aops.c

Go to the documentation of this file.

#include <linux/errno.h>
#include <linux/fs.h>
#include <linux/gfp.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/swap.h>
#include <linux/buffer_head.h>
#include <linux/writeback.h>
#include <linux/bit_spinlock.h>

#include "aops.h"
#include "attrib.h"
#include "debug.h"
#include "inode.h"
#include "mft.h"
#include "runlist.h"
#include "types.h"
#include "ntfs.h"

static void ntfs_end_buffer_async_read(struct buffer_head *bh, int uptodate)
{
    unsigned long flags;
    struct buffer_head *first, *tmp;
    struct page *page;
    struct inode *vi;
    ntfs_inode *ni;
    int page_uptodate = 1;

    page = bh->b_page;
    vi = page->mapping->host;
    ni = NTFS_I(vi);

    if (likely(uptodate)) {
        loff_t i_size;
        s64 file_ofs, init_size;

        set_buffer_uptodate(bh);

        file_ofs = ((s64)page->index << PAGE_CACHE_SHIFT) +
                bh_offset(bh);
        read_lock_irqsave(&ni->size_lock, flags);
        init_size = ni->initialized_size;
        i_size = i_size_read(vi);
        read_unlock_irqrestore(&ni->size_lock, flags);
        if (unlikely(init_size > i_size)) {
            /* Race with shrinking truncate. */
            init_size = i_size;
        }
        /* Check for the current buffer head overflowing. */
        if (unlikely(file_ofs + bh->b_size > init_size)) {
            int ofs;
            void *kaddr;

            ofs = 0;
            if (file_ofs < init_size)
                ofs = init_size - file_ofs;
            local_irq_save(flags);
            kaddr = kmap_atomic(page);
            memset(kaddr + bh_offset(bh) + ofs, 0,
                    bh->b_size - ofs);
            flush_dcache_page(page);
            kunmap_atomic(kaddr);
            local_irq_restore(flags);
        }
    } else {
        clear_buffer_uptodate(bh);
        SetPageError(page);
        ntfs_error(ni->vol->sb, "Buffer I/O error, logical block "
                "0x%llx.", (unsigned long long)bh->b_blocknr);
    }
    first = page_buffers(page);
    local_irq_save(flags);
    bit_spin_lock(BH_Uptodate_Lock, &first->b_state);
    clear_buffer_async_read(bh);
    unlock_buffer(bh);
    tmp = bh;
    do {
        if (!buffer_uptodate(tmp))
            page_uptodate = 0;
        if (buffer_async_read(tmp)) {
            if (likely(buffer_locked(tmp)))
                goto still_busy;
            /* Async buffers must be locked. */
            BUG();
        }
        tmp = tmp->b_this_page;
    } while (tmp != bh);
    bit_spin_unlock(BH_Uptodate_Lock, &first->b_state);
    local_irq_restore(flags);
    /*
     * If none of the buffers had errors then we can set the page uptodate,
     * but we first have to perform the post read mst fixups, if the
     * attribute is mst protected, i.e. if NInoMstProteced(ni) is true.
     * Note we ignore fixup errors as those are detected when
     * map_mft_record() is called which gives us per record granularity
     * rather than per page granularity.
     */
    if (!NInoMstProtected(ni)) {
        if (likely(page_uptodate && !PageError(page)))
            SetPageUptodate(page);
    } else {
        u8 *kaddr;
        unsigned int i, recs;
        u32 rec_size;

        rec_size = ni->itype.index.block_size;
        recs = PAGE_CACHE_SIZE / rec_size;
        /* Should have been verified before we got here... */
        BUG_ON(!recs);
        local_irq_save(flags);
        kaddr = kmap_atomic(page);
        for (i = 0; i < recs; i++)
            post_read_mst_fixup((NTFS_RECORD*)(kaddr +
                    i * rec_size), rec_size);
        kunmap_atomic(kaddr);
        local_irq_restore(flags);
        flush_dcache_page(page);
        if (likely(page_uptodate && !PageError(page)))
            SetPageUptodate(page);
    }
    unlock_page(page);
    return;
still_busy:
    bit_spin_unlock(BH_Uptodate_Lock, &first->b_state);
    local_irq_restore(flags);
    return;
}

static int ntfs_read_block(struct page *page)
{
    loff_t i_size;
    VCN vcn;
    LCN lcn;
    s64 init_size;
    struct inode *vi;
    ntfs_inode *ni;
    ntfs_volume *vol;
    runlist_element *rl;
    struct buffer_head *bh, *head, *arr[MAX_BUF_PER_PAGE];
    sector_t iblock, lblock, zblock;
    unsigned long flags;
    unsigned int blocksize, vcn_ofs;
    int i, nr;
    unsigned char blocksize_bits;

    vi = page->mapping->host;
    ni = NTFS_I(vi);
    vol = ni->vol;

    /* $MFT/$DATA must have its complete runlist in memory at all times. */
    BUG_ON(!ni->runlist.rl && !ni->mft_no && !NInoAttr(ni));

    blocksize = vol->sb->s_blocksize;
    blocksize_bits = vol->sb->s_blocksize_bits;

    if (!page_has_buffers(page)) {
        create_empty_buffers(page, blocksize, 0);
        if (unlikely(!page_has_buffers(page))) {
            unlock_page(page);
            return -ENOMEM;
        }
    }
    bh = head = page_buffers(page);
    BUG_ON(!bh);

    /*
     * We may be racing with truncate.  To avoid some of the problems we
     * now take a snapshot of the various sizes and use those for the whole
     * of the function.  In case of an extending truncate it just means we
     * may leave some buffers unmapped which are now allocated.  This is
     * not a problem since these buffers will just get mapped when a write
     * occurs.  In case of a shrinking truncate, we will detect this later
     * on due to the runlist being incomplete and if the page is being
     * fully truncated, truncate will throw it away as soon as we unlock
     * it so no need to worry what we do with it.
     */
    iblock = (s64)page->index << (PAGE_CACHE_SHIFT - blocksize_bits);
    read_lock_irqsave(&ni->size_lock, flags);
    lblock = (ni->allocated_size + blocksize - 1) >> blocksize_bits;
    init_size = ni->initialized_size;
    i_size = i_size_read(vi);
    read_unlock_irqrestore(&ni->size_lock, flags);
    if (unlikely(init_size > i_size)) {
        /* Race with shrinking truncate. */
        init_size = i_size;
    }
    zblock = (init_size + blocksize - 1) >> blocksize_bits;

