diff options
Diffstat (limited to 'fs/btrfs/inode.c')
| -rw-r--r-- | fs/btrfs/inode.c | 6791 |
1 files changed, 4391 insertions, 2400 deletions
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 558cac2dfa5..3668048e16f 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -32,27 +32,36 @@ #include <linux/writeback.h> #include <linux/statfs.h> #include <linux/compat.h> +#include <linux/aio.h> #include <linux/bit_spinlock.h> #include <linux/xattr.h> #include <linux/posix_acl.h> #include <linux/falloc.h> #include <linux/slab.h> -#include "compat.h" +#include <linux/ratelimit.h> +#include <linux/mount.h> +#include <linux/btrfs.h> +#include <linux/blkdev.h> +#include <linux/posix_acl_xattr.h> #include "ctree.h" #include "disk-io.h" #include "transaction.h" #include "btrfs_inode.h" -#include "ioctl.h" #include "print-tree.h" -#include "volumes.h" #include "ordered-data.h" #include "xattr.h" #include "tree-log.h" +#include "volumes.h" #include "compression.h" #include "locking.h" +#include "free-space-cache.h" +#include "inode-map.h" +#include "backref.h" +#include "hash.h" +#include "props.h" struct btrfs_iget_args { - u64 ino; + struct btrfs_key *location; struct btrfs_root *root; }; @@ -67,9 +76,11 @@ static const struct file_operations btrfs_dir_file_operations; static struct extent_io_ops btrfs_extent_io_ops; static struct kmem_cache *btrfs_inode_cachep; +static struct kmem_cache *btrfs_delalloc_work_cachep; struct kmem_cache *btrfs_trans_handle_cachep; struct kmem_cache *btrfs_transaction_cachep; struct kmem_cache *btrfs_path_cachep; +struct kmem_cache *btrfs_free_space_cachep; #define S_SHIFT 12 static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = { @@ -82,21 +93,30 @@ static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = { [S_IFLNK >> S_SHIFT] = BTRFS_FT_SYMLINK, }; -static void btrfs_truncate(struct inode *inode); -static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end); +static int btrfs_setsize(struct inode *inode, struct iattr *attr); +static int btrfs_truncate(struct inode *inode); +static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent); static noinline int cow_file_range(struct inode *inode, struct page *locked_page, u64 start, u64 end, int *page_started, unsigned long *nr_written, int unlock); +static struct extent_map *create_pinned_em(struct inode *inode, u64 start, + u64 len, u64 orig_start, + u64 block_start, u64 block_len, + u64 orig_block_len, u64 ram_bytes, + int type); + +static int btrfs_dirty_inode(struct inode *inode); static int btrfs_init_inode_security(struct btrfs_trans_handle *trans, - struct inode *inode, struct inode *dir) + struct inode *inode, struct inode *dir, + const struct qstr *qstr) { int err; err = btrfs_init_acl(trans, inode, dir); if (!err) - err = btrfs_xattr_security_init(trans, inode, dir); + err = btrfs_xattr_security_init(trans, inode, dir, qstr); return err; } @@ -105,13 +125,13 @@ static int btrfs_init_inode_security(struct btrfs_trans_handle *trans, * the btree. The caller should have done a btrfs_drop_extents so that * no overlapping inline items exist in the btree */ -static noinline int insert_inline_extent(struct btrfs_trans_handle *trans, +static int insert_inline_extent(struct btrfs_trans_handle *trans, + struct btrfs_path *path, int extent_inserted, struct btrfs_root *root, struct inode *inode, u64 start, size_t size, size_t compressed_size, + int compress_type, struct page **compressed_pages) { - struct btrfs_key key; - struct btrfs_path *path; struct extent_buffer *leaf; struct page *page = NULL; char *kaddr; @@ -120,34 +140,29 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans, int err = 0; int ret; size_t cur_size = size; - size_t datasize; unsigned long offset; - int use_compress = 0; - if (compressed_size && compressed_pages) { - use_compress = 1; + if (compressed_size && compressed_pages) cur_size = compressed_size; - } - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; + inode_add_bytes(inode, size); - path->leave_spinning = 1; - btrfs_set_trans_block_group(trans, inode); + if (!extent_inserted) { + struct btrfs_key key; + size_t datasize; - key.objectid = inode->i_ino; - key.offset = start; - btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY); - datasize = btrfs_file_extent_calc_inline_size(cur_size); + key.objectid = btrfs_ino(inode); + key.offset = start; + btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY); - inode_add_bytes(inode, size); - ret = btrfs_insert_empty_item(trans, root, path, &key, - datasize); - BUG_ON(ret); - if (ret) { - err = ret; - goto fail; + datasize = btrfs_file_extent_calc_inline_size(cur_size); + path->leave_spinning = 1; + ret = btrfs_insert_empty_item(trans, root, path, &key, + datasize); + if (ret) { + err = ret; + goto fail; + } } leaf = path->nodes[0]; ei = btrfs_item_ptr(leaf, path->slots[0], @@ -159,7 +174,7 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans, btrfs_set_file_extent_ram_bytes(leaf, ei, size); ptr = btrfs_file_extent_inline_start(ei); - if (use_compress) { + if (compress_type != BTRFS_COMPRESS_NONE) { struct page *cpage; int i = 0; while (compressed_size > 0) { @@ -167,28 +182,28 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans, cur_size = min_t(unsigned long, compressed_size, PAGE_CACHE_SIZE); - kaddr = kmap_atomic(cpage, KM_USER0); + kaddr = kmap_atomic(cpage); write_extent_buffer(leaf, kaddr, ptr, cur_size); - kunmap_atomic(kaddr, KM_USER0); + kunmap_atomic(kaddr); i++; ptr += cur_size; compressed_size -= cur_size; } btrfs_set_file_extent_compression(leaf, ei, - BTRFS_COMPRESS_ZLIB); + compress_type); } else { page = find_get_page(inode->i_mapping, start >> PAGE_CACHE_SHIFT); btrfs_set_file_extent_compression(leaf, ei, 0); - kaddr = kmap_atomic(page, KM_USER0); + kaddr = kmap_atomic(page); offset = start & (PAGE_CACHE_SIZE - 1); write_extent_buffer(leaf, kaddr + offset, ptr, size); - kunmap_atomic(kaddr, KM_USER0); + kunmap_atomic(kaddr); page_cache_release(page); } btrfs_mark_buffer_dirty(leaf); - btrfs_free_path(path); + btrfs_release_path(path); /* * we're an inline extent, so nobody can @@ -200,11 +215,10 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans, * could end up racing with unlink. */ BTRFS_I(inode)->disk_i_size = inode->i_size; - btrfs_update_inode(trans, root, inode); + ret = btrfs_update_inode(trans, root, inode); - return 0; + return ret; fail: - btrfs_free_path(path); return err; } @@ -214,20 +228,22 @@ fail: * does the checks required to make sure the data is small enough * to fit as an inline extent. */ -static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode, u64 start, u64 end, - size_t compressed_size, - struct page **compressed_pages) +static noinline int cow_file_range_inline(struct btrfs_root *root, + struct inode *inode, u64 start, + u64 end, size_t compressed_size, + int compress_type, + struct page **compressed_pages) { + struct btrfs_trans_handle *trans; u64 isize = i_size_read(inode); u64 actual_end = min(end + 1, isize); u64 inline_len = actual_end - start; - u64 aligned_end = (end + root->sectorsize - 1) & - ~((u64)root->sectorsize - 1); - u64 hint_byte; + u64 aligned_end = ALIGN(end, root->sectorsize); u64 data_len = inline_len; int ret; + struct btrfs_path *path; + int extent_inserted = 0; + u32 extent_item_size; if (compressed_size) data_len = compressed_size; @@ -242,19 +258,53 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans, return 1; } - ret = btrfs_drop_extents(trans, inode, start, aligned_end, - &hint_byte, 1); - BUG_ON(ret); + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) { + btrfs_free_path(path); + return PTR_ERR(trans); + } + trans->block_rsv = &root->fs_info->delalloc_block_rsv; + + if (compressed_size && compressed_pages) + extent_item_size = btrfs_file_extent_calc_inline_size( + compressed_size); + else + extent_item_size = btrfs_file_extent_calc_inline_size( + inline_len); + + ret = __btrfs_drop_extents(trans, root, inode, path, + start, aligned_end, NULL, + 1, 1, extent_item_size, &extent_inserted); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto out; + } if (isize > actual_end) inline_len = min_t(u64, isize, actual_end); - ret = insert_inline_extent(trans, root, inode, start, + ret = insert_inline_extent(trans, path, extent_inserted, + root, inode, start, inline_len, compressed_size, - compressed_pages); - BUG_ON(ret); + compress_type, compressed_pages); + if (ret && ret != -ENOSPC) { + btrfs_abort_transaction(trans, root, ret); + goto out; + } else if (ret == -ENOSPC) { + ret = 1; + goto out; + } + + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags); btrfs_delalloc_release_metadata(inode, end + 1 - start); btrfs_drop_extent_cache(inode, start, aligned_end - 1, 0); - return 0; +out: + btrfs_free_path(path); + btrfs_end_transaction(trans, root); + return ret; } struct async_extent { @@ -263,6 +313,7 @@ struct async_extent { u64 compressed_size; struct page **pages; unsigned long nr_pages; + int compress_type; struct list_head list; }; @@ -280,16 +331,19 @@ static noinline int add_async_extent(struct async_cow *cow, u64 start, u64 ram_size, u64 compressed_size, struct page **pages, - unsigned long nr_pages) + unsigned long nr_pages, + int compress_type) { struct async_extent *async_extent; async_extent = kmalloc(sizeof(*async_extent), GFP_NOFS); + BUG_ON(!async_extent); /* -ENOMEM */ async_extent->start = start; async_extent->ram_size = ram_size; async_extent->compressed_size = compressed_size; async_extent->pages = pages; async_extent->nr_pages = nr_pages; + async_extent->compress_type = compress_type; list_add_tail(&async_extent->list, &cow->extents); return 0; } @@ -308,7 +362,8 @@ static noinline int add_async_extent(struct async_cow *cow, * If this code finds it can't get good compression, it puts an * entry onto the work queue to write the uncompressed bytes. This * makes sure that both compressed inodes and uncompressed inodes - * are written in the same order that pdflush sent them down. + * are written in the same order that the flusher thread sent them + * down. */ static noinline int compress_file_range(struct inode *inode, struct page *locked_page, @@ -317,7 +372,6 @@ static noinline int compress_file_range(struct inode *inode, int *num_added) { struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_trans_handle *trans; u64 num_bytes; u64 blocksize = root->sectorsize; u64 actual_end; @@ -332,6 +386,21 @@ static noinline int compress_file_range(struct inode *inode, unsigned long max_uncompressed = 128 * 1024; int i; int will_compress; + int compress_type = root->fs_info->compress_type; + int redirty = 0; + + /* if this is a small write inside eof, kick off a defrag */ + if ((end - start + 1) < 16 * 1024 && + (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size)) + btrfs_add_inode_defrag(NULL, inode); + + /* + * skip compression for a small file range(<=blocksize) that + * isn't an inline extent, since it dosen't save disk space at all. + */ + if ((end - start + 1) <= blocksize && + (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size)) + goto cleanup_and_bail_uncompressed; actual_end = min_t(u64, isize, end + 1); again: @@ -365,7 +434,7 @@ again: * a compressed extent to 128k. */ total_compressed = min(total_compressed, max_uncompressed); - num_bytes = (end - start + blocksize) & ~(blocksize - 1); + num_bytes = ALIGN(end - start + 1, blocksize); num_bytes = max(blocksize, num_bytes); total_in = 0; ret = 0; @@ -377,16 +446,36 @@ again: */ if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS) && (btrfs_test_opt(root, COMPRESS) || - (BTRFS_I(inode)->force_compress))) { + (BTRFS_I(inode)->force_compress) || + (BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS))) { WARN_ON(pages); pages = kzalloc(sizeof(struct page *) * nr_pages, GFP_NOFS); + if (!pages) { + /* just bail out to the uncompressed code */ + goto cont; + } + + if (BTRFS_I(inode)->force_compress) + compress_type = BTRFS_I(inode)->force_compress; - ret = btrfs_zlib_compress_pages(inode->i_mapping, start, - total_compressed, pages, - nr_pages, &nr_pages_ret, - &total_in, - &total_compressed, - max_compressed); + /* + * we need to call clear_page_dirty_for_io on each + * page in the range. Otherwise applications with the file + * mmap'd can wander in and change the page contents while + * we are compressing them. + * + * If the compression fails for any reason, we set the pages + * dirty again later on. + */ + extent_range_clear_dirty_for_io(inode, start, end); + redirty = 1; + ret = btrfs_compress_pages(compress_type, + inode->i_mapping, start, + total_compressed, pages, + nr_pages, &nr_pages_ret, + &total_in, + &total_compressed, + max_compressed); if (!ret) { unsigned long offset = total_compressed & @@ -398,50 +487,46 @@ again: * sending it down to disk */ if (offset) { - kaddr = kmap_atomic(page, KM_USER0); + kaddr = kmap_atomic(page); memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset); - kunmap_atomic(kaddr, KM_USER0); + kunmap_atomic(kaddr); } will_compress = 1; } } +cont: if (start == 0) { - trans = btrfs_join_transaction(root, 1); - BUG_ON(!trans); - btrfs_set_trans_block_group(trans, inode); - trans->block_rsv = &root->fs_info->delalloc_block_rsv; - /* lets try to make an inline extent */ if (ret || total_in < (actual_end - start)) { /* we didn't compress the entire range, try * to make an uncompressed inline extent. */ - ret = cow_file_range_inline(trans, root, inode, - start, end, 0, NULL); + ret = cow_file_range_inline(root, inode, start, end, + 0, 0, NULL); } else { /* try making a compressed inline extent */ - ret = cow_file_range_inline(trans, root, inode, - start, end, - total_compressed, pages); + ret = cow_file_range_inline(root, inode, start, end, + total_compressed, + compress_type, pages); } - if (ret == 0) { + if (ret <= 0) { + unsigned long clear_flags = EXTENT_DELALLOC | + EXTENT_DEFRAG; + clear_flags |= (ret < 0) ? EXTENT_DO_ACCOUNTING : 0; + /* - * inline extent creation worked, we don't need - * to create any more async work items. Unlock - * and free up our temp pages. + * inline extent creation worked or returned error, + * we don't need to create any more async work items. + * Unlock and free up our temp pages. */ - extent_clear_unlock_delalloc(inode, - &BTRFS_I(inode)->io_tree, - start, end, NULL, - EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY | - EXTENT_CLEAR_DELALLOC | - EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK); - - btrfs_end_transaction(trans, root); + extent_clear_unlock_delalloc(inode, start, end, NULL, + clear_flags, PAGE_UNLOCK | + PAGE_CLEAR_DIRTY | + PAGE_SET_WRITEBACK | + PAGE_END_WRITEBACK); goto free_pages_out; } - btrfs_end_transaction(trans, root); } if (will_compress) { @@ -450,15 +535,13 @@ again: * up to a block size boundary so the allocator