// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
 * Written by Alex Tomas <alex@clusterfs.com>
 */


/*
 * mballoc.c contains the multiblocks allocation routines
 */

#include "ext4_jbd2.h"
#include "mballoc.h"
#include <linux/log2.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/nospec.h>
#include <linux/backing-dev.h>
#include <trace/events/ext4.h>

/*
 * MUSTDO:
 *   - test ext4_ext_search_left() and ext4_ext_search_right()
 *   - search for metadata in few groups
 *
 * TODO v4:
 *   - normalization should take into account whether file is still open
 *   - discard preallocations if no free space left (policy?)
 *   - don't normalize tails
 *   - quota
 *   - reservation for superuser
 *
 * TODO v3:
 *   - bitmap read-ahead (proposed by Oleg Drokin aka green)
 *   - track min/max extents in each group for better group selection
 *   - mb_mark_used() may allocate chunk right after splitting buddy
 *   - tree of groups sorted by number of free blocks
 *   - error handling
 */

/*
 * The allocation request involve request for multiple number of blocks
 * near to the goal(block) value specified.
 *
 * During initialization phase of the allocator we decide to use the
 * group preallocation or inode preallocation depending on the size of
 * the file. The size of the file could be the resulting file size we
 * would have after allocation, or the current file size, which ever
 * is larger. If the size is less than sbi->s_mb_stream_request we
 * select to use the group preallocation. The default value of
 * s_mb_stream_request is 16 blocks. This can also be tuned via
 * /sys/fs/ext4/<partition>/mb_stream_req. The value is represented in
 * terms of number of blocks.
 *
 * The main motivation for having small file use group preallocation is to
 * ensure that we have small files closer together on the disk.
 *
 * First stage the allocator looks at the inode prealloc list,
 * ext4_inode_info->i_prealloc_list, which contains list of prealloc
 * spaces for this particular inode. The inode prealloc space is
 * represented as:
 *
 * pa_lstart -> the logical start block for this prealloc space
 * pa_pstart -> the physical start block for this prealloc space
 * pa_len    -> length for this prealloc space (in clusters)
 * pa_free   ->  free space available in this prealloc space (in clusters)
 *
 * The inode preallocation space is used looking at the _logical_ start
 * block. If only the logical file block falls within the range of prealloc
 * space we will consume the particular prealloc space. This makes sure that
 * we have contiguous physical blocks representing the file blocks
 *
 * The important thing to be noted in case of inode prealloc space is that
 * we don't modify the values associated to inode prealloc space except
 * pa_free.
 *
 * If we are not able to find blocks in the inode prealloc space and if we
 * have the group allocation flag set then we look at the locality group
 * prealloc space. These are per CPU prealloc list represented as
 *
 * ext4_sb_info.s_locality_groups[smp_processor_id()]
 *
 * The reason for having a per cpu locality group is to reduce the contention
 * between CPUs. It is possible to get scheduled at this point.
 *
 * The locality group prealloc space is used looking at whether we have
 * enough free space (pa_free) within the prealloc space.
 *
 * If we can't allocate blocks via inode prealloc or/and locality group
 * prealloc then we look at the buddy cache. The buddy cache is represented
 * by ext4_sb_info.s_buddy_cache (struct inode) whose file offset gets
 * mapped to the buddy and bitmap information regarding different
 * groups. The buddy information is attached to buddy cache inode so that
 * we can access them through the page cache. The information regarding
 * each group is loaded via ext4_mb_load_buddy.  The information involve
 * block bitmap and buddy information. The information are stored in the
 * inode as:
 *
 *  {                        page                        }
 *  [ group 0 bitmap][ group 0 buddy] [group 1][ group 1]...
 *
 *
 * one block each for bitmap and buddy information.  So for each group we
 * take up 2 blocks. A page can contain blocks_per_page (PAGE_SIZE /
 * blocksize) blocks.  So it can have information regarding groups_per_page
 * which is blocks_per_page/2
 *
 * The buddy cache inode is not stored on disk. The inode is thrown
 * away when the filesystem is unmounted.
 *
 * We look for count number of blocks in the buddy cache. If we were able
 * to locate that many free blocks we return with additional information
 * regarding rest of the contiguous physical block available
 *
 * Before allocating blocks via buddy cache we normalize the request
 * blocks. This ensure we ask for more blocks that we needed. The extra
 * blocks that we get after allocation is added to the respective prealloc
 * list. In case of inode preallocation we follow a list of heuristics
 * based on file size. This can be found in ext4_mb_normalize_request. If
 * we are doing a group prealloc we try to normalize the request to
 * sbi->s_mb_group_prealloc.  The default value of s_mb_group_prealloc is
 * dependent on the cluster size; for non-bigalloc file systems, it is
 * 512 blocks. This can be tuned via
 * /sys/fs/ext4/<partition>/mb_group_prealloc. The value is represented in
 * terms of number of blocks. If we have mounted the file system with -O
 * stripe=<value> option the group prealloc request is normalized to the
 * smallest multiple of the stripe value (sbi->s_stripe) which is
 * greater than the default mb_group_prealloc.
 *
 * If "mb_optimize_scan" mount option is set, we maintain in memory group info
 * structures in two data structures:
 *
 * 1) Array of largest free order lists (sbi->s_mb_largest_free_orders)
 *
 *    Locking: sbi->s_mb_largest_free_orders_locks(array of rw locks)
 *
 *    This is an array of lists where the index in the array represents the
 *    largest free order in the buddy bitmap of the participating group infos of
 *    that list. So, there are exactly MB_NUM_ORDERS(sb) (which means total
 *    number of buddy bitmap orders possible) number of lists. Group-infos are
 *    placed in appropriate lists.
 *
 * 2) Average fragment size rb tree (sbi->s_mb_avg_fragment_size_root)
 *
 *    Locking: sbi->s_mb_rb_lock (rwlock)
 *
 *    This is a red black tree consisting of group infos and the tree is sorted
 *    by average fragment sizes (which is calculated as ext4_group_info->bb_free
 *    / ext4_group_info->bb_fragments).
 *
 * When "mb_optimize_scan" mount option is set, mballoc consults the above data
 * structures to decide the order in which groups are to be traversed for
 * fulfilling an allocation request.
 *
 * At CR = 0, we look for groups which have the largest_free_order >= the order
 * of the request. We directly look at the largest free order list in the data
 * structure (1) above where largest_free_order = order of the request. If that
 * list is empty, we look at remaining list in the increasing order of
 * largest_free_order. This allows us to perform CR = 0 lookup in O(1) time.
 *
 * At CR = 1, we only consider groups where average fragment size > request
 * size. So, we lookup a group which has average fragment size just above or
 * equal to request size using our rb tree (data structure 2) in O(log N) time.
 *
 * If "mb_optimize_scan" mount option is not set, mballoc traverses groups in
 * linear order which requires O(N) search time for each CR 0 and CR 1 phase.
 *
 * The regular allocator (using the buddy cache) supports a few tunables.
 *
 * /sys/fs/ext4/<partition>/mb_min_to_scan
 * /sys/fs/ext4/<partition>/mb_max_to_scan
 * /sys/fs/ext4/<partition>/mb_order2_req
 * /sys/fs/ext4/<partition>/mb_linear_limit
 *
 * The regular allocator uses buddy scan only if the request len is power of
 * 2 blocks and the order of allocation is >= sbi->s_mb_order2_reqs. The
 * value of s_mb_order2_reqs can be tuned via
 * /sys/fs/ext4/<partition>/mb_order2_req.  If the request len is equal to
 * stripe size (sbi->s_stripe), we try to search for contiguous block in
 * stripe size. This should result in better allocation on RAID setups. If
 * not, we search in the specific group using bitmap for best extents. The
 * tunable min_to_scan and max_to_scan control the behaviour here.
 * min_to_scan indicate how long the mballoc __must__ look for a best
 * extent and max_to_scan indicates how long the mballoc __can__ look for a
 * best extent in the found extents. Searching for the blocks starts with
 * the group specified as the goal value in allocation context via
 * ac_g_ex. Each group is first checked based on the criteria whether it
 * can be used for allocation. ext4_mb_good_group explains how the groups are
 * checked.
 *
 * When "mb_optimize_scan" is turned on, as mentioned above, the groups may not
 * get traversed linearly. That may result in subsequent allocations being not
 * close to each other. And so, the underlying device may get filled up in a
 * non-linear fashion. While that may not matter on non-rotational devices, for
 * rotational devices that may result in higher seek times. "mb_linear_limit"
 * tells mballoc how many groups mballoc should search linearly before
 * performing consulting above data structures for more efficient lookups. For
 * non rotational devices, this value defaults to 0 and for rotational devices
 * this is set to MB_DEFAULT_LINEAR_LIMIT.
 *
 * Both the prealloc space are getting populated as above. So for the first
 * request we will hit the buddy cache which will result in this prealloc
 * space getting filled. The prealloc space is then later used for the
 * subsequent request.
 */

/*
 * mballoc operates on the following data:
 *  - on-disk bitmap
 *  - in-core buddy (actually includes buddy and bitmap)
 *  - preallocation descriptors (PAs)
 *
 * there are two types of preallocations:
 *  - inode
 *    assiged to specific inode and can be used for this inode only.
 *    it describes part of inode's space preallocated to specific
 *    physical blocks. any block from that preallocated can be used
 *    independent. the descriptor just tracks number of blocks left
 *    unused. so, before taking some block from descriptor, one must
 *    make sure corresponded logical block isn't allocated yet. this
 *    also means that freeing any block within descriptor's range
 *    must discard all preallocated blocks.
 *  - locality group
 *    assigned to specific locality group which does not translate to
 *    permanent set of inodes: inode can join and leave group. space
 *    from this type of preallocation can be used for any inode. thus
 *    it's consumed from the beginning to the end.
 *
 * relation between them can be expressed as:
 *    in-core buddy = on-disk bitmap + preallocation descriptors
 *
 * this mean blocks mballoc considers used are:
 *  - allocated blocks (persistent)
 *  - preallocated blocks (non-persistent)
 *
 * consistency in mballoc world means that at any time a block is either
 * free or used in ALL structures. notice: "any time" should not be read
 * literally -- time is discrete and delimited by locks.
 *
 *  to keep it simple, we don't use block numbers, instead we count number of
 *  blocks: how many blocks marked used/free in on-disk bitmap, buddy and PA.
 *
 * all operations can be expressed as:
 *  - init buddy:                       buddy = on-disk + PAs
 *  - new PA:                           buddy += N; PA = N
 *  - use inode PA:                     on-disk += N; PA -= N
 *  - discard inode PA                  buddy -= on-disk - PA; PA = 0
 *  - use locality group PA             on-disk += N; PA -= N
 *  - discard locality group PA         buddy -= PA; PA = 0
 *  note: 'buddy -= on-disk - PA' is used to show that on-disk bitmap
 *        is used in real operation because we can't know actual used
 *        bits from PA, only from on-disk bitmap
 *
 * if we follow this strict logic, then all operations above should be atomic.
 * given some of them can block, we'd have to use something like semaphores
 * killing performance on high-end SMP hardware. let's try to relax it using
 * the following knowledge:
 *  1) if buddy is referenced, it's already initialized
 *  2) while block is used in buddy and the buddy is referenced,
 *     nobody can re-allocate that block
 *  3) we work on bitmaps and '+' actually means 'set bits'. if on-disk has
 *     bit set and PA claims same block, it's OK. IOW, one can set bit in
 *     on-disk bitmap if buddy has same bit set or/and PA covers corresponded
 *     block
 *
 * so, now we're building a concurrency table:
 *  - init buddy vs.
 *    - new PA
 *      blocks for PA are allocated in the buddy, buddy must be referenced
 *      until PA is linked to allocation group to avoid concurrent buddy init
 *    - use inode PA
 *      we need to make sure that either on-disk bitmap or PA has uptodate data
 *      given (3) we care that PA-=N operation doesn't interfere with init
 *    - discard inode PA
 *      the simplest way would be to have buddy initialized by the discard
 *    - use locality group PA
 *      again PA-=N must be serialized with init
 *    - discard locality group PA
 *      the simplest way would be to have buddy initialized by the discard
 *  - new PA vs.
 *    - use inode PA
 *      i_data_sem serializes them
 *    - discard inode PA
 *      discard process must wait until PA isn't used by another process
 *    - use locality group PA
 *      some mutex should serialize them
 *    - discard locality group PA
 *      discard process must wait until PA isn't used by another process
 *  - use inode PA
 *    - use inode PA
 *      i_data_sem or another mutex should serializes them
 *    - discard inode PA
 *      discard process must wait until PA isn't used by another process
 *    - use locality group PA
 *      nothing wrong here -- they're different PAs covering different blocks
 *    - discard locality group PA
 *      discard process must wait until PA isn't used by another process
 *
 * now we're ready to make few consequences:
 *  - PA is referenced and while it is no discard is possible
 *  - PA is referenced until block isn't marked in on-disk bitmap
 *  - PA changes only after on-disk bitmap
 *  - discard must not compete with init. either init is done before
 *    any discard or they're serialized somehow
 *  - buddy init as sum of on-disk bitmap and PAs is done atomically
 *
 * a special case when we've used PA to emptiness. no need to modify buddy
 * in this case, but we should care about concurrent init
 *
 */

