diff options
| -rw-r--r-- | fs/btrfs/Makefile | 4 | ||||
| -rw-r--r-- | fs/btrfs/TODO | 1 | ||||
| -rw-r--r-- | fs/btrfs/ctree.h | 25 | ||||
| -rw-r--r-- | fs/btrfs/disk-io.c | 46 | ||||
| -rw-r--r-- | fs/btrfs/file.c | 783 | ||||
| -rw-r--r-- | fs/btrfs/inode.c | 2210 | ||||
| -rw-r--r-- | fs/btrfs/super.c | 2879 |
7 files changed, 3070 insertions, 2878 deletions
diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile index 5346f706b2c..5af2426116d 100644 --- a/fs/btrfs/Makefile +++ b/fs/btrfs/Makefile @@ -4,7 +4,7 @@ ifneq ($(KERNELRELEASE),) obj-m := btrfs.o btrfs-y := super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \ hash.o file-item.o inode-item.o inode-map.o disk-io.o \ - transaction.o bit-radix.o + transaction.o bit-radix.o inode.o file.o #btrfs-y := ctree.o disk-io.o radix-tree.o extent-tree.o print-tree.o \ # root-tree.o dir-item.o hash.o file-item.o inode-item.o \ @@ -16,7 +16,7 @@ else KERNELDIR := /lib/modules/`uname -r`/build all:: - $(MAKE) C=1 -C $(KERNELDIR) M=`pwd` modules + $(MAKE) -C $(KERNELDIR) M=`pwd` modules clean:: rm *.o btrfs.ko endif diff --git a/fs/btrfs/TODO b/fs/btrfs/TODO index d9b6d38c603..4a48a2365d5 100644 --- a/fs/btrfs/TODO +++ b/fs/btrfs/TODO @@ -1,3 +1,4 @@ +* fix printk warnings * cleanup, add more error checking, get rid of BUG_ONs * Fix ENOSPC handling * Make allocator smarter diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 4e136b7b03a..24410d42e3c 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1123,5 +1123,28 @@ struct btrfs_csum_item *btrfs_lookup_csum(struct btrfs_trans_handle *trans, int btrfs_csum_truncate(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, u64 isize); -/* super.c */ +/* inode.c */ +void btrfs_delete_inode(struct inode *inode); +void btrfs_read_locked_inode(struct inode *inode); +int btrfs_write_inode(struct inode *inode, int wait); +void btrfs_dirty_inode(struct inode *inode); +struct inode *btrfs_alloc_inode(struct super_block *sb); +void btrfs_destroy_inode(struct inode *inode); +int btrfs_init_cachep(void); +void btrfs_destroy_cachep(void); +int btrfs_ioctl(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg); +long btrfs_compat_ioctl(struct file *file, unsigned int cmd, + unsigned long arg); +struct inode *btrfs_iget_locked(struct super_block *s, u64 objectid, + struct btrfs_root *root); +int btrfs_commit_write(struct file *file, struct page *page, + unsigned from, unsigned to); +int btrfs_get_block(struct inode *inode, sector_t iblock, + struct buffer_head *result, int create); +/* file.c */ +extern struct file_operations btrfs_file_operations; +int btrfs_drop_extents(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct inode *inode, + u64 start, u64 end, u64 *hint_block); #endif diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 8a88404525e..96bf3ef3a79 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -22,7 +22,7 @@ static int check_tree_block(struct btrfs_root *root, struct buffer_head *buf) if (bh_blocknr(buf) != btrfs_header_blocknr(&node->header)) { printk(KERN_CRIT "bh_blocknr(buf) is %Lu, header is %Lu\n", bh_blocknr(buf), btrfs_header_blocknr(&node->header)); - BUG(); + return 1; } return 0; } @@ -253,7 +253,7 @@ uptodate: set_buffer_checked(bh); } if (check_tree_block(root, bh)) - BUG(); + goto fail; return bh; fail: brelse(bh); @@ -398,8 +398,13 @@ struct btrfs_root *open_ctree(struct super_block *sb) struct btrfs_fs_info *fs_info = kmalloc(sizeof(*fs_info), GFP_NOFS); int ret; + int err = -EIO; struct btrfs_super_block *disk_super; + if (!extent_root || !tree_root || !fs_info) { + err = -ENOMEM; + goto fail; + } init_bit_radix(&fs_info->pinned_radix); init_bit_radix(&fs_info->pending_del_radix); init_bit_radix(&fs_info->extent_map_radix); @@ -431,9 +436,11 @@ struct btrfs_root *open_ctree(struct super_block *sb) mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS); fs_info->hash_tfm = crypto_alloc_hash("crc32c", 0, CRYPTO_ALG_ASYNC); spin_lock_init(&fs_info->hash_lock); + if (!fs_info->hash_tfm || IS_ERR(fs_info->hash_tfm)) { - printk("failed to allocate digest hash\n"); - return NULL; + printk("btrfs: failed hash setup, modprobe cryptomgr?\n"); + err = -ENOMEM; + goto fail_iput; } mutex_init(&fs_info->trans_mutex); mutex_init(&fs_info->fs_mutex); @@ -446,30 +453,53 @@ struct btrfs_root *open_ctree(struct super_block *sb) sb->s_blocksize); if (!fs_info->sb_buffer) - return NULL; + goto fail_iput; disk_super = (struct btrfs_super_block *)fs_info->sb_buffer->b_data; + if (!btrfs_super_root(disk_super)) - return NULL; + goto fail_sb_buffer; i_size_write(fs_info->btree_inode, btrfs_super_total_blocks(disk_super) << fs_info->btree_inode->i_blkbits); fs_info->disk_super = disk_super; + + if (strncmp((char *)(&disk_super->magic), BTRFS_MAGIC, + sizeof(disk_super->magic))) { + printk("btrfs: valid FS not found on %s\n", sb->s_id); + goto fail_sb_buffer; + } tree_root->node = read_tree_block(tree_root, btrfs_super_root(disk_super)); - BUG_ON(!tree_root->node); + if (!tree_root->node) + goto fail_sb_buffer; mutex_lock(&fs_info->fs_mutex); ret = find_and_setup_root(sb->s_blocksize, tree_root, fs_info, BTRFS_EXTENT_TREE_OBJECTID, extent_root); - BUG_ON(ret); + if (ret) { + mutex_unlock(&fs_info->fs_mutex); + goto fail_tree_root; + } btrfs_read_block_groups(extent_root); fs_info->generation = btrfs_super_generation(disk_super) + 1; mutex_unlock(&fs_info->fs_mutex); return tree_root; + +fail_tree_root: + btrfs_block_release(tree_root, tree_root->node); +fail_sb_buffer: + btrfs_block_release(tree_root, fs_info->sb_buffer); +fail_iput: + iput(fs_info->btree_inode); +fail: + kfree(extent_root); + kfree(tree_root); + kfree(fs_info); + return ERR_PTR(err); } int write_ctree_super(struct btrfs_trans_handle *trans, struct btrfs_root diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c new file mode 100644 index 00000000000..7f8e3035d99 --- /dev/null +++ b/fs/btrfs/file.c @@ -0,0 +1,783 @@ +#include <linux/module.h> +#include <linux/buffer_head.h> +#include <linux/fs.h> +#include <linux/pagemap.h> +#include <linux/highmem.h> +#include <linux/time.h> +#include <linux/init.h> +#include <linux/string.h> +#include <linux/smp_lock.h> +#include <linux/backing-dev.h> +#include <linux/mpage.h> +#include <linux/swap.h> +#include <linux/writeback.h> +#include <linux/statfs.h> +#include <linux/compat.h> +#include "ctree.h" +#include "disk-io.h" +#include "transaction.h" +#include "btrfs_inode.h" +#include "ioctl.h" +#include "print-tree.h" + + +static int btrfs_copy_from_user(loff_t pos, int num_pages, int write_bytes, + struct page **prepared_pages, + const char __user * buf) +{ + long page_fault = 0; + int i; + int offset = pos & (PAGE_CACHE_SIZE - 1); + + for (i = 0; i < num_pages && write_bytes > 0; i++, offset = 0) { + size_t count = min_t(size_t, + PAGE_CACHE_SIZE - offset, write_bytes); + struct page *page = prepared_pages[i]; + fault_in_pages_readable(buf, count); + + /* Copy data from userspace to the current page */ + kmap(page); + page_fault = __copy_from_user(page_address(page) + offset, + buf, count); + /* Flush processor's dcache for this page */ + flush_dcache_page(page); + kunmap(page); + buf += count; + write_bytes -= count; + + if (page_fault) + break; + } + return page_fault ? -EFAULT : 0; +} + +static void btrfs_drop_pages(struct page **pages, size_t num_pages) +{ + size_t i; + for (i = 0; i < num_pages; i++) { + if (!pages[i]) + break; + unlock_page(pages[i]); + mark_page_accessed(pages[i]); + page_cache_release(pages[i]); + } +} + +static int dirty_and_release_pages(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct file *file, + struct page **pages, + size_t num_pages, + loff_t pos, + size_t write_bytes) +{ + int i; + int offset; + int err = 0; + int ret; + int this_write; + struct inode *inode = file->f_path.dentry->d_inode; + struct buffer_head *bh; + struct btrfs_file_extent_item *ei; + + for (i = 0; i < num_pages; i++) { + offset = pos & (PAGE_CACHE_SIZE -1); + this_write = min(PAGE_CACHE_SIZE - offset, write_bytes); + /* FIXME, one block at a time */ + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + + bh = page_buffers(pages[i]); + + if (buffer_mapped(bh) && bh->b_blocknr == 0) { + struct btrfs_key key; + struct btrfs_path *path; + char *ptr; + u32 datasize; + + /* create an inline extent, and copy the data in */ + path = btrfs_alloc_path(); + BUG_ON(!path); + key.objectid = inode->i_ino; + key.offset = pages[i]->index << PAGE_CACHE_SHIFT; + key.flags = 0; + btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY); + BUG_ON(write_bytes >= PAGE_CACHE_SIZE); + datasize = offset + + btrfs_file_extent_calc_inline_size(write_bytes); + + ret = btrfs_insert_empty_item(trans, root, path, &key, + datasize); + BUG_ON(ret); + ei = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), + path->slots[0], struct btrfs_file_extent_item); + btrfs_set_file_extent_generation(ei, trans->transid); + btrfs_set_file_extent_type(ei, + BTRFS_FILE_EXTENT_INLINE); + ptr = btrfs_file_extent_inline_start(ei); + btrfs_memcpy(root, path->nodes[0]->b_data, + ptr, bh->b_data, offset + write_bytes); + mark_buffer_dirty(path->nodes[0]); + btrfs_free_path(path); + } else if (buffer_mapped(bh)) { + /* csum the file data */ + btrfs_csum_file_block(trans, root, inode->i_ino, + pages[i]->index << PAGE_CACHE_SHIFT, + kmap(pages[i]), PAGE_CACHE_SIZE); + kunmap(pages[i]); + } + SetPageChecked(pages[i]); + ret = btrfs_end_transaction(trans, root); + BUG_ON(ret); + mutex_unlock(&root->fs_info->fs_mutex); + + ret = btrfs_commit_write(file, pages[i], offset, + offset + this_write); + pos += this_write; + if (ret) { + err = ret; + goto failed; + } + WARN_ON(this_write > write_bytes); + write_bytes -= this_write; + } +failed: + return err; +} + +/* + * this is very complex, but the basic idea is to drop all extents + * in the range start - end. hint_block is filled in with a block number + * that would be a good hint to the block allocator for this file. + * + * If an extent intersects the range but is not entirely inside the range + * it is either truncated or split. Anything entirely inside the range + * is deleted from the tree. + */ +int btrfs_drop_extents(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct inode *inode, + u64 start, u64 end, u64 *hint_block) +{ + int ret; + struct btrfs_key key; + struct btrfs_leaf *leaf; + int slot; + struct btrfs_file_extent_item *extent; + u64 extent_end = 0; + int keep; + struct btrfs_file_extent_item old; + struct btrfs_path *path; + u64 search_start = start; + int bookend; + int found_type; + int found_extent; + int found_inline; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + while(1) { + btrfs_release_path(root, path); + ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino, + search_start, -1); + if (ret < 0) + goto out; + if (ret > 0) { + if (path->slots[0] == 0) { + ret = 0; + goto out; + } + path->slots[0]--; + } + keep = 0; + bookend = 0; + found_extent = 0; + found_inline = 0; + extent = NULL; + leaf = btrfs_buffer_leaf(path->nodes[0]); + slot = path->slots[0]; + btrfs_disk_key_to_cpu(&key, &leaf->items[slot].key); + if (key.offset >= end || key.objectid != inode->i_ino) { + ret = 0; + goto out; + } + if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY) { + ret = 0; + goto out; + } + extent = btrfs_item_ptr(leaf, slot, + struct btrfs_file_extent_item); + found_type = btrfs_file_extent_type(extent); + if (found_type == BTRFS_FILE_EXTENT_REG) { + extent_end = key.offset + + (btrfs_file_extent_num_blocks(extent) << + inode->i_blkbits); + found_extent = 1; + } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { + found_inline = 1; + extent_end = key.offset + + btrfs_file_extent_inline_len(leaf->items + slot); + } + + /* we found nothing we can drop */ + if (!found_extent && !found_inline) { + ret = 0; + goto out; + } + + /* we found nothing inside the range */ + if (search_start >= extent_end) { + ret = 0; + goto out; + } + + /* FIXME, there's only one inline extent allowed right now */ + if (found_inline) { + u64 mask = root->blocksize - 1; + search_start = (extent_end + mask) & ~mask; + } else + search_start = extent_end; + + if (end < extent_end && end >= key.offset) { + if (found_extent) { + u64 disk_blocknr = + btrfs_file_extent_disk_blocknr(extent); + u64 disk_num_blocks = + btrfs_file_extent_disk_num_blocks(extent); + memcpy(&old, extent, sizeof(old)); + if (disk_blocknr != 0) { + ret = btrfs_inc_extent_ref(trans, root, + disk_blocknr, disk_num_blocks); + BUG_ON(ret); + } + } + WARN_ON(found_inline); + bookend = 1; + } + + /* truncate existing extent */ + if (start > key.offset) { + u64 new_num; + u64 old_num; + keep = 1; + WARN_ON(start & (root->blocksize - 1)); + if (found_extent) { + new_num = (start - key.offset) >> + inode->i_blkbits; + old_num = btrfs_file_extent_num_blocks(extent); + *hint_block = + btrfs_file_extent_disk_blocknr(extent); + if (btrfs_file_extent_disk_blocknr(extent)) { + inode->i_blocks -= + (old_num - new_num) << 3; + } + btrfs_set_file_extent_num_blocks(extent, + new_num); + mark_buffer_dirty(path->nodes[0]); + } else { + WARN_ON(1); + } + } + /* delete the entire extent */ + if (!keep) { + u64 disk_blocknr = 0; + u64 disk_num_blocks = 0; + u64 extent_num_blocks = 0; + if (found_extent) { + disk_blocknr = + btrfs_file_extent_disk_blocknr(extent); + disk_num_blocks = + btrfs_file_extent_disk_num_blocks(extent); + extent_num_blocks = + btrfs_file_extent_num_blocks(extent); + *hint_block = + btrfs_file_extent_disk_blocknr(extent); + } + ret = btrfs_del_item(trans, root, path); + BUG_ON(ret); + btrfs_release_path(root, path); + extent = NULL; + if (found_extent && disk_blocknr != 0) { + inode->i_blocks -= extent_num_blocks << 3; + ret = btrfs_free_extent(trans, root, + disk_blocknr, + disk_num_blocks, 0); + } + + BUG_ON(ret); + if (!bookend && search_start >= end) { + ret = 0; + goto out; + } + if (!bookend) + continue; + } + /* create bookend, splitting the extent in two */ + if (bookend && found_extent) { + struct btrfs_key ins; + ins.objectid = inode->i_ino; + ins.offset = end; + ins.flags = 0; + btrfs_set_key_type(&ins, BTRFS_EXTENT_DATA_KEY); + + btrfs_release_path(root, path); + ret = btrfs_insert_empty_item(trans, root, path, &ins, + sizeof(*extent)); + BUG_ON(ret); + extent = btrfs_item_ptr( + btrfs_buffer_leaf(path->nodes[0]), + path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_disk_blocknr(extent, + btrfs_file_extent_disk_blocknr(&old)); + btrfs_set_file_extent_disk_num_blocks(extent, + btrfs_file_extent_disk_num_blocks(&old)); + + btrfs_set_file_extent_offset(extent, + btrfs_file_extent_offset(&old) + + ((end - key.offset) >> inode->i_blkbits)); + WARN_ON(btrfs_file_extent_num_blocks(&old) < + (extent_end - end) >> inode->i_blkbits); + btrfs_set_file_extent_num_blocks(extent, + (extent_end - end) >> inode->i_blkbits); + + btrfs_set_file_extent_type(extent, + BTRFS_FILE_EXTENT_REG); + btrfs_set_file_extent_generation(extent, + btrfs_file_extent_generation(&old)); + btrfs_mark_buffer_dirty(path->nodes[0]); + if (btrfs_file_extent_disk_blocknr(&old) != 0) { + inode->i_blocks += + btrfs_file_extent_num_blocks(extent) << 3; + } + ret = 0; + goto out; + } + } +out: + btrfs_free_path(path); + return ret; +} + +/* + * this gets pages into the page cache and locks them down + */ +static int prepare_pages(struct btrfs_root *root, + struct file *file, + struct page **pages, + size_t num_pages, + loff_t pos, + unsigned long first_index, + unsigned long last_index, + size_t write_bytes, + u64 alloc_extent_start) +{ + int i; + unsigned long index = pos >> PAGE_CACHE_SHIFT; + struct inode *inode = file->f_path.dentry->d_inode; + int offset; + int err = 0; + int this_write; + struct buffer_head *bh; + struct buffer_head *head; + loff_t isize = i_size_read(inode); + + memset(pages, 0, num_pages * sizeof(struct page *)); + + for (i = 0; i < num_pages; i++) { + pages[i] = grab_cache_page(inode->i_mapping, index + i); + if (!pages[i]) { + err = -ENOMEM; + goto failed_release; + } + cancel_dirty_page(pages[i], PAGE_CACHE_SIZE); + wait_on_page_writeback(pages[i]); + offset = pos & (PAGE_CACHE_SIZE -1); + this_write = min(PAGE_CACHE_SIZE - offset, write_bytes); + if (!page_has_buffers(pages[i])) { + create_empty_buffers(pages[i], + root->fs_info->sb->s_blocksize, + (1 << BH_Uptodate)); + } + head = page_buffers(pages[i]); + bh = head; + do { + err = btrfs_map_bh_to_logical(root, bh, + alloc_extent_start); + BUG_ON(err); + if (err) + goto failed_truncate; + bh = bh->b_this_page; + if (alloc_extent_start) + alloc_extent_start++; + } while (bh != head); + pos += this_write; + WARN_ON(this_write > write_bytes); + write_bytes -= this_write; + } + return 0; + +failed_release: + btrfs_drop_pages(pages, num_pages); + return err; + +failed_truncate: + btrfs_drop_pages(pages, num_pages); + if (pos > isize) + vmtruncate(inode, isize); + return err; +} + +static ssize_t btrfs_file_write(struct file *file, const char __user *buf, + size_t count, loff_t *ppos) +{ + loff_t pos; + size_t num_written = 0; + int err = 0; + int ret = 0; + struct inode *inode = file->f_path.dentry->d_inode; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct page *pages[8]; + struct page *pinned[2]; + unsigned long first_index; + unsigned long last_index; + u64 start_pos; + u64 num_blocks; + u64 alloc_extent_start; + u64 hint_block; + struct btrfs_trans_handle *trans; + struct btrfs_key ins; + pinned[0] = NULL; + pinned[1] = NULL; + if (file->f_flags & O_DIRECT) + return -EINVAL; + pos = *ppos; + vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE); + current->backing_dev_info = inode->i_mapping->backing_dev_info; + err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode)); + if (err) + goto out; + if (count == 0) + goto out; + err = remove_suid(file->f_path.dentry); + if (err) + goto out; + file_update_time(file); + + start_pos = pos & ~((u64)PAGE_CACHE_SIZE - 1); + num_blocks = (count + pos - start_pos + root->blocksize - 1) >> + inode->i_blkbits; + + mutex_lock(&inode->i_mutex); + first_index = pos >> PAGE_CACHE_SHIFT; + last_index = (pos + count) >> PAGE_CACHE_SHIFT; + + /* + * there are lots of better ways to do this, but this code + * makes sure the first and last page in the file range are + * up to date and ready for cow + */ + if ((pos & (PAGE_CACHE_SIZE - 1))) { + pinned[0] = grab_cache_page(inode->i_mapping, first_index); + if (!PageUptodate(pinned[0])) { + ret = mpage_readpage(pinned[0], btrfs_get_block); + BUG_ON(ret); + wait_on_page_locked(pinned[0]); + } else { + unlock_page(pinned[0]); + } + } + if ((pos + count) & (PAGE_CACHE_SIZE - 1)) { + pinned[1] = grab_cache_page(inode->i_mapping, last_index); + if (!PageUptodate(pinned[1])) { + ret = mpage_readpage(pinned[1], btrfs_get_block); + BUG_ON(ret); + wait_on_page_locked(pinned[1]); + } else { + unlock_page(pinned[1]); + } + } + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + if (!trans) { + err = -ENOMEM; + mutex_unlock(&root->fs_info->fs_mutex); + goto out_unlock; + } + btrfs_set_trans_block_group(trans, inode); + /* FIXME blocksize != 4096 */ + inode->i_blocks += num_blocks << 3; + hint_block = 0; + + /* FIXME...EIEIO, ENOSPC and more */ + + /* step one, delete the existing extents in this range */ + if (start_pos < inode->i_size) { + /* FIXME blocksize != pagesize */ + ret = btrfs_drop_extents(trans, root, inode, + start_pos, + (pos + count + root->blocksize -1) & + ~((u64)root->blocksize - 1), + &hint_block); + BUG_ON(ret); + } + + /* insert any holes we need to create */ + if (inode->i_size < start_pos) { + u64 last_pos_in_file; + u64 hole_size; + u64 mask = root->blocksize - 1; + last_pos_in_file = (inode->i_size + mask) & ~mask; + hole_size = (start_pos - last_pos_in_file + mask) & ~mask; + hole_size >>= inode->i_blkbits; + if (last_pos_in_file < start_pos) { + ret = btrfs_insert_file_extent(trans, root, + inode->i_ino, + last_pos_in_file, + 0, 0, hole_size); + } + BUG_ON(ret); + } + + /* + * either allocate an extent for the new bytes