does sane * things */ - total_compressed = (total_compressed + blocksize - 1) & - ~(blocksize - 1); + total_compressed = ALIGN(total_compressed, blocksize); /* * one last check to make sure the compression is really a * win, compare the page count read with the blocks on disk */ - total_in = (total_in + PAGE_CACHE_SIZE - 1) & - ~(PAGE_CACHE_SIZE - 1); + total_in = ALIGN(total_in, PAGE_CACHE_SIZE); if (total_compressed >= total_in) { will_compress = 0; } else { @@ -493,9 +576,10 @@ again: * and will submit them to the elevator. */ add_async_extent(async_cow, start, num_bytes, - total_compressed, pages, nr_pages_ret); + total_compressed, pages, nr_pages_ret, + compress_type); - if (start + num_bytes < end && start + num_bytes < actual_end) { + if (start + num_bytes < end) { start += num_bytes; pages = NULL; cond_resched(); @@ -515,12 +599,15 @@ cleanup_and_bail_uncompressed: __set_page_dirty_nobuffers(locked_page); /* unlocked later on in the async handlers */ } - add_async_extent(async_cow, start, end - start + 1, 0, NULL, 0); + if (redirty) + extent_range_redirty_for_io(inode, start, end); + add_async_extent(async_cow, start, end - start + 1, + 0, NULL, 0, BTRFS_COMPRESS_NONE); *num_added += 1; } out: - return 0; + return ret; free_pages_out: for (i = 0; i < nr_pages_ret; i++) { @@ -543,7 +630,6 @@ static noinline int submit_compressed_extents(struct inode *inode, { struct async_extent *async_extent; u64 alloc_hint = 0; - struct btrfs_trans_handle *trans; struct btrfs_key ins; struct extent_map *em; struct btrfs_root *root = BTRFS_I(inode)->root; @@ -554,7 +640,7 @@ static noinline int submit_compressed_extents(struct inode *inode, if (list_empty(&async_cow->extents)) return 0; - +again: while (!list_empty(&async_cow->extents)) { async_extent = list_entry(async_cow->extents.next, struct async_extent, list); @@ -570,7 +656,7 @@ retry: lock_extent(io_tree, async_extent->start, async_extent->start + - async_extent->ram_size - 1, GFP_NOFS); + async_extent->ram_size - 1); /* allocate blocks */ ret = cow_file_range(inode, async_cow->locked_page, @@ -579,6 +665,8 @@ retry: async_extent->ram_size - 1, &page_started, &nr_written, 0); + /* JDM XXX */ + /* * if page_started, cow_file_range inserted an * inline extent and took care of all the unlocking @@ -592,25 +680,23 @@ retry: async_extent->ram_size - 1, btrfs_get_extent, WB_SYNC_ALL); + else if (ret) + unlock_page(async_cow->locked_page); kfree(async_extent); cond_resched(); continue; } lock_extent(io_tree, async_extent->start, - async_extent->start + async_extent->ram_size - 1, - GFP_NOFS); + async_extent->start + async_extent->ram_size - 1); - trans = btrfs_join_transaction(root, 1); - ret = btrfs_reserve_extent(trans, root, + ret = btrfs_reserve_extent(root, async_extent->compressed_size, async_extent->compressed_size, - 0, alloc_hint, - (u64)-1, &ins, 1); - btrfs_end_transaction(trans, root); - + 0, alloc_hint, &ins, 1, 1); if (ret) { int i; + for (i = 0; i < async_extent->nr_pages; i++) { WARN_ON(async_extent->pages[i]->mapping); page_cache_release(async_extent->pages[i]); @@ -618,10 +704,14 @@ retry: kfree(async_extent->pages); async_extent->nr_pages = 0; async_extent->pages = NULL; - unlock_extent(io_tree, async_extent->start, - async_extent->start + - async_extent->ram_size - 1, GFP_NOFS); - goto retry; + + if (ret == -ENOSPC) { + unlock_extent(io_tree, async_extent->start, + async_extent->start + + async_extent->ram_size - 1); + goto retry; + } + goto out_free; } /* @@ -632,20 +722,30 @@ retry: async_extent->start + async_extent->ram_size - 1, 0); - em = alloc_extent_map(GFP_NOFS); + em = alloc_extent_map(); + if (!em) { + ret = -ENOMEM; + goto out_free_reserve; + } em->start = async_extent->start; em->len = async_extent->ram_size; em->orig_start = em->start; + em->mod_start = em->start; + em->mod_len = em->len; em->block_start = ins.objectid; em->block_len = ins.offset; + em->orig_block_len = ins.offset; + em->ram_bytes = async_extent->ram_size; em->bdev = root->fs_info->fs_devices->latest_bdev; + em->compress_type = async_extent->compress_type; set_bit(EXTENT_FLAG_PINNED, &em->flags); set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); + em->generation = -1; while (1) { write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, em); + ret = add_extent_mapping(em_tree, em, 1); write_unlock(&em_tree->lock); if (ret != -EEXIST) { free_extent_map(em); @@ -656,40 +756,55 @@ retry: async_extent->ram_size - 1, 0); } - ret = btrfs_add_ordered_extent(inode, async_extent->start, - ins.objectid, - async_extent->ram_size, - ins.offset, - BTRFS_ORDERED_COMPRESSED); - BUG_ON(ret); + if (ret) + goto out_free_reserve; + + ret = btrfs_add_ordered_extent_compress(inode, + async_extent->start, + ins.objectid, + async_extent->ram_size, + ins.offset, + BTRFS_ORDERED_COMPRESSED, + async_extent->compress_type); + if (ret) + goto out_free_reserve; /* * clear dirty, set writeback and unlock the pages. */ - extent_clear_unlock_delalloc(inode, - &BTRFS_I(inode)->io_tree, - async_extent->start, + extent_clear_unlock_delalloc(inode, async_extent->start, async_extent->start + async_extent->ram_size - 1, - NULL, EXTENT_CLEAR_UNLOCK_PAGE | - EXTENT_CLEAR_UNLOCK | - EXTENT_CLEAR_DELALLOC | - EXTENT_CLEAR_DIRTY | EXTENT_SET_WRITEBACK); - + NULL, EXTENT_LOCKED | EXTENT_DELALLOC, + PAGE_UNLOCK | PAGE_CLEAR_DIRTY | + PAGE_SET_WRITEBACK); ret = btrfs_submit_compressed_write(inode, async_extent->start, async_extent->ram_size, ins.objectid, ins.offset, async_extent->pages, async_extent->nr_pages); - - BUG_ON(ret); alloc_hint = ins.objectid + ins.offset; kfree(async_extent); + if (ret) + goto out; cond_resched(); } - - return 0; + ret = 0; +out: + return ret; +out_free_reserve: + btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1); +out_free: + extent_clear_unlock_delalloc(inode, async_extent->start, + async_extent->start + + async_extent->ram_size - 1, + NULL, EXTENT_LOCKED | EXTENT_DELALLOC | + EXTENT_DEFRAG | EXTENT_DO_ACCOUNTING, + PAGE_UNLOCK | PAGE_CLEAR_DIRTY | + PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK); + kfree(async_extent); + goto again; } static u64 get_extent_allocation_hint(struct inode *inode, u64 start, @@ -744,7 +859,6 @@ static noinline int cow_file_range(struct inode *inode, int unlock) { struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_trans_handle *trans; u64 alloc_hint = 0; u64 num_bytes; unsigned long ram_size; @@ -756,42 +870,43 @@ static noinline int cow_file_range(struct inode *inode, struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; int ret = 0; - BUG_ON(root == root->fs_info->tree_root); - trans = btrfs_join_transaction(root, 1); - BUG_ON(!trans); - btrfs_set_trans_block_group(trans, inode); - trans->block_rsv = &root->fs_info->delalloc_block_rsv; + if (btrfs_is_free_space_inode(inode)) { + WARN_ON_ONCE(1); + ret = -EINVAL; + goto out_unlock; + } - num_bytes = (end - start + blocksize) & ~(blocksize - 1); + num_bytes = ALIGN(end - start + 1, blocksize); num_bytes = max(blocksize, num_bytes); disk_num_bytes = num_bytes; - ret = 0; + + /* if this is a small write inside eof, kick off defrag */ + if (num_bytes < 64 * 1024 && + (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size)) + btrfs_add_inode_defrag(NULL, inode); if (start == 0) { /* lets try to make an inline extent */ - ret = cow_file_range_inline(trans, root, inode, - start, end, 0, NULL); + ret = cow_file_range_inline(root, inode, start, end, 0, 0, + NULL); if (ret == 0) { - extent_clear_unlock_delalloc(inode, - &BTRFS_I(inode)->io_tree, - start, end, NULL, - EXTENT_CLEAR_UNLOCK_PAGE | - EXTENT_CLEAR_UNLOCK | - EXTENT_CLEAR_DELALLOC | - EXTENT_CLEAR_DIRTY | - EXTENT_SET_WRITEBACK | - EXTENT_END_WRITEBACK); + extent_clear_unlock_delalloc(inode, start, end, NULL, + EXTENT_LOCKED | EXTENT_DELALLOC | + EXTENT_DEFRAG, PAGE_UNLOCK | + PAGE_CLEAR_DIRTY | PAGE_SET_WRITEBACK | + PAGE_END_WRITEBACK); *nr_written = *nr_written + (end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE; *page_started = 1; - ret = 0; goto out; + } else if (ret < 0) { + goto out_unlock; } } BUG_ON(disk_num_bytes > - btrfs_super_total_bytes(&root->fs_info->super_copy)); + btrfs_super_total_bytes(root->fs_info->super_copy)); alloc_hint = get_extent_allocation_hint(inode, start, num_bytes); btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0); @@ -800,25 +915,35 @@ static noinline int cow_file_range(struct inode *inode, unsigned long op; cur_alloc_size = disk_num_bytes; - ret = btrfs_reserve_extent(trans, root, cur_alloc_size, + ret = btrfs_reserve_extent(root, cur_alloc_size, root->sectorsize, 0, alloc_hint, - (u64)-1, &ins, 1); - BUG_ON(ret); + &ins, 1, 1); + if (ret < 0) + goto out_unlock; - em = alloc_extent_map(GFP_NOFS); + em = alloc_extent_map(); + if (!em) { + ret = -ENOMEM; + goto out_reserve; + } em->start = start; em->orig_start = em->start; ram_size = ins.offset; em->len = ins.offset; + em->mod_start = em->start; + em->mod_len = em->len; em->block_start = ins.objectid; em->block_len = ins.offset; + em->orig_block_len = ins.offset; + em->ram_bytes = ram_size; em->bdev = root->fs_info->fs_devices->latest_bdev; set_bit(EXTENT_FLAG_PINNED, &em->flags); + em->generation = -1; while (1) { write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, em); + ret = add_extent_mapping(em_tree, em, 1); write_unlock(&em_tree->lock); if (ret != -EEXIST) { free_extent_map(em); @@ -827,17 +952,21 @@ static noinline int cow_file_range(struct inode *inode, btrfs_drop_extent_cache(inode, start, start + ram_size - 1, 0); } + if (ret) + goto out_reserve; cur_alloc_size = ins.offset; ret = btrfs_add_ordered_extent(inode, start, ins.objectid, ram_size, cur_alloc_size, 0); - BUG_ON(ret); + if (ret) + goto out_reserve; if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID) { ret = btrfs_reloc_clone_csums(inode, start, cur_alloc_size); - BUG_ON(ret); + if (ret) + goto out_reserve; } if (disk_num_bytes < cur_alloc_size) @@ -850,23 +979,30 @@ static noinline int cow_file_range(struct inode *inode, * Do set the Private2 bit so we know this page was properly * setup for writepage */ - op = unlock ? EXTENT_CLEAR_UNLOCK_PAGE : 0; - op |= EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC | - EXTENT_SET_PRIVATE2; + op = unlock ? PAGE_UNLOCK : 0; + op |= PAGE_SET_PRIVATE2; - extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree, - start, start + ram_size - 1, - locked_page, op); + extent_clear_unlock_delalloc(inode, start, + start + ram_size - 1, locked_page, + EXTENT_LOCKED | EXTENT_DELALLOC, + op); disk_num_bytes -= cur_alloc_size; num_bytes -= cur_alloc_size; alloc_hint = ins.objectid + ins.offset; start += cur_alloc_size; } out: - ret = 0; - btrfs_end_transaction(trans, root); - return ret; + +out_reserve: + btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1); +out_unlock: + extent_clear_unlock_delalloc(inode, start, end, locked_page, + EXTENT_LOCKED | EXTENT_DO_ACCOUNTING | + EXTENT_DELALLOC | EXTENT_DEFRAG, + PAGE_UNLOCK | PAGE_CLEAR_DIRTY | + PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK); + goto out; } /* @@ -881,8 +1017,10 @@ static noinline void async_cow_start(struct btrfs_work *work) compress_file_range(async_cow->inode, async_cow->locked_page, async_cow->start, async_cow->end, async_cow, &num_added); - if (num_added == 0) + if (num_added == 0) { + btrfs_add_delayed_iput(async_cow->inode); async_cow->inode = NULL; + } } /* @@ -900,10 +1038,8 @@ static noinline void async_cow_submit(struct btrfs_work *work) nr_pages = (async_cow->end - async_cow->start + PAGE_CACHE_SIZE) >> PAGE_CACHE_SHIFT; - atomic_sub(nr_pages, &root->fs_info->async_delalloc_pages); - - if (atomic_read(&root->fs_info->async_delalloc_pages) < - 5 * 1042 * 1024 && + if (atomic_sub_return(nr_pages, &root->fs_info->async_delalloc_pages) < + 5 * 1024 * 1024 && waitqueue_active(&root->fs_info->async_submit_wait)) wake_up(&root->fs_info->async_submit_wait); @@ -915,6 +1051,8 @@ static noinline void async_cow_free(struct btrfs_work *work) { struct async_cow *async_cow; async_cow = container_of(work, struct async_cow, work); + if (async_cow->inode) + btrfs_add_delayed_iput(async_cow->inode); kfree(async_cow); } @@ -926,13 +1064,14 @@ static int cow_file_range_async(struct inode *inode, struct page *locked_page, struct btrfs_root *root = BTRFS_I(inode)->root; unsigned long nr_pages; u64 cur_end; - int limit = 10 * 1024 * 1042; + int limit = 10 * 1024 * 1024; clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED, 1, 0, NULL, GFP_NOFS); while (start < end) { async_cow = kmalloc(sizeof(*async_cow), GFP_NOFS); - async_cow->inode = inode; + BUG_ON(!async_cow); /* -ENOMEM */ + async_cow->inode = igrab(inode); async_cow->root = root; async_cow->locked_page = locked_page; async_cow->start = start; @@ -945,17 +1084,15 @@ static int cow_file_range_async(struct inode *inode, struct page *locked_page, async_cow->end = cur_end; INIT_LIST_HEAD(&async_cow->extents); - async_cow->work.func = async_cow_start; - async_cow->work.ordered_func = async_cow_submit; - async_cow->work.ordered_free = async_cow_free; - async_cow->work.flags = 0; + btrfs_init_work(&async_cow->work, async_cow_start, + async_cow_submit, async_cow_free); nr_pages = (cur_end - start + PAGE_CACHE_SIZE) >> PAGE_CACHE_SHIFT; atomic_add(nr_pages, &root->fs_info->async_delalloc_pages); - btrfs_queue_worker(&root->fs_info->delalloc_workers, - &async_cow->work); + btrfs_queue_work(root->fs_info->delalloc_workers, + &async_cow->work); if (atomic_read(&root->fs_info->async_delalloc_pages) > limit) { wait_event(root->fs_info->async_submit_wait, @@ -985,7 +1122,7 @@ static noinline int csum_exist_in_range(struct btrfs_root *root, LIST_HEAD(list); ret = btrfs_lookup_csums_range(root->fs_info->csum_root, bytenr, - bytenr + num_bytes - 1, &list); + bytenr + num_bytes - 1, &list, 0); if (ret == 0 && list_empty(&list)) return 0; @@ -1021,34 +1158,61 @@ static noinline int run_delalloc_nocow(struct inode *inode, u64 extent_offset; u64 disk_bytenr; u64 num_bytes; + u64 disk_num_bytes; + u64 ram_bytes; int extent_type; - int ret; + int ret, err; int type; int nocow; int check_prev = 1; - bool nolock = false; + bool nolock; + u64 ino = btrfs_ino(inode); path = btrfs_alloc_path(); - BUG_ON(!path); - if (root == root->fs_info->tree_root) { - nolock = true; - trans = btrfs_join_transaction_nolock(root, 1); - } else { - trans = btrfs_join_transaction(root, 1); + if (!path) { + extent_clear_unlock_delalloc(inode, start, end, locked_page, + EXTENT_LOCKED | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING | + EXTENT_DEFRAG, PAGE_UNLOCK | + PAGE_CLEAR_DIRTY | + PAGE_SET_WRITEBACK | + PAGE_END_WRITEBACK); + return -ENOMEM; + } + + nolock = btrfs_is_free_space_inode(inode); + + if (nolock) + trans = btrfs_join_transaction_nolock(root); + else + trans = btrfs_join_transaction(root); + + if (IS_ERR(trans)) { + extent_clear_unlock_delalloc(inode, start, end, locked_page, + EXTENT_LOCKED | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING | + EXTENT_DEFRAG, PAGE_UNLOCK | + PAGE_CLEAR_DIRTY | + PAGE_SET_WRITEBACK | + PAGE_END_WRITEBACK); + btrfs_free_path(path); + return PTR_ERR(trans); } - BUG_ON(!trans); + + trans->block_rsv = &root->fs_info->delalloc_block_rsv; cow_start = (u64)-1; cur_offset = start; while (1) { - ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino, + ret = btrfs_lookup_file_extent(trans, root, path, ino, cur_offset, 0); - BUG_ON(ret < 0); + if (ret < 0) + goto error; if (ret > 0 && path->slots[0] > 0 && check_prev) { leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0] - 1); - if (found_key.objectid == inode->i_ino && + if (found_key.objectid == ino && found_key.type == BTRFS_EXTENT_DATA_KEY) path->slots[0]--; } @@ -1058,7 +1222,7 @@ next_slot: if (path->slots[0] >= btrfs_header_nritems(leaf)) { ret = btrfs_next_leaf(root, path); if (ret < 0) - BUG_ON(1); + goto error; if (ret > 0) break; leaf = path->nodes[0]; @@ -1069,7 +1233,7 @@ next_slot: num_bytes = 0; btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - if (found_key.objectid > inode->i_ino || + if (found_key.objectid > ino || found_key.type > BTRFS_EXTENT_DATA_KEY || found_key.offset > end) break; @@ -1084,12 +1248,15 @@ next_slot: struct btrfs_file_extent_item); extent_type = btrfs_file_extent_type(leaf, fi); + ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi); if (extent_type == BTRFS_FILE_EXTENT_REG || extent_type == BTRFS_FILE_EXTENT_PREALLOC) { disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); extent_offset = btrfs_file_extent_offset(leaf, fi); extent_end = found_key.offset + btrfs_file_extent_num_bytes(leaf, fi); + disk_num_bytes = + btrfs_file_extent_disk_num_bytes(leaf, fi); if (extent_end <= start) { path->slots[0]++; goto next_slot; @@ -1104,7 +1271,7 @@ next_slot: goto out_check; if (btrfs_extent_readonly(root, disk_bytenr)) goto out_check; - if (btrfs_cross_ref_exist(trans, root, inode->i_ino, + if (btrfs_cross_ref_exist(trans, root, ino, found_key.offset - extent_offset, disk_bytenr)) goto out_check; @@ -1112,6 +1279,15 @@ next_slot: disk_bytenr += cur_offset - found_key.offset; num_bytes = min(end + 1, extent_end) - cur_offset; /* + * if there are pending snapshots for this root, + * we fall into common COW way. + */ + if (!nolock) { + err = btrfs_start_nocow_write(root); + if (!err) + goto out_check; + } + /* * force cow if csum exists in the range. * this ensure that csum for a given extent are * either valid or do not exist. @@ -1121,7 +1297,8 @@ next_slot: nocow = 1; } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { extent_end = found_key.offset + - btrfs_file_extent_inline_len(leaf, fi); + btrfs_file_extent_inline_len(leaf, + path->slots[0], fi); extent_end = ALIGN(extent_end, root->sectorsize); } else { BUG_ON(1); @@ -1129,6 +1306,8 @@ next_slot: out_check: if (extent_end <= start) { path->slots[0]++; + if (!nolock && nocow) + btrfs_end_nocow_write(root); goto next_slot; } if (!nocow) { @@ -1141,12 +1320,16 @@ out_check: goto next_slot; } - btrfs_release_path(root, path); + btrfs_release_path(path); if (cow_start != (u64)-1) { - ret = cow_file_range(inode, locked_page, cow_start, - found_key.offset - 1, page_started, - nr_written, 1); - BUG_ON(ret); + ret = cow_file_range(inode, locked_page, + cow_start, found_key.offset - 1, + page_started, nr_written, 1); + if (ret) { + if (!nolock && nocow) + btrfs_end_nocow_write(root); + goto error; + } cow_start = (u64)-1; } @@ -1154,17 +1337,24 @@ out_check: struct extent_map *em; struct extent_map_tree *em_tree; em_tree = &BTRFS_I(inode)->extent_tree; - em = alloc_extent_map(GFP_NOFS); + em = alloc_extent_map(); + BUG_ON(!em); /* -ENOMEM */ em->start = cur_offset; - em->orig_start = em->start; + em->orig_start = found_key.offset - extent_offset; em->len = num_bytes; em->block_len = num_bytes; em->block_start = disk_bytenr; + em->orig_block_len = disk_num_bytes; + em->ram_bytes = ram_bytes; em->bdev = root->fs_info->fs_devices->latest_bdev; + em->mod_start = em->start; + em->mod_len = em->len; set_bit(EXTENT_FLAG_PINNED, &em->flags); + set_bit(EXTENT_FLAG_FILLING, &em->flags); + em->generation = -1; while (1) { write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, em); + ret = add_extent_mapping(em_tree, em, 1); write_unlock(&em_tree->lock); if (ret != -EEXIST) { free_extent_map(em); @@ -1180,43 +1370,59 @@ out_check: ret = btrfs_add_ordered_extent(inode, cur_offset, disk_bytenr, num_bytes, num_bytes, type); - BUG_ON(ret); + BUG_ON(ret); /* -ENOMEM */ if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID) { ret = btrfs_reloc_clone_csums(inode, cur_offset, num_bytes); - BUG_ON(ret); + if (ret) { + if (!nolock && nocow) + btrfs_end_nocow_write(root); + goto error; + } } - extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree, - cur_offset, cur_offset + num_bytes - 1, - locked_page, EXTENT_CLEAR_UNLOCK_PAGE | - EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC | - EXTENT_SET_PRIVATE2); + extent_clear_unlock_delalloc(inode, cur_offset, + cur_offset + num_bytes - 1, + locked_page, EXTENT_LOCKED | + EXTENT_DELALLOC, PAGE_UNLOCK | + PAGE_SET_PRIVATE2); + if (!nolock && nocow) + btrfs_end_nocow_write(root); cur_offset = extent_end; if (cur_offset > end) break; } - btrfs_release_path(root, path); + btrfs_release_path(path); - if (cur_offset <= end && cow_start == (u64)-1) + if (cur_offset <= end && cow_start == (u64)-1) { cow_start = cur_offset; + cur_offset = end; + } + if (cow_start != (u64)-1) { ret = cow_file_range(inode, locked_page, cow_start, end, page_started, nr_written, 1); - BUG_ON(ret); + if (ret) + goto error; } - if (nolock) { - ret = btrfs_end_transaction_nolock(trans, root); - BUG_ON(ret); - } else { - ret = btrfs_end_transaction(trans, root); - BUG_ON(ret); - } +error: + err = btrfs_end_transaction(trans, root); + if (!ret) + ret = err; + + if (ret && cur_offset < end) + extent_clear_unlock_delalloc(inode, cur_offset, end, + locked_page, EXTENT_LOCKED | + EXTENT_DELALLOC | EXTENT_DEFRAG | + EXTENT_DO_ACCOUNTING, PAGE_UNLOCK | + PAGE_CLEAR_DIRTY | + PAGE_SET_WRITEBACK | + PAGE_END_WRITEBACK); btrfs_free_path(path); - return 0; + return ret; } /* @@ -1229,31 +1435,36 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page, int ret; struct btrfs_root *root = BTRFS_I(inode)->root; - if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW) + if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW) { ret = run_delalloc_nocow(inode, locked_page, start, end, page_started, 1, nr_written); - else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC) + } else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC) { ret = run_delalloc_nocow(inode, locked_page, start, end, page_started, 0, nr_written); - else if (!btrfs_test_opt(root, COMPRESS) && - !(BTRFS_I(inode)->force_compress)) + } else if (!btrfs_test_opt(root, COMPRESS) && + !(BTRFS_I(inode)->force_compress) && + !(BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS)) { ret = cow_file_range(inode, locked_page, start, end, page_started, nr_written, 1); - else + } else { + set_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, + &BTRFS_I(inode)->runtime_flags); ret = cow_file_range_async(inode, locked_page, start, end, page_started, nr_written); + } return ret; } -static int btrfs_split_extent_hook(struct inode *inode, - struct extent_state *orig, u64 split) +static void btrfs_split_extent_hook(struct inode *inode, + struct extent_state *orig, u64 split) { /* not delalloc, ignore it */ if (!(orig->state & EXTENT_DELALLOC)) - return 0; + return; - atomic_inc(&BTRFS_I(inode)->outstanding_extents); - return 0; + spin_lock(&BTRFS_I(inode)->lock); + BTRFS_I(inode)->outstanding_extents++; + spin_unlock(&BTRFS_I(inode)->lock); } /* @@ -1262,16 +1473,57 @@ static int btrfs_split_extent_hook(struct inode *inode, * extents, such as when we are doing sequential writes, so we can properly * account for the metadata space we'll need. */ -static int btrfs_merge_extent_hook(struct inode *inode, - struct extent_state *new, - struct extent_state *other) +static void btrfs_merge_extent_hook(struct inode *inode, + struct extent_state *new, + struct extent_state *other) { /* not delalloc, ignore it */ if (!(other->state & EXTENT_DELALLOC)) - return 0; + return; - atomic_dec(&BTRFS_I(inode)->outstanding_extents); - return 0; + spin_lock(&BTRFS_I(inode)->lock); + BTRFS_I(inode)->outstanding_extents--; + spin_unlock(&BTRFS_I(inode)->lock); +} + +static void btrfs_add_delalloc_inodes(struct btrfs_root *root, + struct inode *inode) +{ + spin_lock(&root->delalloc_lock); + if (list_empty(&BTRFS_I(inode)->delalloc_inodes)) { + list_add_tail(&BTRFS_I(inode)->delalloc_inodes, + &root->delalloc_inodes); + set_bit(BTRFS_INODE_IN_DELALLOC_LIST, + &BTRFS_I(inode)->runtime_flags); + root->nr_delalloc_inodes++; + if (root->nr_delalloc_inodes == 1) { + spin_lock(&root->fs_info->delalloc_root_lock); + BUG_ON(!list_empty(&root->delalloc_root)); + list_add_tail(&root->delalloc_root, + &root->fs_info->delalloc_roots); + spin_unlock(&root->fs_info->delalloc_root_lock); + } + } + spin_unlock(&root->delalloc_lock); +} + +static void btrfs_del_delalloc_inode(struct btrfs_root *root, + struct inode *inode) +{ + spin_lock(&root->delalloc_lock); + if (!list_empty(&BTRFS_I(inode)->delalloc_inodes)) { + list_del_init(&BTRFS_I(inode)->delalloc_inodes); + clear_bit(BTRFS_INODE_IN_DELALLOC_LIST, + &BTRFS_I(inode)->runtime_flags); + root->nr_delalloc_inodes--; + if (!root->nr_delalloc_inodes) { + spin_lock(&root->fs_info->delalloc_root_lock); + BUG_ON(list_empty(&root->delalloc_root)); + list_del_init(&root->delalloc_root); + spin_unlock(&root->fs_info->delalloc_root_lock); + } + } + spin_unlock(&root->delalloc_lock); } /* @@ -1279,91 +1531,99 @@ static int btrfs_merge_extent_hook(struct inode *inode, * bytes in this file, and to maintain the list of inodes that * have pending delalloc work to be done. */ -static int btrfs_set_bit_hook(struct inode *inode, - struct extent_state *state, int *bits) +static void btrfs_set_bit_hook(struct inode *inode, + struct extent_state *state, unsigned long *bits) { /* * set_bit and clear bit hooks normally require _irqsave/restore - * but in this case, we are only testeing for the DELALLOC + * but in this case, we are only testing for the DELALLOC * bit, which is only set or cleared with irqs on */ if (!(state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) { struct btrfs_root *root = BTRFS_I(inode)->root; u64 len = state->end + 1 - state->start; - int do_list = (root->root_key.objectid != - BTRFS_ROOT_TREE_OBJECTID); + bool do_list = !btrfs_is_free_space_inode(inode); - if (*bits & EXTENT_FIRST_DELALLOC) + if (*bits & EXTENT_FIRST_DELALLOC) { *bits &= ~EXTENT_FIRST_DELALLOC; - else - atomic_inc(&BTRFS_I(inode)->outstanding_extents); + } else { + spin_lock(&BTRFS_I(inode)->lock); + BTRFS_I(inode)->outstanding_extents++; + spin_unlock(&BTRFS_I(inode)->lock); + } - spin_lock(&root->fs_info->delalloc_lock); + __percpu_counter_add(&root->fs_info->delalloc_bytes, len, + root->fs_info->delalloc_batch); + spin_lock(&BTRFS_I(inode)->lock); BTRFS_I(inode)->delalloc_bytes += len; - root->fs_info->delalloc_bytes += len; - if (do_list && list_empty(&BTRFS_I(inode)->delalloc_inodes)) { - list_add_tail(&BTRFS_I(inode)->delalloc_inodes, - &root->fs_info->delalloc_inodes); - } - spin_unlock(&root->fs_info->delalloc_lock); + if (do_list && !test_bit(BTRFS_INODE_IN_DELALLOC_LIST, + &BTRFS_I(inode)->runtime_flags)) + btrfs_add_delalloc_inodes(root, inode); + spin_unlock(&BTRFS_I(inode)->lock); } - return 0; } /* * extent_io.c clear_bit_hook, see set_bit_hook for why */ -static int btrfs_clear_bit_hook(struct inode *inode, - struct extent_state *state, int *bits) +static void btrfs_clear_bit_hook(struct inode *inode, + struct extent_state *state, + unsigned long *bits) { /* * set_bit and clear bit hooks normally require _irqsave/restore - * but in this case, we are only testeing for the DELALLOC + * but in this case, we are only testing for the DELALLOC * bit, which is only set or cleared with irqs on */ if ((state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) { struct btrfs_root *root = BTRFS_I(inode)->root; u64 len = state->end + 1 - state->start; - int do_list = (root->root_key.objectid != - BTRFS_ROOT_TREE_OBJECTID); + bool do_list = !btrfs_is_free_space_inode(inode); - if (*bits & EXTENT_FIRST_DELALLOC) + if (*bits & EXTENT_FIRST_DELALLOC) { *bits &= ~EXTENT_FIRST_DELALLOC; - else if (!(*bits & EXTENT_DO_ACCOUNTING)) - atomic_dec(&BTRFS_I(inode)->outstanding_extents); + } else if (!(*bits & EXTENT_DO_ACCOUNTING)) { + spin_lock(&BTRFS_I(inode)->lock); + BTRFS_I(inode)->outstanding_extents--; + spin_unlock(&BTRFS_I(inode)->lock); + } - if (*bits & EXTENT_DO_ACCOUNTING) + /* + * We don't reserve metadata space for space cache inodes so we + * don't need to call dellalloc_release_metadata if there is an + * error. + */ + if (*bits & EXTENT_DO_ACCOUNTING && + root != root->fs_info->tree_root) btrfs_delalloc_release_metadata(inode, len); if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID - && do_list) + && do_list && !(state->state & EXTENT_NORESERVE)) btrfs_free_reserved_data_space(inode, len); - spin_lock(&root->fs_info->delalloc_lock); - root->fs_info->delalloc_bytes -= len; + __percpu_counter_add(&root->fs_info->delalloc_bytes, -len, + root->fs_info->delalloc_batch); + spin_lock(&BTRFS_I(inode)->lock); BTRFS_I(inode)->delalloc_bytes -= len; - if (do_list && BTRFS_I(inode)->delalloc_bytes == 0 && - !list_empty(&BTRFS_I(inode)->delalloc_inodes)) { - list_del_init(&BTRFS_I(inode)->delalloc_inodes); - } - spin_unlock(&root->fs_info->delalloc_lock); + test_bit(BTRFS_INODE_IN_DELALLOC_LIST, + &BTRFS_I(inode)->runtime_flags)) + btrfs_del_delalloc_inode(root, inode); + spin_unlock(&BTRFS_I(inode)->lock); } - return 0; } /* * extent_io.c merge_bio_hook, this must check the chunk tree to make sure * we don't create bios that span stripes or chunks */ -int btrfs_merge_bio_hook(struct page *page, unsigned long offset, +int btrfs_merge_bio_hook(int rw, struct page *page, unsigned long offset, size_t size, struct bio *bio, unsigned long bio_flags) { struct btrfs_root *root = BTRFS_I(page->mapping->host)->root; - struct btrfs_mapping_tree *map_tree; - u64 logical = (u64)bio->bi_sector << 9; + u64 logical = (u64)bio->bi_iter.bi_sector << 9; u64 length = 0; u64 map_length; int ret; @@ -1371,15 +1631,15 @@ int btrfs_merge_bio_hook(struct page *page, unsigned long offset, if (bio_flags & EXTENT_BIO_COMPRESSED) return 0; - length = bio->bi_size; - map_tree = &root->fs_info->mapping_tree; + length = bio->bi_iter.bi_size; map_length = length; - ret = btrfs_map_block(map_tree, READ, logical, + ret = btrfs_map_block(root->fs_info, rw, logical, &map_length, NULL, 0); - + /* Will always return 0 with map_multi == NULL */ + BUG_ON(ret < 0); if (map_length < length + size) return 1; - return ret; + return 0; } /* @@ -1399,7 +1659,7 @@ static int __btrfs_submit_bio_start(struct inode *inode, int rw, int ret = 0; ret = btrfs_csum_one_bio(root, inode, bio, 0, 0); - BUG_ON(ret); + BUG_ON(ret); /* -ENOMEM */ return 0; } @@ -1416,7 +1676,12 @@ static int __btrfs_submit_bio_done(struct inode *inode, int rw, struct bio *bio, u64 bio_offset) { struct btrfs_root *root = BTRFS_I(inode)->root; - return btrfs_map_bio(root, rw, bio, mirror_num, 1); + int ret; + + ret = btrfs_map_bio(root, rw, bio, mirror_num, 1); + if (ret) + bio_endio(bio, ret); + return ret; } /* @@ -1430,36 +1695,54 @@ static int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, struct btrfs_root *root = BTRFS_I(inode)->root; int ret = 0; int skip_sum; + int metadata = 0; + int async = !atomic_read(&BTRFS_I(inode)->sync_writers); skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM; - if (root == root->fs_info->tree_root) - ret = btrfs_bio_wq_end_io(root->fs_info, bio, 2); - else - ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0); - BUG_ON(ret); + if (btrfs_is_free_space_inode(inode)) + metadata = 2; if (!