 /*
 * Logic in few words:
 *
 *  - allocation:
 *    load group
 *    find blocks
 *    mark bits in on-disk bitmap
 *    release group
 *
 *  - use preallocation:
 *    find proper PA (per-inode or group)
 *    load group
 *    mark bits in on-disk bitmap
 *    release group
 *    release PA
 *
 *  - free:
 *    load group
 *    mark bits in on-disk bitmap
 *    release group
 *
 *  - discard preallocations in group:
 *    mark PAs deleted
 *    move them onto local list
 *    load on-disk bitmap
 *    load group
 *    remove PA from object (inode or locality group)
 *    mark free blocks in-core
 *
 *  - discard inode's preallocations:
 */

/*
 * Locking rules
 *
 * Locks:
 *  - bitlock on a group        (group)
 *  - object (inode/locality)   (object)
 *  - per-pa lock               (pa)
 *  - cr0 lists lock            (cr0)
 *  - cr1 tree lock             (cr1)
 *
 * Paths:
 *  - new pa
 *    object
 *    group
 *
 *  - find and use pa:
 *    pa
 *
 *  - release consumed pa:
 *    pa
 *    group
 *    object
 *
 *  - generate in-core bitmap:
 *    group
 *        pa
 *
 *  - discard all for given object (inode, locality group):
 *    object
 *        pa
 *    group
 *
 *  - discard all for given group:
 *    group
 *        pa
 *    group
 *        object
 *
 *  - allocation path (ext4_mb_regular_allocator)
 *    group
 *    cr0/cr1
 */
static struct kmem_cache *ext4_pspace_cachep;
static struct kmem_cache *ext4_ac_cachep;
static struct kmem_cache *ext4_free_data_cachep;

/* We create slab caches for groupinfo data structures based on the
 * superblock block size.  There will be one per mounted filesystem for
 * each unique s_blocksize_bits */
#define NR_GRPINFO_CACHES 8
static struct kmem_cache *ext4_groupinfo_caches[NR_GRPINFO_CACHES];

static const char * const ext4_groupinfo_slab_names[NR_GRPINFO_CACHES] = {
        "ext4_groupinfo_1k", "ext4_groupinfo_2k", "ext4_groupinfo_4k",
        "ext4_groupinfo_8k", "ext4_groupinfo_16k", "ext4_groupinfo_32k",
        "ext4_groupinfo_64k", "ext4_groupinfo_128k"
};

static void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap,
                                        ext4_group_t group);
static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap,
                                                ext4_group_t group);
static void ext4_mb_new_preallocation(struct ext4_allocation_context *ac);

static bool ext4_mb_good_group(struct ext4_allocation_context *ac,
                               ext4_group_t group, int cr);

/*
 * The algorithm using this percpu seq counter goes below:
 * 1. We sample the percpu discard_pa_seq counter before trying for block
 *    allocation in ext4_mb_new_blocks().
 * 2. We increment this percpu discard_pa_seq counter when we either allocate
 *    or free these blocks i.e. while marking those blocks as used/free in
 *    mb_mark_used()/mb_free_blocks().
 * 3. We also increment this percpu seq counter when we successfully identify
 *    that the bb_prealloc_list is not empty and hence proceed for discarding
 *    of those PAs inside ext4_mb_discard_group_preallocations().
 *
 * Now to make sure that the regular fast path of block allocation is not
 * affected, as a small optimization we only sample the percpu seq counter
 * on that cpu. Only when the block allocation fails and when freed blocks
 * found were 0, that is when we sample percpu seq counter for all cpus using
 * below function ext4_get_discard_pa_seq_sum(). This happens after making
 * sure that all the PAs on grp->bb_prealloc_list got freed or if it's empty.
 */
static DEFINE_PER_CPU(u64, discard_pa_seq);
static inline u64 ext4_get_discard_pa_seq_sum(void)
{
        int __cpu;
        u64 __seq = 0;

        for_each_possible_cpu(__cpu)
                __seq += per_cpu(discard_pa_seq, __cpu);
        return __seq;
}

static inline void *mb_correct_addr_and_bit(int *bit, void *addr)
{
#if BITS_PER_LONG == 64
        *bit += ((unsigned long) addr & 7UL) << 3;
        addr = (void *) ((unsigned long) addr & ~7UL);
#elif BITS_PER_LONG == 32
        *bit += ((unsigned long) addr & 3UL) << 3;
        addr = (void *) ((unsigned long) addr & ~3UL);
#else
#error "how many bits you are?!"
#endif
        return addr;
}

static inline int mb_test_bit(int bit, void *addr)
{
        /*
         * ext4_test_bit on architecture like powerpc
         * needs unsigned long aligned address
         */
        addr = mb_correct_addr_and_bit(&bit, addr);
        return ext4_test_bit(bit, addr);
}

static inline void mb_set_bit(int bit, void *addr)
{
        addr = mb_correct_addr_and_bit(&bit, addr);
        ext4_set_bit(bit, addr);
}

static inline void mb_clear_bit(int bit, void *addr)
{
        addr = mb_correct_addr_and_bit(&bit, addr);
        ext4_clear_bit(bit, addr);
}

static inline int mb_test_and_clear_bit(int bit, void *addr)
{
        addr = mb_correct_addr_and_bit(&bit, addr);
        return ext4_test_and_clear_bit(bit, addr);
}

static inline int mb_find_next_zero_bit(void *addr, int max, int start)
{
        int fix = 0, ret, tmpmax;
        addr = mb_correct_addr_and_bit(&fix, addr);
        tmpmax = max + fix;
        start += fix;

        ret = ext4_find_next_zero_bit(addr, tmpmax, start) - fix;
        if (ret > max)
                return max;
        return ret;
}

static inline int mb_find_next_bit(void *addr, int max, int start)
{
        int fix = 0, ret, tmpmax;
        addr = mb_correct_addr_and_bit(&fix, addr);
        tmpmax = max + fix;
        start += fix;

        ret = ext4_find_next_bit(addr, tmpmax, start) - fix;
        if (ret > max)
                return max;
        return ret;
}

static void *mb_find_buddy(struct ext4_buddy *e4b, int order, int *max)
{
        char *bb;

        BUG_ON(e4b->bd_bitmap == e4b->bd_buddy);
        BUG_ON(max == NULL);

        if (order > e4b->bd_blkbits + 1) {
                *max = 0;
                return NULL;
        }

        /* at order 0 we see each particular block */
        if (order == 0) {
                *max = 1 << (e4b->bd_blkbits + 3);
                return e4b->bd_bitmap;
        }

        bb = e4b->bd_buddy + EXT4_SB(e4b->bd_sb)->s_mb_offsets[order];
        *max = EXT4_SB(e4b->bd_sb)->s_mb_maxs[order];

        return bb;
}

#ifdef DOUBLE_CHECK
static void mb_free_blocks_double(struct inode *inode, struct ext4_buddy *e4b,
                           int first, int count)
{
        int i;
        struct super_block *sb = e4b->bd_sb;

        if (unlikely(e4b->bd_info->bb_bitmap == NULL))
                return;
        assert_spin_locked(ext4_group_lock_ptr(sb, e4b->bd_group));
        for (i = 0; i < count; i++) {
                if (!mb_test_bit(first + i, e4b->bd_info->bb_bitmap)) {
                        ext4_fsblk_t blocknr;

                        blocknr = ext4_group_first_block_no(sb, e4b->bd_group);
                        blocknr += EXT4_C2B(EXT4_SB(sb), first + i);
                        ext4_grp_locked_error(sb, e4b->bd_group,
                                              inode ? inode->i_ino : 0,
                                              blocknr,
                                              "freeing block already freed "
                                              "(bit %u)",
                                              first + i);
                        ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group,
                                        EXT4_GROUP_INFO_BBITMAP_CORRUPT);
                }
                mb_clear_bit(first + i, e4b->bd_info->bb_bitmap);
        }
}

static void mb_mark_used_double(struct ext4_buddy *e4b, int first, int count)
{
        int i;

        if (unlikely(e4b->bd_info->bb_bitmap == NULL))
                return;
        assert_spin_locked(ext4_group_lock_ptr(e4b->bd_sb, e4b->bd_group));
        for (i = 0; i < count; i++) {
                BUG_ON(mb_test_bit(first + i, e4b->bd_info->bb_bitmap));
                mb_set_bit(first + i, e4b->bd_info->bb_bitmap);
        }
}

static void mb_cmp_bitmaps(struct ext4_buddy *e4b, void *bitmap)
{
        if (unlikely(e4b->bd_info->bb_bitmap == NULL))
                return;
        if (memcmp(e4b->bd_info->bb_bitmap, bitmap, e4b->bd_sb->s_blocksize)) {
                unsigned char *b1, *b2;
                int i;
                b1 = (unsigned char *) e4b->bd_info->bb_bitmap;
                b2 = (unsigned char *) bitmap;
                for (i = 0; i < e4b->bd_sb->s_blocksize; i++) {
                        if (b1[i] != b2[i]) {
                                ext4_msg(e4b->bd_sb, KERN_ERR,
                                         "corruption in group %u "
                                         "at byte %u(%u): %x in copy != %x "
                                         "on disk/prealloc",
                                         e4b->bd_group, i, i * 8, b1[i], b2[i]);
                                BUG();
                        }
                }
        }
}

static void mb_group_bb_bitmap_alloc(struct super_block *sb,
                        struct ext4_group_info *grp, ext4_group_t group)
{
        struct buffer_head *bh;

        grp->bb_bitmap = kmalloc(sb->s_blocksize, GFP_NOFS);
        if (!grp->bb_bitmap)
                return;

        bh = ext4_read_block_bitmap(sb, group);
        if (IS_ERR_OR_NULL(bh)) {
                kfree(grp->bb_bitmap);
                grp->bb_bitmap = NULL;
                return;
        }

        memcpy(grp->bb_bitmap, bh->b_data, sb->s_blocksize);
        put_bh(bh);
}

static void mb_group_bb_bitmap_free(struct ext4_group_info *grp)
{
        kfree(grp->bb_bitmap);
}

#else
static inline void mb_free_blocks_double(struct inode *inode,
                                struct ext4_buddy *e4b, int first, int count)
{
        return;
}
static inline void mb_mark_used_double(struct ext4_buddy *e4b,
                                                int first, int count)
{
        return;
}
static inline void mb_cmp_bitmaps(struct ext4_buddy *e4b, void *bitmap)
{
        return;
}

static inline void mb_group_bb_bitmap_alloc(struct super_block *sb,
                        struct ext4_group_info *grp, ext4_group_t group)
{
        return;
}

static inline void mb_group_bb_bitmap_free(struct ext4_group_info *grp)
{
        return;
}
#endif

#ifdef AGGRESSIVE_CHECK

#define MB_CHECK_ASSERT(assert)                                         \
do {                                                                    \
        if (!(assert)) {                                                \
                printk(KERN_EMERG                                       \
                        "Assertion failure in %s() at %s:%d: \"%s\"\n", \
                        function, file, line, # assert);                \
                BUG();                                                  \
        }                                                               \
} while (0)

static int __mb_check_buddy(struct ext4_buddy *e4b, char *file,
                                const char *function, int line)
{
        struct super_block *sb = e4b->bd_sb;
        int order = e4b->bd_blkbits + 1;
        int max;
        int max2;
        int i;
        int j;
        int k;
        int count;
        struct ext4_group_info *grp;
        int fragments = 0;
        int fstart;
        struct list_head *cur;
        void *buddy;
        void *buddy2;

        if (e4b->bd_info->bb_check_counter++ % 10)
                return 0;

        while (order > 1) {
                buddy = mb_find_buddy(e4b, order, &max);
                MB_CHECK_ASSERT(buddy);
                buddy2 = mb_find_buddy(e4b, order - 1, &max2);
                MB_CHECK_ASSERT(buddy2);
                MB_CHECK_ASSERT(buddy != buddy2);
                MB_CHECK_ASSERT(max * 2 == max2);

                count = 0;
                for (i = 0; i < max; i++) {

                        if (mb_test_bit(i, buddy)) {
                                /* only single bit in buddy2 may be 1 */
                                if (!mb_test_bit(i << 1, buddy2)) {
                                        MB_CHECK_ASSERT(
                                                mb_test_bit((i<<1)+1, buddy2));
                                } else if (!mb_test_bit((i << 1) + 1, buddy2)) {
                                        MB_CHECK_ASSERT(
                                                mb_test_bit(i << 1, buddy2));
                                }
                                continue;
                        }

                        /* both bits in buddy2 must be 1 */
                        MB_CHECK_ASSERT(mb_test_bit(i << 1, buddy2));
                        MB_CHECK_ASSERT(mb_test_bit((i << 1) + 1, buddy2));

