[mirror_ubuntu-bionic-kernel.git] / arch / x86 / platform / efi / quirks.c

#define pr_fmt(fmt) "efi: " fmt

#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/time.h>
#include <linux/types.h>
#include <linux/efi.h>
#include <linux/slab.h>
#include <linux/memblock.h>
#include <linux/bootmem.h>
#include <linux/acpi.h>
#include <linux/dmi.h>
#include <asm/efi.h>
#include <asm/uv/uv.h>

#define EFI_MIN_RESERVE 5120

#define EFI_DUMMY_GUID \
	EFI_GUID(0x4424ac57, 0xbe4b, 0x47dd, 0x9e, 0x97, 0xed, 0x50, 0xf0, 0x9f, 0x92, 0xa9)

static efi_char16_t efi_dummy_name[6] = { 'D', 'U', 'M', 'M', 'Y', 0 };

static bool efi_no_storage_paranoia;

/*
 * Some firmware implementations refuse to boot if there's insufficient
 * space in the variable store. The implementation of garbage collection
 * in some FW versions causes stale (deleted) variables to take up space
 * longer than intended and space is only freed once the store becomes
 * almost completely full.
 *
 * Enabling this option disables the space checks in
 * efi_query_variable_store() and forces garbage collection.
 *
 * Only enable this option if deleting EFI variables does not free up
 * space in your variable store, e.g. if despite deleting variables
 * you're unable to create new ones.
 */
static int __init setup_storage_paranoia(char *arg)
{
	efi_no_storage_paranoia = true;
	return 0;
}
early_param("efi_no_storage_paranoia", setup_storage_paranoia);

/*
 * Deleting the dummy variable which kicks off garbage collection
*/
void efi_delete_dummy_variable(void)
{
	efi.set_variable(efi_dummy_name, &EFI_DUMMY_GUID,
			 EFI_VARIABLE_NON_VOLATILE |
			 EFI_VARIABLE_BOOTSERVICE_ACCESS |
			 EFI_VARIABLE_RUNTIME_ACCESS,
			 0, NULL);
}

/*
 * In the nonblocking case we do not attempt to perform garbage
 * collection if we do not have enough free space. Rather, we do the
 * bare minimum check and give up immediately if the available space
 * is below EFI_MIN_RESERVE.
 *
 * This function is intended to be small and simple because it is
 * invoked from crash handler paths.
 */
static efi_status_t
query_variable_store_nonblocking(u32 attributes, unsigned long size)
{
	efi_status_t status;
	u64 storage_size, remaining_size, max_size;

	status = efi.query_variable_info_nonblocking(attributes, &storage_size,
						     &remaining_size,
						     &max_size);
	if (status != EFI_SUCCESS)
		return status;

	if (remaining_size - size < EFI_MIN_RESERVE)
		return EFI_OUT_OF_RESOURCES;

	return EFI_SUCCESS;
}

/*
 * Some firmware implementations refuse to boot if there's insufficient space
 * in the variable store. Ensure that we never use more than a safe limit.
 *
 * Return EFI_SUCCESS if it is safe to write 'size' bytes to the variable
 * store.
 */
efi_status_t efi_query_variable_store(u32 attributes, unsigned long size,
				      bool nonblocking)
{
	efi_status_t status;
	u64 storage_size, remaining_size, max_size;

	if (!(attributes & EFI_VARIABLE_NON_VOLATILE))
		return 0;

	if (nonblocking)
		return query_variable_store_nonblocking(attributes, size);

	status = efi.query_variable_info(attributes, &storage_size,
					 &remaining_size, &max_size);
	if (status != EFI_SUCCESS)
		return status;

	/*
	 * We account for that by refusing the write if permitting it would
	 * reduce the available space to under 5KB. This figure was provided by
	 * Samsung, so should be safe.
	 */
	if ((remaining_size - size < EFI_MIN_RESERVE) &&
		!efi_no_storage_paranoia) {

		/*
		 * Triggering garbage collection may require that the firmware
		 * generate a real EFI_OUT_OF_RESOURCES error. We can force
		 * that by attempting to use more space than is available.
		 */
		unsigned long dummy_size = remaining_size + 1024;
		void *dummy = kzalloc(dummy_size, GFP_ATOMIC);

		if (!dummy)
			return EFI_OUT_OF_RESOURCES;

		status = efi.set_variable(efi_dummy_name, &EFI_DUMMY_GUID,
					  EFI_VARIABLE_NON_VOLATILE |
					  EFI_VARIABLE_BOOTSERVICE_ACCESS |
					  EFI_VARIABLE_RUNTIME_ACCESS,
					  dummy_size, dummy);

		if (status == EFI_SUCCESS) {
			/*
			 * This should have failed, so if it didn't make sure
			 * that we delete it...
			 */
			efi_delete_dummy_variable();
		}

		kfree(dummy);

