dma-mapping.c 15.2 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2
 *  linux/arch/arm/mm/dma-mapping.c
L
Linus Torvalds 已提交
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
 *
 *  Copyright (C) 2000-2004 Russell King
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 *  DMA uncached mapping support.
 */
#include <linux/module.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/errno.h>
#include <linux/list.h>
#include <linux/init.h>
#include <linux/device.h>
#include <linux/dma-mapping.h>

21
#include <asm/memory.h>
22
#include <asm/highmem.h>
L
Linus Torvalds 已提交
23 24
#include <asm/cacheflush.h>
#include <asm/tlbflush.h>
25 26 27 28 29 30
#include <asm/sizes.h>

/* Sanity check size */
#if (CONSISTENT_DMA_SIZE % SZ_2M)
#error "CONSISTENT_DMA_SIZE must be multiple of 2MiB"
#endif
L
Linus Torvalds 已提交
31 32

#define CONSISTENT_END	(0xffe00000)
33 34
#define CONSISTENT_BASE	(CONSISTENT_END - CONSISTENT_DMA_SIZE)

L
Linus Torvalds 已提交
35
#define CONSISTENT_OFFSET(x)	(((unsigned long)(x) - CONSISTENT_BASE) >> PAGE_SHIFT)
36 37 38
#define CONSISTENT_PTE_INDEX(x) (((unsigned long)(x) - CONSISTENT_BASE) >> PGDIR_SHIFT)
#define NUM_CONSISTENT_PTES (CONSISTENT_DMA_SIZE >> PGDIR_SHIFT)

39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61
static u64 get_coherent_dma_mask(struct device *dev)
{
	u64 mask = ISA_DMA_THRESHOLD;

	if (dev) {
		mask = dev->coherent_dma_mask;

		/*
		 * Sanity check the DMA mask - it must be non-zero, and
		 * must be able to be satisfied by a DMA allocation.
		 */
		if (mask == 0) {
			dev_warn(dev, "coherent DMA mask is unset\n");
			return 0;
		}

		if ((~mask) & ISA_DMA_THRESHOLD) {
			dev_warn(dev, "coherent DMA mask %#llx is smaller "
				 "than system GFP_DMA mask %#llx\n",
				 mask, (unsigned long long)ISA_DMA_THRESHOLD);
			return 0;
		}
	}
L
Linus Torvalds 已提交
62

63 64 65
	return mask;
}

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 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
/*
 * Allocate a DMA buffer for 'dev' of size 'size' using the
 * specified gfp mask.  Note that 'size' must be page aligned.
 */
static struct page *__dma_alloc_buffer(struct device *dev, size_t size, gfp_t gfp)
{
	unsigned long order = get_order(size);
	struct page *page, *p, *e;
	void *ptr;
	u64 mask = get_coherent_dma_mask(dev);

#ifdef CONFIG_DMA_API_DEBUG
	u64 limit = (mask + 1) & ~mask;
	if (limit && size >= limit) {
		dev_warn(dev, "coherent allocation too big (requested %#x mask %#llx)\n",
			size, mask);
		return NULL;
	}
#endif

	if (!mask)
		return NULL;

	if (mask < 0xffffffffULL)
		gfp |= GFP_DMA;

	page = alloc_pages(gfp, order);
	if (!page)
		return NULL;

	/*
	 * Now split the huge page and free the excess pages
	 */
	split_page(page, order);
	for (p = page + (size >> PAGE_SHIFT), e = page + (1 << order); p < e; p++)
		__free_page(p);

	/*
	 * Ensure that the allocated pages are zeroed, and that any data
	 * lurking in the kernel direct-mapped region is invalidated.
	 */
	ptr = page_address(page);
	memset(ptr, 0, size);
	dmac_flush_range(ptr, ptr + size);
	outer_flush_range(__pa(ptr), __pa(ptr) + size);

	return page;
}

/*
 * Free a DMA buffer.  'size' must be page aligned.
 */
static void __dma_free_buffer(struct page *page, size_t size)
{
	struct page *e = page + (size >> PAGE_SHIFT);

	while (page < e) {
		__free_page(page);
		page++;
	}
}

128
#ifdef CONFIG_MMU
L
Linus Torvalds 已提交
129
/*
130
 * These are the page tables (2MB each) covering uncached, DMA consistent allocations
L
Linus Torvalds 已提交
131
 */
132
static pte_t *consistent_pte[NUM_CONSISTENT_PTES];
L
Linus Torvalds 已提交
133