                        for (j = 0; j < (1 << order); j++) {
                                k = (i * (1 << order)) + j;
                                MB_CHECK_ASSERT(
                                        !mb_test_bit(k, e4b->bd_bitmap));
                        }
                        count++;
                }
                MB_CHECK_ASSERT(e4b->bd_info->bb_counters[order] == count);
                order--;
        }

        fstart = -1;
        buddy = mb_find_buddy(e4b, 0, &max);
        for (i = 0; i < max; i++) {
                if (!mb_test_bit(i, buddy)) {
                        MB_CHECK_ASSERT(i >= e4b->bd_info->bb_first_free);
                        if (fstart == -1) {
                                fragments++;
                                fstart = i;
                        }
                        continue;
                }
                fstart = -1;
                /* check used bits only */
                for (j = 0; j < e4b->bd_blkbits + 1; j++) {
                        buddy2 = mb_find_buddy(e4b, j, &max2);
                        k = i >> j;
                        MB_CHECK_ASSERT(k < max2);
                        MB_CHECK_ASSERT(mb_test_bit(k, buddy2));
                }
        }
        MB_CHECK_ASSERT(!EXT4_MB_GRP_NEED_INIT(e4b->bd_info));
        MB_CHECK_ASSERT(e4b->bd_info->bb_fragments == fragments);

        grp = ext4_get_group_info(sb, e4b->bd_group);
        list_for_each(cur, &grp->bb_prealloc_list) {
                ext4_group_t groupnr;
                struct ext4_prealloc_space *pa;
                pa = list_entry(cur, struct ext4_prealloc_space, pa_group_list);
                ext4_get_group_no_and_offset(sb, pa->pa_pstart, &groupnr, &k);
                MB_CHECK_ASSERT(groupnr == e4b->bd_group);
                for (i = 0; i < pa->pa_len; i++)
                        MB_CHECK_ASSERT(mb_test_bit(k + i, buddy));
        }
        return 0;
}
#undef MB_CHECK_ASSERT
#define mb_check_buddy(e4b) __mb_check_buddy(e4b,       \
                                        __FILE__, __func__, __LINE__)
#else
#define mb_check_buddy(e4b)
#endif

/*
 * Divide blocks started from @first with length @len into
 * smaller chunks with power of 2 blocks.
 * Clear the bits in bitmap which the blocks of the chunk(s) covered,
 * then increase bb_counters[] for corresponded chunk size.
 */
static void ext4_mb_mark_free_simple(struct super_block *sb,
                                void *buddy, ext4_grpblk_t first, ext4_grpblk_t len,
                                        struct ext4_group_info *grp)
{
        struct ext4_sb_info *sbi = EXT4_SB(sb);
        ext4_grpblk_t min;
        ext4_grpblk_t max;
        ext4_grpblk_t chunk;
        unsigned int border;

        BUG_ON(len > EXT4_CLUSTERS_PER_GROUP(sb));

        border = 2 << sb->s_blocksize_bits;

        while (len > 0) {
                /* find how many blocks can be covered since this position */
                max = ffs(first | border) - 1;

                /* find how many blocks of power 2 we need to mark */
                min = fls(len) - 1;

                if (max < min)
                        min = max;
                chunk = 1 << min;

                /* mark multiblock chunks only */
                grp->bb_counters[min]++;
                if (min > 0)
                        mb_clear_bit(first >> min,
                                     buddy + sbi->s_mb_offsets[min]);

                len -= chunk;
                first += chunk;
        }
}

static void ext4_mb_rb_insert(struct rb_root *root, struct rb_node *new,
                        int (*cmp)(struct rb_node *, struct rb_node *))
{
        struct rb_node **iter = &root->rb_node, *parent = NULL;

        while (*iter) {
                parent = *iter;
                if (cmp(new, *iter) > 0)
                        iter = &((*iter)->rb_left);
                else
                        iter = &((*iter)->rb_right);
        }

        rb_link_node(new, parent, iter);
        rb_insert_color(new, root);
}

static int
ext4_mb_avg_fragment_size_cmp(struct rb_node *rb1, struct rb_node *rb2)
{
        struct ext4_group_info *grp1 = rb_entry(rb1,
                                                struct ext4_group_info,
                                                bb_avg_fragment_size_rb);
        struct ext4_group_info *grp2 = rb_entry(rb2,
                                                struct ext4_group_info,
                                                bb_avg_fragment_size_rb);
        int num_frags_1, num_frags_2;

        num_frags_1 = grp1->bb_fragments ?
                grp1->bb_free / grp1->bb_fragments : 0;
        num_frags_2 = grp2->bb_fragments ?
                grp2->bb_free / grp2->bb_fragments : 0;

        return (num_frags_2 - num_frags_1);
}

/*
 * Reinsert grpinfo into the avg_fragment_size tree with new average
 * fragment size.
 */
static void
mb_update_avg_fragment_size(struct super_block *sb, struct ext4_group_info *grp)
{
        struct ext4_sb_info *sbi = EXT4_SB(sb);

        if (!test_opt2(sb, MB_OPTIMIZE_SCAN) || grp->bb_free == 0)
                return;

        write_lock(&sbi->s_mb_rb_lock);
        if (!RB_EMPTY_NODE(&grp->bb_avg_fragment_size_rb)) {
                rb_erase(&grp->bb_avg_fragment_size_rb,
                                &sbi->s_mb_avg_fragment_size_root);
                RB_CLEAR_NODE(&grp->bb_avg_fragment_size_rb);
        }

        ext4_mb_rb_insert(&sbi->s_mb_avg_fragment_size_root,
                &grp->bb_avg_fragment_size_rb,
                ext4_mb_avg_fragment_size_cmp);
        write_unlock(&sbi->s_mb_rb_lock);
}

/*
 * Choose next group by traversing largest_free_order lists. Updates *new_cr if
 * cr level needs an update.
 */
static void ext4_mb_choose_next_group_cr0(struct ext4_allocation_context *ac,
                        int *new_cr, ext4_group_t *group, ext4_group_t ngroups)
{
        struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb);
        struct ext4_group_info *iter, *grp;
        int i;

        if (ac->ac_status == AC_STATUS_FOUND)
                return;

        if (unlikely(sbi->s_mb_stats && ac->ac_flags & EXT4_MB_CR0_OPTIMIZED))
                atomic_inc(&sbi->s_bal_cr0_bad_suggestions);

        grp = NULL;
        for (i = ac->ac_2order; i < MB_NUM_ORDERS(ac->ac_sb); i++) {
                if (list_empty(&sbi->s_mb_largest_free_orders[i]))
                        continue;
                read_lock(&sbi->s_mb_largest_free_orders_locks[i]);
                if (list_empty(&sbi->s_mb_largest_free_orders[i])) {
                        read_unlock(&sbi->s_mb_largest_free_orders_locks[i]);
                        continue;
                }
                grp = NULL;
                list_for_each_entry(iter, &sbi->s_mb_largest_free_orders[i],
                                    bb_largest_free_order_node) {
                        if (sbi->s_mb_stats)
                                atomic64_inc(&sbi->s_bal_cX_groups_considered[0]);
                        if (likely(ext4_mb_good_group(ac, iter->bb_group, 0))) {
                                grp = iter;
                                break;
                        }
                }
                read_unlock(&sbi->s_mb_largest_free_orders_locks[i]);
                if (grp)
                        break;
        }

        if (!grp) {
                /* Increment cr and search again */
                *new_cr = 1;
        } else {
                *group = grp->bb_group;
                ac->ac_last_optimal_group = *group;
                ac->ac_flags |= EXT4_MB_CR0_OPTIMIZED;
        }
}

/*
 * Choose next group by traversing average fragment size tree. Updates *new_cr
 * if cr lvel needs an update. Sets EXT4_MB_SEARCH_NEXT_LINEAR to indicate that
 * the linear search should continue for one iteration since there's lock
 * contention on the rb tree lock.
 */
static void ext4_mb_choose_next_group_cr1(struct ext4_allocation_context *ac,
                int *new_cr, ext4_group_t *group, ext4_group_t ngroups)
{
        struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb);
        int avg_fragment_size, best_so_far;
        struct rb_node *node, *found;
        struct ext4_group_info *grp;

        /*
         * If there is contention on the lock, instead of waiting for the lock
         * to become available, just continue searching lineraly. We'll resume
         * our rb tree search later starting at ac->ac_last_optimal_group.
         */
        if (!read_trylock(&sbi->s_mb_rb_lock)) {
                ac->ac_flags |= EXT4_MB_SEARCH_NEXT_LINEAR;
                return;
        }

        if (unlikely(ac->ac_flags & EXT4_MB_CR1_OPTIMIZED)) {
                if (sbi->s_mb_stats)
                        atomic_inc(&sbi->s_bal_cr1_bad_suggestions);
                /* We have found something at CR 1 in the past */
                grp = ext4_get_group_info(ac->ac_sb, ac->ac_last_optimal_group);
                for (found = rb_next(&grp->bb_avg_fragment_size_rb); found != NULL;
                     found = rb_next(found)) {
                        grp = rb_entry(found, struct ext4_group_info,
                                       bb_avg_fragment_size_rb);
                        if (sbi->s_mb_stats)
                                atomic64_inc(&sbi->s_bal_cX_groups_considered[1]);
                        if (likely(ext4_mb_good_group(ac, grp->bb_group, 1)))
                                break;
                }
                goto done;
        }

        node = sbi->s_mb_avg_fragment_size_root.rb_node;
        best_so_far = 0;
        found = NULL;

        while (node) {
                grp = rb_entry(node, struct ext4_group_info,
                               bb_avg_fragment_size_rb);
                avg_fragment_size = 0;
                if (ext4_mb_good_group(ac, grp->bb_group, 1)) {
                        avg_fragment_size = grp->bb_fragments ?
                                grp->bb_free / grp->bb_fragments : 0;
                        if (!best_so_far || avg_fragment_size < best_so_far) {
                                best_so_far = avg_fragment_size;
                                found = node;
                        }
                }
                if (avg_fragment_size > ac->ac_g_ex.fe_len)
                        node = node->rb_right;
                else
                        node = node->rb_left;
        }

done:
        if (found) {
                grp = rb_entry(found, struct ext4_group_info,
                               bb_avg_fragment_size_rb);
                *group = grp->bb_group;
                ac->ac_flags |= EXT4_MB_CR1_OPTIMIZED;
        } else {
                *new_cr = 2;
        }

        read_unlock(&sbi->s_mb_rb_lock);
        ac->ac_last_optimal_group = *group;
}

static inline int should_optimize_scan(struct ext4_allocation_context *ac)
{
        if (unlikely(!test_opt2(ac->ac_sb, MB_OPTIMIZE_SCAN)))
                return 0;
        if (ac->ac_criteria >= 2)
                return 0;
        if (ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS))
                return 0;

        return 1;
}

/*
 * Return next linear group for allocation. If linear traversal should not be
 * performed, this function just returns the same group
 */
static int
next_linear_group(struct ext4_allocation_context *ac, int group, int ngroups)
{
        if (!should_optimize_scan(ac))
                goto inc_and_return;

        if (ac->ac_groups_linear_remaining) {
                ac->ac_groups_linear_remaining--;
                goto inc_and_return;
        }

        if (ac->ac_flags & EXT4_MB_SEARCH_NEXT_LINEAR) {
                ac->ac_flags &= ~EXT4_MB_SEARCH_NEXT_LINEAR;
                goto inc_and_return;
        }

        return group;
inc_and_return:
        /*
         * Artificially restricted ngroups for non-extent
         * files makes group > ngroups possible on first loop.
         */
        return group + 1 >= ngroups ? 0 : group + 1;
}

/*
 * ext4_mb_choose_next_group: choose next group for allocation.
 *
 * @ac        Allocation Context
 * @new_cr    This is an output parameter. If the there is no good group
 *            available at current CR level, this field is updated to indicate
 *            the new cr level that should be used.
 * @group     This is an input / output parameter. As an input it indicates the
 *            next group that the allocator intends to use for allocation. As
 *            output, this field indicates the next group that should be used as
 *            determined by the optimization functions.
 * @ngroups   Total number of groups
 */
static void ext4_mb_choose_next_group(struct ext4_allocation_context *ac,
                int *new_cr, ext4_group_t *group, ext4_group_t ngroups)
{
        *new_cr = ac->ac_criteria;

        if (!should_optimize_scan(ac) || ac->ac_groups_linear_remaining)
                return;

        if (*new_cr == 0) {
                ext4_mb_choose_next_group_cr0(ac, new_cr, group, ngroups);
        } else if (*new_cr == 1) {
                ext4_mb_choose_next_group_cr1(ac, new_cr, group, ngroups);
        } else {
                /*
                 * TODO: For CR=2, we can arrange groups in an rb tree sorted by
                 * bb_free. But until that happens, we should never come here.
                 */
                WARN_ON(1);
        }
}

