[deliverable/linux.git] / arch / x86 / kernel / pci-dma.c

#include <linux/dma-mapping.h>
#include <linux/dma-debug.h>
#include <linux/dmar.h>
#include <linux/export.h>
#include <linux/bootmem.h>
#include <linux/gfp.h>
#include <linux/pci.h>
#include <linux/kmemleak.h>

#include <asm/proto.h>
#include <asm/dma.h>
#include <asm/iommu.h>
#include <asm/gart.h>
#include <asm/calgary.h>
#include <asm/x86_init.h>
#include <asm/iommu_table.h>

static int forbid_dac __read_mostly;

struct dma_map_ops *dma_ops = &nommu_dma_ops;
EXPORT_SYMBOL(dma_ops);

static int iommu_sac_force __read_mostly;

#ifdef CONFIG_IOMMU_DEBUG
int panic_on_overflow __read_mostly = 1;
int force_iommu __read_mostly = 1;
#else
int panic_on_overflow __read_mostly = 0;
int force_iommu __read_mostly = 0;
#endif

int iommu_merge __read_mostly = 0;

int no_iommu __read_mostly;
/* Set this to 1 if there is a HW IOMMU in the system */
int iommu_detected __read_mostly = 0;

/*
 * This variable becomes 1 if iommu=pt is passed on the kernel command line.
 * If this variable is 1, IOMMU implementations do no DMA translation for
 * devices and allow every device to access to whole physical memory. This is
 * useful if a user wants to use an IOMMU only for KVM device assignment to
 * guests and not for driver dma translation.
 */
int iommu_pass_through __read_mostly;

extern struct iommu_table_entry __iommu_table[], __iommu_table_end[];

/* Dummy device used for NULL arguments (normally ISA). */
struct device x86_dma_fallback_dev = {
	.init_name = "fallback device",
	.coherent_dma_mask = ISA_DMA_BIT_MASK,
	.dma_mask = &x86_dma_fallback_dev.coherent_dma_mask,
};
EXPORT_SYMBOL(x86_dma_fallback_dev);

/* Number of entries preallocated for DMA-API debugging */
#define PREALLOC_DMA_DEBUG_ENTRIES       65536

int dma_set_mask(struct device *dev, u64 mask)
{
	if (!dev->dma_mask || !dma_supported(dev, mask))
		return -EIO;

	*dev->dma_mask = mask;

	return 0;
}
EXPORT_SYMBOL(dma_set_mask);

void __init pci_iommu_alloc(void)
{
	struct iommu_table_entry *p;

	sort_iommu_table(__iommu_table, __iommu_table_end);
	check_iommu_entries(__iommu_table, __iommu_table_end);

	for (p = __iommu_table; p < __iommu_table_end; p++) {
		if (p && p->detect && p->detect() > 0) {
			p->flags |= IOMMU_DETECTED;
			if (p->early_init)
				p->early_init();
			if (p->flags & IOMMU_FINISH_IF_DETECTED)
				break;
		}
	}
}
void *dma_generic_alloc_coherent(struct device *dev, size_t size,
				 dma_addr_t *dma_addr, gfp_t flag,
				 struct dma_attrs *attrs)
{
	unsigned long dma_mask;
	struct page *page;
	unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
	dma_addr_t addr;

	dma_mask = dma_alloc_coherent_mask(dev, flag);

	flag &= ~__GFP_ZERO;
again:
	page = NULL;
	/* CMA can be used only in the context which permits sleeping */
	if (flag & __GFP_WAIT) {
		page = dma_alloc_from_contiguous(dev, count, get_order(size));
		if (page && page_to_phys(page) + size > dma_mask) {
			dma_release_from_contiguous(dev, page, count);
			page = NULL;
		}
	}
	/* fallback */
	if (!page)
		page = alloc_pages_node(dev_to_node(dev), flag, get_order(size));
	if (!page)
		return NULL;

	addr = page_to_phys(page);
	if (addr + size > dma_mask) {
		__free_pages(page, get_order(size));

		if (dma_mask < DMA_BIT_MASK(32) && !(flag & GFP_DMA)) {
			flag = (flag & ~GFP_DMA32) | GFP_DMA;
			goto again;
		}

		return NULL;
	}
	memset(page_address(page), 0, size);
	*dma_addr = addr;
	return page_address(page);
}

void dma_generic_free_coherent(struct device *dev, size_t size, void *vaddr,
			       dma_addr_t dma_addr, struct dma_attrs *attrs)
{
	unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
	struct page *page = virt_to_page(vaddr);

	if (!dma_release_from_contiguous(dev, page, count))
		free_pages((unsigned long)vaddr, get_order(size));
}

