[mirror_ubuntu-bionic-kernel.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 <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/system.h>
#include <asm/tlbflush.h>
#include <asm/tlb.h>
#include <asm/proto.h>

DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);

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
;

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

	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;
#else
	start = 0x8000;
#endif
	base = memblock_find_in_range(start, max_pfn_mapped<<PAGE_SHIFT,
					tables, PAGE_SIZE);
	if (base == MEMBLOCK_ERROR)
		panic("Cannot find space for the kernel page tables");

	e820_table_start = base >> 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;
}

/*
 * 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 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.
	 */
	use_pse = use_gbpages = 0;
#else
	use_pse = cpu_has_pse;
	use_gbpages = direct_gbpages;
#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;
	}

	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);

	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

#ifdef CONFIG_X86_64
	if (!after_bootmem && !start) {
		pud_t *pud;
		pmd_t *pmd;

		mmu_cr4_features = read_cr4();

		/*
		 * _brk_end cannot change anymore, but it and _end may be
		 * located on different 2M pages. cleanup_highmap(), however,
		 * can only consider _end when it runs, so destroy any
		 * mappings beyond _brk_end here.
		 */
		pud = pud_offset(pgd_offset_k(_brk_end), _brk_end);
		pmd = pmd_offset(pud, _brk_end - 1);
		while (++pmd <= pmd_offset(pud, (unsigned long)_end - 1))
			pmd_clear(pmd);
	}
#endif
	__flush_tlb_all();