		/*
		 * The runtime code may now have triggered a garbage collection
		 * run, so check the variable info again
		 */
		status = efi.query_variable_info(attributes, &storage_size,
						 &remaining_size, &max_size);

		if (status != EFI_SUCCESS)
			return status;

		/*
		 * There still isn't enough room, so return an error
		 */
		if (remaining_size - size < EFI_MIN_RESERVE)
			return EFI_OUT_OF_RESOURCES;
	}

	return EFI_SUCCESS;
}
EXPORT_SYMBOL_GPL(efi_query_variable_store);

/*
 * The UEFI specification makes it clear that the operating system is
 * free to do whatever it wants with boot services code after
 * ExitBootServices() has been called. Ignoring this recommendation a
 * significant bunch of EFI implementations continue calling into boot
 * services code (SetVirtualAddressMap). In order to work around such
 * buggy implementations we reserve boot services region during EFI
 * init and make sure it stays executable. Then, after
 * SetVirtualAddressMap(), it is discarded.
 *
 * However, some boot services regions contain data that is required
 * by drivers, so we need to track which memory ranges can never be
 * freed. This is done by tagging those regions with the
 * EFI_MEMORY_RUNTIME attribute.
 *
 * Any driver that wants to mark a region as reserved must use
 * efi_mem_reserve() which will insert a new EFI memory descriptor
 * into efi.memmap (splitting existing regions if necessary) and tag
 * it with EFI_MEMORY_RUNTIME.
 */
void __init efi_arch_mem_reserve(phys_addr_t addr, u64 size)
{
	phys_addr_t new_phys, new_size;
	struct efi_mem_range mr;
	efi_memory_desc_t md;
	int num_entries;
	void *new;

	if (efi_mem_desc_lookup(addr, &md)) {
		pr_err("Failed to lookup EFI memory descriptor for %pa\n", &addr);
		return;
	}

	if (addr + size > md.phys_addr + (md.num_pages << EFI_PAGE_SHIFT)) {
		pr_err("Region spans EFI memory descriptors, %pa\n", &addr);
		return;
	}

	/* No need to reserve regions that will never be freed. */
	if (md.attribute & EFI_MEMORY_RUNTIME)
		return;

	size += addr % EFI_PAGE_SIZE;
	size = round_up(size, EFI_PAGE_SIZE);
	addr = round_down(addr, EFI_PAGE_SIZE);

	mr.range.start = addr;
	mr.range.end = addr + size - 1;
	mr.attribute = md.attribute | EFI_MEMORY_RUNTIME;

	num_entries = efi_memmap_split_count(&md, &mr.range);
	num_entries += efi.memmap.nr_map;

	new_size = efi.memmap.desc_size * num_entries;

	new_phys = efi_memmap_alloc(num_entries);
	if (!new_phys) {
		pr_err("Could not allocate boot services memmap\n");
		return;
	}

	new = early_memremap(new_phys, new_size);
	if (!new) {
		pr_err("Failed to map new boot services memmap\n");
		return;
	}

	efi_memmap_insert(&efi.memmap, new, &mr);
	early_memunmap(new, new_size);

	efi_memmap_install(new_phys, num_entries);
}

/*
 * Helper function for efi_reserve_boot_services() to figure out if we
 * can free regions in efi_free_boot_services().
 *
 * Use this function to ensure we do not free regions owned by somebody
 * else. We must only reserve (and then free) regions:
 *
 * - Not within any part of the kernel
 * - Not the BIOS reserved area (E820_RESERVED, E820_NVS, etc)
 */
static bool can_free_region(u64 start, u64 size)
{
	if (start + size > __pa_symbol(_text) && start <= __pa_symbol(_end))
		return false;

	if (!e820_all_mapped(start, start+size, E820_RAM))
		return false;

	return true;
}

void __init efi_reserve_boot_services(void)
{
	efi_memory_desc_t *md;

	for_each_efi_memory_desc(md) {
		u64 start = md->phys_addr;
		u64 size = md->num_pages << EFI_PAGE_SHIFT;
		bool already_reserved;

		if (md->type != EFI_BOOT_SERVICES_CODE &&
		    md->type != EFI_BOOT_SERVICES_DATA)
			continue;

		already_reserved = memblock_is_region_reserved(start, size);

