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

#include <linux/gfp.h>
#include <linux/initrd.h>
#include <linux/ioport.h>
#include <linux/swap.h>
#include <linux/memblock.h>
#include <linux/bootmem.h>	/* for max_low_pfn */

#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/setup.h>
#include <asm/tlbflush.h>
#include <asm/tlb.h>
#include <asm/proto.h>
#include <asm/dma.h>		/* for MAX_DMA_PFN */

unsigned long __initdata pgt_buf_start;
unsigned long __meminitdata pgt_buf_end;
unsigned long __meminitdata pgt_buf_top;

int after_bootmem;

int direct_gbpages
#ifdef CONFIG_DIRECT_GBPAGES
				= 1
#endif
;

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

static int page_size_mask;
/*
 * First calculate space needed for kernel direct mapping page tables to cover
 * mr[0].start to mr[nr_range - 1].end, while accounting for possible 2M and 1GB
 * pages. Then find enough contiguous space for those page tables.
 */
static void __init find_early_table_space(struct map_range *mr, int nr_range)
{
	int i;
	unsigned long puds = 0, pmds = 0, ptes = 0, tables;
	unsigned long start = 0, good_end;
	phys_addr_t base;

	for (i = 0; i < nr_range; i++) {
		unsigned long range, extra;

		range = mr[i].end - mr[i].start;
		puds += (range + PUD_SIZE - 1) >> PUD_SHIFT;

		if (mr[i].page_size_mask & (1 << PG_LEVEL_1G)) {
			extra = range - ((range >> PUD_SHIFT) << PUD_SHIFT);
			pmds += (extra + PMD_SIZE - 1) >> PMD_SHIFT;
		} else {
			pmds += (range + PMD_SIZE - 1) >> PMD_SHIFT;
		}

		if (mr[i].page_size_mask & (1 << PG_LEVEL_2M)) {
			extra = range - ((range >> PMD_SHIFT) << PMD_SHIFT);
#ifdef CONFIG_X86_32
			extra += PMD_SIZE;
#endif
			ptes += (extra + PAGE_SIZE - 1) >> PAGE_SHIFT;
		} else {
			ptes += (range + PAGE_SIZE - 1) >> PAGE_SHIFT;
		}
	}

	tables = roundup(puds * sizeof(pud_t), PAGE_SIZE);
	tables += roundup(pmds * sizeof(pmd_t), PAGE_SIZE);
	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
	good_end = max_pfn_mapped << PAGE_SHIFT;

	base = memblock_find_in_range(start, good_end, tables, PAGE_SIZE);
	if (!base)
		panic("Cannot find space for the kernel page tables");

	pgt_buf_start = base >> PAGE_SHIFT;
	pgt_buf_end = pgt_buf_start;
	pgt_buf_top = pgt_buf_start + (tables >> PAGE_SHIFT);

	printk(KERN_DEBUG "kernel direct mapping tables up to %#lx @ [mem %#010lx-%#010lx]\n",
		mr[nr_range - 1].end - 1, pgt_buf_start << PAGE_SHIFT,
		(pgt_buf_top << PAGE_SHIFT) - 1);
}

void probe_page_size_mask(void)
{
#if !defined(CONFIG_DEBUG_PAGEALLOC) && !defined(CONFIG_KMEMCHECK)
	/*
	 * 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.
	 */
	if (direct_gbpages)
		page_size_mask |= 1 << PG_LEVEL_1G;
	if (cpu_has_pse)
		page_size_mask |= 1 << PG_LEVEL_2M;
#endif

	/* 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;
	}
}
void __init native_pagetable_reserve(u64 start, u64 end)
{
	memblock_reserve(start, end - start);
}

#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;
}

/*
 * 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 start_pfn, end_pfn;
	unsigned long ret = 0;
	unsigned long pos;
	struct map_range mr[NR_RANGE_MR];
	int nr_range, i;

	printk(KERN_INFO "init_memory_mapping: [mem %#010lx-%#010lx]\n",
	       start, end - 1);