/*
 * Cache the order of the largest free extent we have available in this block
 * group.
 */
static void
mb_set_largest_free_order(struct super_block *sb, struct ext4_group_info *grp)
{
        struct ext4_sb_info *sbi = EXT4_SB(sb);
        int i;

        if (test_opt2(sb, MB_OPTIMIZE_SCAN) && grp->bb_largest_free_order >= 0) {
                write_lock(&sbi->s_mb_largest_free_orders_locks[
                                              grp->bb_largest_free_order]);
                list_del_init(&grp->bb_largest_free_order_node);
                write_unlock(&sbi->s_mb_largest_free_orders_locks[
                                              grp->bb_largest_free_order]);
        }
        grp->bb_largest_free_order = -1; /* uninit */

        for (i = MB_NUM_ORDERS(sb) - 1; i >= 0; i--) {
                if (grp->bb_counters[i] > 0) {
                        grp->bb_largest_free_order = i;
                        break;
                }
        }
        if (test_opt2(sb, MB_OPTIMIZE_SCAN) &&
            grp->bb_largest_free_order >= 0 && grp->bb_free) {
                write_lock(&sbi->s_mb_largest_free_orders_locks[
                                              grp->bb_largest_free_order]);
                list_add_tail(&grp->bb_largest_free_order_node,
                      &sbi->s_mb_largest_free_orders[grp->bb_largest_free_order]);
                write_unlock(&sbi->s_mb_largest_free_orders_locks[
                                              grp->bb_largest_free_order]);
        }
}

static noinline_for_stack
void ext4_mb_generate_buddy(struct super_block *sb,
                                void *buddy, void *bitmap, ext4_group_t group)
{
        struct ext4_group_info *grp = ext4_get_group_info(sb, group);
        struct ext4_sb_info *sbi = EXT4_SB(sb);
        ext4_grpblk_t max = EXT4_CLUSTERS_PER_GROUP(sb);
        ext4_grpblk_t i = 0;
        ext4_grpblk_t first;
        ext4_grpblk_t len;
        unsigned free = 0;
        unsigned fragments = 0;
        unsigned long long period = get_cycles();

        /* initialize buddy from bitmap which is aggregation
         * of on-disk bitmap and preallocations */
        i = mb_find_next_zero_bit(bitmap, max, 0);
        grp->bb_first_free = i;
        while (i < max) {
                fragments++;
                first = i;
                i = mb_find_next_bit(bitmap, max, i);
                len = i - first;
                free += len;
                if (len > 1)
                        ext4_mb_mark_free_simple(sb, buddy, first, len, grp);
                else
                        grp->bb_counters[0]++;
                if (i < max)
                        i = mb_find_next_zero_bit(bitmap, max, i);
        }
        grp->bb_fragments = fragments;

        if (free != grp->bb_free) {
                ext4_grp_locked_error(sb, group, 0, 0,
                                      "block bitmap and bg descriptor "
                                      "inconsistent: %u vs %u free clusters",
                                      free, grp->bb_free);
                /*
                 * If we intend to continue, we consider group descriptor
                 * corrupt and update bb_free using bitmap value
                 */
                grp->bb_free = free;
                ext4_mark_group_bitmap_corrupted(sb, group,
                                        EXT4_GROUP_INFO_BBITMAP_CORRUPT);
        }
        mb_set_largest_free_order(sb, grp);

        clear_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &(grp->bb_state));

        period = get_cycles() - period;
        atomic_inc(&sbi->s_mb_buddies_generated);
        atomic64_add(period, &sbi->s_mb_generation_time);
        mb_update_avg_fragment_size(sb, grp);
}

/* The buddy information is attached the buddy cache inode
 * for convenience. The information regarding each group
 * is loaded via ext4_mb_load_buddy. The information involve
 * block bitmap and buddy information. The information are
 * stored in the inode as
 *
 * {                        page                        }
 * [ group 0 bitmap][ group 0 buddy] [group 1][ group 1]...
 *
 *
 * one block each for bitmap and buddy information.
 * So for each group we take up 2 blocks. A page can
 * contain blocks_per_page (PAGE_SIZE / blocksize)  blocks.
 * So it can have information regarding groups_per_page which
 * is blocks_per_page/2
 *
 * Locking note:  This routine takes the block group lock of all groups
 * for this page; do not hold this lock when calling this routine!
 */

static int ext4_mb_init_cache(struct page *page, char *incore, gfp_t gfp)
{
        ext4_group_t ngroups;
        int blocksize;
        int blocks_per_page;
        int groups_per_page;
        int err = 0;
        int i;
        ext4_group_t first_group, group;
        int first_block;
        struct super_block *sb;
        struct buffer_head *bhs;
        struct buffer_head **bh = NULL;
        struct inode *inode;
        char *data;
        char *bitmap;
        struct ext4_group_info *grinfo;

        inode = page->mapping->host;
        sb = inode->i_sb;
        ngroups = ext4_get_groups_count(sb);
        blocksize = i_blocksize(inode);
        blocks_per_page = PAGE_SIZE / blocksize;

        mb_debug(sb, "init page %lu\n", page->index);

        groups_per_page = blocks_per_page >> 1;
        if (groups_per_page == 0)
                groups_per_page = 1;

        /* allocate buffer_heads to read bitmaps */
        if (groups_per_page > 1) {
                i = sizeof(struct buffer_head *) * groups_per_page;
                bh = kzalloc(i, gfp);
                if (bh == NULL) {
                        err = -ENOMEM;
                        goto out;
                }
        } else
                bh = &bhs;

        first_group = page->index * blocks_per_page / 2;

        /* read all groups the page covers into the cache */
        for (i = 0, group = first_group; i < groups_per_page; i++, group++) {
                if (group >= ngroups)
                        break;

                grinfo = ext4_get_group_info(sb, group);
                /*
                 * If page is uptodate then we came here after online resize
                 * which added some new uninitialized group info structs, so
                 * we must skip all initialized uptodate buddies on the page,
                 * which may be currently in use by an allocating task.
                 */
                if (PageUptodate(page) && !EXT4_MB_GRP_NEED_INIT(grinfo)) {
                        bh[i] = NULL;
                        continue;
                }
                bh[i] = ext4_read_block_bitmap_nowait(sb, group, false);
                if (IS_ERR(bh[i])) {
                        err = PTR_ERR(bh[i]);
                        bh[i] = NULL;
                        goto out;
                }
                mb_debug(sb, "read bitmap for group %u\n", group);
        }

        /* wait for I/O completion */
        for (i = 0, group = first_group; i < groups_per_page; i++, group++) {
                int err2;

                if (!bh[i])
                        continue;
                err2 = ext4_wait_block_bitmap(sb, group, bh[i]);
                if (!err)
                        err = err2;
        }

        first_block = page->index * blocks_per_page;
        for (i = 0; i < blocks_per_page; i++) {
                group = (first_block + i) >> 1;
                if (group >= ngroups)
                        break;

                if (!bh[group - first_group])
                        /* skip initialized uptodate buddy */
                        continue;

                if (!buffer_verified(bh[group - first_group]))
                        /* Skip faulty bitmaps */
                        continue;
                err = 0;

                /*
                 * data carry information regarding this
                 * particular group in the format specified
                 * above
                 *
                 */
                data = page_address(page) + (i * blocksize);
                bitmap = bh[group - first_group]->b_data;

                /*
                 * We place the buddy block and bitmap block
                 * close together
                 */
                if ((first_block + i) & 1) {
                        /* this is block of buddy */
                        BUG_ON(incore == NULL);
                        mb_debug(sb, "put buddy for group %u in page %lu/%x\n",
                                group, page->index, i * blocksize);
                        trace_ext4_mb_buddy_bitmap_load(sb, group);
                        grinfo = ext4_get_group_info(sb, group);
                        grinfo->bb_fragments = 0;
                        memset(grinfo->bb_counters, 0,
                               sizeof(*grinfo->bb_counters) *
                               (MB_NUM_ORDERS(sb)));
                        /*
                         * incore got set to the group block bitmap below
                         */
                        ext4_lock_group(sb, group);
                        /* init the buddy */
                        memset(data, 0xff, blocksize);
                        ext4_mb_generate_buddy(sb, data, incore, group);
                        ext4_unlock_group(sb, group);
                        incore = NULL;
                } else {
                        /* this is block of bitmap */
                        BUG_ON(incore != NULL);
                        mb_debug(sb, "put bitmap for group %u in page %lu/%x\n",
                                group, page->index, i * blocksize);
                        trace_ext4_mb_bitmap_load(sb, group);

                        /* see comments in ext4_mb_put_pa() */
                        ext4_lock_group(sb, group);
                        memcpy(data, bitmap, blocksize);

                        /* mark all preallocated blks used in in-core bitmap */
                        ext4_mb_generate_from_pa(sb, data, group);
                        ext4_mb_generate_from_freelist(sb, data, group);
                        ext4_unlock_group(sb, group);

                        /* set incore so that the buddy information can be
                         * generated using this
                         */
                        incore = data;
                }
        }
        SetPageUptodate(page);

out:
        if (bh) {
                for (i = 0; i < groups_per_page; i++)
                        brelse(bh[i]);
                if (bh != &bhs)
                        kfree(bh);
        }
        return err;
}

/*
 * Lock the buddy and bitmap pages. This make sure other parallel init_group
 * on the same buddy page doesn't happen whild holding the buddy page lock.
 * Return locked buddy and bitmap pages on e4b struct. If buddy and bitmap
 * are on the same page e4b->bd_buddy_page is NULL and return value is 0.
 */
static int ext4_mb_get_buddy_page_lock(struct super_block *sb,
                ext4_group_t group, struct ext4_buddy *e4b, gfp_t gfp)
{
        struct inode *inode = EXT4_SB(sb)->s_buddy_cache;
        int block, pnum, poff;
        int blocks_per_page;
        struct page *page;

        e4b->bd_buddy_page = NULL;
        e4b->bd_bitmap_page = NULL;

        blocks_per_page = PAGE_SIZE / sb->s_blocksize;
        /*
         * the buddy cache inode stores the block bitmap
         * and buddy information in consecutive blocks.
         * So for each group we need two blocks.
         */
        block = group * 2;
        pnum = block / blocks_per_page;
        poff = block % blocks_per_page;
        page = find_or_create_page(inode->i_mapping, pnum, gfp);
        if (!page)
                return -ENOMEM;
        BUG_ON(page->mapping != inode->i_mapping);
        e4b->bd_bitmap_page = page;
        e4b->bd_bitmap = page_address(page) + (poff * sb->s_blocksize);

        if (blocks_per_page >= 2) {
                /* buddy and bitmap are on the same page */
                return 0;
        }

        block++;
        pnum = block / blocks_per_page;
        page = find_or_create_page(inode->i_mapping, pnum, gfp);
        if (!page)
                return -ENOMEM;
        BUG_ON(page->mapping != inode->i_mapping);
        e4b->bd_buddy_page = page;
        return 0;
}

static void ext4_mb_put_buddy_page_lock(struct ext4_buddy *e4b)
{
        if (e4b->bd_bitmap_page) {
                unlock_page(e4b->bd_bitmap_page);
                put_page(e4b->bd_bitmap_page);
        }
        if (e4b->bd_buddy_page) {
                unlock_page(e4b->bd_buddy_page);
                put_page(e4b->bd_buddy_page);
        }
}

/*
 * Locking note:  This routine calls ext4_mb_init_cache(), which takes the
 * block group lock of all groups for this page; do not hold the BG lock when
 * calling this routine!
 */
static noinline_for_stack
int ext4_mb_init_group(struct super_block *sb, ext4_group_t group, gfp_t gfp)
{

        struct ext4_group_info *this_grp;
        struct ext4_buddy e4b;
        struct page *page;
        int ret = 0;

        might_sleep();
        mb_debug(sb, "init group %u\n", group);
        this_grp = ext4_get_group_info(sb, group);
        /*
         * This ensures that we don't reinit the buddy cache
         * page which map to the group from which we are already
         * allocating. If we are looking at the buddy cache we would
         * have taken a reference using ext4_mb_load_buddy and that
         * would have pinned buddy page to page cache.
         * The call to ext4_mb_get_buddy_page_lock will mark the
         * page accessed.
         */
        ret = ext4_mb_get_buddy_page_lock(sb, group, &e4b, gfp);
        if (ret || !EXT4_MB_GRP_NEED_INIT(this_grp)) {
                /*
                 * somebody initialized the group
                 * return without doing anything
                 */
                goto err;
        }

        page = e4b.bd_bitmap_page;
        ret = ext4_mb_init_cache(page, NULL, gfp);
        if (ret)
                goto err;
        if (!PageUptodate(page)) {
                ret = -EIO;
                goto err;
        }

        if (e4b.bd_buddy_page == NULL) {
                /*
                 * If both the bitmap and buddy are in
                 * the same page we don't need to force
                 * init the buddy
                 */
                ret = 0;
                goto err;
        }
        /* init buddy cache */
        page = e4b.bd_buddy_page;
        ret = ext4_mb_init_cache(page, e4b.bd_bitmap, gfp);
        if (ret)
                goto err;
        if (!PageUptodate(page)) {
                ret = -EIO;
                goto err;
        }
err:
        ext4_mb_put_buddy_page_lock(&e4b);
        return ret;
}

/*
 * Locking note:  This routine calls ext4_mb_init_cache(), which takes the
 * block group lock of all groups for this page; do not hold the BG lock when
 * calling this routine!
 */
static noinline_for_stack int
ext4_mb_load_buddy_gfp(struct super_block *sb, ext4_group_t group,
                       struct ext4_buddy *e4b, gfp_t gfp)
{
        int blocks_per_page;
        int block;
        int pnum;
        int poff;
        struct page *page;
        int ret;
        struct ext4_group_info *grp;
        struct ext4_sb_info *sbi = EXT4_SB(sb);
        struct inode *inode = sbi->s_buddy_cache;

        might_sleep();
        mb_debug(sb, "load group %u\n", group);

        blocks_per_page = PAGE_SIZE / sb->s_blocksize;
        grp = ext4_get_group_info(sb, group);

        e4b->bd_blkbits = sb->s_blocksize_bits;
        e4b->bd_info = grp;
        e4b->bd_sb = sb;
        e4b->bd_group = group;
        e4b->bd_buddy_page = NULL;
        e4b->bd_bitmap_page = NULL;

        if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) {
                /*
                 * we need full data about the group
                 * to make a good selection
                 */
                ret = ext4_mb_init_group(sb, group, gfp);
                if (ret)
                        return ret;
        }

