fallocate support in ext4
authorAmit Arora <aarora@in.ibm.com>
Wed, 18 Jul 2007 01:42:41 +0000 (21:42 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Wed, 18 Jul 2007 01:42:41 +0000 (21:42 -0400)
This patch implements ->fallocate() inode operation in ext4. With this
patch users of ext4 file systems will be able to use fallocate() system
call for persistent preallocation. Current implementation only supports
preallocation for regular files (directories not supported as of date)
with extent maps. This patch does not support block-mapped files currently.
Only FALLOC_ALLOCATE and FALLOC_RESV_SPACE modes are being supported as of
now.

Signed-off-by: Amit Arora <aarora@in.ibm.com>
fs/ext4/extents.c
fs/ext4/file.c
include/linux/ext4_fs.h
include/linux/ext4_fs_extents.h

index b9ce24129070bff5cee547a333139d9a4645d2c7..ba25832a756c4cd5729eeacd284ecbe9c25defe3 100644 (file)
@@ -39,6 +39,7 @@
 #include <linux/quotaops.h>
 #include <linux/string.h>
 #include <linux/slab.h>
+#include <linux/falloc.h>
 #include <linux/ext4_fs_extents.h>
 #include <asm/uaccess.h>
 
@@ -282,7 +283,7 @@ static void ext4_ext_show_path(struct inode *inode, struct ext4_ext_path *path)
                } else if (path->p_ext) {
                        ext_debug("  %d:%d:%llu ",
                                  le32_to_cpu(path->p_ext->ee_block),
-                                 le16_to_cpu(path->p_ext->ee_len),
+                                 ext4_ext_get_actual_len(path->p_ext),
                                  ext_pblock(path->p_ext));
                } else
                        ext_debug("  []");
@@ -305,7 +306,7 @@ static void ext4_ext_show_leaf(struct inode *inode, struct ext4_ext_path *path)
 
        for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ex++) {
                ext_debug("%d:%d:%llu ", le32_to_cpu(ex->ee_block),
-                         le16_to_cpu(ex->ee_len), ext_pblock(ex));
+                         ext4_ext_get_actual_len(ex), ext_pblock(ex));
        }
        ext_debug("\n");
 }
@@ -425,7 +426,7 @@ ext4_ext_binsearch(struct inode *inode, struct ext4_ext_path *path, int block)
        ext_debug("  -> %d:%llu:%d ",
                        le32_to_cpu(path->p_ext->ee_block),
                        ext_pblock(path->p_ext),
-                       le16_to_cpu(path->p_ext->ee_len));
+                       ext4_ext_get_actual_len(path->p_ext));
 
 #ifdef CHECK_BINSEARCH
        {
@@ -686,7 +687,7 @@ static int ext4_ext_split(handle_t *handle, struct inode *inode,
                ext_debug("move %d:%llu:%d in new leaf %llu\n",
                                le32_to_cpu(path[depth].p_ext->ee_block),
                                ext_pblock(path[depth].p_ext),
-                               le16_to_cpu(path[depth].p_ext->ee_len),
+                               ext4_ext_get_actual_len(path[depth].p_ext),
                                newblock);
                /*memmove(ex++, path[depth].p_ext++,
                                sizeof(struct ext4_extent));
@@ -1106,7 +1107,19 @@ static int
 ext4_can_extents_be_merged(struct inode *inode, struct ext4_extent *ex1,
                                struct ext4_extent *ex2)
 {
-       if (le32_to_cpu(ex1->ee_block) + le16_to_cpu(ex1->ee_len) !=
+       unsigned short ext1_ee_len, ext2_ee_len;
+
+       /*
+        * Make sure that either both extents are uninitialized, or
+        * both are _not_.
+        */
+       if (ext4_ext_is_uninitialized(ex1) ^ ext4_ext_is_uninitialized(ex2))
+               return 0;
+
+       ext1_ee_len = ext4_ext_get_actual_len(ex1);
+       ext2_ee_len = ext4_ext_get_actual_len(ex2);
+
+       if (le32_to_cpu(ex1->ee_block) + ext1_ee_len !=
                        le32_to_cpu(ex2->ee_block))
                return 0;
 
