[net-next-2.6.git] / arch / x86 / pci / acpi.c

#include <linux/pci.h>
#include <linux/acpi.h>
#include <linux/init.h>
#include <linux/irq.h>
#include <linux/dmi.h>
#include <asm/numa.h>
#include <asm/pci_x86.h>

struct pci_root_info {
	struct acpi_device *bridge;
	char *name;
	unsigned int res_num;
	struct resource *res;
	struct pci_bus *bus;
	int busnum;
};

static bool pci_use_crs = true;

static int __init set_use_crs(const struct dmi_system_id *id)
{
	pci_use_crs = true;
	return 0;
}

static const struct dmi_system_id pci_use_crs_table[] __initconst = {
	/* http://bugzilla.kernel.org/show_bug.cgi?id=14183 */
	{
		.callback = set_use_crs,
		.ident = "IBM System x3800",
		.matches = {
			DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
			DMI_MATCH(DMI_PRODUCT_NAME, "x3800"),
		},
	},
	{}
};

void __init pci_acpi_crs_quirks(void)
{
	int year;

	if (dmi_get_date(DMI_BIOS_DATE, &year, NULL, NULL) && year < 2008)
		pci_use_crs = false;

	dmi_check_system(pci_use_crs_table);

	/*
	 * If the user specifies "pci=use_crs" or "pci=nocrs" explicitly, that
	 * takes precedence over anything we figured out above.
	 */
	if (pci_probe & PCI_ROOT_NO_CRS)
		pci_use_crs = false;
	else if (pci_probe & PCI_USE__CRS)
		pci_use_crs = true;

	printk(KERN_INFO "PCI: %s host bridge windows from ACPI; "
	       "if necessary, use \"pci=%s\" and report a bug\n",
	       pci_use_crs ? "Using" : "Ignoring",
	       pci_use_crs ? "nocrs" : "use_crs");
}

static acpi_status
resource_to_addr(struct acpi_resource *resource,
			struct acpi_resource_address64 *addr)
{
	acpi_status status;

	status = acpi_resource_to_address64(resource, addr);
	if (ACPI_SUCCESS(status) &&
	    (addr->resource_type == ACPI_MEMORY_RANGE ||
	    addr->resource_type == ACPI_IO_RANGE) &&
	    addr->address_length > 0 &&
	    addr->producer_consumer == ACPI_PRODUCER) {
		return AE_OK;
	}
	return AE_ERROR;
}

static acpi_status
count_resource(struct acpi_resource *acpi_res, void *data)
{
	struct pci_root_info *info = data;
	struct acpi_resource_address64 addr;
	acpi_status status;

	status = resource_to_addr(acpi_res, &addr);
	if (ACPI_SUCCESS(status))
		info->res_num++;
	return AE_OK;
}

static void
align_resource(struct acpi_device *bridge, struct resource *res)
{
	int align = (res->flags & IORESOURCE_MEM) ? 16 : 4;

	/*
	 * Host bridge windows are not BARs, but the decoders on the PCI side
	 * that claim this address space have starting alignment and length
	 * constraints, so fix any obvious BIOS goofs.
	 */
	if (!IS_ALIGNED(res->start, align)) {
		dev_printk(KERN_DEBUG, &bridge->dev,
			   "host bridge window %pR invalid; "
			   "aligning start to %d-byte boundary\n", res, align);
		res->start &= ~(align - 1);
	}
	if (!IS_ALIGNED(res->end + 1, align)) {
		dev_printk(KERN_DEBUG, &bridge->dev,
			   "host bridge window %pR invalid; "
			   "aligning end to %d-byte boundary\n", res, align);
		res->end = ALIGN(res->end, align) - 1;
	}
}

static acpi_status
setup_resource(struct acpi_resource *acpi_res, void *data)
{
	struct pci_root_info *info = data;
	struct resource *res;
	struct acpi_resource_address64 addr;
	acpi_status status;
	unsigned long flags;
	struct resource *root;
	u64 start, end;