    /* Loop through all the buffers in the page. */
    rl = NULL;
    nr = i = 0;
    do {
        int err = 0;

        if (unlikely(buffer_uptodate(bh)))
            continue;
        if (unlikely(buffer_mapped(bh))) {
            arr[nr++] = bh;
            continue;
        }
        bh->b_bdev = vol->sb->s_bdev;
        /* Is the block within the allowed limits? */
        if (iblock < lblock) {
            bool is_retry = false;

            /* Convert iblock into corresponding vcn and offset. */
            vcn = (VCN)iblock << blocksize_bits >>
                    vol->cluster_size_bits;
            vcn_ofs = ((VCN)iblock << blocksize_bits) &
                    vol->cluster_size_mask;
            if (!rl) {
lock_retry_remap:
                down_read(&ni->runlist.lock);
                rl = ni->runlist.rl;
            }
            if (likely(rl != NULL)) {
                /* Seek to element containing target vcn. */
                while (rl->length && rl[1].vcn <= vcn)
                    rl++;
                lcn = ntfs_rl_vcn_to_lcn(rl, vcn);
            } else
                lcn = LCN_RL_NOT_MAPPED;
            /* Successful remap. */
            if (lcn >= 0) {
                /* Setup buffer head to correct block. */
                bh->b_blocknr = ((lcn << vol->cluster_size_bits)
                        + vcn_ofs) >> blocksize_bits;
                set_buffer_mapped(bh);
                /* Only read initialized data blocks. */
                if (iblock < zblock) {
                    arr[nr++] = bh;
                    continue;
                }
                /* Fully non-initialized data block, zero it. */
                goto handle_zblock;
            }
            /* It is a hole, need to zero it. */
            if (lcn == LCN_HOLE)
                goto handle_hole;
            /* If first try and runlist unmapped, map and retry. */
            if (!is_retry && lcn == LCN_RL_NOT_MAPPED) {
                is_retry = true;
                /*
                 * Attempt to map runlist, dropping lock for
                 * the duration.
                 */
                up_read(&ni->runlist.lock);
                err = ntfs_map_runlist(ni, vcn);
                if (likely(!err))
                    goto lock_retry_remap;
                rl = NULL;
            } else if (!rl)
                up_read(&ni->runlist.lock);
            /*
             * If buffer is outside the runlist, treat it as a
             * hole.  This can happen due to concurrent truncate
             * for example.
             */
            if (err == -ENOENT || lcn == LCN_ENOENT) {
                err = 0;
                goto handle_hole;
            }
            /* Hard error, zero out region. */
            if (!err)
                err = -EIO;
            bh->b_blocknr = -1;
            SetPageError(page);
            ntfs_error(vol->sb, "Failed to read from inode 0x%lx, "
                    "attribute type 0x%x, vcn 0x%llx, "
                    "offset 0x%x because its location on "
                    "disk could not be determined%s "
                    "(error code %i).", ni->mft_no,
                    ni->type, (unsigned long long)vcn,
                    vcn_ofs, is_retry ? " even after "
                    "retrying" : "", err);
        }
        /*
         * Either iblock was outside lblock limits or
         * ntfs_rl_vcn_to_lcn() returned error.  Just zero that portion
         * of the page and set the buffer uptodate.
         */
handle_hole:
        bh->b_blocknr = -1UL;
        clear_buffer_mapped(bh);
handle_zblock:
        zero_user(page, i * blocksize, blocksize);
        if (likely(!err))
            set_buffer_uptodate(bh);
    } while (i++, iblock++, (bh = bh->b_this_page) != head);

    /* Release the lock if we took it. */
    if (rl)
        up_read(&ni->runlist.lock);

    /* Check we have at least one buffer ready for i/o. */
    if (nr) {
        struct buffer_head *tbh;

        /* Lock the buffers. */
        for (i = 0; i < nr; i++) {
            tbh = arr[i];
            lock_buffer(tbh);
            tbh->b_end_io = ntfs_end_buffer_async_read;
            set_buffer_async_read(tbh);
        }
        /* Finally, start i/o on the buffers. */
        for (i = 0; i < nr; i++) {
            tbh = arr[i];
            if (likely(!buffer_uptodate(tbh)))
                submit_bh(READ, tbh);
            else
                ntfs_end_buffer_async_read(tbh, 1);
        }
        return 0;
    }
    /* No i/o was scheduled on any of the buffers. */
    if (likely(!PageError(page)))
        SetPageUptodate(page);
    else /* Signal synchronous i/o error. */
        nr = -EIO;
    unlock_page(page);
    return nr;
}

static int ntfs_readpage(struct file *file, struct page *page)
{
    loff_t i_size;
    struct inode *vi;
    ntfs_inode *ni, *base_ni;
    u8 *addr;
    ntfs_attr_search_ctx *ctx;
    MFT_RECORD *mrec;
    unsigned long flags;
    u32 attr_len;
    int err = 0;