@@ -1115,14 +1128,14 @@ ext4_can_extents_be_merged(struct inode *inode, struct ext4_extent *ex1,
         * as an RO_COMPAT feature, refuse to merge to extents if
         * this can result in the top bit of ee_len being set.
         */
-       if (le16_to_cpu(ex1->ee_len) + le16_to_cpu(ex2->ee_len) > EXT_MAX_LEN)
+       if (ext1_ee_len + ext2_ee_len > EXT_MAX_LEN)
                return 0;
 #ifdef AGGRESSIVE_TEST
        if (le16_to_cpu(ex1->ee_len) >= 4)
                return 0;
 #endif
 
-       if (ext_pblock(ex1) + le16_to_cpu(ex1->ee_len) == ext_pblock(ex2))
+       if (ext_pblock(ex1) + ext1_ee_len == ext_pblock(ex2))
                return 1;
        return 0;
 }
@@ -1144,7 +1157,7 @@ unsigned int ext4_ext_check_overlap(struct inode *inode,
        unsigned int ret = 0;
 
        b1 = le32_to_cpu(newext->ee_block);
-       len1 = le16_to_cpu(newext->ee_len);
+       len1 = ext4_ext_get_actual_len(newext);
        depth = ext_depth(inode);
        if (!path[depth].p_ext)
                goto out;
@@ -1191,8 +1204,9 @@ int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
        struct ext4_extent *nearex; /* nearest extent */
        struct ext4_ext_path *npath = NULL;
        int depth, len, err, next;
+       unsigned uninitialized = 0;
 