/*
 * See <Documentation/x86/x86_64/boot-options.txt> for the iommu kernel
 * parameter documentation.
 */
static __init int iommu_setup(char *p)
{
	iommu_merge = 1;

	if (!p)
		return -EINVAL;

	while (*p) {
		if (!strncmp(p, "off", 3))
			no_iommu = 1;
		/* gart_parse_options has more force support */
		if (!strncmp(p, "force", 5))
			force_iommu = 1;
		if (!strncmp(p, "noforce", 7)) {
			iommu_merge = 0;
			force_iommu = 0;
		}

		if (!strncmp(p, "biomerge", 8)) {
			iommu_merge = 1;
			force_iommu = 1;
		}
		if (!strncmp(p, "panic", 5))
			panic_on_overflow = 1;
		if (!strncmp(p, "nopanic", 7))
			panic_on_overflow = 0;
		if (!strncmp(p, "merge", 5)) {
			iommu_merge = 1;
			force_iommu = 1;
		}
		if (!strncmp(p, "nomerge", 7))
			iommu_merge = 0;
		if (!strncmp(p, "forcesac", 8))
			iommu_sac_force = 1;
		if (!strncmp(p, "allowdac", 8))
			forbid_dac = 0;
		if (!strncmp(p, "nodac", 5))
			forbid_dac = 1;
		if (!strncmp(p, "usedac", 6)) {
			forbid_dac = -1;
			return 1;
		}
#ifdef CONFIG_SWIOTLB
		if (!strncmp(p, "soft", 4))
			swiotlb = 1;
#endif
		if (!strncmp(p, "pt", 2))
			iommu_pass_through = 1;

		gart_parse_options(p);

#ifdef CONFIG_CALGARY_IOMMU
		if (!strncmp(p, "calgary", 7))
			use_calgary = 1;
#endif /* CONFIG_CALGARY_IOMMU */

		p += strcspn(p, ",");
		if (*p == ',')
			++p;
	}
	return 0;
}
early_param("iommu", iommu_setup);

int dma_supported(struct device *dev, u64 mask)
{
	struct dma_map_ops *ops = get_dma_ops(dev);

#ifdef CONFIG_PCI
	if (mask > 0xffffffff && forbid_dac > 0) {
		dev_info(dev, "PCI: Disallowing DAC for device\n");
		return 0;
	}
#endif

	if (ops->dma_supported)
		return ops->dma_supported(dev, mask);

	/* Copied from i386. Doesn't make much sense, because it will
	   only work for pci_alloc_coherent.
	   The caller just has to use GFP_DMA in this case. */
	if (mask < DMA_BIT_MASK(24))
		return 0;

	/* Tell the device to use SAC when IOMMU force is on.  This
	   allows the driver to use cheaper accesses in some cases.

	   Problem with this is that if we overflow the IOMMU area and
	   return DAC as fallback address the device may not handle it
	   correctly.

	   As a special case some controllers have a 39bit address
	   mode that is as efficient as 32bit (aic79xx). Don't force
	   SAC for these.  Assume all masks <= 40 bits are of this
	   type. Normally this doesn't make any difference, but gives
	   more gentle handling of IOMMU overflow. */
	if (iommu_sac_force && (mask >= DMA_BIT_MASK(40))) {
		dev_info(dev, "Force SAC with mask %Lx\n", mask);
		return 0;
	}

	return 1;
}
EXPORT_SYMBOL(dma_supported);

static int __init pci_iommu_init(void)
{
	struct iommu_table_entry *p;
	dma_debug_init(PREALLOC_DMA_DEBUG_ENTRIES);

#ifdef CONFIG_PCI
	dma_debug_add_bus(&pci_bus_type);
#endif
	x86_init.iommu.iommu_init();

	for (p = __iommu_table; p < __iommu_table_end; p++) {
		if (p && (p->flags & IOMMU_DETECTED) && p->late_init)
			p->late_init();
	}

	return 0;
}
/* Must execute after PCI subsystem */
rootfs_initcall(pci_iommu_init);

