[deliverable/linux.git] / arch / x86 / mm / init.c

#include <linux/ioport.h>
#include <linux/swap.h>

#include <asm/cacheflush.h>
#include <asm/e820.h>
#include <asm/init.h>
#include <asm/page.h>
#include <asm/page_types.h>
#include <asm/sections.h>
#include <asm/system.h>
#include <asm/tlbflush.h>

unsigned long __initdata e820_table_start;
unsigned long __meminitdata e820_table_end;
unsigned long __meminitdata e820_table_top;

int after_bootmem;

int direct_gbpages
#ifdef CONFIG_DIRECT_GBPAGES
				= 1
#endif
;

int nx_enabled;

#if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
static int disable_nx __cpuinitdata;

/*
 * noexec = on|off
 *
 * Control non-executable mappings for processes.
 *
 * on      Enable
 * off     Disable
 */
static int __init noexec_setup(char *str)
{
	if (!str)
		return -EINVAL;
	if (!strncmp(str, "on", 2)) {
		__supported_pte_mask |= _PAGE_NX;
		disable_nx = 0;
	} else if (!strncmp(str, "off", 3)) {
		disable_nx = 1;
		__supported_pte_mask &= ~_PAGE_NX;
	}
	return 0;
}
early_param("noexec", noexec_setup);
#endif

#ifdef CONFIG_X86_PAE
static void __init set_nx(void)
{
	unsigned int v[4], l, h;

	if (cpu_has_pae && (cpuid_eax(0x80000000) > 0x80000001)) {
		cpuid(0x80000001, &v[0], &v[1], &v[2], &v[3]);

		if ((v[3] & (1 << 20)) && !disable_nx) {
			rdmsr(MSR_EFER, l, h);
			l |= EFER_NX;
			wrmsr(MSR_EFER, l, h);
			nx_enabled = 1;
			__supported_pte_mask |= _PAGE_NX;
		}
	}
}
#else
static inline void set_nx(void)
{
}
#endif

#ifdef CONFIG_X86_64
void __cpuinit check_efer(void)
{
	unsigned long efer;

	rdmsrl(MSR_EFER, efer);
	if (!(efer & EFER_NX) || disable_nx)
		__supported_pte_mask &= ~_PAGE_NX;
}
#endif

static void __init find_early_table_space(unsigned long end, int use_pse,
					  int use_gbpages)
{
	unsigned long puds, pmds, ptes, tables, start;

	puds = (end + PUD_SIZE - 1) >> PUD_SHIFT;
	tables = roundup(puds * sizeof(pud_t), PAGE_SIZE);

	if (use_gbpages) {
		unsigned long extra;

		extra = end - ((end>>PUD_SHIFT) << PUD_SHIFT);
		pmds = (extra + PMD_SIZE - 1) >> PMD_SHIFT;
	} else
		pmds = (end + PMD_SIZE - 1) >> PMD_SHIFT;

	tables += roundup(pmds * sizeof(pmd_t), PAGE_SIZE);

	if (use_pse) {
		unsigned long extra;

		extra = end - ((end>>PMD_SHIFT) << PMD_SHIFT);
#ifdef CONFIG_X86_32
		extra += PMD_SIZE;
#endif
		ptes = (extra + PAGE_SIZE - 1) >> PAGE_SHIFT;
	} else
		ptes = (end + PAGE_SIZE - 1) >> PAGE_SHIFT;

	tables += roundup(ptes * sizeof(pte_t), PAGE_SIZE);

#ifdef CONFIG_X86_32
	/* for fixmap */
	tables += roundup(__end_of_fixed_addresses * sizeof(pte_t), PAGE_SIZE);
#endif

