sync		= bdev_sync,
	.put_device	= bdev_put_device,
};

int logfs_get_sb_bdev(struct logfs_super *p, struct file_system_type *type,
		const char *devname)
{
	struct block_device *bdev;

	bdev = blkdev_get_by_path(devname, FMODE_READ|FMODE_WRITE|FMODE_EXCL,
				  type);
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);

	if (MAJOR(bdev->bd_dev) == MTD_BLOCK_MAJOR) {
		int mtdnr = MINOR(bdev->bd_dev);
		blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
		return logfs_get_sb_mtd(p, mtdnr);
	}

	p->s_bdev = bdev;
	p->s_mtd = NULL;
	p->s_devops = &bd_devops;
	return 0;
}
Commit	Line	Data
5db53f3e JE	1	/*
	2	* fs/logfs/dev_bdev.c - Device access methods for block devices
	3	*
	4	* As should be obvious for Linux kernel code, license is GPLv2
	5	*
	6	* Copyright (c) 2005-2008 Joern Engel <joern@logfs.org>
	7	*/
	8	#include "logfs.h"
	9	#include <linux/bio.h>
	10	#include <linux/blkdev.h>
	11	#include <linux/buffer_head.h>
5a0e3ad6	12	#include <linux/gfp.h>
268bb0ce	13	#include <linux/prefetch.h>
5db53f3e JE	14
	15	#define PAGE_OFS(ofs) ((ofs) & (PAGE_SIZE-1))
	16
5db53f3e JE	17	static int sync_request(struct page page, struct block_device bdev, int rw)
	18	{
	19	struct bio bio;
	20	struct bio_vec bio_vec;
5db53f3e JE	21
5db53f3e JE	22	bio_init(&bio);
cd8bfa9c	23	bio.bi_max_vecs = 1;
5db53f3e JE	24	bio.bi_io_vec = &bio_vec;
	25	bio_vec.bv_page = page;
	26	bio_vec.bv_len = PAGE_SIZE;
	27	bio_vec.bv_offset = 0;
	28	bio.bi_vcnt = 1;
5db53f3e	29	bio.bi_bdev = bdev;
4f024f37 KO	30	bio.bi_iter.bi_sector = page->index * (PAGE_SIZE >> 9);
4f024f37 KO	31	bio.bi_iter.bi_size = PAGE_SIZE;
5db53f3e	32
33879d45	33	return submit_bio_wait(rw, &bio);
5db53f3e JE	34	}
	35
	36	static int bdev_readpage(void _sb, struct page page)
	37	{
	38	struct super_block *sb = _sb;
	39	struct block_device *bdev = logfs_super(sb)->s_bdev;
	40	int err;
	41
	42	err = sync_request(page, bdev, READ);
	43	if (err) {
	44	ClearPageUptodate(page);
	45	SetPageError(page);
	46	} else {
	47	SetPageUptodate(page);
	48	ClearPageError(page);
	49	}
	50	unlock_page(page);
	51	return err;
	52	}
	53
	54	static DECLARE_WAIT_QUEUE_HEAD(wq);
	55
4246a0b6	56	static void writeseg_end_io(struct bio *bio)
5db53f3e	57	{
2c30c71b KO	58	struct bio_vec *bvec;
2c30c71b KO	59	int i;
5db53f3e JE	60	struct super_block *sb = bio->bi_private;
5db53f3e JE	61	struct logfs_super *super = logfs_super(sb);
5db53f3e	62
4246a0b6	63	BUG_ON(bio->bi_error); /* FIXME: Retry io or write elsewhere */
5db53f3e	64
2c30c71b KO	65	bio_for_each_segment_all(bvec, bio, i) {
2c30c71b KO	66	end_page_writeback(bvec->bv_page);
09cbfeaf	67	put_page(bvec->bv_page);
2c30c71b	68	}
5db53f3e JE	69	bio_put(bio);
	70	if (atomic_dec_and_test(&super->s_pending_writes))
	71	wake_up(&wq);
	72	}
	73
	74	static int __bdev_writeseg(struct super_block *sb, u64 ofs, pgoff_t index,
	75	size_t nr_pages)
	76	{
	77	struct logfs_super *super = logfs_super(sb);
	78	struct address_space *mapping = super->s_mapping_inode->i_mapping;
	79	struct bio *bio;
	80	struct page *page;
9f0bbd8c	81	unsigned int max_pages;
5db53f3e JE	82	int i;
5db53f3e JE	83
ce4f2fd7	84	max_pages = min_t(size_t, nr_pages, BIO_MAX_PAGES);
9f0bbd8c	85
5db53f3e	86	bio = bio_alloc(GFP_NOFS, max_pages);
59fe27c0	87	BUG_ON(!bio);
5db53f3e JE	88
	89	for (i = 0; i < nr_pages; i++) {
	90	if (i >= max_pages) {
	91	/* Block layer cannot split bios :( */
	92	bio->bi_vcnt = i;
4f024f37	93	bio->bi_iter.bi_size = i * PAGE_SIZE;
5db53f3e	94	bio->bi_bdev = super->s_bdev;
4f024f37	95	bio->bi_iter.bi_sector = ofs >> 9;
5db53f3e JE	96	bio->bi_private = sb;
	97	bio->bi_end_io = writeseg_end_io;
	98	atomic_inc(&super->s_pending_writes);
	99	submit_bio(WRITE, bio);
	100
	101	ofs += i * PAGE_SIZE;
	102	index += i;
	103	nr_pages -= i;