retry_readpage:
    BUG_ON(!PageLocked(page));
    vi = page->mapping->host;
    i_size = i_size_read(vi);
    /* Is the page fully outside i_size? (truncate in progress) */
    if (unlikely(page->index >= (i_size + PAGE_CACHE_SIZE - 1) >>
            PAGE_CACHE_SHIFT)) {
        zero_user(page, 0, PAGE_CACHE_SIZE);
        ntfs_debug("Read outside i_size - truncated?");
        goto done;
    }
    /*
     * This can potentially happen because we clear PageUptodate() during
     * ntfs_writepage() of MstProtected() attributes.
     */
    if (PageUptodate(page)) {
        unlock_page(page);
        return 0;
    }
    ni = NTFS_I(vi);
    /*
     * Only $DATA attributes can be encrypted and only unnamed $DATA
     * attributes can be compressed.  Index root can have the flags set but
     * this means to create compressed/encrypted files, not that the
     * attribute is compressed/encrypted.  Note we need to check for
     * AT_INDEX_ALLOCATION since this is the type of both directory and
     * index inodes.
     */
    if (ni->type != AT_INDEX_ALLOCATION) {
        /* If attribute is encrypted, deny access, just like NT4. */
        if (NInoEncrypted(ni)) {
            BUG_ON(ni->type != AT_DATA);
            err = -EACCES;
            goto err_out;
        }
        /* Compressed data streams are handled in compress.c. */
        if (NInoNonResident(ni) && NInoCompressed(ni)) {
            BUG_ON(ni->type != AT_DATA);
            BUG_ON(ni->name_len);
            return ntfs_read_compressed_block(page);
        }
    }
    /* NInoNonResident() == NInoIndexAllocPresent() */
    if (NInoNonResident(ni)) {
        /* Normal, non-resident data stream. */
        return ntfs_read_block(page);
    }
    /*
     * Attribute is resident, implying it is not compressed or encrypted.
     * This also means the attribute is smaller than an mft record and
     * hence smaller than a page, so can simply zero out any pages with
     * index above 0.  Note the attribute can actually be marked compressed
     * but if it is resident the actual data is not compressed so we are
     * ok to ignore the compressed flag here.
     */
    if (unlikely(page->index > 0)) {
        zero_user(page, 0, PAGE_CACHE_SIZE);
        goto done;
    }
    if (!NInoAttr(ni))
        base_ni = ni;
    else
        base_ni = ni->ext.base_ntfs_ino;
    /* Map, pin, and lock the mft record. */
    mrec = map_mft_record(base_ni);
    if (IS_ERR(mrec)) {
        err = PTR_ERR(mrec);
        goto err_out;
    }
    /*
     * If a parallel write made the attribute non-resident, drop the mft
     * record and retry the readpage.
     */
    if (unlikely(NInoNonResident(ni))) {
        unmap_mft_record(base_ni);
        goto retry_readpage;
    }
    ctx = ntfs_attr_get_search_ctx(base_ni, mrec);
    if (unlikely(!ctx)) {
        err = -ENOMEM;
        goto unm_err_out;
    }
    err = ntfs_attr_lookup(ni->type, ni->name, ni->name_len,
            CASE_SENSITIVE, 0, NULL, 0, ctx);
    if (unlikely(err))
        goto put_unm_err_out;
    attr_len = le32_to_cpu(ctx->attr->data.resident.value_length);
    read_lock_irqsave(&ni->size_lock, flags);
    if (unlikely(attr_len > ni->initialized_size))
        attr_len = ni->initialized_size;
    i_size = i_size_read(vi);
    read_unlock_irqrestore(&ni->size_lock, flags);
    if (unlikely(attr_len > i_size)) {
        /* Race with shrinking truncate. */
        attr_len = i_size;
    }
    addr = kmap_atomic(page);
    /* Copy the data to the page. */
    memcpy(addr, (u8*)ctx->attr +
            le16_to_cpu(ctx->attr->data.resident.value_offset),
            attr_len);
    /* Zero the remainder of the page. */
    memset(addr + attr_len, 0, PAGE_CACHE_SIZE - attr_len);
    flush_dcache_page(page);
    kunmap_atomic(addr);
put_unm_err_out:
    ntfs_attr_put_search_ctx(ctx);
unm_err_out:
    unmap_mft_record(base_ni);
done:
    SetPageUptodate(page);
err_out:
    unlock_page(page);
    return err;
}

#ifdef NTFS_RW

static int ntfs_write_block(struct page *page, struct writeback_control *wbc)
{
    VCN vcn;
    LCN lcn;
    s64 initialized_size;
    loff_t i_size;
    sector_t block, dblock, iblock;
    struct inode *vi;
    ntfs_inode *ni;
    ntfs_volume *vol;
    runlist_element *rl;
    struct buffer_head *bh, *head;
    unsigned long flags;
    unsigned int blocksize, vcn_ofs;
    int err;
    bool need_end_writeback;
    unsigned char blocksize_bits;

    vi = page->mapping->host;
    ni = NTFS_I(vi);
    vol = ni->vol;

    ntfs_debug("Entering for inode 0x%lx, attribute type 0x%x, page index "
            "0x%lx.", ni->mft_no, ni->type, page->index);

    BUG_ON(!NInoNonResident(ni));
    BUG_ON(NInoMstProtected(ni));
    blocksize = vol->sb->s_blocksize;
    blocksize_bits = vol->sb->s_blocksize_bits;
    if (!page_has_buffers(page)) {
        BUG_ON(!PageUptodate(page));
        create_empty_buffers(page, blocksize,
                (1 << BH_Uptodate) | (1 << BH_Dirty));
        if (unlikely(!page_has_buffers(page))) {
            ntfs_warning(vol->sb, "Error allocating page "
                    "buffers.  Redirtying page so we try "
                    "again later.");
            /*
             * Put the page back on mapping->dirty_pages, but leave
             * its buffers' dirty state as-is.
             */
            redirty_page_for_writepage(wbc, page);
            unlock_page(page);
            return 0;
        }
    }
    bh = head = page_buffers(page);
    BUG_ON(!bh);

    /* NOTE: Different naming scheme to ntfs_read_block()! */

    /* The first block in the page. */
    block = (s64)page->index << (PAGE_CACHE_SHIFT - blocksize_bits);

    read_lock_irqsave(&ni->size_lock, flags);
    i_size = i_size_read(vi);
    initialized_size = ni->initialized_size;
    read_unlock_irqrestore(&ni->size_lock, flags);

    /* The first out of bounds block for the data size. */
    dblock = (i_size + blocksize - 1) >> blocksize_bits;

    /* The last (fully or partially) initialized block. */
    iblock = initialized_size >> blocksize_bits;

    /*
     * Be very careful.  We have no exclusion from __set_page_dirty_buffers
     * here, and the (potentially unmapped) buffers may become dirty at
     * any time.  If a buffer becomes dirty here after we've inspected it
     * then we just miss that fact, and the page stays dirty.
     *
     * Buffers outside i_size may be dirtied by __set_page_dirty_buffers;
     * handle that here by just cleaning them.
     */

    /*
     * Loop through all the buffers in the page, mapping all the dirty
     * buffers to disk addresses and handling any aliases from the
     * underlying block device's mapping.
     */
    rl = NULL;
    err = 0;
    do {
        bool is_retry = false;

        if (unlikely(block >= dblock)) {
            /*
             * Mapped buffers outside i_size will occur, because
             * this page can be outside i_size when there is a
             * truncate in progress. The contents of such buffers
             * were zeroed by ntfs_writepage().
             *
             * FIXME: What about the small race window where
             * ntfs_writepage() has not done any clearing because
             * the page was within i_size but before we get here,
             * vmtruncate() modifies i_size?
             */
            clear_buffer_dirty(bh);
            set_buffer_uptodate(bh);
            continue;
        }

        /* Clean buffers are not written out, so no need to map them. */
        if (!buffer_dirty(bh))
            continue;

