Btrfs: add hole punching

This patch adds hole punching via fallocate.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>

1 files changed, 328 insertions, 4 deletions

diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 58598c249951..57026a6e9c94 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -39,6 +39,7 @@
 #include "tree-log.h"
 #include "locking.h"
 #include "compat.h"
+#include "volumes.h"
 
 /*
  * when auto defrag is enabled we
@@ -584,7 +585,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
                          struct btrfs_root *root, struct inode *inode,
                          struct btrfs_path *path, u64 start, u64 end,
-                         int drop_cache)
+                         u64 *drop_end, int drop_cache)
 {
         struct extent_buffer *leaf;
         struct btrfs_file_extent_item *fi;
@@ -822,6 +823,8 @@ next_slot:
                         btrfs_abort_transaction(trans, root, ret);
         }
 
+        if (drop_end)
+                *drop_end = min(end, extent_end);
         btrfs_release_path(path);
         return ret;
 }
@@ -836,7 +839,7 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans,
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
-        ret = __btrfs_drop_extents(trans, root, inode, path, start, end,
+        ret = __btrfs_drop_extents(trans, root, inode, path, start, end, NULL,
                                    drop_cache);
         btrfs_free_path(path);
         return ret;
@@ -1645,6 +1648,324 @@ static int btrfs_file_mmap(struct file	*filp, struct vm_area_struct *vma)
         return 0;
 }
 