	104	i = 0;
	105
	106	bio = bio_alloc(GFP_NOFS, max_pages);
	107	BUG_ON(!bio);
	108	}
	109	page = find_lock_page(mapping, index + i);
	110	BUG_ON(!page);
	111	bio->bi_io_vec[i].bv_page = page;
	112	bio->bi_io_vec[i].bv_len = PAGE_SIZE;
	113	bio->bi_io_vec[i].bv_offset = 0;
	114
	115	BUG_ON(PageWriteback(page));
	116	set_page_writeback(page);
	117	unlock_page(page);
	118	}
	119	bio->bi_vcnt = nr_pages;
4f024f37	120	bio->bi_iter.bi_size = nr_pages * PAGE_SIZE;
5db53f3e	121	bio->bi_bdev = super->s_bdev;
4f024f37	122	bio->bi_iter.bi_sector = ofs >> 9;
5db53f3e JE	123	bio->bi_private = sb;
	124	bio->bi_end_io = writeseg_end_io;
	125	atomic_inc(&super->s_pending_writes);
	126	submit_bio(WRITE, bio);
	127	return 0;
	128	}
	129
	130	static void bdev_writeseg(struct super_block *sb, u64 ofs, size_t len)
	131	{
	132	struct logfs_super *super = logfs_super(sb);
	133	int head;
	134
	135	BUG_ON(super->s_flags & LOGFS_SB_FLAG_RO);
	136
	137	if (len == 0) {
	138	/* This can happen when the object fit perfectly into a
	139	* segment, the segment gets written per sync and subsequently
	140	* closed.
	141	*/
	142	return;
	143	}
	144	head = ofs & (PAGE_SIZE - 1);
	145	if (head) {
	146	ofs -= head;
	147	len += head;
	148	}
	149	len = PAGE_ALIGN(len);
	150	__bdev_writeseg(sb, ofs, ofs >> PAGE_SHIFT, len >> PAGE_SHIFT);
5db53f3e JE	151	}
5db53f3e JE	152
9421502b	153
4246a0b6	154	static void erase_end_io(struct bio *bio)
9421502b	155	{
9421502b JE	156	struct super_block *sb = bio->bi_private;
	157	struct logfs_super *super = logfs_super(sb);
	158
4246a0b6	159	BUG_ON(bio->bi_error); /* FIXME: Retry io or write elsewhere */
9421502b JE	160	BUG_ON(bio->bi_vcnt == 0);
	161	bio_put(bio);
	162	if (atomic_dec_and_test(&super->s_pending_writes))
	163	wake_up(&wq);
	164	}
	165
	166	static int do_erase(struct super_block *sb, u64 ofs, pgoff_t index,
	167	size_t nr_pages)
	168	{
	169	struct logfs_super *super = logfs_super(sb);
	170	struct bio *bio;
9f0bbd8c	171	unsigned int max_pages;
9421502b JE	172	int i;
9421502b JE	173
ce4f2fd7	174	max_pages = min_t(size_t, nr_pages, BIO_MAX_PAGES);
9f0bbd8c	175
9421502b	176	bio = bio_alloc(GFP_NOFS, max_pages);
59fe27c0	177	BUG_ON(!bio);
9421502b JE	178
	179	for (i = 0; i < nr_pages; i++) {
	180	if (i >= max_pages) {
	181	/* Block layer cannot split bios :( */
	182	bio->bi_vcnt = i;
4f024f37	183	bio->bi_iter.bi_size = i * PAGE_SIZE;
9421502b	184	bio->bi_bdev = super->s_bdev;
4f024f37	185	bio->bi_iter.bi_sector = ofs >> 9;
9421502b JE	186	bio->bi_private = sb;
	187	bio->bi_end_io = erase_end_io;
	188	atomic_inc(&super->s_pending_writes);
	189	submit_bio(WRITE, bio);
	190
	191	ofs += i * PAGE_SIZE;
	192	index += i;
	193	nr_pages -= i;
	194	i = 0;
	195
	196	bio = bio_alloc(GFP_NOFS, max_pages);
	197	BUG_ON(!bio);
	198	}
	199	bio->bi_io_vec[i].bv_page = super->s_erase_page;
	200	bio->bi_io_vec[i].bv_len = PAGE_SIZE;
	201	bio->bi_io_vec[i].bv_offset = 0;
	202	}
	203	bio->bi_vcnt = nr_pages;
4f024f37	204	bio->bi_iter.bi_size = nr_pages * PAGE_SIZE;
9421502b	205	bio->bi_bdev = super->s_bdev;
4f024f37	206	bio->bi_iter.bi_sector = ofs >> 9;
9421502b JE	207	bio->bi_private = sb;
	208	bio->bi_end_io = erase_end_io;
	209	atomic_inc(&super->s_pending_writes);
	210	submit_bio(WRITE, bio);
	211	return 0;
	212	}
	213
	214	static int bdev_erase(struct super_block *sb, loff_t to, size_t len,
	215	int ensure_write)
5db53f3e JE	216	{
5db53f3e JE	217	struct logfs_super *super = logfs_super(sb);
5db53f3e JE	218
	219	BUG_ON(to & (PAGE_SIZE - 1));
	220	BUG_ON(len & (PAGE_SIZE - 1));
	221
9421502b	222	if (super->s_flags & LOGFS_SB_FLAG_RO)
5db53f3e JE	223	return -EROFS;
5db53f3e JE	224
9421502b JE	225	if (ensure_write) {
	226	/*
	227	* Object store doesn't care whether erases happen or not.