134
#include "vmregion.h"
L
Linus Torvalds 已提交
135

136 137
static struct arm_vmregion_head consistent_head = {
	.vm_lock	= __SPIN_LOCK_UNLOCKED(&consistent_head.vm_lock),
L
Linus Torvalds 已提交
138 139 140 141 142 143 144 145 146
	.vm_list	= LIST_HEAD_INIT(consistent_head.vm_list),
	.vm_start	= CONSISTENT_BASE,
	.vm_end		= CONSISTENT_END,
};

#ifdef CONFIG_HUGETLB_PAGE
#error ARM Coherent DMA allocator does not (yet) support huge TLB
#endif

147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
/*
 * Initialise the consistent memory allocation.
 */
static int __init consistent_init(void)
{
	int ret = 0;
	pgd_t *pgd;
	pmd_t *pmd;
	pte_t *pte;
	int i = 0;
	u32 base = CONSISTENT_BASE;

	do {
		pgd = pgd_offset(&init_mm, base);
		pmd = pmd_alloc(&init_mm, pgd, base);
		if (!pmd) {
			printk(KERN_ERR "%s: no pmd tables\n", __func__);
			ret = -ENOMEM;
			break;
		}
		WARN_ON(!pmd_none(*pmd));

		pte = pte_alloc_kernel(pmd, base);
		if (!pte) {
			printk(KERN_ERR "%s: no pte tables\n", __func__);
			ret = -ENOMEM;
			break;
		}

		consistent_pte[i++] = pte;
		base += (1 << PGDIR_SHIFT);
	} while (base < CONSISTENT_END);

	return ret;
}

core_initcall(consistent_init);

L
Linus Torvalds 已提交
185
static void *
186
__dma_alloc_remap(struct page *page, size_t size, gfp_t gfp, pgprot_t prot)
L
Linus Torvalds 已提交
187
{
188
	struct arm_vmregion *c;
L
Linus Torvalds 已提交
189

190 191 192 193 194 195
	if (!consistent_pte[0]) {
		printk(KERN_ERR "%s: not initialised\n", __func__);
		dump_stack();
		return NULL;
	}

L
Linus Torvalds 已提交
196 197 198
	/*
	 * Allocate a virtual address in the consistent mapping region.
	 */
199
	c = arm_vmregion_alloc(&consistent_head, size,
L
Linus Torvalds 已提交
200 201
			    gfp & ~(__GFP_DMA | __GFP_HIGHMEM));
	if (c) {
202 203 204
		pte_t *pte;
		int idx = CONSISTENT_PTE_INDEX(c->vm_start);
		u32 off = CONSISTENT_OFFSET(c->vm_start) & (PTRS_PER_PTE-1);
L
Linus Torvalds 已提交
205

206
		pte = consistent_pte[idx] + off;
L
Linus Torvalds 已提交
207 208 209 210 211
		c->vm_pages = page;

		do {
			BUG_ON(!pte_none(*pte));

R
Russell King 已提交
212
			set_pte_ext(pte, mk_pte(page, prot), 0);
L
Linus Torvalds 已提交
213 214
			page++;
			pte++;
215 216 217 218 219
			off++;
			if (off >= PTRS_PER_PTE) {
				off = 0;
				pte = consistent_pte[++idx];
			}
L
Linus Torvalds 已提交
220 221 222 223 224 225
		} while (size -= PAGE_SIZE);

		return (void *)c->vm_start;
	}
	return NULL;
}
226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264

static void __dma_free_remap(void *cpu_addr, size_t size)
{
	struct arm_vmregion *c;
	unsigned long addr;
	pte_t *ptep;
	int idx;
	u32 off;

	c = arm_vmregion_find_remove(&consistent_head, (unsigned long)cpu_addr);
	if (!c) {
		printk(KERN_ERR "%s: trying to free invalid coherent area: %p\n",
		       __func__, cpu_addr);
		dump_stack();
		return;
	}

	if ((c->vm_end - c->vm_start) != size) {
		printk(KERN_ERR "%s: freeing wrong coherent size (%ld != %d)\n",
		       __func__, c->vm_end - c->vm_start, size);
		dump_stack();
		size = c->vm_end - c->vm_start;
	}

	idx = CONSISTENT_PTE_INDEX(c->vm_start);
	off = CONSISTENT_OFFSET(c->vm_start) & (PTRS_PER_PTE-1);
	ptep = consistent_pte[idx] + off;
	addr = c->vm_start;
	do {
		pte_t pte = ptep_get_and_clear(&init_mm, addr, ptep);

		ptep++;
		addr += PAGE_SIZE;
		off++;
		if (off >= PTRS_PER_PTE) {
			off = 0;
			ptep = consistent_pte[++idx];
		}