-       BUG_ON(newext->ee_len == 0);
+       BUG_ON(ext4_ext_get_actual_len(newext) == 0);
        depth = ext_depth(inode);
        ex = path[depth].p_ext;
        BUG_ON(path[depth].p_hdr == NULL);
@@ -1200,14 +1214,24 @@ int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
        /* try to insert block into found extent and return */
        if (ex && ext4_can_extents_be_merged(inode, ex, newext)) {
                ext_debug("append %d block to %d:%d (from %llu)\n",
-                               le16_to_cpu(newext->ee_len),
+                               ext4_ext_get_actual_len(newext),
                                le32_to_cpu(ex->ee_block),
-                               le16_to_cpu(ex->ee_len), ext_pblock(ex));
+                               ext4_ext_get_actual_len(ex), ext_pblock(ex));
                err = ext4_ext_get_access(handle, inode, path + depth);
                if (err)
                        return err;
-               ex->ee_len = cpu_to_le16(le16_to_cpu(ex->ee_len)
-                                        + le16_to_cpu(newext->ee_len));
+
+               /*
+                * ext4_can_extents_be_merged should have checked that either
+                * both extents are uninitialized, or both aren't. Thus we
+                * need to check only one of them here.
+                */
+               if (ext4_ext_is_uninitialized(ex))
+                       uninitialized = 1;
+               ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
+                                       + ext4_ext_get_actual_len(newext));
+               if (uninitialized)
+                       ext4_ext_mark_uninitialized(ex);
                eh = path[depth].p_hdr;
                nearex = ex;
                goto merge;
@@ -1263,7 +1287,7 @@ has_space:
                ext_debug("first extent in the leaf: %d:%llu:%d\n",
                                le32_to_cpu(newext->ee_block),
                                ext_pblock(newext),
-                               le16_to_cpu(newext->ee_len));
+                               ext4_ext_get_actual_len(newext));
                path[depth].p_ext = EXT_FIRST_EXTENT(eh);
        } else if (le32_to_cpu(newext->ee_block)
                           > le32_to_cpu(nearex->ee_block)) {
@@ -1276,7 +1300,7 @@ has_space:
                                        "move %d from 0x%p to 0x%p\n",
                                        le32_to_cpu(newext->ee_block),
                                        ext_pblock(newext),
-                                       le16_to_cpu(newext->ee_len),
+                                       ext4_ext_get_actual_len(newext),
                                        nearex, len, nearex + 1, nearex + 2);
                        memmove(nearex + 2, nearex + 1, len);
                }
@@ -1289,7 +1313,7 @@ has_space:
                                "move %d from 0x%p to 0x%p\n",
                                le32_to_cpu(newext->ee_block),
                                ext_pblock(newext),
-                               le16_to_cpu(newext->ee_len),
+                               ext4_ext_get_actual_len(newext),
                                nearex, len, nearex + 1, nearex + 2);
                memmove(nearex + 1, nearex, len);
                path[depth].p_ext = nearex;
@@ -1308,8 +1332,13 @@ merge:
                if (!ext4_can_extents_be_merged(inode, nearex, nearex + 1))
                        break;
                /* merge with next extent! */
-               nearex->ee_len = cpu_to_le16(le16_to_cpu(nearex->ee_len)
-                                            + le16_to_cpu(nearex[1].ee_len));
+               if (ext4_ext_is_uninitialized(nearex))
+                       uninitialized = 1;
+               nearex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(nearex)
+                                       + ext4_ext_get_actual_len(nearex + 1));
+               if (uninitialized)
+                       ext4_ext_mark_uninitialized(nearex);
+
                if (nearex + 1 < EXT_LAST_EXTENT(eh)) {
                        len = (EXT_LAST_EXTENT(eh) - nearex - 1)
                                        * sizeof(struct ext4_extent);
@@ -1379,8 +1408,8 @@ int ext4_ext_walk_space(struct inode *inode, unsigned long block,
                        end = le32_to_cpu(ex->ee_block);
                        if (block + num < end)
                                end = block + num;
-               } else if (block >=
-                            le32_to_cpu(ex->ee_block) + le16_to_cpu(ex->ee_len)) {
+               } else if (block >= le32_to_cpu(ex->ee_block)
+                                       + ext4_ext_get_actual_len(ex)) {
                        /* need to allocate space after found extent */
                        start = block;
                        end = block + num;
@@ -1392,7 +1421,8 @@ int ext4_ext_walk_space(struct inode *inode, unsigned long block,
                         * by found extent
                         */
                        start = block;
-                       end = le32_to_cpu(ex->ee_block) + le16_to_cpu(ex->ee_len);
+                       end = le32_to_cpu(ex->ee_block)
+                               + ext4_ext_get_actual_len(ex);
                        if (block + num < end)
                                end = block + num;
                        exists = 1;
@@ -1408,7 +1438,7 @@ int ext4_ext_walk_space(struct inode *inode, unsigned long block,
                        cbex.ec_type = EXT4_EXT_CACHE_GAP;
                } else {
                        cbex.ec_block = le32_to_cpu(ex->ee_block);
-                       cbex.ec_len = le16_to_cpu(ex->ee_len);
+                       cbex.ec_len = ext4_ext_get_actual_len(ex);
                        cbex.ec_start = ext_pblock(ex);
                        cbex.ec_type = EXT4_EXT_CACHE_EXTENT;
                }
@@ -1481,15 +1511,15 @@ ext4_ext_put_gap_in_cache(struct inode *inode, struct ext4_ext_path *path,
                ext_debug("cache gap(before): %lu [%lu:%lu]",
                                (unsigned long) block,
                                (unsigned long) le32_to_cpu(ex->ee_block),
-                               (unsigned long) le16_to_cpu(ex->ee_len));
+                               (unsigned long) ext4_ext_get_actual_len(ex));
        } else if (block >= le32_to_cpu(ex->ee_block)
-                           + le16_to_cpu(ex->ee_len)) {
+                       + ext4_ext_get_actual_len(ex)) {
                lblock = le32_to_cpu(ex->ee_block)
-                        + le16_to_cpu(ex->ee_len);
+                       + ext4_ext_get_actual_len(ex);
                len = ext4_ext_next_allocated_block(path);
                ext_debug("cache gap(after): [%lu:%lu] %lu",
                                (unsigned long) le32_to_cpu(ex->ee_block),
-                               (unsigned long) le16_to_cpu(ex->ee_len),
+                               (unsigned long) ext4_ext_get_actual_len(ex),
                                (unsigned long) block);
                BUG_ON(len == lblock);
                len = len - lblock;
@@ -1619,12 +1649,12 @@ static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
                                unsigned long from, unsigned long to)
 {
        struct buffer_head *bh;
+       unsigned short ee_len =  ext4_ext_get_actual_len(ex);
        int i;
 
