[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;

/*
 * Group multi-function PCI devices into a single device-group for the
 * iommu_device_group interface.  This tells the iommu driver to pretend
 * it cannot distinguish between functions of a device, exposing only one
 * group for the device.  Useful for disallowing use of individual PCI
 * functions from userspace drivers.
 */
int iommu_group_mf __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       32768

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;
	if (!(flag & GFP_ATOMIC))
		page = dma_alloc_from_contiguous(dev, count, get_order(size));
	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;
	}

	*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;
		if (!strncmp(p, "group_mf", 8))
			iommu_group_mf = 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 __devinit 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
bcb71abe AW	48	/*
	49	* Group multi-function PCI devices into a single device-group for the
	50	* iommu_device_group interface. This tells the iommu driver to pretend
	51	* it cannot distinguish between functions of a device, exposing only one
	52	* group for the device. Useful for disallowing use of individual PCI
	53	* functions from userspace drivers.
	54	*/
	55	int iommu_group_mf __read_mostly;
	56
ee1f284f KRW	57	extern struct iommu_table_entry __iommu_table[], __iommu_table_end[];
ee1f284f KRW	58
eb647138	59	/* Dummy device used for NULL arguments (normally ISA). */
6c505ce3	60	struct device x86_dma_fallback_dev = {
1a927133	61	.init_name = "fallback device",
eb647138	62	.coherent_dma_mask = ISA_DMA_BIT_MASK,
6c505ce3	63	.dma_mask = &x86_dma_fallback_dev.coherent_dma_mask,
098cb7f2	64	};
6c505ce3	65	EXPORT_SYMBOL(x86_dma_fallback_dev);
098cb7f2	66
2118d0c5 JR	67	/* Number of entries preallocated for DMA-API debugging */
	68	#define PREALLOC_DMA_DEBUG_ENTRIES 32768
	69
459121c9 GC	70	int dma_set_mask(struct device *dev, u64 mask)
	71	{
	72	if (!dev->dma_mask \|\| !dma_supported(dev, mask))
	73	return -EIO;
	74
	75	*dev->dma_mask = mask;
	76
	77	return 0;
	78	}
	79	EXPORT_SYMBOL(dma_set_mask);
	80
116890d5 GC	81	void __init pci_iommu_alloc(void)
116890d5 GC	82	{
ee1f284f KRW	83	struct iommu_table_entry *p;
ee1f284f KRW	84
ee1f284f KRW	85	sort_iommu_table(__iommu_table, __iommu_table_end);
ee1f284f KRW	86	check_iommu_entries(__iommu_table, __iommu_table_end);
116890d5	87
ee1f284f KRW	88	for (p = __iommu_table; p < __iommu_table_end; p++) {
	89	if (p && p->detect && p->detect() > 0) {
	90	p->flags \|= IOMMU_DETECTED;
	91	if (p->early_init)
	92	p->early_init();
	93	if (p->flags & IOMMU_FINISH_IF_DETECTED)
	94	break;
	95	}
	96	}
116890d5	97	}
9f6ac577	98	void dma_generic_alloc_coherent(struct device dev, size_t size,
baa676fc AP	99	dma_addr_t *dma_addr, gfp_t flag,
baa676fc AP	100	struct dma_attrs *attrs)
9f6ac577 FT	101	{
9f6ac577 FT	102	unsigned long dma_mask;
c080e26e	103	struct page *page;
0a2b9a6e	104	unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
9f6ac577 FT	105	dma_addr_t addr;
	106
	107	dma_mask = dma_alloc_coherent_mask(dev, flag);
	108
	109	flag \|= __GFP_ZERO;
	110	again:
c080e26e	111	page = NULL;
0a2b9a6e MS	112	if (!(flag & GFP_ATOMIC))
	113	page = dma_alloc_from_contiguous(dev, count, get_order(size));
	114	if (!page)
	115	page = alloc_pages_node(dev_to_node(dev), flag, get_order(size));
9f6ac577 FT	116	if (!page)
	117	return NULL;
	118
	119	addr = page_to_phys(page);
a4c2baa6	120	if (addr + size > dma_mask) {
9f6ac577 FT	121	__free_pages(page, get_order(size));
9f6ac577 FT	122
284901a9	123	if (dma_mask < DMA_BIT_MASK(32) && !(flag & GFP_DMA)) {
9f6ac577 FT	124	flag = (flag & ~GFP_DMA32) \| GFP_DMA;
	125	goto again;
	126	}
	127
	128	return NULL;
	129	}
	130
	131	*dma_addr = addr;
	132	return page_address(page);
	133	}
	134
0a2b9a6e MS	135	void dma_generic_free_coherent(struct device dev, size_t size, void vaddr,
	136	dma_addr_t dma_addr, struct dma_attrs *attrs)
	137	{
	138	unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
	139	struct page *page = virt_to_page(vaddr);
	140
	141	if (!dma_release_from_contiguous(dev, page, count))
	142	free_pages((unsigned long)vaddr, get_order(size));
	143	}
	144
fae9a0d8	145	/*
395cf969 PB	146	* See <Documentation/x86/x86_64/boot-options.txt> for the iommu kernel