	/*
	 * RED-PEN putting page tables only on node 0 could
	 * cause a hotspot and fill up ZONE_DMA. The page tables
	 * need roughly 0.5KB per GB.
	 */
#ifdef CONFIG_X86_32
	start = 0x7000;
	e820_table_start = find_e820_area(start, max_pfn_mapped<<PAGE_SHIFT,
					tables, PAGE_SIZE);
#else /* CONFIG_X86_64 */
	start = 0x8000;
	e820_table_start = find_e820_area(start, end, tables, PAGE_SIZE);
#endif
	if (e820_table_start == -1UL)
		panic("Cannot find space for the kernel page tables");

	e820_table_start >>= PAGE_SHIFT;
	e820_table_end = e820_table_start;
	e820_table_top = e820_table_start + (tables >> PAGE_SHIFT);

	printk(KERN_DEBUG "kernel direct mapping tables up to %lx @ %lx-%lx\n",
		end, e820_table_start << PAGE_SHIFT, e820_table_top << PAGE_SHIFT);
}

struct map_range {
	unsigned long start;
	unsigned long end;
	unsigned page_size_mask;
};

#ifdef CONFIG_X86_32
#define NR_RANGE_MR 3
#else /* CONFIG_X86_64 */
#define NR_RANGE_MR 5
#endif

static int __meminit save_mr(struct map_range *mr, int nr_range,
			     unsigned long start_pfn, unsigned long end_pfn,
			     unsigned long page_size_mask)
{
	if (start_pfn < end_pfn) {
		if (nr_range >= NR_RANGE_MR)
			panic("run out of range for init_memory_mapping\n");
		mr[nr_range].start = start_pfn<<PAGE_SHIFT;
		mr[nr_range].end   = end_pfn<<PAGE_SHIFT;
		mr[nr_range].page_size_mask = page_size_mask;
		nr_range++;
	}

	return nr_range;
}

#ifdef CONFIG_X86_64
static void __init init_gbpages(void)
{
	if (direct_gbpages && cpu_has_gbpages)
		printk(KERN_INFO "Using GB pages for direct mapping\n");
	else
		direct_gbpages = 0;
}
#else
static inline void init_gbpages(void)
{
}
#endif

/*
 * Setup the direct mapping of the physical memory at PAGE_OFFSET.
 * This runs before bootmem is initialized and gets pages directly from
 * the physical memory. To access them they are temporarily mapped.
 */
unsigned long __init_refok init_memory_mapping(unsigned long start,
					       unsigned long end)
{
	unsigned long page_size_mask = 0;
	unsigned long start_pfn, end_pfn;
	unsigned long ret = 0;
	unsigned long pos;

	struct map_range mr[NR_RANGE_MR];
	int nr_range, i;
	int use_pse, use_gbpages;

	printk(KERN_INFO "init_memory_mapping: %016lx-%016lx\n", start, end);

	if (!after_bootmem)
		init_gbpages();

#ifdef CONFIG_DEBUG_PAGEALLOC
	/*
	 * For CONFIG_DEBUG_PAGEALLOC, identity mapping will use small pages.
	 * This will simplify cpa(), which otherwise needs to support splitting
	 * large pages into small in interrupt context, etc.
	 */
	use_pse = use_gbpages = 0;
#else
	use_pse = cpu_has_pse;
	use_gbpages = direct_gbpages;
#endif

	set_nx();
	if (nx_enabled)
		printk(KERN_INFO "NX (Execute Disable) protection: active\n");

	/* Enable PSE if available */
	if (cpu_has_pse)
		set_in_cr4(X86_CR4_PSE);

	/* Enable PGE if available */
	if (cpu_has_pge) {
		set_in_cr4(X86_CR4_PGE);
		__supported_pte_mask |= _PAGE_GLOBAL;
	}

	if (use_gbpages)
		page_size_mask |= 1 << PG_LEVEL_1G;
	if (use_pse)
		page_size_mask |= 1 << PG_LEVEL_2M;

	memset(mr, 0, sizeof(mr));
	nr_range = 0;

	/* head if not big page alignment ? */
	start_pfn = start >> PAGE_SHIFT;
	pos = start_pfn << PAGE_SHIFT;
#ifdef CONFIG_X86_32
	/*
	 * Don't use a large page for the first 2/4MB of memory
	 * because there are often fixed size MTRRs in there
	 * and overlapping MTRRs into large pages can cause
	 * slowdowns.
	 */
	if (pos == 0)
		end_pfn = 1<<(PMD_SHIFT - PAGE_SHIFT);
	else
		end_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
				 << (PMD_SHIFT - PAGE_SHIFT);
#else /* CONFIG_X86_64 */
	end_pfn = ((pos + (PMD_SIZE - 1)) >> PMD_SHIFT)
			<< (PMD_SHIFT - PAGE_SHIFT);
#endif
	if (end_pfn > (end >> PAGE_SHIFT))
		end_pfn = end >> PAGE_SHIFT;
	if (start_pfn < end_pfn) {
		nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);
		pos = end_pfn << PAGE_SHIFT;
	}