	status = resource_to_addr(acpi_res, &addr);
	if (!ACPI_SUCCESS(status))
		return AE_OK;

	if (addr.resource_type == ACPI_MEMORY_RANGE) {
		root = &iomem_resource;
		flags = IORESOURCE_MEM;
		if (addr.info.mem.caching == ACPI_PREFETCHABLE_MEMORY)
			flags |= IORESOURCE_PREFETCH;
	} else if (addr.resource_type == ACPI_IO_RANGE) {
		root = &ioport_resource;
		flags = IORESOURCE_IO;
	} else
		return AE_OK;

	start = addr.minimum + addr.translation_offset;
	end = start + addr.address_length - 1;

	res = &info->res[info->res_num];
	res->name = info->name;
	res->flags = flags;
	res->start = start;
	res->end = end;
	res->child = NULL;
	align_resource(info->bridge, res);

	if (!pci_use_crs) {
		dev_printk(KERN_DEBUG, &info->bridge->dev,
			   "host bridge window %pR (ignored)\n", res);
		return AE_OK;
	}

	if (insert_resource(root, res)) {
		dev_err(&info->bridge->dev,
			"can't allocate host bridge window %pR\n", res);
	} else {
		pci_bus_add_resource(info->bus, res, 0);
		info->res_num++;
		if (addr.translation_offset)
			dev_info(&info->bridge->dev, "host bridge window %pR "
				 "(PCI address [%#llx-%#llx])\n",
				 res, res->start - addr.translation_offset,
				 res->end - addr.translation_offset);
		else
			dev_info(&info->bridge->dev,
				 "host bridge window %pR\n", res);
	}
	return AE_OK;
}

static void
get_current_resources(struct acpi_device *device, int busnum,
			int domain, struct pci_bus *bus)
{
	struct pci_root_info info;
	size_t size;

	if (pci_use_crs)
		pci_bus_remove_resources(bus);

	info.bridge = device;
	info.bus = bus;
	info.res_num = 0;
	acpi_walk_resources(device->handle, METHOD_NAME__CRS, count_resource,
				&info);
	if (!info.res_num)
		return;

	size = sizeof(*info.res) * info.res_num;
	info.res = kmalloc(size, GFP_KERNEL);
	if (!info.res)
		goto res_alloc_fail;

	info.name = kmalloc(16, GFP_KERNEL);
	if (!info.name)
		goto name_alloc_fail;
	sprintf(info.name, "PCI Bus %04x:%02x", domain, busnum);

	info.res_num = 0;
	acpi_walk_resources(device->handle, METHOD_NAME__CRS, setup_resource,
				&info);

	return;

name_alloc_fail:
	kfree(info.res);
res_alloc_fail:
	return;
}

struct pci_bus * __devinit pci_acpi_scan_root(struct acpi_device *device, int domain, int busnum)
{
	struct pci_bus *bus;
	struct pci_sysdata *sd;
	int node;
#ifdef CONFIG_ACPI_NUMA
	int pxm;
#endif

	if (domain && !pci_domains_supported) {
		printk(KERN_WARNING "pci_bus %04x:%02x: "
		       "ignored (multiple domains not supported)\n",
		       domain, busnum);
		return NULL;
	}

	node = -1;
#ifdef CONFIG_ACPI_NUMA
	pxm = acpi_get_pxm(device->handle);
	if (pxm >= 0)
		node = pxm_to_node(pxm);
	if (node != -1)
		set_mp_bus_to_node(busnum, node);
	else
#endif
		node = get_mp_bus_to_node(busnum);

	if (node != -1 && !node_online(node))
		node = -1;