265 266 267
		if (pte_none(pte) || !pte_present(pte))
			printk(KERN_CRIT "%s: bad page in kernel page table\n",
			       __func__);
268 269 270 271 272 273 274
	} while (size -= PAGE_SIZE);

	flush_tlb_kernel_range(c->vm_start, c->vm_end);

	arm_vmregion_free(&consistent_head, c);
}

275
#else	/* !CONFIG_MMU */
276

277 278 279 280 281
#define __dma_alloc_remap(page, size, gfp, prot)	page_address(page)
#define __dma_free_remap(addr, size)			do { } while (0)

#endif	/* CONFIG_MMU */

282 283 284 285
static void *
__dma_alloc(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp,
	    pgprot_t prot)
{
286
	struct page *page;
287
	void *addr;
288

289 290
	*handle = ~0;
	size = PAGE_ALIGN(size);
291

292 293 294
	page = __dma_alloc_buffer(dev, size, gfp);
	if (!page)
		return NULL;
295

296 297 298 299
	if (!arch_is_coherent())
		addr = __dma_alloc_remap(page, size, gfp, prot);
	else
		addr = page_address(page);
300

301 302
	if (addr)
		*handle = page_to_dma(dev, page);
303

304 305
	return addr;
}
L
Linus Torvalds 已提交
306 307 308 309 310 311

/*
 * Allocate DMA-coherent memory space and return both the kernel remapped
 * virtual and bus address for that space.
 */
void *
A
Al Viro 已提交
312
dma_alloc_coherent(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp)
L
Linus Torvalds 已提交
313
{
314 315 316 317 318
	void *memory;

	if (dma_alloc_from_coherent(dev, size, handle, &memory))
		return memory;

L
Linus Torvalds 已提交
319
	return __dma_alloc(dev, size, handle, gfp,
320
			   pgprot_dmacoherent(pgprot_kernel));
L
Linus Torvalds 已提交
321 322 323 324 325 326 327 328
}
EXPORT_SYMBOL(dma_alloc_coherent);

/*
 * Allocate a writecombining region, in much the same way as
 * dma_alloc_coherent above.
 */
void *
A
Al Viro 已提交
329
dma_alloc_writecombine(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp)
L
Linus Torvalds 已提交
330 331 332 333 334 335 336 337 338
{
	return __dma_alloc(dev, size, handle, gfp,
			   pgprot_writecombine(pgprot_kernel));
}
EXPORT_SYMBOL(dma_alloc_writecombine);

static int dma_mmap(struct device *dev, struct vm_area_struct *vma,
		    void *cpu_addr, dma_addr_t dma_addr, size_t size)
{
339 340
	int ret = -ENXIO;
#ifdef CONFIG_MMU
341 342
	unsigned long user_size, kern_size;
	struct arm_vmregion *c;
L
Linus Torvalds 已提交
343 344 345

	user_size = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;

346
	c = arm_vmregion_find(&consistent_head, (unsigned long)cpu_addr);
L
Linus Torvalds 已提交
347 348 349 350 351 352 353 354 355 356 357 358 359
	if (c) {
		unsigned long off = vma->vm_pgoff;

		kern_size = (c->vm_end - c->vm_start) >> PAGE_SHIFT;

		if (off < kern_size &&
		    user_size <= (kern_size - off)) {
			ret = remap_pfn_range(vma, vma->vm_start,
					      page_to_pfn(c->vm_pages) + off,
					      user_size << PAGE_SHIFT,
					      vma->vm_page_prot);
		}
	}
360
#endif	/* CONFIG_MMU */
L
Linus Torvalds 已提交
361 362 363 364 365 366 367

	return ret;
}

int dma_mmap_coherent(struct device *dev, struct vm_area_struct *vma,
		      void *cpu_addr, dma_addr_t dma_addr, size_t size)
{
368
	vma->vm_page_prot = pgprot_dmacoherent(vma->vm_page_prot);
L
Linus Torvalds 已提交
369 370 371 372 373 374 375 376 377 378 379 380 381 382
	return dma_mmap(dev, vma, cpu_addr, dma_addr, size);
}
EXPORT_SYMBOL(dma_mmap_coherent);

int dma_mmap_writecombine(struct device *dev, struct vm_area_struct *vma,
			  void *cpu_addr, dma_addr_t dma_addr, size_t size)
{
	vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
	return dma_mmap(dev, vma, cpu_addr, dma_addr, size);
}
EXPORT_SYMBOL(dma_mmap_writecombine);

