bounce.c 6.47 KB
Newer Older
1 2 3 4 5
/* bounce buffer handling for block devices
 *
 * - Split from highmem.c
 */

6 7
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

8
#include <linux/mm.h>
9
#include <linux/export.h>
10
#include <linux/swap.h>
11
#include <linux/gfp.h>
12 13 14 15 16 17 18
#include <linux/bio.h>
#include <linux/pagemap.h>
#include <linux/mempool.h>
#include <linux/blkdev.h>
#include <linux/init.h>
#include <linux/hash.h>
#include <linux/highmem.h>
19
#include <linux/bootmem.h>
20
#include <linux/printk.h>
21 22
#include <asm/tlbflush.h>

23 24
#include <trace/events/block.h>

25 26 27 28 29
#define POOL_SIZE	64
#define ISA_POOL_SIZE	16

static mempool_t *page_pool, *isa_page_pool;

30
#if defined(CONFIG_HIGHMEM) || defined(CONFIG_NEED_BOUNCE_POOL)
31 32
static __init int init_emergency_pool(void)
{
33
#if defined(CONFIG_HIGHMEM) && !defined(CONFIG_MEMORY_HOTPLUG)
34
	if (max_pfn <= max_low_pfn)
35
		return 0;
36
#endif
37 38 39

	page_pool = mempool_create_page_pool(POOL_SIZE, 0);
	BUG_ON(!page_pool);
40
	pr_info("pool size: %d pages\n", POOL_SIZE);
41 42 43 44 45

	return 0;
}

__initcall(init_emergency_pool);
46
#endif
47

48
#ifdef CONFIG_HIGHMEM
49 50 51 52 53 54 55 56 57
/*
 * highmem version, map in to vec
 */
static void bounce_copy_vec(struct bio_vec *to, unsigned char *vfrom)
{
	unsigned long flags;
	unsigned char *vto;

	local_irq_save(flags);
58
	vto = kmap_atomic(to->bv_page);
59
	memcpy(vto + to->bv_offset, vfrom, to->bv_len);
60
	kunmap_atomic(vto);
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
	local_irq_restore(flags);
}

#else /* CONFIG_HIGHMEM */

#define bounce_copy_vec(to, vfrom)	\
	memcpy(page_address((to)->bv_page) + (to)->bv_offset, vfrom, (to)->bv_len)

#endif /* CONFIG_HIGHMEM */

/*
 * allocate pages in the DMA region for the ISA pool
 */
static void *mempool_alloc_pages_isa(gfp_t gfp_mask, void *data)
{
	return mempool_alloc_pages(gfp_mask | GFP_DMA, data);
}

/*
 * gets called "every" time someone init's a queue with BLK_BOUNCE_ISA
 * as the max address, so check if the pool has already been created.
 */
int init_emergency_isa_pool(void)
{
	if (isa_page_pool)
		return 0;

	isa_page_pool = mempool_create(ISA_POOL_SIZE, mempool_alloc_pages_isa,
				       mempool_free_pages, (void *) 0);
	BUG_ON(!isa_page_pool);

92
	pr_info("isa pool size: %d pages\n", ISA_POOL_SIZE);
93 94 95 96 97 98 99 100 101 102 103
	return 0;
}

/*
 * Simple bounce buffer support for highmem pages. Depending on the
 * queue gfp mask set, *to may or may not be a highmem page. kmap it
 * always, it will do the Right Thing
 */
static void copy_to_high_bio_irq(struct bio *to, struct bio *from)
{
	unsigned char *vfrom;
104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
	struct bio_vec tovec, *fromvec = from->bi_io_vec;
	struct bvec_iter iter;

	bio_for_each_segment(tovec, to, iter) {
		if (tovec.bv_page != fromvec->bv_page) {
			/*
			 * fromvec->bv_offset and fromvec->bv_len might have
			 * been modified by the block layer, so use the original
			 * copy, bounce_copy_vec already uses tovec->bv_len
			 */
			vfrom = page_address(fromvec->bv_page) +
				tovec.bv_offset;

			bounce_copy_vec(&tovec, vfrom);
			flush_dcache_page(tovec.bv_page);
		}
120

121
		fromvec++;
122 123 124 125 126 127 128 129 130 131 132 133 134 135 136
	}
}

static void bounce_end_io(struct bio *bio, mempool_t *pool, int err)
{
	struct bio *bio_orig = bio->bi_private;
	struct bio_vec *bvec, *org_vec;
	int i;

	if (test_bit(BIO_EOPNOTSUPP, &bio->bi_flags))
		set_bit(BIO_EOPNOTSUPP, &bio_orig->bi_flags);

	/*
	 * free up bounce indirect pages used
	 */
137
	bio_for_each_segment_all(bvec, bio, i) {
138 139 140 141 142 143 144 145
		org_vec = bio_orig->bi_io_vec + i;
		if (bvec->bv_page == org_vec->bv_page)
			continue;

		dec_zone_page_state(bvec->bv_page, NR_BOUNCE);
		mempool_free(bvec->bv_page, pool);
	}