	228	* But for the journal they are required. Otherwise a scan
	229	* can find an old commit entry and assume it is the current
	230	* one, travelling back in time.
	231	*/
	232	do_erase(sb, to, to >> PAGE_SHIFT, len >> PAGE_SHIFT);
5db53f3e	233	}
9421502b	234
5db53f3e JE	235	return 0;
	236	}
	237
	238	static void bdev_sync(struct super_block *sb)
	239	{
	240	struct logfs_super *super = logfs_super(sb);
	241
	242	wait_event(wq, atomic_read(&super->s_pending_writes) == 0);
	243	}
	244
	245	static struct page bdev_find_first_sb(struct super_block sb, u64 *ofs)
	246	{
	247	struct logfs_super *super = logfs_super(sb);
	248	struct address_space *mapping = super->s_mapping_inode->i_mapping;
	249	filler_t *filler = bdev_readpage;
	250
	251	*ofs = 0;
	252	return read_cache_page(mapping, 0, filler, sb);
	253	}
	254
	255	static struct page bdev_find_last_sb(struct super_block sb, u64 *ofs)
	256	{
	257	struct logfs_super *super = logfs_super(sb);
	258	struct address_space *mapping = super->s_mapping_inode->i_mapping;
	259	filler_t *filler = bdev_readpage;
	260	u64 pos = (super->s_bdev->bd_inode->i_size & ~0xfffULL) - 0x1000;
	261	pgoff_t index = pos >> PAGE_SHIFT;
	262
	263	*ofs = pos;
	264	return read_cache_page(mapping, index, filler, sb);
	265	}
	266
	267	static int bdev_write_sb(struct super_block sb, struct page page)
	268	{
	269	struct block_device *bdev = logfs_super(sb)->s_bdev;
	270
	271	/* Nothing special to do for block devices. */
	272	return sync_request(page, bdev, WRITE);
	273	}
	274
e5a0726a	275	static void bdev_put_device(struct logfs_super *s)
5db53f3e	276	{
e525fd89	277	blkdev_put(s->s_bdev, FMODE_READ\|FMODE_WRITE\|FMODE_EXCL);
5db53f3e JE	278	}
5db53f3e JE	279
6f485b41 JE	280	static int bdev_can_write_buf(struct super_block *sb, u64 ofs)
	281	{
	282	return 0;
	283	}
	284
5db53f3e JE	285	static const struct logfs_device_ops bd_devops = {
	286	.find_first_sb = bdev_find_first_sb,
	287	.find_last_sb = bdev_find_last_sb,
	288	.write_sb = bdev_write_sb,
	289	.readpage = bdev_readpage,
	290	.writeseg = bdev_writeseg,
	291	.erase = bdev_erase,
6f485b41	292	.can_write_buf = bdev_can_write_buf,
5db53f3e JE	293	.sync = bdev_sync,
	294	.put_device = bdev_put_device,
	295	};
	296
7d945a3a AV	297	int logfs_get_sb_bdev(struct logfs_super p, struct file_system_type type,
7d945a3a AV	298	const char *devname)
5db53f3e JE	299	{
	300	struct block_device *bdev;
	301
d4d77629 TH	302	bdev = blkdev_get_by_path(devname, FMODE_READ\|FMODE_WRITE\|FMODE_EXCL,
d4d77629 TH	303	type);
7d945a3a	304	if (IS_ERR(bdev))
5db53f3e JE	305	return PTR_ERR(bdev);
	306
	307	if (MAJOR(bdev->bd_dev) == MTD_BLOCK_MAJOR) {
	308	int mtdnr = MINOR(bdev->bd_dev);
e525fd89	309	blkdev_put(bdev, FMODE_READ\|FMODE_WRITE\|FMODE_EXCL);
7d945a3a	310	return logfs_get_sb_mtd(p, mtdnr);
5db53f3e JE	311	}
5db53f3e JE	312
0d85c799 AV	313	p->s_bdev = bdev;
	314	p->s_mtd = NULL;
	315	p->s_devops = &bd_devops;
7d945a3a	316	return 0;
5db53f3e	317	}