or setup the key + * to show we are doing inline data in the extent + */ + if (inode->i_size >= PAGE_CACHE_SIZE || pos + count < inode->i_size || + pos + count - start_pos > BTRFS_MAX_INLINE_DATA_SIZE(root)) { + ret = btrfs_alloc_extent(trans, root, inode->i_ino, + num_blocks, hint_block, (u64)-1, + &ins, 1); + BUG_ON(ret); + ret = btrfs_insert_file_extent(trans, root, inode->i_ino, + start_pos, ins.objectid, ins.offset, + ins.offset); + BUG_ON(ret); + } else { + ins.offset = 0; + ins.objectid = 0; + } + BUG_ON(ret); + alloc_extent_start = ins.objectid; + ret = btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + + while(count > 0) { + size_t offset = pos & (PAGE_CACHE_SIZE - 1); + size_t write_bytes = min(count, PAGE_CACHE_SIZE - offset); + size_t num_pages = (write_bytes + PAGE_CACHE_SIZE - 1) >> + PAGE_CACHE_SHIFT; + + memset(pages, 0, sizeof(pages)); + ret = prepare_pages(root, file, pages, num_pages, + pos, first_index, last_index, + write_bytes, alloc_extent_start); + BUG_ON(ret); + + /* FIXME blocks != pagesize */ + if (alloc_extent_start) + alloc_extent_start += num_pages; + ret = btrfs_copy_from_user(pos, num_pages, + write_bytes, pages, buf); + BUG_ON(ret); + + ret = dirty_and_release_pages(NULL, root, file, pages, + num_pages, pos, write_bytes); + BUG_ON(ret); + btrfs_drop_pages(pages, num_pages); + + buf += write_bytes; + count -= write_bytes; + pos += write_bytes; + num_written += write_bytes; + + balance_dirty_pages_ratelimited(inode->i_mapping); + btrfs_btree_balance_dirty(root); + cond_resched(); + } +out_unlock: + mutex_unlock(&inode->i_mutex); +out: + if (pinned[0]) + page_cache_release(pinned[0]); + if (pinned[1]) + page_cache_release(pinned[1]); + *ppos = pos; + current->backing_dev_info = NULL; + mark_inode_dirty(inode); + return num_written ? num_written : err; +} + +/* + * FIXME, do this by stuffing the csum we want in the info hanging off + * page->private. For now, verify file csums on read + */ +static int btrfs_read_actor(read_descriptor_t *desc, struct page *page, + unsigned long offset, unsigned long size) +{ + char *kaddr; + unsigned long left, count = desc->count; + struct inode *inode = page->mapping->host; + + if (size > count) + size = count; + + if (!PageChecked(page)) { + /* FIXME, do it per block */ + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + struct buffer_head *bh; + + if (page_has_buffers(page)) { + bh = page_buffers(page); + if (!buffer_mapped(bh)) { + SetPageChecked(page); + goto checked; + } + } + + ret = btrfs_csum_verify_file_block(root, + page->mapping->host->i_ino, + page->index << PAGE_CACHE_SHIFT, + kmap(page), PAGE_CACHE_SIZE); + if (ret) { + if (ret != -ENOENT) { + printk("failed to verify ino %lu page %lu ret %d\n", + page->mapping->host->i_ino, + page->index, ret); + memset(page_address(page), 1, PAGE_CACHE_SIZE); + flush_dcache_page(page); + } + } + SetPageChecked(page); + kunmap(page); + } +checked: + /* + * Faults on the destination of a read are common, so do it before + * taking the kmap. + */ + if (!fault_in_pages_writeable(desc->arg.buf, size)) { + kaddr = kmap_atomic(page, KM_USER0); + left = __copy_to_user_inatomic(desc->arg.buf, + kaddr + offset, size); + kunmap_atomic(kaddr, KM_USER0); + if (left == 0) + goto success; + } + + /* Do it the slow way */ + kaddr = kmap(page); + left = __copy_to_user(desc->arg.buf, kaddr + offset, size); + kunmap(page); + + if (left) { + size -= left; + desc->error = -EFAULT; + } +success: + desc->count = count - size; + desc->written += size; + desc->arg.buf += size; + return size; +} + +/** + * btrfs_file_aio_read - filesystem read routine, with a mod to csum verify + * @iocb: kernel I/O control block + * @iov: io vector request + * @nr_segs: number of segments in the iovec + * @pos: current file position + */ +static ssize_t btrfs_file_aio_read(struct kiocb *iocb, const struct iovec *iov, + unsigned long nr_segs, loff_t pos) +{ + struct file *filp = iocb->ki_filp; + ssize_t retval; + unsigned long seg; + size_t count; + loff_t *ppos = &iocb->ki_pos; + + count = 0; + for (seg = 0; seg < nr_segs; seg++) { + const struct iovec *iv = &iov[seg]; + + /* + * If any segment has a negative length, or the cumulative + * length ever wraps negative then return -EINVAL. + */ + count += iv->iov_len; + if (unlikely((ssize_t)(count|iv->iov_len) < 0)) + return -EINVAL; + if (access_ok(VERIFY_WRITE, iv->iov_base, iv->iov_len)) + continue; + if (seg == 0) + return -EFAULT; + nr_segs = seg; + count -= iv->iov_len; /* This segment is no good */ + break; + } + retval = 0; + if (count) { + for (seg = 0; seg < nr_segs; seg++) { + read_descriptor_t desc; + + desc.written = 0; + desc.arg.buf = iov[seg].iov_base; + desc.count = iov[seg].iov_len; + if (desc.count == 0) + continue; + desc.error = 0; + do_generic_file_read(filp, ppos, &desc, + btrfs_read_actor); + retval += desc.written; + if (desc.error) { + retval = retval ?: desc.error; + break; + } + } + } + return retval; +} + +static int btrfs_sync_file(struct file *file, + struct dentry *dentry, int datasync) +{ + struct inode *inode = dentry->d_inode; + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + struct btrfs_trans_handle *trans; + + /* + * FIXME, use inode generation number to check if we can skip the + * commit + */ + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + if (!trans) { + ret = -ENOMEM; + goto out; + } + ret = btrfs_commit_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); +out: + return ret > 0 ? EIO : ret; +} + +struct file_operations btrfs_file_operations = { + .llseek = generic_file_llseek, + .read = do_sync_read, + .aio_read = btrfs_file_aio_read, + .write = btrfs_file_write, + .mmap = generic_file_mmap, + .open = generic_file_open, + .ioctl = btrfs_ioctl, + .fsync = btrfs_sync_file, +#ifdef CONFIG_COMPAT + .compat_ioctl = btrfs_compat_ioctl, +#endif +}; + diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c new file mode 100644 index 00000000000..62a3a778d37 --- /dev/null +++ b/fs/btrfs/inode.c @@ -0,0 +1,2210 @@ +#include <linux/module.h> +#include <linux/buffer_head.h> +#include <linux/fs.h> +#include <linux/pagemap.h> +#include <linux/highmem.h> +#include <linux/time.h> +#include <linux/init.h> +#include <linux/string.h> +#include <linux/smp_lock.h> +#include <linux/backing-dev.h> +#include <linux/mpage.h> +#include <linux/swap.h> +#include <linux/writeback.h> +#include <linux/statfs.h> +#include <linux/compat.h> +#include "ctree.h" +#include "disk-io.h" +#include "transaction.h" +#include "btrfs_inode.h" +#include "ioctl.h" +#include "print-tree.h" + +struct btrfs_iget_args { + u64 ino; + struct btrfs_root *root; +}; + +static struct inode_operations btrfs_dir_inode_operations; +static struct inode_operations btrfs_symlink_inode_operations; +static struct inode_operations btrfs_dir_ro_inode_operations; +static struct inode_operations btrfs_file_inode_operations; +static struct address_space_operations btrfs_aops; +static struct address_space_operations btrfs_symlink_aops; +static struct file_operations btrfs_dir_file_operations; + +static struct kmem_cache *btrfs_inode_cachep; +struct kmem_cache *btrfs_trans_handle_cachep; +struct kmem_cache *btrfs_transaction_cachep; +struct kmem_cache *btrfs_bit_radix_cachep; +struct kmem_cache *btrfs_path_cachep; + +#define S_SHIFT 12 +static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = { + [S_IFREG >> S_SHIFT] = BTRFS_FT_REG_FILE, + [S_IFDIR >> S_SHIFT] = BTRFS_FT_DIR, + [S_IFCHR >> S_SHIFT] = BTRFS_FT_CHRDEV, + [S_IFBLK >> S_SHIFT] = BTRFS_FT_BLKDEV, + [S_IFIFO >> S_SHIFT] = BTRFS_FT_FIFO, + [S_IFSOCK >> S_SHIFT] = BTRFS_FT_SOCK, + [S_IFLNK >> S_SHIFT] = BTRFS_FT_SYMLINK, +}; + +void btrfs_read_locked_inode(struct inode *inode) +{ + struct btrfs_path *path; + struct btrfs_inode_item *inode_item; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_key location; + u64 alloc_group_block; + int ret; + + path = btrfs_alloc_path(); + BUG_ON(!path); + btrfs_init_path(path); + mutex_lock(&root->fs_info->fs_mutex); + + memcpy(&location, &BTRFS_I(inode)->location, sizeof(location)); + ret = btrfs_lookup_inode(NULL, root, path, &location, 0); + if (ret) { + btrfs_free_path(path); + goto make_bad; + } + inode_item = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), + path->slots[0], + struct btrfs_inode_item); + + inode->i_mode = btrfs_inode_mode(inode_item); + inode->i_nlink = btrfs_inode_nlink(inode_item); + inode->i_uid = btrfs_inode_uid(inode_item); + inode->i_gid = btrfs_inode_gid(inode_item); + inode->i_size = btrfs_inode_size(inode_item); + inode->i_atime.tv_sec = btrfs_timespec_sec(&inode_item->atime); + inode->i_atime.tv_nsec = btrfs_timespec_nsec(&inode_item->atime); + inode->i_mtime.tv_sec = btrfs_timespec_sec(&inode_item->mtime); + inode->i_mtime.tv_nsec = btrfs_timespec_nsec(&inode_item->mtime); + inode->i_ctime.tv_sec = btrfs_timespec_sec(&inode_item->ctime); + inode->i_ctime.tv_nsec = btrfs_timespec_nsec(&inode_item->ctime); + inode->i_blocks = btrfs_inode_nblocks(inode_item); + inode->i_generation = btrfs_inode_generation(inode_item); + alloc_group_block = btrfs_inode_block_group(inode_item); + BTRFS_I(inode)->block_group = btrfs_lookup_block_group(root->fs_info, + alloc_group_block); + + btrfs_free_path(path); + inode_item = NULL; + + mutex_unlock(&root->fs_info->fs_mutex); + + switch (inode->i_mode & S_IFMT) { +#if 0 + default: + init_special_inode(inode, inode->i_mode, + btrfs_inode_rdev(inode_item)); + break; +#endif + case S_IFREG: + inode->i_mapping->a_ops = &btrfs_aops; + inode->i_fop = &btrfs_file_operations; + inode->i_op = &btrfs_file_inode_operations; + break; + case S_IFDIR: + inode->i_fop = &btrfs_dir_file_operations; + if (root == root->fs_info->tree_root) + inode->i_op = &btrfs_dir_ro_inode_operations; + else + inode->i_op = &btrfs_dir_inode_operations; + break; + case S_IFLNK: + inode->i_op = &btrfs_symlink_inode_operations; + inode->i_mapping->a_ops = &btrfs_symlink_aops; + break; + } + return; + +make_bad: + btrfs_release_path(root, path); + btrfs_free_path(path); + mutex_unlock(&root->fs_info->fs_mutex); + make_bad_inode(inode); +} + +static void fill_inode_item(struct btrfs_inode_item *item, + struct inode *inode) +{ + btrfs_set_inode_uid(item, inode->i_uid); + btrfs_set_inode_gid(item, inode->i_gid); + btrfs_set_inode_size(item, inode->i_size); + btrfs_set_inode_mode(item, inode->i_mode); + btrfs_set_inode_nlink(item, inode->i_nlink); + btrfs_set_timespec_sec(&item->atime, inode->i_atime.tv_sec); + btrfs_set_timespec_nsec(&item->atime, inode->i_atime.tv_nsec); + btrfs_set_timespec_sec(&item->mtime, inode->i_mtime.tv_sec); + btrfs_set_timespec_nsec(&item->mtime, inode->i_mtime.tv_nsec); + btrfs_set_timespec_sec(&item->ctime, inode->i_ctime.tv_sec); + btrfs_set_timespec_nsec(&item->ctime, inode->i_ctime.tv_nsec); + btrfs_set_inode_nblocks(item, inode->i_blocks); + btrfs_set_inode_generation(item, inode->i_generation); + btrfs_set_inode_block_group(item, + BTRFS_I(inode)->block_group->key.objectid); +} + +static int btrfs_update_inode(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode) +{ + struct btrfs_inode_item *inode_item; + struct btrfs_path *path; + int ret; + + path = btrfs_alloc_path(); + BUG_ON(!path); + btrfs_init_path(path); + ret = btrfs_lookup_inode(trans, root, path, + &BTRFS_I(inode)->location, 1); + if (ret) { + if (ret > 0) + ret = -ENOENT; + goto failed; + } + + inode_item = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), + path->slots[0], + struct btrfs_inode_item); + + fill_inode_item(inode_item, inode); + btrfs_mark_buffer_dirty(path->nodes[0]); + ret = 0; +failed: + btrfs_release_path(root, path); + btrfs_free_path(path); + return ret; +} + + +static int btrfs_unlink_trans(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *dir, + struct dentry *dentry) +{ + struct btrfs_path *path; + const char *name = dentry->d_name.name; + int name_len = dentry->d_name.len; + int ret = 0; + u64 objectid; + struct btrfs_dir_item *di; + + path = btrfs_alloc_path(); + BUG_ON(!path); + btrfs_init_path(path); + di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino, + name, name_len, -1); + if (IS_ERR(di)) { + ret = PTR_ERR(di); + goto err; + } + if (!di) { + ret = -ENOENT; + goto err; + } + objectid = btrfs_disk_key_objectid(&di->location); + ret = btrfs_delete_one_dir_name(trans, root, path, di); + BUG_ON(ret); + btrfs_release_path(root, path); + + di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino, + objectid, name, name_len, -1); + if (IS_ERR(di)) { + ret = PTR_ERR(di); + goto err; + } + if (!di) { + ret = -ENOENT; + goto err; + } + ret = btrfs_delete_one_dir_name(trans, root, path, di); + BUG_ON(ret); + + dentry->d_inode->i_ctime = dir->i_ctime; +err: + btrfs_free_path(path); + if (!ret) { + dir->i_size -= name_len * 2; + btrfs_update_inode(trans, root, dir); + drop_nlink(dentry->d_inode); + btrfs_update_inode(trans, root, dentry->d_inode); + dir->i_sb->s_dirt = 1; + } + return ret; +} + +static int btrfs_unlink(struct inode *dir, struct dentry *dentry) +{ + struct btrfs_root *root; + struct btrfs_trans_handle *trans; + int ret; + + root = BTRFS_I(dir)->root; + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, dir); + ret = btrfs_unlink_trans(trans, root, dir, dentry); + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + btrfs_btree_balance_dirty(root); + return ret; +} + +static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) +{ + struct inode *inode = dentry->d_inode; + int err; + int ret; + struct btrfs_root *root = BTRFS_I(dir)->root; + struct btrfs_path *path; + struct btrfs_key key; + struct btrfs_trans_handle *trans; + struct btrfs_key found_key; + int found_type; + struct btrfs_leaf *leaf; + char *goodnames = ".."; + + path = btrfs_alloc_path(); + BUG_ON(!path); + btrfs_init_path(path); + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, dir); + key.objectid = inode->i_ino; + key.offset = (u64)-1; + key.flags = (u32)-1; + while(1) { + ret = btrfs_search_slot(trans, root, &key, path, -1, 1); + if (ret < 0) { + err = ret; + goto out; + } + BUG_ON(ret == 0); + if (path->slots[0] == 0) { + err = -ENOENT; + goto out; + } + path->slots[0]--; + leaf = btrfs_buffer_leaf(path->nodes[0]); + btrfs_disk_key_to_cpu(&found_key, + &leaf->items[path->slots[0]].key); + found_type = btrfs_key_type(&found_key); + if (found_key.objectid != inode->i_ino) { + err = -ENOENT; + goto out; + } + if ((found_type != BTRFS_DIR_ITEM_KEY && + found_type != BTRFS_DIR_INDEX_KEY) || + (!btrfs_match_dir_item_name(root, path, goodnames, 2) && + !btrfs_match_dir_item_name(root, path, goodnames, 1))) { + err = -ENOTEMPTY; + goto out; + } + ret = btrfs_del_item(trans, root, path); + BUG_ON(ret); + + if (found_type == BTRFS_DIR_ITEM_KEY && found_key.offset == 1) + break; + btrfs_release_path(root, path); + } + ret = 0; + btrfs_release_path(root, path); + + /* now the directory is empty */ + err = btrfs_unlink_trans(trans, root, dir, dentry); + if (!err) { + inode->i_size = 0; + } +out: + btrfs_release_path(root, path); + btrfs_free_path(path); + mutex_unlock(&root->fs_info->fs_mutex); + ret = btrfs_end_transaction(trans, root); + btrfs_btree_balance_dirty(root); + if (ret && !err) + err = ret; + return err; +} + +static int btrfs_free_inode(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode) +{ + struct btrfs_path *path; + int ret; + + clear_inode(inode); + + path = btrfs_alloc_path(); + BUG_ON(!path); + btrfs_init_path(path); + ret = btrfs_lookup_inode(trans, root, path, + &BTRFS_I(inode)->location, -1); + BUG_ON(ret); + ret = btrfs_del_item(trans, root, path); + BUG_ON(ret); + btrfs_free_path(path); + return ret; +} + +/* + * truncates go from a high offset to a low offset. So, walk + * from hi to lo in the node and issue readas. Stop when you find + * keys from a different objectid + */ +static void reada_truncate(struct btrfs_root *root, struct btrfs_path *path, + u64 objectid) +{ + struct btrfs_node *node; + int i; + int nritems; + u64 item_objectid; + u64 blocknr; + int slot; + int ret; + + if (!path->nodes[1]) + return; + node = btrfs_buffer_node(path->nodes[1]); + slot = path->slots[1]; + if (slot == 0) + return; + nritems = btrfs_header_nritems(&node->header); + for (i = slot - 1; i >= 0; i--) { + item_objectid = btrfs_disk_key_objectid(&node->ptrs[i].key); + if (item_objectid != objectid) + break; + blocknr = btrfs_node_blockptr(node, i); + ret = readahead_tree_block(root, blocknr); + if (ret) + break; + } +} + +/* + * this can truncate away extent items, csum items and directory items. + * It starts at a high offset and removes keys until it can't find + * any higher than i_size. + * + * csum items that cross the new i_size are truncated to the new size + * as well. + */ +static int btrfs_truncate_in_trans(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode) +{ + int ret; + struct btrfs_path *path; + struct btrfs_key key; + struct btrfs_disk_key *found_key; + u32 found_type; + struct btrfs_leaf *leaf; + struct btrfs_file_extent_item *fi; + u64 extent_start = 0; + u64 extent_num_blocks = 0; + u64 item_end = 0; + int found_extent; + int del_item; + + path = btrfs_alloc_path(); + BUG_ON(!path); + /* FIXME, add redo link to tree so we don't leak on crash */ + key.objectid = inode->i_ino; + key.offset = (u64)-1; + key.flags = (u32)-1; + while(1) { + btrfs_init_path(path); + fi = NULL; + ret = btrfs_search_slot(trans, root, &key, path, -1, 1); + if (ret < 0) { + goto error; + } + if (ret > 0) { + BUG_ON(path->slots[0] == 0); + path->slots[0]--; + } + reada_truncate(root, path, inode->i_ino); + leaf = btrfs_buffer_leaf(path->nodes[0]); + found_key = &leaf->items[path->slots[0]].key; + found_type = btrfs_disk_key_type(found_key); + + if (btrfs_disk_key_objectid(found_key) != inode->i_ino) + break; + if (found_type != BTRFS_CSUM_ITEM_KEY && + found_type != BTRFS_DIR_ITEM_KEY && + found_type != BTRFS_DIR_INDEX_KEY && + found_type != BTRFS_EXTENT_DATA_KEY) + break; + + item_end = btrfs_disk_key_offset(found_key); + if (found_type == BTRFS_EXTENT_DATA_KEY) { + fi = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), + path->slots[0], + struct btrfs_file_extent_item); + if (btrfs_file_extent_type(fi) != + BTRFS_FILE_EXTENT_INLINE) { + item_end += btrfs_file_extent_num_blocks(fi) << + inode->i_blkbits; + } + } + if (found_type == BTRFS_CSUM_ITEM_KEY) { + ret = btrfs_csum_truncate(trans, root, path, + inode->i_size); + BUG_ON(ret); + } + if (item_end < inode->i_size) { + if (found_type) { + btrfs_set_key_type(&key, found_type - 1); + continue; + } + break; + } + if (btrfs_disk_key_offset(found_key) >= inode->i_size) + del_item = 1; + else + del_item = 0; + found_extent = 0; + + /* FIXME, shrink the extent if the ref count is only 1 */ + if (found_type == BTRFS_EXTENT_DATA_KEY && + btrfs_file_extent_type(fi) != + BTRFS_FILE_EXTENT_INLINE) { + u64 num_dec; + if (!del_item) { + u64 orig_num_blocks = + btrfs_file_extent_num_blocks(fi); + extent_num_blocks = inode->i_size - + btrfs_disk_key_offset(found_key) + + root->blocksize - 1; + extent_num_blocks >>= inode->i_blkbits; + btrfs_set_file_extent_num_blocks(fi, + extent_num_blocks); + inode->i_blocks -= (orig_num_blocks - + extent_num_blocks) << 3; + mark_buffer_dirty(path->nodes[0]); + } else { + extent_start = + btrfs_file_extent_disk_blocknr(fi); + extent_num_blocks = + btrfs_file_extent_disk_num_blocks(fi); + /* FIXME blocksize != 4096 */ + num_dec = btrfs_file_extent_num_blocks(fi) << 3; + if (extent_start != 0) { + found_extent = 1; + inode->i_blocks -= num_dec; + } + } + } + if (del_item) { + ret = btrfs_del_item(trans, root, path); + BUG_ON(ret); + } else { + break; + } + btrfs_release_path(root, path); + if (found_extent) { + ret = btrfs_free_extent(trans, root, extent_start, + extent_num_blocks, 0); + BUG_ON(ret); + } + } + ret = 0; +error: + btrfs_release_path(root, path); + btrfs_free_path(path); + inode->i_sb->s_dirt = 1; + return ret; +} + +/* + * taken from block_truncate_page, but does cow as it zeros out + * any bytes left