Subversion Repositories shark

Rev

Go to most recent revision | Blame | Compare with Previous | Last modification | View Log | RSS feed

#ifndef _LINUX_MMZONE_H
#define _LINUX_MMZONE_H

#ifdef __KERNEL__
#ifndef __ASSEMBLY__

#include <linux/config.h>
#include <linux/spinlock.h>
#include <linux/list.h>
#include <linux/wait.h>
#include <linux/cache.h>
#include <linux/threads.h>
#include <linux/numa.h>
#include <asm/atomic.h>

/* Free memory management - zoned buddy allocator.  */
#ifndef CONFIG_FORCE_MAX_ZONEORDER
#define MAX_ORDER 11
#else
#define MAX_ORDER CONFIG_FORCE_MAX_ZONEORDER
#endif

struct free_area {
        struct list_head        free_list;
        unsigned long           *map;
};

struct pglist_data;

/*
 * zone->lock and zone->lru_lock are two of the hottest locks in the kernel.
 * So add a wild amount of padding here to ensure that they fall into separate
 * cachelines.  There are very few zone structures in the machine, so space
 * consumption is not a concern here.
 */

#if defined(CONFIG_SMP)
struct zone_padding {
        int x;
} ____cacheline_maxaligned_in_smp;
#define ZONE_PADDING(name)      struct zone_padding name;
#else
#define ZONE_PADDING(name)
#endif

struct per_cpu_pages {
        int count;              /* number of pages in the list */
        int low;                /* low watermark, refill needed */
        int high;               /* high watermark, emptying needed */
        int batch;              /* chunk size for buddy add/remove */
        struct list_head list;  /* the list of pages */
};

struct per_cpu_pageset {
        struct per_cpu_pages pcp[2];    /* 0: hot.  1: cold */
} ____cacheline_aligned_in_smp;

/*
 * On machines where it is needed (eg PCs) we divide physical memory
 * into multiple physical zones. On a PC we have 3 zones:
 *
 * ZONE_DMA       < 16 MB       ISA DMA capable memory
 * ZONE_NORMAL  16-896 MB       direct mapped by the kernel
 * ZONE_HIGHMEM  > 896 MB       only page cache and user processes
 */


struct zone {
        /*
         * Commonly accessed fields:
         */

        spinlock_t              lock;
        unsigned long           free_pages;
        unsigned long           pages_min, pages_low, pages_high;

        ZONE_PADDING(_pad1_)

        spinlock_t              lru_lock;      
        struct list_head        active_list;
        struct list_head        inactive_list;
        atomic_t                refill_counter;
        unsigned long           nr_active;
        unsigned long           nr_inactive;
        int                     all_unreclaimable; /* All pages pinned */
        unsigned long           pages_scanned;     /* since last reclaim */

        ZONE_PADDING(_pad2_)

        /*
         * prev_priority holds the scanning priority for this zone.  It is
         * defined as the scanning priority at which we achieved our reclaim
         * target at the previous try_to_free_pages() or balance_pgdat()
         * invokation.
         *
         * We use prev_priority as a measure of how much stress page reclaim is
         * under - it drives the swappiness decision: whether to unmap mapped
         * pages.
         *
         * temp_priority is used to remember the scanning priority at which
         * this zone was successfully refilled to free_pages == pages_high.
         *
         * Access to both these fields is quite racy even on uniprocessor.  But
         * it is expected to average out OK.
         */

        int temp_priority;
        int prev_priority;

        /*
         * free areas of different sizes
         */

        struct free_area        free_area[MAX_ORDER];

        /*
         * wait_table           -- the array holding the hash table
         * wait_table_size      -- the size of the hash table array
         * wait_table_bits      -- wait_table_size == (1 << wait_table_bits)
         *
         * The purpose of all these is to keep track of the people
         * waiting for a page to become available and make them
         * runnable again when possible. The trouble is that this
         * consumes a lot of space, especially when so few things
         * wait on pages at a given time. So instead of using
         * per-page waitqueues, we use a waitqueue hash table.
         *
         * The bucket discipline is to sleep on the same queue when
         * colliding and wake all in that wait queue when removing.
         * When something wakes, it must check to be sure its page is
         * truly available, a la thundering herd. The cost of a
         * collision is great, but given the expected load of the
         * table, they should be so rare as to be outweighed by the
         * benefits from the saved space.
         *
         * __wait_on_page_locked() and unlock_page() in mm/filemap.c, are the
         * primary users of these fields, and in mm/page_alloc.c
         * free_area_init_core() performs the initialization of them.
         */

        wait_queue_head_t       * wait_table;
        unsigned long           wait_table_size;
        unsigned long           wait_table_bits;

