btrfs: rename struct btrfs_clone_extent_info to a more generic name
authorFilipe Manana <fdmanana@suse.com>
Tue, 8 Sep 2020 10:27:22 +0000 (11:27 +0100)
committerDavid Sterba <dsterba@suse.com>
Wed, 7 Oct 2020 10:13:16 +0000 (12:13 +0200)
Now that we can use btrfs_clone_extent_info to convey information for a
new prealloc extent as well, and not just for existing extents that are
being cloned, rename it to btrfs_replace_extent_info, which reflects the
fact that this is now more generic and it is used to replace all existing
extents in a file range with the extent described by the structure.

Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
fs/btrfs/ctree.h
fs/btrfs/file.c
fs/btrfs/inode.c
fs/btrfs/reflink.c

index 6200e43..0e4034d 100644 (file)
@@ -1194,7 +1194,11 @@ struct btrfs_root {
 #endif
 };
 
-struct btrfs_clone_extent_info {
+/*
+ * Structure that conveys information about an extent that is going to replace
+ * all the extents in a file range.
+ */
+struct btrfs_replace_extent_info {
        u64 disk_offset;
        u64 disk_len;
        u64 data_offset;
@@ -3086,7 +3090,7 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans,
                       u64 end, int drop_cache);
 int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
                           const u64 start, const u64 end,
-                          struct btrfs_clone_extent_info *clone_info,
+                          struct btrfs_replace_extent_info *extent_info,
                           struct btrfs_trans_handle **trans_out);
 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
                              struct btrfs_inode *inode, u64 start, u64 end);
index 28794a9..7ac0a20 100644 (file)
@@ -2573,8 +2573,8 @@ static int btrfs_punch_hole_lock_range(struct inode *inode,
 static int btrfs_insert_clone_extent(struct btrfs_trans_handle *trans,
                                     struct inode *inode,
                                     struct btrfs_path *path,
-                                    struct btrfs_clone_extent_info *clone_info,
-                                    const u64 clone_len)
+                                    struct btrfs_replace_extent_info *extent_info,
+                                    const u64 replace_len)
 {
        struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
        struct btrfs_root *root = BTRFS_I(inode)->root;
@@ -2585,67 +2585,67 @@ static int btrfs_insert_clone_extent(struct btrfs_trans_handle *trans,
        struct btrfs_ref ref = { 0 };
        int ret;
 
-       if (clone_len == 0)
+       if (replace_len == 0)
                return 0;
 
-       if (clone_info->disk_offset == 0 &&
+       if (extent_info->disk_offset == 0 &&
            btrfs_fs_incompat(fs_info, NO_HOLES))
                return 0;
 
        key.objectid = btrfs_ino(BTRFS_I(inode));
        key.type = BTRFS_EXTENT_DATA_KEY;
-       key.offset = clone_info->file_offset;
+       key.offset = extent_info->file_offset;
        ret = btrfs_insert_empty_item(trans, root, path, &key,
                                      sizeof(struct btrfs_file_extent_item));
        if (ret)
                return ret;
        leaf = path->nodes[0];
        slot = path->slots[0];
-       write_extent_buffer(leaf, clone_info->extent_buf,
+       write_extent_buffer(leaf, extent_info->extent_buf,
                            btrfs_item_ptr_offset(leaf, slot),
                            sizeof(struct btrfs_file_extent_item));
        extent = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
        ASSERT(btrfs_file_extent_type(leaf, extent) != BTRFS_FILE_EXTENT_INLINE);
-       btrfs_set_file_extent_offset(leaf, extent, clone_info->data_offset);
-       btrfs_set_file_extent_num_bytes(leaf, extent, clone_len);
-       if (clone_info->is_new_extent)
+       btrfs_set_file_extent_offset(leaf, extent, extent_info->data_offset);
+       btrfs_set_file_extent_num_bytes(leaf, extent, replace_len);
+       if (extent_info->is_new_extent)
                btrfs_set_file_extent_generation(leaf, extent, trans->transid);
        btrfs_mark_buffer_dirty(leaf);
        btrfs_release_path(path);
 
        ret = btrfs_inode_set_file_extent_range(BTRFS_I(inode),
-                       clone_info->file_offset, clone_len);
+                       extent_info->file_offset, replace_len);
        if (ret)
                return ret;
 