/*
 * free a page as defined by the above mapping.
383
 * Must not be called with IRQs disabled.
L
Linus Torvalds 已提交
384 385 386
 */
void dma_free_coherent(struct device *dev, size_t size, void *cpu_addr, dma_addr_t handle)
{
387 388
	WARN_ON(irqs_disabled());

389 390 391
	if (dma_release_from_coherent(dev, get_order(size), cpu_addr))
		return;

392 393
	size = PAGE_ALIGN(size);

394 395
	if (!arch_is_coherent())
		__dma_free_remap(cpu_addr, size);
396 397

	__dma_free_buffer(dma_to_page(dev, handle), size);
L
Linus Torvalds 已提交
398 399 400 401 402
}
EXPORT_SYMBOL(dma_free_coherent);

/*
 * Make an area consistent for devices.
403 404 405
 * Note: Drivers should NOT use this function directly, as it will break
 * platforms with CONFIG_DMABOUNCE.
 * Use the driver DMA support - see dma-mapping.h (dma_sync_*)
L
Linus Torvalds 已提交
406
 */
407
static void dma_cache_maint(const void *start, size_t size, int direction)
L
Linus Torvalds 已提交
408
{
409
	void (*outer_op)(unsigned long, unsigned long);
L
Linus Torvalds 已提交
410 411 412

	switch (direction) {
	case DMA_FROM_DEVICE:		/* invalidate only */
413
		outer_op = outer_inv_range;
L
Linus Torvalds 已提交
414 415
		break;
	case DMA_TO_DEVICE:		/* writeback only */
416
		outer_op = outer_clean_range;
L
Linus Torvalds 已提交
417 418
		break;
	case DMA_BIDIRECTIONAL:		/* writeback and invalidate */
419
		outer_op = outer_flush_range;
L
Linus Torvalds 已提交
420 421 422 423
		break;
	default:
		BUG();
	}
424 425

	outer_op(__pa(start), __pa(start) + size);
L
Linus Torvalds 已提交
426
}
427 428 429 430

void ___dma_single_cpu_to_dev(const void *kaddr, size_t size,
	enum dma_data_direction dir)
{
431 432 433
	BUG_ON(!virt_addr_valid(kaddr) || !virt_addr_valid(kaddr + size - 1));

	dmac_map_area(kaddr, size, dir);
434 435 436 437 438 439 440
	dma_cache_maint(kaddr, size, dir);
}
EXPORT_SYMBOL(___dma_single_cpu_to_dev);

void ___dma_single_dev_to_cpu(const void *kaddr, size_t size,
	enum dma_data_direction dir)
{
441 442 443
	BUG_ON(!virt_addr_valid(kaddr) || !virt_addr_valid(kaddr + size - 1));

	dmac_unmap_area(kaddr, size, dir);
444 445
}
EXPORT_SYMBOL(___dma_single_dev_to_cpu);
446

447
static void dma_cache_maint_page(struct page *page, unsigned long offset,
448 449
	size_t size, enum dma_data_direction dir,
	void (*op)(const void *, size_t, int))
450 451 452 453 454 455 456 457 458 459
{
	/*
	 * A single sg entry may refer to multiple physically contiguous
	 * pages.  But we still need to process highmem pages individually.
	 * If highmem is not configured then the bulk of this loop gets
	 * optimized out.
	 */
	size_t left = size;
	do {
		size_t len = left;
460 461 462 463 464 465 466 467 468 469 470 471 472
		void *vaddr;

		if (PageHighMem(page)) {
			if (len + offset > PAGE_SIZE) {
				if (offset >= PAGE_SIZE) {
					page += offset / PAGE_SIZE;
					offset %= PAGE_SIZE;
				}
				len = PAGE_SIZE - offset;
			}
			vaddr = kmap_high_get(page);
			if (vaddr) {
				vaddr += offset;
473
				op(vaddr, len, dir);
474
				kunmap_high(page);
475
			}
476 477
		} else {
			vaddr = page_address(page) + offset;
478
			op(vaddr, len, dir);
479 480 481 482 483 484
		}
		offset = 0;
		page++;
		left -= len;
	} while (left);
}
485 486 487 488

void ___dma_page_cpu_to_dev(struct page *page, unsigned long off,
	size_t size, enum dma_data_direction dir)
{
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505
	unsigned long paddr;
	void (*outer_op)(unsigned long, unsigned long);

	switch (direction) {
	case DMA_FROM_DEVICE:		/* invalidate only */
		outer_op = outer_inv_range;
		break;
	case DMA_TO_DEVICE:		/* writeback only */
		outer_op = outer_clean_range;
		break;
	case DMA_BIDIRECTIONAL:		/* writeback and invalidate */
		outer_op = outer_flush_range;
		break;
	default:
		BUG();
	}