	/* big page (2M) range */
	start_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
			 << (PMD_SHIFT - PAGE_SHIFT);
#ifdef CONFIG_X86_32
	end_pfn = (end>>PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT);
#else /* CONFIG_X86_64 */
	end_pfn = ((pos + (PUD_SIZE - 1))>>PUD_SHIFT)
			 << (PUD_SHIFT - PAGE_SHIFT);
	if (end_pfn > ((end>>PMD_SHIFT)<<(PMD_SHIFT - PAGE_SHIFT)))
		end_pfn = ((end>>PMD_SHIFT)<<(PMD_SHIFT - PAGE_SHIFT));
#endif

	if (start_pfn < end_pfn) {
		nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
				page_size_mask & (1<<PG_LEVEL_2M));
		pos = end_pfn << PAGE_SHIFT;
	}

#ifdef CONFIG_X86_64
	/* big page (1G) range */
	start_pfn = ((pos + (PUD_SIZE - 1))>>PUD_SHIFT)
			 << (PUD_SHIFT - PAGE_SHIFT);
	end_pfn = (end >> PUD_SHIFT) << (PUD_SHIFT - PAGE_SHIFT);
	if (start_pfn < end_pfn) {
		nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
				page_size_mask &
				 ((1<<PG_LEVEL_2M)|(1<<PG_LEVEL_1G)));
		pos = end_pfn << PAGE_SHIFT;
	}

	/* tail is not big page (1G) alignment */
	start_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
			 << (PMD_SHIFT - PAGE_SHIFT);
	end_pfn = (end >> PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT);
	if (start_pfn < end_pfn) {
		nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
				page_size_mask & (1<<PG_LEVEL_2M));
		pos = end_pfn << PAGE_SHIFT;
	}
#endif

	/* tail is not big page (2M) alignment */
	start_pfn = pos>>PAGE_SHIFT;
	end_pfn = end>>PAGE_SHIFT;
	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);

	/* try to merge same page size and continuous */
	for (i = 0; nr_range > 1 && i < nr_range - 1; i++) {
		unsigned long old_start;
		if (mr[i].end != mr[i+1].start ||
		    mr[i].page_size_mask != mr[i+1].page_size_mask)
			continue;
		/* move it */
		old_start = mr[i].start;
		memmove(&mr[i], &mr[i+1],
			(nr_range - 1 - i) * sizeof(struct map_range));
		mr[i--].start = old_start;
		nr_range--;
	}

	for (i = 0; i < nr_range; i++)
		printk(KERN_DEBUG " %010lx - %010lx page %s\n",
				mr[i].start, mr[i].end,
			(mr[i].page_size_mask & (1<<PG_LEVEL_1G))?"1G":(
			 (mr[i].page_size_mask & (1<<PG_LEVEL_2M))?"2M":"4k"));

	/*
	 * Find space for the kernel direct mapping tables.
	 *
	 * Later we should allocate these tables in the local node of the
	 * memory mapped. Unfortunately this is done currently before the
	 * nodes are discovered.
	 */
	if (!after_bootmem)
		find_early_table_space(end, use_pse, use_gbpages);

#ifdef CONFIG_X86_32
	for (i = 0; i < nr_range; i++)
		kernel_physical_mapping_init(mr[i].start, mr[i].end,
					     mr[i].page_size_mask);
	ret = end;
#else /* CONFIG_X86_64 */
	for (i = 0; i < nr_range; i++)
		ret = kernel_physical_mapping_init(mr[i].start, mr[i].end,
						   mr[i].page_size_mask);
#endif

