bounce.c 6.5 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2 3 4 5 6
/* bounce buffer handling for block devices
 *
 * - Split from highmem.c
 */

7 8
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

9
#include <linux/mm.h>
10
#include <linux/export.h>
11
#include <linux/swap.h>
12
#include <linux/gfp.h>
13 14 15 16
#include <linux/bio.h>
#include <linux/pagemap.h>
#include <linux/mempool.h>
#include <linux/blkdev.h>
17
#include <linux/backing-dev.h>
18 19 20
#include <linux/init.h>
#include <linux/hash.h>
#include <linux/highmem.h>
21
#include <linux/bootmem.h>
22
#include <linux/printk.h>
23 24
#include <asm/tlbflush.h>

25
#include <trace/events/block.h>
26
#include "blk.h"
27

28 29 30
#define POOL_SIZE	64
#define ISA_POOL_SIZE	16

31
static struct bio_set *bounce_bio_set, *bounce_bio_split;
32 33
static mempool_t *page_pool, *isa_page_pool;

34
#if defined(CONFIG_HIGHMEM) || defined(CONFIG_NEED_BOUNCE_POOL)
35 36
static __init int init_emergency_pool(void)
{
37
#if defined(CONFIG_HIGHMEM) && !defined(CONFIG_MEMORY_HOTPLUG)
38
	if (max_pfn <= max_low_pfn)
39
		return 0;
40
#endif
41 42 43

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

46 47 48 49 50 51 52 53
	bounce_bio_set = bioset_create(BIO_POOL_SIZE, 0, BIOSET_NEED_BVECS);
	BUG_ON(!bounce_bio_set);
	if (bioset_integrity_create(bounce_bio_set, BIO_POOL_SIZE))
		BUG_ON(1);

	bounce_bio_split = bioset_create(BIO_POOL_SIZE, 0, 0);
	BUG_ON(!bounce_bio_split);

54 55 56 57
	return 0;
}

__initcall(init_emergency_pool);
58
#endif
59

60
#ifdef CONFIG_HIGHMEM
61 62 63 64 65 66 67 68 69
/*
 * 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);
70
	vto = kmap_atomic(to->bv_page);
71
	memcpy(vto + to->bv_offset, vfrom, to->bv_len);
72
	kunmap_atomic(vto);
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103
	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);

104
	pr_info("isa pool size: %d pages\n", ISA_POOL_SIZE);
105 106 107 108 109 110 111 112 113 114 115
	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;
116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
	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);
		}
132

133
		fromvec++;
134 135 136
	}
}

137
static void bounce_end_io(struct bio *bio, mempool_t *pool)
138 139 140 141
{
	struct bio *bio_orig = bio->bi_private;
	struct bio_vec *bvec, *org_vec;
	int i;
M
Ming Lei 已提交
142
	int start = bio_orig->bi_iter.bi_idx;
143 144 145 146

	/*
	 * free up bounce indirect pages used
	 */
147
	bio_for_each_segment_all(bvec, bio, i) {
M
Ming Lei 已提交
148 149
		org_vec = bio_orig->bi_io_vec + i + start;

150 151 152 153 154 155 156
		if (bvec->bv_page == org_vec->bv_page)
			continue;

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

157
	bio_orig->bi_status = bio->bi_status;
158
	bio_endio(bio_orig);
159 160 161
	bio_put(bio);
}

162
static void bounce_end_io_write(struct bio *bio)
163
{
164
	bounce_end_io(bio, page_pool);
165 166
}

167
static void bounce_end_io_write_isa(struct bio *bio)
168 169
{

170
	bounce_end_io(bio, isa_page_pool);
171 172
}

173
static void __bounce_end_io_read(struct bio *bio, mempool_t *pool)
174 175 176
{
	struct bio *bio_orig = bio->bi_private;

177
	if (!bio->bi_status)
178 179
		copy_to_high_bio_irq(bio_orig, bio);

180
	bounce_end_io(bio, pool);
181 182
}

183
static void bounce_end_io_read(struct bio *bio)
184
{
185
	__bounce_end_io_read(bio, page_pool);
186 187
}

188
static void bounce_end_io_read_isa(struct bio *bio)
189
{
190
	__bounce_end_io_read(bio, isa_page_pool);
191 192
}

193
static void __blk_queue_bounce(struct request_queue *q, struct bio **bio_orig,
194
			       mempool_t *pool)
195
{
196 197
	struct bio *bio;
	int rw = bio_data_dir(*bio_orig);
198 199
	struct bio_vec *to, from;
	struct bvec_iter iter;
200 201 202
	unsigned i = 0;
	bool bounce = false;
	int sectors = 0;
203

204 205 206
	bio_for_each_segment(from, *bio_orig, iter) {
		if (i++ < BIO_MAX_PAGES)
			sectors += from.bv_len >> 9;
207
		if (page_to_pfn(from.bv_page) > q->limits.bounce_pfn)
208 209 210 211 212 213 214 215 216 217 218 219
			bounce = true;
	}
	if (!bounce)
		return;

	if (sectors < bio_sectors(*bio_orig)) {
		bio = bio_split(*bio_orig, sectors, GFP_NOIO, bounce_bio_split);
		bio_chain(bio, *bio_orig);
		generic_make_request(*bio_orig);
		*bio_orig = bio;
	}
	bio = bio_clone_bioset(*bio_orig, GFP_NOIO, bounce_bio_set);
220

221
	bio_for_each_segment_all(to, bio, i) {
222
		struct page *page = to->bv_page;
223

224
		if (page_to_pfn(page) <= q->limits.bounce_pfn)
225
			continue;
226

227
		to->bv_page = mempool_alloc(pool, q->bounce_gfp);
228
		inc_zone_page_state(to->bv_page, NR_BOUNCE);
229 230 231 232

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

233 234
			flush_dcache_page(page);

235
			vto = page_address(to->bv_page) + to->bv_offset;
236
			vfrom = kmap_atomic(page) + to->bv_offset;
237
			memcpy(vto, vfrom, to->bv_len);
238
			kunmap_atomic(vfrom);
239 240 241
		}
	}

242
	trace_block_bio_bounce(q, *bio_orig);
243

244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
	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;
}

260
void blk_queue_bounce(struct request_queue *q, struct bio **bio_orig)
261 262 263
{
	mempool_t *pool;

264 265 266
	/*
	 * Data-less bio, nothing to bounce
	 */
267
	if (!bio_has_data(*bio_orig))
268 269
		return;

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 (q->limits.bounce_pfn >= blk_max_pfn)
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);
288
}