        /* Make sure we have enough initialized size. */
        if (unlikely((block >= iblock) &&
                (initialized_size < i_size))) {
            /*
             * If this page is fully outside initialized size, zero
             * out all pages between the current initialized size
             * and the current page. Just use ntfs_readpage() to do
             * the zeroing transparently.
             */
            if (block > iblock) {
                // TODO:
                // For each page do:
                // - read_cache_page()
                // Again for each page do:
                // - wait_on_page_locked()
                // - Check (PageUptodate(page) &&
                //          !PageError(page))
                // Update initialized size in the attribute and
                // in the inode.
                // Again, for each page do:
                //  __set_page_dirty_buffers();
                // page_cache_release()
                // We don't need to wait on the writes.
                // Update iblock.
            }
            /*
             * The current page straddles initialized size. Zero
             * all non-uptodate buffers and set them uptodate (and
             * dirty?). Note, there aren't any non-uptodate buffers
             * if the page is uptodate.
             * FIXME: For an uptodate page, the buffers may need to
             * be written out because they were not initialized on
             * disk before.
             */
            if (!PageUptodate(page)) {
                // TODO:
                // Zero any non-uptodate buffers up to i_size.
                // Set them uptodate and dirty.
            }
            // TODO:
            // Update initialized size in the attribute and in the
            // inode (up to i_size).
            // Update iblock.
            // FIXME: This is inefficient. Try to batch the two
            // size changes to happen in one go.
            ntfs_error(vol->sb, "Writing beyond initialized size "
                    "is not supported yet. Sorry.");
            err = -EOPNOTSUPP;
            break;
            // Do NOT set_buffer_new() BUT DO clear buffer range
            // outside write request range.
            // set_buffer_uptodate() on complete buffers as well as
            // set_buffer_dirty().
        }

        /* No need to map buffers that are already mapped. */
        if (buffer_mapped(bh))
            continue;

        /* Unmapped, dirty buffer. Need to map it. */
        bh->b_bdev = vol->sb->s_bdev;

        /* Convert block into corresponding vcn and offset. */
        vcn = (VCN)block << blocksize_bits;
        vcn_ofs = vcn & vol->cluster_size_mask;
        vcn >>= vol->cluster_size_bits;
        if (!rl) {
lock_retry_remap:
            down_read(&ni->runlist.lock);
            rl = ni->runlist.rl;
        }
        if (likely(rl != NULL)) {
            /* Seek to element containing target vcn. */
            while (rl->length && rl[1].vcn <= vcn)
                rl++;
            lcn = ntfs_rl_vcn_to_lcn(rl, vcn);
        } else
            lcn = LCN_RL_NOT_MAPPED;
        /* Successful remap. */
        if (lcn >= 0) {
            /* Setup buffer head to point to correct block. */
            bh->b_blocknr = ((lcn << vol->cluster_size_bits) +
                    vcn_ofs) >> blocksize_bits;
            set_buffer_mapped(bh);
            continue;
        }
        /* It is a hole, need to instantiate it. */
        if (lcn == LCN_HOLE) {
            u8 *kaddr;
            unsigned long *bpos, *bend;

            /* Check if the buffer is zero. */
            kaddr = kmap_atomic(page);
            bpos = (unsigned long *)(kaddr + bh_offset(bh));
            bend = (unsigned long *)((u8*)bpos + blocksize);
            do {
                if (unlikely(*bpos))
                    break;
            } while (likely(++bpos < bend));
            kunmap_atomic(kaddr);
            if (bpos == bend) {
                /*
                 * Buffer is zero and sparse, no need to write
                 * it.
                 */
                bh->b_blocknr = -1;
                clear_buffer_dirty(bh);
                continue;
            }
            // TODO: Instantiate the hole.
            // clear_buffer_new(bh);
            // unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
            ntfs_error(vol->sb, "Writing into sparse regions is "
                    "not supported yet. Sorry.");
            err = -EOPNOTSUPP;
            break;
        }
        /* If first try and runlist unmapped, map and retry. */
        if (!is_retry && lcn == LCN_RL_NOT_MAPPED) {
            is_retry = true;
            /*
             * Attempt to map runlist, dropping lock for
             * the duration.
             */
            up_read(&ni->runlist.lock);
            err = ntfs_map_runlist(ni, vcn);
            if (likely(!err))
                goto lock_retry_remap;
            rl = NULL;
        } else if (!rl)
            up_read(&ni->runlist.lock);
        /*
         * If buffer is outside the runlist, truncate has cut it out
         * of the runlist.  Just clean and clear the buffer and set it
         * uptodate so it can get discarded by the VM.
         */
        if (err == -ENOENT || lcn == LCN_ENOENT) {
            bh->b_blocknr = -1;
            clear_buffer_dirty(bh);
            zero_user(page, bh_offset(bh), blocksize);
            set_buffer_uptodate(bh);
            err = 0;
            continue;
        }
        /* Failed to map the buffer, even after retrying. */
        if (!err)
            err = -EIO;
        bh->b_blocknr = -1;
        ntfs_error(vol->sb, "Failed to write to inode 0x%lx, "
                "attribute type 0x%x, vcn 0x%llx, offset 0x%x "
                "because its location on disk could not be "
                "determined%s (error code %i).", ni->mft_no,
                ni->type, (unsigned long long)vcn,
                vcn_ofs, is_retry ? " even after "
                "retrying" : "", err);
        break;
    } while (block++, (bh = bh->b_this_page) != head);

    /* Release the lock if we took it. */
    if (rl)
        up_read(&ni->runlist.lock);

    /* For the error case, need to reset bh to the beginning. */
    bh = head;

    /* Just an optimization, so ->readpage() is not called later. */
    if (unlikely(!PageUptodate(page))) {
        int uptodate = 1;
        do {
            if (!buffer_uptodate(bh)) {
                uptodate = 0;
                bh = head;
                break;
            }
        } while ((bh = bh->b_this_page) != head);
        if (uptodate)
            SetPageUptodate(page);
    }