+static int hole_mergeable(struct inode *inode, struct extent_buffer *leaf,
+                          int slot, u64 start, u64 end)
+{
+        struct btrfs_file_extent_item *fi;
+        struct btrfs_key key;
+
+        if (slot < 0 || slot >= btrfs_header_nritems(leaf))
+                return 0;
+
+        btrfs_item_key_to_cpu(leaf, &key, slot);
+        if (key.objectid != btrfs_ino(inode) ||
+            key.type != BTRFS_EXTENT_DATA_KEY)
+                return 0;
+
+        fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
+
+        if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG)
+                return 0;
+
+        if (btrfs_file_extent_disk_bytenr(leaf, fi))
+                return 0;
+
+        if (key.offset == end)
+                return 1;
+        if (key.offset + btrfs_file_extent_num_bytes(leaf, fi) == start)
+                return 1;
+        return 0;
+}
+
+static int fill_holes(struct btrfs_trans_handle *trans, struct inode *inode,
+                      struct btrfs_path *path, u64 offset, u64 end)
+{
+        struct btrfs_root *root = BTRFS_I(inode)->root;
+        struct extent_buffer *leaf;
+        struct btrfs_file_extent_item *fi;
+        struct extent_map *hole_em;
+        struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+        struct btrfs_key key;
+        int ret;
+
+        key.objectid = btrfs_ino(inode);
+        key.type = BTRFS_EXTENT_DATA_KEY;
+        key.offset = offset;
+
+
+        ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
+        if (ret < 0)
+                return ret;
+        BUG_ON(!ret);
+
+        leaf = path->nodes[0];
+        if (hole_mergeable(inode, leaf, path->slots[0]-1, offset, end)) {
+                u64 num_bytes;
+
+                path->slots[0]--;
+                fi = btrfs_item_ptr(leaf, path->slots[0],
+                                    struct btrfs_file_extent_item);
+                num_bytes = btrfs_file_extent_num_bytes(leaf, fi) +
+                        end - offset;
+                btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
+                btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
+                btrfs_set_file_extent_offset(leaf, fi, 0);
+                btrfs_mark_buffer_dirty(leaf);
+                goto out;
+        }
+
+        if (hole_mergeable(inode, leaf, path->slots[0]+1, offset, end)) {
+                u64 num_bytes;
+
+                path->slots[0]++;
+                key.offset = offset;
+                btrfs_set_item_key_safe(trans, root, path, &key);
+                fi = btrfs_item_ptr(leaf, path->slots[0],
+                                    struct btrfs_file_extent_item);
+                num_bytes = btrfs_file_extent_num_bytes(leaf, fi) + end -
+                        offset;
+                btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
+                btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
+                btrfs_set_file_extent_offset(leaf, fi, 0);
+                btrfs_mark_buffer_dirty(leaf);
+                goto out;
+        }
+        btrfs_release_path(path);
+
+        ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode), offset,
+                                       0, 0, end - offset, 0, end - offset,
+                                       0, 0, 0);
+        if (ret)
+                return ret;
+
+out:
+        btrfs_release_path(path);
+
+        hole_em = alloc_extent_map();
+        if (!hole_em) {
+                btrfs_drop_extent_cache(inode, offset, end - 1, 0);
+                set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+                        &BTRFS_I(inode)->runtime_flags);
+        } else {
+                hole_em->start = offset;
+                hole_em->len = end - offset;
+                hole_em->orig_start = offset;
+
+                hole_em->block_start = EXTENT_MAP_HOLE;
+                hole_em->block_len = 0;
+                hole_em->bdev = root->fs_info->fs_devices->latest_bdev;
+                hole_em->compress_type = BTRFS_COMPRESS_NONE;
+                hole_em->generation = trans->transid;
+
+                do {
+                        btrfs_drop_extent_cache(inode, offset, end - 1, 0);
+                        write_lock(&em_tree->lock);
+                        ret = add_extent_mapping(em_tree, hole_em);
+                        if (!ret)
+                                list_move(&hole_em->list,
+                                          &em_tree->modified_extents);
+                        write_unlock(&em_tree->lock);
+                } while (ret == -EEXIST);
+                free_extent_map(hole_em);
+                if (ret)
+                        set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
+                                &BTRFS_I(inode)->runtime_flags);
+        }
+
+        return 0;
+}
+
+static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
+{
+        struct btrfs_root *root = BTRFS_I(inode)->root;
+        struct extent_state *cached_state = NULL;
+        struct btrfs_path *path;
+        struct btrfs_block_rsv *rsv;
+        struct btrfs_trans_handle *trans;
+        u64 mask = BTRFS_I(inode)->root->sectorsize - 1;
+        u64 lockstart = (offset + mask) & ~mask;
+        u64 lockend = ((offset + len) & ~mask) - 1;
+        u64 cur_offset = lockstart;
+        u64 min_size = btrfs_calc_trunc_metadata_size(root, 1);
+        u64 drop_end;
+        unsigned long nr;
+        int ret = 0;
+        int err = 0;
+        bool same_page = (offset >> PAGE_CACHE_SHIFT) ==
+                ((offset + len) >> PAGE_CACHE_SHIFT);
+
+        btrfs_wait_ordered_range(inode, offset, len);
+
+        mutex_lock(&inode->i_mutex);
+        if (offset >= inode->i_size) {
+                mutex_unlock(&inode->i_mutex);
+                return 0;
+        }
+
+        /*
+         * Only do this if we are in the same page and we aren't doing the
+         * entire page.
+         */
+        if (same_page && len < PAGE_CACHE_SIZE) {
+                ret = btrfs_truncate_page(inode, offset, len, 0);
+                mutex_unlock(&inode->i_mutex);