	if (!after_bootmem && e820_table_end > e820_table_start)
		memblock_x86_reserve_range(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;
	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 memory %08lx..%08lx\n",
		begin, 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, 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)
{
	/*
	 * 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
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>
540aca06	6
e5b2bb55	7	#include <asm/cacheflush.h>
f765090a	8	#include <asm/e820.h>
4fcb2083	9	#include <asm/init.h>
e5b2bb55	10	#include <asm/page.h>
540aca06	11	#include <asm/page_types.h>
e5b2bb55	12	#include <asm/sections.h>
49834396	13	#include <asm/setup.h>
e5b2bb55	14	#include <asm/system.h>
f765090a	15	#include <asm/tlbflush.h>
9518e0e4	16	#include <asm/tlb.h>
76c06927	17	#include <asm/proto.h>
9518e0e4 PE	18
9518e0e4 PE	19	DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
f765090a	20
298af9d8 PE	21	unsigned long __initdata e820_table_start;
	22	unsigned long __meminitdata e820_table_end;
	23	unsigned long __meminitdata e820_table_top;
f765090a PE	24
	25	int after_bootmem;
	26
	27	int direct_gbpages
	28	#ifdef CONFIG_DIRECT_GBPAGES
	29	= 1
	30	#endif
	31	;
	32
	33	static void __init find_early_table_space(unsigned long end, int use_pse,
	34	int use_gbpages)
	35	{
	36	unsigned long puds, pmds, ptes, tables, start;
a9ce6bc1	37	phys_addr_t base;
f765090a PE	38
	39	puds = (end + PUD_SIZE - 1) >> PUD_SHIFT;
	40	tables = roundup(puds * sizeof(pud_t), PAGE_SIZE);
	41
	42	if (use_gbpages) {
	43	unsigned long extra;
	44
	45	extra = end - ((end>>PUD_SHIFT) << PUD_SHIFT);
	46	pmds = (extra + PMD_SIZE - 1) >> PMD_SHIFT;
	47	} else
	48	pmds = (end + PMD_SIZE - 1) >> PMD_SHIFT;
	49
	50	tables += roundup(pmds * sizeof(pmd_t), PAGE_SIZE);
	51
	52	if (use_pse) {
	53	unsigned long extra;
	54
	55	extra = end - ((end>>PMD_SHIFT) << PMD_SHIFT);
	56	#ifdef CONFIG_X86_32
	57	extra += PMD_SIZE;
	58	#endif
	59	ptes = (extra + PAGE_SIZE - 1) >> PAGE_SHIFT;
	60	} else
	61	ptes = (end + PAGE_SIZE - 1) >> PAGE_SHIFT;
	62
	63	tables += roundup(ptes * sizeof(pte_t), PAGE_SIZE);
	64
	65	#ifdef CONFIG_X86_32
	66	/* for fixmap */
	67	tables += roundup(__end_of_fixed_addresses * sizeof(pte_t), PAGE_SIZE);
	68	#endif
	69
	70	/*
	71	* RED-PEN putting page tables only on node 0 could
	72	* cause a hotspot and fill up ZONE_DMA. The page tables
	73	* need roughly 0.5KB per GB.
	74	*/
	75	#ifdef CONFIG_X86_32
	76	start = 0x7000;
80989ce0	77	#else
f765090a	78	start = 0x8000;
f765090a	79	#endif
a9ce6bc1	80	base = memblock_find_in_range(start, max_pfn_mapped<<PAGE_SHIFT,
80989ce0	81	tables, PAGE_SIZE);
a9ce6bc1	82	if (base == MEMBLOCK_ERROR)
f765090a PE	83	panic("Cannot find space for the kernel page tables");
f765090a PE	84
a9ce6bc1	85	e820_table_start = base >> PAGE_SHIFT;
298af9d8 PE	86	e820_table_end = e820_table_start;
298af9d8 PE	87	e820_table_top = e820_table_start + (tables >> PAGE_SHIFT);
f765090a PE	88
f765090a PE	89	printk(KERN_DEBUG "kernel direct mapping tables up to %lx @ %lx-%lx\n",
298af9d8	90	end, e820_table_start << PAGE_SHIFT, e820_table_top << PAGE_SHIFT);
f765090a PE	91	}
	92
	93	struct map_range {
	94	unsigned long start;
	95	unsigned long end;
	96	unsigned page_size_mask;
	97	};
	98
	99	#ifdef CONFIG_X86_32
	100	#define NR_RANGE_MR 3
	101	#else /* CONFIG_X86_64 */
	102	#define NR_RANGE_MR 5
	103	#endif
	104
dc9dd5cc JB	105	static int __meminit save_mr(struct map_range *mr, int nr_range,
	106	unsigned long start_pfn, unsigned long end_pfn,
	107	unsigned long page_size_mask)
f765090a PE	108	{
	109	if (start_pfn < end_pfn) {
	110	if (nr_range >= NR_RANGE_MR)
	111	panic("run out of range for init_memory_mapping\n");
	112	mr[nr_range].start = start_pfn<<PAGE_SHIFT;
	113	mr[nr_range].end = end_pfn<<PAGE_SHIFT;
	114	mr[nr_range].page_size_mask = page_size_mask;
	115	nr_range++;
	116	}
	117
	118	return nr_range;
	119	}
	120
f765090a PE	121	/*
	122	* Setup the direct mapping of the physical memory at PAGE_OFFSET.
	123	* This runs before bootmem is initialized and gets pages directly from
	124	* the physical memory. To access them they are temporarily mapped.
	125	*/
	126	unsigned long __init_refok init_memory_mapping(unsigned long start,
	127	unsigned long end)
	128	{
	129	unsigned long page_size_mask = 0;
	130	unsigned long start_pfn, end_pfn;
c77a3b59	131	unsigned long ret = 0;
f765090a	132	unsigned long pos;
f765090a PE	133