	/* Allocate per-root-bus (not per bus) arch-specific data.
	 * TODO: leak; this memory is never freed.
	 * It's arguable whether it's worth the trouble to care.
	 */
	sd = kzalloc(sizeof(*sd), GFP_KERNEL);
	if (!sd) {
		printk(KERN_WARNING "pci_bus %04x:%02x: "
		       "ignored (out of memory)\n", domain, busnum);
		return NULL;
	}

	sd->domain = domain;
	sd->node = node;
	/*
	 * Maybe the desired pci bus has been already scanned. In such case
	 * it is unnecessary to scan the pci bus with the given domain,busnum.
	 */
	bus = pci_find_bus(domain, busnum);
	if (bus) {
		/*
		 * If the desired bus exits, the content of bus->sysdata will
		 * be replaced by sd.
		 */
		memcpy(bus->sysdata, sd, sizeof(*sd));
		kfree(sd);
	} else {
		bus = pci_create_bus(NULL, busnum, &pci_root_ops, sd);
		if (bus) {
			get_current_resources(device, busnum, domain, bus);
			bus->subordinate = pci_scan_child_bus(bus);
		}
	}

	if (!bus)
		kfree(sd);

	if (bus && node != -1) {
#ifdef CONFIG_ACPI_NUMA
		if (pxm >= 0)
			dev_printk(KERN_DEBUG, &bus->dev,
				   "on NUMA node %d (pxm %d)\n", node, pxm);
#else
		dev_printk(KERN_DEBUG, &bus->dev, "on NUMA node %d\n", node);
#endif
	}

	return bus;
}

int __init pci_acpi_init(void)
{
	struct pci_dev *dev = NULL;

	if (acpi_noirq)
		return -ENODEV;

	printk(KERN_INFO "PCI: Using ACPI for IRQ routing\n");
	acpi_irq_penalty_init();
	pcibios_enable_irq = acpi_pci_irq_enable;
	pcibios_disable_irq = acpi_pci_irq_disable;
	x86_init.pci.init_irq = x86_init_noop;

	if (pci_routeirq) {
		/*
		 * PCI IRQ routing is set up by pci_enable_device(), but we
		 * also do it here in case there are still broken drivers that
		 * don't use pci_enable_device().
		 */
		printk(KERN_INFO "PCI: Routing PCI interrupts for all devices because \"pci=routeirq\" specified\n");
		for_each_pci_dev(dev)
			acpi_pci_irq_enable(dev);
	}

