[linux-2.6-block.git] / arch / ia64 / mm / contig.c

/*
 * This file is subject to the terms and conditions of the GNU General Public
 * License.  See the file "COPYING" in the main directory of this archive
 * for more details.
 *
 * Copyright (C) 1998-2003 Hewlett-Packard Co
 *	David Mosberger-Tang <davidm@hpl.hp.com>
 *	Stephane Eranian <eranian@hpl.hp.com>
 * Copyright (C) 2000, Rohit Seth <rohit.seth@intel.com>
 * Copyright (C) 1999 VA Linux Systems
 * Copyright (C) 1999 Walt Drummond <drummond@valinux.com>
 * Copyright (C) 2003 Silicon Graphics, Inc. All rights reserved.
 *
 * Routines used by ia64 machines with contiguous (or virtually contiguous)
 * memory.
 */
#include <linux/bootmem.h>
#include <linux/efi.h>
#include <linux/memblock.h>
#include <linux/mm.h>
#include <linux/nmi.h>
#include <linux/swap.h>

#include <asm/meminit.h>
#include <asm/pgalloc.h>
#include <asm/pgtable.h>
#include <asm/sections.h>
#include <asm/mca.h>

#ifdef CONFIG_VIRTUAL_MEM_MAP
static unsigned long max_gap;
#endif

/* physical address where the bootmem map is located */
unsigned long bootmap_start;

/**
 * find_bootmap_location - callback to find a memory area for the bootmap
 * @start: start of region
 * @end: end of region
 * @arg: unused callback data
 *
 * Find a place to put the bootmap and return its starting address in
 * bootmap_start.  This address must be page-aligned.
 */
static int __init
find_bootmap_location (u64 start, u64 end, void *arg)
{
	u64 needed = *(unsigned long *)arg;
	u64 range_start, range_end, free_start;
	int i;

#if IGNORE_PFN0
	if (start == PAGE_OFFSET) {
		start += PAGE_SIZE;
		if (start >= end)
			return 0;
	}
#endif

	free_start = PAGE_OFFSET;

	for (i = 0; i < num_rsvd_regions; i++) {
		range_start = max(start, free_start);
		range_end   = min(end, rsvd_region[i].start & PAGE_MASK);

		free_start = PAGE_ALIGN(rsvd_region[i].end);

		if (range_end <= range_start)
			continue; /* skip over empty range */

		if (range_end - range_start >= needed) {
			bootmap_start = __pa(range_start);
			return -1;	/* done */
		}

		/* nothing more available in this segment */
		if (range_end == end)
			return 0;
	}
	return 0;
}

#ifdef CONFIG_SMP
static void *cpu_data;
/**
 * per_cpu_init - setup per-cpu variables
 *
 * Allocate and setup per-cpu data areas.
 */
void *per_cpu_init(void)
{
	static bool first_time = true;
	void *cpu0_data = __cpu0_per_cpu;
	unsigned int cpu;

	if (!first_time)
		goto skip;
	first_time = false;

	/*
	 * get_free_pages() cannot be used before cpu_init() done.
	 * BSP allocates PERCPU_PAGE_SIZE bytes for all possible CPUs
	 * to avoid that AP calls get_zeroed_page().
	 */
	for_each_possible_cpu(cpu) {
		void *src = cpu == 0 ? cpu0_data : __phys_per_cpu_start;

		memcpy(cpu_data, src, __per_cpu_end - __per_cpu_start);
		__per_cpu_offset[cpu] = (char *)cpu_data - __per_cpu_start;
		per_cpu(local_per_cpu_offset, cpu) = __per_cpu_offset[cpu];

		/*
		 * percpu area for cpu0 is moved from the __init area
		 * which is setup by head.S and used till this point.
		 * Update ar.k3.  This move is ensures that percpu
		 * area for cpu0 is on the correct node and its
		 * virtual address isn't insanely far from other
		 * percpu areas which is important for congruent
		 * percpu allocator.
		 */
		if (cpu == 0)
			ia64_set_kr(IA64_KR_PER_CPU_DATA, __pa(cpu_data) -
				    (unsigned long)__per_cpu_start);

		cpu_data += PERCPU_PAGE_SIZE;
	}
skip:
	return __per_cpu_start + __per_cpu_offset[smp_processor_id()];
}

static inline void
alloc_per_cpu_data(void)
{
	cpu_data = __alloc_bootmem(PERCPU_PAGE_SIZE * num_possible_cpus(),
				   PERCPU_PAGE_SIZE, __pa(MAX_DMA_ADDRESS));
}