(rw & REQ_WRITE)) { + ret = btrfs_bio_wq_end_io(root->fs_info, bio, metadata); + if (ret) + goto out; + if (bio_flags & EXTENT_BIO_COMPRESSED) { - return btrfs_submit_compressed_read(inode, bio, - mirror_num, bio_flags); - } else if (!skip_sum) - btrfs_lookup_bio_sums(root, inode, bio, NULL); + ret = btrfs_submit_compressed_read(inode, bio, + mirror_num, + bio_flags); + goto out; + } else if (!skip_sum) { + ret = btrfs_lookup_bio_sums(root, inode, bio, NULL); + if (ret) + goto out; + } goto mapit; - } else if (!skip_sum) { + } else if (async && !skip_sum) { /* csum items have already been cloned */ if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID) goto mapit; /* we're doing a write, do the async checksumming */ - return btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info, + ret = btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info, inode, rw, bio, mirror_num, bio_flags, bio_offset, __btrfs_submit_bio_start, __btrfs_submit_bio_done); + goto out; + } else if (!skip_sum) { + ret = btrfs_csum_one_bio(root, inode, bio, 0, 0); + if (ret) + goto out; } mapit: - return btrfs_map_bio(root, rw, bio, mirror_num, 0); + ret = btrfs_map_bio(root, rw, bio, mirror_num, 0); + +out: + if (ret < 0) + bio_endio(bio, ret); + return ret; } /* @@ -1472,11 +1755,11 @@ static noinline int add_pending_csums(struct btrfs_trans_handle *trans, { struct btrfs_ordered_sum *sum; - btrfs_set_trans_block_group(trans, inode); - list_for_each_entry(sum, list, list) { + trans->adding_csums = 1; btrfs_csum_file_blocks(trans, BTRFS_I(inode)->root->fs_info->csum_root, sum); + trans->adding_csums = 0; } return 0; } @@ -1484,8 +1767,7 @@ static noinline int add_pending_csums(struct btrfs_trans_handle *trans, int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end, struct extent_state **cached_state) { - if ((end & (PAGE_CACHE_SIZE - 1)) == 0) - WARN_ON(1); + WARN_ON((end & (PAGE_CACHE_SIZE - 1)) == 0); return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end, cached_state, GFP_NOFS); } @@ -1505,6 +1787,7 @@ static void btrfs_writepage_fixup_worker(struct btrfs_work *work) struct inode *inode; u64 page_start; u64 page_end; + int ret; fixup = container_of(work, struct btrfs_writepage_fixup, work); page = fixup->page; @@ -1520,7 +1803,7 @@ again: page_end = page_offset(page) + PAGE_CACHE_SIZE - 1; lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, 0, - &cached_state, GFP_NOFS); + &cached_state); /* already ordered? We're done */ if (PagePrivate2(page)) @@ -1532,18 +1815,28 @@ again: page_end, &cached_state, GFP_NOFS); unlock_page(page); btrfs_start_ordered_extent(inode, ordered, 1); + btrfs_put_ordered_extent(ordered); goto again; } - BUG(); + ret = btrfs_delalloc_reserve_space(inode, PAGE_CACHE_SIZE); + if (ret) { + mapping_set_error(page->mapping, ret); + end_extent_writepage(page, ret, page_start, page_end); + ClearPageChecked(page); + goto out; + } + btrfs_set_extent_delalloc(inode, page_start, page_end, &cached_state); ClearPageChecked(page); + set_page_dirty(page); out: unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start, page_end, &cached_state, GFP_NOFS); out_page: unlock_page(page); page_cache_release(page); + kfree(fixup); } /* @@ -1576,10 +1869,10 @@ static int btrfs_writepage_start_hook(struct page *page, u64 start, u64 end) SetPageChecked(page); page_cache_get(page); - fixup->work.func = btrfs_writepage_fixup_worker; + btrfs_init_work(&fixup->work, btrfs_writepage_fixup_worker, NULL, NULL); fixup->page = page; - btrfs_queue_worker(&root->fs_info->fixup_workers, &fixup->work); - return -EAGAIN; + btrfs_queue_work(root->fs_info->fixup_workers, &fixup->work); + return -EBUSY; } static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, @@ -1594,13 +1887,12 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, struct btrfs_path *path; struct extent_buffer *leaf; struct btrfs_key ins; - u64 hint; + int extent_inserted = 0; int ret; path = btrfs_alloc_path(); - BUG_ON(!path); - - path->leave_spinning = 1; + if (!path) + return -ENOMEM; /* * we may be replacing one extent in the tree with another. @@ -1611,15 +1903,23 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, * the caller is expected to unpin it and allow it to be merged * with the others. */ - ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes, - &hint, 0); - BUG_ON(ret); + ret = __btrfs_drop_extents(trans, root, inode, path, file_pos, + file_pos + num_bytes, NULL, 0, + 1, sizeof(*fi), &extent_inserted); + if (ret) + goto out; - ins.objectid = inode->i_ino; - ins.offset = file_pos; - ins.type = BTRFS_EXTENT_DATA_KEY; - ret = btrfs_insert_empty_item(trans, root, path, &ins, sizeof(*fi)); - BUG_ON(ret); + if (!extent_inserted) { + ins.objectid = btrfs_ino(inode); + ins.offset = file_pos; + ins.type = BTRFS_EXTENT_DATA_KEY; + + path->leave_spinning = 1; + ret = btrfs_insert_empty_item(trans, root, path, &ins, + sizeof(*fi)); + if (ret) + goto out; + } leaf = path->nodes[0]; fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); @@ -1634,10 +1934,8 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, btrfs_set_file_extent_encryption(leaf, fi, encryption); btrfs_set_file_extent_other_encoding(leaf, fi, other_encoding); - btrfs_unlock_up_safe(path, 1); - btrfs_set_lock_blocking(leaf); - btrfs_mark_buffer_dirty(leaf); + btrfs_release_path(path); inode_add_bytes(inode, num_bytes); @@ -1646,112 +1944,849 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, ins.type = BTRFS_EXTENT_ITEM_KEY; ret = btrfs_alloc_reserved_file_extent(trans, root, root->root_key.objectid, - inode->i_ino, file_pos, &ins); - BUG_ON(ret); + btrfs_ino(inode), file_pos, &ins); +out: btrfs_free_path(path); + return ret; +} + +/* snapshot-aware defrag */ +struct sa_defrag_extent_backref { + struct rb_node node; + struct old_sa_defrag_extent *old; + u64 root_id; + u64 inum; + u64 file_pos; + u64 extent_offset; + u64 num_bytes; + u64 generation; +}; + +struct old_sa_defrag_extent { + struct list_head list; + struct new_sa_defrag_extent *new; + + u64 extent_offset; + u64 bytenr; + u64 offset; + u64 len; + int count; +}; + +struct new_sa_defrag_extent { + struct rb_root root; + struct list_head head; + struct btrfs_path *path; + struct inode *inode; + u64 file_pos; + u64 len; + u64 bytenr; + u64 disk_len; + u8 compress_type; +}; + +static int backref_comp(struct sa_defrag_extent_backref *b1, + struct sa_defrag_extent_backref *b2) +{ + if (b1->root_id < b2->root_id) + return -1; + else if (b1->root_id > b2->root_id) + return 1; + + if (b1->inum < b2->inum) + return -1; + else if (b1->inum > b2->inum) + return 1; + + if (b1->file_pos < b2->file_pos) + return -1; + else if (b1->file_pos > b2->file_pos) + return 1; + + /* + * [------------------------------] ===> (a range of space) + * |<--->| |<---->| =============> (fs/file tree A) + * |<---------------------------->| ===> (fs/file tree B) + * + * A range of space can refer to two file extents in one tree while + * refer to only one file extent in another tree. + * + * So we may process a disk offset more than one time(two extents in A) + * and locate at the same extent(one extent in B), then insert two same + * backrefs(both refer to the extent in B). + */ return 0; } +static void backref_insert(struct rb_root *root, + struct sa_defrag_extent_backref *backref) +{ + struct rb_node **p = &root->rb_node; + struct rb_node *parent = NULL; + struct sa_defrag_extent_backref *entry; + int ret; + + while (*p) { + parent = *p; + entry = rb_entry(parent, struct sa_defrag_extent_backref, node); + + ret = backref_comp(backref, entry); + if (ret < 0) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&backref->node, parent, p); + rb_insert_color(&backref->node, root); +} + /* - * helper function for btrfs_finish_ordered_io, this - * just reads in some of the csum leaves to prime them into ram - * before we start the transaction. It limits the amount of btree - * reads required while inside the transaction. + * Note the backref might has changed, and in this case we just return 0. */ +static noinline int record_one_backref(u64 inum, u64 offset, u64 root_id, + void *ctx) +{ + struct btrfs_file_extent_item *extent; + struct btrfs_fs_info *fs_info; + struct old_sa_defrag_extent *old = ctx; + struct new_sa_defrag_extent *new = old->new; + struct btrfs_path *path = new->path; + struct btrfs_key key; + struct btrfs_root *root; + struct sa_defrag_extent_backref *backref; + struct extent_buffer *leaf; + struct inode *inode = new->inode; + int slot; + int ret; + u64 extent_offset; + u64 num_bytes; + + if (BTRFS_I(inode)->root->root_key.objectid == root_id && + inum == btrfs_ino(inode)) + return 0; + + key.objectid = root_id; + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + + fs_info = BTRFS_I(inode)->root->fs_info; + root = btrfs_read_fs_root_no_name(fs_info, &key); + if (IS_ERR(root)) { + if (PTR_ERR(root) == -ENOENT) + return 0; + WARN_ON(1); + pr_debug("inum=%llu, offset=%llu, root_id=%llu\n", + inum, offset, root_id); + return PTR_ERR(root); + } + + key.objectid = inum; + key.type = BTRFS_EXTENT_DATA_KEY; + if (offset > (u64)-1 << 32) + key.offset = 0; + else + key.offset = offset; + + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (WARN_ON(ret < 0)) + return ret; + ret = 0; + + while (1) { + cond_resched(); + + leaf = path->nodes[0]; + slot = path->slots[0]; + + if (slot >= btrfs_header_nritems(leaf)) { + ret = btrfs_next_leaf(root, path); + if (ret < 0) { + goto out; + } else if (ret > 0) { + ret = 0; + goto out; + } + continue; + } + + path->slots[0]++; + + btrfs_item_key_to_cpu(leaf, &key, slot); + + if (key.objectid > inum) + goto out; + + if (key.objectid < inum || key.type != BTRFS_EXTENT_DATA_KEY) + continue; + + extent = btrfs_item_ptr(leaf, slot, + struct btrfs_file_extent_item); + + if (btrfs_file_extent_disk_bytenr(leaf, extent) != old->bytenr) + continue; + + /* + * 'offset' refers to the exact key.offset, + * NOT the 'offset' field in btrfs_extent_data_ref, ie. + * (key.offset - extent_offset). + */ + if (key.offset != offset) + continue; + + extent_offset = btrfs_file_extent_offset(leaf, extent); + num_bytes = btrfs_file_extent_num_bytes(leaf, extent); + + if (extent_offset >= old->extent_offset + old->offset + + old->len || extent_offset + num_bytes <= + old->extent_offset + old->offset) + continue; + break; + } + + backref = kmalloc(sizeof(*backref), GFP_NOFS); + if (!backref) { + ret = -ENOENT; + goto out; + } + + backref->root_id = root_id; + backref->inum = inum; + backref->file_pos = offset; + backref->num_bytes = num_bytes; + backref->extent_offset = extent_offset; + backref->generation = btrfs_file_extent_generation(leaf, extent); + backref->old = old; + backref_insert(&new->root, backref); + old->count++; +out: + btrfs_release_path(path); + WARN_ON(ret); + return ret; +} + +static noinline bool record_extent_backrefs(struct btrfs_path *path, + struct new_sa_defrag_extent *new) +{ + struct btrfs_fs_info *fs_info = BTRFS_I(new->inode)->root->fs_info; + struct old_sa_defrag_extent *old, *tmp; + int ret; + + new->path = path; + + list_for_each_entry_safe(old, tmp, &new->head, list) { + ret = iterate_inodes_from_logical(old->bytenr + + old->extent_offset, fs_info, + path, record_one_backref, + old); + if (ret < 0 && ret != -ENOENT) + return false; + + /* no backref to be processed for this extent */ + if (!old->count) { + list_del(&old->list); + kfree(old); + } + } + + if (list_empty(&new->head)) + return false; + + return true; +} + +static int relink_is_mergable(struct extent_buffer *leaf, + struct btrfs_file_extent_item *fi, + struct new_sa_defrag_extent *new) +{ + if (btrfs_file_extent_disk_bytenr(leaf, fi) != new->bytenr) + return 0; + + if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG) + return 0; + + if (btrfs_file_extent_compression(leaf, fi) != new->compress_type) + return 0; + + if (btrfs_file_extent_encryption(leaf, fi) || + btrfs_file_extent_other_encoding(leaf, fi)) + return 0; + + return 1; +} + +/* + * Note the backref might has changed, and in this case we just return 0. + */ +static noinline int relink_extent_backref(struct btrfs_path *path, + struct sa_defrag_extent_backref *prev, + struct sa_defrag_extent_backref *backref) +{ + struct btrfs_file_extent_item *extent; + struct btrfs_file_extent_item *item; + struct btrfs_ordered_extent *ordered; + struct btrfs_trans_handle *trans; + struct btrfs_fs_info *fs_info; + struct btrfs_root *root; + struct btrfs_key key; + struct extent_buffer *leaf; + struct old_sa_defrag_extent *old = backref->old; + struct new_sa_defrag_extent *new = old->new; + struct inode *src_inode = new->inode; + struct inode *inode; + struct extent_state *cached = NULL; + int ret = 0; + u64 start; + u64 len; + u64 lock_start; + u64 lock_end; + bool merge = false; + int index; + + if (prev && prev->root_id == backref->root_id && + prev->inum == backref->inum && + prev->file_pos + prev->num_bytes == backref->file_pos) + merge = true; + + /* step 1: get root */ + key.objectid = backref->root_id; + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + + fs_info = BTRFS_I(src_inode)->root->fs_info; + index = srcu_read_lock(&fs_info->subvol_srcu); + + root = btrfs_read_fs_root_no_name(fs_info, &key); + if (IS_ERR(root)) { + srcu_read_unlock(&fs_info->subvol_srcu, index); + if (PTR_ERR(root) == -ENOENT) + return 0; + return PTR_ERR(root); + } + + if (btrfs_root_readonly(root)) { + srcu_read_unlock(&fs_info->subvol_srcu, index); + return 