#ifdef CONFIG_PCI
/* Many VIA bridges seem to corrupt data for DAC. Disable it here */

static void via_no_dac(struct pci_dev *dev)
{
	if (forbid_dac == 0) {
		dev_info(&dev->dev, "disabling DAC on VIA PCI bridge\n");
		forbid_dac = 1;
	}
}
DECLARE_PCI_FIXUP_CLASS_FINAL(PCI_VENDOR_ID_VIA, PCI_ANY_ID,
				PCI_CLASS_BRIDGE_PCI, 8, via_no_dac);
#endif
Commit	Line	Data
459121c9	1	#include <linux/dma-mapping.h>
2118d0c5	2	#include <linux/dma-debug.h>
cb5867a5	3	#include <linux/dmar.h>
69c60c88	4	#include <linux/export.h>
116890d5	5	#include <linux/bootmem.h>
5a0e3ad6	6	#include <linux/gfp.h>
bca5c096	7	#include <linux/pci.h>
acde31dc	8	#include <linux/kmemleak.h>
cb5867a5	9
116890d5 GC	10	#include <asm/proto.h>
116890d5 GC	11	#include <asm/dma.h>
46a7fa27	12	#include <asm/iommu.h>
1d9b16d1	13	#include <asm/gart.h>
cb5867a5	14	#include <asm/calgary.h>
b4941a9a	15	#include <asm/x86_init.h>
ee1f284f	16	#include <asm/iommu_table.h>
459121c9	17
3b15e581 FY	18	static int forbid_dac __read_mostly;
3b15e581 FY	19
a3b28ee1	20	struct dma_map_ops *dma_ops = &nommu_dma_ops;
85c246ee GC	21	EXPORT_SYMBOL(dma_ops);
85c246ee GC	22
b4cdc430	23	static int iommu_sac_force __read_mostly;
8e0c3797	24
f9c258de GC	25	#ifdef CONFIG_IOMMU_DEBUG
	26	int panic_on_overflow __read_mostly = 1;
	27	int force_iommu __read_mostly = 1;
	28	#else
	29	int panic_on_overflow __read_mostly = 0;
	30	int force_iommu __read_mostly = 0;
	31	#endif
	32
fae9a0d8 GC	33	int iommu_merge __read_mostly = 0;
	34
	35	int no_iommu __read_mostly;
	36	/* Set this to 1 if there is a HW IOMMU in the system */
	37	int iommu_detected __read_mostly = 0;
	38
ac0101d3 JR	39	/*
ac0101d3 JR	40	* This variable becomes 1 if iommu=pt is passed on the kernel command line.
e3be785f	41	* If this variable is 1, IOMMU implementations do no DMA translation for
ac0101d3	42	* devices and allow every device to access to whole physical memory. This is
fb637f3c	43	* useful if a user wants to use an IOMMU only for KVM device assignment to
ac0101d3 JR	44	* guests and not for driver dma translation.
	45	*/
	46	int iommu_pass_through __read_mostly;
aed5d5f4	47
ee1f284f KRW	48	extern struct iommu_table_entry __iommu_table[], __iommu_table_end[];
ee1f284f KRW	49
eb647138	50	/* Dummy device used for NULL arguments (normally ISA). */
6c505ce3	51	struct device x86_dma_fallback_dev = {
1a927133	52	.init_name = "fallback device",
eb647138	53	.coherent_dma_mask = ISA_DMA_BIT_MASK,
6c505ce3	54	.dma_mask = &x86_dma_fallback_dev.coherent_dma_mask,
098cb7f2	55	};
6c505ce3	56	EXPORT_SYMBOL(x86_dma_fallback_dev);
098cb7f2	57
2118d0c5	58	/* Number of entries preallocated for DMA-API debugging */
73b664ce	59	#define PREALLOC_DMA_DEBUG_ENTRIES 65536
2118d0c5	60
459121c9 GC	61	int dma_set_mask(struct device *dev, u64 mask)
	62	{
	63	if (!dev->dma_mask \|\| !dma_supported(dev, mask))
	64	return -EIO;
	65
	66	*dev->dma_mask = mask;
	67
	68	return 0;
	69	}
	70	EXPORT_SYMBOL(dma_set_mask);
	71
116890d5 GC	72	void __init pci_iommu_alloc(void)
116890d5 GC	73	{
ee1f284f KRW	74	struct iommu_table_entry *p;
ee1f284f KRW	75
ee1f284f KRW	76	sort_iommu_table(__iommu_table, __iommu_table_end);
ee1f284f KRW	77	check_iommu_entries(__iommu_table, __iommu_table_end);
116890d5	78