/**
 * setup_per_cpu_areas - setup percpu areas
 *
 * Arch code has already allocated and initialized percpu areas.  All
 * this function has to do is to teach the determined layout to the
 * dynamic percpu allocator, which happens to be more complex than
 * creating whole new ones using helpers.
 */
void __init
setup_per_cpu_areas(void)
{
	struct pcpu_alloc_info *ai;
	struct pcpu_group_info *gi;
	unsigned int cpu;
	ssize_t static_size, reserved_size, dyn_size;
	int rc;

	ai = pcpu_alloc_alloc_info(1, num_possible_cpus());
	if (!ai)
		panic("failed to allocate pcpu_alloc_info");
	gi = &ai->groups[0];

	/* units are assigned consecutively to possible cpus */
	for_each_possible_cpu(cpu)
		gi->cpu_map[gi->nr_units++] = cpu;

	/* set parameters */
	static_size = __per_cpu_end - __per_cpu_start;
	reserved_size = PERCPU_MODULE_RESERVE;
	dyn_size = PERCPU_PAGE_SIZE - static_size - reserved_size;
	if (dyn_size < 0)
		panic("percpu area overflow static=%zd reserved=%zd\n",
		      static_size, reserved_size);

	ai->static_size		= static_size;
	ai->reserved_size	= reserved_size;
	ai->dyn_size		= dyn_size;
	ai->unit_size		= PERCPU_PAGE_SIZE;
	ai->atom_size		= PAGE_SIZE;
	ai->alloc_size		= PERCPU_PAGE_SIZE;

	rc = pcpu_setup_first_chunk(ai, __per_cpu_start + __per_cpu_offset[0]);
	if (rc)
		panic("failed to setup percpu area (err=%d)", rc);

	pcpu_free_alloc_info(ai);
}
#else
#define alloc_per_cpu_data() do { } while (0)
#endif /* CONFIG_SMP */

/**
 * find_memory - setup memory map
 *
 * Walk the EFI memory map and find usable memory for the system, taking
 * into account reserved areas.
 */
void __init
find_memory (void)
{
	unsigned long bootmap_size;

	reserve_memory();

	/* first find highest page frame number */
	min_low_pfn = ~0UL;
	max_low_pfn = 0;
	efi_memmap_walk(find_max_min_low_pfn, NULL);
	max_pfn = max_low_pfn;
	/* how many bytes to cover all the pages */
	bootmap_size = bootmem_bootmap_pages(max_pfn) << PAGE_SHIFT;

	/* look for a location to hold the bootmap */
	bootmap_start = ~0UL;
	efi_memmap_walk(find_bootmap_location, &bootmap_size);
	if (bootmap_start == ~0UL)
		panic("Cannot find %ld bytes for bootmap\n", bootmap_size);

	bootmap_size = init_bootmem_node(NODE_DATA(0),
			(bootmap_start >> PAGE_SHIFT), 0, max_pfn);

	/* Free all available memory, then mark bootmem-map as being in use. */
	efi_memmap_walk(filter_rsvd_memory, free_bootmem);
	reserve_bootmem(bootmap_start, bootmap_size, BOOTMEM_DEFAULT);

	find_initrd();

	alloc_per_cpu_data();
}

/*
 * Set up the page tables.
 */

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

	memset(max_zone_pfns, 0, sizeof(max_zone_pfns));
#ifdef CONFIG_ZONE_DMA32
	max_dma = virt_to_phys((void *) MAX_DMA_ADDRESS) >> PAGE_SHIFT;
	max_zone_pfns[ZONE_DMA32] = max_dma;
#endif
	max_zone_pfns[ZONE_NORMAL] = max_low_pfn;

#ifdef CONFIG_VIRTUAL_MEM_MAP
	efi_memmap_walk(filter_memory, register_active_ranges);
	efi_memmap_walk(find_largest_hole, (u64 *)&max_gap);
	if (max_gap < LARGE_GAP) {
		vmem_map = (struct page *) 0;
		free_area_init_nodes(max_zone_pfns);
	} else {
		unsigned long map_size;

		/* allocate virtual_mem_map */

		map_size = PAGE_ALIGN(ALIGN(max_low_pfn, MAX_ORDER_NR_PAGES) *
			sizeof(struct page));
		VMALLOC_END -= map_size;
		vmem_map = (struct page *) VMALLOC_END;
		efi_memmap_walk(create_mem_map_page_table, NULL);