0; + } + + /* step 2: get inode */ + key.objectid = backref->inum; + key.type = BTRFS_INODE_ITEM_KEY; + key.offset = 0; + + inode = btrfs_iget(fs_info->sb, &key, root, NULL); + if (IS_ERR(inode)) { + srcu_read_unlock(&fs_info->subvol_srcu, index); + return 0; + } + + srcu_read_unlock(&fs_info->subvol_srcu, index); + + /* step 3: relink backref */ + lock_start = backref->file_pos; + lock_end = backref->file_pos + backref->num_bytes - 1; + lock_extent_bits(&BTRFS_I(inode)->io_tree, lock_start, lock_end, + 0, &cached); + + ordered = btrfs_lookup_first_ordered_extent(inode, lock_end); + if (ordered) { + btrfs_put_ordered_extent(ordered); + goto out_unlock; + } + + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + goto out_unlock; + } + + key.objectid = backref->inum; + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = backref->file_pos; + + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) { + goto out_free_path; + } else if (ret > 0) { + ret = 0; + goto out_free_path; + } + + extent = btrfs_item_ptr(path->nodes[0], path->slots[0], + struct btrfs_file_extent_item); + + if (btrfs_file_extent_generation(path->nodes[0], extent) != + backref->generation) + goto out_free_path; + + btrfs_release_path(path); + + start = backref->file_pos; + if (backref->extent_offset < old->extent_offset + old->offset) + start += old->extent_offset + old->offset - + backref->extent_offset; + + len = min(backref->extent_offset + backref->num_bytes, + old->extent_offset + old->offset + old->len); + len -= max(backref->extent_offset, old->extent_offset + old->offset); + + ret = btrfs_drop_extents(trans, root, inode, start, + start + len, 1); + if (ret) + goto out_free_path; +again: + key.objectid = btrfs_ino(inode); + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = start; + + path->leave_spinning = 1; + if (merge) { + struct btrfs_file_extent_item *fi; + u64 extent_len; + struct btrfs_key found_key; + + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + if (ret < 0) + goto out_free_path; + + path->slots[0]--; + leaf = path->nodes[0]; + btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); + + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + extent_len = btrfs_file_extent_num_bytes(leaf, fi); + + if (extent_len + found_key.offset == start && + relink_is_mergable(leaf, fi, new)) { + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_len + len); + btrfs_mark_buffer_dirty(leaf); + inode_add_bytes(inode, len); + + ret = 1; + goto out_free_path; + } else { + merge = false; + btrfs_release_path(path); + goto again; + } + } + + ret = btrfs_insert_empty_item(trans, root, path, &key, + sizeof(*extent)); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto out_free_path; + } + + leaf = path->nodes[0]; + item = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_disk_bytenr(leaf, item, new->bytenr); + btrfs_set_file_extent_disk_num_bytes(leaf, item, new->disk_len); + btrfs_set_file_extent_offset(leaf, item, start - new->file_pos); + btrfs_set_file_extent_num_bytes(leaf, item, len); + btrfs_set_file_extent_ram_bytes(leaf, item, new->len); + btrfs_set_file_extent_generation(leaf, item, trans->transid); + btrfs_set_file_extent_type(leaf, item, BTRFS_FILE_EXTENT_REG); + btrfs_set_file_extent_compression(leaf, item, new->compress_type); + btrfs_set_file_extent_encryption(leaf, item, 0); + btrfs_set_file_extent_other_encoding(leaf, item, 0); + + btrfs_mark_buffer_dirty(leaf); + inode_add_bytes(inode, len); + btrfs_release_path(path); + + ret = btrfs_inc_extent_ref(trans, root, new->bytenr, + new->disk_len, 0, + backref->root_id, backref->inum, + new->file_pos, 0); /* start - extent_offset */ + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto out_free_path; + } + + ret = 1; +out_free_path: + btrfs_release_path(path); + path->leave_spinning = 0; + btrfs_end_transaction(trans, root); +out_unlock: + unlock_extent_cached(&BTRFS_I(inode)->io_tree, lock_start, lock_end, + &cached, GFP_NOFS); + iput(inode); + return ret; +} + +static void free_sa_defrag_extent(struct new_sa_defrag_extent *new) +{ + struct old_sa_defrag_extent *old, *tmp; + + if (!new) + return; + + list_for_each_entry_safe(old, tmp, &new->head, list) { + list_del(&old->list); + kfree(old); + } + kfree(new); +} + +static void relink_file_extents(struct new_sa_defrag_extent *new) +{ + struct btrfs_path *path; + struct sa_defrag_extent_backref *backref; + struct sa_defrag_extent_backref *prev = NULL; + struct inode *inode; + struct btrfs_root *root; + struct rb_node *node; + int ret; + + inode = new->inode; + root = BTRFS_I(inode)->root; + + path = btrfs_alloc_path(); + if (!path) + return; + + if (!record_extent_backrefs(path, new)) { + btrfs_free_path(path); + goto out; + } + btrfs_release_path(path); + + while (1) { + node = rb_first(&new->root); + if (!node) + break; + rb_erase(node, &new->root); + + backref = rb_entry(node, struct sa_defrag_extent_backref, node); + + ret = relink_extent_backref(path, prev, backref); + WARN_ON(ret < 0); + + kfree(prev); + + if (ret == 1) + prev = backref; + else + prev = NULL; + cond_resched(); + } + kfree(prev); + + btrfs_free_path(path); +out: + free_sa_defrag_extent(new); + + atomic_dec(&root->fs_info->defrag_running); + wake_up(&root->fs_info->transaction_wait); +} + +static struct new_sa_defrag_extent * +record_old_file_extents(struct inode *inode, + struct btrfs_ordered_extent *ordered) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_path *path; + struct btrfs_key key; + struct old_sa_defrag_extent *old; + struct new_sa_defrag_extent *new; + int ret; + + new = kmalloc(sizeof(*new), GFP_NOFS); + if (!new) + return NULL; + + new->inode = inode; + new->file_pos = ordered->file_offset; + new->len = ordered->len; + new->bytenr = ordered->start; + new->disk_len = ordered->disk_len; + new->compress_type = ordered->compress_type; + new->root = RB_ROOT; + INIT_LIST_HEAD(&new->head); + + path = btrfs_alloc_path(); + if (!path) + goto out_kfree; + + key.objectid = btrfs_ino(inode); + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = new->file_pos; + + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) + goto out_free_path; + if (ret > 0 && path->slots[0] > 0) + path->slots[0]--; + + /* find out all the old extents for the file range */ + while (1) { + struct btrfs_file_extent_item *extent; + struct extent_buffer *l; + int slot; + u64 num_bytes; + u64 offset; + u64 end; + u64 disk_bytenr; + u64 extent_offset; + + l = path->nodes[0]; + slot = path->slots[0]; + + if (slot >= btrfs_header_nritems(l)) { + ret = btrfs_next_leaf(root, path); + if (ret < 0) + goto out_free_path; + else if (ret > 0) + break; + continue; + } + + btrfs_item_key_to_cpu(l, &key, slot); + + if (key.objectid != btrfs_ino(inode)) + break; + if (key.type != BTRFS_EXTENT_DATA_KEY) + break; + if (key.offset >= new->file_pos + new->len) + break; + + extent = btrfs_item_ptr(l, slot, struct btrfs_file_extent_item); + + num_bytes = btrfs_file_extent_num_bytes(l, extent); + if (key.offset + num_bytes < new->file_pos) + goto next; + + disk_bytenr = btrfs_file_extent_disk_bytenr(l, extent); + if (!disk_bytenr) + goto next; + + extent_offset = btrfs_file_extent_offset(l, extent); + + old = kmalloc(sizeof(*old), GFP_NOFS); + if (!old) + goto out_free_path; + + offset = max(new->file_pos, key.offset); + end = min(new->file_pos + new->len, key.offset + num_bytes); + + old->bytenr = disk_bytenr; + old->extent_offset = extent_offset; + old->offset = offset - key.offset; + old->len = end - offset; + old->new = new; + old->count = 0; + list_add_tail(&old->list, &new->head); +next: + path->slots[0]++; + cond_resched(); + } + + btrfs_free_path(path); + atomic_inc(&root->fs_info->defrag_running); + + return new; + +out_free_path: + btrfs_free_path(path); +out_kfree: + free_sa_defrag_extent(new); + return NULL; +} + +static void btrfs_release_delalloc_bytes(struct btrfs_root *root, + u64 start, u64 len) +{ + struct btrfs_block_group_cache *cache; + + cache = btrfs_lookup_block_group(root->fs_info, start); + ASSERT(cache); + + spin_lock(&cache->lock); + cache->delalloc_bytes -= len; + spin_unlock(&cache->lock); + + btrfs_put_block_group(cache); +} + /* as ordered data IO finishes, this gets called so we can finish * an ordered extent if the range of bytes in the file it covers are * fully written. */ -static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) +static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent) { + struct inode *inode = ordered_extent->inode; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans = NULL; - struct btrfs_ordered_extent *ordered_extent = NULL; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_state *cached_state = NULL; - int compressed = 0; - int ret; - bool nolock = false; + struct new_sa_defrag_extent *new = NULL; + int compress_type = 0; + int ret = 0; + u64 logical_len = ordered_extent->len; + bool nolock; + bool truncated = false; - ret = btrfs_dec_test_ordered_pending(inode, &ordered_extent, start, - end - start + 1); - if (!ret) - return 0; - BUG_ON(!ordered_extent); + nolock = btrfs_is_free_space_inode(inode); + + if (test_bit(BTRFS_ORDERED_IOERR, &ordered_extent->flags)) { + ret = -EIO; + goto out; + } - nolock = (root == root->fs_info->tree_root); + if (test_bit(BTRFS_ORDERED_TRUNCATED, &ordered_extent->flags)) { + truncated = true; + logical_len = ordered_extent->truncated_len; + /* Truncated the entire extent, don't bother adding */ + if (!logical_len) + goto out; + } if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) { - BUG_ON(!list_empty(&ordered_extent->list)); - ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent); - if (!ret) { - if (nolock) - trans = btrfs_join_transaction_nolock(root, 1); - else - trans = btrfs_join_transaction(root, 1); - BUG_ON(!trans); - btrfs_set_trans_block_group(trans, inode); - trans->block_rsv = &root->fs_info->delalloc_block_rsv; - ret = btrfs_update_inode(trans, root, inode); - BUG_ON(ret); + BUG_ON(!list_empty(&ordered_extent->list)); /* Logic error */ + btrfs_ordered_update_i_size(inode, 0, ordered_extent); + if (nolock) + trans = btrfs_join_transaction_nolock(root); + else + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + trans = NULL; + goto out; } + trans->block_rsv = &root->fs_info->delalloc_block_rsv; + ret = btrfs_update_inode_fallback(trans, root, inode); + if (ret) /* -ENOMEM or corruption */ + btrfs_abort_transaction(trans, root, ret); goto out; } lock_extent_bits(io_tree, ordered_extent->file_offset, ordered_extent->file_offset + ordered_extent->len - 1, - 0, &cached_state, GFP_NOFS); + 0, &cached_state); + + ret = test_range_bit(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + ordered_extent->len - 1, + EXTENT_DEFRAG, 1, cached_state); + if (ret) { + u64 last_snapshot = btrfs_root_last_snapshot(&root->root_item); + if (0 && last_snapshot >= BTRFS_I(inode)->generation) + /* the inode is shared */ + new = record_old_file_extents(inode, ordered_extent); + + clear_extent_bit(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + ordered_extent->len - 1, + EXTENT_DEFRAG, 0, 0, &cached_state, GFP_NOFS); + } if (nolock) - trans = btrfs_join_transaction_nolock(root, 1); + trans = btrfs_join_transaction_nolock(root); else - trans = btrfs_join_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + trans = NULL; + goto out_unlock; + } + trans->block_rsv = &root->fs_info->delalloc_block_rsv; if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags)) - compressed = 1; + compress_type = ordered_extent->compress_type; if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) { - BUG_ON(compressed); + BUG_ON(compress_type); ret = btrfs_mark_extent_written(trans, inode, ordered_extent->file_offset, ordered_extent->file_offset + - ordered_extent->len); - BUG_ON(ret); + logical_len); } else { BUG_ON(root == root->fs_info->tree_root); ret = insert_reserved_file_extent(trans, inode, ordered_extent->file_offset, ordered_extent->start, ordered_extent->disk_len, - ordered_extent->len, - ordered_extent->len, - compressed, 0, 0, + logical_len, logical_len, + compress_type, 0, 0, BTRFS_FILE_EXTENT_REG); - unpin_extent_cache(&BTRFS_I(inode)->extent_tree, - ordered_extent->file_offset, - ordered_extent->len); - BUG_ON(ret); + if (!ret) + btrfs_release_delalloc_bytes(root, + ordered_extent->start, + ordered_extent->disk_len); + } + unpin_extent_cache(&BTRFS_I(inode)->extent_tree, + ordered_extent->file_offset, ordered_extent->len, + trans->transid); + if (ret < 0) { + btrfs_abort_transaction(trans, root, ret); + goto out_unlock; } - unlock_extent_cached(io_tree, ordered_extent->file_offset, - ordered_extent->file_offset + - ordered_extent->len - 1, &cached_state, GFP_NOFS); add_pending_csums(trans, inode, ordered_extent->file_offset, &ordered_extent->list); btrfs_ordered_update_i_size(inode, 0, ordered_extent); - ret = btrfs_update_inode(trans, root, inode); - BUG_ON(ret); + ret = btrfs_update_inode_fallback(trans, root, inode); + if (ret) { /* -ENOMEM or corruption */ + btrfs_abort_transaction(trans, root, ret); + goto out_unlock; + } + ret = 0; +out_unlock: + unlock_extent_cached(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + + ordered_extent->len - 1, &cached_state, GFP_NOFS); out: - if (nolock) { - if (trans) - btrfs_end_transaction_nolock(trans, root); - } else { + if (root != root->fs_info->tree_root) btrfs_delalloc_release_metadata(inode, ordered_extent->len); - if (trans) - btrfs_end_transaction(trans, root); + if (trans) + btrfs_end_transaction(trans, root); + + if (ret || truncated) { + u64 start, end; + + if (truncated) + start = ordered_extent->file_offset + logical_len; + else + start = ordered_extent->file_offset; + end = ordered_extent->file_offset + ordered_extent->len - 1; + clear_extent_uptodate(io_tree, start, end, NULL, GFP_NOFS); + + /* Drop the cache for the part of the extent we didn't write. */ + btrfs_drop_extent_cache(inode, start, end, 0); + + /* + * If the ordered extent had an IOERR or something else went + * wrong we need to return the space for this ordered extent + * back to the allocator. We only free the extent in the + * truncated case if we didn't write out the extent at all. + */ + if ((ret || !logical_len) && + !test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags) && + !