		/*
		 * Because the following memblock_reserve() is paired
		 * with free_bootmem_late() for this region in
		 * efi_free_boot_services(), we must be extremely
		 * careful not to reserve, and subsequently free,
		 * critical regions of memory (like the kernel image) or
		 * those regions that somebody else has already
		 * reserved.
		 *
		 * A good example of a critical region that must not be
		 * freed is page zero (first 4Kb of memory), which may
		 * contain boot services code/data but is marked
		 * E820_RESERVED by trim_bios_range().
		 */
		if (!already_reserved) {
			memblock_reserve(start, size);

			/*
			 * If we are the first to reserve the region, no
			 * one else cares about it. We own it and can
			 * free it later.
			 */
			if (can_free_region(start, size))
				continue;
		}

		/*
		 * We don't own the region. We must not free it.
		 *
		 * Setting this bit for a boot services region really
		 * doesn't make sense as far as the firmware is
		 * concerned, but it does provide us with a way to tag
		 * those regions that must not be paired with
		 * free_bootmem_late().
		 */
		md->attribute |= EFI_MEMORY_RUNTIME;
	}
}

void __init efi_free_boot_services(void)
{
	phys_addr_t new_phys, new_size;
	efi_memory_desc_t *md;
	int num_entries = 0;
	void *new, *new_md;

	for_each_efi_memory_desc(md) {
		unsigned long long start = md->phys_addr;
		unsigned long long size = md->num_pages << EFI_PAGE_SHIFT;
		size_t rm_size;

		if (md->type != EFI_BOOT_SERVICES_CODE &&
		    md->type != EFI_BOOT_SERVICES_DATA) {
			num_entries++;
			continue;
		}

		/* Do not free, someone else owns it: */
		if (md->attribute & EFI_MEMORY_RUNTIME) {
			num_entries++;
			continue;
		}

		/*
		 * Nasty quirk: if all sub-1MB memory is used for boot
		 * services, we can get here without having allocated the
		 * real mode trampoline.  It's too late to hand boot services
		 * memory back to the memblock allocator, so instead
		 * try to manually allocate the trampoline if needed.
		 *
		 * I've seen this on a Dell XPS 13 9350 with firmware
		 * 1.4.4 with SGX enabled booting Linux via Fedora 24's
		 * grub2-efi on a hard disk.  (And no, I don't know why
		 * this happened, but Linux should still try to boot rather
		 * panicing early.)
		 */
		rm_size = real_mode_size_needed();
		if (rm_size && (start + rm_size) < (1<<20) && size >= rm_size) {
			set_real_mode_mem(start, rm_size);
			start += rm_size;
			size -= rm_size;
		}

		free_bootmem_late(start, size);
	}

	new_size = efi.memmap.desc_size * num_entries;
	new_phys = efi_memmap_alloc(num_entries);
	if (!new_phys) {
		pr_err("Failed to allocate new EFI memmap\n");
		return;
	}

	new = memremap(new_phys, new_size, MEMREMAP_WB);
	if (!new) {
		pr_err("Failed to map new EFI memmap\n");
		return;
	}

	/*
	 * Build a new EFI memmap that excludes any boot services
	 * regions that are not tagged EFI_MEMORY_RUNTIME, since those
	 * regions have now been freed.
	 */
	new_md = new;
	for_each_efi_memory_desc(md) {
		if (!(md->attribute & EFI_MEMORY_RUNTIME) &&
		    (md->type == EFI_BOOT_SERVICES_CODE ||
		     md->type == EFI_BOOT_SERVICES_DATA))
			continue;

		memcpy(new_md, md, efi.memmap.desc_size);
		new_md += efi.memmap.desc_size;
	}

	memunmap(new);

	if (efi_memmap_install(new_phys, num_entries)) {
		pr_err("Could not install new EFI memmap\n");
		return;
	}
}

/*
 * A number of config table entries get remapped to virtual addresses
 * after entering EFI virtual mode. However, the kexec kernel requires
 * their physical addresses therefore we pass them via setup_data and
 * correct those entries to their respective physical addresses here.
 *
 * Currently only handles smbios which is necessary for some firmware
 * implementation.
 */
int __init efi_reuse_config(u64 tables, int nr_tables)
{
	int i, sz, ret = 0;
	void *p, *tablep;
	struct efi_setup_data *data;

	if (!efi_setup)
		return 0;

	if (!efi_enabled(EFI_64BIT))
		return 0;

	data = early_memremap(efi_setup, sizeof(*data));
	if (!data) {
		ret = -ENOMEM;
		goto out;
	}