	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 " [mem %#010lx-%#010lx] page %s\n",
				mr[i].start, mr[i].end - 1,
			(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(mr, nr_range);

	for (i = 0; i < nr_range; i++)
		ret = kernel_physical_mapping_init(mr[i].start, mr[i].end,
						   mr[i].page_size_mask);

#ifdef CONFIG_X86_32
	early_ioremap_page_table_range_init();

	load_cr3(swapper_pg_dir);
#endif

	__flush_tlb_all();

	/*
	 * Reserve the kernel pagetable pages we used (pgt_buf_start -
	 * pgt_buf_end) and free the other ones (pgt_buf_end - pgt_buf_top)
	 * so that they can be reused for other purposes.
	 *
	 * On native it just means calling memblock_reserve, on Xen it also
	 * means marking RW the pagetable pages that we allocated before
	 * but that haven't been used.
	 *
	 * In fact on xen we mark RO the whole range pgt_buf_start -
	 * pgt_buf_top, because we have to make sure that when
	 * init_memory_mapping reaches the pagetable pages area, it maps
	 * RO all the pagetable pages, including the ones that are beyond
	 * pgt_buf_end at that time.
	 */
	if (!after_bootmem && pgt_buf_end > pgt_buf_start)
		x86_init.mapping.pagetable_reserve(PFN_PHYS(pgt_buf_start),
				PFN_PHYS(pgt_buf_end));

	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;
	unsigned long begin_aligned, end_aligned;

	/* Make sure boundaries are page aligned */
	begin_aligned = PAGE_ALIGN(begin);
	end_aligned   = end & PAGE_MASK;

	if (WARN_ON(begin_aligned != begin || end_aligned != end)) {
		begin = begin_aligned;
		end   = end_aligned;
	}

	if (begin >= end)
		return;

	addr = begin;

	/*
	 * 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 [mem %#010lx-%#010lx]\n",
		begin, end - 1);
	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 and non-executable first.
	 */
	set_memory_nx(begin, (end - begin) >> PAGE_SHIFT);
	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, 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 __init free_initrd_mem(unsigned long start, unsigned long end)
{
	/*
	 * end could be not aligned, and We can not align that,
	 * decompresser could be confused by aligned initrd_end
	 * We already reserve the end partial page before in
	 *   - i386_start_kernel()
	 *   - x86_64_start_kernel()
	 *   - relocate_initrd()
	 * So here We can do PAGE_ALIGN() safely to get partial page to be freed
	 */
	free_init_pages("initrd memory", start, PAGE_ALIGN(end));
}
#endif

void __init zone_sizes_init(void)
{
	unsigned long max_zone_pfns[MAX_NR_ZONES];

	memset(max_zone_pfns, 0, sizeof(max_zone_pfns));

#ifdef CONFIG_ZONE_DMA
	max_zone_pfns[ZONE_DMA]		= MAX_DMA_PFN;
#endif
#ifdef CONFIG_ZONE_DMA32
	max_zone_pfns[ZONE_DMA32]	= MAX_DMA32_PFN;
#endif
	max_zone_pfns[ZONE_NORMAL]	= max_low_pfn;
#ifdef CONFIG_HIGHMEM
	max_zone_pfns[ZONE_HIGHMEM]	= max_pfn;
#endif