#ifdef CONFIG_X86_32
	early_ioremap_page_table_range_init();

	load_cr3(swapper_pg_dir);
#endif

#ifdef CONFIG_X86_64
	if (!after_bootmem)
		mmu_cr4_features = read_cr4();
#endif
	__flush_tlb_all();

	if (!after_bootmem && e820_table_end > e820_table_start)
		reserve_early(e820_table_start << PAGE_SHIFT,
				 e820_table_end << PAGE_SHIFT, "PGTABLE");

	if (!after_bootmem)
		early_memtest(start, end);

	return ret >> PAGE_SHIFT;
}


/*
 * devmem_is_allowed() checks to see if /dev/mem access to a certain address
 * is valid. The argument is a physical page number.
 *
 *
 * On x86, access has to be given to the first megabyte of ram because that area
 * contains bios code and data regions used by X and dosemu and similar apps.
 * Access has to be given to non-kernel-ram areas as well, these contain the PCI
 * mmio resources as well as potential bios/acpi data regions.
 */
int devmem_is_allowed(unsigned long pagenr)
{
	if (pagenr <= 256)
		return 1;
	if (iomem_is_exclusive(pagenr << PAGE_SHIFT))
		return 0;
	if (!page_is_ram(pagenr))
		return 1;
	return 0;
}

void free_init_pages(char *what, unsigned long begin, unsigned long end)
{
	unsigned long addr = begin;

	if (addr >= end)
		return;

	/*
	 * If debugging page accesses then do not free this memory but
	 * mark them not present - any buggy init-section access will
	 * create a kernel page fault:
	 */
#ifdef CONFIG_DEBUG_PAGEALLOC
	printk(KERN_INFO "debug: unmapping init memory %08lx..%08lx\n",
		begin, PAGE_ALIGN(end));
	set_memory_np(begin, (end - begin) >> PAGE_SHIFT);
#else
	/*
	 * We just marked the kernel text read only above, now that
	 * we are going to free part of that, we need to make that
	 * writeable first.
	 */
	set_memory_rw(begin, (end - begin) >> PAGE_SHIFT);

	printk(KERN_INFO "Freeing %s: %luk freed\n", what, (end - begin) >> 10);

	for (; addr < end; addr += PAGE_SIZE) {
		ClearPageReserved(virt_to_page(addr));
		init_page_count(virt_to_page(addr));
		memset((void *)(addr & ~(PAGE_SIZE-1)),
			POISON_FREE_INITMEM, PAGE_SIZE);
		free_page(addr);
		totalram_pages++;
	}
#endif
}

void free_initmem(void)
{
	free_init_pages("unused kernel memory",
			(unsigned long)(&__init_begin),
			(unsigned long)(&__init_end));
}