 #ifdef EXTENTS_STATS
        {
                struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
-               unsigned short ee_len =  le16_to_cpu(ex->ee_len);
                spin_lock(&sbi->s_ext_stats_lock);
                sbi->s_ext_blocks += ee_len;
                sbi->s_ext_extents++;
@@ -1638,12 +1668,12 @@ static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
        }
 #endif
        if (from >= le32_to_cpu(ex->ee_block)
-           && to == le32_to_cpu(ex->ee_block) + le16_to_cpu(ex->ee_len) - 1) {
+           && to == le32_to_cpu(ex->ee_block) + ee_len - 1) {
                /* tail removal */
                unsigned long num;
                ext4_fsblk_t start;
-               num = le32_to_cpu(ex->ee_block) + le16_to_cpu(ex->ee_len) - from;
-               start = ext_pblock(ex) + le16_to_cpu(ex->ee_len) - num;
+               num = le32_to_cpu(ex->ee_block) + ee_len - from;
+               start = ext_pblock(ex) + ee_len - num;
                ext_debug("free last %lu blocks starting %llu\n", num, start);
                for (i = 0; i < num; i++) {
                        bh = sb_find_get_block(inode->i_sb, start + i);
@@ -1651,12 +1681,12 @@ static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
                }
                ext4_free_blocks(handle, inode, start, num);
        } else if (from == le32_to_cpu(ex->ee_block)
-                  && to <= le32_to_cpu(ex->ee_block) + le16_to_cpu(ex->ee_len) - 1) {
+                  && to <= le32_to_cpu(ex->ee_block) + ee_len - 1) {
                printk("strange request: removal %lu-%lu from %u:%u\n",
-                      from, to, le32_to_cpu(ex->ee_block), le16_to_cpu(ex->ee_len));
+                       from, to, le32_to_cpu(ex->ee_block), ee_len);
        } else {
                printk("strange request: removal(2) %lu-%lu from %u:%u\n",
-                      from, to, le32_to_cpu(ex->ee_block), le16_to_cpu(ex->ee_len));
+                       from, to, le32_to_cpu(ex->ee_block), ee_len);
        }
        return 0;
 }
@@ -1671,6 +1701,7 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
        unsigned a, b, block, num;
        unsigned long ex_ee_block;
        unsigned short ex_ee_len;
+       unsigned uninitialized = 0;
        struct ext4_extent *ex;
 
        ext_debug("truncate since %lu in leaf\n", start);
@@ -1685,7 +1716,9 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
        ex = EXT_LAST_EXTENT(eh);
 
        ex_ee_block = le32_to_cpu(ex->ee_block);
-       ex_ee_len = le16_to_cpu(ex->ee_len);
+       if (ext4_ext_is_uninitialized(ex))
+               uninitialized = 1;
+       ex_ee_len = ext4_ext_get_actual_len(ex);
 
        while (ex >= EXT_FIRST_EXTENT(eh) &&
                        ex_ee_block + ex_ee_len > start) {
@@ -1753,6 +1786,8 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
 
                ex->ee_block = cpu_to_le32(block);
                ex->ee_len = cpu_to_le16(num);
+               if (uninitialized)
+                       ext4_ext_mark_uninitialized(ex);
 
                err = ext4_ext_dirty(handle, inode, path + depth);
                if (err)
@@ -1762,7 +1797,7 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
                                ext_pblock(ex));
                ex--;
                ex_ee_block = le32_to_cpu(ex->ee_block);
-               ex_ee_len = le16_to_cpu(ex->ee_len);
+               ex_ee_len = ext4_ext_get_actual_len(ex);
        }
 
        if (correct_index && eh->eh_entries)
@@ -2038,7 +2073,7 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
        if (ex) {
                unsigned long ee_block = le32_to_cpu(ex->ee_block);
                ext4_fsblk_t ee_start = ext_pblock(ex);
-               unsigned short ee_len  = le16_to_cpu(ex->ee_len);
+               unsigned short ee_len;
 
                /*
                 * Allow future support for preallocated extents to be added
@@ -2046,8 +2081,9 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
                 * Uninitialized extents are treated as holes, except that
                 * we avoid (fail) allocating new blocks during a write.
                 */
-               if (ee_len > EXT_MAX_LEN)
+               if (le16_to_cpu(ex->ee_len) > EXT_MAX_LEN)
                        goto out2;
+               ee_len = ext4_ext_get_actual_len(ex);
                /* if found extent covers block, simply return it */
                if (iblock >= ee_block && iblock < ee_block + ee_len) {
                        newblock = iblock - ee_block + ee_start;
@@ -2055,8 +2091,11 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
                        allocated = ee_len - (iblock - ee_block);
                        ext_debug("%d fit into %lu:%d -> %llu\n", (int) iblock,
                                        ee_block, ee_len, newblock);
-                       ext4_ext_put_in_cache(inode, ee_block, ee_len,
-                                               ee_start, EXT4_EXT_CACHE_EXTENT);
+                       /* Do not put uninitialized extent in the cache */
+                       if (!ext4_ext_is_uninitialized(ex))
+                               ext4_ext_put_in_cache(inode, ee_block,
+                                                       ee_len, ee_start,
+                                                       EXT4_EXT_CACHE_EXTENT);
                        goto out;
                }
        }
@@ -2098,6 +2137,8 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
        /* try to insert new extent into found leaf and return */
        ext4_ext_store_pblock(&newex, newblock);
        newex.ee_len = cpu_to_le16(allocated);
+       if (create == EXT4_CREATE_UNINITIALIZED_EXT)  /* Mark uninitialized */
+               ext4_ext_mark_uninitialized(&newex);
        err = ext4_ext_insert_extent(handle, inode, path, &newex);
        if (err) {
                /* free data blocks we just allocated */
@@ -2113,8 +2154,10 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
        newblock = ext_pblock(&newex);
        __set_bit(BH_New, &bh_result->b_state);
 