	free_area_init_nodes(max_zone_pfns);
}
Commit	Line	Data
5a0e3ad6	1	#include <linux/gfp.h>
2c1b284e	2	#include <linux/initrd.h>
540aca06	3	#include <linux/ioport.h>
e5b2bb55	4	#include <linux/swap.h>
a9ce6bc1	5	#include <linux/memblock.h>
17623915	6	#include <linux/bootmem.h> /* for max_low_pfn */
540aca06	7
e5b2bb55	8	#include <asm/cacheflush.h>
f765090a	9	#include <asm/e820.h>
4fcb2083	10	#include <asm/init.h>
e5b2bb55	11	#include <asm/page.h>
540aca06	12	#include <asm/page_types.h>
e5b2bb55	13	#include <asm/sections.h>
49834396	14	#include <asm/setup.h>
f765090a	15	#include <asm/tlbflush.h>
9518e0e4	16	#include <asm/tlb.h>
76c06927	17	#include <asm/proto.h>
17623915	18	#include <asm/dma.h> /* for MAX_DMA_PFN */
9518e0e4	19
d1b19426 YL	20	unsigned long __initdata pgt_buf_start;
	21	unsigned long __meminitdata pgt_buf_end;
	22	unsigned long __meminitdata pgt_buf_top;
f765090a PE	23
	24	int after_bootmem;
	25
	26	int direct_gbpages
	27	#ifdef CONFIG_DIRECT_GBPAGES
	28	= 1
	29	#endif
	30	;
	31
844ab6f9 JS	32	struct map_range {
	33	unsigned long start;
	34	unsigned long end;
	35	unsigned page_size_mask;
	36	};
	37
fa62aafe	38	static int page_size_mask;
844ab6f9 JS	39	/*
	40	* First calculate space needed for kernel direct mapping page tables to cover
	41	* mr[0].start to mr[nr_range - 1].end, while accounting for possible 2M and 1GB
	42	* pages. Then find enough contiguous space for those page tables.
	43	*/
	44	static void __init find_early_table_space(struct map_range *mr, int nr_range)
f765090a	45	{
844ab6f9 JS	46	int i;
	47	unsigned long puds = 0, pmds = 0, ptes = 0, tables;
	48	unsigned long start = 0, good_end;
a9ce6bc1	49	phys_addr_t base;
f765090a	50
844ab6f9 JS	51	for (i = 0; i < nr_range; i++) {
844ab6f9 JS	52	unsigned long range, extra;
f765090a	53
844ab6f9 JS	54	range = mr[i].end - mr[i].start;
844ab6f9 JS	55	puds += (range + PUD_SIZE - 1) >> PUD_SHIFT;
f765090a	56
844ab6f9 JS	57	if (mr[i].page_size_mask & (1 << PG_LEVEL_1G)) {
	58	extra = range - ((range >> PUD_SHIFT) << PUD_SHIFT);
	59	pmds += (extra + PMD_SIZE - 1) >> PMD_SHIFT;
	60	} else {
	61	pmds += (range + PMD_SIZE - 1) >> PMD_SHIFT;
	62	}
f765090a	63
844ab6f9 JS	64	if (mr[i].page_size_mask & (1 << PG_LEVEL_2M)) {
844ab6f9 JS	65	extra = range - ((range >> PMD_SHIFT) << PMD_SHIFT);
f765090a	66	#ifdef CONFIG_X86_32
844ab6f9	67	extra += PMD_SIZE;
f765090a	68	#endif
844ab6f9 JS	69	ptes += (extra + PAGE_SIZE - 1) >> PAGE_SHIFT;
	70	} else {
	71	ptes += (range + PAGE_SIZE - 1) >> PAGE_SHIFT;
	72	}
	73	}
f765090a	74
844ab6f9 JS	75	tables = roundup(puds * sizeof(pud_t), PAGE_SIZE);
844ab6f9 JS	76	tables += roundup(pmds * sizeof(pmd_t), PAGE_SIZE);
f765090a PE	77	tables += roundup(ptes * sizeof(pte_t), PAGE_SIZE);
	78
	79	#ifdef CONFIG_X86_32
	80	/* for fixmap */
	81	tables += roundup(__end_of_fixed_addresses * sizeof(pte_t), PAGE_SIZE);
f765090a	82	#endif
8548c84d	83	good_end = max_pfn_mapped << PAGE_SHIFT;