	if (!data->smbios)
		goto out_memremap;

	sz = sizeof(efi_config_table_64_t);

	p = tablep = early_memremap(tables, nr_tables * sz);
	if (!p) {
		pr_err("Could not map Configuration table!\n");
		ret = -ENOMEM;
		goto out_memremap;
	}

	for (i = 0; i < efi.systab->nr_tables; i++) {
		efi_guid_t guid;

		guid = ((efi_config_table_64_t *)p)->guid;

		if (!efi_guidcmp(guid, SMBIOS_TABLE_GUID))
			((efi_config_table_64_t *)p)->table = data->smbios;
		p += sz;
	}
	early_memunmap(tablep, nr_tables * sz);

out_memremap:
	early_memunmap(data, sizeof(*data));
out:
	return ret;
}

static const struct dmi_system_id sgi_uv1_dmi[] = {
	{ NULL, "SGI UV1",
		{	DMI_MATCH(DMI_PRODUCT_NAME,	"Stoutland Platform"),
			DMI_MATCH(DMI_PRODUCT_VERSION,	"1.0"),
			DMI_MATCH(DMI_BIOS_VENDOR,	"SGI.COM"),
		}
	},
	{ } /* NULL entry stops DMI scanning */
};

void __init efi_apply_memmap_quirks(void)
{
	/*
	 * Once setup is done earlier, unmap the EFI memory map on mismatched
	 * firmware/kernel architectures since there is no support for runtime
	 * services.
	 */
	if (!efi_runtime_supported()) {
		pr_info("Setup done, disabling due to 32/64-bit mismatch\n");
		efi_memmap_unmap();
	}

	/* UV2+ BIOS has a fix for this issue.  UV1 still needs the quirk. */
	if (dmi_check_system(sgi_uv1_dmi))
		set_bit(EFI_OLD_MEMMAP, &efi.flags);
}

/*
 * For most modern platforms the preferred method of powering off is via
 * ACPI. However, there are some that are known to require the use of
 * EFI runtime services and for which ACPI does not work at all.
 *
 * Using EFI is a last resort, to be used only if no other option
 * exists.
 */
bool efi_reboot_required(void)
{
	if (!acpi_gbl_reduced_hardware)
		return false;