		/*
		 * alloc_node_mem_map makes an adjustment for mem_map
		 * which isn't compatible with vmem_map.
		 */
		NODE_DATA(0)->node_mem_map = vmem_map +
			find_min_pfn_with_active_regions();
		free_area_init_nodes(max_zone_pfns);

		printk("Virtual mem_map starts at 0x%p\n", mem_map);
	}
#else /* !CONFIG_VIRTUAL_MEM_MAP */
	memblock_add_node(0, PFN_PHYS(max_low_pfn), 0);
	free_area_init_nodes(max_zone_pfns);
#endif /* !CONFIG_VIRTUAL_MEM_MAP */
	zero_page_memmap_ptr = virt_to_page(ia64_imva(empty_zero_page));
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* This file is subject to the terms and conditions of the GNU General Public
	3	* License. See the file "COPYING" in the main directory of this archive
	4	* for more details.
	5	*
	6	* Copyright (C) 1998-2003 Hewlett-Packard Co
	7	* David Mosberger-Tang <davidm@hpl.hp.com>
	8	* Stephane Eranian <eranian@hpl.hp.com>
	9	* Copyright (C) 2000, Rohit Seth <rohit.seth@intel.com>
	10	* Copyright (C) 1999 VA Linux Systems
	11	* Copyright (C) 1999 Walt Drummond <drummond@valinux.com>
	12	* Copyright (C) 2003 Silicon Graphics, Inc. All rights reserved.
	13	*
	14	* Routines used by ia64 machines with contiguous (or virtually contiguous)
	15	* memory.
	16	*/
1da177e4 LT	17	#include <linux/bootmem.h>
1da177e4 LT	18	#include <linux/efi.h>
98e4ae8a	19	#include <linux/memblock.h>
1da177e4	20	#include <linux/mm.h>
99a19cf1	21	#include <linux/nmi.h>
1da177e4 LT	22	#include <linux/swap.h>
	23
	24	#include <asm/meminit.h>
	25	#include <asm/pgalloc.h>
	26	#include <asm/pgtable.h>
	27	#include <asm/sections.h>
	28	#include <asm/mca.h>
	29
	30	#ifdef CONFIG_VIRTUAL_MEM_MAP
e44e41d0	31	static unsigned long max_gap;
1da177e4 LT	32	#endif
1da177e4 LT	33
1da177e4 LT	34	/* physical address where the bootmem map is located */
	35	unsigned long bootmap_start;
	36
1da177e4 LT	37	/**
	38	* find_bootmap_location - callback to find a memory area for the bootmap
	39	* @start: start of region
	40	* @end: end of region
	41	* @arg: unused callback data
	42	*
	43	* Find a place to put the bootmap and return its starting address in
	44	* bootmap_start. This address must be page-aligned.
	45	*/
dae28066	46	static int __init
e088a4ad	47	find_bootmap_location (u64 start, u64 end, void *arg)
1da177e4	48	{
e088a4ad MW	49	u64 needed = (unsigned long )arg;
e088a4ad MW	50	u64 range_start, range_end, free_start;
1da177e4 LT	51	int i;
	52
	53	#if IGNORE_PFN0
	54	if (start == PAGE_OFFSET) {
	55	start += PAGE_SIZE;
	56	if (start >= end)
	57	return 0;
	58	}
	59	#endif
	60
	61	free_start = PAGE_OFFSET;
	62
	63	for (i = 0; i < num_rsvd_regions; i++) {
	64	range_start = max(start, free_start);
	65	range_end = min(end, rsvd_region[i].start & PAGE_MASK);
	66
	67	free_start = PAGE_ALIGN(rsvd_region[i].end);
	68
	69	if (range_end <= range_start)
	70	continue; /* skip over empty range */
	71
	72	if (range_end - range_start >= needed) {
	73	bootmap_start = __pa(range_start);
	74	return -1; /* done */
	75	}
	76
	77	/* nothing more available in this segment */
	78	if (range_end == end)
	79	return 0;
	80	}
	81	return 0;
	82	}
	83
4b9ddc7c TL	84	#ifdef CONFIG_SMP
	85	static void *cpu_data;
	86	/**
	87	* per_cpu_init - setup per-cpu variables
	88	*
	89	* Allocate and setup per-cpu data areas.
	90	*/