1411e0ec	84
4b239f45	85	base = memblock_find_in_range(start, good_end, tables, PAGE_SIZE);
1f5026a7	86	if (!base)
f765090a PE	87	panic("Cannot find space for the kernel page tables");
f765090a PE	88
d1b19426 YL	89	pgt_buf_start = base >> PAGE_SHIFT;
	90	pgt_buf_end = pgt_buf_start;
	91	pgt_buf_top = pgt_buf_start + (tables >> PAGE_SHIFT);
f765090a	92
365811d6	93	printk(KERN_DEBUG "kernel direct mapping tables up to %#lx @ [mem %#010lx-%#010lx]\n",
844ab6f9	94	mr[nr_range - 1].end - 1, pgt_buf_start << PAGE_SHIFT,
365811d6	95	(pgt_buf_top << PAGE_SHIFT) - 1);
f765090a PE	96	}
f765090a PE	97
fa62aafe YL	98	void probe_page_size_mask(void)
	99	{
	100	#if !defined(CONFIG_DEBUG_PAGEALLOC) && !defined(CONFIG_KMEMCHECK)
	101	/*
	102	* For CONFIG_DEBUG_PAGEALLOC, identity mapping will use small pages.
	103	* This will simplify cpa(), which otherwise needs to support splitting
	104	* large pages into small in interrupt context, etc.
	105	*/
	106	if (direct_gbpages)
	107	page_size_mask \|= 1 << PG_LEVEL_1G;
	108	if (cpu_has_pse)
	109	page_size_mask \|= 1 << PG_LEVEL_2M;
	110	#endif
	111
	112	/* Enable PSE if available */
	113	if (cpu_has_pse)
	114	set_in_cr4(X86_CR4_PSE);
	115
	116	/* Enable PGE if available */
	117	if (cpu_has_pge) {
	118	set_in_cr4(X86_CR4_PGE);
	119	__supported_pte_mask \|= _PAGE_GLOBAL;
	120	}
	121	}
53f8023f	122	void __init native_pagetable_reserve(u64 start, u64 end)
279b706b	123	{
24aa0788	124	memblock_reserve(start, end - start);
279b706b SS	125	}
279b706b SS	126
f765090a PE	127	#ifdef CONFIG_X86_32
	128	#define NR_RANGE_MR 3
	129	#else /* CONFIG_X86_64 */
	130	#define NR_RANGE_MR 5
	131	#endif
	132
dc9dd5cc JB	133	static int __meminit save_mr(struct map_range *mr, int nr_range,
	134	unsigned long start_pfn, unsigned long end_pfn,
	135	unsigned long page_size_mask)
f765090a PE	136	{
	137	if (start_pfn < end_pfn) {
	138	if (nr_range >= NR_RANGE_MR)
	139	panic("run out of range for init_memory_mapping\n");
	140	mr[nr_range].start = start_pfn<<PAGE_SHIFT;
	141	mr[nr_range].end = end_pfn<<PAGE_SHIFT;
	142	mr[nr_range].page_size_mask = page_size_mask;
	143	nr_range++;
	144	}
	145
	146	return nr_range;
	147	}
	148
f765090a PE	149	/*
	150	* Setup the direct mapping of the physical memory at PAGE_OFFSET.
	151	* This runs before bootmem is initialized and gets pages directly from
	152	* the physical memory. To access them they are temporarily mapped.
	153	*/
	154	unsigned long __init_refok init_memory_mapping(unsigned long start,
	155	unsigned long end)
	156	{
f765090a	157	unsigned long start_pfn, end_pfn;
c77a3b59	158	unsigned long ret = 0;
f765090a	159	unsigned long pos;
f765090a PE	160	struct map_range mr[NR_RANGE_MR];
f765090a PE	161	int nr_range, i;
f765090a	162
365811d6 BH	163	printk(KERN_INFO "init_memory_mapping: [mem %#010lx-%#010lx]\n",
365811d6 BH	164	start, end - 1);
f765090a	165
f765090a PE	166	memset(mr, 0, sizeof(mr));
	167	nr_range = 0;
	168