-       ext4_ext_put_in_cache(inode, iblock, allocated, newblock,
-                               EXT4_EXT_CACHE_EXTENT);
+       /* Cache only when it is _not_ an uninitialized extent */
+       if (create != EXT4_CREATE_UNINITIALIZED_EXT)
+               ext4_ext_put_in_cache(inode, iblock, allocated, newblock,
+                                               EXT4_EXT_CACHE_EXTENT);
 out:
        if (allocated > max_blocks)
                allocated = max_blocks;
@@ -2217,3 +2260,127 @@ int ext4_ext_writepage_trans_blocks(struct inode *inode, int num)
 
        return needed;
 }
+
+/*
+ * preallocate space for a file. This implements ext4's fallocate inode
+ * operation, which gets called from sys_fallocate system call.
+ * For block-mapped files, posix_fallocate should fall back to the method
+ * of writing zeroes to the required new blocks (the same behavior which is
+ * expected for file systems which do not support fallocate() system call).
+ */
+long ext4_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len)
+{
+       handle_t *handle;
+       ext4_fsblk_t block, max_blocks;
+       ext4_fsblk_t nblocks = 0;
+       int ret = 0;
+       int ret2 = 0;
+       int retries = 0;
+       struct buffer_head map_bh;
+       unsigned int credits, blkbits = inode->i_blkbits;
+
+       /*
+        * currently supporting (pre)allocate mode for extent-based
+        * files _only_
+        */
+       if (!(EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL))
+               return -EOPNOTSUPP;
+
+       /* preallocation to directories is currently not supported */
+       if (S_ISDIR(inode->i_mode))
+               return -ENODEV;
+
+       block = offset >> blkbits;
+       max_blocks = (EXT4_BLOCK_ALIGN(len + offset, blkbits) >> blkbits)
+                       - block;
+
+       /*
+        * credits to insert 1 extent into extent tree + buffers to be able to
+        * modify 1 super block, 1 block bitmap and 1 group descriptor.
+        */
+       credits = EXT4_DATA_TRANS_BLOCKS(inode->i_sb) + 3;
+retry:
+       while (ret >= 0 && ret < max_blocks) {
+               block = block + ret;
+               max_blocks = max_blocks - ret;
+               handle = ext4_journal_start(inode, credits);
+               if (IS_ERR(handle)) {
+                       ret = PTR_ERR(handle);
+                       break;
+               }
+
+               ret = ext4_ext_get_blocks(handle, inode, block,
+                                         max_blocks, &map_bh,
+                                         EXT4_CREATE_UNINITIALIZED_EXT, 0);
+               WARN_ON(!ret);
+               if (!ret) {
+                       ext4_error(inode->i_sb, "ext4_fallocate",
+                                  "ext4_ext_get_blocks returned 0! inode#%lu"
+                                  ", block=%llu, max_blocks=%llu",
+                                  inode->i_ino, block, max_blocks);
+                       ret = -EIO;
+                       ext4_mark_inode_dirty(handle, inode);
+                       ret2 = ext4_journal_stop(handle);
+                       break;
+               }
+               if (ret > 0) {
+                       /* check wrap through sign-bit/zero here */
+                       if ((block + ret) < 0 || (block + ret) < block) {
+                               ret = -EIO;
+                               ext4_mark_inode_dirty(handle, inode);
+                               ret2 = ext4_journal_stop(handle);
+                               break;
+                       }
+                       if (buffer_new(&map_bh) && ((block + ret) >
+                           (EXT4_BLOCK_ALIGN(i_size_read(inode), blkbits)
+                           >> blkbits)))
+                                       nblocks = nblocks + ret;
+               }
+
+               /* Update ctime if new blocks get allocated */
+               if (nblocks) {
+                       struct timespec now;
+
+                       now = current_fs_time(inode->i_sb);
+                       if (!timespec_equal(&inode->i_ctime, &now))
+                               inode->i_ctime = now;
+               }
+
+               ext4_mark_inode_dirty(handle, inode);
+               ret2 = ext4_journal_stop(handle);
+               if (ret2)
+                       break;
+       }
+
+       if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
+               goto retry;
+
+       /*
+        * Time to update the file size.
+        * Update only when preallocation was requested beyond the file size.
+        */
+       if (!(mode & FALLOC_FL_KEEP_SIZE) &&
+           (offset + len) > i_size_read(inode)) {
+               if (ret > 0) {
+                       /*
+                        * if no error, we assume preallocation succeeded
+                        * completely
+                        */
+                       mutex_lock(&inode->i_mutex);
+                       i_size_write(inode, offset + len);
+                       EXT4_I(inode)->i_disksize = i_size_read(inode);
+                       mutex_unlock(&inode->i_mutex);
+               } else if (ret < 0 && nblocks) {
+                       /* Handle partial allocation scenario */
+                       loff_t newsize;
+
+                       mutex_lock(&inode->i_mutex);
+                       newsize  = (nblocks << blkbits) + i_size_read(inode);
+                       i_size_write(inode, EXT4_BLOCK_ALIGN(newsize, blkbits));
+                       EXT4_I(inode)->i_disksize = i_size_read(inode);
+                       mutex_unlock(&inode->i_mutex);
+               }
+       }
+
+       return ret > 0 ? ret2 : ret;
+}
index d4c8186aed646f6a4439b17d0b0cda9655aadf55..1a81cd66d63b2b2371e4c35025f004657ca66adf 100644 (file)
@@ -134,5 +134,6 @@ const struct inode_operations ext4_file_inode_operations = {
        .removexattr    = generic_removexattr,
 #endif
        .permission     = ext4_permission,
+       .fallocate      = ext4_fallocate,
 };
 