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) + btrfs_free_reserved_extent(root, ordered_extent->start, + ordered_extent->disk_len, 1); + } + + + /* + * This needs to be done to make sure anybody waiting knows we are done + * updating everything for this ordered extent. + */ + btrfs_remove_ordered_extent(inode, ordered_extent); + + /* for snapshot-aware defrag */ + if (new) { + if (ret) { + free_sa_defrag_extent(new); + atomic_dec(&root->fs_info->defrag_running); + } else { + relink_file_extents(new); + } } /* once for us */ @@ -1759,174 +2794,60 @@ out: /* once for the tree */ btrfs_put_ordered_extent(ordered_extent); - return 0; + return ret; } -static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end, - struct extent_state *state, int uptodate) +static void finish_ordered_fn(struct btrfs_work *work) { - ClearPagePrivate2(page); - return btrfs_finish_ordered_io(page->mapping->host, start, end); + struct btrfs_ordered_extent *ordered_extent; + ordered_extent = container_of(work, struct btrfs_ordered_extent, work); + btrfs_finish_ordered_io(ordered_extent); } -/* - * When IO fails, either with EIO or csum verification fails, we - * try other mirrors that might have a good copy of the data. This - * io_failure_record is used to record state as we go through all the - * mirrors. If another mirror has good data, the page is set up to date - * and things continue. If a good mirror can't be found, the original - * bio end_io callback is called to indicate things have failed. - */ -struct io_failure_record { - struct page *page; - u64 start; - u64 len; - u64 logical; - unsigned long bio_flags; - int last_mirror; -}; - -static int btrfs_io_failed_hook(struct bio *failed_bio, - struct page *page, u64 start, u64 end, - struct extent_state *state) +static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end, + struct extent_state *state, int uptodate) { - struct io_failure_record *failrec = NULL; - u64 private; - struct extent_map *em; struct inode *inode = page->mapping->host; - struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree; - struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; - struct bio *bio; - int num_copies; - int ret; - int rw; - u64 logical; - - ret = get_state_private(failure_tree, start, &private); - if (ret) { - failrec = kmalloc(sizeof(*failrec), GFP_NOFS); - if (!failrec) - return -ENOMEM; - failrec->start = start; - failrec->len = end - start + 1; - failrec->last_mirror = 0; - failrec->bio_flags = 0; - - read_lock(&em_tree->lock); - em = lookup_extent_mapping(em_tree, start, failrec->len); - if (em->start > start || em->start + em->len < start) { - free_extent_map(em); - em = NULL; - } - read_unlock(&em_tree->lock); + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_ordered_extent *ordered_extent = NULL; + struct btrfs_workqueue *workers; - if (!em || IS_ERR(em)) { - kfree(failrec); - return -EIO; - } - logical = start - em->start; - logical = em->block_start + logical; - if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) { - logical = em->block_start; - failrec->bio_flags = EXTENT_BIO_COMPRESSED; - } - failrec->logical = logical; - free_extent_map(em); - set_extent_bits(failure_tree, start, end, EXTENT_LOCKED | - EXTENT_DIRTY, GFP_NOFS); - set_state_private(failure_tree, start, - (u64)(unsigned long)failrec); - } else { - failrec = (struct io_failure_record *)(unsigned long)private; - } - num_copies = btrfs_num_copies( - &BTRFS_I(inode)->root->fs_info->mapping_tree, - failrec->logical, failrec->len); - failrec->last_mirror++; - if (!state) { - spin_lock(&BTRFS_I(inode)->io_tree.lock); - state = find_first_extent_bit_state(&BTRFS_I(inode)->io_tree, - failrec->start, - EXTENT_LOCKED); - if (state && state->start != failrec->start) - state = NULL; - spin_unlock(&BTRFS_I(inode)->io_tree.lock); - } - if (!state || failrec->last_mirror > num_copies) { - set_state_private(failure_tree, failrec->start, 0); - clear_extent_bits(failure_tree, failrec->start, - failrec->start + failrec->len - 1, - EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS); - kfree(failrec); - return -EIO; - } - bio = bio_alloc(GFP_NOFS, 1); - bio->bi_private = state; - bio->bi_end_io = failed_bio->bi_end_io; - bio->bi_sector = failrec->logical >> 9; - bio->bi_bdev = failed_bio->bi_bdev; - bio->bi_size = 0; + trace_btrfs_writepage_end_io_hook(page, start, end, uptodate); - bio_add_page(bio, page, failrec->len, start - page_offset(page)); - if (failed_bio->bi_rw & REQ_WRITE) - rw = WRITE; - else - rw = READ; + ClearPagePrivate2(page); + if (!btrfs_dec_test_ordered_pending(inode, &ordered_extent, start, + end - start + 1, uptodate)) + return 0; - BTRFS_I(inode)->io_tree.ops->submit_bio_hook(inode, rw, bio, - failrec->last_mirror, - failrec->bio_flags, 0); - return 0; -} + btrfs_init_work(&ordered_extent->work, finish_ordered_fn, NULL, NULL); -/* - * each time an IO finishes, we do a fast check in the IO failure tree - * to see if we need to process or clean up an io_failure_record - */ -static int btrfs_clean_io_failures(struct inode *inode, u64 start) -{ - u64 private; - u64 private_failure; - struct io_failure_record *failure; - int ret; + if (btrfs_is_free_space_inode(inode)) + workers = root->fs_info->endio_freespace_worker; + else + workers = root->fs_info->endio_write_workers; + btrfs_queue_work(workers, &ordered_extent->work); - private = 0; - if (count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private, - (u64)-1, 1, EXTENT_DIRTY)) { - ret = get_state_private(&BTRFS_I(inode)->io_failure_tree, - start, &private_failure); - if (ret == 0) { - failure = (struct io_failure_record *)(unsigned long) - private_failure; - set_state_private(&BTRFS_I(inode)->io_failure_tree, - failure->start, 0); - clear_extent_bits(&BTRFS_I(inode)->io_failure_tree, - failure->start, - failure->start + failure->len - 1, - EXTENT_DIRTY | EXTENT_LOCKED, - GFP_NOFS); - kfree(failure); - } - } return 0; } /* * when reads are done, we need to check csums to verify the data is correct - * if there's a match, we allow the bio to finish. If not, we go through - * the io_failure_record routines to find good copies + * if there's a match, we allow the bio to finish. If not, the code in + * extent_io.c will try to find good copies for us. */ -static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end, - struct extent_state *state) +static int btrfs_readpage_end_io_hook(struct btrfs_io_bio *io_bio, + u64 phy_offset, struct page *page, + u64 start, u64 end, int mirror) { - size_t offset = start - ((u64)page->index << PAGE_CACHE_SHIFT); + size_t offset = start - page_offset(page); struct inode *inode = page->mapping->host; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; char *kaddr; - u64 private = ~(u32)0; - int ret; struct btrfs_root *root = BTRFS_I(inode)->root; + u32 csum_expected; u32 csum = ~(u32)0; + static DEFINE_RATELIMIT_STATE(_rs, DEFAULT_RATELIMIT_INTERVAL, + DEFAULT_RATELIMIT_BURST); if (PageChecked(page)) { ClearPageChecked(page); @@ -1934,7 +2855,7 @@ static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end, } if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM) - return 0; + goto good; if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID && test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1, NULL)) { @@ -1943,40 +2864,27 @@ static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end, return 0; } - if (state && state->start == start) { - private = state->private; - ret = 0; - } else { - ret = get_state_private(io_tree, start, &private); - } - kaddr = kmap_atomic(page, KM_USER0); - if (ret) - goto zeroit; + phy_offset >>= inode->i_sb->s_blocksize_bits; + csum_expected = *(((u32 *)io_bio->csum) + phy_offset); - csum = btrfs_csum_data(root, kaddr + offset, csum, end - start + 1); + kaddr = kmap_atomic(page); + csum = btrfs_csum_data(kaddr + offset, csum, end - start + 1); btrfs_csum_final(csum, (char *)&csum); - if (csum != private) + if (csum != csum_expected) goto zeroit; - kunmap_atomic(kaddr, KM_USER0); + kunmap_atomic(kaddr); good: - /* if the io failure tree for this inode is non-empty, - * check to see if we've recovered from a failed IO - */ - btrfs_clean_io_failures(inode, start); return 0; zeroit: - if (printk_ratelimit()) { - printk(KERN_INFO "btrfs csum failed ino %lu off %llu csum %u " - "private %llu\n", page->mapping->host->i_ino, - (unsigned long long)start, csum, - (unsigned long long)private); - } + if (__ratelimit(&_rs)) + btrfs_info(root->fs_info, "csum failed ino %llu off %llu csum %u expected csum %u", + btrfs_ino(page->mapping->host), start, csum, csum_expected); memset(kaddr + offset, 1, end - start + 1); flush_dcache_page(page); - kunmap_atomic(kaddr, KM_USER0); - if (private == 0) + kunmap_atomic(kaddr); + if (csum_expected == 0) return 0; return -EIO; } @@ -1986,6 +2894,8 @@ struct delayed_iput { struct inode *inode; }; +/* JDM: If this is fs-wide, why can't we add a pointer to + * btrfs_inode instead and avoid the allocation? */ void btrfs_add_delayed_iput(struct inode *inode) { struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; @@ -2015,7 +2925,6 @@ void btrfs_run_delayed_iputs(struct btrfs_root *root) if (empty) return; - down_read(&root->fs_info->cleanup_work_sem); spin_lock(&fs_info->delayed_iput_lock); list_splice_init(&fs_info->delayed_iputs, &list); spin_unlock(&fs_info->delayed_iput_lock); @@ -2026,123 +2935,52 @@ void btrfs_run_delayed_iputs(struct btrfs_root *root) iput(delayed->inode); kfree(delayed); } - up_read(&root->fs_info->cleanup_work_sem); } /* - * calculate extra metadata reservation when snapshotting a subvolume - * contains orphan files. + * This is called in transaction commit time. If there are no orphan + * files in the subvolume, it removes orphan item and frees block_rsv + * structure. */ -void btrfs_orphan_pre_snapshot(struct btrfs_trans_handle *trans, - struct btrfs_pending_snapshot *pending, - u64 *bytes_to_reserve) +void btrfs_orphan_commit_root(struct btrfs_trans_handle *trans, + struct btrfs_root *root) { - struct btrfs_root *root; struct btrfs_block_rsv *block_rsv; - u64 num_bytes; - int index; + int ret; - root = pending->root; - if (!root->orphan_block_rsv || list_empty(&root->orphan_list)) + if (atomic_read(&root->orphan_inodes) || + root->orphan_cleanup_state != ORPHAN_CLEANUP_DONE) return; - block_rsv = root->orphan_block_rsv; - - /* orphan block reservation for the snapshot */ - num_bytes = block_rsv->size; - - /* - * after the snapshot is created, COWing tree blocks may use more - * space than it frees. So we should make sure there is enough - * reserved space. - */ - index = trans->transid & 0x1; - if (block_rsv->reserved + block_rsv->freed[index] < block_rsv->size) { - num_bytes += block_rsv->size - - (block_rsv->reserved + block_rsv->freed[index]); + spin_lock(&root->orphan_lock); + if (atomic_read(&root->orphan_inodes)) { + spin_unlock(&root->orphan_lock); + return; } - *bytes_to_reserve += num_bytes; -} - -void btrfs_orphan_post_snapshot(struct btrfs_trans_handle *trans, - struct btrfs_pending_snapshot *pending) -{ - struct btrfs_root *root = pending->root; - struct btrfs_root *snap = pending->snap; - struct btrfs_block_rsv *block_rsv; - u64 num_bytes; - int index; - int ret; - - if (!root->orphan_block_rsv || list_empty(&root->orphan_list)) + if (root->orphan_cleanup_state != ORPHAN_CLEANUP_DONE) { + spin_unlock(&root->orphan_lock); return; - - /* refill source subvolume's orphan block reservation */ - block_rsv = root->orphan_block_rsv; - index = trans->transid & 0x1; - if (block_rsv->reserved + block_rsv->freed[index] < block_rsv->size) { - num_bytes = block_rsv->size - - (block_rsv->reserved + block_rsv->freed[index]); - ret = btrfs_block_rsv_migrate(&pending->block_rsv, - root->orphan_block_rsv, - num_bytes); - BUG_ON(ret); } - /* setup orphan block reservation for the snapshot */ - block_rsv = btrfs_alloc_block_rsv(snap); - BUG_ON(!block_rsv); - - btrfs_add_durable_block_rsv(root->fs_info, block_rsv); - snap->orphan_block_rsv = block_rsv; - - num_bytes = root->orphan_block_rsv->size; - ret = btrfs_block_rsv_migrate(&pending->block_rsv, - block_rsv, num_bytes); - BUG_ON(ret); - -#if 0 - /* insert orphan item for the snapshot */ - WARN_ON(!root->orphan_item_inserted); - ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root, - snap->root_key.objectid); - BUG_ON(ret); - snap->orphan_item_inserted = 1; -#endif -} - -enum btrfs_orphan_cleanup_state { - ORPHAN_CLEANUP_STARTED = 1, - ORPHAN_CLEANUP_DONE = 2, -}; - -/* - * This is called in transaction commmit time. If there are no orphan - * files in the subvolume, it removes orphan item and frees block_rsv - * structure. - */ -void btrfs_orphan_commit_root(struct btrfs_trans_handle *trans, - struct btrfs_root *root) -{ - int ret; - - if (!list_empty(&root->orphan_list) || - root->orphan_cleanup_state != ORPHAN_CLEANUP_DONE) - return; + block_rsv = root->orphan_block_rsv; + root->orphan_block_rsv = NULL; + spin_unlock(&root->orphan_lock); - if (root->orphan_item_inserted && + if (test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state) && btrfs_root_refs(&root->root_item) > 0) { ret = btrfs_del_orphan_item(trans, root->fs_info->tree_root, root->root_key.objectid); - BUG_ON(ret); - root->orphan_item_inserted = 0; + if (ret) + btrfs_abort_transaction(trans, root, ret); + else + clear_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, + &root->state); } - if (root->orphan_block_rsv) { - WARN_ON(root->orphan_block_rsv->size > 0); - btrfs_free_block_rsv(root, root->orphan_block_rsv); - root->orphan_block_rsv = NULL; + if (block_rsv) { + WARN_ON(block_rsv->size > 0); + btrfs_free_block_rsv(root, block_rsv); } } @@ -2162,8 +3000,9 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) int ret; if (!root->orphan_block_rsv) { - block_rsv = btrfs_alloc_block_rsv(root); - BUG_ON(!block_rsv); + block_rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP); + if (!block_rsv) + return -ENOMEM; } spin_lock(&root->orphan_lock); @@ -2174,8 +3013,8 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) block_rsv = NULL; } - if (list_empty(&BTRFS_I(inode)->i_orphan)) { - list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list); + if (!test_and_set_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags)) { #if 0 /* * For proper ENOSPC handling, we should do orphan @@ -2188,36 +3027,48 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) insert = 1; #endif insert = 1; - } else { - WARN_ON(!BTRFS_I(inode)->orphan_meta_reserved); + atomic_inc(&root->orphan_inodes); } - if (!BTRFS_I(inode)->orphan_meta_reserved) { - BTRFS_I(inode)->orphan_meta_reserved = 1; + if (!test_and_set_bit(BTRFS_INODE_ORPHAN_META_RESERVED, + &BTRFS_I(inode)->runtime_flags)) reserve = 1; - } spin_unlock(&root->orphan_lock); - if (block_rsv) - btrfs_add_durable_block_rsv(root->fs_info, block_rsv); - /* grab metadata reservation from transaction handle */ if (reserve) { ret = btrfs_orphan_reserve_metadata(trans, inode); - BUG_ON(ret); + BUG_ON(ret); /* -ENOSPC in reservation; Logic error? JDM */ } /* insert an orphan item to track this unlinked/truncated file */ if (insert >= 1) { - ret = btrfs_insert_orphan_item(trans, root, inode->i_ino); - BUG_ON(ret); + ret = btrfs_insert_orphan_item(trans, root, btrfs_ino(inode)); + if (ret) { + atomic_dec(&root->orphan_inodes); + if (reserve) { + clear_bit(BTRFS_INODE_ORPHAN_META_RESERVED, + &BTRFS_I(inode)->runtime_flags); + btrfs_orphan_release_metadata(inode); + } + if (ret != -EEXIST) { + clear_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags); + btrfs_abort_transaction(trans, root, ret); + return ret; + } + } + ret = 0; } /* insert an orphan item to track subvolume contains orphan files */ if (insert >= 2) { ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root, root->root_key.objectid); - BUG_ON(ret); + if (ret && ret != -EEXIST) { + btrfs_abort_transaction(trans, root, ret); + return ret; + } } return 0; } @@ -2226,7 +3077,8 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) * We have done the truncate/delete so we can go ahead and remove the orphan * item for this particular inode. */ -int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode) +static int btrfs_orphan_del(struct btrfs_trans_handle *trans, + struct inode *inode) { struct btrfs_root *root = BTRFS_I(inode)->root; int delete_item = 0; @@ -2234,46 +3086,50 @@ int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode) int ret = 0; spin_lock(&root->orphan_lock); - if (!list_empty(&BTRFS_I(inode)->i_orphan)) { - list_del_init(&BTRFS_I(inode)->i_orphan); + if (test_and_clear_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags)) delete_item = 1; - } - if (BTRFS_I(inode)->orphan_meta_reserved) { - BTRFS_I(inode)->orphan_meta_reserved = 0; + if (test_and_clear_bit(BTRFS_INODE_ORPHAN_META_RESERVED, + &BTRFS_I(inode)->runtime_flags)) release_rsv = 1; - } spin_unlock(&root->orphan_lock); - if (trans && delete_item) { - ret = btrfs_del_orphan_item(trans, root, inode->i_ino); - BUG_ON(ret); + if (delete_item) { + atomic_dec(&root->orphan_inodes); + if (trans) + ret = btrfs_del_orphan_item(trans, root, + btrfs_ino(inode)); } if (release_rsv) btrfs_orphan_release_metadata(inode); - return 0; + return ret; } /* * this cleans up any orphans that may be left on the list from the last use * of this root. */ -void btrfs_orphan_cleanup(struct btrfs_root *root) +int btrfs_orphan_cleanup(struct btrfs_root *root) { struct btrfs_path *path; struct extent_buffer *leaf; struct btrfs_key key, found_key; struct btrfs_trans_handle *trans; struct inode *inode; + u64 last_objectid = 0; int ret = 0, nr_unlink = 0, nr_truncate = 0; if (cmpxchg(&root->orphan_cleanup_state, 0, ORPHAN_CLEANUP_STARTED)) - return; + return 0; path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) { + ret = -ENOMEM; + goto out; + } path->reada = -1; key.objectid = BTRFS_ORPHAN_OBJECTID; @@ -2282,18 +3138,16 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) while (1) { ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); - if (ret < 0) { - printk(KERN_ERR "Error searching slot for orphan: %d" - "\n", ret); - break; - } + if (ret < 0) + goto out; /* * if ret == 0 means we found what we were searching for, which - * is weird, but possible, so only screw with path if we didnt + * is weird, but possible, so only screw with path if we didn't * find the key and see if we have stuff that matches */ if (ret > 0) { + ret = 0; if (path->slots[0] == 0) break; path->slots[0]--; @@ -2310,53 +3164,127 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) break; /* release the path since we're done with it */ - btrfs_release_path(root, path); + btrfs_release_path(path); /* * this is where we are basically btrfs_lookup, without the * crossing root thing. we store the inode number in the * offset of the orphan item. */ + + if (found_key.offset == last_objectid) { + btrfs_err(root->fs_info, + "Error removing orphan entry, stopping orphan cleanup"); + ret = -EINVAL; + goto out; + } + + last_objectid = found_key.offset; + found_key.objectid = found_key.offset; found_key.type = BTRFS_INODE_ITEM_KEY; found_key.offset = 0; inode = btrfs_iget(root->fs_info->sb, &found_key, root, NULL); - BUG_ON(IS_ERR(inode)); + ret = PTR_ERR_OR_ZERO(inode); + if (ret && ret != -ESTALE) + goto out; - /* - * add this inode to the orphan list so btrfs_orphan_del does - * the proper thing when we hit it - */ - spin_lock(&root->orphan_lock); - list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list); - spin_unlock(&root->orphan_lock); + if (ret == -ESTALE && root == root->fs_info->tree_root) { + struct btrfs_root *dead_root; + struct btrfs_fs_info *fs_info = root->fs_info; + int is_dead_root = 0; + /* + * this is an orphan in the tree root. Currently these + * could come from 2 sources: + * a) a snapshot deletion in progress + * b) a free space cache inode + * We need to distinguish those two, as the snapshot + * orphan must not get deleted. + * find_dead_roots already ran before us, so if this + * is a snapshot deletion, we should find the root + * in the dead_roots list + */ + spin_lock(&fs_info->trans_lock); + list_for_each_entry(dead_root, &fs_info->dead_roots, + root_list) { + if (dead_root->root_key.objectid == + found_key.objectid) { + is_dead_root = 1; + break; + } + } + spin_unlock(&fs_info->trans_lock); + if (is_dead_root) { + /* prevent this orphan from being found again */ + key.offset = found_key.objectid - 1; + continue; + } + } /* - * if this is a bad inode, means we actually succeeded in - * removing the inode, but not the orphan record, which means - * we need to manually delete the orphan since iput will just - * do a destroy_inode + * Inode is already gone but the orphan item is still there, + * kill the orphan item. */ - if (is_bad_inode(inode)) { - trans = btrfs_start_transaction(root, 0); - btrfs_orphan_del(trans, inode); + if (ret == -ESTALE) { + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + goto out; + } + btrfs_debug(root->fs_info, "auto deleting %Lu", + found_key.objectid); + ret = btrfs_del_orphan_item(trans, root, + found_key.objectid); btrfs_end_transaction(trans, root); - iput(inode); + if (ret) + goto out; continue; } + /* + * add this inode to the orphan list so btrfs_orphan_del does + * the proper thing when we hit it + */ + set_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags); + atomic_inc(&root->orphan_inodes); + /* if we have links, this was a truncate, lets do that */ if (inode->i_nlink) { + if (WARN_ON(!S_ISREG(inode->i_mode))) { + iput(inode); + continue; + } nr_truncate++; - btrfs_truncate(inode); + + /* 1 for the orphan item deletion. */ + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + iput(inode); + ret = PTR_ERR(trans); + goto out; + } + ret = btrfs_orphan_add(trans, inode); + btrfs_end_transaction(trans, root); + if (ret) { + iput(inode); + goto out; + } + + ret = btrfs_truncate(inode); + if (ret) + btrfs_orphan_del(NULL, inode); } else { nr_unlink++; } /* this will do delete_inode and everything for us */ iput(inode); + if (ret) + goto out; } - btrfs_free_path(path); + /* release the path since we're done with it */ + btrfs_release_path(path); root->orphan_cleanup_state = ORPHAN_CLEANUP_DONE; @@ -2364,15 +3292,24 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) btrfs_block_rsv_release(root, root->orphan_block_rsv, (u64)-1); - if (root->orphan_block_rsv || root->orphan_item_inserted) { - trans = btrfs_join_transaction(root, 1); - btrfs_end_transaction(trans, root); + if (root->orphan_block_rsv || + test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state)) { + trans = btrfs_join_transaction(root); + if (!IS_ERR(trans)) + btrfs_end_transaction(trans, root); } if (nr_unlink) - printk(KERN_INFO "btrfs: unlinked %d orphans\n", nr_unlink); + btrfs_debug(root->fs_info, "unlinked %d orphans", nr_unlink); if (nr_truncate) - printk(KERN_INFO "btrfs: truncated %d orphans\n", nr_truncate); + btrfs_debug(root->fs_info, "truncated %d orphans", nr_truncate); + +out: + if (ret) + btrfs_crit(root->fs_info, + "could not do orphan cleanup %d", ret); + btrfs_free_path(path); + return ret; } /* @@ -2382,13 +3319,24 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) * slot is the slot the inode is in, objectid is the objectid of the inode */ static noinline int acls_after_inode_item(struct extent_buffer *leaf, - int slot, u64 objectid) + int slot, u64 objectid, + int *first_xattr_slot) { u32 nritems = btrfs_header_nritems(leaf); struct btrfs_key found_key; + static u64 xattr_access = 0; + static u64 xattr_default = 0; int scanned = 0; + if (!xattr_access) { + xattr_access = btrfs_name_hash(POSIX_ACL_XATTR_ACCESS, + strlen(POSIX_ACL_XATTR_ACCESS)); + xattr_default = btrfs_name_hash(POSIX_ACL_XATTR_DEFAULT, + strlen(POSIX_ACL_XATTR_DEFAULT)); + } + slot++; + *first_xattr_slot = -1; while (slot < nritems) { btrfs_item_key_to_cpu(leaf, &found_key, slot); @@ -2397,8 +3345,13 @@ static noinline int acls_after_inode_item(struct extent_buffer *leaf, return 0; /* we found an xattr, assume we've got an acl */ - if (found_key.type == BTRFS_XATTR_ITEM_KEY) - return 1; + if (found_key.type == BTRFS_XATTR_ITEM_KEY) { + if (*first_xattr_slot == -1) + *first_xattr_slot = slot; + if (found_key.offset == xattr_access || + found_key.offset == xattr_default) + return 1; + } /* * we found a key greater than an xattr key, there can't @@ -2423,6 +3376,8 @@ static noinline int acls_after_inode_item(struct extent_buffer *leaf, * something larger than an xattr. We have to assume the inode * has acls */ + if (*first_xattr_slot == -1) + *first_xattr_slot = slot; return 1; } @@ -2437,13 +3392,21 @@ static void btrfs_read_locked_inode(struct inode *inode) struct btrfs_timespec *tspec; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_key location; + unsigned long ptr; int maybe_acls; - u64 alloc_group_block; u32 rdev; int ret; + bool filled = false; + int first_xattr_slot; + + ret = btrfs_fill_inode(inode, &rdev); + if (!ret) + filled = true; path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) + goto make_bad; + memcpy(&location, &BTRFS_I(inode)->location, sizeof(location)); ret = btrfs_lookup_inode(NULL, root, path, &location, 0); @@ -2451,13 +3414,16 @@ static void btrfs_read_locked_inode(struct inode *inode) goto make_bad; leaf = path->nodes[0]; + + if (filled) + goto cache_index; + inode_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_inode_item); - inode->i_mode = btrfs_inode_mode(leaf, inode_item); - inode->i_nlink = btrfs_inode_nlink(leaf, inode_item); - inode->i_uid = btrfs_inode_uid(leaf, inode_item); - inode->i_gid = btrfs_inode_gid(leaf, inode_item); + set_nlink(inode, btrfs_inode_nlink(leaf, inode_item)); + i_uid_write(inode, btrfs_inode_uid(leaf, inode_item)); + i_gid_write(inode, btrfs_inode_gid(leaf, inode_item)); btrfs_i_size_write(inode, btrfs_inode_size(leaf, inode_item)); tspec = btrfs_inode_atime(inode_item); @@ -2474,7 +3440,19 @@ static void btrfs_read_locked_inode(struct inode *inode) inode_set_bytes(inode, btrfs_inode_nbytes(leaf, inode_item)); BTRFS_I(inode)->generation = btrfs_inode_generation(leaf, inode_item); - BTRFS_I(inode)->sequence = btrfs_inode_sequence(leaf, inode_item); + BTRFS_I(inode)->last_trans = btrfs_inode_transid(leaf, inode_item); + + /* + * If we were modified in the current generation and evicted from memory + * and then re-read we need to do a full sync since we don't have any + * idea about which extents were modified before we were evicted from + * cache. + */ + if (BTRFS_I(inode)->last_trans == root->fs_info->generation) + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + + inode->i_version = btrfs_inode_sequence(leaf, inode_item); inode->i_generation = BTRFS_I(inode)->generation; inode->i_rdev = 0; rdev = btrfs_inode_rdev(leaf, inode_item); @@ -2482,21 +3460,50 @@ static void btrfs_read_locked_inode(struct inode *inode) BTRFS_I(inode)->index_cnt = (u64)-1; BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item); - alloc_group_block = btrfs_inode_block_group(leaf, inode_item); +cache_index: + path->slots[0]++; + if (inode->i_nlink != 1 || + path->slots[0] >= btrfs_header_nritems(leaf)) + goto cache_acl; + btrfs_item_key_to_cpu(leaf, &location, path->slots[0]); + if (location.objectid != btrfs_ino(inode)) + goto cache_acl; + + ptr = btrfs_item_ptr_offset(leaf, path->slots[0]); + if (location.type == BTRFS_INODE_REF_KEY) { + struct btrfs_inode_ref *ref; + + ref = (struct btrfs_inode_ref *)ptr; + BTRFS_I(inode)->dir_index = btrfs_inode_ref_index(leaf, ref); + } else if (location.type == BTRFS_INODE_EXTREF_KEY) { + struct btrfs_inode_extref *extref; + + extref = (struct btrfs_inode_extref *)ptr; + BTRFS_I(inode)->dir_index = btrfs_inode_extref_index(leaf, + extref); + } +cache_acl: /* * try to precache a NULL acl entry for files that don't have * any xattrs or acls */ - maybe_acls = acls_after_inode_item(leaf, path->slots[0], inode->i_ino); + maybe_acls = acls_after_inode_item(leaf, path->slots[0], + btrfs_ino(inode), &first_xattr_slot); + if (first_xattr_slot != -1) { + path->slots[0] = first_xattr_slot; + ret = btrfs_load_inode_props(inode, path); + if (ret) + btrfs_err(root->fs_info, + "error loading props for ino %llu (root %llu): %d", + btrfs_ino(inode), + root->root_key.objectid, ret); + } + btrfs_free_path(path); + if (!maybe_acls) cache_no_acl(inode); - BTRFS_I(inode)->block_group = btrfs_find_block_group(root, 0, - alloc_group_block, 0); - btrfs_free_path(path); - inode_item = NULL; - switch (inode->i_mode & S_IFMT) { case S_IFREG: inode->i_mapping->a_ops = &btrfs_aops; @@ -2539,40 +3546,47 @@ static void fill_inode_item(struct btrfs_trans_handle *trans, struct btrfs_inode_item *item, struct inode *inode) { - btrfs_set_inode_uid(leaf, item, inode->i_uid); - btrfs_set_inode_gid(leaf, item, inode->i_gid); - btrfs_set_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size); - btrfs_set_inode_mode(leaf, item, inode->i_mode); - btrfs_set_inode_nlink(leaf, item, inode->i_nlink); - - btrfs_set_timespec_sec(leaf, btrfs_inode_atime(item), - inode->i_atime.tv_sec); - btrfs_set_timespec_nsec(leaf, btrfs_inode_atime(item), - inode->i_atime.tv_nsec); - - btrfs_set_timespec_sec(leaf, btrfs_inode_mtime(item), - inode->i_mtime.tv_sec); - btrfs_set_timespec_nsec(leaf, btrfs_inode_mtime(item), - inode->i_mtime.tv_nsec); - - btrfs_set_timespec_sec(leaf, btrfs_inode_ctime(item), - inode->i_ctime.tv_sec); - btrfs_set_timespec_nsec(leaf, btrfs_inode_ctime(item), - inode->i_ctime.tv_nsec); - - btrfs_set_inode_nbytes(leaf, item, inode_get_bytes(inode)); - btrfs_set_inode_generation(leaf, item, BTRFS_I(inode)->generation); - btrfs_set_inode_sequence(leaf, item, BTRFS_I(inode)->sequence); - btrfs_set_inode_transid(leaf, item, trans->transid); - btrfs_set_inode_rdev(leaf, item, inode->i_rdev); - btrfs_set_inode_flags(leaf, item, BTRFS_I(inode)->flags); - btrfs_set_inode_block_group(leaf, item, BTRFS_I(inode)->block_group); + struct btrfs_map_token