	efi_reboot_quirk_mode = EFI_RESET_WARM;
	return true;
}

bool efi_poweroff_required(void)
{
	return acpi_gbl_reduced_hardware || acpi_no_s5;
}
Commit	Line	Data
26d7f65f MF	1	#define pr_fmt(fmt) "efi: " fmt
26d7f65f MF	2
eeb9db09 ST	3	#include <linux/init.h>
	4	#include <linux/kernel.h>
	5	#include <linux/string.h>
	6	#include <linux/time.h>
	7	#include <linux/types.h>
	8	#include <linux/efi.h>
	9	#include <linux/slab.h>
	10	#include <linux/memblock.h>
	11	#include <linux/bootmem.h>
44be28e9	12	#include <linux/acpi.h>
d394f2d9	13	#include <linux/dmi.h>
eeb9db09 ST	14	#include <asm/efi.h>
	15	#include <asm/uv/uv.h>
	16
	17	#define EFI_MIN_RESERVE 5120
	18
	19	#define EFI_DUMMY_GUID \
	20	EFI_GUID(0x4424ac57, 0xbe4b, 0x47dd, 0x9e, 0x97, 0xed, 0x50, 0xf0, 0x9f, 0x92, 0xa9)
	21
	22	static efi_char16_t efi_dummy_name[6] = { 'D', 'U', 'M', 'M', 'Y', 0 };
	23
	24	static bool efi_no_storage_paranoia;
	25
	26	/*
	27	* Some firmware implementations refuse to boot if there's insufficient
	28	* space in the variable store. The implementation of garbage collection
	29	* in some FW versions causes stale (deleted) variables to take up space
	30	* longer than intended and space is only freed once the store becomes
	31	* almost completely full.
	32	*
	33	* Enabling this option disables the space checks in
	34	* efi_query_variable_store() and forces garbage collection.
	35	*
	36	* Only enable this option if deleting EFI variables does not free up
	37	* space in your variable store, e.g. if despite deleting variables
	38	* you're unable to create new ones.
	39	*/
	40	static int __init setup_storage_paranoia(char *arg)
	41	{
	42	efi_no_storage_paranoia = true;
	43	return 0;
	44	}
	45	early_param("efi_no_storage_paranoia", setup_storage_paranoia);
	46
	47	/*
	48	* Deleting the dummy variable which kicks off garbage collection
	49	*/
	50	void efi_delete_dummy_variable(void)
	51	{
	52	efi.set_variable(efi_dummy_name, &EFI_DUMMY_GUID,
	53	EFI_VARIABLE_NON_VOLATILE \|
	54	EFI_VARIABLE_BOOTSERVICE_ACCESS \|
	55	EFI_VARIABLE_RUNTIME_ACCESS,
	56	0, NULL);
	57	}
	58
ca0e30dc AB	59	/*
	60	* In the nonblocking case we do not attempt to perform garbage
	61	* collection if we do not have enough free space. Rather, we do the
	62	* bare minimum check and give up immediately if the available space
	63	* is below EFI_MIN_RESERVE.
	64	*
	65	* This function is intended to be small and simple because it is
	66	* invoked from crash handler paths.
	67	*/
	68	static efi_status_t
	69	query_variable_store_nonblocking(u32 attributes, unsigned long size)
	70	{
	71	efi_status_t status;
	72	u64 storage_size, remaining_size, max_size;
	73
	74	status = efi.query_variable_info_nonblocking(attributes, &storage_size,
	75	&remaining_size,
	76	&max_size);
	77	if (status != EFI_SUCCESS)
	78	return status;
	79
	80	if (remaining_size - size < EFI_MIN_RESERVE)
	81	return EFI_OUT_OF_RESOURCES;
	82
	83	return EFI_SUCCESS;
	84	}
	85
eeb9db09 ST	86	/*
	87	* Some firmware implementations refuse to boot if there's insufficient space
	88	* in the variable store. Ensure that we never use more than a safe limit.
	89	*
	90	* Return EFI_SUCCESS if it is safe to write 'size' bytes to the variable
	91	* store.
	92	*/
ca0e30dc AB	93	efi_status_t efi_query_variable_store(u32 attributes, unsigned long size,
ca0e30dc AB	94	bool nonblocking)
eeb9db09 ST	95	{
	96	efi_status_t status;
	97	u64 storage_size, remaining_size, max_size;
	98
	99	if (!(attributes & EFI_VARIABLE_NON_VOLATILE))
	100	return 0;
	101
ca0e30dc AB	102	if (nonblocking)
	103	return query_variable_store_nonblocking(attributes, size);
	104
eeb9db09 ST	105	status = efi.query_variable_info(attributes, &storage_size,
	106	&remaining_size, &max_size);
	107	if (status != EFI_SUCCESS)
	108	return status;
	109
	110	/*
	111	* We account for that by refusing the write if permitting it would
	112	* reduce the available space to under 5KB. This figure was provided by
	113	* Samsung, so should be safe.
	114	*/
	115	if ((remaining_size - size < EFI_MIN_RESERVE) &&
	116	!efi_no_storage_paranoia) {
	117
	118	/*