506
	dma_cache_maint_page(page, off, size, dir, dmac_map_area);
507 508 509

	paddr = page_to_phys(page) + off;
	outer_op(paddr, paddr + size);
510 511 512 513 514 515
}
EXPORT_SYMBOL(___dma_page_cpu_to_dev);

void ___dma_page_dev_to_cpu(struct page *page, unsigned long off,
	size_t size, enum dma_data_direction dir)
{
516
	dma_cache_maint_page(page, off, size, dir, dmac_unmap_area);
517 518
}
EXPORT_SYMBOL(___dma_page_dev_to_cpu);
519

520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
/**
 * dma_map_sg - map a set of SG buffers for streaming mode DMA
 * @dev: valid struct device pointer, or NULL for ISA and EISA-like devices
 * @sg: list of buffers
 * @nents: number of buffers to map
 * @dir: DMA transfer direction
 *
 * Map a set of buffers described by scatterlist in streaming mode for DMA.
 * This is the scatter-gather version of the dma_map_single interface.
 * Here the scatter gather list elements are each tagged with the
 * appropriate dma address and length.  They are obtained via
 * sg_dma_{address,length}.
 *
 * Device ownership issues as mentioned for dma_map_single are the same
 * here.
 */
int dma_map_sg(struct device *dev, struct scatterlist *sg, int nents,
		enum dma_data_direction dir)
{
	struct scatterlist *s;
540
	int i, j;
541 542

	for_each_sg(sg, s, nents, i) {
543 544 545 546
		s->dma_address = dma_map_page(dev, sg_page(s), s->offset,
						s->length, dir);
		if (dma_mapping_error(dev, s->dma_address))
			goto bad_mapping;
547 548
	}
	return nents;
549 550 551 552 553

 bad_mapping:
	for_each_sg(sg, s, i, j)
		dma_unmap_page(dev, sg_dma_address(s), sg_dma_len(s), dir);
	return 0;
554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569
}
EXPORT_SYMBOL(dma_map_sg);

/**
 * dma_unmap_sg - unmap a set of SG buffers mapped by dma_map_sg
 * @dev: valid struct device pointer, or NULL for ISA and EISA-like devices
 * @sg: list of buffers
 * @nents: number of buffers to unmap (returned from dma_map_sg)
 * @dir: DMA transfer direction (same as was passed to dma_map_sg)
 *
 * Unmap a set of streaming mode DMA translations.  Again, CPU access
 * rules concerning calls here are the same as for dma_unmap_single().
 */
void dma_unmap_sg(struct device *dev, struct scatterlist *sg, int nents,
		enum dma_data_direction dir)
{
570 571 572 573 574
	struct scatterlist *s;
	int i;

	for_each_sg(sg, s, nents, i)
		dma_unmap_page(dev, sg_dma_address(s), sg_dma_len(s), dir);
575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
}
EXPORT_SYMBOL(dma_unmap_sg);

/**
 * dma_sync_sg_for_cpu
 * @dev: valid struct device pointer, or NULL for ISA and EISA-like devices
 * @sg: list of buffers
 * @nents: number of buffers to map (returned from dma_map_sg)
 * @dir: DMA transfer direction (same as was passed to dma_map_sg)
 */
void dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg,
			int nents, enum dma_data_direction dir)
{
	struct scatterlist *s;
	int i;

	for_each_sg(sg, s, nents, i) {
592 593 594 595 596 597
		if (!dmabounce_sync_for_cpu(dev, sg_dma_address(s), 0,
					    sg_dma_len(s), dir))
			continue;

		__dma_page_dev_to_cpu(sg_page(s), s->offset,
				      s->length, dir);
598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615
	}
}
EXPORT_SYMBOL(dma_sync_sg_for_cpu);

/**
 * dma_sync_sg_for_device
 * @dev: valid struct device pointer, or NULL for ISA and EISA-like devices
 * @sg: list of buffers
 * @nents: number of buffers to map (returned from dma_map_sg)
 * @dir: DMA transfer direction (same as was passed to dma_map_sg)
 */
void dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
			int nents, enum dma_data_direction dir)
{
	struct scatterlist *s;
	int i;

	for_each_sg(sg, s, nents, i) {
616 617 618 619
		if (!dmabounce_sync_for_device(dev, sg_dma_address(s), 0,
					sg_dma_len(s), dir))
			continue;

620 621
		__dma_page_cpu_to_dev(sg_page(s), s->offset,
				      s->length, dir);
622 623 624
	}
}
EXPORT_SYMBOL(dma_sync_sg_for_device);