ee1f284f KRW	79	for (p = __iommu_table; p < __iommu_table_end; p++) {
	80	if (p && p->detect && p->detect() > 0) {
	81	p->flags \|= IOMMU_DETECTED;
	82	if (p->early_init)
	83	p->early_init();
	84	if (p->flags & IOMMU_FINISH_IF_DETECTED)
	85	break;
	86	}
	87	}
116890d5	88	}
9f6ac577	89	void dma_generic_alloc_coherent(struct device dev, size_t size,
baa676fc AP	90	dma_addr_t *dma_addr, gfp_t flag,
baa676fc AP	91	struct dma_attrs *attrs)
9f6ac577 FT	92	{
9f6ac577 FT	93	unsigned long dma_mask;
c080e26e	94	struct page *page;
0a2b9a6e	95	unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
9f6ac577 FT	96	dma_addr_t addr;
	97
	98	dma_mask = dma_alloc_coherent_mask(dev, flag);
	99
d92ef66c	100	flag &= ~__GFP_ZERO;
9f6ac577	101	again:
c080e26e	102	page = NULL;
c091c71a	103	/* CMA can be used only in the context which permits sleeping */
38f7ea5a	104	if (flag & __GFP_WAIT) {
0a2b9a6e	105	page = dma_alloc_from_contiguous(dev, count, get_order(size));
38f7ea5a AM	106	if (page && page_to_phys(page) + size > dma_mask) {
	107	dma_release_from_contiguous(dev, page, count);
	108	page = NULL;
	109	}
	110	}
c091c71a	111	/* fallback */
0a2b9a6e MS	112	if (!page)
0a2b9a6e MS	113	page = alloc_pages_node(dev_to_node(dev), flag, get_order(size));
9f6ac577 FT	114	if (!page)
	115	return NULL;
	116
	117	addr = page_to_phys(page);
a4c2baa6	118	if (addr + size > dma_mask) {
9f6ac577 FT	119	__free_pages(page, get_order(size));
9f6ac577 FT	120
284901a9	121	if (dma_mask < DMA_BIT_MASK(32) && !(flag & GFP_DMA)) {
9f6ac577 FT	122	flag = (flag & ~GFP_DMA32) \| GFP_DMA;
	123	goto again;
	124	}
	125
	126	return NULL;
	127	}
d92ef66c	128	memset(page_address(page), 0, size);
9f6ac577 FT	129	*dma_addr = addr;
	130	return page_address(page);
	131	}
	132
0a2b9a6e MS	133	void dma_generic_free_coherent(struct device dev, size_t size, void vaddr,
	134	dma_addr_t dma_addr, struct dma_attrs *attrs)
	135	{
	136	unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
	137	struct page *page = virt_to_page(vaddr);
	138
	139	if (!dma_release_from_contiguous(dev, page, count))
	140	free_pages((unsigned long)vaddr, get_order(size));
	141	}
	142
fae9a0d8	143	/*
395cf969 PB	144	* See <Documentation/x86/x86_64/boot-options.txt> for the iommu kernel
395cf969 PB	145	* parameter documentation.
fae9a0d8 GC	146	*/
	147	static __init int iommu_setup(char *p)
	148	{
	149	iommu_merge = 1;
	150
	151	if (!p)
	152	return -EINVAL;
	153
	154	while (*p) {
	155	if (!strncmp(p, "off", 3))
	156	no_iommu = 1;
	157	/* gart_parse_options has more force support */
	158	if (!strncmp(p, "force", 5))
	159	force_iommu = 1;
	160	if (!strncmp(p, "noforce", 7)) {
	161	iommu_merge = 0;
	162	force_iommu = 0;
	163	}
	164
	165	if (!strncmp(p, "biomerge", 8)) {
fae9a0d8 GC	166	iommu_merge = 1;
	167	force_iommu = 1;
	168	}
	169	if (!strncmp(p, "panic", 5))
	170	panic_on_overflow = 1;
	171	if (!strncmp(p, "nopanic", 7))
	172	panic_on_overflow = 0;
	173	if (!strncmp(p, "merge", 5)) {
	174	iommu_merge = 1;
	175	force_iommu = 1;
	176	}
	177	if (!strncmp(p, "nomerge", 7))
	178	iommu_merge = 0;
	179	if (!strncmp(p, "forcesac", 8))
	180	iommu_sac_force = 1;
	181	if (!strncmp(p, "allowdac", 8))
	182	forbid_dac = 0;
	183	if (!strncmp(p, "nodac", 5))
2ae8bb75	184	forbid_dac = 1;