        /* If it's a hole, nothing more needs to be done. */
-       if (clone_info->disk_offset == 0)
+       if (extent_info->disk_offset == 0)
                return 0;
 
-       inode_add_bytes(inode, clone_len);
+       inode_add_bytes(inode, replace_len);
 
-       if (clone_info->is_new_extent && clone_info->insertions == 0) {
-               key.objectid = clone_info->disk_offset;
+       if (extent_info->is_new_extent && extent_info->insertions == 0) {
+               key.objectid = extent_info->disk_offset;
                key.type = BTRFS_EXTENT_ITEM_KEY;
-               key.offset = clone_info->disk_len;
+               key.offset = extent_info->disk_len;
                ret = btrfs_alloc_reserved_file_extent(trans, root,
                                                       btrfs_ino(BTRFS_I(inode)),
-                                                      clone_info->file_offset,
-                                                      clone_info->qgroup_reserved,
+                                                      extent_info->file_offset,
+                                                      extent_info->qgroup_reserved,
                                                       &key);
        } else {
                u64 ref_offset;
 
                btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF,
-                                      clone_info->disk_offset,
-                                      clone_info->disk_len, 0);
-               ref_offset = clone_info->file_offset - clone_info->data_offset;
+                                      extent_info->disk_offset,
+                                      extent_info->disk_len, 0);
+               ref_offset = extent_info->file_offset - extent_info->data_offset;
                btrfs_init_data_ref(&ref, root->root_key.objectid,
                                    btrfs_ino(BTRFS_I(inode)), ref_offset);
                ret = btrfs_inc_extent_ref(trans, &ref);
        }
 
-       clone_info->insertions++;
+       extent_info->insertions++;
 
        return ret;
 }
@@ -2653,15 +2653,15 @@ static int btrfs_insert_clone_extent(struct btrfs_trans_handle *trans,
 /*
  * The respective range must have been previously locked, as well as the inode.
  * The end offset is inclusive (last byte of the range).
- * @clone_info is NULL for fallocate's hole punching and non-NULL for extent
- * cloning.
- * When cloning, we don't want to end up in a state where we dropped extents
- * without inserting a new one, so we must abort the transaction to avoid a
- * corruption.
+ * @extent_info is NULL for fallocate's hole punching and non-NULL when replacing
+ * the file range with an extent.
+ * When not punching a hole, we don't want to end up in a state where we dropped
+ * extents without inserting a new one, so we must abort the transaction to avoid
+ * corruption.
  */
 int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
                           const u64 start, const u64 end,
-                          struct btrfs_clone_extent_info *clone_info,
+                          struct btrfs_replace_extent_info *extent_info,
                           struct btrfs_trans_handle **trans_out)
 {
        struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
@@ -2690,10 +2690,10 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
        /*
         * 1 - update the inode
         * 1 - removing the extents in the range
-        * 1 - adding the hole extent if no_holes isn't set or if we are cloning
-        *     an extent
+        * 1 - adding the hole extent if no_holes isn't set or if we are
+        *     replacing the range with a new extent
         */
-       if (!btrfs_fs_incompat(fs_info, NO_HOLES) || clone_info)
+       if (!btrfs_fs_incompat(fs_info, NO_HOLES) || extent_info)
                rsv_count = 3;
        else
                rsv_count = 2;
@@ -2723,7 +2723,7 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
                         * returned by __btrfs_drop_extents() without having
                         * changed anything in the file.
                         */
-                       if (clone_info && !clone_info->is_new_extent &&
+                       if (extent_info && !extent_info->is_new_extent &&
                            ret && ret != -EOPNOTSUPP)
                                btrfs_abort_transaction(trans, ret);
                        break;
@@ -2731,7 +2731,7 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
 
                trans->block_rsv = &fs_info->trans_block_rsv;
 
-               if (!clone_info && cur_offset < drop_end &&
+               if (!extent_info && cur_offset < drop_end &&
                    cur_offset < ino_size) {
                        ret = fill_holes(trans, BTRFS_I(inode), path,
                                        cur_offset, drop_end);
@@ -2745,7 +2745,7 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
                                btrfs_abort_transaction(trans, ret);
                                break;
                        }
-               } else if (!clone_info && cur_offset < drop_end) {
+               } else if (!extent_info && cur_offset < drop_end) {
                        /*
                         * We are past the i_size here, but since we didn't
                         * insert holes we need to clear the mapped area so we
@@ -2765,18 +2765,18 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
                        }
                }
 