	169	/* head if not big page alignment ? */
	170	start_pfn = start >> PAGE_SHIFT;
	171	pos = start_pfn << PAGE_SHIFT;
	172	#ifdef CONFIG_X86_32
	173	/*
	174	* Don't use a large page for the first 2/4MB of memory
	175	* because there are often fixed size MTRRs in there
	176	* and overlapping MTRRs into large pages can cause
	177	* slowdowns.
	178	*/
	179	if (pos == 0)
	180	end_pfn = 1<<(PMD_SHIFT - PAGE_SHIFT);
	181	else
	182	end_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
	183	<< (PMD_SHIFT - PAGE_SHIFT);
	184	#else /* CONFIG_X86_64 */
	185	end_pfn = ((pos + (PMD_SIZE - 1)) >> PMD_SHIFT)
	186	<< (PMD_SHIFT - PAGE_SHIFT);
	187	#endif
	188	if (end_pfn > (end >> PAGE_SHIFT))
	189	end_pfn = end >> PAGE_SHIFT;
	190	if (start_pfn < end_pfn) {
	191	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);
	192	pos = end_pfn << PAGE_SHIFT;
	193	}
	194
	195	/* big page (2M) range */
	196	start_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
	197	<< (PMD_SHIFT - PAGE_SHIFT);
	198	#ifdef CONFIG_X86_32
	199	end_pfn = (end>>PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT);
	200	#else /* CONFIG_X86_64 */
	201	end_pfn = ((pos + (PUD_SIZE - 1))>>PUD_SHIFT)
	202	<< (PUD_SHIFT - PAGE_SHIFT);
	203	if (end_pfn > ((end>>PMD_SHIFT)<<(PMD_SHIFT - PAGE_SHIFT)))
	204	end_pfn = ((end>>PMD_SHIFT)<<(PMD_SHIFT - PAGE_SHIFT));
	205	#endif
	206
	207	if (start_pfn < end_pfn) {
	208	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
	209	page_size_mask & (1<<PG_LEVEL_2M));
	210	pos = end_pfn << PAGE_SHIFT;
	211	}
	212
	213	#ifdef CONFIG_X86_64
	214	/* big page (1G) range */
	215	start_pfn = ((pos + (PUD_SIZE - 1))>>PUD_SHIFT)
	216	<< (PUD_SHIFT - PAGE_SHIFT);
	217	end_pfn = (end >> PUD_SHIFT) << (PUD_SHIFT - PAGE_SHIFT);
	218	if (start_pfn < end_pfn) {
	219	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
	220	page_size_mask &
	221	((1<<PG_LEVEL_2M)\|(1<<PG_LEVEL_1G)));
	222	pos = end_pfn << PAGE_SHIFT;
	223	}
	224
	225	/* tail is not big page (1G) alignment */
	226	start_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
	227	<< (PMD_SHIFT - PAGE_SHIFT);
	228	end_pfn = (end >> PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT);
	229	if (start_pfn < end_pfn) {
230	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
231	page_size_mask & (1<<PG_LEVEL_2M));
232	pos = end_pfn << PAGE_SHIFT;
233	}
234	#endif
235
236	/* tail is not big page (2M) alignment */
237	start_pfn = pos>>PAGE_SHIFT;
238	end_pfn = end>>PAGE_SHIFT;
239	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);
240
241	/* try to merge same page size and continuous */
242	for (i = 0; nr_range > 1 && i < nr_range - 1; i++) {
243	unsigned long old_start;
244	if (mr[i].end != mr[i+1].start \|\|
245	mr[i].page_size_mask != mr[i+1].page_size_mask)
246	continue;
247	/* move it */
248	old_start = mr[i].start;
249	memmove(&mr[i], &mr[i+1],
250	(nr_range - 1 - i) * sizeof(struct map_range));
251	mr[i--].start = old_start;