395cf969 PB	147	* parameter documentation.
fae9a0d8 GC	148	*/
	149	static __init int iommu_setup(char *p)
	150	{
	151	iommu_merge = 1;
	152
	153	if (!p)
	154	return -EINVAL;
	155
	156	while (*p) {
	157	if (!strncmp(p, "off", 3))
	158	no_iommu = 1;
	159	/* gart_parse_options has more force support */
	160	if (!strncmp(p, "force", 5))
	161	force_iommu = 1;
	162	if (!strncmp(p, "noforce", 7)) {
	163	iommu_merge = 0;
	164	force_iommu = 0;
	165	}
	166
	167	if (!strncmp(p, "biomerge", 8)) {
fae9a0d8 GC	168	iommu_merge = 1;
	169	force_iommu = 1;
	170	}
	171	if (!strncmp(p, "panic", 5))
	172	panic_on_overflow = 1;
	173	if (!strncmp(p, "nopanic", 7))
	174	panic_on_overflow = 0;
	175	if (!strncmp(p, "merge", 5)) {
	176	iommu_merge = 1;
	177	force_iommu = 1;
	178	}
	179	if (!strncmp(p, "nomerge", 7))
	180	iommu_merge = 0;
	181	if (!strncmp(p, "forcesac", 8))
	182	iommu_sac_force = 1;
	183	if (!strncmp(p, "allowdac", 8))
	184	forbid_dac = 0;
	185	if (!strncmp(p, "nodac", 5))
2ae8bb75	186	forbid_dac = 1;
fae9a0d8 GC	187	if (!strncmp(p, "usedac", 6)) {
	188	forbid_dac = -1;
	189	return 1;
	190	}
	191	#ifdef CONFIG_SWIOTLB
	192	if (!strncmp(p, "soft", 4))
	193	swiotlb = 1;
3238c0c4	194	#endif
80286879	195	if (!strncmp(p, "pt", 2))
4ed0d3e6	196	iommu_pass_through = 1;
bcb71abe AW	197	if (!strncmp(p, "group_mf", 8))
bcb71abe AW	198	iommu_group_mf = 1;
fae9a0d8	199
fae9a0d8	200	gart_parse_options(p);
fae9a0d8 GC	201
	202	#ifdef CONFIG_CALGARY_IOMMU
	203	if (!strncmp(p, "calgary", 7))
	204	use_calgary = 1;
	205	#endif /* CONFIG_CALGARY_IOMMU */
	206
	207	p += strcspn(p, ",");
	208	if (*p == ',')
	209	++p;
	210	}
	211	return 0;
	212	}
	213	early_param("iommu", iommu_setup);
	214
8e0c3797 GC	215	int dma_supported(struct device *dev, u64 mask)
8e0c3797 GC	216	{
160c1d8e	217	struct dma_map_ops *ops = get_dma_ops(dev);
8d8bb39b	218
8e0c3797 GC	219	#ifdef CONFIG_PCI
8e0c3797 GC	220	if (mask > 0xffffffff && forbid_dac > 0) {
fc3a8828	221	dev_info(dev, "PCI: Disallowing DAC for device\n");
8e0c3797 GC	222	return 0;
	223	}
	224	#endif
	225
8d8bb39b FT	226	if (ops->dma_supported)
8d8bb39b FT	227	return ops->dma_supported(dev, mask);
8e0c3797 GC	228
	229	/* Copied from i386. Doesn't make much sense, because it will
	230	only work for pci_alloc_coherent.
	231	The caller just has to use GFP_DMA in this case. */
2f4f27d4	232	if (mask < DMA_BIT_MASK(24))
8e0c3797 GC	233	return 0;
	234
	235	/* Tell the device to use SAC when IOMMU force is on. This
	236	allows the driver to use cheaper accesses in some cases.
	237
	238	Problem with this is that if we overflow the IOMMU area and
	239	return DAC as fallback address the device may not handle it
	240	correctly.
	241
	242	As a special case some controllers have a 39bit address
	243	mode that is as efficient as 32bit (aic79xx). Don't force
	244	SAC for these. Assume all masks <= 40 bits are of this
	245	type. Normally this doesn't make any difference, but gives
	246	more gentle handling of IOMMU overflow. */
50cf156a	247	if (iommu_sac_force && (mask >= DMA_BIT_MASK(40))) {
fc3a8828	248	dev_info(dev, "Force SAC with mask %Lx\n", mask);
8e0c3797 GC	249	return 0;
	250	}
	251
	252	return 1;
	253	}
	254	EXPORT_SYMBOL(dma_supported);
	255
cb5867a5 GC	256	static int __init pci_iommu_init(void)
cb5867a5 GC	257	{
ee1f284f	258	struct iommu_table_entry *p;
2118d0c5 JR	259	dma_debug_init(PREALLOC_DMA_DEBUG_ENTRIES);
2118d0c5 JR	260
86f31952 JR	261	#ifdef CONFIG_PCI
	262	dma_debug_add_bus(&pci_bus_type);
	263	#endif
d07c1be0 FT	264	x86_init.iommu.iommu_init();
d07c1be0 FT	265
ee1f284f KRW	266	for (p = __iommu_table; p < __iommu_table_end; p++) {
	267	if (p && (p->flags & IOMMU_DETECTED) && p->late_init)
	268	p->late_init();
	269	}
75f1cdf1	270
cb5867a5 GC	271	return 0;
cb5867a5 GC	272	}
cb5867a5	273	/* Must execute after PCI subsystem */
9a821b23	274	rootfs_initcall(pci_iommu_init);
3b15e581 FY	275
	276	#ifdef CONFIG_PCI
	277	/* Many VIA bridges seem to corrupt data for DAC. Disable it here */
	278
	279	static __devinit void via_no_dac(struct pci_dev *dev)
	280	{
c484b241	281	if (forbid_dac == 0) {
13bf7576	282	dev_info(&dev->dev, "disabling DAC on VIA PCI bridge\n");
3b15e581 FY	283	forbid_dac = 1;
	284	}
	285	}
c484b241 YL	286	DECLARE_PCI_FIXUP_CLASS_FINAL(PCI_VENDOR_ID_VIA, PCI_ANY_ID,
c484b241 YL	287	PCI_CLASS_BRIDGE_PCI, 8, via_no_dac);
3b15e581	288	#endif