#ifdef CONFIG_BLK_DEV_INITRD
void free_initrd_mem(unsigned long start, unsigned long end)
{
	free_init_pages("initrd memory", start, end);
}
#endif
Commit	Line	Data
540aca06	1	#include <linux/ioport.h>
e5b2bb55	2	#include <linux/swap.h>
540aca06	3
e5b2bb55	4	#include <asm/cacheflush.h>
f765090a	5	#include <asm/e820.h>
4fcb2083	6	#include <asm/init.h>
e5b2bb55	7	#include <asm/page.h>
540aca06	8	#include <asm/page_types.h>
e5b2bb55 PE	9	#include <asm/sections.h>
e5b2bb55 PE	10	#include <asm/system.h>
f765090a PE	11	#include <asm/tlbflush.h>
f765090a PE	12
298af9d8 PE	13	unsigned long __initdata e820_table_start;
	14	unsigned long __meminitdata e820_table_end;
	15	unsigned long __meminitdata e820_table_top;
f765090a PE	16
	17	int after_bootmem;
	18
	19	int direct_gbpages
	20	#ifdef CONFIG_DIRECT_GBPAGES
	21	= 1
	22	#endif
	23	;
	24
89388913 PE	25	int nx_enabled;
	26
	27	#if defined(CONFIG_X86_64) \|\| defined(CONFIG_X86_PAE)
	28	static int disable_nx __cpuinitdata;
	29
	30	/*
	31	* noexec = on\|off
	32	*
	33	* Control non-executable mappings for processes.
	34	*
	35	* on Enable
	36	* off Disable
	37	*/
	38	static int __init noexec_setup(char *str)
	39	{
	40	if (!str)
	41	return -EINVAL;
	42	if (!strncmp(str, "on", 2)) {
	43	__supported_pte_mask \|= _PAGE_NX;
	44	disable_nx = 0;
	45	} else if (!strncmp(str, "off", 3)) {
	46	disable_nx = 1;
	47	__supported_pte_mask &= ~_PAGE_NX;
	48	}
	49	return 0;
	50	}
	51	early_param("noexec", noexec_setup);
	52	#endif
	53
	54	#ifdef CONFIG_X86_PAE
	55	static void __init set_nx(void)
	56	{
	57	unsigned int v[4], l, h;
	58
	59	if (cpu_has_pae && (cpuid_eax(0x80000000) > 0x80000001)) {
	60	cpuid(0x80000001, &v[0], &v[1], &v[2], &v[3]);
	61
	62	if ((v[3] & (1 << 20)) && !disable_nx) {
	63	rdmsr(MSR_EFER, l, h);
	64	l \|= EFER_NX;
	65	wrmsr(MSR_EFER, l, h);
	66	nx_enabled = 1;
	67	__supported_pte_mask \|= _PAGE_NX;
	68	}
	69	}
	70	}
	71	#else
	72	static inline void set_nx(void)
	73	{
	74	}
	75	#endif
	76
	77	#ifdef CONFIG_X86_64
	78	void __cpuinit check_efer(void)
	79	{
	80	unsigned long efer;
	81
	82	rdmsrl(MSR_EFER, efer);
	83	if (!(efer & EFER_NX) \|\| disable_nx)
	84	__supported_pte_mask &= ~_PAGE_NX;
	85	}
	86	#endif
	87
f765090a PE	88	static void __init find_early_table_space(unsigned long end, int use_pse,
	89	int use_gbpages)
	90	{
	91	unsigned long puds, pmds, ptes, tables, start;
	92
	93	puds = (end + PUD_SIZE - 1) >> PUD_SHIFT;
	94	tables = roundup(puds * sizeof(pud_t), PAGE_SIZE);
	95
	96	if (use_gbpages) {
	97	unsigned long extra;
	98
	99	extra = end - ((end>>PUD_SHIFT) << PUD_SHIFT);
	100	pmds = (extra + PMD_SIZE - 1) >> PMD_SHIFT;
	101	} else
	102	pmds = (end + PMD_SIZE - 1) >> PMD_SHIFT;
	103
	104	tables += roundup(pmds * sizeof(pmd_t), PAGE_SIZE);
	105
	106	if (use_pse) {
	107	unsigned long extra;
	108
	109	extra = end - ((end>>PMD_SHIFT) << PMD_SHIFT);
	110	#ifdef CONFIG_X86_32
	111	extra += PMD_SIZE;
	112	#endif
	113	ptes = (extra + PAGE_SIZE - 1) >> PAGE_SHIFT;
	114	} else
	115	ptes = (end + PAGE_SIZE - 1) >> PAGE_SHIFT;
	116
	117	tables += roundup(ptes * sizeof(pte_t), PAGE_SIZE);
	118
	119	#ifdef CONFIG_X86_32
	120	/* for fixmap */
	121	tables += roundup(__end_of_fixed_addresses * sizeof(pte_t), PAGE_SIZE);
	122	#endif
	123
	124	/*
	125	* RED-PEN putting page tables only on node 0 could
	126	* cause a hotspot and fill up ZONE_DMA. The page tables
	127	* need roughly 0.5KB per GB.
	128	*/