146
	bio_endio(bio_orig, err);
147 148 149
	bio_put(bio);
}

150
static void bounce_end_io_write(struct bio *bio, int err)
151 152 153 154
{
	bounce_end_io(bio, page_pool, err);
}

155
static void bounce_end_io_write_isa(struct bio *bio, int err)
156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
{

	bounce_end_io(bio, isa_page_pool, err);
}

static void __bounce_end_io_read(struct bio *bio, mempool_t *pool, int err)
{
	struct bio *bio_orig = bio->bi_private;

	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
		copy_to_high_bio_irq(bio_orig, bio);

	bounce_end_io(bio, pool, err);
}

171
static void bounce_end_io_read(struct bio *bio, int err)
172 173 174 175
{
	__bounce_end_io_read(bio, page_pool, err);
}

176
static void bounce_end_io_read_isa(struct bio *bio, int err)
177 178 179 180
{
	__bounce_end_io_read(bio, isa_page_pool, err);
}

181 182 183 184 185 186 187 188 189
#ifdef CONFIG_NEED_BOUNCE_POOL
static int must_snapshot_stable_pages(struct request_queue *q, struct bio *bio)
{
	if (bio_data_dir(bio) != WRITE)
		return 0;

	if (!bdi_cap_stable_pages_required(&q->backing_dev_info))
		return 0;

190
	return test_bit(BIO_SNAP_STABLE, &bio->bi_flags);
191 192 193 194 195 196 197 198
}
#else
static int must_snapshot_stable_pages(struct request_queue *q, struct bio *bio)
{
	return 0;
}
#endif /* CONFIG_NEED_BOUNCE_POOL */

199
static void __blk_queue_bounce(struct request_queue *q, struct bio **bio_orig,
200
			       mempool_t *pool, int force)
201
{
202 203
	struct bio *bio;
	int rw = bio_data_dir(*bio_orig);
204 205
	struct bio_vec *to, from;
	struct bvec_iter iter;
206
	unsigned i;
207

208 209
	if (force)
		goto bounce;
210 211
	bio_for_each_segment(from, *bio_orig, iter)
		if (page_to_pfn(from.bv_page) > queue_bounce_pfn(q))
212
			goto bounce;
213

214 215 216
	return;
bounce:
	bio = bio_clone_bioset(*bio_orig, GFP_NOIO, fs_bio_set);
217

218
	bio_for_each_segment_all(to, bio, i) {
219
		struct page *page = to->bv_page;
220

221 222
		if (page_to_pfn(page) <= queue_bounce_pfn(q) && !force)
			continue;
223 224

		inc_zone_page_state(to->bv_page, NR_BOUNCE);
225
		to->bv_page = mempool_alloc(pool, q->bounce_gfp);
226 227 228 229

		if (rw == WRITE) {
			char *vto, *vfrom;

230 231
			flush_dcache_page(page);

232
			vto = page_address(to->bv_page) + to->bv_offset;
233
			vfrom = kmap_atomic(page) + to->bv_offset;
234
			memcpy(vto, vfrom, to->bv_len);
235
			kunmap_atomic(vfrom);
236 237 238
		}
	}

239
	trace_block_bio_bounce(q, *bio_orig);
240

241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
	bio->bi_flags |= (1 << BIO_BOUNCED);

	if (pool == page_pool) {
		bio->bi_end_io = bounce_end_io_write;
		if (rw == READ)
			bio->bi_end_io = bounce_end_io_read;
	} else {
		bio->bi_end_io = bounce_end_io_write_isa;
		if (rw == READ)
			bio->bi_end_io = bounce_end_io_read_isa;
	}

	bio->bi_private = *bio_orig;
	*bio_orig = bio;
}

257
void blk_queue_bounce(struct request_queue *q, struct bio **bio_orig)
258
{
259
	int must_bounce;
260 261
	mempool_t *pool;

262 263 264
	/*
	 * Data-less bio, nothing to bounce
	 */
265
	if (!bio_has_data(*bio_orig))
266 267
		return;

268 269
	must_bounce = must_snapshot_stable_pages(q, *bio_orig);

270 271 272 273 274 275
	/*
	 * for non-isa bounce case, just check if the bounce pfn is equal
	 * to or bigger than the highest pfn in the system -- in that case,
	 * don't waste time iterating over bio segments
	 */
	if (!(q->bounce_gfp & GFP_DMA)) {
276
		if (queue_bounce_pfn(q) >= blk_max_pfn && !must_bounce)
277 278 279 280 281 282 283 284 285 286
			return;
		pool = page_pool;
	} else {
		BUG_ON(!isa_page_pool);
		pool = isa_page_pool;
	}

	/*
	 * slow path
	 */
287
	__blk_queue_bounce(q, bio_orig, pool, must_bounce);
288 289 290
}

EXPORT_SYMBOL(blk_queue_bounce);