        ZONE_PADDING(_pad3_)

        struct per_cpu_pageset  pageset[NR_CPUS];

        /*
         * Discontig memory support fields.
         */

        struct pglist_data      *zone_pgdat;
        struct page             *zone_mem_map;
        /* zone_start_pfn == zone_start_paddr >> PAGE_SHIFT */
        unsigned long           zone_start_pfn;

        /*
         * rarely used fields:
         */

        char                    *name;
        unsigned long           spanned_pages;  /* total size, including holes */
        unsigned long           present_pages;  /* amount of memory (excluding holes) */
} ____cacheline_maxaligned_in_smp;

#define ZONE_DMA                0
#define ZONE_NORMAL             1
#define ZONE_HIGHMEM            2
#define MAX_NR_ZONES            3
#define GFP_ZONEMASK    0x03

/*
 * One allocation request operates on a zonelist. A zonelist
 * is a list of zones, the first one is the 'goal' of the
 * allocation, the other zones are fallback zones, in decreasing
 * priority.
 *
 * Right now a zonelist takes up less than a cacheline. We never
 * modify it apart from boot-up, and only a few indices are used,
 * so despite the zonelist table being relatively big, the cache
 * footprint of this construct is very small.
 */

struct zonelist {
        struct zone *zones[MAX_NUMNODES * MAX_NR_ZONES + 1]; // NULL delimited
};


/*
 * The pg_data_t structure is used in machines with CONFIG_DISCONTIGMEM
 * (mostly NUMA machines?) to denote a higher-level memory zone than the
 * zone denotes.
 *
 * On NUMA machines, each NUMA node would have a pg_data_t to describe
 * it's memory layout.
 *
 * Memory statistics and page replacement data structures are maintained on a
 * per-zone basis.
 */

struct bootmem_data;
typedef struct pglist_data {
        struct zone node_zones[MAX_NR_ZONES];
        struct zonelist node_zonelists[MAX_NR_ZONES];
        int nr_zones;
        struct page *node_mem_map;
        unsigned long *valid_addr_bitmap;
        struct bootmem_data *bdata;
        unsigned long node_start_pfn;
        unsigned long node_present_pages; /* total number of physical pages */
        unsigned long node_spanned_pages; /* total size of physical page
                                             range, including holes */

        int node_id;
        struct pglist_data *pgdat_next;
        wait_queue_head_t       kswapd_wait;
} pg_data_t;

#define node_present_pages(nid) (NODE_DATA(nid)->node_present_pages)
#define node_spanned_pages(nid) (NODE_DATA(nid)->node_spanned_pages)

extern int numnodes;
extern struct pglist_data *pgdat_list;

void get_zone_counts(unsigned long *active, unsigned long *inactive,
                        unsigned long *free);
void build_all_zonelists(void);
void wakeup_kswapd(struct zone *zone);

/**
 * for_each_pgdat - helper macro to iterate over all nodes
 * @pgdat - pointer to a pg_data_t variable
 *
 * Meant to help with common loops of the form
 * pgdat = pgdat_list;
 * while(pgdat) {
 *      ...
 *      pgdat = pgdat->pgdat_next;
 * }
 */

#define for_each_pgdat(pgdat) \
        for (pgdat = pgdat_list; pgdat; pgdat = pgdat->pgdat_next)


/*
 * next_zone - helper magic for for_each_zone()
 * Thanks to William Lee Irwin III for this piece of ingenuity.
 */

static inline struct zone *next_zone(struct zone *zone)
{
        pg_data_t *pgdat = zone->zone_pgdat;

        if (zone - pgdat->node_zones < MAX_NR_ZONES - 1)
                zone++;
        else if (pgdat->pgdat_next) {
                pgdat = pgdat->pgdat_next;
                zone = pgdat->node_zones;
        } else
                zone = NULL;

        return zone;
}

/**
 * for_each_zone - helper macro to iterate over all memory zones
 * @zone - pointer to struct zone variable
 *
 * The user only needs to declare the zone variable, for_each_zone
 * fills it in. This basically means for_each_zone() is an
 * easier to read version of this piece of code:
 *
 * for (pgdat = pgdat_list; pgdat; pgdat = pgdat->node_next)
 *      for (i = 0; i < MAX_NR_ZONES; ++i) {
 *              struct zone * z = pgdat->node_zones + i;
 *              ...
 *      }
 * }
 */