	119	* Triggering garbage collection may require that the firmware
	120	* generate a real EFI_OUT_OF_RESOURCES error. We can force
	121	* that by attempting to use more space than is available.
	122	*/
	123	unsigned long dummy_size = remaining_size + 1024;
	124	void *dummy = kzalloc(dummy_size, GFP_ATOMIC);
	125
	126	if (!dummy)
	127	return EFI_OUT_OF_RESOURCES;
	128
	129	status = efi.set_variable(efi_dummy_name, &EFI_DUMMY_GUID,
	130	EFI_VARIABLE_NON_VOLATILE \|
	131	EFI_VARIABLE_BOOTSERVICE_ACCESS \|
	132	EFI_VARIABLE_RUNTIME_ACCESS,
	133	dummy_size, dummy);
	134
	135	if (status == EFI_SUCCESS) {
	136	/*
	137	* This should have failed, so if it didn't make sure
	138	* that we delete it...
	139	*/
	140	efi_delete_dummy_variable();
	141	}
	142
	143	kfree(dummy);
	144
	145	/*
	146	* The runtime code may now have triggered a garbage collection
	147	* run, so check the variable info again
	148	*/
	149	status = efi.query_variable_info(attributes, &storage_size,
	150	&remaining_size, &max_size);
	151
	152	if (status != EFI_SUCCESS)
	153	return status;
	154
	155	/*
	156	* There still isn't enough room, so return an error
	157	*/
	158	if (remaining_size - size < EFI_MIN_RESERVE)
	159	return EFI_OUT_OF_RESOURCES;
	160	}
	161
	162	return EFI_SUCCESS;
	163	}
	164	EXPORT_SYMBOL_GPL(efi_query_variable_store);
	165
816e7612 MF	166	/*
	167	* The UEFI specification makes it clear that the operating system is
	168	* free to do whatever it wants with boot services code after
	169	* ExitBootServices() has been called. Ignoring this recommendation a
	170	* significant bunch of EFI implementations continue calling into boot
	171	* services code (SetVirtualAddressMap). In order to work around such
	172	* buggy implementations we reserve boot services region during EFI
	173	* init and make sure it stays executable. Then, after
	174	* SetVirtualAddressMap(), it is discarded.
	175	*
	176	* However, some boot services regions contain data that is required
	177	* by drivers, so we need to track which memory ranges can never be
	178	* freed. This is done by tagging those regions with the
	179	* EFI_MEMORY_RUNTIME attribute.
	180	*
	181	* Any driver that wants to mark a region as reserved must use
	182	* efi_mem_reserve() which will insert a new EFI memory descriptor
	183	* into efi.memmap (splitting existing regions if necessary) and tag
	184	* it with EFI_MEMORY_RUNTIME.
	185	*/
	186	void __init efi_arch_mem_reserve(phys_addr_t addr, u64 size)
	187	{
	188	phys_addr_t new_phys, new_size;
	189	struct efi_mem_range mr;
	190	efi_memory_desc_t md;
	191	int num_entries;
	192	void *new;
	193
	194	if (efi_mem_desc_lookup(addr, &md)) {
	195	pr_err("Failed to lookup EFI memory descriptor for %pa\n", &addr);
	196	return;
	197	}
	198
	199	if (addr + size > md.phys_addr + (md.num_pages << EFI_PAGE_SHIFT)) {
	200	pr_err("Region spans EFI memory descriptors, %pa\n", &addr);
	201	return;
	202	}
	203
6f6266a5 OS	204	/* No need to reserve regions that will never be freed. */
	205	if (md.attribute & EFI_MEMORY_RUNTIME)
	206	return;
	207
92dc3350 MF	208	size += addr % EFI_PAGE_SIZE;
	209	size = round_up(size, EFI_PAGE_SIZE);
	210	addr = round_down(addr, EFI_PAGE_SIZE);
	211
816e7612	212	mr.range.start = addr;
92dc3350	213	mr.range.end = addr + size - 1;
816e7612 MF	214	mr.attribute = md.attribute \| EFI_MEMORY_RUNTIME;
	215
	216	num_entries = efi_memmap_split_count(&md, &mr.range);
	217	num_entries += efi.memmap.nr_map;
	218
	219	new_size = efi.memmap.desc_size * num_entries;
	220
20b1e22d	221	new_phys = efi_memmap_alloc(num_entries);
816e7612 MF	222	if (!new_phys) {
	223	pr_err("Could not allocate boot services memmap\n");
	224	return;
	225	}
	226
	227	new = early_memremap(new_phys, new_size);
	228	if (!new) {
	229	pr_err("Failed to map new boot services memmap\n");
	230	return;
	231	}
	232
	233	efi_memmap_insert(&efi.memmap, new, &mr);
	234	early_memunmap(new, new_size);
	235
	236	efi_memmap_install(new_phys, num_entries);
	237	}
	238
452308de MF	239	/*
	240	* Helper function for efi_reserve_boot_services() to figure out if we
	241	* can free regions in efi_free_boot_services().
	242	*