252	nr_range--;
253	}
254
255	for (i = 0; i < nr_range; i++)
365811d6 BH	256	printk(KERN_DEBUG " [mem %#010lx-%#010lx] page %s\n",
365811d6 BH	257	mr[i].start, mr[i].end - 1,
f765090a PE	258	(mr[i].page_size_mask & (1<<PG_LEVEL_1G))?"1G":(
	259	(mr[i].page_size_mask & (1<<PG_LEVEL_2M))?"2M":"4k"));
	260
	261	/*
	262	* Find space for the kernel direct mapping tables.
	263	*
	264	* Later we should allocate these tables in the local node of the
	265	* memory mapped. Unfortunately this is done currently before the
	266	* nodes are discovered.
	267	*/
	268	if (!after_bootmem)
844ab6f9	269	find_early_table_space(mr, nr_range);
f765090a	270
f765090a PE	271	for (i = 0; i < nr_range; i++)
	272	ret = kernel_physical_mapping_init(mr[i].start, mr[i].end,
	273	mr[i].page_size_mask);
f765090a PE	274
	275	#ifdef CONFIG_X86_32
	276	early_ioremap_page_table_range_init();
	277
	278	load_cr3(swapper_pg_dir);
	279	#endif
	280
f765090a PE	281	__flush_tlb_all();
f765090a PE	282
279b706b SS	283	/*
	284	* Reserve the kernel pagetable pages we used (pgt_buf_start -
	285	* pgt_buf_end) and free the other ones (pgt_buf_end - pgt_buf_top)
	286	* so that they can be reused for other purposes.
	287	*
24aa0788 TH	288	* On native it just means calling memblock_reserve, on Xen it also
24aa0788 TH	289	* means marking RW the pagetable pages that we allocated before
279b706b SS	290	* but that haven't been used.
	291	*
	292	* In fact on xen we mark RO the whole range pgt_buf_start -
	293	* pgt_buf_top, because we have to make sure that when
	294	* init_memory_mapping reaches the pagetable pages area, it maps
	295	* RO all the pagetable pages, including the ones that are beyond
	296	* pgt_buf_end at that time.
	297	*/
d1b19426	298	if (!after_bootmem && pgt_buf_end > pgt_buf_start)
279b706b SS	299	x86_init.mapping.pagetable_reserve(PFN_PHYS(pgt_buf_start),
279b706b SS	300	PFN_PHYS(pgt_buf_end));
f765090a PE	301
	302	if (!after_bootmem)
	303	early_memtest(start, end);
	304
	305	return ret >> PAGE_SHIFT;
	306	}
	307
e5b2bb55	308
540aca06 PE	309	/*
	310	* devmem_is_allowed() checks to see if /dev/mem access to a certain address
	311	* is valid. The argument is a physical page number.
	312	*
	313	*
	314	* On x86, access has to be given to the first megabyte of ram because that area
	315	* contains bios code and data regions used by X and dosemu and similar apps.
	316	* Access has to be given to non-kernel-ram areas as well, these contain the PCI
	317	* mmio resources as well as potential bios/acpi data regions.
	318	*/
	319	int devmem_is_allowed(unsigned long pagenr)
	320	{
73e8f3d7	321	if (pagenr < 256)
540aca06 PE	322	return 1;
	323	if (iomem_is_exclusive(pagenr << PAGE_SHIFT))
	324	return 0;
	325	if (!page_is_ram(pagenr))
	326	return 1;
	327	return 0;
	328	}
	329
e5b2bb55 PE	330	void free_init_pages(char *what, unsigned long begin, unsigned long end)
e5b2bb55 PE	331	{
c967da6a YL	332	unsigned long addr;