	134	struct map_range mr[NR_RANGE_MR];
	135	int nr_range, i;
	136	int use_pse, use_gbpages;
	137
	138	printk(KERN_INFO "init_memory_mapping: %016lx-%016lx\n", start, end);
	139
f8561296	140	#if defined(CONFIG_DEBUG_PAGEALLOC) \|\| defined(CONFIG_KMEMCHECK)
f765090a PE	141	/*
	142	* For CONFIG_DEBUG_PAGEALLOC, identity mapping will use small pages.
	143	* This will simplify cpa(), which otherwise needs to support splitting
	144	* large pages into small in interrupt context, etc.
	145	*/
	146	use_pse = use_gbpages = 0;
	147	#else
	148	use_pse = cpu_has_pse;
	149	use_gbpages = direct_gbpages;
	150	#endif
	151
f765090a PE	152	/* Enable PSE if available */
	153	if (cpu_has_pse)
	154	set_in_cr4(X86_CR4_PSE);
	155
	156	/* Enable PGE if available */
	157	if (cpu_has_pge) {
	158	set_in_cr4(X86_CR4_PGE);
	159	__supported_pte_mask \|= _PAGE_GLOBAL;
	160	}
f765090a PE	161
	162	if (use_gbpages)
	163	page_size_mask \|= 1 << PG_LEVEL_1G;
	164	if (use_pse)
	165	page_size_mask \|= 1 << PG_LEVEL_2M;
	166
	167	memset(mr, 0, sizeof(mr));
	168	nr_range = 0;
	169
	170	/* head if not big page alignment ? */
	171	start_pfn = start >> PAGE_SHIFT;
	172	pos = start_pfn << PAGE_SHIFT;
	173	#ifdef CONFIG_X86_32
	174	/*
	175	* Don't use a large page for the first 2/4MB of memory
	176	* because there are often fixed size MTRRs in there
	177	* and overlapping MTRRs into large pages can cause
	178	* slowdowns.
	179	*/
	180	if (pos == 0)
	181	end_pfn = 1<<(PMD_SHIFT - PAGE_SHIFT);
	182	else
	183	end_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
	184	<< (PMD_SHIFT - PAGE_SHIFT);
	185	#else /* CONFIG_X86_64 */
	186	end_pfn = ((pos + (PMD_SIZE - 1)) >> PMD_SHIFT)
	187	<< (PMD_SHIFT - PAGE_SHIFT);
	188	#endif
	189	if (end_pfn > (end >> PAGE_SHIFT))
	190	end_pfn = end >> PAGE_SHIFT;
	191	if (start_pfn < end_pfn) {
	192	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);
	193	pos = end_pfn << PAGE_SHIFT;
	194	}
	195
	196	/* big page (2M) range */
	197	start_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
	198	<< (PMD_SHIFT - PAGE_SHIFT);
	199	#ifdef CONFIG_X86_32
	200	end_pfn = (end>>PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT);
	201	#else /* CONFIG_X86_64 */
	202	end_pfn = ((pos + (PUD_SIZE - 1))>>PUD_SHIFT)
	203	<< (PUD_SHIFT - PAGE_SHIFT);
	204	if (end_pfn > ((end>>PMD_SHIFT)<<(PMD_SHIFT - PAGE_SHIFT)))
	205	end_pfn = ((end>>PMD_SHIFT)<<(PMD_SHIFT - PAGE_SHIFT));
	206	#endif
	207
	208	if (start_pfn < end_pfn) {
	209	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
	210	page_size_mask & (1<<PG_LEVEL_2M));
	211	pos = end_pfn << PAGE_SHIFT;
	212	}
	213
	214	#ifdef CONFIG_X86_64
	215	/* big page (1G) range */
	216	start_pfn = ((pos + (PUD_SIZE - 1))>>PUD_SHIFT)
	217	<< (PUD_SHIFT - PAGE_SHIFT);
	218	end_pfn = (end >> PUD_SHIFT) << (PUD_SHIFT - PAGE_SHIFT);
	219	if (start_pfn < end_pfn) {
	220	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
	221	page_size_mask &
	222	((1<<PG_LEVEL_2M)\|(1<<PG_LEVEL_1G)));
	223	pos = end_pfn << PAGE_SHIFT;
	224	}
225
226	/* tail is not big page (1G) alignment */
227	start_pfn = ((pos + (PMD_SIZE - 1))>>PMD_SHIFT)
228	<< (PMD_SHIFT - PAGE_SHIFT);
229	end_pfn = (end >> PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT);
230	if (start_pfn < end_pfn) {
231	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn,
232	page_size_mask & (1<<PG_LEVEL_2M));
233	pos = end_pfn << PAGE_SHIFT;
234	}
235	#endif
236
237	/* tail is not big page (2M) alignment */
238	start_pfn = pos>>PAGE_SHIFT;
239	end_pfn = end>>PAGE_SHIFT;
240	nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);
241
242	/* try to merge same page size and continuous */
243	for (i = 0; nr_range > 1 && i < nr_range - 1; i++) {
244	unsigned long old_start;
245	if (mr[i].end != mr[i+1].start \|\|
246	mr[i].page_size_mask != mr[i+1].page_size_mask)
247	continue;
248	/* move it */
249	old_start = mr[i].start;
250	memmove(&mr[i], &mr[i+1],
251	(nr_range - 1 - i) * sizeof(struct map_range));
252	mr[i--].start = old_start;
253	nr_range--;
254	}
255
256	for (i = 0; i < nr_range; i++)
257	printk(KERN_DEBUG " %010lx - %010lx page %s\n",
258	mr[i].start, mr[i].end,
259	(mr[i].page_size_mask & (1<<PG_LEVEL_1G))?"1G":(
260	(mr[i].page_size_mask & (1<<PG_LEVEL_2M))?"2M":"4k"));
261
262	/*
263	* Find space for the kernel direct mapping tables.
264	*
265	* Later we should allocate these tables in the local node of the
266	* memory mapped. Unfortunately this is done currently before the
267	* nodes are discovered.