    /* Setup all mapped, dirty buffers for async write i/o. */
    do {
        if (buffer_mapped(bh) && buffer_dirty(bh)) {
            lock_buffer(bh);
            if (test_clear_buffer_dirty(bh)) {
                BUG_ON(!buffer_uptodate(bh));
                mark_buffer_async_write(bh);
            } else
                unlock_buffer(bh);
        } else if (unlikely(err)) {
            /*
             * For the error case. The buffer may have been set
             * dirty during attachment to a dirty page.
             */
            if (err != -ENOMEM)
                clear_buffer_dirty(bh);
        }
    } while ((bh = bh->b_this_page) != head);

    if (unlikely(err)) {
        // TODO: Remove the -EOPNOTSUPP check later on...
        if (unlikely(err == -EOPNOTSUPP))
            err = 0;
        else if (err == -ENOMEM) {
            ntfs_warning(vol->sb, "Error allocating memory. "
                    "Redirtying page so we try again "
                    "later.");
            /*
             * Put the page back on mapping->dirty_pages, but
             * leave its buffer's dirty state as-is.
             */
            redirty_page_for_writepage(wbc, page);
            err = 0;
        } else
            SetPageError(page);
    }

    BUG_ON(PageWriteback(page));
    set_page_writeback(page);   /* Keeps try_to_free_buffers() away. */

    /* Submit the prepared buffers for i/o. */
    need_end_writeback = true;
    do {
        struct buffer_head *next = bh->b_this_page;
        if (buffer_async_write(bh)) {
            submit_bh(WRITE, bh);
            need_end_writeback = false;
        }
        bh = next;
    } while (bh != head);
    unlock_page(page);

    /* If no i/o was started, need to end_page_writeback(). */
    if (unlikely(need_end_writeback))
        end_page_writeback(page);

    ntfs_debug("Done.");
    return err;
}

static int ntfs_write_mst_block(struct page *page,
        struct writeback_control *wbc)
{
    sector_t block, dblock, rec_block;
    struct inode *vi = page->mapping->host;
    ntfs_inode *ni = NTFS_I(vi);
    ntfs_volume *vol = ni->vol;
    u8 *kaddr;
    unsigned int rec_size = ni->itype.index.block_size;
    ntfs_inode *locked_nis[PAGE_CACHE_SIZE / rec_size];
    struct buffer_head *bh, *head, *tbh, *rec_start_bh;
    struct buffer_head *bhs[MAX_BUF_PER_PAGE];
    runlist_element *rl;
    int i, nr_locked_nis, nr_recs, nr_bhs, max_bhs, bhs_per_rec, err, err2;
    unsigned bh_size, rec_size_bits;
    bool sync, is_mft, page_is_dirty, rec_is_dirty;
    unsigned char bh_size_bits;

    ntfs_debug("Entering for inode 0x%lx, attribute type 0x%x, page index "
            "0x%lx.", vi->i_ino, ni->type, page->index);
    BUG_ON(!NInoNonResident(ni));
    BUG_ON(!NInoMstProtected(ni));
    is_mft = (S_ISREG(vi->i_mode) && !vi->i_ino);
    /*
     * NOTE: ntfs_write_mst_block() would be called for $MFTMirr if a page
     * in its page cache were to be marked dirty.  However this should
     * never happen with the current driver and considering we do not
     * handle this case here we do want to BUG(), at least for now.
     */
    BUG_ON(!(is_mft || S_ISDIR(vi->i_mode) ||
            (NInoAttr(ni) && ni->type == AT_INDEX_ALLOCATION)));
    bh_size = vol->sb->s_blocksize;
    bh_size_bits = vol->sb->s_blocksize_bits;
    max_bhs = PAGE_CACHE_SIZE / bh_size;
    BUG_ON(!max_bhs);
    BUG_ON(max_bhs > MAX_BUF_PER_PAGE);

    /* Were we called for sync purposes? */
    sync = (wbc->sync_mode == WB_SYNC_ALL);

    /* Make sure we have mapped buffers. */
    bh = head = page_buffers(page);
    BUG_ON(!bh);

    rec_size_bits = ni->itype.index.block_size_bits;
    BUG_ON(!(PAGE_CACHE_SIZE >> rec_size_bits));
    bhs_per_rec = rec_size >> bh_size_bits;
    BUG_ON(!bhs_per_rec);

    /* The first block in the page. */
    rec_block = block = (sector_t)page->index <<
            (PAGE_CACHE_SHIFT - bh_size_bits);

    /* The first out of bounds block for the data size. */
    dblock = (i_size_read(vi) + bh_size - 1) >> bh_size_bits;

    rl = NULL;
    err = err2 = nr_bhs = nr_recs = nr_locked_nis = 0;
    page_is_dirty = rec_is_dirty = false;
    rec_start_bh = NULL;
    do {
        bool is_retry = false;

        if (likely(block < rec_block)) {
            if (unlikely(block >= dblock)) {
                clear_buffer_dirty(bh);
                set_buffer_uptodate(bh);
                continue;
            }
            /*
             * This block is not the first one in the record.  We
             * ignore the buffer's dirty state because we could
             * have raced with a parallel mark_ntfs_record_dirty().
             */
            if (!rec_is_dirty)
                continue;
            if (unlikely(err2)) {
                if (err2 != -ENOMEM)
                    clear_buffer_dirty(bh);
                continue;
            }
        } else /* if (block == rec_block) */ {
            BUG_ON(block > rec_block);
            /* This block is the first one in the record. */
            rec_block += bhs_per_rec;
            err2 = 0;
            if (unlikely(block >= dblock)) {
                clear_buffer_dirty(bh);
                continue;
            }
            if (!buffer_dirty(bh)) {
                /* Clean records are not written out. */
                rec_is_dirty = false;
                continue;
            }
            rec_is_dirty = true;
            rec_start_bh = bh;
        }
        /* Need to map the buffer if it is not mapped already. */
        if (unlikely(!buffer_mapped(bh))) {
            VCN vcn;
            LCN lcn;
            unsigned int vcn_ofs;