        /*
         * the buddy cache inode stores the block bitmap
         * and buddy information in consecutive blocks.
         * So for each group we need two blocks.
         */
        block = group * 2;
        pnum = block / blocks_per_page;
        poff = block % blocks_per_page;

        /* we could use find_or_create_page(), but it locks page
         * what we'd like to avoid in fast path ... */
        page = find_get_page_flags(inode->i_mapping, pnum, FGP_ACCESSED);
        if (page == NULL || !PageUptodate(page)) {
                if (page)
                        /*
                         * drop the page reference and try
                         * to get the page with lock. If we
                         * are not uptodate that implies
                         * somebody just created the page but
                         * is yet to initialize the same. So
                         * wait for it to initialize.
                         */
                        put_page(page);
                page = find_or_create_page(inode->i_mapping, pnum, gfp);
                if (page) {
                        BUG_ON(page->mapping != inode->i_mapping);
                        if (!PageUptodate(page)) {
                                ret = ext4_mb_init_cache(page, NULL, gfp);
                                if (ret) {
                                        unlock_page(page);
                                        goto err;
                                }
                                mb_cmp_bitmaps(e4b, page_address(page) +
                                               (poff * sb->s_blocksize));
                        }
                        unlock_page(page);
                }
        }
        if (page == NULL) {
                ret = -ENOMEM;
                goto err;
        }
        if (!PageUptodate(page)) {
                ret = -EIO;
                goto err;
        }

        /* Pages marked accessed already */
        e4b->bd_bitmap_page = page;
        e4b->bd_bitmap = page_address(page) + (poff * sb->s_blocksize);

        block++;
        pnum = block / blocks_per_page;
        poff = block % blocks_per_page;

        page = find_get_page_flags(inode->i_mapping, pnum, FGP_ACCESSED);
        if (page == NULL || !PageUptodate(page)) {
                if (page)
                        put_page(page);
                page = find_or_create_page(inode->i_mapping, pnum, gfp);
                if (page) {
                        BUG_ON(page->mapping != inode->i_mapping);
                        if (!PageUptodate(page)) {
                                ret = ext4_mb_init_cache(page, e4b->bd_bitmap,
                                                         gfp);
                                if (ret) {
                                        unlock_page(page);
                                        goto err;
                                }
                        }
                        unlock_page(page);
                }
        }
        if (page == NULL) {
                ret = -ENOMEM;
                goto err;
        }
        if (!PageUptodate(page)) {
                ret = -EIO;
                goto err;
        }

        /* Pages marked accessed already */
        e4b->bd_buddy_page = page;
        e4b->bd_buddy = page_address(page) + (poff * sb->s_blocksize);

        return 0;

err:
        if (page)
                put_page(page);
        if (e4b->bd_bitmap_page)
                put_page(e4b->bd_bitmap_page);
        if (e4b->bd_buddy_page)
                put_page(e4b->bd_buddy_page);
        e4b->bd_buddy = NULL;
        e4b->bd_bitmap = NULL;
        return ret;
}

static int ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
                              struct ext4_buddy *e4b)
{
        return ext4_mb_load_buddy_gfp(sb, group, e4b, GFP_NOFS);
}

static void ext4_mb_unload_buddy(struct ext4_buddy *e4b)
{
        if (e4b->bd_bitmap_page)
                put_page(e4b->bd_bitmap_page);
        if (e4b->bd_buddy_page)
                put_page(e4b->bd_buddy_page);
}


static int mb_find_order_for_block(struct ext4_buddy *e4b, int block)
{
        int order = 1, max;
        void *bb;

        BUG_ON(e4b->bd_bitmap == e4b->bd_buddy);
        BUG_ON(block >= (1 << (e4b->bd_blkbits + 3)));

        while (order <= e4b->bd_blkbits + 1) {
                bb = mb_find_buddy(e4b, order, &max);
                if (!mb_test_bit(block >> order, bb)) {
                        /* this block is part of buddy of order 'order' */
                        return order;
                }
                order++;
        }
        return 0;
}

static void mb_clear_bits(void *bm, int cur, int len)
{
        __u32 *addr;

        len = cur + len;
        while (cur < len) {
                if ((cur & 31) == 0 && (len - cur) >= 32) {
                        /* fast path: clear whole word at once */
                        addr = bm + (cur >> 3);
                        *addr = 0;
                        cur += 32;
                        continue;
                }
                mb_clear_bit(cur, bm);
                cur++;
        }
}

/* clear bits in given range
 * will return first found zero bit if any, -1 otherwise
 */
static int mb_test_and_clear_bits(void *bm, int cur, int len)
{
        __u32 *addr;
        int zero_bit = -1;

        len = cur + len;
        while (cur < len) {
                if ((cur & 31) == 0 && (len - cur) >= 32) {
                        /* fast path: clear whole word at once */
                        addr = bm + (cur >> 3);
                        if (*addr != (__u32)(-1) && zero_bit == -1)
                                zero_bit = cur + mb_find_next_zero_bit(addr, 32, 0);
                        *addr = 0;
                        cur += 32;
                        continue;
                }
                if (!mb_test_and_clear_bit(cur, bm) && zero_bit == -1)
                        zero_bit = cur;
                cur++;
        }

        return zero_bit;
}

void ext4_set_bits(void *bm, int cur, int len)
{
        __u32 *addr;

        len = cur + len;
        while (cur < len) {
                if ((cur & 31) == 0 && (len - cur) >= 32) {
                        /* fast path: set whole word at once */
                        addr = bm + (cur >> 3);
                        *addr = 0xffffffff;
                        cur += 32;
                        continue;
                }
                mb_set_bit(cur, bm);
                cur++;
        }
}

static inline int mb_buddy_adjust_border(int* bit, void* bitmap, int side)
{
        if (mb_test_bit(*bit + side, bitmap)) {
                mb_clear_bit(*bit, bitmap);
                (*bit) -= side;
                return 1;
        }
        else {
                (*bit) += side;
                mb_set_bit(*bit, bitmap);
                return -1;
        }
}

static void mb_buddy_mark_free(struct ext4_buddy *e4b, int first, int last)
{
        int max;
        int order = 1;
        void *buddy = mb_find_buddy(e4b, order, &max);

        while (buddy) {
                void *buddy2;

                /* Bits in range [first; last] are known to be set since
                 * corresponding blocks were allocated. Bits in range
                 * (first; last) will stay set because they form buddies on
                 * upper layer. We just deal with borders if they don't
                 * align with upper layer and then go up.
                 * Releasing entire group is all about clearing
                 * single bit of highest order buddy.
                 */

                /* Example:
                 * ---------------------------------
                 * |   1   |   1   |   1   |   1   |
                 * ---------------------------------
                 * | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
                 * ---------------------------------
                 *   0   1   2   3   4   5   6   7
                 *      \_____________________/
                 *
                 * Neither [1] nor [6] is aligned to above layer.
                 * Left neighbour [0] is free, so mark it busy,
                 * decrease bb_counters and extend range to
                 * [0; 6]
                 * Right neighbour [7] is busy. It can't be coaleasced with [6], so
                 * mark [6] free, increase bb_counters and shrink range to
                 * [0; 5].
                 * Then shift range to [0; 2], go up and do the same.
                 */


                if (first & 1)
                        e4b->bd_info->bb_counters[order] += mb_buddy_adjust_border(&first, buddy, -1);
                if (!(last & 1))
                        e4b->bd_info->bb_counters[order] += mb_buddy_adjust_border(&last, buddy, 1);
                if (first > last)
                        break;
                order++;

                if (first == last || !(buddy2 = mb_find_buddy(e4b, order, &max))) {
                        mb_clear_bits(buddy, first, last - first + 1);
                        e4b->bd_info->bb_counters[order - 1] += last - first + 1;
                        break;
                }
                first >>= 1;
                last >>= 1;
                buddy = buddy2;
        }
}

static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b,
                           int first, int count)
{
        int left_is_free = 0;
        int right_is_free = 0;
        int block;
        int last = first + count - 1;
        struct super_block *sb = e4b->bd_sb;

        if (WARN_ON(count == 0))
                return;
        BUG_ON(last >= (sb->s_blocksize << 3));
        assert_spin_locked(ext4_group_lock_ptr(sb, e4b->bd_group));
        /* Don't bother if the block group is corrupt. */
        if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info)))
                return;

        mb_check_buddy(e4b);
        mb_free_blocks_double(inode, e4b, first, count);

        this_cpu_inc(discard_pa_seq);
        e4b->bd_info->bb_free += count;
        if (first < e4b->bd_info->bb_first_free)
                e4b->bd_info->bb_first_free = first;

        /* access memory sequentially: check left neighbour,
         * clear range and then check right neighbour
         */
        if (first != 0)
                left_is_free = !mb_test_bit(first - 1, e4b->bd_bitmap);
        block = mb_test_and_clear_bits(e4b->bd_bitmap, first, count);
        if (last + 1 < EXT4_SB(sb)->s_mb_maxs[0])
                right_is_free = !mb_test_bit(last + 1, e4b->bd_bitmap);

        if (unlikely(block != -1)) {
                struct ext4_sb_info *sbi = EXT4_SB(sb);
                ext4_fsblk_t blocknr;

                blocknr = ext4_group_first_block_no(sb, e4b->bd_group);
                blocknr += EXT4_C2B(sbi, block);
                if (!(sbi->s_mount_state & EXT4_FC_REPLAY)) {
                        ext4_grp_locked_error(sb, e4b->bd_group,
                                              inode ? inode->i_ino : 0,
                                              blocknr,
                                              "freeing already freed block (bit %u); block bitmap corrupt.",
                                              block);
                        ext4_mark_group_bitmap_corrupted(
                                sb, e4b->bd_group,
                                EXT4_GROUP_INFO_BBITMAP_CORRUPT);
                }
                goto done;
        }

        /* let's maintain fragments counter */
        if (left_is_free && right_is_free)
                e4b->bd_info->bb_fragments--;
        else if (!left_is_free && !right_is_free)
                e4b->bd_info->bb_fragments++;

        /* buddy[0] == bd_bitmap is a special case, so handle
         * it right away and let mb_buddy_mark_free stay free of
         * zero order checks.
         * Check if neighbours are to be coaleasced,
         * adjust bitmap bb_counters and borders appropriately.
         */
        if (first & 1) {
                first += !left_is_free;
                e4b->bd_info->bb_counters[0] += left_is_free ? -1 : 1;
        }
        if (!(last & 1)) {
                last -= !right_is_free;
                e4b->bd_info->bb_counters[0] += right_is_free ? -1 : 1;
        }

        if (first <= last)
                mb_buddy_mark_free(e4b, first >> 1, last >> 1);

done:
        mb_set_largest_free_order(sb, e4b->bd_info);
        mb_update_avg_fragment_size(sb, e4b->bd_info);
        mb_check_buddy(e4b);
}

static int mb_find_extent(struct ext4_buddy *e4b, int block,
                                int needed, struct ext4_free_extent *ex)
{
        int next = block;
        int max, order;
        void *buddy;

        assert_spin_locked(ext4_group_lock_ptr(e4b->bd_sb, e4b->bd_group));
        BUG_ON(ex == NULL);

        buddy = mb_find_buddy(e4b, 0, &max);
        BUG_ON(buddy == NULL);
        BUG_ON(block >= max);
        if (mb_test_bit(block, buddy)) {
                ex->fe_len = 0;
                ex->fe_start = 0;
                ex->fe_group = 0;
                return 0;
        }

        /* find actual order */
        order = mb_find_order_for_block(e4b, block);
        block = block >> order;

        ex->fe_len = 1 << order;
        ex->fe_start = block << order;
        ex->fe_group = e4b->bd_group;

        /* calc difference from given start */
        next = next - ex->fe_start;
        ex->fe_len -= next;
        ex->fe_start += next;

        while (needed > ex->fe_len &&
               mb_find_buddy(e4b, order, &max)) {

                if (block + 1 >= max)
                        break;

                next = (block + 1) * (1 << order);
                if (mb_test_bit(next, e4b->bd_bitmap))
                        break;

                order = mb_find_order_for_block(e4b, next);

                block = next >> order;
                ex->fe_len += 1 << order;
        }

        if (ex->fe_start + ex->fe_len > EXT4_CLUSTERS_PER_GROUP(e4b->bd_sb)) {
                /* Should never happen! (but apparently sometimes does?!?) */
                WARN_ON(1);
                ext4_grp_locked_error(e4b->bd_sb, e4b->bd_group, 0, 0,
                        "corruption or bug in mb_find_extent "
                        "block=%d, order=%d needed=%d ex=%u/%d/%d@%u",
                        block, order, needed, ex->fe_group, ex->fe_start,
                        ex->fe_len, ex->fe_logical);
                ex->fe_len = 0;
                ex->fe_start = 0;
                ex->fe_group = 0;
        }
        return ex->fe_len;
}

static int mb_mark_used(struct ext4_buddy *e4b, struct ext4_free_extent *ex)
{
        int ord;
        int mlen = 0;
        int max = 0;
        int cur;
        int start = ex->fe_start;
        int len = ex->fe_len;
        unsigned ret = 0;
        int len0 = len;
        void *buddy;