268	*/
269	if (!after_bootmem)
270	find_early_table_space(end, use_pse, use_gbpages);
271
f765090a PE	272	for (i = 0; i < nr_range; i++)
	273	ret = kernel_physical_mapping_init(mr[i].start, mr[i].end,
	274	mr[i].page_size_mask);
f765090a PE	275
	276	#ifdef CONFIG_X86_32
	277	early_ioremap_page_table_range_init();
	278
	279	load_cr3(swapper_pg_dir);
	280	#endif
	281
	282	#ifdef CONFIG_X86_64
49834396 JB	283	if (!after_bootmem && !start) {
	284	pud_t *pud;
	285	pmd_t *pmd;
	286
f765090a	287	mmu_cr4_features = read_cr4();
49834396 JB	288
	289	/*
	290	* _brk_end cannot change anymore, but it and _end may be
	291	* located on different 2M pages. cleanup_highmap(), however,
	292	* can only consider _end when it runs, so destroy any
	293	* mappings beyond _brk_end here.
	294	*/
	295	pud = pud_offset(pgd_offset_k(_brk_end), _brk_end);
	296	pmd = pmd_offset(pud, _brk_end - 1);
	297	while (++pmd <= pmd_offset(pud, (unsigned long)_end - 1))
	298	pmd_clear(pmd);
	299	}
f765090a PE	300	#endif
	301	__flush_tlb_all();
	302
298af9d8	303	if (!after_bootmem && e820_table_end > e820_table_start)
a9ce6bc1	304	memblock_x86_reserve_range(e820_table_start << PAGE_SHIFT,
298af9d8	305	e820_table_end << PAGE_SHIFT, "PGTABLE");
f765090a PE	306
	307	if (!after_bootmem)
	308	early_memtest(start, end);
	309
	310	return ret >> PAGE_SHIFT;
	311	}
	312
e5b2bb55	313
540aca06 PE	314	/*
	315	* devmem_is_allowed() checks to see if /dev/mem access to a certain address
	316	* is valid. The argument is a physical page number.
	317	*
	318	*
	319	* On x86, access has to be given to the first megabyte of ram because that area
	320	* contains bios code and data regions used by X and dosemu and similar apps.
	321	* Access has to be given to non-kernel-ram areas as well, these contain the PCI
	322	* mmio resources as well as potential bios/acpi data regions.
	323	*/
	324	int devmem_is_allowed(unsigned long pagenr)
	325	{
	326	if (pagenr <= 256)
	327	return 1;
	328	if (iomem_is_exclusive(pagenr << PAGE_SHIFT))
	329	return 0;
	330	if (!page_is_ram(pagenr))
	331	return 1;
	332	return 0;
	333	}
	334
e5b2bb55 PE	335	void free_init_pages(char *what, unsigned long begin, unsigned long end)
e5b2bb55 PE	336	{
c967da6a YL	337	unsigned long addr;
c967da6a YL	338	unsigned long begin_aligned, end_aligned;
e5b2bb55	339
c967da6a YL	340	/* Make sure boundaries are page aligned */
	341	begin_aligned = PAGE_ALIGN(begin);
	342	end_aligned = end & PAGE_MASK;
	343
	344	if (WARN_ON(begin_aligned != begin \|\| end_aligned != end)) {
	345	begin = begin_aligned;
	346	end = end_aligned;
	347	}
	348
	349	if (begin >= end)
e5b2bb55 PE	350	return;
e5b2bb55 PE	351
c967da6a YL	352	addr = begin;
c967da6a YL	353
e5b2bb55 PE	354	/*
	355	* If debugging page accesses then do not free this memory but
	356	* mark them not present - any buggy init-section access will
	357	* create a kernel page fault:
	358	*/
	359	#ifdef CONFIG_DEBUG_PAGEALLOC
	360	printk(KERN_INFO "debug: unmapping init memory %08lx..%08lx\n",
c967da6a	361	begin, end);
e5b2bb55 PE	362	set_memory_np(begin, (end - begin) >> PAGE_SHIFT);
	363	#else
	364	/*
	365	* We just marked the kernel text read only above, now that
	366	* we are going to free part of that, we need to make that
	367	* writeable first.
	368	*/
	369	set_memory_rw(begin, (end - begin) >> PAGE_SHIFT);
	370
	371	printk(KERN_INFO "Freeing %s: %luk freed\n", what, (end - begin) >> 10);
	372
	373	for (; addr < end; addr += PAGE_SIZE) {
	374	ClearPageReserved(virt_to_page(addr));
	375	init_page_count(virt_to_page(addr));
c967da6a	376	memset((void *)addr, POISON_FREE_INITMEM, PAGE_SIZE);
e5b2bb55 PE	377	free_page(addr);
	378	totalram_pages++;
	379	}
	380	#endif
	381	}
	382
	383	void free_initmem(void)
	384	{
	385	free_init_pages("unused kernel memory",
	386	(unsigned long)(&__init_begin),
	387	(unsigned long)(&__init_end));
	388	}
731ddea6 PE	389
	390	#ifdef CONFIG_BLK_DEV_INITRD
	391	void free_initrd_mem(unsigned long start, unsigned long end)
	392	{
c967da6a YL	393	/*
	394	* end could be not aligned, and We can not align that,
	395	* decompresser could be confused by aligned initrd_end
	396	* We already reserve the end partial page before in
	397	* - i386_start_kernel()
	398	* - x86_64_start_kernel()
	399	* - relocate_initrd()
	400	* So here We can do PAGE_ALIGN() safely to get partial page to be freed
	401	*/
	402	free_init_pages("initrd memory", start, PAGE_ALIGN(end));
731ddea6 PE	403	}
731ddea6 PE	404	#endif