ccce9bb8	91	void *per_cpu_init(void)
4b9ddc7c	92	{
36886478 TH	93	static bool first_time = true;
	94	void *cpu0_data = __cpu0_per_cpu;
	95	unsigned int cpu;
	96
	97	if (!first_time)
	98	goto skip;
	99	first_time = false;
4b9ddc7c TL	100
4b9ddc7c TL	101	/*
52594762 TH	102	* get_free_pages() cannot be used before cpu_init() done.
	103	* BSP allocates PERCPU_PAGE_SIZE bytes for all possible CPUs
	104	* to avoid that AP calls get_zeroed_page().
4b9ddc7c	105	*/
52594762	106	for_each_possible_cpu(cpu) {
36886478 TH	107	void *src = cpu == 0 ? cpu0_data : __phys_per_cpu_start;
	108
	109	memcpy(cpu_data, src, __per_cpu_end - __per_cpu_start);
	110	__per_cpu_offset[cpu] = (char *)cpu_data - __per_cpu_start;
	111	per_cpu(local_per_cpu_offset, cpu) = __per_cpu_offset[cpu];
	112
	113	/*
	114	* percpu area for cpu0 is moved from the __init area
	115	* which is setup by head.S and used till this point.
	116	* Update ar.k3. This move is ensures that percpu
	117	* area for cpu0 is on the correct node and its
	118	* virtual address isn't insanely far from other
	119	* percpu areas which is important for congruent
	120	* percpu allocator.
	121	*/
	122	if (cpu == 0)
	123	ia64_set_kr(IA64_KR_PER_CPU_DATA, __pa(cpu_data) -
	124	(unsigned long)__per_cpu_start);
	125
	126	cpu_data += PERCPU_PAGE_SIZE;
4b9ddc7c	127	}
36886478	128	skip:
4b9ddc7c TL	129	return __per_cpu_start + __per_cpu_offset[smp_processor_id()];
	130	}
	131
	132	static inline void
	133	alloc_per_cpu_data(void)
	134	{
52594762	135	cpu_data = __alloc_bootmem(PERCPU_PAGE_SIZE * num_possible_cpus(),
4b9ddc7c TL	136	PERCPU_PAGE_SIZE, __pa(MAX_DMA_ADDRESS));
4b9ddc7c TL	137	}
52594762 TH	138
	139	/**
	140	* setup_per_cpu_areas - setup percpu areas
	141	*
	142	* Arch code has already allocated and initialized percpu areas. All
	143	* this function has to do is to teach the determined layout to the
	144	* dynamic percpu allocator, which happens to be more complex than
	145	* creating whole new ones using helpers.
	146	*/
	147	void __init
	148	setup_per_cpu_areas(void)
	149	{
	150	struct pcpu_alloc_info *ai;
	151	struct pcpu_group_info *gi;
	152	unsigned int cpu;
	153	ssize_t static_size, reserved_size, dyn_size;
	154	int rc;
	155
	156	ai = pcpu_alloc_alloc_info(1, num_possible_cpus());
	157	if (!ai)
	158	panic("failed to allocate pcpu_alloc_info");
	159	gi = &ai->groups[0];
	160
	161	/* units are assigned consecutively to possible cpus */
	162	for_each_possible_cpu(cpu)
	163	gi->cpu_map[gi->nr_units++] = cpu;
	164
	165	/* set parameters */
	166	static_size = __per_cpu_end - __per_cpu_start;
	167	reserved_size = PERCPU_MODULE_RESERVE;
	168	dyn_size = PERCPU_PAGE_SIZE - static_size - reserved_size;
	169	if (dyn_size < 0)
	170	panic("percpu area overflow static=%zd reserved=%zd\n",
	171	static_size, reserved_size);
	172
	173	ai->static_size = static_size;
	174	ai->reserved_size = reserved_size;
	175	ai->dyn_size = dyn_size;
	176	ai->unit_size = PERCPU_PAGE_SIZE;
	177	ai->atom_size = PAGE_SIZE;
	178	ai->alloc_size = PERCPU_PAGE_SIZE;
	179
	180	rc = pcpu_setup_first_chunk(ai, __per_cpu_start + __per_cpu_offset[0]);
	181	if (rc)
	182	panic("failed to setup percpu area (err=%d)", rc);
	183
	184	pcpu_free_alloc_info(ai);
	185	}
4b9ddc7c TL	186	#else
	187	#define alloc_per_cpu_data() do { } while (0)