        BUG_ON(start + len > (e4b->bd_sb->s_blocksize << 3));
        BUG_ON(e4b->bd_group != ex->fe_group);
        assert_spin_locked(ext4_group_lock_ptr(e4b->bd_sb, e4b->bd_group));
        mb_check_buddy(e4b);
        mb_mark_used_double(e4b, start, len);

        this_cpu_inc(discard_pa_seq);
        e4b->bd_info->bb_free -= len;
        if (e4b->bd_info->bb_first_free == start)
                e4b->bd_info->bb_first_free += len;

        /* let's maintain fragments counter */
        if (start != 0)
                mlen = !mb_test_bit(start - 1, e4b->bd_bitmap);
        if (start + len < EXT4_SB(e4b->bd_sb)->s_mb_maxs[0])
                max = !mb_test_bit(start + len, e4b->bd_bitmap);
        if (mlen && max)
                e4b->bd_info->bb_fragments++;
        else if (!mlen && !max)
                e4b->bd_info->bb_fragments--;

        /* let's maintain buddy itself */
        while (len) {
                ord = mb_find_order_for_block(e4b, start);

                if (((start >> ord) << ord) == start && len >= (1 << ord)) {
                        /* the whole chunk may be allocated at once! */
                        mlen = 1 << ord;
                        buddy = mb_find_buddy(e4b, ord, &max);
                        BUG_ON((start >> ord) >= max);
                        mb_set_bit(start >> ord, buddy);
                        e4b->bd_info->bb_counters[ord]--;
                        start += mlen;
                        len -= mlen;
                        BUG_ON(len < 0);
                        continue;
                }

                /* store for history */
                if (ret == 0)
                        ret = len | (ord << 16);

                /* we have to split large buddy */
                BUG_ON(ord <= 0);
                buddy = mb_find_buddy(e4b, ord, &max);
                mb_set_bit(start >> ord, buddy);
                e4b->bd_info->bb_counters[ord]--;

                ord--;
                cur = (start >> ord) & ~1U;
                buddy = mb_find_buddy(e4b, ord, &max);
                mb_clear_bit(cur, buddy);
                mb_clear_bit(cur + 1, buddy);
                e4b->bd_info->bb_counters[ord]++;
                e4b->bd_info->bb_counters[ord]++;
        }
        mb_set_largest_free_order(e4b->bd_sb, e4b->bd_info);

        mb_update_avg_fragment_size(e4b->bd_sb, e4b->bd_info);
        ext4_set_bits(e4b->bd_bitmap, ex->fe_start, len0);
        mb_check_buddy(e4b);

        return ret;
}

/*
 * Must be called under group lock!
 */
static void ext4_mb_use_best_found(struct ext4_allocation_context *ac,
                                        struct ext4_buddy *e4b)
{
        struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb);
        int ret;

        BUG_ON(ac->ac_b_ex.fe_group != e4b->bd_group);
        BUG_ON(ac->ac_status == AC_STATUS_FOUND);

        ac->ac_b_ex.fe_len = min(ac->ac_b_ex.fe_len, ac->ac_g_ex.fe_len);
        ac->ac_b_ex.fe_logical = ac->ac_g_ex.fe_logical;
        ret = mb_mark_used(e4b, &ac->ac_b_ex);

        /* preallocation can change ac_b_ex, thus we store actually
         * allocated blocks for history */
        ac->ac_f_ex = ac->ac_b_ex;

        ac->ac_status = AC_STATUS_FOUND;
        ac->ac_tail = ret & 0xffff;
        ac->ac_buddy = ret >> 16;

        /*
         * take the page reference. We want the page to be pinned
         * so that we don't get a ext4_mb_init_cache_call for this
         * group until we update the bitmap. That would mean we
         * double allocate blocks. The reference is dropped
         * in ext4_mb_release_context
         */
        ac->ac_bitmap_page = e4b->bd_bitmap_page;
        get_page(ac->ac_bitmap_page);
        ac->ac_buddy_page = e4b->bd_buddy_page;
        get_page(ac->ac_buddy_page);
        /* store last allocated for subsequent stream allocation */
        if (ac->ac_flags & EXT4_MB_STREAM_ALLOC) {
                spin_lock(&sbi->s_md_lock);
                sbi->s_mb_last_group = ac->ac_f_ex.fe_group;
                sbi->s_mb_last_start = ac->ac_f_ex.fe_start;
                spin_unlock(&sbi->s_md_lock);
        }
        /*
         * As we've just preallocated more space than
         * user requested originally, we store allocated
         * space in a special descriptor.
         */
        if (ac->ac_o_ex.fe_len < ac->ac_b_ex.fe_len)
                ext4_mb_new_preallocation(ac);

}

static void ext4_mb_check_limits(struct ext4_allocation_context *ac,
                                        struct ext4_buddy *e4b,
                                        int finish_group)
{
        struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb);
        struct ext4_free_extent *bex = &ac->ac_b_ex;
        struct ext4_free_extent *gex = &ac->ac_g_ex;
        struct ext4_free_extent ex;
        int max;

        if (ac->ac_status == AC_STATUS_FOUND)
                return;
        /*
         * We don't want to scan for a whole year
         */
        if (ac->ac_found > sbi->s_mb_max_to_scan &&
                        !(ac->ac_flags & EXT4_MB_HINT_FIRST)) {
                ac->ac_status = AC_STATUS_BREAK;
                return;
        }

        /*
         * Haven't found good chunk so far, let's continue
         */
        if (bex->fe_len < gex->fe_len)
                return;

        if ((finish_group || ac->ac_found > sbi->s_mb_min_to_scan)
                        && bex->fe_group == e4b->bd_group) {
                /* recheck chunk's availability - we don't know
                 * when it was found (within this lock-unlock
                 * period or not) */
                max = mb_find_extent(e4b, bex->fe_start, gex->fe_len, &ex);
                if (max >= gex->fe_len) {
                        ext4_mb_use_best_found(ac, e4b);
                        return;
                }
        }
}

/*
 * The routine checks whether found extent is good enough. If it is,
 * then the extent gets marked used and flag is set to the context
 * to stop scanning. Otherwise, the extent is compared with the
 * previous found extent and if new one is better, then it's stored
 * in the context. Later, the best found extent will be used, if
 * mballoc can't find good enough extent.
 *
 * FIXME: real allocation policy is to be designed yet!
 */
static void ext4_mb_measure_extent(struct ext4_allocation_context *ac,
                                        struct ext4_free_extent *ex,
                                        struct ext4_buddy *e4b)
{
        struct ext4_free_extent *bex = &ac->ac_b_ex;
        struct ext4_free_extent *gex = &ac->ac_g_ex;

        BUG_ON(ex->fe_len <= 0);
        BUG_ON(ex->fe_len > EXT4_CLUSTERS_PER_GROUP(ac->ac_sb));
        BUG_ON(ex->fe_start >= EXT4_CLUSTERS_PER_GROUP(ac->ac_sb));
        BUG_ON(ac->ac_status != AC_STATUS_CONTINUE);

        ac->ac_found++;

        /*
         * The special case - take what you catch first
         */
        if (unlikely(ac->ac_flags & EXT4_MB_HINT_FIRST)) {
                *bex = *ex;
                ext4_mb_use_best_found(ac, e4b);
                return;
        }

        /*
         * Let's check whether the chuck is good enough
         */
        if (ex->fe_len == gex->fe_len) {
                *bex = *ex;
                ext4_mb_use_best_found(ac, e4b);
                return;
        }

        /*
         * If this is first found extent, just store it in the context
         */
        if (bex->fe_len == 0) {
                *bex = *ex;
                return;
        }

        /*
         * If new found extent is better, store it in the context
         */
        if (bex->fe_len < gex->fe_len) {
                /* if the request isn't satisfied, any found extent
                 * larger than previous best one is better */
                if (ex->fe_len > bex->fe_len)
                        *bex = *ex;
        } else if (ex->fe_len > gex->fe_len) {
                /* if the request is satisfied, then we try to find
                 * an extent that still satisfy the request, but is
                 * smaller than previous one */
                if (ex->fe_len < bex->fe_len)
                        *bex = *ex;
        }

        ext4_mb_check_limits(ac, e4b, 0);
}

static noinline_for_stack
int ext4_mb_try_best_found(struct ext4_allocation_context *ac,
                                        struct ext4_buddy *e4b)
{
        struct ext4_free_extent ex = ac->ac_b_ex;
        ext4_group_t group = ex.fe_group;
        int max;
        int err;

        BUG_ON(ex.fe_len <= 0);
        err = ext4_mb_load_buddy(ac->ac_sb, group, e4b);
        if (err)
                return err;

        ext4_lock_group(ac->ac_sb, group);
        max = mb_find_extent(e4b, ex.fe_start, ex.fe_len, &ex);

        if (max > 0) {
                ac->ac_b_ex = ex;
                ext4_mb_use_best_found(ac, e4b);
        }

        ext4_unlock_group(ac->ac_sb, group);
        ext4_mb_unload_buddy(e4b);

        return 0;
}

static noinline_for_stack
int ext4_mb_find_by_goal(struct ext4_allocation_context *ac,
                                struct ext4_buddy *e4b)
{
        ext4_group_t group = ac->ac_g_ex.fe_group;
        int max;
        int err;
        struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb);
        struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group);
        struct ext4_free_extent ex;

        if (!(ac->ac_flags & EXT4_MB_HINT_TRY_GOAL))
                return 0;
        if (grp->bb_free == 0)
                return 0;

        err = ext4_mb_load_buddy(ac->ac_sb, group, e4b);
        if (err)
                return err;

        if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info))) {
                ext4_mb_unload_buddy(e4b);
                return 0;
        }

        ext4_lock_group(ac->ac_sb, group);
        max = mb_find_extent(e4b, ac->ac_g_ex.fe_start,
                             ac->ac_g_ex.fe_len, &ex);
        ex.fe_logical = 0xDEADFA11; /* debug value */

        if (max >= ac->ac_g_ex.fe_len && ac->ac_g_ex.fe_len == sbi->s_stripe) {
                ext4_fsblk_t start;

                start = ext4_group_first_block_no(ac->ac_sb, e4b->bd_group) +
                        ex.fe_start;
                /* use do_div to get remainder (would be 64-bit modulo) */
                if (do_div(start, sbi->s_stripe) == 0) {
                        ac->ac_found++;
                        ac->ac_b_ex = ex;
                        ext4_mb_use_best_found(ac, e4b);
                }
        } else if (max >= ac->ac_g_ex.fe_len) {
                BUG_ON(ex.fe_len <= 0);
                BUG_ON(ex.fe_group != ac->ac_g_ex.fe_group);
                BUG_ON(ex.fe_start != ac->ac_g_ex.fe_start);
                ac->ac_found++;
                ac->ac_b_ex = ex;
                ext4_mb_use_best_found(ac, e4b);
        } else if (max > 0 && (ac->ac_flags & EXT4_MB_HINT_MERGE)) {
                /* Sometimes, caller may want to merge even small
                 * number of blocks to an existing extent */
                BUG_ON(ex.fe_len <= 0);
                BUG_ON(ex.fe_group != ac->ac_g_ex.fe_group);
                BUG_ON(ex.fe_start != ac->ac_g_ex.fe_start);
                ac->ac_found++;
                ac->ac_b_ex = ex;
                ext4_mb_use_best_found(ac, e4b);
        }
        ext4_unlock_group(ac->ac_sb, group);
        ext4_mb_unload_buddy(e4b);

        return 0;
}

/*
 * The routine scans buddy structures (not bitmap!) from given order
 * to max order and tries to find big enough chunk to satisfy the req
 */
static noinline_for_stack
void ext4_mb_simple_scan_group(struct ext4_allocation_context *ac,
                                        struct ext4_buddy *e4b)
{
        struct super_block *sb = ac->ac_sb;
        struct ext4_group_info *grp = e4b->bd_info;
        void *buddy;
        int i;
        int k;
        int max;

        BUG_ON(ac->ac_2order <= 0);
        for (i = ac->ac_2order; i < MB_NUM_ORDERS(sb); i++) {
                if (grp->bb_counters[i] == 0)
                        continue;

                buddy = mb_find_buddy(e4b, i, &max);
                BUG_ON(buddy == NULL);

                k = mb_find_next_zero_bit(buddy, max, 0);
                if (k >= max) {
                        ext4_grp_locked_error(ac->ac_sb, e4b->bd_group, 0, 0,
                                "%d free clusters of order %d. But found 0",
                                grp->bb_counters[i], i);
                        ext4_mark_group_bitmap_corrupted(ac->ac_sb,
                                         e4b->bd_group,
                                        EXT4_GROUP_INFO_BBITMAP_CORRUPT);
                        break;
                }
                ac->ac_found++;

                ac->ac_b_ex.fe_len = 1 << i;
                ac->ac_b_ex.fe_start = k << i;
                ac->ac_b_ex.fe_group = e4b->bd_group;

                ext4_mb_use_best_found(ac, e4b);