#define for_each_zone(zone) \
        for (zone = pgdat_list->node_zones; zone; zone = next_zone(zone))


/**
 * is_highmem - helper function to quickly check if a struct zone is a
 *              highmem zone or not.  This is an attempt to keep references
 *              to ZONE_{DMA/NORMAL/HIGHMEM/etc} in general code to a minimum.
 * @zone - pointer to struct zone variable
 */

static inline int is_highmem(struct zone *zone)
{
        return (zone - zone->zone_pgdat->node_zones == ZONE_HIGHMEM);
}

/* These two functions are used to setup the per zone pages min values */
struct ctl_table;
struct file;
int min_free_kbytes_sysctl_handler(struct ctl_table *, int, struct file *,
                                          void *, size_t *);
extern void setup_per_zone_pages_min(void);


#ifdef CONFIG_NUMA
#define MAX_NR_MEMBLKS  BITS_PER_LONG /* Max number of Memory Blocks */
#else /* !CONFIG_NUMA */
#define MAX_NR_MEMBLKS  1
#endif /* CONFIG_NUMA */

#include <linux/topology.h>
/* Returns the number of the current Node. */
#define numa_node_id()          (cpu_to_node(smp_processor_id()))

#ifndef CONFIG_DISCONTIGMEM

extern struct pglist_data contig_page_data;
#define NODE_DATA(nid)          (&contig_page_data)
#define NODE_MEM_MAP(nid)       mem_map
#define MAX_NODES_SHIFT         0

#else /* CONFIG_DISCONTIGMEM */

#include <asm/mmzone.h>

#if BITS_PER_LONG == 32
/*
 * with 32 bit flags field, page->zone is currently 8 bits.
 * there are 3 zones (2 bits) and this leaves 8-2=6 bits for nodes.
 */

#define MAX_NODES_SHIFT         6
#elif BITS_PER_LONG == 64
/*
 * with 64 bit flags field, there's plenty of room.
 */

#define MAX_NODES_SHIFT         10
#endif

#endif /* !CONFIG_DISCONTIGMEM */

#if NODES_SHIFT > MAX_NODES_SHIFT
#error NODES_SHIFT > MAX_NODES_SHIFT
#endif

extern DECLARE_BITMAP(node_online_map, MAX_NUMNODES);
extern DECLARE_BITMAP(memblk_online_map, MAX_NR_MEMBLKS);

#if defined(CONFIG_DISCONTIGMEM) || defined(CONFIG_NUMA)

#define node_online(node)       test_bit(node, node_online_map)
#define node_set_online(node)   set_bit(node, node_online_map)
#define node_set_offline(node)  clear_bit(node, node_online_map)
static inline unsigned int num_online_nodes(void)
{
        int i, num = 0;

        for(i = 0; i < MAX_NUMNODES; i++){
                if (node_online(i))
                        num++;
        }
        return num;
}

#define memblk_online(memblk)           test_bit(memblk, memblk_online_map)
#define memblk_set_online(memblk)       set_bit(memblk, memblk_online_map)
#define memblk_set_offline(memblk)      clear_bit(memblk, memblk_online_map)
static inline unsigned int num_online_memblks(void)
{
        int i, num = 0;

        for(i = 0; i < MAX_NR_MEMBLKS; i++){
                if (memblk_online(i))
                        num++;
        }
        return num;
}

#else /* !CONFIG_DISCONTIGMEM && !CONFIG_NUMA */

#define node_online(node) \
        ({ BUG_ON((node) != 0); test_bit(node, node_online_map); })

#define node_set_online(node) \
        ({ BUG_ON((node) != 0); set_bit(node, node_online_map); })

#define node_set_offline(node) \
        ({ BUG_ON((node) != 0); clear_bit(node, node_online_map); })

#define num_online_nodes()      1

#define memblk_online(memblk) \
        ({ BUG_ON((memblk) != 0); test_bit(memblk, memblk_online_map); })

#define memblk_set_online(memblk) \
        ({ BUG_ON((memblk) != 0); set_bit(memblk, memblk_online_map); })

#define memblk_set_offline(memblk) \
        ({ BUG_ON((memblk) != 0); clear_bit(memblk, memblk_online_map); })

#define num_online_memblks()            1

#endif /* CONFIG_DISCONTIGMEM || CONFIG_NUMA */
#endif /* !__ASSEMBLY__ */
#endif /* __KERNEL__ */
#endif /* _LINUX_MMZONE_H */