	243	* Use this function to ensure we do not free regions owned by somebody
	244	* else. We must only reserve (and then free) regions:
	245	*
	246	* - Not within any part of the kernel
	247	* - Not the BIOS reserved area (E820_RESERVED, E820_NVS, etc)
	248	*/
	249	static bool can_free_region(u64 start, u64 size)
	250	{
	251	if (start + size > __pa_symbol(_text) && start <= __pa_symbol(_end))
	252	return false;
	253
	254	if (!e820_all_mapped(start, start+size, E820_RAM))
	255	return false;
	256
	257	return true;
	258	}
	259
eeb9db09 ST	260	void __init efi_reserve_boot_services(void)
eeb9db09 ST	261	{
78ce248f	262	efi_memory_desc_t *md;
eeb9db09	263
78ce248f	264	for_each_efi_memory_desc(md) {
eeb9db09 ST	265	u64 start = md->phys_addr;
eeb9db09 ST	266	u64 size = md->num_pages << EFI_PAGE_SHIFT;
452308de	267	bool already_reserved;
eeb9db09 ST	268
	269	if (md->type != EFI_BOOT_SERVICES_CODE &&
	270	md->type != EFI_BOOT_SERVICES_DATA)
	271	continue;
452308de MF	272
	273	already_reserved = memblock_is_region_reserved(start, size);
	274
	275	/*
	276	* Because the following memblock_reserve() is paired
	277	* with free_bootmem_late() for this region in
	278	* efi_free_boot_services(), we must be extremely
	279	* careful not to reserve, and subsequently free,
	280	* critical regions of memory (like the kernel image) or
	281	* those regions that somebody else has already
	282	* reserved.
	283	*
	284	* A good example of a critical region that must not be
	285	* freed is page zero (first 4Kb of memory), which may
	286	* contain boot services code/data but is marked
	287	* E820_RESERVED by trim_bios_range().
	288	*/
	289	if (!already_reserved) {
eeb9db09	290	memblock_reserve(start, size);
452308de MF	291
	292	/*
	293	* If we are the first to reserve the region, no
	294	* one else cares about it. We own it and can
	295	* free it later.
	296	*/
	297	if (can_free_region(start, size))
	298	continue;
	299	}
	300
	301	/*
	302	* We don't own the region. We must not free it.
	303	*
	304	* Setting this bit for a boot services region really
	305	* doesn't make sense as far as the firmware is
	306	* concerned, but it does provide us with a way to tag
	307	* those regions that must not be paired with
	308	* free_bootmem_late().
	309	*/
	310	md->attribute \|= EFI_MEMORY_RUNTIME;
eeb9db09 ST	311	}
	312	}
	313
	314	void __init efi_free_boot_services(void)
	315	{
816e7612	316	phys_addr_t new_phys, new_size;
78ce248f	317	efi_memory_desc_t *md;
816e7612 MF	318	int num_entries = 0;
816e7612 MF	319	void new, new_md;
eeb9db09	320
78ce248f	321	for_each_efi_memory_desc(md) {
eeb9db09 ST	322	unsigned long long start = md->phys_addr;
eeb9db09 ST	323	unsigned long long size = md->num_pages << EFI_PAGE_SHIFT;
5bc653b7	324	size_t rm_size;
eeb9db09 ST	325
eeb9db09 ST	326	if (md->type != EFI_BOOT_SERVICES_CODE &&
816e7612 MF	327	md->type != EFI_BOOT_SERVICES_DATA) {
816e7612 MF	328	num_entries++;
eeb9db09	329	continue;
816e7612	330	}
eeb9db09	331
452308de	332	/* Do not free, someone else owns it: */
816e7612 MF	333	if (md->attribute & EFI_MEMORY_RUNTIME) {
816e7612 MF	334	num_entries++;
eeb9db09	335	continue;
816e7612	336	}
eeb9db09	337
5bc653b7 AL	338	/*
	339	* Nasty quirk: if all sub-1MB memory is used for boot
	340	* services, we can get here without having allocated the
	341	* real mode trampoline. It's too late to hand boot services
	342	* memory back to the memblock allocator, so instead
	343	* try to manually allocate the trampoline if needed.
	344	*
	345	* I've seen this on a Dell XPS 13 9350 with firmware
	346	* 1.4.4 with SGX enabled booting Linux via Fedora 24's
	347	* grub2-efi on a hard disk. (And no, I don't know why
	348	* this happened, but Linux should still try to boot rather
	349	* panicing early.)
	350	*/
	351	rm_size = real_mode_size_needed();
	352	if (rm_size && (start + rm_size) < (1<<20) && size >= rm_size) {
	353	set_real_mode_mem(start, rm_size);
	354	start += rm_size;
	355	size -= rm_size;
	356	}
	357
eeb9db09 ST	358	free_bootmem_late(start, size);
eeb9db09 ST	359	}
816e7612 MF	360
816e7612 MF	361	new_size = efi.memmap.desc_size * num_entries;