	129	#ifdef CONFIG_X86_32
	130	start = 0x7000;
298af9d8	131	e820_table_start = find_e820_area(start, max_pfn_mapped<<PAGE_SHIFT,
f765090a PE	132	tables, PAGE_SIZE);
	133	#else /* CONFIG_X86_64 */
	134	start = 0x8000;
298af9d8	135	e820_table_start = find_e820_area(start, end, tables, PAGE_SIZE);
f765090a	136	#endif
298af9d8	137	if (e820_table_start == -1UL)
f765090a PE	138	panic("Cannot find space for the kernel page tables");
f765090a PE	139
298af9d8 PE	140	e820_table_start >>= PAGE_SHIFT;
	141	e820_table_end = e820_table_start;
	142	e820_table_top = e820_table_start + (tables >> PAGE_SHIFT);
f765090a PE	143
f765090a PE	144	printk(KERN_DEBUG "kernel direct mapping tables up to %lx @ %lx-%lx\n",
298af9d8	145	end, e820_table_start << PAGE_SHIFT, e820_table_top << PAGE_SHIFT);
f765090a PE	146	}
	147
	148	struct map_range {
	149	unsigned long start;
	150	unsigned long end;
	151	unsigned page_size_mask;
	152	};
	153
	154	#ifdef CONFIG_X86_32
	155	#define NR_RANGE_MR 3
	156	#else /* CONFIG_X86_64 */
	157	#define NR_RANGE_MR 5
	158	#endif
	159
dc9dd5cc JB	160	static int __meminit save_mr(struct map_range *mr, int nr_range,
	161	unsigned long start_pfn, unsigned long end_pfn,
	162	unsigned long page_size_mask)
f765090a PE	163	{
	164	if (start_pfn < end_pfn) {
	165	if (nr_range >= NR_RANGE_MR)
	166	panic("run out of range for init_memory_mapping\n");
	167	mr[nr_range].start = start_pfn<<PAGE_SHIFT;
	168	mr[nr_range].end = end_pfn<<PAGE_SHIFT;
	169	mr[nr_range].page_size_mask = page_size_mask;
	170	nr_range++;
	171	}
	172
	173	return nr_range;
	174	}
	175
	176	#ifdef CONFIG_X86_64
	177	static void __init init_gbpages(void)
	178	{
	179	if (direct_gbpages && cpu_has_gbpages)
	180	printk(KERN_INFO "Using GB pages for direct mapping\n");
	181	else
	182	direct_gbpages = 0;
	183	}
	184	#else
	185	static inline void init_gbpages(void)
	186	{
	187	}
	188	#endif
	189
	190	/*
	191	* Setup the direct mapping of the physical memory at PAGE_OFFSET.
	192	* This runs before bootmem is initialized and gets pages directly from
	193	* the physical memory. To access them they are temporarily mapped.
	194	*/
	195	unsigned long __init_refok init_memory_mapping(unsigned long start,
	196	unsigned long end)
	197	{
	198	unsigned long page_size_mask = 0;
	199	unsigned long start_pfn, end_pfn;
c77a3b59	200	unsigned long ret = 0;
f765090a	201	unsigned long pos;
f765090a PE	202
	203	struct map_range mr[NR_RANGE_MR];
	204	int nr_range, i;
	205	int use_pse, use_gbpages;
	206
	207	printk(KERN_INFO "init_memory_mapping: %016lx-%016lx\n", start, end);
	208
	209	if (!after_bootmem)
	210	init_gbpages();
	211
	212	#ifdef CONFIG_DEBUG_PAGEALLOC
	213	/*
	214	* For CONFIG_DEBUG_PAGEALLOC, identity mapping will use small pages.
	215	* This will simplify cpa(), which otherwise needs to support splitting
	216	* large pages into small in interrupt context, etc.
	217	*/
	218	use_pse = use_gbpages = 0;
	219	#else
	220	use_pse = cpu_has_pse;
	221	use_gbpages = direct_gbpages;
	222	#endif
	223
f765090a PE	224	set_nx();
	225	if (nx_enabled)
	226	printk(KERN_INFO "NX (Execute Disable) protection: active\n");
f765090a PE	227
	228	/* Enable PSE if available */
	229	if (cpu_has_pse)
	230	set_in_cr4(X86_CR4_PSE);
	231
	232	/* Enable PGE if available */
	233	if (cpu_has_pge) {
	234	set_in_cr4(X86_CR4_PGE);
	235	__supported_pte_mask \|= _PAGE_GLOBAL;
	236	}
f765090a PE	237
	238	if (use_gbpages)
	239	page_size_mask \|= 1 << PG_LEVEL_1G;