            bh->b_bdev = vol->sb->s_bdev;
            /* Obtain the vcn and offset of the current block. */
            vcn = (VCN)block << bh_size_bits;
            vcn_ofs = vcn & vol->cluster_size_mask;
            vcn >>= vol->cluster_size_bits;
            if (!rl) {
lock_retry_remap:
                down_read(&ni->runlist.lock);
                rl = ni->runlist.rl;
            }
            if (likely(rl != NULL)) {
                /* Seek to element containing target vcn. */
                while (rl->length && rl[1].vcn <= vcn)
                    rl++;
                lcn = ntfs_rl_vcn_to_lcn(rl, vcn);
            } else
                lcn = LCN_RL_NOT_MAPPED;
            /* Successful remap. */
            if (likely(lcn >= 0)) {
                /* Setup buffer head to correct block. */
                bh->b_blocknr = ((lcn <<
                        vol->cluster_size_bits) +
                        vcn_ofs) >> bh_size_bits;
                set_buffer_mapped(bh);
            } else {
                /*
                 * Remap failed.  Retry to map the runlist once
                 * unless we are working on $MFT which always
                 * has the whole of its runlist in memory.
                 */
                if (!is_mft && !is_retry &&
                        lcn == LCN_RL_NOT_MAPPED) {
                    is_retry = true;
                    /*
                     * Attempt to map runlist, dropping
                     * lock for the duration.
                     */
                    up_read(&ni->runlist.lock);
                    err2 = ntfs_map_runlist(ni, vcn);
                    if (likely(!err2))
                        goto lock_retry_remap;
                    if (err2 == -ENOMEM)
                        page_is_dirty = true;
                    lcn = err2;
                } else {
                    err2 = -EIO;
                    if (!rl)
                        up_read(&ni->runlist.lock);
                }
                /* Hard error.  Abort writing this record. */
                if (!err || err == -ENOMEM)
                    err = err2;
                bh->b_blocknr = -1;
                ntfs_error(vol->sb, "Cannot write ntfs record "
                        "0x%llx (inode 0x%lx, "
                        "attribute type 0x%x) because "
                        "its location on disk could "
                        "not be determined (error "
                        "code %lli).",
                        (long long)block <<
                        bh_size_bits >>
                        vol->mft_record_size_bits,
                        ni->mft_no, ni->type,
                        (long long)lcn);
                /*
                 * If this is not the first buffer, remove the
                 * buffers in this record from the list of
                 * buffers to write and clear their dirty bit
                 * if not error -ENOMEM.
                 */
                if (rec_start_bh != bh) {
                    while (bhs[--nr_bhs] != rec_start_bh)
                        ;
                    if (err2 != -ENOMEM) {
                        do {
                            clear_buffer_dirty(
                                rec_start_bh);
                        } while ((rec_start_bh =
                                rec_start_bh->
                                b_this_page) !=
                                bh);
                    }
                }
                continue;
            }
        }
        BUG_ON(!buffer_uptodate(bh));
        BUG_ON(nr_bhs >= max_bhs);
        bhs[nr_bhs++] = bh;
    } while (block++, (bh = bh->b_this_page) != head);
    if (unlikely(rl))
        up_read(&ni->runlist.lock);
    /* If there were no dirty buffers, we are done. */
    if (!nr_bhs)
        goto done;
    /* Map the page so we can access its contents. */
    kaddr = kmap(page);
    /* Clear the page uptodate flag whilst the mst fixups are applied. */
    BUG_ON(!PageUptodate(page));
    ClearPageUptodate(page);
    for (i = 0; i < nr_bhs; i++) {
        unsigned int ofs;

        /* Skip buffers which are not at the beginning of records. */
        if (i % bhs_per_rec)
            continue;
        tbh = bhs[i];
        ofs = bh_offset(tbh);
        if (is_mft) {
            ntfs_inode *tni;
            unsigned long mft_no;

            /* Get the mft record number. */
            mft_no = (((s64)page->index << PAGE_CACHE_SHIFT) + ofs)
                    >> rec_size_bits;
            /* Check whether to write this mft record. */
            tni = NULL;
            if (!ntfs_may_write_mft_record(vol, mft_no,
                    (MFT_RECORD*)(kaddr + ofs), &tni)) {
                /*
                 * The record should not be written.  This
                 * means we need to redirty the page before
                 * returning.
                 */
                page_is_dirty = true;
                /*
                 * Remove the buffers in this mft record from
                 * the list of buffers to write.
                 */
                do {
                    bhs[i] = NULL;
                } while (++i % bhs_per_rec);
                continue;
            }
            /*
             * The record should be written.  If a locked ntfs
             * inode was returned, add it to the array of locked
             * ntfs inodes.
             */
            if (tni)
                locked_nis[nr_locked_nis++] = tni;
        }
        /* Apply the mst protection fixups. */
        err2 = pre_write_mst_fixup((NTFS_RECORD*)(kaddr + ofs),
                rec_size);
        if (unlikely(err2)) {
            if (!err || err == -ENOMEM)
                err = -EIO;
            ntfs_error(vol->sb, "Failed to apply mst fixups "
                    "(inode 0x%lx, attribute type 0x%x, "
                    "page index 0x%lx, page offset 0x%x)!"
                    "  Unmount and run chkdsk.", vi->i_ino,
                    ni->type, page->index, ofs);
            /*
             * Mark all the buffers in this record clean as we do
             * not want to write corrupt data to disk.
             */
            do {
                clear_buffer_dirty(bhs[i]);
                bhs[i] = NULL;
            } while (++i % bhs_per_rec);
            continue;
        }
        nr_recs++;
    }
    /* If no records are to be written out, we are done. */
    if (!nr_recs)
        goto unm_done;
    flush_dcache_page(page);
    /* Lock buffers and start synchronous write i/o on them. */
    for (i = 0; i < nr_bhs; i++) {
        tbh = bhs[i];
        if (!tbh)
            continue;
        if (!trylock_buffer(tbh))
            BUG();
        /* The buffer dirty state is now irrelevant, just clean it. */
        clear_buffer_dirty(tbh);
        BUG_ON(!buffer_uptodate(tbh));
        BUG_ON(!buffer_mapped(tbh));
        get_bh(tbh);
        tbh->b_end_io = end_buffer_write_sync;
        submit_bh(WRITE, tbh);
    }
    /* Synchronize the mft mirror now if not @sync. */
    if (is_mft && !sync)
        goto do_mirror;
do_wait:
    /* Wait on i/o completion of buffers. */
    for (i = 0; i < nr_bhs; i++) {
        tbh = bhs[i];
        if (!tbh)
            continue;
        wait_on_buffer(tbh);
        if (unlikely(!buffer_uptodate(tbh))) {
            ntfs_error(vol->sb, "I/O error while writing ntfs "
                    "record buffer (inode 0x%lx, "
                    "attribute type 0x%x, page index "
                    "0x%lx, page offset 0x%lx)!  Unmount "
                    "and run chkdsk.", vi->i_ino, ni->type,
                    page->index, bh_offset(tbh));
            if (!err || err == -ENOMEM)
                err = -EIO;
            /*
             * Set the buffer uptodate so the page and buffer
             * states do not become out of sync.
             */
            set_buffer_uptodate(tbh);
        }
    }
    /* If @sync, now synchronize the mft mirror. */
    if (is_mft && sync) {
do_mirror:
        for (i = 0; i < nr_bhs; i++) {
            unsigned long mft_no;
            unsigned int ofs;