-               if (clone_info && drop_end > clone_info->file_offset) {
-                       u64 clone_len = drop_end - clone_info->file_offset;
+               if (extent_info && drop_end > extent_info->file_offset) {
+                       u64 replace_len = drop_end - extent_info->file_offset;
 
                        ret = btrfs_insert_clone_extent(trans, inode, path,
-                                                       clone_info, clone_len);
+                                                       extent_info, replace_len);
                        if (ret) {
                                btrfs_abort_transaction(trans, ret);
                                break;
                        }
-                       clone_info->data_len -= clone_len;
-                       clone_info->data_offset += clone_len;
-                       clone_info->file_offset += clone_len;
+                       extent_info->data_len -= replace_len;
+                       extent_info->data_offset += replace_len;
+                       extent_info->file_offset += replace_len;
                }
 
                cur_offset = drop_end;
@@ -2800,7 +2800,7 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
                BUG_ON(ret);    /* shouldn't happen */
                trans->block_rsv = rsv;
 
-               if (!clone_info) {
+               if (!extent_info) {
                        ret = find_first_non_hole(inode, &cur_offset, &len);
                        if (unlikely(ret < 0))
                                break;
@@ -2819,7 +2819,7 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
         * than 16Mb would force the full fsync any way (when
         * try_release_extent_mapping() is invoked during page cache truncation.
         */
-       if (clone_info && !clone_info->is_new_extent)
+       if (extent_info && !extent_info->is_new_extent)
                set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
                        &BTRFS_I(inode)->runtime_flags);
 
@@ -2845,7 +2845,7 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
         * (because it's useless) or if it represents a 0 bytes range (when
         * cur_offset == drop_end).
         */
-       if (!clone_info && cur_offset < ino_size && cur_offset < drop_end) {
+       if (!extent_info && cur_offset < ino_size && cur_offset < drop_end) {
                ret = fill_holes(trans, BTRFS_I(inode), path,
                                cur_offset, drop_end);
                if (ret) {
@@ -2853,7 +2853,7 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
                        btrfs_abort_transaction(trans, ret);
                        goto out_trans;
                }
-       } else if (!clone_info && cur_offset < drop_end) {
+       } else if (!extent_info && cur_offset < drop_end) {
                /* See the comment in the loop above for the reasoning here. */
                ret = btrfs_inode_clear_file_extent_range(BTRFS_I(inode),
                                        cur_offset, drop_end - cur_offset);
@@ -2863,9 +2863,9 @@ int btrfs_punch_hole_range(struct inode *inode, struct btrfs_path *path,
                }
 
        }
-       if (clone_info) {
-               ret = btrfs_insert_clone_extent(trans, inode, path, clone_info,
-                                               clone_info->data_len);
+       if (extent_info) {
+               ret = btrfs_insert_clone_extent(trans, inode, path, extent_info,
+                                               extent_info->data_len);
                if (ret) {
                        btrfs_abort_transaction(trans, ret);
                        goto out_trans;
index a9f4875..77ab9ea 100644 (file)
@@ -9582,7 +9582,7 @@ static struct btrfs_trans_handle *insert_prealloc_file_extent(
                                       u64 file_offset)
 {
        struct btrfs_file_extent_item stack_fi;
-       struct btrfs_clone_extent_info extent_info;
+       struct btrfs_replace_extent_info extent_info;
        struct btrfs_trans_handle *trans = trans_in;
        struct btrfs_path *path;
        u64 start = ins->objectid;
index 020da4d..dc8b539 100644 (file)
@@ -439,7 +439,7 @@ process_slot:
 
                if (type == BTRFS_FILE_EXTENT_REG ||
                    type == BTRFS_FILE_EXTENT_PREALLOC) {
-                       struct btrfs_clone_extent_info clone_info;
+                       struct btrfs_replace_extent_info clone_info;
 
                        /*
                         *    a  | --- range to clone ---|  b