	240	if (use_pse)
	241	page_size_mask \|= 1 << PG_LEVEL_2M;
	242
	243	memset(mr, 0, sizeof(mr));
	244	nr_range = 0;
	245
	246	/* head if not big page alignment ? */
	247	start_pfn = start >> PAGE_SHIFT;
	248	pos = start_pfn << PAGE_SHIFT;
	249	#ifdef CONFIG_X86_32
	250	/*
	251	* Don't use a large page for the first 2/4MB of memory
	252	* because there are often fixed size MTRRs in there
	253	* and overlapping MTRRs into large pages can cause
	254	* slowdowns.
	255	*/
	256	if (pos == 0)
	257	end_pfn = 1<<(PMD_SHIFT - PAGE_SHIFT);
	258	else
	259	end_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
	260	<< (PMD_SHIFT - PAGE_SHIFT);
	261	#else /* CONFIG_X86_64 */
	262	end_pfn = ((pos + (PMD_SIZE - 1)) >> PMD_SHIFT)
	263	<< (PMD_SHIFT - PAGE_SHIFT);
	264	#endif
	265	if (end_pfn > (end >> PAGE_SHIFT))
	266	end_pfn = end >> PAGE_SHIFT;
	267	if (start_pfn < end_pfn) {
	268	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);
	269	pos = end_pfn << PAGE_SHIFT;
	270	}
	271
	272	/* big page (2M) range */
	273	start_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
	274	<< (PMD_SHIFT - PAGE_SHIFT);
	275	#ifdef CONFIG_X86_32
	276	end_pfn = (end>>PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT);
	277	#else /* CONFIG_X86_64 */
	278	end_pfn = ((pos + (PUD_SIZE - 1))>>PUD_SHIFT)
	279	<< (PUD_SHIFT - PAGE_SHIFT);
	280	if (end_pfn > ((end>>PMD_SHIFT)<<(PMD_SHIFT - PAGE_SHIFT)))
	281	end_pfn = ((end>>PMD_SHIFT)<<(PMD_SHIFT - PAGE_SHIFT));
	282	#endif
	283
	284	if (start_pfn < end_pfn) {
	285	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
	286	page_size_mask & (1<<PG_LEVEL_2M));
	287	pos = end_pfn << PAGE_SHIFT;
	288	}
	289
	290	#ifdef CONFIG_X86_64
	291	/* big page (1G) range */
	292	start_pfn = ((pos + (PUD_SIZE - 1))>>PUD_SHIFT)
	293	<< (PUD_SHIFT - PAGE_SHIFT);
	294	end_pfn = (end >> PUD_SHIFT) << (PUD_SHIFT - PAGE_SHIFT);
	295	if (start_pfn < end_pfn) {
	296	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
	297	page_size_mask &
	298	((1<<PG_LEVEL_2M)\|(1<<PG_LEVEL_1G)));
	299	pos = end_pfn << PAGE_SHIFT;
	300	}
301
302	/* tail is not big page (1G) alignment */
303	start_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
304	<< (PMD_SHIFT - PAGE_SHIFT);
305	end_pfn = (end >> PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT);
306	if (start_pfn < end_pfn) {
307	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
308	page_size_mask & (1<<PG_LEVEL_2M));
309	pos = end_pfn << PAGE_SHIFT;
310	}
311	#endif
312
313	/* tail is not big page (2M) alignment */
314	start_pfn = pos>>PAGE_SHIFT;
315	end_pfn = end>>PAGE_SHIFT;
316	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);
317
318	/* try to merge same page size and continuous */
319	for (i = 0; nr_range > 1 && i < nr_range - 1; i++) {
320	unsigned long old_start;
321	if (mr[i].end != mr[i+1].start \|\|
322	mr[i].page_size_mask != mr[i+1].page_size_mask)
323	continue;
324	/* move it */
325	old_start = mr[i].start;
326	memmove(&mr[i], &mr[i+1],
327	(nr_range - 1 - i) * sizeof(struct map_range));
328	mr[i--].start = old_start;
329	nr_range--;
330	}
331
332	for (i = 0; i < nr_range; i++)
333	printk(KERN_DEBUG " %010lx - %010lx page %s\n",
334	mr[i].start, mr[i].end,
335	(mr[i].page_size_mask & (1<<PG_LEVEL_1G))?"1G":(
336	(mr[i].page_size_mask & (1<<PG_LEVEL_2M))?"2M":"4k"));
337
338	/*
339	* Find space for the kernel direct mapping tables.
340	*