index de1f9f78625a3279bf5ddb580d57cc1003eef152..87c2d7a05b01af12cd2da161d4db81f387e479c5 100644 (file)
                                 EXT4_GOOD_OLD_FIRST_INO : \
                                 (s)->s_first_ino)
 #endif
+#define EXT4_BLOCK_ALIGN(size, blkbits)                ALIGN((size), (1 << (blkbits)))
 
 /*
  * Macro-instructions used to manage fragments
@@ -225,6 +226,11 @@ struct ext4_new_group_data {
        __u32 free_blocks_count;
 };
 
+/*
+ * Following is used by preallocation code to tell get_blocks() that we
+ * want uninitialzed extents.
+ */
+#define EXT4_CREATE_UNINITIALIZED_EXT          2
 
 /*
  * ioctl commands
@@ -983,6 +989,8 @@ extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
 extern void ext4_ext_truncate(struct inode *, struct page *);
 extern void ext4_ext_init(struct super_block *);
 extern void ext4_ext_release(struct super_block *);
+extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
+                         loff_t len);
 static inline int
 ext4_get_blocks_wrap(handle_t *handle, struct inode *inode, sector_t block,
                        unsigned long max_blocks, struct buffer_head *bh,
index acfe59740b032d1a5adcfebaa69ca491c048cafa..e3d5afc6f23e00e0e65feb5310a0a8e79aa9f8ad 100644 (file)
@@ -188,6 +188,21 @@ ext4_ext_invalidate_cache(struct inode *inode)
        EXT4_I(inode)->i_cached_extent.ec_type = EXT4_EXT_CACHE_NO;
 }
 
+static inline void ext4_ext_mark_uninitialized(struct ext4_extent *ext)
+{
+       ext->ee_len |= cpu_to_le16(0x8000);
+}
+
+static inline int ext4_ext_is_uninitialized(struct ext4_extent *ext)
+{
+       return (int)(le16_to_cpu((ext)->ee_len) & 0x8000);
+}
+
+static inline int ext4_ext_get_actual_len(struct ext4_extent *ext)
+{
+       return (int)(le16_to_cpu((ext)->ee_len) & 0x7FFF);
+}
+
 extern int ext4_extent_tree_init(handle_t *, struct inode *);
 extern int ext4_ext_calc_credits_for_insert(struct inode *, struct ext4_ext_path *);
 extern unsigned int ext4_ext_check_overlap(struct inode *, struct ext4_extent *, struct ext4_ext_path *);