in the last page in the file. + */ +static int btrfs_truncate_page(struct address_space *mapping, loff_t from) +{ + struct inode *inode = mapping->host; + unsigned blocksize = 1 << inode->i_blkbits; + pgoff_t index = from >> PAGE_CACHE_SHIFT; + unsigned offset = from & (PAGE_CACHE_SIZE-1); + struct page *page; + char *kaddr; + int ret = 0; + struct btrfs_root *root = BTRFS_I(inode)->root; + u64 alloc_hint = 0; + struct btrfs_key ins; + struct btrfs_trans_handle *trans; + + if ((offset & (blocksize - 1)) == 0) + goto out; + + ret = -ENOMEM; + page = grab_cache_page(mapping, index); + if (!page) + goto out; + + if (!PageUptodate(page)) { + ret = mpage_readpage(page, btrfs_get_block); + lock_page(page); + if (!PageUptodate(page)) { + ret = -EIO; + goto out; + } + } + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + + ret = btrfs_drop_extents(trans, root, inode, + page->index << PAGE_CACHE_SHIFT, + (page->index + 1) << PAGE_CACHE_SHIFT, + &alloc_hint); + BUG_ON(ret); + ret = btrfs_alloc_extent(trans, root, inode->i_ino, 1, + alloc_hint, (u64)-1, &ins, 1); + BUG_ON(ret); + ret = btrfs_insert_file_extent(trans, root, inode->i_ino, + page->index << PAGE_CACHE_SHIFT, + ins.objectid, 1, 1); + BUG_ON(ret); + SetPageChecked(page); + kaddr = kmap(page); + memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset); + flush_dcache_page(page); + btrfs_csum_file_block(trans, root, inode->i_ino, + page->index << PAGE_CACHE_SHIFT, + kaddr, PAGE_CACHE_SIZE); + kunmap(page); + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + + set_page_dirty(page); + unlock_page(page); + page_cache_release(page); +out: + return ret; +} + +static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) +{ + struct inode *inode = dentry->d_inode; + int err; + + err = inode_change_ok(inode, attr); + if (err) + return err; + + if (S_ISREG(inode->i_mode) && + attr->ia_valid & ATTR_SIZE && attr->ia_size > inode->i_size) { + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(inode)->root; + u64 mask = root->blocksize - 1; + u64 pos = (inode->i_size + mask) & ~mask; + u64 hole_size; + + if (attr->ia_size <= pos) + goto out; + + btrfs_truncate_page(inode->i_mapping, inode->i_size); + + hole_size = (attr->ia_size - pos + mask) & ~mask; + hole_size >>= inode->i_blkbits; + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + err = btrfs_insert_file_extent(trans, root, inode->i_ino, + pos, 0, 0, hole_size); + BUG_ON(err); + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + } +out: + err = inode_setattr(inode, attr); + + return err; +} +void btrfs_delete_inode(struct inode *inode) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + + truncate_inode_pages(&inode->i_data, 0); + if (is_bad_inode(inode)) { + goto no_delete; + } + inode->i_size = 0; + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + ret = btrfs_truncate_in_trans(trans, root, inode); + BUG_ON(ret); + btrfs_free_inode(trans, root, inode); + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + btrfs_btree_balance_dirty(root); + return; +no_delete: + clear_inode(inode); +} + +/* + * this returns the key found in the dir entry in the location pointer. + * If no dir entries were found, location->objectid is 0. + */ +static int btrfs_inode_by_name(struct inode *dir, struct dentry *dentry, + struct btrfs_key *location) +{ + const char *name = dentry->d_name.name; + int namelen = dentry->d_name.len; + struct btrfs_dir_item *di; + struct btrfs_path *path; + struct btrfs_root *root = BTRFS_I(dir)->root; + int ret; + + path = btrfs_alloc_path(); + BUG_ON(!path); + btrfs_init_path(path); + di = btrfs_lookup_dir_item(NULL, root, path, dir->i_ino, name, + namelen, 0); + if (!di || IS_ERR(di)) { + location->objectid = 0; + ret = 0; + goto out; + } + btrfs_disk_key_to_cpu(location, &di->location); +out: + btrfs_release_path(root, path); + btrfs_free_path(path); + return ret; +} + +/* + * when we hit a tree root in a directory, the btrfs part of the inode + * needs to be changed to reflect the root directory of the tree root. This + * is kind of like crossing a mount point. + */ +static int fixup_tree_root_location(struct btrfs_root *root, + struct btrfs_key *location, + struct btrfs_root **sub_root) +{ + struct btrfs_path *path; + struct btrfs_root_item *ri; + + if (btrfs_key_type(location) != BTRFS_ROOT_ITEM_KEY) + return 0; + if (location->objectid == BTRFS_ROOT_TREE_OBJECTID) + return 0; + + path = btrfs_alloc_path(); + BUG_ON(!path); + mutex_lock(&root->fs_info->fs_mutex); + + *sub_root = btrfs_read_fs_root(root->fs_info, location); + if (IS_ERR(*sub_root)) + return PTR_ERR(*sub_root); + + ri = &(*sub_root)->root_item; + location->objectid = btrfs_root_dirid(ri); + location->flags = 0; + btrfs_set_key_type(location, BTRFS_INODE_ITEM_KEY); + location->offset = 0; + + btrfs_free_path(path); + mutex_unlock(&root->fs_info->fs_mutex); + return 0; +} + +static int btrfs_init_locked_inode(struct inode *inode, void *p) +{ + struct btrfs_iget_args *args = p; + inode->i_ino = args->ino; + BTRFS_I(inode)->root = args->root; + return 0; +} + +static int btrfs_find_actor(struct inode *inode, void *opaque) +{ + struct btrfs_iget_args *args = opaque; + return (args->ino == inode->i_ino && + args->root == BTRFS_I(inode)->root); +} + +struct inode *btrfs_iget_locked(struct super_block *s, u64 objectid, + struct btrfs_root *root) +{ + struct inode *inode; + struct btrfs_iget_args args; + args.ino = objectid; + args.root = root; + + inode = iget5_locked(s, objectid, btrfs_find_actor, + btrfs_init_locked_inode, + (void *)&args); + return inode; +} + +static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry, + struct nameidata *nd) +{ + struct inode * inode; + struct btrfs_inode *bi = BTRFS_I(dir); + struct btrfs_root *root = bi->root; + struct btrfs_root *sub_root = root; + struct btrfs_key location; + int ret; + + if (dentry->d_name.len > BTRFS_NAME_LEN) + return ERR_PTR(-ENAMETOOLONG); + mutex_lock(&root->fs_info->fs_mutex); + ret = btrfs_inode_by_name(dir, dentry, &location); + mutex_unlock(&root->fs_info->fs_mutex); + if (ret < 0) + return ERR_PTR(ret); + inode = NULL; + if (location.objectid) { + ret = fixup_tree_root_location(root, &location, &sub_root); + if (ret < 0) + return ERR_PTR(ret); + if (ret > 0) + return ERR_PTR(-ENOENT); + inode = btrfs_iget_locked(dir->i_sb, location.objectid, + sub_root); + if (!inode) + return ERR_PTR(-EACCES); + if (inode->i_state & I_NEW) { + /* the inode and parent dir are two different roots */ + if (sub_root != root) { + igrab(inode); + sub_root->inode = inode; + } + BTRFS_I(inode)->root = sub_root; + memcpy(&BTRFS_I(inode)->location, &location, + sizeof(location)); + btrfs_read_locked_inode(inode); + unlock_new_inode(inode); + } + } + return d_splice_alias(inode, dentry); +} + +/* + * readahead one full node of leaves as long as their keys include + * the objectid supplied + */ +static void reada_leaves(struct btrfs_root *root, struct btrfs_path *path, + u64 objectid) +{ + struct btrfs_node *node; + int i; + u32 nritems; + u64 item_objectid; + u64 blocknr; + int slot; + int ret; + + if (!path->nodes[1]) + return; + node = btrfs_buffer_node(path->nodes[1]); + slot = path->slots[1]; + nritems = btrfs_header_nritems(&node->header); + for (i = slot + 1; i < nritems; i++) { + item_objectid = btrfs_disk_key_objectid(&node->ptrs[i].key); + if (item_objectid != objectid) + break; + blocknr = btrfs_node_blockptr(node, i); + ret = readahead_tree_block(root, blocknr); + if (ret) + break; + } +} +static unsigned char btrfs_filetype_table[] = { + DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK +}; + +static int btrfs_readdir(struct file *filp, void *dirent, filldir_t filldir) +{ + struct inode *inode = filp->f_path.dentry->d_inode; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_item *item; + struct btrfs_dir_item *di; + struct btrfs_key key; + struct btrfs_path *path; + int ret; + u32 nritems; + struct btrfs_leaf *leaf; + int slot; + int advance; + unsigned char d_type; + int over = 0; + u32 di_cur; + u32 di_total; + u32 di_len; + int key_type = BTRFS_DIR_INDEX_KEY; + + /* FIXME, use a real flag for deciding about the key type */ + if (root->fs_info->tree_root == root) + key_type = BTRFS_DIR_ITEM_KEY; + mutex_lock(&root->fs_info->fs_mutex); + key.objectid = inode->i_ino; + key.flags = 0; + btrfs_set_key_type(&key, key_type); + key.offset = filp->f_pos; + path = btrfs_alloc_path(); + btrfs_init_path(path); + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) + goto err; + advance = 0; + reada_leaves(root, path, inode->i_ino); + while(1) { + leaf = btrfs_buffer_leaf(path->nodes[0]); + nritems = btrfs_header_nritems(&leaf->header); + slot = path->slots[0]; + if (advance || slot >= nritems) { + if (slot >= nritems -1) { + reada_leaves(root, path, inode->i_ino); + ret = btrfs_next_leaf(root, path); + if (ret) + break; + leaf = btrfs_buffer_leaf(path->nodes[0]); + nritems = btrfs_header_nritems(&leaf->header); + slot = path->slots[0]; + } else { + slot++; + path->slots[0]++; + } + } + advance = 1; + item = leaf->items + slot; + if (btrfs_disk_key_objectid(&item->key) != key.objectid) + break; + if (btrfs_disk_key_type(&item->key) != key_type) + break; + if (btrfs_disk_key_offset(&item->key) < filp->f_pos) + continue; + filp->f_pos = btrfs_disk_key_offset(&item->key); + advance = 1; + di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item); + di_cur = 0; + di_total = btrfs_item_size(leaf->items + slot); + while(di_cur < di_total) { + d_type = btrfs_filetype_table[btrfs_dir_type(di)]; + over = filldir(dirent, (const char *)(di + 1), + btrfs_dir_name_len(di), + btrfs_disk_key_offset(&item->key), + btrfs_disk_key_objectid(&di->location), + d_type); + if (over) + goto nopos; + di_len = btrfs_dir_name_len(di) + sizeof(*di); + di_cur += di_len; + di = (struct btrfs_dir_item *)((char *)di + di_len); + } + } + filp->f_pos++; +nopos: + ret = 0; +err: + btrfs_release_path(root, path); + btrfs_free_path(path); + mutex_unlock(&root->fs_info->fs_mutex); + return ret; +} + +int btrfs_write_inode(struct inode *inode, int wait) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_trans_handle *trans; + int ret = 0; + + if (wait) { + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + ret = btrfs_commit_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + } + return ret; +} + +/* + * This is somewhat expense, updating the tree every time the + * inode changes. But, it is most likely to find the inode in cache. + * FIXME, needs more benchmarking...there are no reasons other than performance + * to keep or drop this code. + */ +void btrfs_dirty_inode(struct inode *inode) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_trans_handle *trans; + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + btrfs_update_inode(trans, root, inode); + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + btrfs_btree_balance_dirty(root); +} + +static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + u64 objectid, + struct btrfs_block_group_cache *group, + int mode) +{ + struct inode *inode; + struct btrfs_inode_item inode_item; + struct btrfs_key *location; + int ret; + int owner; + + inode = new_inode(root->fs_info->sb); + if (!inode) + return ERR_PTR(-ENOMEM); + + BTRFS_I(inode)->root = root; + if (mode & S_IFDIR) + owner = 0; + else + owner = 1; + group = btrfs_find_block_group(root, group, 0, 0, owner); + BTRFS_I(inode)->block_group = group; + + inode->i_uid = current->fsuid; + inode->i_gid = current->fsgid; + inode->i_mode = mode; + inode->i_ino = objectid; + inode->i_blocks = 0; + inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME; + fill_inode_item(&inode_item, inode); + location = &BTRFS_I(inode)->location; + location->objectid = objectid; + location->flags = 0; + location->offset = 0; + btrfs_set_key_type(location, BTRFS_INODE_ITEM_KEY); + + ret = btrfs_insert_inode(trans, root, objectid, &inode_item); + BUG_ON(ret); + + insert_inode_hash(inode); + return inode; +} + +static inline u8 btrfs_inode_type(struct inode *inode) +{ + return btrfs_type_by_mode[(inode->i_mode & S_IFMT) >> S_SHIFT]; +} + +static int btrfs_add_link(struct btrfs_trans_handle *trans, + struct dentry *dentry, struct inode *inode) +{ + int ret; + struct btrfs_key key; + struct btrfs_root *root = BTRFS_I(dentry->d_parent->d_inode)->root; + key.objectid = inode->i_ino; + key.flags = 0; + btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); + key.offset = 0; + + ret = btrfs_insert_dir_item(trans, root, + dentry->d_name.name, dentry->d_name.len, + dentry->d_parent->d_inode->i_ino, + &key, btrfs_inode_type(inode)); + if (ret == 0) { + dentry->d_parent->d_inode->i_size += dentry->d_name.len * 2; + ret = btrfs_update_inode(trans, root, + dentry->d_parent->d_inode); + } + return ret; +} + +static int btrfs_add_nondir(struct btrfs_trans_handle *trans, + struct dentry *dentry, struct inode *inode) +{ + int err = btrfs_add_link(trans, dentry, inode); + if (!err) { + d_instantiate(dentry, inode); + return 0; + } + if (err > 0) + err = -EEXIST; + return err; +} + +static int btrfs_create(struct inode *dir, struct dentry *dentry, + int mode, struct nameidata *nd) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(dir)->root; + struct inode *inode; + int err; + int drop_inode = 0; + u64 objectid; + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, dir); + + err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); + if (err) { + err = -ENOSPC; + goto out_unlock; + } + + inode = btrfs_new_inode(trans, root, objectid, + BTRFS_I(dir)->block_group, mode); + err = PTR_ERR(inode); + if (IS_ERR(inode)) + goto out_unlock; + + btrfs_set_trans_block_group(trans, inode); + err = btrfs_add_nondir(trans, dentry, inode); + if (err) + drop_inode = 1; + else { + inode->i_mapping->a_ops = &btrfs_aops; + inode->i_fop = &btrfs_file_operations; + inode->i_op = &btrfs_file_inode_operations; + } + dir->i_sb->s_dirt = 1; + btrfs_update_inode_block_group(trans, inode); + btrfs_update_inode_block_group(trans, dir); +out_unlock: + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + + if (drop_inode) { + inode_dec_link_count(inode); + iput(inode); + } + btrfs_btree_balance_dirty(root); + return err; +} + +static int btrfs_link(struct dentry *old_dentry, struct inode *dir, + struct dentry *dentry) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(dir)->root; + struct inode *inode = old_dentry->d_inode; + int err; + int drop_inode = 0; + + if (inode->i_nlink == 0) + return -ENOENT; + + inc_nlink(inode); + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, dir); + atomic_inc(&inode->i_count); + err = btrfs_add_nondir(trans, dentry, inode); + if (err) + drop_inode = 1; + dir->i_sb->s_dirt = 1; + btrfs_update_inode_block_group(trans, dir); + btrfs_update_inode(trans, root, inode); + + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + + if (drop_inode) { + inode_dec_link_count(inode); + iput(inode); + } + btrfs_btree_balance_dirty(root); + return err; +} + +static int btrfs_make_empty_dir(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + u64 objectid, u64 dirid) +{ + int ret; + char buf[2]; + struct btrfs_key key; + + buf[0] = '.'; + buf[1] = '.'; + + key.objectid = objectid; + key.offset = 0; + key.flags = 0; + btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); + + ret = btrfs_insert_dir_item(trans, root, buf, 1, objectid, + &key, BTRFS_FT_DIR); + if (ret) + goto error; + key.objectid = dirid; + ret = btrfs_insert_dir_item(trans, root, buf, 2, objectid, + &key, BTRFS_FT_DIR); + if (ret) + goto error; +error: + return ret; +} + +static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) +{ + struct inode *inode; + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(dir)->root; + int err = 0; + int drop_on_err = 0; + u64 objectid; + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, dir); + if (IS_ERR(trans)) { + err = PTR_ERR(trans); + goto out_unlock; + } + + err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); + if (err) { + err = -ENOSPC; + goto out_unlock; + } + + inode = btrfs_new_inode(trans, root, objectid, + BTRFS_I(dir)->block_group, S_IFDIR | mode); + if (IS_ERR(inode)) { + err = PTR_ERR(inode); + goto out_fail; + } + drop_on_err = 1; + inode->i_op = &btrfs_dir_inode_operations; + inode->i_fop = &btrfs_dir_file_operations; + btrfs_set_trans_block_group(trans, inode); + + err = btrfs_make_empty_dir(trans, root, inode->i_ino, dir->i_ino); + if (err) + goto out_fail; + + inode->i_size = 6; + err = btrfs_update_inode(trans, root, inode); + if (err) + goto out_fail; + err = btrfs_add_link(trans, dentry, inode); + if (err) + goto out_fail; + d_instantiate(dentry, inode); + drop_on_err = 0; + dir->i_sb->s_dirt = 1; + btrfs_update_inode_block_group(trans, inode); + btrfs_update_inode_block_group(trans, dir); + +out_fail: + btrfs_end_transaction(trans, root); +out_unlock: + mutex_unlock(&root->fs_info->fs_mutex); + if (drop_on_err) + iput(inode); + btrfs_btree_balance_dirty(root); + return err; +} + +/* + * FIBMAP and others want to pass in a fake buffer head. They need to + * use BTRFS_GET_BLOCK_NO_DIRECT to make sure we don't try to memcpy + * any packed file data into the fake bh + */ +#define BTRFS_GET_BLOCK_NO_CREATE 0 +#define BTRFS_GET_BLOCK_CREATE 1 +#define BTRFS_GET_BLOCK_NO_DIRECT 2 + +/* + * FIXME create==1 doe not work. + */ +static int btrfs_get_block_lock(struct inode *inode, sector_t iblock, + struct buffer_head *result, int create) +{ + int ret; + int err = 0; + u64 blocknr; + u64 extent_start = 0; + u64 extent_end = 0; + u64 objectid = inode->i_ino; + u32 found_type; + u64 alloc_hint = 0; + struct btrfs_path *path; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_file_extent_item *item; + struct btrfs_leaf *leaf; + struct btrfs_disk_key *found_key; + struct btrfs_trans_handle *trans = NULL; + + path = btrfs_alloc_path(); + BUG_ON(!path); + btrfs_init_path(path); + if (create & BTRFS_GET_BLOCK_CREATE) { + WARN_ON(1); + /* this almost but not quite works */ + trans = btrfs_start_transaction(root, 1); + if (!trans) { + err = -ENOMEM; + goto out; + } + ret = btrfs_drop_extents(trans, root, inode, + iblock << inode->i_blkbits, + (iblock + 1) << inode->i_blkbits, + &alloc_hint); + BUG_ON(ret); + } + + ret = btrfs_lookup_file_extent(NULL, root, path, + inode->i_ino, + iblock << inode->i_blkbits, 0); + if (ret < 0) { + err = ret; + goto out; + } + + if (ret != 0) { + if (path->slots[0] == 0) { + btrfs_release_path(root, path); + goto not_found; + } + path->slots[0]--; + } + + item = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), path->slots[0], + struct btrfs_file_extent_item); + leaf = btrfs_buffer_leaf(path->nodes[0]); + blocknr = btrfs_file_extent_disk_blocknr(item); + blocknr += btrfs_file_extent_offset(item); + + /* are we inside the extent that was found? */ + found_key = &leaf->items[path->slots[0]].key; + found_type = btrfs_disk_key_type(found_key); + if (btrfs_disk_key_objectid(found_key) != objectid || + found_type != BTRFS_EXTENT_DATA_KEY) { + extent_end = 0; + extent_start = 0; + goto not_found; + } + found_type = btrfs_file_extent_type(item); + extent_start = btrfs_disk_key_offset(&leaf->items[path->slots[0]].key); + if (found_type == BTRFS_FILE_EXTENT_REG) { + extent_start = extent_start >> inode->i_blkbits; + extent_end = extent_start + btrfs_file_extent_num_blocks(item); + err = 0; + if (btrfs_file_extent_disk_blocknr(item) == 0) + goto out; + if (iblock >= extent_start && iblock < extent_end) { + btrfs_map_bh_to_logical(root, result, blocknr + + iblock - extent_start); + goto out; + } + } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { + char *ptr; + char *map; + u32 size; + + if (create & BTRFS_GET_BLOCK_NO_DIRECT) { + err = -EINVAL; + goto out; + } + size = btrfs_file_extent_inline_len(leaf->items + + path->slots[0]); + extent_end = (extent_start + size) >> inode->i_blkbits; + extent_start >>= inode->i_blkbits; + if (iblock < extent_start || iblock > extent_end) { + goto not_found; + } + ptr = btrfs_file_extent_inline_start(item); + map = kmap(result->b_page); + memcpy(map, ptr, size); + memset(map + size, 0, PAGE_CACHE_SIZE - size); + flush_dcache_page(result->b_page); + kunmap(result->b_page); + set_buffer_uptodate(result); + SetPageChecked(result->b_page); + btrfs_map_bh_to_logical(root, result, 0); + } +not_found: + if (create & BTRFS_GET_BLOCK_CREATE) { + struct btrfs_key ins; + ret = btrfs_alloc_extent(trans, root, inode->i_ino, + 1, alloc_hint, (u64)-1, + &ins, 1); + BUG_ON(ret); + ret = btrfs_insert_file_extent(trans, root, inode->i_ino, + iblock << inode->i_blkbits, + ins.objectid, ins.offset, + ins.offset); + BUG_ON(ret); + SetPageChecked(result->b_page); + btrfs_map_bh_to_logical(root, result, ins.objectid); + } +out: + if (trans) + err = btrfs_end_transaction(trans, root); + btrfs_free_path(path); + return err; +} + +int btrfs_get_block(struct inode *inode, sector_t iblock, + struct buffer_head *result, int create) +{ + int err; + struct btrfs_root *root = BTRFS_I(inode)->root; + mutex_lock(&root->fs_info->fs_mutex); + err = btrfs_get_block_lock(inode, iblock, result, create); + mutex_unlock(&root->fs_info->fs_mutex); + return err; +} + +static int btrfs_get_block_bmap(struct inode *inode, sector_t iblock, + struct buffer_head *result, int create) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + mutex_lock(&root->fs_info->fs_mutex); + btrfs_get_block_lock(inode, iblock, result, BTRFS_GET_BLOCK_NO_DIRECT); + mutex_unlock(&root->fs_info->fs_mutex); + return 0; +} + +static sector_t btrfs_bmap(struct address_space *as, sector_t block) +{ + return generic_block_bmap(as, block, btrfs_get_block_bmap); +} + +static int btrfs_prepare_write(struct file *file, struct page *page, + unsigned from, unsigned to) +{ + return block_prepare_write(page, from, to, btrfs_get_block); +} + +static int btrfs_readpage(struct file *file, struct page *page) +{ + return mpage_readpage(page, btrfs_get_block); +} + +/* + * Aside from a tiny bit of packed file data handling, this is the + * same as the generic code. + * + * While block_write_full_page is writing back the dirty buffers under + * the page lock, whoever dirtied the buffers may decide to clean them + * again at any time. We handle that by only looking at the buffer + * state inside lock_buffer(). + * + * If block_write_full_page() is called for regular writeback + * (wbc->sync_mode == WB_SYNC_NONE) then it will redirty a page which has a + * locked buffer. This only can happen if someone has written the buffer + * directly, with submit_bh(). At the address_space level PageWriteback + * prevents this contention from occurring. + */ +static int __btrfs_write_full_page(struct inode *inode, struct page *page, + struct writeback_control *wbc) +{ + int err; + sector_t block; + sector_t last_block; + struct buffer_head *bh, *head; + const unsigned blocksize = 1 << inode->i_blkbits; + int nr_underway = 0; + + BUG_ON(!PageLocked(page)); + + last_block = (i_size_read(inode) - 1) >> inode->i_blkbits; + + if (!page_has_buffers(page)) { + create_empty_buffers(page, blocksize, + (1 << BH_Dirty)|(1 << BH_Uptodate)); + } + + /* + * 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. + */ + + block = (sector_t)page->index << (PAGE_CACHE_SHIFT - inode->i_blkbits); + head = page_buffers(page); + bh = head; + + /* + * Get all the dirty buffers mapped to disk addresses and + * handle any aliases from the underlying blockdev's mapping. + */ + do { + if (block > last_block) { + /* + * mapped buffers outside i_size will occur, because + * this page can be outside i_size when there is a + * truncate in progress. + */ + /* + * The buffer was zeroed by block_write_full_page() + */ + clear_buffer_dirty(bh); + set_buffer_uptodate(bh); + } else if (!buffer_mapped(bh) && buffer_dirty(bh)) { + WARN_ON(bh->b_size != blocksize); + err = btrfs_get_block(inode, block, bh, 0); + if (err) { + goto recover; + } + if (buffer_new(bh)) { + /* blockdev mappings never come here */ + clear_buffer_new(bh); + } + } + bh = bh->b_this_page; + block++; + } while (bh != head); + + do { + if (!buffer_mapped(bh)) + continue; + /* + * If it's a fully non-blocking write attempt and we cannot + * lock the buffer then redirty the page. Note that this can + * potentially cause a busy-wait loop from pdflush and kswapd + * activity, but those code paths have their own higher-level + * throttling. + */ + if (wbc->sync_mode != WB_SYNC_NONE || !wbc->nonblocking) { + lock_buffer(bh); + } else if (test_set_buffer_locked(bh)) { + redirty_page_for_writepage(wbc, page); + continue; + } + if (test_clear_buffer_dirty(bh) && bh->b_blocknr != 0) { + mark_buffer_async_write(bh); + } else { + unlock_buffer(bh); + } + } while ((bh = bh->b_this_page) != head); + + /* + * The page and its buffers are protected by PageWriteback(), so we can + * drop the bh refcounts early. + */ + BUG_ON(PageWriteback(page)); + set_page_writeback(page); + + do { + struct buffer_head *next = bh->b_this_page; + if (buffer_async_write(bh)) { + submit_bh(WRITE, bh); + nr_underway++; + } + bh = next; + } while (bh != head); + unlock_page(page); + + err = 0; +done: + if (nr_underway == 0) { + /* + * The page was marked dirty, but the buffers were + * clean. Someone wrote them back by hand with + * ll_rw_block/submit_bh. A rare case. + */ + int uptodate = 1; + do { + if (!buffer_uptodate(bh)) { + uptodate = 0; + break; + } + bh = bh->b_this_page; + } while (bh != head); + if (uptodate) + SetPageUptodate(page); + end_page_writeback(page); + } + return err; + +recover: + /* + * ENOSPC, or some other error. We may already have added some + * blocks to the file, so we need to write these out to avoid + * exposing stale data. + * The page is currently locked and not marked for writeback + */ + bh = head; + /* Recovery: lock and submit the mapped buffers */ + do { + if (buffer_mapped(bh) && buffer_dirty(bh)) { + lock_buffer(bh); + mark_buffer_async_write(bh); + } else { + /* + * The buffer may have been set dirty during + * attachment to a dirty page. + */ + clear_buffer_dirty(bh); + } + } while ((bh = bh->b_this_page) != head); + SetPageError(page); + BUG_ON(PageWriteback(page)); + set_page_writeback(page); + do { + struct buffer_head *next = bh->b_this_page; + if (buffer_async_write(bh)) { + clear_buffer_dirty(bh); + submit_bh(WRITE, bh); + nr_underway++; + } + bh = next; + } while (bh != head); + unlock_page(page); + goto done; +} + +static int btrfs_writepage(struct page *page, struct writeback_control *wbc) +{ + struct inode * const inode = page->mapping->host; + loff_t i_size = i_size_read(inode); + const pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT; + unsigned offset; + void *kaddr; + + /* Is the page fully inside i_size? */ + if (page->index < end_index) + return __btrfs_write_full_page(inode, page, wbc); + + /* Is the page fully outside i_size? (truncate in progress) */ + offset = i_size & (PAGE_CACHE_SIZE-1); + if (page->index >= end_index+1 || !offset) { + /* + * The page may have dirty, unmapped buffers. For example, + * they may have been added in ext3_writepage(). Make them + * freeable here, so the page does not leak. + */ + block_invalidatepage(page, 0); + unlock_page(page); + return 0; /* don't care */ + } + + /* + * The page straddles i_size. It must be zeroed out on each and every + * writepage invokation because it may be mmapped. "A file is mapped + * in multiples of the page size. For a file that is not a multiple of + * the page size, the remaining memory is zeroed when mapped, and + * writes to that region are not written out to the file." + */ + kaddr = kmap_atomic(page, KM_USER0); + memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset); + flush_dcache_page(page); + kunmap_atomic(kaddr, KM_USER0); + return __btrfs_write_full_page(inode, page, wbc); +} + +static void btrfs_truncate(struct inode *inode) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + struct btrfs_trans_handle *trans; + + if (!S_ISREG(inode->i_mode)) + return; + if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) + return; + + btrfs_truncate_page(inode->i_mapping, inode->i_size); + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); + + /* FIXME, add redo link to tree so we don't leak on crash */ + ret = btrfs_truncate_in_trans(trans, root, inode); + BUG_ON(ret); + btrfs_update_inode(trans, root, inode); + ret = btrfs_end_transaction(trans, root); + BUG_ON(ret); + mutex_unlock(&root->fs_info->fs_mutex); + btrfs_btree_balance_dirty(root); +} + +int btrfs_commit_write(struct file *file, struct page *page, + unsigned from, unsigned to) +{ + struct inode *inode = page->mapping->host; + struct buffer_head *bh; + loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to; + + SetPageUptodate(page); + bh = page_buffers(page); + set_buffer_uptodate(bh); + if (buffer_mapped(bh) && bh->b_blocknr != 0) { + set_page_dirty(page); + } + if (pos > inode->i_size) { + i_size_write(inode, pos); + mark_inode_dirty(inode); + } + return 0; +} + +static int create_subvol(struct btrfs_root *root, char *name, int namelen) +{ + struct btrfs_trans_handle *trans; + struct btrfs_key key; + struct btrfs_root_item root_item; + struct btrfs_inode_item *inode_item; + struct buffer_head *subvol; + struct btrfs_leaf *leaf; + struct btrfs_root *new_root; + struct inode *inode; + struct inode *dir; + int ret; + u64 objectid; + u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID; + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + BUG_ON(!trans); + + subvol = btrfs_alloc_free_block(trans, root, 0); + if (subvol == NULL) + return -ENOSPC; + leaf = btrfs_buffer_leaf(subvol); + btrfs_set_header_nritems(&leaf->header, 0); + btrfs_set_header_level(&leaf->header, 0); + btrfs_set_header_blocknr(&leaf->header, bh_blocknr(subvol)); + btrfs_set_header_generation(&leaf->header, trans->transid); + btrfs_set_header_owner(&leaf->header, root->root_key.objectid); + memcpy(leaf->header.fsid, root->fs_info->disk_super->fsid, + sizeof(leaf->header.fsid)); + mark_buffer_dirty(subvol); + + inode_item = &root_item.inode; + memset(inode_item, 0, sizeof(*inode_item)); + btrfs_set_inode_generation(inode_item, 1); + btrfs_set_inode_size(inode_item, 3); + btrfs_set_inode_nlink(inode_item, 1); + btrfs_set_inode_nblocks(inode_item, 1); + btrfs_set_inode_mode(inode_item, S_IFDIR | 0755); + + btrfs_set_root_blocknr(&root_item, bh_blocknr(subvol)); + btrfs_set_root_refs(&root_item, 1); + brelse(subvol); + subvol = NULL; + + ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root, + 0, &objectid); + BUG_ON(ret); + + btrfs_set_root_dirid(&root_item, new_dirid); + + key.objectid = objectid; + key.offset = 1; + key.flags = 0; + btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); + ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key, + &root_item); + BUG_ON(ret); + + /* + * insert the directory item + */ + key.offset = (u64)-1; + dir = root->fs_info->sb->s_root->d_inode; + ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root, + name, namelen, dir->i_ino, &key, + BTRFS_FT_DIR); + BUG_ON(ret); + + ret = btrfs_commit_transaction(trans, root); + BUG_ON(ret); + + new_root = btrfs_read_fs_root(root->fs_info, &key); + BUG_ON(!new_root); + + trans = btrfs_start_transaction(new_root, 1); + BUG_ON(!trans); + + inode = btrfs_new_inode(trans, new_root, new_dirid, + BTRFS_I(dir)->block_group, S_IFDIR | 0700); + inode->i_op = &btrfs_dir_inode_operations; + inode->i_fop = &btrfs_dir_file_operations; + + ret = btrfs_make_empty_dir(trans, new_root, new_dirid, new_dirid); + BUG_ON(ret); + + inode->i_nlink = 1; + inode->i_size = 6; + ret = btrfs_update_inode(trans, new_root, inode); + BUG_ON(ret); + + ret = btrfs_commit_transaction(trans, new_root); + BUG_ON(ret); + + iput(inode); + + mutex_unlock(&root->fs_info->fs_mutex); + btrfs_btree_balance_dirty(root); + return 0; +} + +static int create_snapshot(struct btrfs_root *root, char *name, int namelen) +{ + struct btrfs_trans_handle *trans; + struct btrfs_key key; + struct btrfs_root_item new_root_item; + int ret; + u64 objectid; + + if (!root->ref_cows) + return -EINVAL; + + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + BUG_ON(!trans); + + ret = btrfs_update_inode(trans, root, root->inode); + BUG_ON(ret); + + ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root, + 0, &objectid); + BUG_ON(ret); + + memcpy(&new_root_item, &root->root_item, + sizeof(new_root_item)); + + key.objectid = objectid; + key.offset = 1; + key.flags = 0; + btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); + btrfs_set_root_blocknr(&new_root_item, bh_blocknr(root->node)); + + ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key, + &new_root_item); + BUG_ON(ret); + + /* + * insert the directory item + */ + key.offset = (u64)-1; + ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root, + name, namelen, + root->fs_info->sb->s_root->d_inode->i_ino, + &key, BTRFS_FT_DIR); + + BUG_ON(ret); + + ret = btrfs_inc_root_ref(trans, root); + BUG_ON(ret); + + ret = btrfs_commit_transaction(trans, root); + BUG_ON(ret); + mutex_unlock(&root->fs_info->fs_mutex); + btrfs_btree_balance_dirty(root); + return 0; +} + +int btrfs_ioctl(struct inode *inode, struct file *filp, unsigned int + cmd, unsigned long arg) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_ioctl_vol_args vol_args; + int ret = 0; + struct btrfs_dir_item *di; + int namelen; + struct btrfs_path *path; + u64 root_dirid; + + switch (cmd) { + case BTRFS_IOC_SNAP_CREATE: + if (copy_from_user(&vol_args, + (struct btrfs_ioctl_vol_args __user *)arg, + sizeof(vol_args))) + return -EFAULT; + namelen = strlen(vol_args.name); + if (namelen > BTRFS_VOL_NAME_MAX) + return -EINVAL; + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + root_dirid = root->fs_info->sb->s_root->d_inode->i_ino, + mutex_lock(&root->fs_info->fs_mutex); + di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root, + path, root_dirid, + vol_args.name, namelen, 0); + mutex_unlock(&root->fs_info->fs_mutex); + btrfs_free_path(path); + if (di && !IS_ERR(di)) + return -EEXIST; + + if (root == root->fs_info->tree_root) + ret = create_subvol(root, vol_args.name, namelen); + else + ret = create_snapshot(root, vol_args.name, namelen); + WARN_ON(ret); + break; + default: + return -ENOTTY; + } + return ret; +} + +#ifdef CONFIG_COMPAT +long btrfs_compat_ioctl(struct file *file, unsigned int cmd, + unsigned long arg) +{ + struct inode *inode = file->f_path.dentry->d_inode; + int ret; + lock_kernel(); + ret = btrfs_ioctl(inode, file, cmd, (unsigned long) compat_ptr(arg)); + unlock_kernel(); + return ret; + +} +#endif + +/* + * Called inside transaction, so use GFP_NOFS + */ +struct inode *btrfs_alloc_inode(struct super_block *sb) +{ + struct btrfs_inode *ei; + + ei = kmem_cache_alloc(btrfs_inode_cachep, GFP_NOFS); + if (!ei) + return NULL; + return &ei->vfs_inode; +} + +void btrfs_destroy_inode(struct inode *inode) +{ + WARN_ON(!list_empty(&inode->i_dentry)); + WARN_ON(inode->i_data.nrpages); + + kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode)); +} + +static void init_once(void * foo, struct kmem_cache * cachep, + unsigned long flags) +{ + struct btrfs_inode *ei = (struct btrfs_inode *) foo; + + inode_init_once(&ei->vfs_inode); +} + +void btrfs_destroy_cachep(void) +{ + if (btrfs_inode_cachep) + kmem_cache_destroy(btrfs_inode_cachep); + if (btrfs_trans_handle_cachep) + kmem_cache_destroy(btrfs_trans_handle_cachep); + if (btrfs_transaction_cachep) + kmem_cache_destroy(btrfs_transaction_cachep); + if (btrfs_bit_radix_cachep) + kmem_cache_destroy(btrfs_bit_radix_cachep); + if (btrfs_path_cachep) + kmem_cache_destroy(btrfs_path_cachep); +} + +int btrfs_init_cachep(void) +{ + btrfs_inode_cachep = kmem_cache_create("btrfs_inode_cache", + sizeof(struct btrfs_inode), + 0, (SLAB_RECLAIM_ACCOUNT| + SLAB_MEM_SPREAD), + init_once, NULL); + if (!btrfs_inode_cachep) + goto fail; + btrfs_trans_handle_cachep = kmem_cache_create("btrfs_trans_handle_cache", + sizeof(struct btrfs_trans_handle), + 0, (SLAB_RECLAIM_ACCOUNT| + SLAB_MEM_SPREAD), + NULL, NULL); + if (!btrfs_trans_handle_cachep) + goto fail; + btrfs_transaction_cachep = kmem_cache_create("btrfs_transaction_cache", + sizeof(struct btrfs_transaction), + 0, (SLAB_RECLAIM_ACCOUNT| + SLAB_MEM_SPREAD), + NULL, NULL); + if (!btrfs_transaction_cachep) + goto fail; + btrfs_path_cachep = kmem_cache_create("btrfs_path_cache", + sizeof(struct btrfs_transaction), + 0, (SLAB_RECLAIM_ACCOUNT| + SLAB_MEM_SPREAD), + NULL, NULL); + if (!btrfs_path_cachep) + goto fail; + btrfs_bit_radix_cachep = kmem_cache_create("btrfs_radix", + 256, + 0, (SLAB_RECLAIM_ACCOUNT| + SLAB_MEM_SPREAD | + SLAB_DESTROY_BY_RCU), + NULL, NULL); + if (!btrfs_bit_radix_cachep) + goto fail; + return 0; +fail: + btrfs_destroy_cachep(); + return -ENOMEM; +} + +static int btrfs_getattr(struct vfsmount *mnt, + struct dentry *dentry, struct kstat *stat) +{ + struct inode *inode = dentry->d_inode; + generic_fillattr(inode, stat); + stat->blksize = 256 * 1024; + return 0; +} + +static int btrfs_rename(struct inode * old_dir, struct dentry *old_dentry, + struct inode * new_dir,struct dentry *new_dentry) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(old_dir)->root; + struct inode *new_inode = new_dentry->d_inode; + struct inode *old_inode = old_dentry->d_inode; + struct timespec ctime = CURRENT_TIME; + struct btrfs_path *path; + struct btrfs_dir_item *di; + int ret; + + if (S_ISDIR(old_inode->i_mode) && new_inode && + new_inode->i_size > BTRFS_EMPTY_DIR_SIZE) { + return -ENOTEMPTY; + } + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, new_dir); + path = btrfs_alloc_path(); + if (!path) { + ret = -ENOMEM; + goto out_fail; + } + + old_dentry->d_inode->i_nlink++; + old_dir->i_ctime = old_dir->i_mtime = ctime; + new_dir->i_ctime = new_dir->i_mtime = ctime; + old_inode->i_ctime = ctime; + if (S_ISDIR(old_inode->i_mode) && old_dir != new_dir) { + struct btrfs_key *location = &BTRFS_I(new_dir)->location; + u64 old_parent_oid; + di = btrfs_lookup_dir_item(trans, root, path, old_inode->i_ino, + "..", 2, -1); + if (IS_ERR(di)) { + ret = PTR_ERR(di); + goto out_fail; + } + if (!di) { + ret = -ENOENT; + goto out_fail; + } + old_parent_oid = btrfs_disk_key_objectid(&di->location); + ret = btrfs_del_item(trans, root, path); + if (ret) { + ret = -EIO; + goto out_fail; + } + btrfs_release_path(root, path); + + di = btrfs_lookup_dir_index_item(trans, root, path, + old_inode->i_ino, + old_parent_oid, + "..", 2, -1); + if (IS_ERR(di)) { + ret = PTR_ERR(di); + goto out_fail; + } + if (!di) { + ret = -ENOENT; + goto out_fail; + } + ret = btrfs_del_item(trans, root, path); + if (ret) { + ret = -EIO; + goto out_fail; + } + btrfs_release_path(root, path); + + ret = btrfs_insert_dir_item(trans, root, "..", 2, + old_inode->i_ino, location, + BTRFS_FT_DIR); + if (ret) + goto out_fail; + } + + + ret = btrfs_unlink_trans(trans, root, old_dir, old_dentry); + if (ret) + goto out_fail; + + if (new_inode) { + new_inode->i_ctime = CURRENT_TIME; + ret = btrfs_unlink_trans(trans, root, new_dir, new_dentry); + if (ret) + goto out_fail; + if (S_ISDIR(new_inode->i_mode)) + clear_nlink(new_inode); + else + drop_nlink(new_inode); + btrfs_update_inode(trans, root, new_inode); + } + ret = btrfs_add_link(trans, new_dentry, old_inode); + if (ret) + goto out_fail; + +out_fail: + btrfs_free_path(path); + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + return ret; +} + +static int btrfs_symlink(struct inode *dir, struct dentry *dentry, + const char *symname) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(dir)->root; + struct btrfs_path *path; + struct btrfs_key key; + struct inode *inode; + int err; + int drop_inode = 0; + u64 objectid; + int name_len; + int datasize; + char *ptr; + struct btrfs_file_extent_item *ei; + + name_len = strlen(symname) + 1; + if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root)) + return -ENAMETOOLONG; + mutex_lock(&root->fs_info->fs_mutex); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, dir); + + err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); + if (err) { + err = -ENOSPC; + goto out_unlock; + } + + inode = btrfs_new_inode(trans, root, objectid, + BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO); + err = PTR_ERR(inode); + if (IS_ERR(inode)) + goto out_unlock; + + btrfs_set_trans_block_group(trans, inode); + err = btrfs_add_nondir(trans, dentry, inode); + if (err) + drop_inode = 1; + else { + inode->i_mapping->a_ops = &btrfs_aops; + inode->i_fop = &btrfs_file_operations; + inode->i_op = &btrfs_file_inode_operations; + } + dir->i_sb->s_dirt = 1; + btrfs_update_inode_block_group(trans, inode); + btrfs_update_inode_block_group(trans, dir); + if (drop_inode) + goto out_unlock; + + path = btrfs_alloc_path(); + BUG_ON(!path); + key.objectid = inode->i_ino; + key.offset = 0; + key.flags = 0; + btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY); + datasize = btrfs_file_extent_calc_inline_size(name_len); + err = btrfs_insert_empty_item(trans, root, path, &key, + datasize); + BUG_ON(err); + ei = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), + path->slots[0], struct btrfs_file_extent_item); + btrfs_set_file_extent_generation(ei, trans->transid); + btrfs_set_file_extent_type(ei, + BTRFS_FILE_EXTENT_INLINE); + ptr = btrfs_file_extent_inline_start(ei); + btrfs_memcpy(root, path->nodes[0]->b_data, + ptr, symname, name_len); + mark_buffer_dirty(path->nodes[0]); + btrfs_free_path(path); + inode->i_op = &btrfs_symlink_inode_operations; + inode->i_mapping->a_ops = &btrfs_symlink_aops; + inode->i_size = name_len - 1; + btrfs_update_inode(trans, root, inode); + err = 0; + +out_unlock: + btrfs_end_transaction(trans, root); + mutex_unlock(&root->fs_info->fs_mutex); + + if (drop_inode) { + inode_dec_link_count(inode); + iput(inode); + } + btrfs_btree_balance_dirty(root); + return err; +} + +static struct inode_operations btrfs_dir_inode_operations = { + .lookup = btrfs_lookup, + .create = btrfs_create, + .unlink = btrfs_unlink, + .link = btrfs_link, + .mkdir = btrfs_mkdir, + .rmdir = btrfs_rmdir, + .rename = btrfs_rename, + .symlink = btrfs_symlink, + .setattr = btrfs_setattr, +}; + +static struct inode_operations btrfs_dir_ro_inode_operations = { + .lookup = btrfs_lookup, +}; + +static struct file_operations btrfs_dir_file_operations = { + .llseek = generic_file_llseek, + .read = generic_read_dir, + .readdir = btrfs_readdir, + .ioctl = btrfs_ioctl, +#ifdef CONFIG_COMPAT + .compat_ioctl = btrfs_compat_ioctl, +#endif +}; + +static struct address_space_operations btrfs_aops = { + .readpage = btrfs_readpage, + .writepage = btrfs_writepage, + .sync_page = block_sync_page, + .prepare_write = btrfs_prepare_write, + .commit_write = btrfs_commit_write, + .bmap = btrfs_bmap, +}; + +static struct address_space_operations btrfs_symlink_aops = { + .readpage = btrfs_readpage, + .writepage = btrfs_writepage, +}; + +static struct inode_operations btrfs_file_inode_operations = { + .truncate = btrfs_truncate, + .getattr = btrfs_getattr, + .setattr = btrfs_setattr, +}; + +static struct inode_operations btrfs_symlink_inode_operations = { + .readlink = generic_readlink, + .follow_link = page_follow_link_light, + .put_link = page_put_link, +}; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 9749090cfbb..94b3535a2a1 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -20,873 +20,9 @@ #include "ioctl.h" #include "print-tree.h" -struct btrfs_iget_args { - u64 ino; - struct btrfs_root *root; -}; - #define BTRFS_SUPER_MAGIC 0x9123682E -static struct inode_operations btrfs_dir_inode_operations; -static struct inode_operations btrfs_symlink_inode_operations; -static struct inode_operations btrfs_dir_ro_inode_operations; static struct super_operations btrfs_super_ops; -static struct file_operations btrfs_dir_file_operations; -static struct inode_operations btrfs_file_inode_operations; -static struct address_space_operations btrfs_aops; -static struct address_space_operations btrfs_symlink_aops; -static struct file_operations btrfs_file_operations; - -static int drop_extents(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode, - u64 start, u64 end, u64 *hint_block); -static int btrfs_get_block(struct inode *inode, sector_t iblock, - struct buffer_head *result, int create); - - -#define S_SHIFT 12 -static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = { - [S_IFREG >> S_SHIFT] = BTRFS_FT_REG_FILE, - [S_IFDIR >> S_SHIFT] = BTRFS_FT_DIR, - [S_IFCHR >> S_SHIFT] = BTRFS_FT_CHRDEV, - [S_IFBLK >> S_SHIFT] = BTRFS_FT_BLKDEV, - [S_IFIFO >> S_SHIFT] = BTRFS_FT_FIFO, - [S_IFSOCK >> S_SHIFT] = BTRFS_FT_SOCK, - [S_IFLNK >> S_SHIFT] = BTRFS_FT_SYMLINK, -}; - -static void btrfs_read_locked_inode(struct inode *inode) -{ - struct btrfs_path *path; - struct btrfs_inode_item *inode_item; - struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_key location; - u64 alloc_group_block; - int ret; - - path = btrfs_alloc_path(); - BUG_ON(!path); - btrfs_init_path(path); - mutex_lock(&root->fs_info->fs_mutex); - - memcpy(&location, &BTRFS_I(inode)->location, sizeof(location)); - ret = btrfs_lookup_inode(NULL, root, path, &location, 0); - if (ret) { - btrfs_free_path(path); - goto make_bad; - } - inode_item = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), - path->slots[0], - struct btrfs_inode_item); - - inode->i_mode = btrfs_inode_mode(inode_item); - inode->i_nlink = btrfs_inode_nlink(inode_item); - inode->i_uid = btrfs_inode_uid(inode_item); - inode->i_gid = btrfs_inode_gid(inode_item); - inode->i_size = btrfs_inode_size(inode_item); - inode->i_atime.tv_sec = btrfs_timespec_sec(&inode_item->atime); - inode->i_atime.tv_nsec = btrfs_timespec_nsec(&inode_item->atime); - inode->i_mtime.tv_sec = btrfs_timespec_sec(&inode_item->mtime); - inode->i_mtime.tv_nsec = btrfs_timespec_nsec(&inode_item->mtime); - inode->i_ctime.tv_sec = btrfs_timespec_sec(&inode_item->ctime); - inode->i_ctime.tv_nsec = btrfs_timespec_nsec(&inode_item->ctime); - inode->i_blocks = btrfs_inode_nblocks(inode_item); - inode->i_generation = btrfs_inode_generation(inode_item); - alloc_group_block = btrfs_inode_block_group(inode_item); - BTRFS_I(inode)->block_group = btrfs_lookup_block_group(root->fs_info, - alloc_group_block); - - btrfs_free_path(path); - inode_item = NULL; - - mutex_unlock(&root->fs_info->fs_mutex); - - switch (inode->i_mode & S_IFMT) { -#if 0 - default: - init_special_inode(inode, inode->i_mode, - btrfs_inode_rdev(inode_item)); - break; -#endif - case S_IFREG: - inode->i_mapping->a_ops = &btrfs_aops; - inode->i_fop = &btrfs_file_operations; - inode->i_op = &btrfs_file_inode_operations; - break; - case S_IFDIR: - inode->i_fop = &btrfs_dir_file_operations; - if (root == root->fs_info->tree_root) - inode->i_op = &btrfs_dir_ro_inode_operations; - else - inode->i_op = &btrfs_dir_inode_operations; - break; - case S_IFLNK: - inode->i_op = &btrfs_symlink_inode_operations; - inode->i_mapping->a_ops = &btrfs_symlink_aops; - break; - } - return; - -make_bad: - btrfs_release_path(root, path); - btrfs_free_path(path); - mutex_unlock(&root->fs_info->fs_mutex); - make_bad_inode(inode); -} - -static void fill_inode_item(struct btrfs_inode_item *item, - struct inode *inode) -{ - btrfs_set_inode_uid(item, inode->i_uid); - btrfs_set_inode_gid(item, inode->i_gid); - btrfs_set_inode_size(item, inode->i_size); - btrfs_set_inode_mode(item, inode->i_mode); - btrfs_set_inode_nlink(item, inode->i_nlink); - btrfs_set_timespec_sec(&item->atime, inode->i_atime.tv_sec); - btrfs_set_timespec_nsec(&item->atime, inode->i_atime.tv_nsec); - btrfs_set_timespec_sec(&item->mtime, inode->i_mtime.tv_sec); - btrfs_set_timespec_nsec(&item->mtime, inode->i_mtime.tv_nsec); - btrfs_set_timespec_sec(&item->ctime, inode->i_ctime.tv_sec); - btrfs_set_timespec_nsec(&item->ctime, inode->i_ctime.tv_nsec); - btrfs_set_inode_nblocks(item, inode->i_blocks); - btrfs_set_inode_generation(item, inode->i_generation); - btrfs_set_inode_block_group(item, - BTRFS_I(inode)->block_group->key.objectid); -} - -static int btrfs_update_inode(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode) -{ - struct btrfs_inode_item *inode_item; - struct btrfs_path *path; - int ret; - - path = btrfs_alloc_path(); - BUG_ON(!path); - btrfs_init_path(path); - ret = btrfs_lookup_inode(trans, root, path, - &BTRFS_I(inode)->location, 1); - if (ret) { - if (ret > 0) - ret = -ENOENT; - goto failed; - } - - inode_item = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), - path->slots[0], - struct btrfs_inode_item); - - fill_inode_item(inode_item, inode); - btrfs_mark_buffer_dirty(path->nodes[0]); - ret = 0; -failed: - btrfs_release_path(root, path); - btrfs_free_path(path); - return ret; -} - - -static int btrfs_unlink_trans(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *dir, - struct dentry *dentry) -{ - struct btrfs_path *path; - const char *name = dentry->d_name.name; - int name_len = dentry->d_name.len; - int ret = 0; - u64 objectid; - struct btrfs_dir_item *di; - - path = btrfs_alloc_path(); - BUG_ON(!path); - btrfs_init_path(path); - di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino, - name, name_len, -1); - if (IS_ERR(di)) { - ret = PTR_ERR(di); - goto err; - } - if (!di) { - ret = -ENOENT; - goto err; - } - objectid = btrfs_disk_key_objectid(&di->location); - ret = btrfs_delete_one_dir_name(trans, root, path, di); - BUG_ON(ret); - btrfs_release_path(root, path); - - di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino, - objectid, name, name_len, -1); - if (IS_ERR(di)) { - ret = PTR_ERR(di); - goto err; - } - if (!di) { - ret = -ENOENT; - goto err; - } - ret = btrfs_delete_one_dir_name(trans, root, path, di); - BUG_ON(ret); - - dentry->d_inode->i_ctime = dir->i_ctime; -err: - btrfs_free_path(path); - if (!ret) { - dir->i_size -= name_len * 2; - btrfs_update_inode(trans, root, dir); - drop_nlink(dentry->d_inode); - btrfs_update_inode(trans, root, dentry->d_inode); - dir->i_sb->s_dirt = 1; - } - return ret; -} - -static int btrfs_unlink(struct inode *dir, struct dentry *dentry) -{ - struct btrfs_root *root; - struct btrfs_trans_handle *trans; - int ret; - - root = BTRFS_I(dir)->root; - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, dir); - ret = btrfs_unlink_trans(trans, root, dir, dentry); - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - btrfs_btree_balance_dirty(root); - return ret; -} - -static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) -{ - struct inode *inode = dentry->d_inode; - int err; - int ret; - struct btrfs_root *root = BTRFS_I(dir)->root; - struct btrfs_path *path; - struct btrfs_key key; - struct btrfs_trans_handle *trans; - struct btrfs_key found_key; - int found_type; - struct btrfs_leaf *leaf; - char *goodnames = ".."; - - path = btrfs_alloc_path(); - BUG_ON(!path); - btrfs_init_path(path); - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, dir); - key.objectid = inode->i_ino; - key.offset = (u64)-1; - key.flags = (u32)-1; - while(1) { - ret = btrfs_search_slot(trans, root, &key, path, -1, 1); - if (ret < 0) { - err = ret; - goto out; - } - BUG_ON(ret == 0); - if (path->slots[0] == 0) { - err = -ENOENT; - goto out; - } - path->slots[0]--; - leaf = btrfs_buffer_leaf(path->nodes[0]); - btrfs_disk_key_to_cpu(&found_key, - &leaf->items[path->slots[0]].key); - found_type = btrfs_key_type(&found_key); - if (found_key.objectid != inode->i_ino) { - err = -ENOENT; - goto out; - } - if ((found_type != BTRFS_DIR_ITEM_KEY && - found_type != BTRFS_DIR_INDEX_KEY) || - (!btrfs_match_dir_item_name(root, path, goodnames, 2) && - !btrfs_match_dir_item_name(root, path, goodnames, 1))) { - err = -ENOTEMPTY; - goto out; - } - ret = btrfs_del_item(trans, root, path); - BUG_ON(ret); - - if (found_type == BTRFS_DIR_ITEM_KEY && found_key.offset == 1) - break; - btrfs_release_path(root, path); - } - ret = 0; - btrfs_release_path(root, path); - - /* now the directory is empty */ - err = btrfs_unlink_trans(trans, root, dir, dentry); - if (!err) { - inode->i_size = 0; - } -out: - btrfs_release_path(root, path); - btrfs_free_path(path); - mutex_unlock(&root->fs_info->fs_mutex); - ret = btrfs_end_transaction(trans, root); - btrfs_btree_balance_dirty(root); - if (ret && !err) - err = ret; - return err; -} - -static int btrfs_free_inode(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode) -{ - struct btrfs_path *path; - int ret; - - clear_inode(inode); - - path = btrfs_alloc_path(); - BUG_ON(!path); - btrfs_init_path(path); - ret = btrfs_lookup_inode(trans, root, path, - &BTRFS_I(inode)->location, -1); - BUG_ON(ret); - ret = btrfs_del_item(trans, root, path); - BUG_ON(ret); - btrfs_free_path(path); - return ret; -} - -static void reada_truncate(struct btrfs_root *root, struct btrfs_path *path, - u64 objectid) -{ - struct btrfs_node *node; - int i; - int nritems; - u64 item_objectid; - u64 blocknr; - int slot; - int ret; - - if (!path->nodes[1]) - return; - node = btrfs_buffer_node(path->nodes[1]); - slot = path->slots[1]; - if (slot == 0) - return; - nritems = btrfs_header_nritems(&node->header); - for (i = slot - 1; i >= 0; i--) { - item_objectid = btrfs_disk_key_objectid(&node->ptrs[i].key); - if (item_objectid != objectid) - break; - blocknr = btrfs_node_blockptr(node, i); - ret = readahead_tree_block(root, blocknr); - if (ret) - break; - } -} - -static int btrfs_truncate_in_trans(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode) -{ - int ret; - struct btrfs_path *path; - struct btrfs_key key; - struct btrfs_disk_key *found_key; - u32 found_type; - struct btrfs_leaf *leaf; - struct btrfs_file_extent_item *fi; - u64 extent_start = 0; - u64 extent_num_blocks = 0; - u64 item_end = 0; - int found_extent; - int del_item; - - path = btrfs_alloc_path(); - BUG_ON(!path); - /* FIXME, add redo link to tree so we don't leak on crash */ - key.objectid = inode->i_ino; - key.offset = (u64)-1; - key.flags = (u32)-1; - while(1) { - btrfs_init_path(path); - fi = NULL; - ret = btrfs_search_slot(trans, root, &key, path, -1, 1); - if (ret < 0) { - goto error; - } - if (ret > 0) { - BUG_ON(path->slots[0] == 0); - path->slots[0]--; - } - reada_truncate(root, path, inode->i_ino); - leaf = btrfs_buffer_leaf(path->nodes[0]); - found_key = &leaf->items[path->slots[0]].key; - found_type = btrfs_disk_key_type(found_key); - if (btrfs_disk_key_objectid(found_key) != inode->i_ino) - break; - if (found_type != BTRFS_CSUM_ITEM_KEY && - found_type != BTRFS_DIR_ITEM_KEY && - found_type != BTRFS_DIR_INDEX_KEY && - found_type != BTRFS_EXTENT_DATA_KEY) - break; - item_end = btrfs_disk_key_offset(found_key); - if (found_type == BTRFS_EXTENT_DATA_KEY) { - fi = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), - path->slots[0], - struct btrfs_file_extent_item); - if (btrfs_file_extent_type(fi) != - BTRFS_FILE_EXTENT_INLINE) { - item_end += btrfs_file_extent_num_blocks(fi) << - inode->i_blkbits; - } - } - if (found_type == BTRFS_CSUM_ITEM_KEY) { - ret = btrfs_csum_truncate(trans, root, path, - inode->i_size); - BUG_ON(ret); - } - if (item_end < inode->i_size) { - if (found_type) { - btrfs_set_key_type(&key, found_type - 1); - continue; - } - break; - } - if (btrfs_disk_key_offset(found_key) >= inode->i_size) - del_item = 1; - else - del_item = 0; - found_extent = 0; - - if (found_type == BTRFS_EXTENT_DATA_KEY && - btrfs_file_extent_type(fi) != - BTRFS_FILE_EXTENT_INLINE) { - u64 num_dec; - if (!del_item) { - u64 orig_num_blocks = - btrfs_file_extent_num_blocks(fi); - extent_num_blocks = inode->i_size - - btrfs_disk_key_offset(found_key) + - root->blocksize - 1; - extent_num_blocks >>= inode->i_blkbits; - btrfs_set_file_extent_num_blocks(fi, - extent_num_blocks); - inode->i_blocks -= (orig_num_blocks - - extent_num_blocks) << 3; - mark_buffer_dirty(path->nodes[0]); - } else { - extent_start = - btrfs_file_extent_disk_blocknr(fi); - extent_num_blocks = - btrfs_file_extent_disk_num_blocks(fi); - /* FIXME blocksize != 4096 */ - num_dec = btrfs_file_extent_num_blocks(fi) << 3; - if (extent_start != 0) { - found_extent = 1; - inode->i_blocks -= num_dec; - } - } - } - if (del_item) { - ret = btrfs_del_item(trans, root, path); - BUG_ON(ret); - } else { - break; - } - btrfs_release_path(root, path); - if (found_extent) { - ret = btrfs_free_extent(trans, root, extent_start, - extent_num_blocks, 0); - BUG_ON(ret); - } - } - ret = 0; -error: - btrfs_release_path(root, path); - btrfs_free_path(path); - inode->i_sb->s_dirt = 1; - return ret; -} - -static int btrfs_truncate_page(struct address_space *mapping, loff_t from) -{ - struct inode *inode = mapping->host; - unsigned blocksize = 1 << inode->i_blkbits; - pgoff_t index = from >> PAGE_CACHE_SHIFT; - unsigned offset = from & (PAGE_CACHE_SIZE-1); - struct page *page; - char *kaddr; - int ret = 0; - struct btrfs_root *root = BTRFS_I(inode)->root; - u64 alloc_hint; - struct btrfs_key ins; - struct btrfs_trans_handle *trans; - - if ((offset & (blocksize - 1)) == 0) - goto out; - - ret = -ENOMEM; - page = grab_cache_page(mapping, index); - if (!page) - goto out; - - if (!PageUptodate(page)) { - ret = mpage_readpage(page, btrfs_get_block); - lock_page(page); - if (!PageUptodate(page)) { - ret = -EIO; - goto out; - } - } - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - - ret = drop_extents(trans, root, inode, page->index << PAGE_CACHE_SHIFT, - (page->index + 1) << PAGE_CACHE_SHIFT, &alloc_hint); - BUG_ON(ret); - ret = btrfs_alloc_extent(trans, root, inode->i_ino, 1, - alloc_hint, (u64)-1, &ins, 1); - BUG_ON(ret); - ret = btrfs_insert_file_extent(trans, root, inode->i_ino, - page->index << PAGE_CACHE_SHIFT, - ins.objectid, 1, 1); - BUG_ON(ret); - SetPageChecked(page); - kaddr = kmap(page); - memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset); - flush_dcache_page(page); - btrfs_csum_file_block(trans, root, inode->i_ino, - page->index << PAGE_CACHE_SHIFT, - kaddr, PAGE_CACHE_SIZE); - kunmap(page); - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - - set_page_dirty(page); - unlock_page(page); - page_cache_release(page); -out: - return ret; -} - -static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) -{ - struct inode *inode = dentry->d_inode; - int err; - - err = inode_change_ok(inode, attr); - if (err) - return err; - - if (S_ISREG(inode->i_mode) && - attr->ia_valid & ATTR_SIZE && attr->ia_size > inode->i_size) { - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(inode)->root; - u64 mask = root->blocksize - 1; - u64 pos = (inode->i_size + mask) & ~mask; - u64 hole_size; - - if (attr->ia_size <= pos) - goto out; - - btrfs_truncate_page(inode->i_mapping, inode->i_size); - - hole_size = (attr->ia_size - pos + mask) & ~mask; - hole_size >>= inode->i_blkbits; - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - err = btrfs_insert_file_extent(trans, root, inode->i_ino, - pos, 0, 0, hole_size); - BUG_ON(err); - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - } -out: - err = inode_setattr(inode, attr); - - return err; -} -static void btrfs_delete_inode(struct inode *inode) -{ - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(inode)->root; - int ret; - - truncate_inode_pages(&inode->i_data, 0); - if (is_bad_inode(inode)) { - goto no_delete; - } - inode->i_size = 0; - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - ret = btrfs_truncate_in_trans(trans, root, inode); - BUG_ON(ret); - btrfs_free_inode(trans, root, inode); - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - btrfs_btree_balance_dirty(root); - return; -no_delete: - clear_inode(inode); -} - -static int btrfs_inode_by_name(struct inode *dir, struct dentry *dentry, - struct btrfs_key *location) -{ - const char *name = dentry->d_name.name; - int namelen = dentry->d_name.len; - struct btrfs_dir_item *di; - struct btrfs_path *path; - struct btrfs_root *root = BTRFS_I(dir)->root; - int ret; - - path = btrfs_alloc_path(); - BUG_ON(!path); - btrfs_init_path(path); - di = btrfs_lookup_dir_item(NULL, root, path, dir->i_ino, name, - namelen, 0); - if (!di || IS_ERR(di)) { - location->objectid = 0; - ret = 0; - goto