341	* Later we should allocate these tables in the local node of the
342	* memory mapped. Unfortunately this is done currently before the
343	* nodes are discovered.
344	*/
345	if (!after_bootmem)
346	find_early_table_space(end, use_pse, use_gbpages);
347
348	#ifdef CONFIG_X86_32
349	for (i = 0; i < nr_range; i++)
e53fb04f PE	350	kernel_physical_mapping_init(mr[i].start, mr[i].end,
e53fb04f PE	351	mr[i].page_size_mask);
f765090a PE	352	ret = end;
	353	#else /* CONFIG_X86_64 */
	354	for (i = 0; i < nr_range; i++)
	355	ret = kernel_physical_mapping_init(mr[i].start, mr[i].end,
	356	mr[i].page_size_mask);
	357	#endif
	358
	359	#ifdef CONFIG_X86_32
	360	early_ioremap_page_table_range_init();
	361
	362	load_cr3(swapper_pg_dir);
	363	#endif
	364
	365	#ifdef CONFIG_X86_64
	366	if (!after_bootmem)
	367	mmu_cr4_features = read_cr4();
	368	#endif
	369	__flush_tlb_all();
	370
298af9d8 PE	371	if (!after_bootmem && e820_table_end > e820_table_start)
	372	reserve_early(e820_table_start << PAGE_SHIFT,
	373	e820_table_end << PAGE_SHIFT, "PGTABLE");
f765090a PE	374
	375	if (!after_bootmem)
	376	early_memtest(start, end);
	377
	378	return ret >> PAGE_SHIFT;
	379	}
	380
e5b2bb55	381
540aca06 PE	382	/*
	383	* devmem_is_allowed() checks to see if /dev/mem access to a certain address
	384	* is valid. The argument is a physical page number.
	385	*
	386	*
	387	* On x86, access has to be given to the first megabyte of ram because that area
	388	* contains bios code and data regions used by X and dosemu and similar apps.
	389	* Access has to be given to non-kernel-ram areas as well, these contain the PCI
	390	* mmio resources as well as potential bios/acpi data regions.
	391	*/
	392	int devmem_is_allowed(unsigned long pagenr)
	393	{
	394	if (pagenr <= 256)
	395	return 1;
	396	if (iomem_is_exclusive(pagenr << PAGE_SHIFT))
	397	return 0;
	398	if (!page_is_ram(pagenr))
	399	return 1;
	400	return 0;
	401	}
	402
e5b2bb55 PE	403	void free_init_pages(char *what, unsigned long begin, unsigned long end)
	404	{
	405	unsigned long addr = begin;
	406
	407	if (addr >= end)
	408	return;
	409
	410	/*
	411	* If debugging page accesses then do not free this memory but
	412	* mark them not present - any buggy init-section access will
	413	* create a kernel page fault:
	414	*/
	415	#ifdef CONFIG_DEBUG_PAGEALLOC
	416	printk(KERN_INFO "debug: unmapping init memory %08lx..%08lx\n",
	417	begin, PAGE_ALIGN(end));
	418	set_memory_np(begin, (end - begin) >> PAGE_SHIFT);
	419	#else
	420	/*
	421	* We just marked the kernel text read only above, now that
	422	* we are going to free part of that, we need to make that
	423	* writeable first.
	424	*/
	425	set_memory_rw(begin, (end - begin) >> PAGE_SHIFT);
	426
	427	printk(KERN_INFO "Freeing %s: %luk freed\n", what, (end - begin) >> 10);
	428
	429	for (; addr < end; addr += PAGE_SIZE) {
	430	ClearPageReserved(virt_to_page(addr));
	431	init_page_count(virt_to_page(addr));
	432	memset((void *)(addr & ~(PAGE_SIZE-1)),
	433	POISON_FREE_INITMEM, PAGE_SIZE);
	434	free_page(addr);
	435	totalram_pages++;
	436	}
	437	#endif
	438	}
	439
	440	void free_initmem(void)
	441	{
	442	free_init_pages("unused kernel memory",
	443	(unsigned long)(&__init_begin),
	444	(unsigned long)(&__init_end));
	445	}
731ddea6 PE	446
	447	#ifdef CONFIG_BLK_DEV_INITRD
	448	void free_initrd_mem(unsigned long start, unsigned long end)
	449	{
	450	free_init_pages("initrd memory", start, end);
	451	}
	452	#endif