+                return ret;
+        }
+
+        /* zero back part of the first page */
+        ret = btrfs_truncate_page(inode, offset, 0, 0);
+        if (ret) {
+                mutex_unlock(&inode->i_mutex);
+                return ret;
+        }
+
+        /* zero the front end of the last page */
+        ret = btrfs_truncate_page(inode, offset + len, 0, 1);
+        if (ret) {
+                mutex_unlock(&inode->i_mutex);
+                return ret;
+        }
+
+        if (lockend < lockstart) {
+                mutex_unlock(&inode->i_mutex);
+                return 0;
+        }
+
+        while (1) {
+                struct btrfs_ordered_extent *ordered;
+
+                truncate_pagecache_range(inode, lockstart, lockend);
+
+                lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+                                 0, &cached_state);
+                ordered = btrfs_lookup_first_ordered_extent(inode, lockend);
+
+                /*
+                 * We need to make sure we have no ordered extents in this range
+                 * and nobody raced in and read a page in this range, if we did
+                 * we need to try again.
+                 */
+                if ((!ordered ||
+                    (ordered->file_offset + ordered->len < lockstart ||
+                     ordered->file_offset > lockend)) &&
+                     !test_range_bit(&BTRFS_I(inode)->io_tree, lockstart,
+                                     lockend, EXTENT_UPTODATE, 0,
+                                     cached_state)) {
+                        if (ordered)
+                                btrfs_put_ordered_extent(ordered);
+                        break;
+                }
+                if (ordered)
+                        btrfs_put_ordered_extent(ordered);
+                unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart,
+                                     lockend, &cached_state, GFP_NOFS);
+                btrfs_wait_ordered_range(inode, lockstart,
+                                         lockend - lockstart + 1);
+        }
+
+        path = btrfs_alloc_path();
+        if (!path) {
+                ret = -ENOMEM;
+                goto out;
+        }
+
+        rsv = btrfs_alloc_block_rsv(root);
+        if (!rsv) {
+                ret = -ENOMEM;
+                goto out_free;
+        }
+        rsv->size = btrfs_calc_trunc_metadata_size(root, 1);
+        rsv->failfast = 1;
+
+        /*
+         * 1 - update the inode
+         * 1 - removing the extents in the range
+         * 1 - adding the hole extent
+         */
+        trans = btrfs_start_transaction(root, 3);
+        if (IS_ERR(trans)) {
+                err = PTR_ERR(trans);
+                goto out_free;
+        }
+
+        ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, rsv,
+                                      min_size);
+        BUG_ON(ret);
+        trans->block_rsv = rsv;
+
+        while (cur_offset < lockend) {
+                ret = __btrfs_drop_extents(trans, root, inode, path,
+                                           cur_offset, lockend + 1,
+                                           &drop_end, 1);
+                if (ret != -ENOSPC)
+                        break;
+
+                trans->block_rsv = &root->fs_info->trans_block_rsv;
+
+                ret = fill_holes(trans, inode, path, cur_offset, drop_end);
+                if (ret) {
+                        err = ret;
+                        break;
+                }
+
+                cur_offset = drop_end;
+
+                ret = btrfs_update_inode(trans, root, inode);
+                if (ret) {
+                        err = ret;
+                        break;
+                }
+
+                nr = trans->blocks_used;
+                btrfs_end_transaction(trans, root);
+                btrfs_btree_balance_dirty(root, nr);
+
+                trans = btrfs_start_transaction(root, 3);
+                if (IS_ERR(trans)) {
+                        ret = PTR_ERR(trans);
+                        trans = NULL;
+                        break;
+                }
+
+                ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv,
+                                              rsv, min_size);
+                BUG_ON(ret);    /* shouldn't happen */
+                trans->block_rsv = rsv;
+        }
+
+        if (ret) {
+                err = ret;
+                goto out_trans;
+        }
+
+        trans->block_rsv = &root->fs_info->trans_block_rsv;
+        ret = fill_holes(trans, inode, path, cur_offset, drop_end);
+        if (ret) {
+                err = ret;
+                goto out_trans;
+        }
+
+out_trans:
+        if (!trans)
+                goto out_free;
+
+        trans->block_rsv = &root->fs_info->trans_block_rsv;
+        ret = btrfs_update_inode(trans, root, inode);
+        nr = trans->blocks_used;
+        btrfs_end_transaction(trans, root);
+        btrfs_btree_balance_dirty(root, nr);
+out_free:
+        btrfs_free_path(path);
+        btrfs_free_block_rsv(root, rsv);
+out:
+        unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+                             &cached_state, GFP_NOFS);
+        mutex_unlock(&inode->i_mutex);
+        if (ret && !err)
+                err = ret;
+        return err;
+}
+
 static long btrfs_fallocate(struct file *file, int mode,
                             loff_t offset, loff_t len)
 {
@@ -1663,10 +1984,13 @@ static long btrfs_fallocate(struct file *file, int mode,
         alloc_start = offset & ~mask;
         alloc_end =  (offset + len + mask) & ~mask;
 
-        /* We only support the FALLOC_FL_KEEP_SIZE mode */
-        if (mode & ~FALLOC_FL_KEEP_SIZE)
+        /* Make sure we aren't being give some crap mode */
+        if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
                 return -EOPNOTSUPP;
 
+        if (mode & FALLOC_FL_PUNCH_HOLE)
+                return btrfs_punch_hole(inode, offset, len);
+
         /*
          * Make sure we have enough space before we do the
          * allocation.