out; - } - btrfs_disk_key_to_cpu(location, &di->location); -out: - btrfs_release_path(root, path); - btrfs_free_path(path); - return ret; -} - -static int fixup_tree_root_location(struct btrfs_root *root, - struct btrfs_key *location, - struct btrfs_root **sub_root) -{ - struct btrfs_path *path; - struct btrfs_root_item *ri; - - if (btrfs_key_type(location) != BTRFS_ROOT_ITEM_KEY) - return 0; - if (location->objectid == BTRFS_ROOT_TREE_OBJECTID) - return 0; - - path = btrfs_alloc_path(); - BUG_ON(!path); - mutex_lock(&root->fs_info->fs_mutex); - - *sub_root = btrfs_read_fs_root(root->fs_info, location); - if (IS_ERR(*sub_root)) - return PTR_ERR(*sub_root); - - ri = &(*sub_root)->root_item; - location->objectid = btrfs_root_dirid(ri); - location->flags = 0; - btrfs_set_key_type(location, BTRFS_INODE_ITEM_KEY); - location->offset = 0; - - btrfs_free_path(path); - mutex_unlock(&root->fs_info->fs_mutex); - return 0; -} - -static int btrfs_init_locked_inode(struct inode *inode, void *p) -{ - struct btrfs_iget_args *args = p; - inode->i_ino = args->ino; - BTRFS_I(inode)->root = args->root; - return 0; -} - -static int btrfs_find_actor(struct inode *inode, void *opaque) -{ - struct btrfs_iget_args *args = opaque; - return (args->ino == inode->i_ino && - args->root == BTRFS_I(inode)->root); -} - -static struct inode *btrfs_iget_locked(struct super_block *s, u64 objectid, - struct btrfs_root *root) -{ - struct inode *inode; - struct btrfs_iget_args args; - args.ino = objectid; - args.root = root; - - inode = iget5_locked(s, objectid, btrfs_find_actor, - btrfs_init_locked_inode, - (void *)&args); - return inode; -} - -static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry, - struct nameidata *nd) -{ - struct inode * inode; - struct btrfs_inode *bi = BTRFS_I(dir); - struct btrfs_root *root = bi->root; - struct btrfs_root *sub_root = root; - struct btrfs_key location; - int ret; - - if (dentry->d_name.len > BTRFS_NAME_LEN) - return ERR_PTR(-ENAMETOOLONG); - mutex_lock(&root->fs_info->fs_mutex); - ret = btrfs_inode_by_name(dir, dentry, &location); - mutex_unlock(&root->fs_info->fs_mutex); - if (ret < 0) - return ERR_PTR(ret); - inode = NULL; - if (location.objectid) { - ret = fixup_tree_root_location(root, &location, &sub_root); - if (ret < 0) - return ERR_PTR(ret); - if (ret > 0) - return ERR_PTR(-ENOENT); - inode = btrfs_iget_locked(dir->i_sb, location.objectid, - sub_root); - if (!inode) - return ERR_PTR(-EACCES); - if (inode->i_state & I_NEW) { - if (sub_root != root) { -printk("adding new root for inode %lu root %p (found %p)\n", inode->i_ino, sub_root, BTRFS_I(inode)->root); - igrab(inode); - sub_root->inode = inode; - } - BTRFS_I(inode)->root = sub_root; - memcpy(&BTRFS_I(inode)->location, &location, - sizeof(location)); - btrfs_read_locked_inode(inode); - unlock_new_inode(inode); - } - } - return d_splice_alias(inode, dentry); -} - -static void reada_leaves(struct btrfs_root *root, struct btrfs_path *path, - u64 objectid) -{ - struct btrfs_node *node; - int i; - u32 nritems; - u64 item_objectid; - u64 blocknr; - int slot; - int ret; - - if (!path->nodes[1]) - return; - node = btrfs_buffer_node(path->nodes[1]); - slot = path->slots[1]; - nritems = btrfs_header_nritems(&node->header); - for (i = slot + 1; i < nritems; i++) { - item_objectid = btrfs_disk_key_objectid(&node->ptrs[i].key); - if (item_objectid != objectid) - break; - blocknr = btrfs_node_blockptr(node, i); - ret = readahead_tree_block(root, blocknr); - if (ret) - break; - } -} -static unsigned char btrfs_filetype_table[] = { - DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK -}; - -static int btrfs_readdir(struct file *filp, void *dirent, filldir_t filldir) -{ - struct inode *inode = filp->f_path.dentry->d_inode; - struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_item *item; - struct btrfs_dir_item *di; - struct btrfs_key key; - struct btrfs_path *path; - int ret; - u32 nritems; - struct btrfs_leaf *leaf; - int slot; - int advance; - unsigned char d_type; - int over = 0; - u32 di_cur; - u32 di_total; - u32 di_len; - int key_type = BTRFS_DIR_INDEX_KEY; - - /* FIXME, use a real flag for deciding about the key type */ - if (root->fs_info->tree_root == root) - key_type = BTRFS_DIR_ITEM_KEY; - mutex_lock(&root->fs_info->fs_mutex); - key.objectid = inode->i_ino; - key.flags = 0; - btrfs_set_key_type(&key, key_type); - key.offset = filp->f_pos; - path = btrfs_alloc_path(); - btrfs_init_path(path); - ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); - if (ret < 0) - goto err; - advance = 0; - reada_leaves(root, path, inode->i_ino); - while(1) { - leaf = btrfs_buffer_leaf(path->nodes[0]); - nritems = btrfs_header_nritems(&leaf->header); - slot = path->slots[0]; - if (advance || slot >= nritems) { - if (slot >= nritems -1) { - reada_leaves(root, path, inode->i_ino); - ret = btrfs_next_leaf(root, path); - if (ret) - break; - leaf = btrfs_buffer_leaf(path->nodes[0]); - nritems = btrfs_header_nritems(&leaf->header); - slot = path->slots[0]; - } else { - slot++; - path->slots[0]++; - } - } - advance = 1; - item = leaf->items + slot; - if (btrfs_disk_key_objectid(&item->key) != key.objectid) - break; - if (btrfs_disk_key_type(&item->key) != key_type) - break; - if (btrfs_disk_key_offset(&item->key) < filp->f_pos) - continue; - filp->f_pos = btrfs_disk_key_offset(&item->key); - advance = 1; - di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item); - di_cur = 0; - di_total = btrfs_item_size(leaf->items + slot); - while(di_cur < di_total) { - d_type = btrfs_filetype_table[btrfs_dir_type(di)]; - over = filldir(dirent, (const char *)(di + 1), - btrfs_dir_name_len(di), - btrfs_disk_key_offset(&item->key), - btrfs_disk_key_objectid(&di->location), - d_type); - if (over) - goto nopos; - di_len = btrfs_dir_name_len(di) + sizeof(*di); - di_cur += di_len; - di = (struct btrfs_dir_item *)((char *)di + di_len); - } - } - filp->f_pos++; -nopos: - ret = 0; -err: - btrfs_release_path(root, path); - btrfs_free_path(path); - mutex_unlock(&root->fs_info->fs_mutex); - return ret; -} static void btrfs_put_super (struct super_block * sb) { @@ -907,6 +43,7 @@ static int btrfs_fill_super(struct super_block * sb, void * data, int silent) struct btrfs_super_block *disk_super; struct btrfs_root *tree_root; struct btrfs_inode *bi; + int err; sb->s_maxbytes = MAX_LFS_FILESIZE; sb->s_magic = BTRFS_SUPER_MAGIC; @@ -915,16 +52,12 @@ static int btrfs_fill_super(struct super_block * sb, void * data, int silent) tree_root = open_ctree(sb); - if (!tree_root) { + if (!tree_root || IS_ERR(tree_root)) { printk("btrfs: open_ctree failed\n"); return -EIO; } sb->s_fs_info = tree_root; disk_super = tree_root->fs_info->disk_super; - printk("read in super total blocks %Lu root %Lu\n", - btrfs_super_total_blocks(disk_super), - btrfs_super_root_dir(disk_super)); - inode = btrfs_iget_locked(sb, btrfs_super_root_dir(disk_super), tree_root); bi = BTRFS_I(inode); @@ -934,8 +67,10 @@ static int btrfs_fill_super(struct super_block * sb, void * data, int silent) bi->root = tree_root; btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY); - if (!inode) - return -ENOMEM; + if (!inode) { + err = -ENOMEM; + goto fail_close; + } if (inode->i_state & I_NEW) { btrfs_read_locked_inode(inode); unlock_new_inode(inode); @@ -944,323 +79,18 @@ static int btrfs_fill_super(struct super_block * sb, void * data, int silent) root_dentry = d_alloc_root(inode); if (!root_dentry) { iput(inode); - return -ENOMEM; + err = -ENOMEM; + goto fail_close; } sb->s_root = root_dentry; btrfs_transaction_queue_work(tree_root, HZ * 30); return 0; -} -static int btrfs_write_inode(struct inode *inode, int wait) -{ - struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_trans_handle *trans; - int ret = 0; - - if (wait) { - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - ret = btrfs_commit_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - } - return ret; -} - -static void btrfs_dirty_inode(struct inode *inode) -{ - struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_trans_handle *trans; - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - btrfs_update_inode(trans, root, inode); - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - btrfs_btree_balance_dirty(root); -} - -static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - u64 objectid, - struct btrfs_block_group_cache *group, - int mode) -{ - struct inode *inode; - struct btrfs_inode_item inode_item; - struct btrfs_key *location; - int ret; - int owner; - - inode = new_inode(root->fs_info->sb); - if (!inode) - return ERR_PTR(-ENOMEM); - - BTRFS_I(inode)->root = root; - if (mode & S_IFDIR) - owner = 0; - else - owner = 1; - group = btrfs_find_block_group(root, group, 0, 0, owner); - BTRFS_I(inode)->block_group = group; - - inode->i_uid = current->fsuid; - inode->i_gid = current->fsgid; - inode->i_mode = mode; - inode->i_ino = objectid; - inode->i_blocks = 0; - inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME; - fill_inode_item(&inode_item, inode); - location = &BTRFS_I(inode)->location; - location->objectid = objectid; - location->flags = 0; - location->offset = 0; - btrfs_set_key_type(location, BTRFS_INODE_ITEM_KEY); - - ret = btrfs_insert_inode(trans, root, objectid, &inode_item); - BUG_ON(ret); - - insert_inode_hash(inode); - return inode; -} - -static inline u8 btrfs_inode_type(struct inode *inode) -{ - return btrfs_type_by_mode[(inode->i_mode & S_IFMT) >> S_SHIFT]; -} - -static int btrfs_add_link(struct btrfs_trans_handle *trans, - struct dentry *dentry, struct inode *inode) -{ - int ret; - struct btrfs_key key; - struct btrfs_root *root = BTRFS_I(dentry->d_parent->d_inode)->root; - key.objectid = inode->i_ino; - key.flags = 0; - btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); - key.offset = 0; - - ret = btrfs_insert_dir_item(trans, root, - dentry->d_name.name, dentry->d_name.len, - dentry->d_parent->d_inode->i_ino, - &key, btrfs_inode_type(inode)); - if (ret == 0) { - dentry->d_parent->d_inode->i_size += dentry->d_name.len * 2; - ret = btrfs_update_inode(trans, root, - dentry->d_parent->d_inode); - } - return ret; -} - -static int btrfs_add_nondir(struct btrfs_trans_handle *trans, - struct dentry *dentry, struct inode *inode) -{ - int err = btrfs_add_link(trans, dentry, inode); - if (!err) { - d_instantiate(dentry, inode); - return 0; - } - if (err > 0) - err = -EEXIST; +fail_close: + close_ctree(tree_root); return err; } -static int btrfs_create(struct inode *dir, struct dentry *dentry, - int mode, struct nameidata *nd) -{ - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(dir)->root; - struct inode *inode; - int err; - int drop_inode = 0; - u64 objectid; - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, dir); - - err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); - if (err) { - err = -ENOSPC; - goto out_unlock; - } - - inode = btrfs_new_inode(trans, root, objectid, - BTRFS_I(dir)->block_group, mode); - err = PTR_ERR(inode); - if (IS_ERR(inode)) - goto out_unlock; - - btrfs_set_trans_block_group(trans, inode); - err = btrfs_add_nondir(trans, dentry, inode); - if (err) - drop_inode = 1; - else { - inode->i_mapping->a_ops = &btrfs_aops; - inode->i_fop = &btrfs_file_operations; - inode->i_op = &btrfs_file_inode_operations; - } - dir->i_sb->s_dirt = 1; - btrfs_update_inode_block_group(trans, inode); - btrfs_update_inode_block_group(trans, dir); -out_unlock: - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - - if (drop_inode) { - inode_dec_link_count(inode); - iput(inode); - } - btrfs_btree_balance_dirty(root); - return err; -} - -static int btrfs_link(struct dentry *old_dentry, struct inode *dir, - struct dentry *dentry) -{ - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(dir)->root; - struct inode *inode = old_dentry->d_inode; - int err; - int drop_inode = 0; - - if (inode->i_nlink == 0) - return -ENOENT; - - inc_nlink(inode); - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, dir); - atomic_inc(&inode->i_count); - err = btrfs_add_nondir(trans, dentry, inode); - if (err) - drop_inode = 1; - dir->i_sb->s_dirt = 1; - btrfs_update_inode_block_group(trans, dir); - btrfs_update_inode(trans, root, inode); - - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - - if (drop_inode) { - inode_dec_link_count(inode); - iput(inode); - } - btrfs_btree_balance_dirty(root); - return err; -} - -static int btrfs_make_empty_dir(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - u64 objectid, u64 dirid) -{ - int ret; - char buf[2]; - struct btrfs_key key; - - buf[0] = '.'; - buf[1] = '.'; - - key.objectid = objectid; - key.offset = 0; - key.flags = 0; - btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); - - ret = btrfs_insert_dir_item(trans, root, buf, 1, objectid, - &key, BTRFS_FT_DIR); - if (ret) - goto error; - key.objectid = dirid; - ret = btrfs_insert_dir_item(trans, root, buf, 2, objectid, - &key, BTRFS_FT_DIR); - if (ret) - goto error; -error: - return ret; -} - -static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) -{ - struct inode *inode; - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(dir)->root; - int err = 0; - int drop_on_err = 0; - u64 objectid; - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, dir); - if (IS_ERR(trans)) { - err = PTR_ERR(trans); - goto out_unlock; - } - - err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); - if (err) { - err = -ENOSPC; - goto out_unlock; - } - - inode = btrfs_new_inode(trans, root, objectid, - BTRFS_I(dir)->block_group, S_IFDIR | mode); - if (IS_ERR(inode)) { - err = PTR_ERR(inode); - goto out_fail; - } - drop_on_err = 1; - inode->i_op = &btrfs_dir_inode_operations; - inode->i_fop = &btrfs_dir_file_operations; - btrfs_set_trans_block_group(trans, inode); - - err = btrfs_make_empty_dir(trans, root, inode->i_ino, dir->i_ino); - if (err) - goto out_fail; - - inode->i_size = 6; - err = btrfs_update_inode(trans, root, inode); - if (err) - goto out_fail; - err = btrfs_add_link(trans, dentry, inode); - if (err) - goto out_fail; - d_instantiate(dentry, inode); - drop_on_err = 0; - dir->i_sb->s_dirt = 1; - btrfs_update_inode_block_group(trans, inode); - btrfs_update_inode_block_group(trans, dir); - -out_fail: - btrfs_end_transaction(trans, root); -out_unlock: - mutex_unlock(&root->fs_info->fs_mutex); - if (drop_on_err) - iput(inode); - btrfs_btree_balance_dirty(root); - return err; -} - -static int btrfs_sync_file(struct file *file, - struct dentry *dentry, int datasync) -{ - struct inode *inode = dentry->d_inode; - struct btrfs_root *root = BTRFS_I(inode)->root; - int ret; - struct btrfs_trans_handle *trans; - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - if (!trans) { - ret = -ENOMEM; - goto out; - } - ret = btrfs_commit_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); -out: - return ret > 0 ? EIO : ret; -} - static int btrfs_sync_fs(struct super_block *sb, int wait) { struct btrfs_trans_handle *trans; @@ -1278,1430 +108,15 @@ static int btrfs_sync_fs(struct super_block *sb, int wait) ret = btrfs_commit_transaction(trans, root); sb->s_dirt = 0; BUG_ON(ret); -printk("btrfs sync_fs\n"); mutex_unlock(&root->fs_info->fs_mutex); return 0; } -#define BTRFS_GET_BLOCK_NO_CREATE 0 -#define BTRFS_GET_BLOCK_CREATE 1 -#define BTRFS_GET_BLOCK_NO_DIRECT 2 - -static int btrfs_get_block_lock(struct inode *inode, sector_t iblock, - struct buffer_head *result, int create) -{ - int ret; - int err = 0; - u64 blocknr; - u64 extent_start = 0; - u64 extent_end = 0; - u64 objectid = inode->i_ino; - u32 found_type; - u64 alloc_hint = 0; - struct btrfs_path *path; - struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_file_extent_item *item; - struct btrfs_leaf *leaf; - struct btrfs_disk_key *found_key; - struct btrfs_trans_handle *trans = NULL; - - path = btrfs_alloc_path(); - BUG_ON(!path); - btrfs_init_path(path); - if (create & BTRFS_GET_BLOCK_CREATE) { - WARN_ON(1); - /* this almost but not quite works */ - trans = btrfs_start_transaction(root, 1); - if (!trans) { - err = -ENOMEM; - goto out; - } - ret = drop_extents(trans, root, inode, - iblock << inode->i_blkbits, - (iblock + 1) << inode->i_blkbits, - &alloc_hint); - BUG_ON(ret); - } - - ret = btrfs_lookup_file_extent(NULL, root, path, - inode->i_ino, - iblock << inode->i_blkbits, 0); - if (ret < 0) { - err = ret; - goto out; - } - - if (ret != 0) { - if (path->slots[0] == 0) { - btrfs_release_path(root, path); - goto not_found; - } - path->slots[0]--; - } - - item = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), path->slots[0], - struct btrfs_file_extent_item); - leaf = btrfs_buffer_leaf(path->nodes[0]); - blocknr = btrfs_file_extent_disk_blocknr(item); - blocknr += btrfs_file_extent_offset(item); - - /* are we inside the extent that was found? */ - found_key = &leaf->items[path->slots[0]].key; - found_type = btrfs_disk_key_type(found_key); - if (btrfs_disk_key_objectid(found_key) != objectid || - found_type != BTRFS_EXTENT_DATA_KEY) { - extent_end = 0; - extent_start = 0; - goto not_found; - } - found_type = btrfs_file_extent_type(item); - extent_start = btrfs_disk_key_offset(&leaf->items[path->slots[0]].key); - if (found_type == BTRFS_FILE_EXTENT_REG) { - extent_start = extent_start >> inode->i_blkbits; - extent_end = extent_start + btrfs_file_extent_num_blocks(item); - err = 0; - if (btrfs_file_extent_disk_blocknr(item) == 0) - goto out; - if (iblock >= extent_start && iblock < extent_end) { - btrfs_map_bh_to_logical(root, result, blocknr + - iblock - extent_start); - goto out; - } - } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { - char *ptr; - char *map; - u32 size; - - if (create & BTRFS_GET_BLOCK_NO_DIRECT) { - err = -EINVAL; - goto out; - } - size = btrfs_file_extent_inline_len(leaf->items + - path->slots[0]); - extent_end = (extent_start + size) >> inode->i_blkbits; - extent_start >>= inode->i_blkbits; - if (iblock < extent_start || iblock > extent_end) { - goto not_found; - } - ptr = btrfs_file_extent_inline_start(item); - map = kmap(result->b_page); - memcpy(map, ptr, size); - memset(map + size, 0, PAGE_CACHE_SIZE - size); - flush_dcache_page(result->b_page); - kunmap(result->b_page); - set_buffer_uptodate(result); - SetPageChecked(result->b_page); - btrfs_map_bh_to_logical(root, result, 0); - } -not_found: - if (create & BTRFS_GET_BLOCK_CREATE) { - struct btrfs_key ins; - ret = btrfs_alloc_extent(trans, root, inode->i_ino, - 1, alloc_hint, (u64)-1, - &ins, 1); - BUG_ON(ret); - ret = btrfs_insert_file_extent(trans, root, inode->i_ino, - iblock << inode->i_blkbits, - ins.objectid, ins.offset, - ins.offset); - BUG_ON(ret); - SetPageChecked(result->b_page); - btrfs_map_bh_to_logical(root, result, ins.objectid); - } -out: - if (trans) - err = btrfs_end_transaction(trans, root); - btrfs_free_path(path); - return err; -} - -static int btrfs_get_block(struct inode *inode, sector_t iblock, - struct buffer_head *result, int create) -{ - int err; - struct btrfs_root *root = BTRFS_I(inode)->root; - mutex_lock(&root->fs_info->fs_mutex); - err = btrfs_get_block_lock(inode, iblock, result, create); - mutex_unlock(&root->fs_info->fs_mutex); - return err; -} - -static int btrfs_get_block_bmap(struct inode *inode, sector_t iblock, - struct buffer_head *result, int create) -{ - struct btrfs_root *root = BTRFS_I(inode)->root; - mutex_lock(&root->fs_info->fs_mutex); - btrfs_get_block_lock(inode, iblock, result, BTRFS_GET_BLOCK_NO_DIRECT); - mutex_unlock(&root->fs_info->fs_mutex); - return 0; -} - -static sector_t btrfs_bmap(struct address_space *as, sector_t block) -{ - return generic_block_bmap(as, block, btrfs_get_block_bmap); -} - -static int btrfs_prepare_write(struct file *file, struct page *page, - unsigned from, unsigned to) -{ - return block_prepare_write(page, from, to, btrfs_get_block); -} - static void btrfs_write_super(struct super_block *sb) { sb->s_dirt = 0; } -static int btrfs_readpage(struct file *file, struct page *page) -{ - return mpage_readpage(page, btrfs_get_block); -} - -/* - * While block_write_full_page is writing back the dirty