20b1e22d	362	new_phys = efi_memmap_alloc(num_entries);
816e7612 MF	363	if (!new_phys) {
	364	pr_err("Failed to allocate new EFI memmap\n");
	365	return;
	366	}
	367
	368	new = memremap(new_phys, new_size, MEMREMAP_WB);
	369	if (!new) {
	370	pr_err("Failed to map new EFI memmap\n");
	371	return;
	372	}
	373
	374	/*
	375	* Build a new EFI memmap that excludes any boot services
	376	* regions that are not tagged EFI_MEMORY_RUNTIME, since those
	377	* regions have now been freed.
	378	*/
	379	new_md = new;
	380	for_each_efi_memory_desc(md) {
	381	if (!(md->attribute & EFI_MEMORY_RUNTIME) &&
	382	(md->type == EFI_BOOT_SERVICES_CODE \|\|
	383	md->type == EFI_BOOT_SERVICES_DATA))
	384	continue;
	385
	386	memcpy(new_md, md, efi.memmap.desc_size);
	387	new_md += efi.memmap.desc_size;
	388	}
	389
	390	memunmap(new);
	391
	392	if (efi_memmap_install(new_phys, num_entries)) {
	393	pr_err("Could not install new EFI memmap\n");
	394	return;
	395	}
eeb9db09 ST	396	}
	397
	398	/*
	399	* A number of config table entries get remapped to virtual addresses
	400	* after entering EFI virtual mode. However, the kexec kernel requires
	401	* their physical addresses therefore we pass them via setup_data and
	402	* correct those entries to their respective physical addresses here.
	403	*
	404	* Currently only handles smbios which is necessary for some firmware
	405	* implementation.
	406	*/
	407	int __init efi_reuse_config(u64 tables, int nr_tables)
	408	{
	409	int i, sz, ret = 0;
	410	void p, tablep;
	411	struct efi_setup_data *data;
	412
	413	if (!efi_setup)
	414	return 0;
	415
	416	if (!efi_enabled(EFI_64BIT))
	417	return 0;
	418
	419	data = early_memremap(efi_setup, sizeof(*data));
	420	if (!data) {
	421	ret = -ENOMEM;
	422	goto out;
	423	}
	424
	425	if (!data->smbios)
	426	goto out_memremap;
	427
	428	sz = sizeof(efi_config_table_64_t);
	429
	430	p = tablep = early_memremap(tables, nr_tables * sz);
	431	if (!p) {
	432	pr_err("Could not map Configuration table!\n");
	433	ret = -ENOMEM;
	434	goto out_memremap;
	435	}
	436
	437	for (i = 0; i < efi.systab->nr_tables; i++) {
	438	efi_guid_t guid;
	439
	440	guid = ((efi_config_table_64_t *)p)->guid;
	441
	442	if (!efi_guidcmp(guid, SMBIOS_TABLE_GUID))
	443	((efi_config_table_64_t *)p)->table = data->smbios;
	444	p += sz;
	445	}
98a716b6	446	early_memunmap(tablep, nr_tables * sz);
eeb9db09 ST	447
eeb9db09 ST	448	out_memremap:
98a716b6	449	early_memunmap(data, sizeof(*data));
eeb9db09 ST	450	out:
	451	return ret;
	452	}
	453
d394f2d9 AT	454	static const struct dmi_system_id sgi_uv1_dmi[] = {
	455	{ NULL, "SGI UV1",
	456	{ DMI_MATCH(DMI_PRODUCT_NAME, "Stoutland Platform"),
	457	DMI_MATCH(DMI_PRODUCT_VERSION, "1.0"),
	458	DMI_MATCH(DMI_BIOS_VENDOR, "SGI.COM"),
	459	}
	460	},
	461	{ } /* NULL entry stops DMI scanning */
	462	};
	463
eeb9db09 ST	464	void __init efi_apply_memmap_quirks(void)
	465	{
	466	/*
	467	* Once setup is done earlier, unmap the EFI memory map on mismatched
	468	* firmware/kernel architectures since there is no support for runtime
	469	* services.
	470	*/
	471	if (!efi_runtime_supported()) {
26d7f65f	472	pr_info("Setup done, disabling due to 32/64-bit mismatch\n");
9479c7ce	473	efi_memmap_unmap();
eeb9db09 ST	474	}
eeb9db09 ST	475
d394f2d9 AT	476	/* UV2+ BIOS has a fix for this issue. UV1 still needs the quirk. */
d394f2d9 AT	477	if (dmi_check_system(sgi_uv1_dmi))
eeb9db09 ST	478	set_bit(EFI_OLD_MEMMAP, &efi.flags);
eeb9db09 ST	479	}
44be28e9 MF	480
	481	/*
	482	* For most modern platforms the preferred method of powering off is via
	483	* ACPI. However, there are some that are known to require the use of
	484	* EFI runtime services and for which ACPI does not work at all.
	485	*
	486	* Using EFI is a last resort, to be used only if no other option
	487	* exists.
	488	*/
	489	bool efi_reboot_required(void)
	490	{
	491	if (!acpi_gbl_reduced_hardware)
	492	return false;
	493
	494	efi_reboot_quirk_mode = EFI_RESET_WARM;
	495	return true;
	496	}
	497
	498	bool efi_poweroff_required(void)
	499	{
13737181	500	return acpi_gbl_reduced_hardware \|\| acpi_no_s5;
44be28e9	501	}