                BUG_ON(ac->ac_f_ex.fe_len != ac->ac_g_ex.fe_len);

                if (EXT4_SB(sb)->s_mb_stats)
                        atomic_inc(&EXT4_SB(sb)->s_bal_2orders);

                break;
        }
}

/*
 * The routine scans the group and measures all found extents.
 * In order to optimize scanning, caller must pass number of
 * free blocks in the group, so the routine can know upper limit.
 */
static noinline_for_stack
void ext4_mb_complex_scan_group(struct ext4_allocation_context *ac,
                                        struct ext4_buddy *e4b)
{
        struct super_block *sb = ac->ac_sb;
        void *bitmap = e4b->bd_bitmap;
        struct ext4_free_extent ex;
        int i;
        int free;

        free = e4b->bd_info->bb_free;
        if (WARN_ON(free <= 0))
                return;

        i = e4b->bd_info->bb_first_free;

        while (free && ac->ac_status == AC_STATUS_CONTINUE) {
                i = mb_find_next_zero_bit(bitmap,
                                                EXT4_CLUSTERS_PER_GROUP(sb), i);
                if (i >= EXT4_CLUSTERS_PER_GROUP(sb)) {
                        /*
                         * IF we have corrupt bitmap, we won't find any
                         * free blocks even though group info says we
                         * have free blocks
                         */
                        ext4_grp_locked_error(sb, e4b->bd_group, 0, 0,
                                        "%d free clusters as per "
                                        "group info. But bitmap says 0",
                                        free);
                        ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group,
                                        EXT4_GROUP_INFO_BBITMAP_CORRUPT);
                        break;
                }

                mb_find_extent(e4b, i, ac->ac_g_ex.fe_len, &ex);
                if (WARN_ON(ex.fe_len <= 0))
                        break;
                if (free < ex.fe_len) {
                        ext4_grp_locked_error(sb, e4b->bd_group, 0, 0,
                                        "%d free clusters as per "
                                        "group info. But got %d blocks",
                                        free, ex.fe_len);
                        ext4_mark_group_bitmap_corrupted(sb, e4b->bd_group,
                                        EXT4_GROUP_INFO_BBITMAP_CORRUPT);
                        /*
                         * The number of free blocks differs. This mostly
                         * indicate that the bitmap is corrupt. So exit
                         * without claiming the space.
                         */
                        break;
                }
                ex.fe_logical = 0xDEADC0DE; /* debug value */
                ext4_mb_measure_extent(ac, &ex, e4b);

                i += ex.fe_len;
                free -= ex.fe_len;
        }

        ext4_mb_check_limits(ac, e4b, 1);
}

/*
 * This is a special case for storages like raid5
 * we try to find stripe-aligned chunks for stripe-size-multiple requests
 */
static noinline_for_stack
void ext4_mb_scan_aligned(struct ext4_allocation_context *ac,
                                 struct ext4_buddy *e4b)
{
        struct super_block *sb = ac->ac_sb;
        struct ext4_sb_info *sbi = EXT4_SB(sb);
        void *bitmap = e4b->bd_bitmap;
        struct ext4_free_extent ex;
        ext4_fsblk_t first_group_block;
        ext4_fsblk_t a;
        ext4_grpblk_t i;
        int max;

        BUG_ON(sbi->s_stripe == 0);

        /* find first stripe-aligned block in group */
        first_group_block = ext4_group_first_block_no(sb, e4b->bd_group);

        a = first_group_block + sbi->s_stripe - 1;
        do_div(a, sbi->s_stripe);
        i = (a * sbi->s_stripe) - first_group_block;

        while (i < EXT4_CLUSTERS_PER_GROUP(sb)) {
                if (!mb_test_bit(i, bitmap)) {
                        max = mb_find_extent(e4b, i, sbi->s_stripe, &ex);
                        if (max >= sbi->s_stripe) {
                                ac->ac_found++;
                                ex.fe_logical = 0xDEADF00D; /* debug value */
                                ac->ac_b_ex = ex;
                                ext4_mb_use_best_found(ac, e4b);
                                break;
                        }
                }
                i += sbi->s_stripe;
        }
}

/*
 * This is also called BEFORE we load the buddy bitmap.
 * Returns either 1 or 0 indicating that the group is either suitable
 * for the allocation or not.
 */
static bool ext4_mb_good_group(struct ext4_allocation_context *ac,
                                ext4_group_t group, int cr)
{
        ext4_grpblk_t free, fragments;
        int flex_size = ext4_flex_bg_size(EXT4_SB(ac->ac_sb));
        struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group);

        BUG_ON(cr < 0 || cr >= 4);

        if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(grp)))
                return false;

        free = grp->bb_free;
        if (free == 0)
                return false;

        fragments = grp->bb_fragments;
        if (fragments == 0)
                return false;

        switch (cr) {
        case 0:
                BUG_ON(ac->ac_2order == 0);

                /* Avoid using the first bg of a flexgroup for data files */
                if ((ac->ac_flags & EXT4_MB_HINT_DATA) &&
                    (flex_size >= EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME) &&
                    ((group % flex_size) == 0))
                        return false;

                if (free < ac->ac_g_ex.fe_len)
                        return false;

                if (ac->ac_2order >= MB_NUM_ORDERS(ac->ac_sb))
                        return true;

                if (grp->bb_largest_free_order < ac->ac_2order)
                        return false;

                return true;
        case 1:
                if ((free / fragments) >= ac->ac_g_ex.fe_len)
                        return true;
                break;
        case 2:
                if (free >= ac->ac_g_ex.fe_len)
                        return true;
                break;
        case 3:
                return true;
        default:
                BUG();
        }

        return false;
}

/*
 * This could return negative error code if something goes wrong
 * during ext4_mb_init_group(). This should not be called with
 * ext4_lock_group() held.
 */
static int ext4_mb_good_group_nolock(struct ext4_allocation_context *ac,
                                     ext4_group_t group, int cr)
{
        struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group);
        struct super_block *sb = ac->ac_sb;
        struct ext4_sb_info *sbi = EXT4_SB(sb);
        bool should_lock = ac->ac_flags & EXT4_MB_STRICT_CHECK;
        ext4_grpblk_t free;
        int ret = 0;

        if (sbi->s_mb_stats)
                atomic64_inc(&sbi->s_bal_cX_groups_considered[ac->ac_criteria]);
        if (should_lock)
                ext4_lock_group(sb, group);
        free = grp->bb_free;
        if (free == 0)
                goto out;
        if (cr <= 2 && free < ac->ac_g_ex.fe_len)
                goto out;
        if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(grp)))
                goto out;
        if (should_lock)
                ext4_unlock_group(sb, group);

        /* We only do this if the grp has never been initialized */
        if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) {
                struct ext4_group_desc *gdp =
                        ext4_get_group_desc(sb, group, NULL);
                int ret;

                /* cr=0/1 is a very optimistic search to find large
                 * good chunks almost for free.  If buddy data is not
                 * ready, then this optimization makes no sense.  But
                 * we never skip the first block group in a flex_bg,
                 * since this gets used for metadata block allocation,
                 * and we want to make sure we locate metadata blocks
                 * in the first block group in the flex_bg if possible.
                 */
                if (cr < 2 &&
                    (!sbi->s_log_groups_per_flex ||
                     ((group & ((1 << sbi->s_log_groups_per_flex) - 1)) != 0)) &&
                    !(ext4_has_group_desc_csum(sb) &&
                      (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT))))
                        return 0;
                ret = ext4_mb_init_group(sb, group, GFP_NOFS);
                if (ret)
                        return ret;
        }

        if (should_lock)
                ext4_lock_group(sb, group);
        ret = ext4_mb_good_group(ac, group, cr);
out:
        if (should_lock)
                ext4_unlock_group(sb, group);
        return ret;
}

/*
 * Start prefetching @nr block bitmaps starting at @group.
 * Return the next group which needs to be prefetched.
 */
ext4_group_t ext4_mb_prefetch(struct super_block *sb, ext4_group_t group,
                              unsigned int nr, int *cnt)
{
        ext4_group_t ngroups = ext4_get_groups_count(sb);
        struct buffer_head *bh;
        struct blk_plug plug;

        blk_start_plug(&plug);
        while (nr-- > 0) {
                struct ext4_group_desc *gdp = ext4_get_group_desc(sb, group,
                                                                  NULL);
                struct ext4_group_info *grp = ext4_get_group_info(sb, group);

                /*
                 * Prefetch block groups with free blocks; but don't
                 * bother if it is marked uninitialized on disk, since
                 * it won't require I/O to read.  Also only try to
                 * prefetch once, so we avoid getblk() call, which can
                 * be expensive.
                 */
                if (!EXT4_MB_GRP_TEST_AND_SET_READ(grp) &&
                    EXT4_MB_GRP_NEED_INIT(grp) &&
                    ext4_free_group_clusters(sb, gdp) > 0 &&
                    !(ext4_has_group_desc_csum(sb) &&
                      (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)))) {
                        bh = ext4_read_block_bitmap_nowait(sb, group, true);
                        if (bh && !IS_ERR(bh)) {
                                if (!buffer_uptodate(bh) && cnt)
                                        (*cnt)++;
                                brelse(bh);
                        }
                }
                if (++group >= ngroups)
                        group = 0;
        }
        blk_finish_plug(&plug);
        return group;
}

/*
 * Prefetching reads the block bitmap into the buffer cache; but we
 * need to make sure that the buddy bitmap in the page cache has been
 * initialized.  Note that ext4_mb_init_group() will block if the I/O
 * is not yet completed, or indeed if it was not initiated by
 * ext4_mb_prefetch did not start the I/O.
 *
 * TODO: We should actually kick off the buddy bitmap setup in a work
 * queue when the buffer I/O is completed, so that we don't block
 * waiting for the block allocation bitmap read to finish when
 * ext4_mb_prefetch_fini is called from ext4_mb_regular_allocator().
 */
void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
                           unsigned int nr)
{
        while (nr-- > 0) {
                struct ext4_group_desc *gdp = ext4_get_group_desc(sb, group,
                                                                  NULL);
                struct ext4_group_info *grp = ext4_get_group_info(sb, group);

                if (!group)
                        group = ext4_get_groups_count(sb);
                group--;
                grp = ext4_get_group_info(sb, group);

                if (EXT4_MB_GRP_NEED_INIT(grp) &&
                    ext4_free_group_clusters(sb, gdp) > 0 &&
                    !(ext4_has_group_desc_csum(sb) &&
                      (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)))) {
                        if (ext4_mb_init_group(sb, group, GFP_NOFS))
                                break;
                }
        }
}

static noinline_for_stack int
ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
{
        ext4_group_t prefetch_grp = 0, ngroups, group, i;
        int cr = -1;
        int err = 0, first_err = 0;
        unsigned int nr = 0, prefetch_ios = 0;
        struct ext4_sb_info *sbi;
        struct super_block *sb;
        struct ext4_buddy e4b;
        int lost;

        sb = ac->ac_sb;
        sbi = EXT4_SB(sb);
        ngroups = ext4_get_groups_count(sb);
        /* non-extent files are limited to low blocks/groups */
        if (!(ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS)))
                ngroups = sbi->s_blockfile_groups;

        BUG_ON(ac->ac_status == AC_STATUS_FOUND);

        /* first, try the goal */
        err = ext4_mb_find_by_goal(ac, &e4b);
        if (err || ac->ac_status == AC_STATUS_FOUND)
                goto out;

        if (unlikely(ac->ac_flags & EXT4_MB_HINT_GOAL_ONLY))
                goto out;

        /*
         * ac->ac_2order is set only if the fe_len is a power of 2
         * if ac->ac_2order is set we also set criteria to 0 so that we
         * try exact allocation using buddy.
         */
        i = fls(ac->ac_g_ex.fe_len);
        ac->ac_2order = 0;
        /*
         * We search using buddy data only if the order of the request
         * is greater than equal to the sbi_s_mb_order2_reqs
         * You can tune it via /sys/fs/ext4/<partition>/mb_order2_req
         * We also support searching for power-of-two requests only for
         * requests upto maximum buddy size we have constructed.
         */
        if (i >= sbi->s_mb_order2_reqs && i <= MB_NUM_ORDERS(sb)) {
                /*
                 * This should tell if fe_len is exactly power of 2
                 */
                if ((ac->ac_g_ex.fe_len & (~(1 << (i - 1)))) == 0)
                        ac->ac_2order = array_index_nospec(i - 1,
                                                           MB_NUM_ORDERS(sb));
        }

        /* if stream allocation is enabled, use global goal */
        if (ac->ac_flags & EXT4_MB_STREAM_ALLOC) {
                /* TBD: may be hot point */
                spin_lock(&sbi->s_md_lock);
                ac->ac_g_ex.fe_group = sbi->s_mb_last_group;
                ac->ac_g_ex.fe_start = sbi->s_mb_last_start;
                spin_unlock(&sbi->s_md_lock);
        }

        /* Let's just scan groups to find more-less suitable blocks */
        cr = ac->ac_2order ? 0 : 1;
        /*
         * cr == 0 try to get exact allocation,
         * cr == 3  try to get anything
         */
repeat:
        for (; cr < 4 && ac->ac_status == AC_STATUS_CONTINUE; cr++) {
                ac->ac_criteria = cr;
                /*
                 * searching for the right group start
                 * from the goal value specified
                 */
                group = ac->ac_g_ex.fe_group;
                ac->ac_last_optimal_group = group;
                ac->ac_groups_linear_remaining = sbi->s_mb_max_linear_groups;
                prefetch_grp = group;

                for (i = 0; i < ngroups; group = next_linear_group(ac, group, ngroups),
                             i++) {
                        int ret = 0, new_cr;