            /*
             * Skip buffers which are not at the beginning of
             * records.
             */
            if (i % bhs_per_rec)
                continue;
            tbh = bhs[i];
            /* Skip removed buffers (and hence records). */
            if (!tbh)
                continue;
            ofs = bh_offset(tbh);
            /* Get the mft record number. */
            mft_no = (((s64)page->index << PAGE_CACHE_SHIFT) + ofs)
                    >> rec_size_bits;
            if (mft_no < vol->mftmirr_size)
                ntfs_sync_mft_mirror(vol, mft_no,
                        (MFT_RECORD*)(kaddr + ofs),
                        sync);
        }
        if (!sync)
            goto do_wait;
    }
    /* Remove the mst protection fixups again. */
    for (i = 0; i < nr_bhs; i++) {
        if (!(i % bhs_per_rec)) {
            tbh = bhs[i];
            if (!tbh)
                continue;
            post_write_mst_fixup((NTFS_RECORD*)(kaddr +
                    bh_offset(tbh)));
        }
    }
    flush_dcache_page(page);
unm_done:
    /* Unlock any locked inodes. */
    while (nr_locked_nis-- > 0) {
        ntfs_inode *tni, *base_tni;
        
        tni = locked_nis[nr_locked_nis];
        /* Get the base inode. */
        mutex_lock(&tni->extent_lock);
        if (tni->nr_extents >= 0)
            base_tni = tni;
        else {
            base_tni = tni->ext.base_ntfs_ino;
            BUG_ON(!base_tni);
        }
        mutex_unlock(&tni->extent_lock);
        ntfs_debug("Unlocking %s inode 0x%lx.",
                tni == base_tni ? "base" : "extent",
                tni->mft_no);
        mutex_unlock(&tni->mrec_lock);
        atomic_dec(&tni->count);
        iput(VFS_I(base_tni));
    }
    SetPageUptodate(page);
    kunmap(page);
done:
    if (unlikely(err && err != -ENOMEM)) {
        /*
         * Set page error if there is only one ntfs record in the page.
         * Otherwise we would loose per-record granularity.
         */
        if (ni->itype.index.block_size == PAGE_CACHE_SIZE)
            SetPageError(page);
        NVolSetErrors(vol);
    }
    if (page_is_dirty) {
        ntfs_debug("Page still contains one or more dirty ntfs "
                "records.  Redirtying the page starting at "
                "record 0x%lx.", page->index <<
                (PAGE_CACHE_SHIFT - rec_size_bits));
        redirty_page_for_writepage(wbc, page);
        unlock_page(page);
    } else {
        /*
         * Keep the VM happy.  This must be done otherwise the
         * radix-tree tag PAGECACHE_TAG_DIRTY remains set even though
         * the page is clean.
         */
        BUG_ON(PageWriteback(page));
        set_page_writeback(page);
        unlock_page(page);
        end_page_writeback(page);
    }
    if (likely(!err))
        ntfs_debug("Done.");
    return err;
}

static int ntfs_writepage(struct page *page, struct writeback_control *wbc)
{
    loff_t i_size;
    struct inode *vi = page->mapping->host;
    ntfs_inode *base_ni = NULL, *ni = NTFS_I(vi);
    char *addr;
    ntfs_attr_search_ctx *ctx = NULL;
    MFT_RECORD *m = NULL;
    u32 attr_len;
    int err;