	188	#endif /* CONFIG_SMP */
	189
1da177e4 LT	190	/**
	191	* find_memory - setup memory map
	192	*
	193	* Walk the EFI memory map and find usable memory for the system, taking
	194	* into account reserved areas.
	195	*/
dae28066	196	void __init
1da177e4 LT	197	find_memory (void)
	198	{
	199	unsigned long bootmap_size;
	200
	201	reserve_memory();
	202
	203	/* first find highest page frame number */
a3f5c338 ZN	204	min_low_pfn = ~0UL;
	205	max_low_pfn = 0;
	206	efi_memmap_walk(find_max_min_low_pfn, NULL);
	207	max_pfn = max_low_pfn;
1da177e4 LT	208	/* how many bytes to cover all the pages */
	209	bootmap_size = bootmem_bootmap_pages(max_pfn) << PAGE_SHIFT;
	210
	211	/* look for a location to hold the bootmap */
	212	bootmap_start = ~0UL;
	213	efi_memmap_walk(find_bootmap_location, &bootmap_size);
	214	if (bootmap_start == ~0UL)
	215	panic("Cannot find %ld bytes for bootmap\n", bootmap_size);
	216
a3f5c338 ZN	217	bootmap_size = init_bootmem_node(NODE_DATA(0),
a3f5c338 ZN	218	(bootmap_start >> PAGE_SHIFT), 0, max_pfn);
1da177e4 LT	219
	220	/* Free all available memory, then mark bootmem-map as being in use. */
	221	efi_memmap_walk(filter_rsvd_memory, free_bootmem);
72a7fe39	222	reserve_bootmem(bootmap_start, bootmap_size, BOOTMEM_DEFAULT);
1da177e4 LT	223
1da177e4 LT	224	find_initrd();
45a98fc6	225
4b9ddc7c	226	alloc_per_cpu_data();
1da177e4 LT	227	}
1da177e4 LT	228
1da177e4 LT	229	/*
	230	* Set up the page tables.
	231	*/
	232
dae28066	233	void __init
1da177e4 LT	234	paging_init (void)
	235	{
	236	unsigned long max_dma;
05e0caad	237	unsigned long max_zone_pfns[MAX_NR_ZONES];
1da177e4	238
6391af17	239	memset(max_zone_pfns, 0, sizeof(max_zone_pfns));
d5c23ebf	240	#ifdef CONFIG_ZONE_DMA32
09ae1f58	241	max_dma = virt_to_phys((void *) MAX_DMA_ADDRESS) >> PAGE_SHIFT;
d5c23ebf	242	max_zone_pfns[ZONE_DMA32] = max_dma;
09ae1f58	243	#endif
05e0caad	244	max_zone_pfns[ZONE_NORMAL] = max_low_pfn;
1da177e4 LT	245
1da177e4 LT	246	#ifdef CONFIG_VIRTUAL_MEM_MAP
98075d24	247	efi_memmap_walk(filter_memory, register_active_ranges);
1da177e4 LT	248	efi_memmap_walk(find_largest_hole, (u64 *)&max_gap);
	249	if (max_gap < LARGE_GAP) {
	250	vmem_map = (struct page *) 0;
05e0caad	251	free_area_init_nodes(max_zone_pfns);
1da177e4 LT	252	} else {
	253	unsigned long map_size;
	254
	255	/* allocate virtual_mem_map */
	256
921eea1c BP	257	map_size = PAGE_ALIGN(ALIGN(max_low_pfn, MAX_ORDER_NR_PAGES) *
921eea1c BP	258	sizeof(struct page));
126b3fcd TH	259	VMALLOC_END -= map_size;
126b3fcd TH	260	vmem_map = (struct page *) VMALLOC_END;
1da177e4 LT	261	efi_memmap_walk(create_mem_map_page_table, NULL);
1da177e4 LT	262
05e0caad MG	263	/*
	264	* alloc_node_mem_map makes an adjustment for mem_map
	265	* which isn't compatible with vmem_map.
	266	*/
	267	NODE_DATA(0)->node_mem_map = vmem_map +
	268	find_min_pfn_with_active_regions();
	269	free_area_init_nodes(max_zone_pfns);
1da177e4 LT	270
	271	printk("Virtual mem_map starts at 0x%p\n", mem_map);
	272	}
	273	#else /* !CONFIG_VIRTUAL_MEM_MAP */
98e4ae8a	274	memblock_add_node(0, PFN_PHYS(max_low_pfn), 0);
05e0caad	275	free_area_init_nodes(max_zone_pfns);
1da177e4 LT	276	#endif /* !CONFIG_VIRTUAL_MEM_MAP */
	277	zero_page_memmap_ptr = virt_to_page(ia64_imva(empty_zero_page));
	278	}