	return 0;
}
Commit	Line	Data
1da177e4 LT	1	#include <linux/pci.h>
	2	#include <linux/acpi.h>
	3	#include <linux/init.h>
b33fa1f3	4	#include <linux/irq.h>
036fff4c	5	#include <linux/dmi.h>
69e1a33f	6	#include <asm/numa.h>
82487711	7	#include <asm/pci_x86.h>
1da177e4	8
62f420f8	9	struct pci_root_info {
42887b29	10	struct acpi_device *bridge;
62f420f8 GH	11	char *name;
	12	unsigned int res_num;
	13	struct resource *res;
	14	struct pci_bus *bus;
	15	int busnum;
	16	};
	17
7bc5e3f2 BH	18	static bool pci_use_crs = true;
	19
	20	static int __init set_use_crs(const struct dmi_system_id *id)
	21	{
	22	pci_use_crs = true;
	23	return 0;
	24	}
	25
	26	static const struct dmi_system_id pci_use_crs_table[] __initconst = {
	27	/* http://bugzilla.kernel.org/show_bug.cgi?id=14183 */
	28	{
	29	.callback = set_use_crs,
	30	.ident = "IBM System x3800",
	31	.matches = {
	32	DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
	33	DMI_MATCH(DMI_PRODUCT_NAME, "x3800"),
	34	},
	35	},
	36	{}
	37	};
	38
	39	void __init pci_acpi_crs_quirks(void)
	40	{
	41	int year;
	42
	43	if (dmi_get_date(DMI_BIOS_DATE, &year, NULL, NULL) && year < 2008)
	44	pci_use_crs = false;
	45
	46	dmi_check_system(pci_use_crs_table);
	47
	48	/*
	49	* If the user specifies "pci=use_crs" or "pci=nocrs" explicitly, that
	50	* takes precedence over anything we figured out above.
	51	*/
	52	if (pci_probe & PCI_ROOT_NO_CRS)
	53	pci_use_crs = false;
	54	else if (pci_probe & PCI_USE__CRS)
	55	pci_use_crs = true;
	56
	57	printk(KERN_INFO "PCI: %s host bridge windows from ACPI; "
	58	"if necessary, use \"pci=%s\" and report a bug\n",
	59	pci_use_crs ? "Using" : "Ignoring",
	60	pci_use_crs ? "nocrs" : "use_crs");
	61	}
	62
62f420f8 GH	63	static acpi_status
	64	resource_to_addr(struct acpi_resource *resource,
	65	struct acpi_resource_address64 *addr)
	66	{
	67	acpi_status status;
	68
	69	status = acpi_resource_to_address64(resource, addr);
	70	if (ACPI_SUCCESS(status) &&
	71	(addr->resource_type == ACPI_MEMORY_RANGE \|\|
	72	addr->resource_type == ACPI_IO_RANGE) &&
	73	addr->address_length > 0 &&
	74	addr->producer_consumer == ACPI_PRODUCER) {
	75	return AE_OK;
	76	}
	77	return AE_ERROR;
	78	}
	79
	80	static acpi_status
	81	count_resource(struct acpi_resource acpi_res, void data)
	82	{
	83	struct pci_root_info *info = data;
	84	struct acpi_resource_address64 addr;
	85	acpi_status status;
	86
	87	status = resource_to_addr(acpi_res, &addr);
	88	if (ACPI_SUCCESS(status))
	89	info->res_num++;
	90	return AE_OK;
	91	}
	92
03db42ad BH	93	static void
	94	align_resource(struct acpi_device bridge, struct resource res)
	95	{
	96	int align = (res->flags & IORESOURCE_MEM) ? 16 : 4;
	97
	98	/*
	99	* Host bridge windows are not BARs, but the decoders on the PCI side
	100	* that claim this address space have starting alignment and length
	101	* constraints, so fix any obvious BIOS goofs.
	102	*/
ea7f1b6e	103	if (!IS_ALIGNED(res->start, align)) {
03db42ad BH	104	dev_printk(KERN_DEBUG, &bridge->dev,
	105	"host bridge window %pR invalid; "
	106	"aligning start to %d-byte boundary\n", res, align);
	107	res->start &= ~(align - 1);
	108	}
ea7f1b6e	109	if (!IS_ALIGNED(res->end + 1, align)) {
03db42ad BH	110	dev_printk(KERN_DEBUG, &bridge->dev,
	111	"host bridge window %pR invalid; "
	112	"aligning end to %d-byte boundary\n", res, align);