                        cond_resched();

                        ext4_mb_choose_next_group(ac, &new_cr, &group, ngroups);
                        if (new_cr != cr) {
                                cr = new_cr;
                                goto repeat;
                        }

                        /*
                         * Batch reads of the block allocation bitmaps
                         * to get multiple READs in flight; limit
                         * prefetching at cr=0/1, otherwise mballoc can
                         * spend a lot of time loading imperfect groups
                         */
                        if ((prefetch_grp == group) &&
                            (cr > 1 ||
                             prefetch_ios < sbi->s_mb_prefetch_limit)) {
                                unsigned int curr_ios = prefetch_ios;

                                nr = sbi->s_mb_prefetch;
                                if (ext4_has_feature_flex_bg(sb)) {
                                        nr = 1 << sbi->s_log_groups_per_flex;
                                        nr -= group & (nr - 1);
                                        nr = min(nr, sbi->s_mb_prefetch);
                                }
                                prefetch_grp = ext4_mb_prefetch(sb, group,
                                                        nr, &prefetch_ios);
                                if (prefetch_ios == curr_ios)
                                        nr = 0;
                        }

                        /* This now checks without needing the buddy page */
                        ret = ext4_mb_good_group_nolock(ac, group, cr);
                        if (ret <= 0) {
                                if (!first_err)
                                        first_err = ret;
                                continue;
                        }

                        err = ext4_mb_load_buddy(sb, group, &e4b);
                        if (err)
                                goto out;

                        ext4_lock_group(sb, group);

                        /*
                         * We need to check again after locking the
                         * block group
                         */
                        ret = ext4_mb_good_group(ac, group, cr);
                        if (ret == 0) {
                                ext4_unlock_group(sb, group);
                                ext4_mb_unload_buddy(&e4b);
                                continue;
                        }

                        ac->ac_groups_scanned++;
                        if (cr == 0)
                                ext4_mb_simple_scan_group(ac, &e4b);
                        else if (cr == 1 && sbi->s_stripe &&
                                        !(ac->ac_g_ex.fe_len % sbi->s_stripe))
                                ext4_mb_scan_aligned(ac, &e4b);
                        else
                                ext4_mb_complex_scan_group(ac, &e4b);

                        ext4_unlock_group(sb, group);
                        ext4_mb_unload_buddy(&e4b);

                        if (ac->ac_status != AC_STATUS_CONTINUE)
                                break;
                }
                /* Processed all groups and haven't found blocks */
                if (sbi->s_mb_stats && i == ngroups)
                        atomic64_inc(&sbi->s_bal_cX_failed[cr]);
        }

        if (ac->ac_b_ex.fe_len > 0 && ac->ac_status != AC_STATUS_FOUND &&
            !(ac->ac_flags & EXT4_MB_HINT_FIRST)) {
                /*
                 * We've been searching too long. Let's try to allocate
                 * the best chunk we've found so far
                 */
                ext4_mb_try_best_found(ac, &e4b);
                if (ac->ac_status != AC_STATUS_FOUND) {
                        /*
                         * Someone more lucky has already allocated it.
                         * The only thing we can do is just take first
                         * found block(s)
                         */
                        lost = atomic_inc_return(&sbi->s_mb_lost_chunks);
                        mb_debug(sb, "lost chunk, group: %u, start: %d, len: %d, lost: %d\n",
                                 ac->ac_b_ex.fe_group, ac->ac_b_ex.fe_start,
                                 ac->ac_b_ex.fe_len, lost);

                        ac->ac_b_ex.fe_group = 0;
                        ac->ac_b_ex.fe_start = 0;
                        ac->ac_b_ex.fe_len = 0;
                        ac->ac_status = AC_STATUS_CONTINUE;
                        ac->ac_flags |= EXT4_MB_HINT_FIRST;
                        cr = 3;
                        goto repeat;
                }
        }

        if (sbi->s_mb_stats && ac->ac_status == AC_STATUS_FOUND)
                atomic64_inc(&sbi->s_bal_cX_hits[ac->ac_criteria]);
out:
        if (!err && ac->ac_status != AC_STATUS_FOUND && first_err)
                err = first_err;

        mb_debug(sb, "Best len %d, origin len %d, ac_status %u, ac_flags 0x%x, cr %d ret %d\n",
                 ac->ac_b_ex.fe_len, ac->ac_o_ex.fe_len, ac->ac_status,
                 ac->ac_flags, cr, err);

        if (nr)
                ext4_mb_prefetch_fini(sb, prefetch_grp, nr);

        return err;
}

static void *ext4_mb_seq_groups_start(struct seq_file *seq, loff_t *pos)
{
        struct super_block *sb = PDE_DATA(file_inode(seq->file));
        ext4_group_t group;

        if (*pos < 0 || *pos >= ext4_get_groups_count(sb))
                return NULL;
        group = *pos + 1;
        return (void *) ((unsigned long) group);
}

static void *ext4_mb_seq_groups_next(struct seq_file *seq, void *v, loff_t *pos)
{
        struct super_block *sb = PDE_DATA(file_inode(seq->file));
        ext4_group_t group;

        ++*pos;
        if (*pos < 0 || *pos >= ext4_get_groups_count(sb))
                return NULL;
        group = *pos + 1;
        return (void *) ((unsigned long) group);
}

static int ext4_mb_seq_groups_show(struct seq_file *seq, void *v)
{
        struct super_block *sb = PDE_DATA(file_inode(seq->file));
        ext4_group_t group = (ext4_group_t) ((unsigned long) v);
        int i;
        int err, buddy_loaded = 0;
        struct ext4_buddy e4b;
        struct ext4_group_info *grinfo;
        unsigned char blocksize_bits = min_t(unsigned char,
                                             sb->s_blocksize_bits,
                                             EXT4_MAX_BLOCK_LOG_SIZE);
        struct sg {
                struct ext4_group_info info;
                ext4_grpblk_t counters[EXT4_MAX_BLOCK_LOG_SIZE + 2];
        } sg;

        group--;
        if (group == 0)
                seq_puts(seq, "#group: free  frags first ["
                              " 2^0   2^1   2^2   2^3   2^4   2^5   2^6  "
                              " 2^7   2^8   2^9   2^10  2^11  2^12  2^13  ]\n");

        i = (blocksize_bits + 2) * sizeof(sg.info.bb_counters[0]) +
                sizeof(struct ext4_group_info);

        grinfo = ext4_get_group_info(sb, group);
        /* Load the group info in memory only if not already loaded. */
        if (unlikely(EXT4_MB_GRP_NEED_INIT(grinfo))) {
                err = ext4_mb_load_buddy(sb, group, &e4b);
                if (err) {
                        seq_printf(seq, "#%-5u: I/O error\n", group);
                        return 0;
                }
                buddy_loaded = 1;
        }

        memcpy(&sg, ext4_get_group_info(sb, group), i);

        if (buddy_loaded)
                ext4_mb_unload_buddy(&e4b);

        seq_printf(seq, "#%-5u: %-5u %-5u %-5u [", group, sg.info.bb_free,
                        sg.info.bb_fragments, sg.info.bb_first_free);
        for (i = 0; i <= 13; i++)
                seq_printf(seq, " %-5u", i <= blocksize_bits + 1 ?
                                sg.info.bb_counters[i] : 0);
        seq_puts(seq, " ]\n");

        return 0;
}

static void ext4_mb_seq_groups_stop(struct seq_file *seq, void *v)
{
}

const struct seq_operations ext4_mb_seq_groups_ops = {
        .start  = ext4_mb_seq_groups_start,
        .next   = ext4_mb_seq_groups_next,
        .stop   = ext4_mb_seq_groups_stop,
        .show   = ext4_mb_seq_groups_show,
};

int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset)
{
        struct super_block *sb = (struct super_block *)seq->private;
        struct ext4_sb_info *sbi = EXT4_SB(sb);

        seq_puts(seq, "mballoc:\n");
        if (!sbi->s_mb_stats) {
                seq_puts(seq, "\tmb stats collection turned off.\n");
                seq_puts(seq, "\tTo enable, please write \"1\" to sysfs file mb_stats.\n");
                return 0;
        }
        seq_printf(seq, "\treqs: %u\n", atomic_read(&sbi->s_bal_reqs));
        seq_printf(seq, "\tsuccess: %u\n", atomic_read(&sbi->s_bal_success));

        seq_printf(seq, "\tgroups_scanned: %u\n",  atomic_read(&sbi->s_bal_groups_scanned));

        seq_puts(seq, "\tcr0_stats:\n");
        seq_printf(seq, "\t\thits: %llu\n", atomic64_read(&sbi->s_bal_cX_hits[0]));
        seq_printf(seq, "\t\tgroups_considered: %llu\n",
                   atomic64_read(&sbi->s_bal_cX_groups_considered[0]));
        seq_printf(seq, "\t\tuseless_loops: %llu\n",
                   atomic64_read(&sbi->s_bal_cX_failed[0]));
        seq_printf(seq, "\t\tbad_suggestions: %u\n",
                   atomic_read(&sbi->s_bal_cr0_bad_suggestions));

        seq_puts(seq, "\tcr1_stats:\n");
        seq_printf(seq, "\t\thits: %llu\n", atomic64_read(&sbi->s_bal_cX_hits[1]));
        seq_printf(seq, "\t\tgroups_considered: %llu\n",
                   atomic64_read(&sbi->s_bal_cX_groups_considered[1]));
        seq_printf(seq, "\t\tuseless_loops: %llu\n",
                   atomic64_read(&sbi->s_bal_cX_failed[1]));
        seq_printf(seq, "\t\tbad_suggestions: %u\n",
                   atomic_read(&sbi->s_bal_cr1_bad_suggestions));

        seq_puts(seq, "\tcr2_stats:\n");
        seq_printf(seq, "\t\thits: %llu\n", atomic64_read(&sbi->s_bal_cX_hits[2]));
        seq_printf(seq, "\t\tgroups_considered: %llu\n",
                   atomic64_read(&sbi->s_bal_cX_groups_considered[2]));
        seq_printf(seq, "\t\tuseless_loops: %llu\n",
                   atomic64_read(&sbi->s_bal_cX_failed[2]));

        seq_puts(seq, "\tcr3_stats:\n");
        seq_printf(seq, "\t\thits: %llu\n", atomic64_read(&sbi->s_bal_cX_hits[3]));
        seq_printf(seq, "\t\tgroups_considered: %llu\n",
                   atomic64_read(&sbi->s_bal_cX_groups_considered[3]));
        seq_printf(seq, "\t\tuseless_loops: %llu\n",
                   atomic64_read(&sbi->s_bal_cX_failed[3]));
        seq_printf(seq, "\textents_scanned: %u\n", atomic_read(&sbi->s_bal_ex_scanned));
        seq_printf(seq, "\t\tgoal_hits: %u\n", atomic_read(&sbi->s_bal_goals));
        seq_printf(seq, "\t\t2^n_hits: %u\n", atomic_read(&sbi->s_bal_2orders));
        seq_printf(seq, "\t\tbreaks: %u\n", atomic_read(&sbi->s_bal_breaks));
        seq_printf(seq, "\t\tlost: %u\n", atomic_read(&sbi->s_mb_lost_chunks));

        seq_printf(seq, "\tbuddies_generated: %u/%u\n",
                   atomic_read(&sbi->s_mb_buddies_generated),
                   ext4_get_groups_count(sb));
        seq_printf(seq, "\tbuddies_time_used: %llu\n",
                   atomic64_read(&sbi->s_mb_generation_time));
        seq_printf(seq, "\tpreallocated: %u\n",
                   atomic_read(&sbi->s_mb_preallocated));
        seq_printf(seq, "\tdiscarded: %u\n",
                   atomic_read(&sbi->s_mb_discarded));
        return 0;
}

static void *ext4_mb_seq_structs_summary_start(struct seq_file *seq, loff_t *pos)
{
        struct super_block *sb = PDE_DATA(file_inode(seq->file));
        unsigned long position;

        read_lock(&EXT4_SB(sb)->s_mb_rb_lock);

        if (*pos < 0 || *pos >= MB_NUM_ORDERS(sb) + 1)
                return NULL;
        position = *pos + 1;
        return (void *) ((unsigned long) position);
}

static void *ext4_mb_seq_structs_summary_next(struct seq_file *seq, void *v, loff_t *pos)
{
        struct super_block *sb = PDE_DATA(file_inode(seq->file));
        unsigned long position;

        ++*pos;
        if (*pos < 0 || *pos >= MB_NUM_ORDERS(sb) + 1)
                return NULL;
        position = *pos + 1;
        return (void *) ((unsigned long) position);
}

static int ext4_mb_seq_structs_summary_show(struct seq_file *seq, void *v)
{
        struct super_block *sb = PDE_DATA(file_inode(seq->file));
        struct ext4_sb_info *sbi = EXT4_SB(sb);
        unsigned long position = ((unsigned long) v);
        struct ext4_group_info *grp;
        struct rb_node *n;
        unsigned int count, min, max;