buffers under - * the page lock, whoever dirtied the buffers may decide to clean them - * again at any time. We handle that by only looking at the buffer - * state inside lock_buffer(). - * - * If block_write_full_page() is called for regular writeback - * (wbc->sync_mode == WB_SYNC_NONE) then it will redirty a page which has a - * locked buffer. This only can happen if someone has written the buffer - * directly, with submit_bh(). At the address_space level PageWriteback - * prevents this contention from occurring. - */ -static int __btrfs_write_full_page(struct inode *inode, struct page *page, - struct writeback_control *wbc) -{ - int err; - sector_t block; - sector_t last_block; - struct buffer_head *bh, *head; - const unsigned blocksize = 1 << inode->i_blkbits; - int nr_underway = 0; - - BUG_ON(!PageLocked(page)); - - last_block = (i_size_read(inode) - 1) >> inode->i_blkbits; - - if (!page_has_buffers(page)) { - create_empty_buffers(page, blocksize, - (1 << BH_Dirty)|(1 << BH_Uptodate)); - } - - /* - * 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. - */ - - block = (sector_t)page->index << (PAGE_CACHE_SHIFT - inode->i_blkbits); - head = page_buffers(page); - bh = head; - - /* - * Get all the dirty buffers mapped to disk addresses and - * handle any aliases from the underlying blockdev's mapping. - */ - do { - if (block > last_block) { - /* - * mapped buffers outside i_size will occur, because - * this page can be outside i_size when there is a - * truncate in progress. - */ - /* - * The buffer was zeroed by block_write_full_page() - */ - clear_buffer_dirty(bh); - set_buffer_uptodate(bh); - } else if (!buffer_mapped(bh) && buffer_dirty(bh)) { - WARN_ON(bh->b_size != blocksize); - err = btrfs_get_block(inode, block, bh, 0); - if (err) { -printk("writepage going to recovery err %d\n", err); - goto recover; - } - if (buffer_new(bh)) { - /* blockdev mappings never come here */ - clear_buffer_new(bh); - } - } - bh = bh->b_this_page; - block++; - } while (bh != head); - - do { - if (!buffer_mapped(bh)) - continue; - /* - * If it's a fully non-blocking write attempt and we cannot - * lock the buffer then redirty the page. Note that this can - * potentially cause a busy-wait loop from pdflush and kswapd - * activity, but those code paths have their own higher-level - * throttling. - */ - if (wbc->sync_mode != WB_SYNC_NONE || !wbc->nonblocking) { - lock_buffer(bh); - } else if (test_set_buffer_locked(bh)) { - redirty_page_for_writepage(wbc, page); - continue; - } - if (test_clear_buffer_dirty(bh) && bh->b_blocknr != 0) { - mark_buffer_async_write(bh); - } else { - unlock_buffer(bh); - } - } while ((bh = bh->b_this_page) != head); - - /* - * The page and its buffers are protected by PageWriteback(), so we can - * drop the bh refcounts early. - */ - BUG_ON(PageWriteback(page)); - set_page_writeback(page); - - do { - struct buffer_head *next = bh->b_this_page; - if (buffer_async_write(bh)) { - submit_bh(WRITE, bh); - nr_underway++; - } - bh = next; - } while (bh != head); - unlock_page(page); - - err = 0; -done: - if (nr_underway == 0) { - /* - * The page was marked dirty, but the buffers were - * clean. Someone wrote them back by hand with - * ll_rw_block/submit_bh. A rare case. - */ - int uptodate = 1; - do { - if (!buffer_uptodate(bh)) { - uptodate = 0; - break; - } - bh = bh->b_this_page; - } while (bh != head); - if (uptodate) - SetPageUptodate(page); - end_page_writeback(page); - } - return err; - -recover: - /* - * ENOSPC, or some other error. We may already have added some - * blocks to the file, so we need to write these out to avoid - * exposing stale data. - * The page is currently locked and not marked for writeback - */ - bh = head; - /* Recovery: lock and submit the mapped buffers */ - do { - if (buffer_mapped(bh) && buffer_dirty(bh)) { - lock_buffer(bh); - mark_buffer_async_write(bh); - } else { - /* - * The buffer may have been set dirty during - * attachment to a dirty page. - */ - clear_buffer_dirty(bh); - } - } while ((bh = bh->b_this_page) != head); - SetPageError(page); - BUG_ON(PageWriteback(page)); - set_page_writeback(page); - do { - struct buffer_head *next = bh->b_this_page; - if (buffer_async_write(bh)) { - clear_buffer_dirty(bh); - submit_bh(WRITE, bh); - nr_underway++; - } - bh = next; - } while (bh != head); - unlock_page(page); - goto done; -} - -/* - * The generic ->writepage function for buffer-backed address_spaces - */ -static int btrfs_writepage(struct page *page, struct writeback_control *wbc) -{ - struct inode * const inode = page->mapping->host; - loff_t i_size = i_size_read(inode); - const pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT; - unsigned offset; - void *kaddr; - - /* Is the page fully inside i_size? */ - if (page->index < end_index) - return __btrfs_write_full_page(inode, page, wbc); - - /* Is the page fully outside i_size? (truncate in progress) */ - offset = i_size & (PAGE_CACHE_SIZE-1); - if (page->index >= end_index+1 || !offset) { - /* - * The page may have dirty, unmapped buffers. For example, - * they may have been added in ext3_writepage(). Make them - * freeable here, so the page does not leak. - */ - block_invalidatepage(page, 0); - unlock_page(page); - return 0; /* don't care */ - } - - /* - * The page straddles i_size. It must be zeroed out on each and every - * writepage invokation because it may be mmapped. "A file is mapped - * in multiples of the page size. For a file that is not a multiple of - * the page size, the remaining memory is zeroed when mapped, and - * writes to that region are not written out to the file." - */ - kaddr = kmap_atomic(page, KM_USER0); - memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset); - flush_dcache_page(page); - kunmap_atomic(kaddr, KM_USER0); - return __btrfs_write_full_page(inode, page, wbc); -} - -static void btrfs_truncate(struct inode *inode) -{ - struct btrfs_root *root = BTRFS_I(inode)->root; - int ret; - struct btrfs_trans_handle *trans; - - if (!S_ISREG(inode->i_mode)) - return; - if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) - return; - - btrfs_truncate_page(inode->i_mapping, inode->i_size); - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - - /* FIXME, add redo link to tree so we don't leak on crash */ - ret = btrfs_truncate_in_trans(trans, root, inode); - BUG_ON(ret); - btrfs_update_inode(trans, root, inode); - ret = btrfs_end_transaction(trans, root); - BUG_ON(ret); - mutex_unlock(&root->fs_info->fs_mutex); - btrfs_btree_balance_dirty(root); -} - -static int btrfs_commit_write(struct file *file, struct page *page, - unsigned from, unsigned to) -{ - struct inode *inode = page->mapping->host; - struct buffer_head *bh; - loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to; - - SetPageUptodate(page); - bh = page_buffers(page); - set_buffer_uptodate(bh); - if (buffer_mapped(bh) && bh->b_blocknr != 0) { - set_page_dirty(page); - } - if (pos > inode->i_size) { - i_size_write(inode, pos); - mark_inode_dirty(inode); - } - return 0; -} - -static int btrfs_copy_from_user(loff_t pos, int num_pages, int write_bytes, - struct page **prepared_pages, - const char __user * buf) -{ - long page_fault = 0; - int i; - int offset = pos & (PAGE_CACHE_SIZE - 1); - - for (i = 0; i < num_pages && write_bytes > 0; i++, offset = 0) { - size_t count = min_t(size_t, - PAGE_CACHE_SIZE - offset, write_bytes); - struct page *page = prepared_pages[i]; - fault_in_pages_readable(buf, count); - - /* Copy data from userspace to the current page */ - kmap(page); - page_fault = __copy_from_user(page_address(page) + offset, - buf, count); - /* Flush processor's dcache for this page */ - flush_dcache_page(page); - kunmap(page); - buf += count; - write_bytes -= count; - - if (page_fault) - break; - } - return page_fault ? -EFAULT : 0; -} - -static void btrfs_drop_pages(struct page **pages, size_t num_pages) -{ - size_t i; - for (i = 0; i < num_pages; i++) { - if (!pages[i]) - break; - unlock_page(pages[i]); - mark_page_accessed(pages[i]); - page_cache_release(pages[i]); - } -} -static int dirty_and_release_pages(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct file *file, - struct page **pages, - size_t num_pages, - loff_t pos, - size_t write_bytes) -{ - int i; - int offset; - int err = 0; - int ret; - int this_write; - struct inode *inode = file->f_path.dentry->d_inode; - struct buffer_head *bh; - struct btrfs_file_extent_item *ei; - - for (i = 0; i < num_pages; i++) { - offset = pos & (PAGE_CACHE_SIZE -1); - this_write = min(PAGE_CACHE_SIZE - offset, write_bytes); - /* FIXME, one block at a time */ - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - - bh = page_buffers(pages[i]); - if (buffer_mapped(bh) && bh->b_blocknr == 0) { - struct btrfs_key key; - struct btrfs_path *path; - char *ptr; - u32 datasize; - - path = btrfs_alloc_path(); - BUG_ON(!path); - key.objectid = inode->i_ino; - key.offset = pages[i]->index << PAGE_CACHE_SHIFT; - key.flags = 0; - btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY); - BUG_ON(write_bytes >= PAGE_CACHE_SIZE); - datasize = offset + - btrfs_file_extent_calc_inline_size(write_bytes); - ret = btrfs_insert_empty_item(trans, root, path, &key, - datasize); - BUG_ON(ret); - ei = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), - path->slots[0], struct btrfs_file_extent_item); - btrfs_set_file_extent_generation(ei, trans->transid); - btrfs_set_file_extent_type(ei, - BTRFS_FILE_EXTENT_INLINE); - ptr = btrfs_file_extent_inline_start(ei); - btrfs_memcpy(root, path->nodes[0]->b_data, - ptr, bh->b_data, offset + write_bytes); - mark_buffer_dirty(path->nodes[0]); - btrfs_free_path(path); - } else if (buffer_mapped(bh)) { - btrfs_csum_file_block(trans, root, inode->i_ino, - pages[i]->index << PAGE_CACHE_SHIFT, - kmap(pages[i]), PAGE_CACHE_SIZE); - kunmap(pages[i]); - } - SetPageChecked(pages[i]); - // btrfs_update_inode_block_group(trans, inode); - ret = btrfs_end_transaction(trans, root); - BUG_ON(ret); - mutex_unlock(&root->fs_info->fs_mutex); - - ret = btrfs_commit_write(file, pages[i], offset, - offset + this_write); - pos += this_write; - if (ret) { - err = ret; - goto failed; - } - WARN_ON(this_write > write_bytes); - write_bytes -= this_write; - } -failed: - return err; -} - -static int drop_extents(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode, - u64 start, u64 end, u64 *hint_block) -{ - int ret; - struct btrfs_key key; - struct btrfs_leaf *leaf; - int slot; - struct btrfs_file_extent_item *extent; - u64 extent_end = 0; - int keep; - struct btrfs_file_extent_item old; - struct btrfs_path *path; - u64 search_start = start; - int bookend; - int found_type; - int found_extent; - int found_inline; - - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; - while(1) { - btrfs_release_path(root, path); - ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino, - search_start, -1); - if (ret < 0) - goto out; - if (ret > 0) { - if (path->slots[0] == 0) { - ret = 0; - goto out; - } - path->slots[0]--; - } - keep = 0; - bookend = 0; - found_extent = 0; - found_inline = 0; - extent = NULL; - leaf = btrfs_buffer_leaf(path->nodes[0]); - slot = path->slots[0]; - btrfs_disk_key_to_cpu(&key, &leaf->items[slot].key); - if (key.offset >= end || key.objectid != inode->i_ino) { - ret = 0; - goto out; - } - if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY) { - ret = 0; - goto out; - } - extent = btrfs_item_ptr(leaf, slot, - struct btrfs_file_extent_item); - found_type = btrfs_file_extent_type(extent); - if (found_type == BTRFS_FILE_EXTENT_REG) { - extent_end = key.offset + - (btrfs_file_extent_num_blocks(extent) << - inode->i_blkbits); - found_extent = 1; - } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { - found_inline = 1; - extent_end = key.offset + - btrfs_file_extent_inline_len(leaf->items + slot); - } - - if (!found_extent && !found_inline) { - ret = 0; - goto out; - } - - if (search_start >= extent_end) { - ret = 0; - goto out; - } - - if (found_inline) { - u64 mask = root->blocksize - 1; - search_start = (extent_end + mask) & ~mask; - } else - search_start = extent_end; - - if (end < extent_end && end >= key.offset) { - if (found_extent) { - u64 disk_blocknr = - btrfs_file_extent_disk_blocknr(extent); - u64 disk_num_blocks = - btrfs_file_extent_disk_num_blocks(extent); - memcpy(&old, extent, sizeof(old)); - if (disk_blocknr != 0) { - ret = btrfs_inc_extent_ref(trans, root, - disk_blocknr, disk_num_blocks); - BUG_ON(ret); - } - } - WARN_ON(found_inline); - bookend = 1; - } - - if (start > key.offset) { - u64 new_num; - u64 old_num; - /* truncate existing extent */ - keep = 1; - WARN_ON(start & (root->blocksize - 1)); - if (found_extent) { - new_num = (start - key.offset) >> - inode->i_blkbits; - old_num = btrfs_file_extent_num_blocks(extent); - *hint_block = - btrfs_file_extent_disk_blocknr(extent); - if (btrfs_file_extent_disk_blocknr(extent)) { - inode->i_blocks -= - (old_num - new_num) << 3; - } - btrfs_set_file_extent_num_blocks(extent, - new_num); - mark_buffer_dirty(path->nodes[0]); - } else { - WARN_ON(1); - } - } - if (!keep) { - u64 disk_blocknr = 0; - u64 disk_num_blocks = 0; - u64 extent_num_blocks = 0; - if (found_extent) { - disk_blocknr = - btrfs_file_extent_disk_blocknr(extent); - disk_num_blocks = - btrfs_file_extent_disk_num_blocks(extent); - extent_num_blocks = - btrfs_file_extent_num_blocks(extent); - *hint_block = - btrfs_file_extent_disk_blocknr(extent); - } - ret = btrfs_del_item(trans, root, path); - BUG_ON(ret); - btrfs_release_path(root, path); - extent = NULL; - if (found_extent && disk_blocknr != 0) { - inode->i_blocks -= extent_num_blocks << 3; - ret = btrfs_free_extent(trans, root, - disk_blocknr, - disk_num_blocks, 0); - } - - BUG_ON(ret); - if (!bookend && search_start >= end) { - ret = 0; - goto out; - } - if (!bookend) - continue; - } - if (bookend && found_extent) { - /* create bookend */ - struct btrfs_key ins; - ins.objectid = inode->i_ino; - ins.offset = end; - ins.flags = 0; - btrfs_set_key_type(&ins, BTRFS_EXTENT_DATA_KEY); - - btrfs_release_path(root, path); - ret = btrfs_insert_empty_item(trans, root, path, &ins, - sizeof(*extent)); - BUG_ON(ret); - extent = btrfs_item_ptr( - btrfs_buffer_leaf(path->nodes[0]), - path->slots[0], - struct btrfs_file_extent_item); - btrfs_set_file_extent_disk_blocknr(extent, - btrfs_file_extent_disk_blocknr(&old)); - btrfs_set_file_extent_disk_num_blocks(extent, - btrfs_file_extent_disk_num_blocks(&old)); - - btrfs_set_file_extent_offset(extent, - btrfs_file_extent_offset(&old) + - ((end - key.offset) >> inode->i_blkbits)); - WARN_ON(btrfs_file_extent_num_blocks(&old) < - (extent_end - end) >> inode->i_blkbits); - btrfs_set_file_extent_num_blocks(extent, - (extent_end - end) >> inode->i_blkbits); - - btrfs_set_file_extent_type(extent, - BTRFS_FILE_EXTENT_REG); - btrfs_set_file_extent_generation(extent, - btrfs_file_extent_generation(&old)); - btrfs_mark_buffer_dirty(path->nodes[0]); - if (btrfs_file_extent_disk_blocknr(&old) != 0) { - inode->i_blocks += - btrfs_file_extent_num_blocks(extent) << 3; - } - ret = 0; - goto out; - } - } -out: - btrfs_free_path(path); - return ret; -} - -static int prepare_pages(struct btrfs_root *root, - struct file *file, - struct page **pages, - size_t num_pages, - loff_t pos, - unsigned long first_index, - unsigned long last_index, - size_t write_bytes, - u64 alloc_extent_start) -{ - int i; - unsigned long index = pos >> PAGE_CACHE_SHIFT; - struct inode *inode = file->f_path.dentry->d_inode; - int offset; - int err = 0; - int this_write; - struct buffer_head *bh; - struct buffer_head *head; - loff_t isize = i_size_read(inode); - - memset(pages, 0, num_pages * sizeof(struct page *)); - - for (i = 0; i < num_pages; i++) { - pages[i] = grab_cache_page(inode->i_mapping, index + i); - if (!pages[i]) { - err = -ENOMEM; - goto failed_release; - } - cancel_dirty_page(pages[i], PAGE_CACHE_SIZE); - wait_on_page_writeback(pages[i]); - offset = pos & (PAGE_CACHE_SIZE -1); - this_write = min(PAGE_CACHE_SIZE - offset, write_bytes); - if (!page_has_buffers(pages[i])) { - create_empty_buffers(pages[i], - root->fs_info->sb->s_blocksize, - (1 << BH_Uptodate)); - } - head = page_buffers(pages[i]); - bh = head; - do { - err = btrfs_map_bh_to_logical(root, bh, - alloc_extent_start); - BUG_ON(err); - if (err) - goto failed_truncate; - bh = bh->b_this_page; - if (alloc_extent_start) - alloc_extent_start++; - } while (bh != head); - pos += this_write; - WARN_ON(this_write > write_bytes); - write_bytes -= this_write; - } - return 0; - -failed_release: - btrfs_drop_pages(pages, num_pages); - return err; - -failed_truncate: - btrfs_drop_pages(pages, num_pages); - if (pos > isize) - vmtruncate(inode, isize); - return err; -} - -static ssize_t btrfs_file_write(struct file *file, const char __user *buf, - size_t count, loff_t *ppos) -{ - loff_t pos; - size_t num_written = 0; - int err = 0; - int ret = 0; - struct inode *inode = file->f_path.dentry->d_inode; - struct btrfs_root *root = BTRFS_I(inode)->root; - struct page *pages[8]; - struct page *pinned[2]; - unsigned long first_index; - unsigned long last_index; - u64 start_pos; - u64 num_blocks; - u64 alloc_extent_start; - u64 hint_block; - struct btrfs_trans_handle *trans; - struct btrfs_key ins; - pinned[0] = NULL; - pinned[1] = NULL; - if (file->f_flags & O_DIRECT) - return -EINVAL; - pos = *ppos; - vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE); - current->backing_dev_info = inode->i_mapping->backing_dev_info; - err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode)); - if (err) - goto out; - if (count == 0) - goto out; - err = remove_suid(file->f_path.dentry); - if (err) - goto out; - file_update_time(file); - - start_pos = pos & ~((u64)PAGE_CACHE_SIZE - 1); - num_blocks = (count + pos - start_pos + root->blocksize - 1) >> - inode->i_blkbits; - - mutex_lock(&inode->i_mutex); - first_index = pos >> PAGE_CACHE_SHIFT; - last_index = (pos + count) >> PAGE_CACHE_SHIFT; - - if ((pos & (PAGE_CACHE_SIZE - 1))) { - pinned[0] = grab_cache_page(inode->i_mapping, first_index); - if (!PageUptodate(pinned[0])) { - ret = mpage_readpage(pinned[0], btrfs_get_block); - BUG_ON(ret); - wait_on_page_locked(pinned[0]); - } else { - unlock_page(pinned[0]); - } - } - if ((pos + count) & (PAGE_CACHE_SIZE - 1)) { - pinned[1] = grab_cache_page(inode->i_mapping, last_index); - if (!PageUptodate(pinned[1])) { - ret = mpage_readpage(pinned[1], btrfs_get_block); - BUG_ON(ret); - wait_on_page_locked(pinned[1]); - } else { - unlock_page(pinned[1]); - } - } - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - if (!trans) { - err = -ENOMEM; - mutex_unlock(&root->fs_info->fs_mutex); - goto out_unlock; - } - btrfs_set_trans_block_group(trans, inode); - /* FIXME blocksize != 4096 */ - inode->i_blocks += num_blocks << 3; - hint_block = 0; - if (start_pos < inode->i_size) { - /* FIXME blocksize != pagesize */ - ret = drop_extents(trans, root, inode, - start_pos, - (pos + count + root->blocksize -1) & - ~((u64)root->blocksize - 1), &hint_block); - BUG_ON(ret); - } - if (inode->i_size < start_pos) { - u64 last_pos_in_file; - u64 hole_size; - u64 mask = root->blocksize - 1; - last_pos_in_file = (inode->i_size + mask) & ~mask; - hole_size = (start_pos - last_pos_in_file + mask) & ~mask; - hole_size >>= inode->i_blkbits; - if (last_pos_in_file < start_pos) { - ret = btrfs_insert_file_extent(trans, root, - inode->i_ino, - last_pos_in_file, - 0, 0, hole_size); - } - BUG_ON(ret); - } - if (inode->i_size >= PAGE_CACHE_SIZE || pos + count < inode->i_size || - pos + count - start_pos > BTRFS_MAX_INLINE_DATA_SIZE(root)) { - ret = btrfs_alloc_extent(trans, root, inode->i_ino, - num_blocks, hint_block, (u64)-1, - &ins, 1); - BUG_ON(ret); - ret = btrfs_insert_file_extent(trans, root, inode->i_ino, - start_pos, ins.objectid, ins.offset, - ins.offset); - BUG_ON(ret); - } else { - ins.offset = 0; - ins.objectid = 0; - } - BUG_ON(ret); - alloc_extent_start = ins.objectid; - // btrfs_update_inode_block_group(trans, inode); - ret = btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - - while(count > 0) { - size_t offset = pos & (PAGE_CACHE_SIZE - 1); - size_t write_bytes = min(count, PAGE_CACHE_SIZE - offset); - size_t num_pages = (write_bytes + PAGE_CACHE_SIZE - 1) >> - PAGE_CACHE_SHIFT; - - memset(pages, 0, sizeof(pages)); - ret = prepare_pages(root, file, pages, num_pages, - pos, first_index, last_index, - write_bytes, alloc_extent_start); - BUG_ON(ret); - - /* FIXME blocks != pagesize */ - if (alloc_extent_start) - alloc_extent_start += num_pages; - ret = btrfs_copy_from_user(pos, num_pages, - write_bytes, pages, buf); - BUG_ON(ret); - - ret = dirty_and_release_pages(NULL, root, file, pages, - num_pages, pos, write_bytes); - BUG_ON(ret); - btrfs_drop_pages(pages, num_pages); - - buf += write_bytes; - count -= write_bytes; - pos += write_bytes; - num_written += write_bytes; - - balance_dirty_pages_ratelimited(inode->i_mapping); - btrfs_btree_balance_dirty(root); - cond_resched(); - } -out_unlock: - mutex_unlock(&inode->i_mutex); -out: - if (pinned[0]) - page_cache_release(pinned[0]); - if (pinned[1]) - page_cache_release(pinned[1]); - *ppos = pos; - current->backing_dev_info = NULL; - mark_inode_dirty(inode); - return num_written ? num_written : err; -} - -static int btrfs_read_actor(read_descriptor_t *desc, struct page *page, - unsigned long offset, unsigned long size) -{ - char *kaddr; - unsigned long left, count = desc->count; - struct inode *inode = page->mapping->host; - - if (size > count) - size = count; - - if (!PageChecked(page)) { - /* FIXME, do it per block */ - struct btrfs_root *root = BTRFS_I(inode)->root; - int ret; - struct buffer_head *bh; - - if (page_has_buffers(page)) { - bh = page_buffers(page); - if (!buffer_mapped(bh)) { - SetPageChecked(page); - goto checked; - } - } - - ret = btrfs_csum_verify_file_block(root, - page->mapping->host->i_ino, - page->index << PAGE_CACHE_SHIFT, - kmap(page), PAGE_CACHE_SIZE); - if (ret) { - if (ret != -ENOENT) { - printk("failed to verify ino %lu page %lu ret %d\n", - page->mapping->host->i_ino, - page->index, ret); - memset(page_address(page), 1, PAGE_CACHE_SIZE); - flush_dcache_page(page); - } - } - SetPageChecked(page); - kunmap(page); - } -checked: - /* - * Faults on the destination of a read are common, so do it before - * taking the kmap. - */ - if (!fault_in_pages_writeable(desc->arg.buf, size)) { - kaddr = kmap_atomic(page, KM_USER0); - left = __copy_to_user_inatomic(desc->arg.buf, - kaddr + offset, size); - kunmap_atomic(kaddr, KM_USER0); - if (left == 0) - goto success; - } - - /* Do it the slow way */ - kaddr = kmap(page); - left = __copy_to_user(desc->arg.buf, kaddr + offset, size); - kunmap(page); - - if (left) { - size -= left; - desc->error = -EFAULT; - } -success: - desc->count = count - size; - desc->written += size; - desc->arg.buf += size; - return size; -} - -/** - * btrfs_file_aio_read - filesystem read routine - * @iocb: kernel I/O control block - * @iov: io vector request - * @nr_segs: number of segments in the iovec - * @pos: current file position - */ -static ssize_t btrfs_file_aio_read(struct kiocb *iocb, const struct iovec *iov, - unsigned long nr_segs, loff_t pos) -{ - struct file *filp = iocb->ki_filp; - ssize_t retval; - unsigned long seg; - size_t count; - loff_t *ppos = &iocb->ki_pos; - - count = 0; - for (seg = 0; seg < nr_segs; seg++) { - const struct iovec *iv = &iov[seg]; - - /* - * If any segment has a negative length, or the cumulative - * length ever wraps negative then return -EINVAL. - */ - count += iv->iov_len; - if (unlikely((ssize_t)(count|iv->iov_len) < 0)) - return -EINVAL; - if (access_ok(VERIFY_WRITE, iv->iov_base, iv->iov_len)) - continue; - if (seg == 0) - return -EFAULT; - nr_segs = seg; - count -= iv->iov_len; /* This segment is no good */ - break; - } - retval = 0; - if (count) { - for (seg = 0; seg < nr_segs; seg++) { - read_descriptor_t desc; - - desc.written = 0; - desc.arg.buf = iov[seg].iov_base; - desc.count = iov[seg].iov_len; - if (desc.count == 0) - continue; - desc.error = 0; - do_generic_file_read(filp, ppos, &desc, - btrfs_read_actor); - retval += desc.written; - if (desc.error) { - retval = retval ?: desc.error; - break; - } - } - } - return retval; -} - -static int create_subvol(struct btrfs_root *root, char *name, int namelen) -{ - struct btrfs_trans_handle *trans; - struct btrfs_key key; - struct btrfs_root_item root_item; - struct btrfs_inode_item *inode_item; - struct buffer_head *subvol; - struct btrfs_leaf *leaf; - struct btrfs_root *new_root; - struct inode *inode; - struct inode *dir; - int ret; - u64 objectid; - u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID; - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - BUG_ON(!trans); - - subvol = btrfs_alloc_free_block(trans, root, 0); - if (subvol == NULL) - return -ENOSPC; - leaf = btrfs_buffer_leaf(subvol); - btrfs_set_header_nritems(&leaf->header, 0); - btrfs_set_header_level(&leaf->header, 0); - btrfs_set_header_blocknr(&leaf->header, bh_blocknr(subvol)); - btrfs_set_header_generation(&leaf->header, trans->transid); - btrfs_set_header_owner(&leaf->header, root->root_key.objectid); - memcpy(leaf->header.fsid, root->fs_info->disk_super->fsid, - sizeof(leaf->header.fsid)); - mark_buffer_dirty(subvol); - - inode_item = &root_item.inode; - memset(inode_item, 0, sizeof(*inode_item)); - btrfs_set_inode_generation(inode_item, 1); - btrfs_set_inode_size(inode_item, 3); - btrfs_set_inode_nlink(inode_item, 1); - btrfs_set_inode_nblocks(inode_item, 1); - btrfs_set_inode_mode(inode_item, S_IFDIR | 0755); - - btrfs_set_root_blocknr(&root_item, bh_blocknr(subvol)); - btrfs_set_root_refs(&root_item, 1); - brelse(subvol); - subvol = NULL; - - ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root, - 0, &objectid); - BUG_ON(ret); - - btrfs_set_root_dirid(&root_item, new_dirid); - - key.objectid = objectid; - key.offset = 1; - key.flags = 0; - btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); - ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key, - &root_item); - BUG_ON(ret); - - /* - * insert the directory item - */ - key.offset = (u64)-1; - dir = root->fs_info->sb->s_root->d_inode; - ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root, - name, namelen, dir->i_ino, &key, - BTRFS_FT_DIR); - BUG_ON(ret); - - ret = btrfs_commit_transaction(trans, root); - BUG_ON(ret); - - new_root = btrfs_read_fs_root(root->fs_info, &key); - BUG_ON(!new_root); - - trans = btrfs_start_transaction(new_root, 1); - BUG_ON(!trans); - - inode = btrfs_new_inode(trans, new_root, new_dirid, - BTRFS_I(dir)->block_group, S_IFDIR | 0700); - inode->i_op = &btrfs_dir_inode_operations; - inode->i_fop = &btrfs_dir_file_operations; - - ret = btrfs_make_empty_dir(trans, new_root, new_dirid, new_dirid); - BUG_ON(ret); - - inode->i_nlink = 1; - inode->i_size = 6; - ret = btrfs_update_inode(trans, new_root, inode); - BUG_ON(ret); - - ret = btrfs_commit_transaction(trans, new_root); - BUG_ON(ret); - - iput(inode); - - mutex_unlock(&root->fs_info->fs_mutex); - btrfs_btree_balance_dirty(root); - return 0; -} - -static int create_snapshot(struct btrfs_root *root, char *name, int namelen) -{ - struct btrfs_trans_handle *trans; - struct btrfs_key key; - struct btrfs_root_item new_root_item; - int ret; - u64 objectid; - - if (!root->ref_cows) - return -EINVAL; - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - BUG_ON(!trans); - - ret = btrfs_update_inode(trans, root, root->inode); - BUG_ON(ret); - - ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root, - 0, &objectid); - BUG_ON(ret); - - memcpy(&new_root_item, &root->root_item, - sizeof(new_root_item)); - - key.objectid = objectid; - key.offset = 1; - key.flags = 0; - btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); - btrfs_set_root_blocknr(&new_root_item, bh_blocknr(root->node)); - - ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key, - &new_root_item); - BUG_ON(ret); - - /* - * insert the directory item - */ - key.offset = (u64)-1; - ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root, - name, namelen, - root->fs_info->sb->s_root->d_inode->i_ino, - &key, BTRFS_FT_DIR); - - BUG_ON(ret); - - ret = btrfs_inc_root_ref(trans, root); - BUG_ON(ret); - - ret = btrfs_commit_transaction(trans, root); - BUG_ON(ret); - mutex_unlock(&root->fs_info->fs_mutex); - btrfs_btree_balance_dirty(root); - return 0; -} - -static int btrfs_ioctl(struct inode *inode, struct file *filp, unsigned int - cmd, unsigned long arg) -{ - struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_ioctl_vol_args vol_args; - int ret = 0; - struct btrfs_dir_item *di; - int namelen; - struct btrfs_path *path; - u64 root_dirid; - - switch (cmd) { - case BTRFS_IOC_SNAP_CREATE: - if (copy_from_user(&vol_args, - (struct btrfs_ioctl_vol_args __user *)arg, - sizeof(vol_args))) - return -EFAULT; - namelen = strlen(vol_args.name); - if (namelen > BTRFS_VOL_NAME_MAX) - return -EINVAL; - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; - root_dirid = root->fs_info->sb->s_root->d_inode->i_ino, - mutex_lock(&root->fs_info->fs_mutex); - di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root, - path, root_dirid, - vol_args.name, namelen, 0); - mutex_unlock(&root->fs_info->fs_mutex); - btrfs_free_path(path); - if (di && !IS_ERR(di)) - return -EEXIST; - - if (root == root->fs_info->tree_root) - ret = create_subvol(root, vol_args.name, namelen); - else - ret = create_snapshot(root, vol_args.name, namelen); - WARN_ON(ret); - break; - default: - return -ENOTTY; - } - return ret; -} - -#ifdef CONFIG_COMPAT -static long btrfs_compat_ioctl(struct file *file, unsigned int cmd, - unsigned long arg) -{ - struct inode *inode = file->f_path.dentry->d_inode; - int ret; - lock_kernel(); - ret = btrfs_ioctl(inode, file, cmd, (unsigned long) compat_ptr(arg)); - unlock_kernel(); - return ret; - -} -#endif - -static struct kmem_cache *btrfs_inode_cachep; -struct kmem_cache *btrfs_trans_handle_cachep; -struct kmem_cache *btrfs_transaction_cachep; -struct kmem_cache *btrfs_bit_radix_cachep; -struct kmem_cache *btrfs_path_cachep; - -/* - * Called inside transaction, so use GFP_NOFS - */ -static struct inode *btrfs_alloc_inode(struct super_block *sb) -{ - struct btrfs_inode *ei; - - ei = kmem_cache_alloc(btrfs_inode_cachep, GFP_NOFS); - if (!ei) - return NULL; - return &ei->vfs_inode; -} - -static void btrfs_destroy_inode(struct inode *inode) -{ - WARN_ON(!list_empty(&inode->i_dentry)); - WARN_ON(inode->i_data.nrpages); - - kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode)); -} - -static void init_once(void * foo, struct kmem_cache * cachep, - unsigned long flags) -{ - struct btrfs_inode *ei = (struct btrfs_inode *) foo; - - if ((flags & (SLAB_CTOR_CONSTRUCTOR)) == - SLAB_CTOR_CONSTRUCTOR) { - inode_init_once(&ei->vfs_inode); - } -} - -static int init_inodecache(void) -{ - btrfs_inode_cachep = kmem_cache_create("btrfs_inode_cache", - sizeof(struct btrfs_inode), - 0, (SLAB_RECLAIM_ACCOUNT| - SLAB_MEM_SPREAD), - init_once, NULL); - btrfs_trans_handle_cachep = kmem_cache_create("btrfs_trans_handle_cache", - sizeof(struct btrfs_trans_handle), - 0, (SLAB_RECLAIM_ACCOUNT| - SLAB_MEM_SPREAD), - NULL, NULL); - btrfs_transaction_cachep = kmem_cache_create("btrfs_transaction_cache", - sizeof(struct btrfs_transaction), - 0, (SLAB_RECLAIM_ACCOUNT| - SLAB_MEM_SPREAD), - NULL, NULL); - btrfs_path_cachep = kmem_cache_create("btrfs_path_cache", - sizeof(struct btrfs_transaction), - 0, (SLAB_RECLAIM_ACCOUNT| - SLAB_MEM_SPREAD), - NULL, NULL); - btrfs_bit_radix_cachep = kmem_cache_create("btrfs_radix", - 256, - 0, (SLAB_RECLAIM_ACCOUNT| - SLAB_MEM_SPREAD | - SLAB_DESTROY_BY_RCU), - NULL, NULL); - if (btrfs_inode_cachep == NULL || btrfs_trans_handle_cachep == NULL || - btrfs_transaction_cachep == NULL || btrfs_bit_radix_cachep == NULL) - return -ENOMEM; - return 0; -} - -static void destroy_inodecache(void) -{ - kmem_cache_destroy(btrfs_inode_cachep); - kmem_cache_destroy(btrfs_trans_handle_cachep); - kmem_cache_destroy(btrfs_transaction_cachep); - kmem_cache_destroy(btrfs_bit_radix_cachep); - kmem_cache_destroy(btrfs_path_cachep); -} - static int btrfs_get_sb(struct file_system_type *fs_type, int flags, const char *dev_name, void *data, struct vfsmount *mnt) { @@ -2709,15 +124,6 @@ static int btrfs_get_sb(struct file_system_type *fs_type, btrfs_fill_super, mnt); } -static int btrfs_getattr(struct vfsmount *mnt, - struct dentry *dentry, struct kstat *stat) -{ - struct inode *inode = dentry->d_inode; - generic_fillattr(inode, stat); - stat->blksize = 256 * 1024; - return 0; -} - static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) { struct btrfs_root *root = btrfs_sb(dentry->d_sb); @@ -2732,197 +138,6 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) return 0; } -static int btrfs_rename(struct inode * old_dir, struct dentry *old_dentry, - struct inode * new_dir,struct dentry *new_dentry) -{ - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(old_dir)->root; - struct inode *new_inode = new_dentry->d_inode; - struct inode *old_inode = old_dentry->d_inode; - struct timespec ctime = CURRENT_TIME; - struct btrfs_path *path; - struct btrfs_dir_item *di; - int ret; - - if (S_ISDIR(old_inode->i_mode) && new_inode && - new_inode->i_size > BTRFS_EMPTY_DIR_SIZE) { - return -ENOTEMPTY; - } - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, new_dir); - path = btrfs_alloc_path(); - if (!path) { - ret = -ENOMEM; - goto out_fail; - } - - old_dentry->d_inode->i_nlink++; - old_dir->i_ctime = old_dir->i_mtime = ctime; - new_dir->i_ctime = new_dir->i_mtime = ctime; - old_inode->i_ctime = ctime; - if (S_ISDIR(old_inode->i_mode) && old_dir != new_dir) { - struct btrfs_key *location = &BTRFS_I(new_dir)->location; - u64 old_parent_oid; - di = btrfs_lookup_dir_item(trans, root, path, old_inode->i_ino, - "..", 2, -1); - if (IS_ERR(di)) { - ret = PTR_ERR(di); - goto out_fail; - } - if (!di) { - ret = -ENOENT; - goto out_fail; - } - old_parent_oid = btrfs_disk_key_objectid(&di->location); - ret = btrfs_del_item(trans, root, path); - if (ret) { - ret = -EIO; - goto out_fail; - } - btrfs_release_path(root, path); - - di = btrfs_lookup_dir_index_item(trans, root, path, - old_inode->i_ino, - old_parent_oid, - "..", 2, -1); - if (IS_ERR(di)) { - ret = PTR_ERR(di); - goto out_fail; - } - if (!di) { - ret = -ENOENT; - goto out_fail; - } - ret = btrfs_del_item(trans, root, path); - if (ret) { - ret = -EIO; - goto out_fail; - } - btrfs_release_path(root, path); - - ret = btrfs_insert_dir_item(trans, root, "..", 2, - old_inode->i_ino, location, - BTRFS_FT_DIR); - if (ret) - goto out_fail; - } - - - ret = btrfs_unlink_trans(trans, root, old_dir, old_dentry); - if (ret) - goto out_fail; - - if (new_inode) { - new_inode->i_ctime = CURRENT_TIME; - ret = btrfs_unlink_trans(trans, root, new_dir, new_dentry); - if (ret) - goto out_fail; - if (S_ISDIR(new_inode->i_mode)) - clear_nlink(new_inode); - else - drop_nlink(new_inode); - btrfs_update_inode(trans, root, new_inode); - } - ret = btrfs_add_link(trans, new_dentry, old_inode); - if (ret) - goto out_fail; - -out_fail: - btrfs_free_path(path); - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - return ret; -} - -static int btrfs_symlink(struct inode *dir, struct dentry *dentry, - const char *symname) -{ - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(dir)->root; - struct btrfs_path *path; - struct btrfs_key key; - struct inode *inode; - int err; - int drop_inode = 0; - u64 objectid; - int name_len; - int datasize; - char *ptr; - struct btrfs_file_extent_item *ei; - - name_len = strlen(symname) + 1; - if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root)) - return -ENAMETOOLONG; - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, dir); - - err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); - if (err) { - err = -ENOSPC; - goto out_unlock; - } - - inode = btrfs_new_inode(trans, root, objectid, - BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO); - err = PTR_ERR(inode); - if (IS_ERR(inode)) - goto out_unlock; - - btrfs_set_trans_block_group(trans, inode); - err = btrfs_add_nondir(trans, dentry, inode); - if (err) - drop_inode = 1; - else { - inode->i_mapping->a_ops = &btrfs_aops; - inode->i_fop = &btrfs_file_operations; - inode->i_op = &btrfs_file_inode_operations; - } - dir->i_sb->s_dirt = 1; - btrfs_update_inode_block_group(trans, inode); - btrfs_update_inode_block_group(trans, dir); - if (drop_inode) - goto out_unlock; - - path = btrfs_alloc_path(); - BUG_ON(!path); - key.objectid = inode->i_ino; - key.offset = 0; - key.flags = 0; - btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY); - datasize = btrfs_file_extent_calc_inline_size(name_len); - err = btrfs_insert_empty_item(trans, root, path, &key, - datasize); - BUG_ON(err); - ei = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), - path->slots[0], struct btrfs_file_extent_item); - btrfs_set_file_extent_generation(ei, trans->transid); - btrfs_set_file_extent_type(ei, - BTRFS_FILE_EXTENT_INLINE); - ptr = btrfs_file_extent_inline_start(ei); - btrfs_memcpy(root, path->nodes[0]->b_data, - ptr, symname, name_len); - mark_buffer_dirty(path->nodes[0]); - btrfs_free_path(path); - inode->i_op = &btrfs_symlink_inode_operations; - inode->i_mapping->a_ops = &btrfs_symlink_aops; - inode->i_size = name_len - 1; - btrfs_update_inode(trans, root, inode); - err = 0; - -out_unlock: - btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - - if (drop_inode) { - inode_dec_link_count(inode); - iput(inode); - } - btrfs_btree_balance_dirty(root); - return err; -} - static struct file_system_type btrfs_fs_type = { .owner = THIS_MODULE, .name = "btrfs", @@ -2944,91 +159,21 @@ static struct super_operations btrfs_super_ops = { .statfs = btrfs_statfs, }; -static struct inode_operations btrfs_dir_inode_operations = { - .lookup = btrfs_lookup, - .create = btrfs_create, - .unlink = btrfs_unlink, - .link = btrfs_link, - .mkdir = btrfs_mkdir, - .rmdir = btrfs_rmdir, - .rename = btrfs_rename, - .symlink = btrfs_symlink, - .setattr = btrfs_setattr, -}; - -static struct inode_operations btrfs_dir_ro_inode_operations = { - .lookup = btrfs_lookup, -}; - -static struct file_operations btrfs_dir_file_operations = { - .llseek = generic_file_llseek, - .read = generic_read_dir, - .readdir = btrfs_readdir, - .ioctl = btrfs_ioctl, -#ifdef CONFIG_COMPAT - .compat_ioctl = btrfs_compat_ioctl, -#endif -}; - -static struct address_space_operations btrfs_aops = { - .readpage = btrfs_readpage, - .writepage = btrfs_writepage, - .sync_page = block_sync_page, - .prepare_write = btrfs_prepare_write, - .commit_write = btrfs_commit_write, - .bmap = btrfs_bmap, -}; - -static struct address_space_operations btrfs_symlink_aops = { - .readpage = btrfs_readpage, - .writepage = btrfs_writepage, -}; - -static struct inode_operations btrfs_file_inode_operations = { - .truncate = btrfs_truncate, - .getattr = btrfs_getattr, - .setattr = btrfs_setattr, -}; - -static struct file_operations btrfs_file_operations = { - .llseek = generic_file_llseek, - .read = do_sync_read, - .aio_read = btrfs_file_aio_read, - .write = btrfs_file_write, - .mmap = generic_file_mmap, - .open = generic_file_open, - .ioctl = btrfs_ioctl, - .fsync = btrfs_sync_file, -#ifdef CONFIG_COMPAT - .compat_ioctl = btrfs_compat_ioctl, -#endif -}; - -static struct inode_operations btrfs_symlink_inode_operations = { - .readlink = generic_readlink, - .follow_link = page_follow_link_light, - .put_link = page_put_link, -}; - static int __init init_btrfs_fs(void) { int err; - printk("btrfs loaded!\n"); btrfs_init_transaction_sys(); - err = init_inodecache(); + err = btrfs_init_cachep(); if (err) return err; return register_filesystem(&btrfs_fs_type); - destroy_inodecache(); - return err; } static void __exit exit_btrfs_fs(void) { btrfs_exit_transaction_sys(); - destroy_inodecache(); + btrfs_destroy_cachep(); unregister_filesystem(&btrfs_fs_type); - printk("btrfs unloaded\n"); } module_init(init_btrfs_fs) |