fae9a0d8 GC	185	if (!strncmp(p, "usedac", 6)) {
	186	forbid_dac = -1;
	187	return 1;
	188	}
	189	#ifdef CONFIG_SWIOTLB
	190	if (!strncmp(p, "soft", 4))
	191	swiotlb = 1;
3238c0c4	192	#endif
80286879	193	if (!strncmp(p, "pt", 2))
4ed0d3e6	194	iommu_pass_through = 1;
fae9a0d8	195
fae9a0d8	196	gart_parse_options(p);
fae9a0d8 GC	197
	198	#ifdef CONFIG_CALGARY_IOMMU
	199	if (!strncmp(p, "calgary", 7))
	200	use_calgary = 1;
	201	#endif /* CONFIG_CALGARY_IOMMU */
	202
	203	p += strcspn(p, ",");
	204	if (*p == ',')
	205	++p;
	206	}
	207	return 0;
	208	}
	209	early_param("iommu", iommu_setup);
	210
8e0c3797 GC	211	int dma_supported(struct device *dev, u64 mask)
8e0c3797 GC	212	{
160c1d8e	213	struct dma_map_ops *ops = get_dma_ops(dev);
8d8bb39b	214
8e0c3797 GC	215	#ifdef CONFIG_PCI
8e0c3797 GC	216	if (mask > 0xffffffff && forbid_dac > 0) {
fc3a8828	217	dev_info(dev, "PCI: Disallowing DAC for device\n");
8e0c3797 GC	218	return 0;
	219	}
	220	#endif
	221
8d8bb39b FT	222	if (ops->dma_supported)
8d8bb39b FT	223	return ops->dma_supported(dev, mask);
8e0c3797 GC	224
	225	/* Copied from i386. Doesn't make much sense, because it will
	226	only work for pci_alloc_coherent.
	227	The caller just has to use GFP_DMA in this case. */
2f4f27d4	228	if (mask < DMA_BIT_MASK(24))
8e0c3797 GC	229	return 0;
	230
	231	/* Tell the device to use SAC when IOMMU force is on. This
	232	allows the driver to use cheaper accesses in some cases.
	233
	234	Problem with this is that if we overflow the IOMMU area and
	235	return DAC as fallback address the device may not handle it
	236	correctly.
	237
	238	As a special case some controllers have a 39bit address
	239	mode that is as efficient as 32bit (aic79xx). Don't force
	240	SAC for these. Assume all masks <= 40 bits are of this
	241	type. Normally this doesn't make any difference, but gives
	242	more gentle handling of IOMMU overflow. */
50cf156a	243	if (iommu_sac_force && (mask >= DMA_BIT_MASK(40))) {
fc3a8828	244	dev_info(dev, "Force SAC with mask %Lx\n", mask);
8e0c3797 GC	245	return 0;
	246	}
	247
	248	return 1;
	249	}
	250	EXPORT_SYMBOL(dma_supported);
	251
cb5867a5 GC	252	static int __init pci_iommu_init(void)
cb5867a5 GC	253	{
ee1f284f	254	struct iommu_table_entry *p;
2118d0c5 JR	255	dma_debug_init(PREALLOC_DMA_DEBUG_ENTRIES);
2118d0c5 JR	256
86f31952 JR	257	#ifdef CONFIG_PCI
	258	dma_debug_add_bus(&pci_bus_type);
	259	#endif
d07c1be0 FT	260	x86_init.iommu.iommu_init();
d07c1be0 FT	261
ee1f284f KRW	262	for (p = __iommu_table; p < __iommu_table_end; p++) {
	263	if (p && (p->flags & IOMMU_DETECTED) && p->late_init)
	264	p->late_init();
	265	}
75f1cdf1	266
cb5867a5 GC	267	return 0;
cb5867a5 GC	268	}
cb5867a5	269	/* Must execute after PCI subsystem */
9a821b23	270	rootfs_initcall(pci_iommu_init);
3b15e581 FY	271
	272	#ifdef CONFIG_PCI
	273	/* Many VIA bridges seem to corrupt data for DAC. Disable it here */
	274
a18e3690	275	static void via_no_dac(struct pci_dev *dev)
3b15e581	276	{
c484b241	277	if (forbid_dac == 0) {
13bf7576	278	dev_info(&dev->dev, "disabling DAC on VIA PCI bridge\n");
3b15e581 FY	279	forbid_dac = 1;
	280	}
	281	}
c484b241 YL	282	DECLARE_PCI_FIXUP_CLASS_FINAL(PCI_VENDOR_ID_VIA, PCI_ANY_ID,
c484b241 YL	283	PCI_CLASS_BRIDGE_PCI, 8, via_no_dac);
3b15e581	284	#endif