c967da6a YL	333	unsigned long begin_aligned, end_aligned;
e5b2bb55	334
c967da6a YL	335	/* Make sure boundaries are page aligned */
	336	begin_aligned = PAGE_ALIGN(begin);
	337	end_aligned = end & PAGE_MASK;
	338
	339	if (WARN_ON(begin_aligned != begin \|\| end_aligned != end)) {
	340	begin = begin_aligned;
	341	end = end_aligned;
	342	}
	343
	344	if (begin >= end)
e5b2bb55 PE	345	return;
e5b2bb55 PE	346
c967da6a YL	347	addr = begin;
c967da6a YL	348
e5b2bb55 PE	349	/*
	350	* If debugging page accesses then do not free this memory but
	351	* mark them not present - any buggy init-section access will
	352	* create a kernel page fault:
	353	*/
	354	#ifdef CONFIG_DEBUG_PAGEALLOC
365811d6 BH	355	printk(KERN_INFO "debug: unmapping init [mem %#010lx-%#010lx]\n",
365811d6 BH	356	begin, end - 1);
e5b2bb55 PE	357	set_memory_np(begin, (end - begin) >> PAGE_SHIFT);
	358	#else
	359	/*
	360	* We just marked the kernel text read only above, now that
	361	* we are going to free part of that, we need to make that
5bd5a452	362	* writeable and non-executable first.
e5b2bb55	363	*/
5bd5a452	364	set_memory_nx(begin, (end - begin) >> PAGE_SHIFT);
e5b2bb55 PE	365	set_memory_rw(begin, (end - begin) >> PAGE_SHIFT);
	366
	367	printk(KERN_INFO "Freeing %s: %luk freed\n", what, (end - begin) >> 10);
	368
	369	for (; addr < end; addr += PAGE_SIZE) {
	370	ClearPageReserved(virt_to_page(addr));
	371	init_page_count(virt_to_page(addr));
c967da6a	372	memset((void *)addr, POISON_FREE_INITMEM, PAGE_SIZE);
e5b2bb55 PE	373	free_page(addr);
	374	totalram_pages++;
	375	}
	376	#endif
	377	}
	378
	379	void free_initmem(void)
	380	{
	381	free_init_pages("unused kernel memory",
	382	(unsigned long)(&__init_begin),
	383	(unsigned long)(&__init_end));
	384	}
731ddea6 PE	385
731ddea6 PE	386	#ifdef CONFIG_BLK_DEV_INITRD
0d26d1d8	387	void __init free_initrd_mem(unsigned long start, unsigned long end)
731ddea6	388	{
c967da6a YL	389	/*
	390	* end could be not aligned, and We can not align that,
	391	* decompresser could be confused by aligned initrd_end
	392	* We already reserve the end partial page before in
	393	* - i386_start_kernel()
	394	* - x86_64_start_kernel()
	395	* - relocate_initrd()
	396	* So here We can do PAGE_ALIGN() safely to get partial page to be freed
	397	*/
	398	free_init_pages("initrd memory", start, PAGE_ALIGN(end));
731ddea6 PE	399	}
731ddea6 PE	400	#endif
17623915 PE	401
	402	void __init zone_sizes_init(void)
	403	{
	404	unsigned long max_zone_pfns[MAX_NR_ZONES];
	405
	406	memset(max_zone_pfns, 0, sizeof(max_zone_pfns));
	407
	408	#ifdef CONFIG_ZONE_DMA
	409	max_zone_pfns[ZONE_DMA] = MAX_DMA_PFN;
	410	#endif
	411	#ifdef CONFIG_ZONE_DMA32
	412	max_zone_pfns[ZONE_DMA32] = MAX_DMA32_PFN;
	413	#endif
	414	max_zone_pfns[ZONE_NORMAL] = max_low_pfn;
	415	#ifdef CONFIG_HIGHMEM
	416	max_zone_pfns[ZONE_HIGHMEM] = max_pfn;
	417	#endif
	418
	419	free_area_init_nodes(max_zone_pfns);
	420	}
	421