retry_writepage:
    BUG_ON(!PageLocked(page));
    i_size = i_size_read(vi);
    /* Is the page fully outside i_size? (truncate in progress) */
    if (unlikely(page->index >= (i_size + PAGE_CACHE_SIZE - 1) >>
            PAGE_CACHE_SHIFT)) {
        /*
         * The page may have dirty, unmapped buffers.  Make them
         * freeable here, so the page does not leak.
         */
        block_invalidatepage(page, 0);
        unlock_page(page);
        ntfs_debug("Write outside i_size - truncated?");
        return 0;
    }
    /*
     * Only $DATA attributes can be encrypted and only unnamed $DATA
     * attributes can be compressed.  Index root can have the flags set but
     * this means to create compressed/encrypted files, not that the
     * attribute is compressed/encrypted.  Note we need to check for
     * AT_INDEX_ALLOCATION since this is the type of both directory and
     * index inodes.
     */
    if (ni->type != AT_INDEX_ALLOCATION) {
        /* If file is encrypted, deny access, just like NT4. */
        if (NInoEncrypted(ni)) {
            unlock_page(page);
            BUG_ON(ni->type != AT_DATA);
            ntfs_debug("Denying write access to encrypted file.");
            return -EACCES;
        }
        /* Compressed data streams are handled in compress.c. */
        if (NInoNonResident(ni) && NInoCompressed(ni)) {
            BUG_ON(ni->type != AT_DATA);
            BUG_ON(ni->name_len);
            // TODO: Implement and replace this with
            // return ntfs_write_compressed_block(page);
            unlock_page(page);
            ntfs_error(vi->i_sb, "Writing to compressed files is "
                    "not supported yet.  Sorry.");
            return -EOPNOTSUPP;
        }
        // TODO: Implement and remove this check.
        if (NInoNonResident(ni) && NInoSparse(ni)) {
            unlock_page(page);
            ntfs_error(vi->i_sb, "Writing to sparse files is not "
                    "supported yet.  Sorry.");
            return -EOPNOTSUPP;
        }
    }
    /* NInoNonResident() == NInoIndexAllocPresent() */
    if (NInoNonResident(ni)) {
        /* We have to zero every time due to mmap-at-end-of-file. */
        if (page->index >= (i_size >> PAGE_CACHE_SHIFT)) {
            /* The page straddles i_size. */
            unsigned int ofs = i_size & ~PAGE_CACHE_MASK;
            zero_user_segment(page, ofs, PAGE_CACHE_SIZE);
        }
        /* Handle mst protected attributes. */
        if (NInoMstProtected(ni))
            return ntfs_write_mst_block(page, wbc);
        /* Normal, non-resident data stream. */
        return ntfs_write_block(page, wbc);
    }
    /*
     * Attribute is resident, implying it is not compressed, encrypted, or
     * mst protected.  This also means the attribute is smaller than an mft
     * record and hence smaller than a page, so can simply return error on
     * any pages with index above 0.  Note the attribute can actually be
     * marked compressed but if it is resident the actual data is not
     * compressed so we are ok to ignore the compressed flag here.
     */
    BUG_ON(page_has_buffers(page));
    BUG_ON(!PageUptodate(page));
    if (unlikely(page->index > 0)) {
        ntfs_error(vi->i_sb, "BUG()! page->index (0x%lx) > 0.  "
                "Aborting write.", page->index);
        BUG_ON(PageWriteback(page));
        set_page_writeback(page);
        unlock_page(page);
        end_page_writeback(page);
        return -EIO;
    }
    if (!NInoAttr(ni))
        base_ni = ni;
    else
        base_ni = ni->ext.base_ntfs_ino;
    /* Map, pin, and lock the mft record. */
    m = map_mft_record(base_ni);
    if (IS_ERR(m)) {
        err = PTR_ERR(m);
        m = NULL;
        ctx = NULL;
        goto err_out;
    }
    /*
     * If a parallel write made the attribute non-resident, drop the mft
     * record and retry the writepage.
     */
    if (unlikely(NInoNonResident(ni))) {
        unmap_mft_record(base_ni);
        goto retry_writepage;
    }
    ctx = ntfs_attr_get_search_ctx(base_ni, m);
    if (unlikely(!ctx)) {
        err = -ENOMEM;
        goto err_out;
    }
    err = ntfs_attr_lookup(ni->type, ni->name, ni->name_len,
            CASE_SENSITIVE, 0, NULL, 0, ctx);
    if (unlikely(err))
        goto err_out;
    /*
     * Keep the VM happy.  This must be done otherwise the radix-tree tag
     * PAGECACHE_TAG_DIRTY remains set even though the page is clean.
     */
    BUG_ON(PageWriteback(page));
    set_page_writeback(page);
    unlock_page(page);
    attr_len = le32_to_cpu(ctx->attr->data.resident.value_length);
    i_size = i_size_read(vi);
    if (unlikely(attr_len > i_size)) {
        /* Race with shrinking truncate or a failed truncate. */
        attr_len = i_size;
        /*
         * If the truncate failed, fix it up now.  If a concurrent
         * truncate, we do its job, so it does not have to do anything.
         */
        err = ntfs_resident_attr_value_resize(ctx->mrec, ctx->attr,
                attr_len);
        /* Shrinking cannot fail. */
        BUG_ON(err);
    }
    addr = kmap_atomic(page);
    /* Copy the data from the page to the mft record. */
    memcpy((u8*)ctx->attr +
            le16_to_cpu(ctx->attr->data.resident.value_offset),
            addr, attr_len);
    /* Zero out of bounds area in the page cache page. */
    memset(addr + attr_len, 0, PAGE_CACHE_SIZE - attr_len);
    kunmap_atomic(addr);
    flush_dcache_page(page);
    flush_dcache_mft_record_page(ctx->ntfs_ino);
    /* We are done with the page. */
    end_page_writeback(page);
    /* Finally, mark the mft record dirty, so it gets written back. */
    mark_mft_record_dirty(ctx->ntfs_ino);
    ntfs_attr_put_search_ctx(ctx);
    unmap_mft_record(base_ni);
    return 0;
err_out:
    if (err == -ENOMEM) {
        ntfs_warning(vi->i_sb, "Error allocating memory. Redirtying "
                "page so we try again later.");
        /*
         * Put the page back on mapping->dirty_pages, but leave its
         * buffers' dirty state as-is.
         */
        redirty_page_for_writepage(wbc, page);
        err = 0;
    } else {
        ntfs_error(vi->i_sb, "Resident attribute write failed with "
                "error %i.", err);
        SetPageError(page);
        NVolSetErrors(ni->vol);
    }
    unlock_page(page);
    if (ctx)
        ntfs_attr_put_search_ctx(ctx);
    if (m)
        unmap_mft_record(base_ni);
    return err;
}

#endif  /* NTFS_RW */

const struct address_space_operations ntfs_aops = {
    .readpage   = ntfs_readpage,    /* Fill page with data. */
#ifdef NTFS_RW
    .writepage  = ntfs_writepage,   /* Write dirty page to disk. */
#endif /* NTFS_RW */
    .migratepage    = buffer_migrate_page,  /* Move a page cache page from
                           one physical page to an
                           other. */
    .error_remove_page = generic_error_remove_page,
};

const struct address_space_operations ntfs_mst_aops = {
    .readpage   = ntfs_readpage,    /* Fill page with data. */
#ifdef NTFS_RW
    .writepage  = ntfs_writepage,   /* Write dirty page to disk. */
    .set_page_dirty = __set_page_dirty_nobuffers,   /* Set the page dirty
                           without touching the buffers
                           belonging to the page. */
#endif /* NTFS_RW */
    .migratepage    = buffer_migrate_page,  /* Move a page cache page from
                           one physical page to an
                           other. */
    .error_remove_page = generic_error_remove_page,
};

#ifdef NTFS_RW

void mark_ntfs_record_dirty(struct page *page, const unsigned int ofs) {
    struct address_space *mapping = page->mapping;
    ntfs_inode *ni = NTFS_I(mapping->host);
    struct buffer_head *bh, *head, *buffers_to_free = NULL;
    unsigned int end, bh_size, bh_ofs;

    BUG_ON(!PageUptodate(page));
    end = ofs + ni->itype.index.block_size;
    bh_size = VFS_I(ni)->i_sb->s_blocksize;
    spin_lock(&mapping->private_lock);
    if (unlikely(!page_has_buffers(page))) {
        spin_unlock(&mapping->private_lock);
        bh = head = alloc_page_buffers(page, bh_size, 1);
        spin_lock(&mapping->private_lock);
        if (likely(!page_has_buffers(page))) {
            struct buffer_head *tail;

            do {
                set_buffer_uptodate(bh);
                tail = bh;
                bh = bh->b_this_page;
            } while (bh);
            tail->b_this_page = head;
            attach_page_buffers(page, head);
        } else
            buffers_to_free = bh;
    }
    bh = head = page_buffers(page);
    BUG_ON(!bh);
    do {
        bh_ofs = bh_offset(bh);
        if (bh_ofs + bh_size <= ofs)
            continue;
        if (unlikely(bh_ofs >= end))
            break;
        set_buffer_dirty(bh);
    } while ((bh = bh->b_this_page) != head);
    spin_unlock(&mapping->private_lock);
    __set_page_dirty_nobuffers(page);
    if (unlikely(buffers_to_free)) {
        do {
            bh = buffers_to_free->b_this_page;
            free_buffer_head(buffers_to_free);
            buffers_to_free = bh;
        } while (buffers_to_free);
    }
}

#endif /* NTFS_RW */