ea7f1b6e	113	res->end = ALIGN(res->end, align) - 1;
03db42ad BH	114	}
	115	}
	116
62f420f8 GH	117	static acpi_status
	118	setup_resource(struct acpi_resource acpi_res, void data)
	119	{
	120	struct pci_root_info *info = data;
	121	struct resource *res;
	122	struct acpi_resource_address64 addr;
	123	acpi_status status;
	124	unsigned long flags;
	125	struct resource *root;
2cdb3f1d YL	126	u64 start, end;
2cdb3f1d YL	127
62f420f8 GH	128	status = resource_to_addr(acpi_res, &addr);
	129	if (!ACPI_SUCCESS(status))
	130	return AE_OK;
	131
	132	if (addr.resource_type == ACPI_MEMORY_RANGE) {
	133	root = &iomem_resource;
	134	flags = IORESOURCE_MEM;
	135	if (addr.info.mem.caching == ACPI_PREFETCHABLE_MEMORY)
	136	flags \|= IORESOURCE_PREFETCH;
	137	} else if (addr.resource_type == ACPI_IO_RANGE) {
	138	root = &ioport_resource;
	139	flags = IORESOURCE_IO;
	140	} else
	141	return AE_OK;
	142
2cdb3f1d YL	143	start = addr.minimum + addr.translation_offset;
2cdb3f1d YL	144	end = start + addr.address_length - 1;
f9cde5ff	145
2cdb3f1d YL	146	res = &info->res[info->res_num];
	147	res->name = info->name;
	148	res->flags = flags;
	149	res->start = start;
	150	res->end = end;
	151	res->child = NULL;
03db42ad	152	align_resource(info->bridge, res);
2cdb3f1d	153
7bc5e3f2	154	if (!pci_use_crs) {
f1db6fde BH	155	dev_printk(KERN_DEBUG, &info->bridge->dev,
	156	"host bridge window %pR (ignored)\n", res);
	157	return AE_OK;
	158	}
	159
62f420f8	160	if (insert_resource(root, res)) {
c7dabef8 BH	161	dev_err(&info->bridge->dev,
c7dabef8 BH	162	"can't allocate host bridge window %pR\n", res);
62f420f8	163	} else {
2fe2abf8	164	pci_bus_add_resource(info->bus, res, 0);
62f420f8	165	info->res_num++;
42887b29	166	if (addr.translation_offset)
c7dabef8	167	dev_info(&info->bridge->dev, "host bridge window %pR "
42887b29 BH	168	"(PCI address [%#llx-%#llx])\n",
	169	res, res->start - addr.translation_offset,
	170	res->end - addr.translation_offset);
	171	else
	172	dev_info(&info->bridge->dev,
c7dabef8	173	"host bridge window %pR\n", res);
62f420f8 GH	174	}
	175	return AE_OK;
	176	}
	177
62f420f8 GH	178	static void
62f420f8 GH	179	get_current_resources(struct acpi_device *device, int busnum,
cb3576fa	180	int domain, struct pci_bus *bus)
62f420f8 GH	181	{
	182	struct pci_root_info info;
	183	size_t size;
	184
7bc5e3f2	185	if (pci_use_crs)
2fe2abf8	186	pci_bus_remove_resources(bus);
f1db6fde	187
42887b29	188	info.bridge = device;
62f420f8 GH	189	info.bus = bus;
	190	info.res_num = 0;
	191	acpi_walk_resources(device->handle, METHOD_NAME__CRS, count_resource,
	192	&info);
	193	if (!info.res_num)
	194	return;
	195
	196	size = sizeof(info.res) info.res_num;
	197	info.res = kmalloc(size, GFP_KERNEL);
	198	if (!info.res)
	199	goto res_alloc_fail;
	200
cb3576fa	201	info.name = kmalloc(16, GFP_KERNEL);
62f420f8 GH	202	if (!info.name)
62f420f8 GH	203	goto name_alloc_fail;
cb3576fa	204	sprintf(info.name, "PCI Bus %04x:%02x", domain, busnum);
62f420f8 GH	205
	206	info.res_num = 0;
	207	acpi_walk_resources(device->handle, METHOD_NAME__CRS, setup_resource,
	208	&info);
62f420f8 GH	209
	210	return;
	211
	212	name_alloc_fail:
	213	kfree(info.res);
	214	res_alloc_fail:
	215	return;
	216	}
	217
1da177e4 LT	218	struct pci_bus * __devinit pci_acpi_scan_root(struct acpi_device *device, int domain, int busnum)
1da177e4 LT	219	{
69e1a33f	220	struct pci_bus *bus;
08f1c192	221	struct pci_sysdata *sd;
871d5f8d YL	222	int node;
871d5f8d YL	223	#ifdef CONFIG_ACPI_NUMA
08f1c192	224	int pxm;
871d5f8d	225	#endif
08f1c192	226
a79e4198	227	if (domain && !pci_domains_supported) {
2a6bed83 BH	228	printk(KERN_WARNING "pci_bus %04x:%02x: "
	229	"ignored (multiple domains not supported)\n",
	230	domain, busnum);
a79e4198 JG	231	return NULL;
	232	}
	233
871d5f8d YL	234	node = -1;
	235	#ifdef CONFIG_ACPI_NUMA
	236	pxm = acpi_get_pxm(device->handle);
	237	if (pxm >= 0)
	238	node = pxm_to_node(pxm);
	239	if (node != -1)
	240	set_mp_bus_to_node(busnum, node);
	241	else
871d5f8d	242	#endif
871d5f8d	243	node = get_mp_bus_to_node(busnum);
b755de8d YL	244
	245	if (node != -1 && !node_online(node))
	246	node = -1;
871d5f8d	247
08f1c192 MBY	248	/* Allocate per-root-bus (not per bus) arch-specific data.
	249	* TODO: leak; this memory is never freed.
	250	* It's arguable whether it's worth the trouble to care.
	251	*/
	252	sd = kzalloc(sizeof(*sd), GFP_KERNEL);
	253	if (!sd) {
2a6bed83 BH	254	printk(KERN_WARNING "pci_bus %04x:%02x: "
2a6bed83 BH	255	"ignored (out of memory)\n", domain, busnum);
08f1c192 MBY	256	return NULL;
08f1c192 MBY	257	}
69e1a33f	258
a79e4198	259	sd->domain = domain;
871d5f8d	260	sd->node = node;
b87e81e5	261	/*
	262	* Maybe the desired pci bus has been already scanned. In such case
	263	* it is unnecessary to scan the pci bus with the given domain,busnum.
	264	*/
	265	bus = pci_find_bus(domain, busnum);
	266	if (bus) {
	267	/*
	268	* If the desired bus exits, the content of bus->sysdata will
	269	* be replaced by sd.
	270	*/
	271	memcpy(bus->sysdata, sd, sizeof(*sd));
	272	kfree(sd);
626fdfec YL	273	} else {
	274	bus = pci_create_bus(NULL, busnum, &pci_root_ops, sd);
	275	if (bus) {
f1db6fde	276	get_current_resources(device, busnum, domain, bus);
626fdfec YL	277	bus->subordinate = pci_scan_child_bus(bus);
	278	}
	279	}
08f1c192	280
08f1c192 MBY	281	if (!bus)
	282	kfree(sd);
	283
dbb6152e	284	if (bus && node != -1) {
69e1a33f	285	#ifdef CONFIG_ACPI_NUMA
dbb6152e	286	if (pxm >= 0)
2b8c2efe BH	287	dev_printk(KERN_DEBUG, &bus->dev,
2b8c2efe BH	288	"on NUMA node %d (pxm %d)\n", node, pxm);
dbb6152e	289	#else
2b8c2efe	290	dev_printk(KERN_DEBUG, &bus->dev, "on NUMA node %d\n", node);
69e1a33f	291	#endif
dbb6152e	292	}
62f420f8	293
69e1a33f	294	return bus;
1da177e4 LT	295	}
1da177e4 LT	296
8dd779b1	297	int __init pci_acpi_init(void)
1da177e4 LT	298	{
	299	struct pci_dev *dev = NULL;
	300
1da177e4	301	if (acpi_noirq)
b72d0db9	302	return -ENODEV;
1da177e4 LT	303
	304	printk(KERN_INFO "PCI: Using ACPI for IRQ routing\n");
	305	acpi_irq_penalty_init();
1da177e4	306	pcibios_enable_irq = acpi_pci_irq_enable;
87bec66b	307	pcibios_disable_irq = acpi_pci_irq_disable;
ab3b3793	308	x86_init.pci.init_irq = x86_init_noop;
1da177e4 LT	309
	310	if (pci_routeirq) {
	311	/*
	312	* PCI IRQ routing is set up by pci_enable_device(), but we
	313	* also do it here in case there are still broken drivers that
	314	* don't use pci_enable_device().
	315	*/
	316	printk(KERN_INFO "PCI: Routing PCI interrupts for all devices because \"pci=routeirq\" specified\n");
fb37fb96	317	for_each_pci_dev(dev)
1da177e4	318	acpi_pci_irq_enable(dev);
657472e9	319	}
1da177e4	320
1da177e4 LT	321	return 0;
1da177e4 LT	322	}