token; + + btrfs_init_map_token(&token); + + btrfs_set_token_inode_uid(leaf, item, i_uid_read(inode), &token); + btrfs_set_token_inode_gid(leaf, item, i_gid_read(inode), &token); + btrfs_set_token_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size, + &token); + btrfs_set_token_inode_mode(leaf, item, inode->i_mode, &token); + btrfs_set_token_inode_nlink(leaf, item, inode->i_nlink, &token); + + btrfs_set_token_timespec_sec(leaf, btrfs_inode_atime(item), + inode->i_atime.tv_sec, &token); + btrfs_set_token_timespec_nsec(leaf, btrfs_inode_atime(item), + inode->i_atime.tv_nsec, &token); + + btrfs_set_token_timespec_sec(leaf, btrfs_inode_mtime(item), + inode->i_mtime.tv_sec, &token); + btrfs_set_token_timespec_nsec(leaf, btrfs_inode_mtime(item), + inode->i_mtime.tv_nsec, &token); + + btrfs_set_token_timespec_sec(leaf, btrfs_inode_ctime(item), + inode->i_ctime.tv_sec, &token); + btrfs_set_token_timespec_nsec(leaf, btrfs_inode_ctime(item), + inode->i_ctime.tv_nsec, &token); + + btrfs_set_token_inode_nbytes(leaf, item, inode_get_bytes(inode), + &token); + btrfs_set_token_inode_generation(leaf, item, BTRFS_I(inode)->generation, + &token); + btrfs_set_token_inode_sequence(leaf, item, inode->i_version, &token); + btrfs_set_token_inode_transid(leaf, item, trans->transid, &token); + btrfs_set_token_inode_rdev(leaf, item, inode->i_rdev, &token); + btrfs_set_token_inode_flags(leaf, item, BTRFS_I(inode)->flags, &token); + btrfs_set_token_inode_block_group(leaf, item, 0, &token); } /* * copy everything in the in-memory inode into the btree. */ -noinline int btrfs_update_inode(struct btrfs_trans_handle *trans, +static noinline int btrfs_update_inode_item(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *inode) { struct btrfs_inode_item *inode_item; @@ -2581,20 +3595,21 @@ noinline int btrfs_update_inode(struct btrfs_trans_handle *trans, int ret; path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) + return -ENOMEM; + path->leave_spinning = 1; - ret = btrfs_lookup_inode(trans, root, path, - &BTRFS_I(inode)->location, 1); + ret = btrfs_lookup_inode(trans, root, path, &BTRFS_I(inode)->location, + 1); if (ret) { if (ret > 0) ret = -ENOENT; goto failed; } - btrfs_unlock_up_safe(path, 1); leaf = path->nodes[0]; inode_item = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_inode_item); + struct btrfs_inode_item); fill_inode_item(trans, leaf, inode_item, inode); btrfs_mark_buffer_dirty(leaf); @@ -2605,16 +3620,55 @@ failed: return ret; } +/* + * copy everything in the in-memory inode into the btree. + */ +noinline int btrfs_update_inode(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct inode *inode) +{ + int ret; + + /* + * If the inode is a free space inode, we can deadlock during commit + * if we put it into the delayed code. + * + * The data relocation inode should also be directly updated + * without delay + */ + if (!btrfs_is_free_space_inode(inode) + && root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID) { + btrfs_update_root_times(trans, root); + + ret = btrfs_delayed_update_inode(trans, root, inode); + if (!ret) + btrfs_set_inode_last_trans(trans, inode); + return ret; + } + + return btrfs_update_inode_item(trans, root, inode); +} + +noinline int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode) +{ + int ret; + + ret = btrfs_update_inode(trans, root, inode); + if (ret == -ENOSPC) + return btrfs_update_inode_item(trans, root, inode); + return ret; +} /* * unlink helper that gets used here in inode.c and in the tree logging * recovery code. It remove a link in a directory with a given name, and * also drops the back refs in the inode to the directory */ -int btrfs_unlink_inode(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *dir, struct inode *inode, - const char *name, int name_len) +static int __btrfs_unlink_inode(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *dir, struct inode *inode, + const char *name, int name_len) { struct btrfs_path *path; int ret = 0; @@ -2622,15 +3676,17 @@ int btrfs_unlink_inode(struct btrfs_trans_handle *trans, struct btrfs_dir_item *di; struct btrfs_key key; u64 index; + u64 ino = btrfs_ino(inode); + u64 dir_ino = btrfs_ino(dir); path = btrfs_alloc_path(); if (!path) { ret = -ENOMEM; - goto err; + goto out; } path->leave_spinning = 1; - di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino, + di = btrfs_lookup_dir_item(trans, root, path, dir_ino, name, name_len, -1); if (IS_ERR(di)) { ret = PTR_ERR(di); @@ -2645,265 +3701,154 @@ int btrfs_unlink_inode(struct btrfs_trans_handle *trans, ret = btrfs_delete_one_dir_name(trans, root, path, di); if (ret) goto err; - btrfs_release_path(root, path); + btrfs_release_path(path); - ret = btrfs_del_inode_ref(trans, root, name, name_len, - inode->i_ino, - dir->i_ino, &index); - if (ret) { - printk(KERN_INFO "btrfs failed to delete reference to %.*s, " - "inode %lu parent %lu\n", name_len, name, - inode->i_ino, dir->i_ino); - goto err; + /* + * If we don't have dir index, we have to get it by looking up + * the inode ref, since we get the inode ref, remove it directly, + * it is unnecessary to do delayed deletion. + * + * But if we have dir index, needn't search inode ref to get it. + * Since the inode ref is close to the inode item, it is better + * that we delay to delete it, and just do this deletion when + * we update the inode item. + */ + if (BTRFS_I(inode)->dir_index) { + ret = btrfs_delayed_delete_inode_ref(inode); + if (!ret) { + index = BTRFS_I(inode)->dir_index; + goto skip_backref; + } } - di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino, - index, name, name_len, -1); - if (IS_ERR(di)) { - ret = PTR_ERR(di); + ret = btrfs_del_inode_ref(trans, root, name, name_len, ino, + dir_ino, &index); + if (ret) { + btrfs_info(root->fs_info, + "failed to delete reference to %.*s, inode %llu parent %llu", + name_len, name, ino, dir_ino); + btrfs_abort_transaction(trans, root, ret); goto err; } - if (!di) { - ret = -ENOENT; +skip_backref: + ret = btrfs_delete_delayed_dir_index(trans, root, dir, index); + if (ret) { + btrfs_abort_transaction(trans, root, ret); goto err; } - ret = btrfs_delete_one_dir_name(trans, root, path, di); - btrfs_release_path(root, path); ret = btrfs_del_inode_ref_in_log(trans, root, name, name_len, - inode, dir->i_ino); - BUG_ON(ret != 0 && ret != -ENOENT); + inode, dir_ino); + if (ret != 0 && ret != -ENOENT) { + btrfs_abort_transaction(trans, root, ret); + goto err; + } ret = btrfs_del_dir_entries_in_log(trans, root, name, name_len, dir, index); if (ret == -ENOENT) ret = 0; + else if (ret) + btrfs_abort_transaction(trans, root, ret); err: btrfs_free_path(path); if (ret) goto out; btrfs_i_size_write(dir, dir->i_size - name_len * 2); + inode_inc_iversion(inode); + inode_inc_iversion(dir); inode->i_ctime = dir->i_mtime = dir->i_ctime = CURRENT_TIME; - btrfs_update_inode(trans, root, dir); - btrfs_drop_nlink(inode); - ret = btrfs_update_inode(trans, root, inode); + ret = btrfs_update_inode(trans, root, dir); out: return ret; } -/* helper to check if there is any shared block in the path */ -static int check_path_shared(struct btrfs_root *root, - struct btrfs_path *path) +int btrfs_unlink_inode(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *dir, struct inode *inode, + const char *name, int name_len) { - struct extent_buffer *eb; - int level; - u64 refs = 1; - int uninitialized_var(ret); - - for (level = 0; level < BTRFS_MAX_LEVEL; level++) { - if (!path->nodes[level]) - break; - eb = path->nodes[level]; - if (!btrfs_block_can_be_shared(root, eb)) - continue; - ret = btrfs_lookup_extent_info(NULL, root, eb->start, eb->len, - &refs, NULL); - if (refs > 1) - return 1; + int ret; + ret = __btrfs_unlink_inode(trans, root, dir, inode, name, name_len); + if (!ret) { + drop_nlink(inode); + ret = btrfs_update_inode(trans, root, inode); } - return ret; /* XXX callers? */ + return ret; } /* * helper to start transaction for unlink and rmdir. * - * unlink and rmdir are special in btrfs, they do not always free space. - * so in enospc case, we should make sure they will free space before - * allowing them to use the global metadata reservation. + * unlink and rmdir are special in btrfs, they do not always free space, so + * if we cannot make our reservations the normal way try and see if there is + * plenty of slack room in the global reserve to migrate, otherwise we cannot + * allow the unlink to occur. */ -static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir, - struct dentry *dentry) +static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir) { struct btrfs_trans_handle *trans; struct btrfs_root *root = BTRFS_I(dir)->root; - struct btrfs_path *path; - struct btrfs_inode_ref *ref; - struct btrfs_dir_item *di; - struct inode *inode = dentry->d_inode; - u64 index; - int check_link = 1; - int err = -ENOSPC; int ret; - trans = btrfs_start_transaction(root, 10); + /* + * 1 for the possible orphan item + * 1 for the dir item + * 1 for the dir index + * 1 for the inode ref + * 1 for the inode + */ + trans = btrfs_start_transaction(root, 5); if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC) return trans; - if (inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID) - return ERR_PTR(-ENOSPC); - - /* check if there is someone else holds reference */ - if (S_ISDIR(inode->i_mode) && atomic_read(&inode->i_count) > 1) - return ERR_PTR(-ENOSPC); - - if (atomic_read(&inode->i_count) > 2) - return ERR_PTR(-ENOSPC); - - if (xchg(&root->fs_info->enospc_unlink, 1)) - return ERR_PTR(-ENOSPC); - - path = btrfs_alloc_path(); - if (!path) { - root->fs_info->enospc_unlink = 0; - return ERR_PTR(-ENOMEM); - } - - trans = btrfs_start_transaction(root, 0); - if (IS_ERR(trans)) { - btrfs_free_path(path); - root->fs_info->enospc_unlink = 0; - return trans; - } - - path->skip_locking = 1; - path->search_commit_root = 1; - - ret = btrfs_lookup_inode(trans, root, path, - &BTRFS_I(dir)->location, 0); - if (ret < 0) { - err = ret; - goto out; - } - if (ret == 0) { - if (check_path_shared(root, path)) - goto out; - } else { - check_link = 0; - } - btrfs_release_path(root, path); - - ret = btrfs_lookup_inode(trans, root, path, - &BTRFS_I(inode)->location, 0); - if (ret < 0) { - err = ret; - goto out; - } - if (ret == 0) { - if (check_path_shared(root, path)) - goto out; - } else { - check_link = 0; - } - btrfs_release_path(root, path); + if (PTR_ERR(trans) == -ENOSPC) { + u64 num_bytes = btrfs_calc_trans_metadata_size(root, 5); - if (ret == 0 && S_ISREG(inode->i_mode)) { - ret = btrfs_lookup_file_extent(trans, root, path, - inode->i_ino, (u64)-1, 0); - if (ret < 0) { - err = ret; - goto out; + trans = btrfs_start_transaction(root, 0); + if (IS_ERR(trans)) + return trans; + ret = btrfs_cond_migrate_bytes(root->fs_info, + &root->fs_info->trans_block_rsv, + num_bytes, 5); + if (ret) { + btrfs_end_transaction(trans, root); + return ERR_PTR(ret); } - BUG_ON(ret == 0); - if (check_path_shared(root, path)) - goto out; - btrfs_release_path(root, path); - } - - if (!check_link) { - err = 0; - goto out; - } - - di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino, - dentry->d_name.name, dentry->d_name.len, 0); - if (IS_ERR(di)) { - err = PTR_ERR(di); - goto out; - } - if (di) { - if (check_path_shared(root, path)) - goto out; - } else { - err = 0; - goto out; - } - btrfs_release_path(root, path); - - ref = btrfs_lookup_inode_ref(trans, root, path, - dentry->d_name.name, dentry->d_name.len, - inode->i_ino, dir->i_ino, 0); - if (IS_ERR(ref)) { - err = PTR_ERR(ref); - goto out; - } - BUG_ON(!ref); - if (check_path_shared(root, path)) - goto out; - index = btrfs_inode_ref_index(path->nodes[0], ref); - btrfs_release_path(root, path); - - di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino, index, - dentry->d_name.name, dentry->d_name.len, 0); - if (IS_ERR(di)) { - err = PTR_ERR(di); - goto out; - } - BUG_ON(ret == -ENOENT); - if (check_path_shared(root, path)) - goto out; - - err = 0; -out: - btrfs_free_path(path); - if (err) { - btrfs_end_transaction(trans, root); - root->fs_info->enospc_unlink = 0; - return ERR_PTR(err); + trans->block_rsv = &root->fs_info->trans_block_rsv; + trans->bytes_reserved = num_bytes; } - - trans->block_rsv = &root->fs_info->global_block_rsv; return trans; } -static void __unlink_end_trans(struct btrfs_trans_handle *trans, - struct btrfs_root *root) -{ - if (trans->block_rsv == &root->fs_info->global_block_rsv) { - BUG_ON(!root->fs_info->enospc_unlink); - root->fs_info->enospc_unlink = 0; - } - btrfs_end_transaction_throttle(trans, root); -} - static int btrfs_unlink(struct inode *dir, struct dentry *dentry) { struct btrfs_root *root = BTRFS_I(dir)->root; struct btrfs_trans_handle *trans; struct inode *inode = dentry->d_inode; int ret; - unsigned long nr = 0; - trans = __unlink_start_trans(dir, dentry); + trans = __unlink_start_trans(dir); if (IS_ERR(trans)) return PTR_ERR(trans); - btrfs_set_trans_block_group(trans, dir); - btrfs_record_unlink_dir(trans, dir, dentry->d_inode, 0); ret = btrfs_unlink_inode(trans, root, dir, dentry->d_inode, dentry->d_name.name, dentry->d_name.len); - BUG_ON(ret); + if (ret) + goto out; if (inode->i_nlink == 0) { ret = btrfs_orphan_add(trans, inode); - BUG_ON(ret); + if (ret) + goto out; } - nr = trans->blocks_used; - __unlink_end_trans(trans, root); - btrfs_btree_balance_dirty(root, nr); +out: + btrfs_end_transaction(trans, root); + btrfs_btree_balance_dirty(root); return ret; } @@ -2918,55 +3863,73 @@ int btrfs_unlink_subvol(struct btrfs_trans_handle *trans, struct btrfs_key key; u64 index; int ret; + u64 dir_ino = btrfs_ino(dir); path = btrfs_alloc_path(); if (!path) return -ENOMEM; - di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino, + di = btrfs_lookup_dir_item(trans, root, path, dir_ino, name, name_len, -1); - BUG_ON(!di || IS_ERR(di)); + if (IS_ERR_OR_NULL(di)) { + if (!di) + ret = -ENOENT; + else + ret = PTR_ERR(di); + goto out; + } leaf = path->nodes[0]; btrfs_dir_item_key_to_cpu(leaf, di, &key); WARN_ON(key.type != BTRFS_ROOT_ITEM_KEY || key.objectid != objectid); ret = btrfs_delete_one_dir_name(trans, root, path, di); - BUG_ON(ret); - btrfs_release_path(root, path); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto out; + } + btrfs_release_path(path); ret = btrfs_del_root_ref(trans, root->fs_info->tree_root, objectid, root->root_key.objectid, - dir->i_ino, &index, name, name_len); + dir_ino, &index, name, name_len); if (ret < 0) { - BUG_ON(ret != -ENOENT); - di = btrfs_search_dir_index_item(root, path, dir->i_ino, + if (ret != -ENOENT) { + btrfs_abort_transaction(trans, root, ret); + goto out; + } + di = btrfs_search_dir_index_item(root, path, dir_ino, name, name_len); - BUG_ON(!di || IS_ERR(di)); + if (IS_ERR_OR_NULL(di)) { + if (!di) + ret = -ENOENT; + else + ret = PTR_ERR(di); + btrfs_abort_transaction(trans, root, ret); + goto out; + } leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); - btrfs_release_path(root, path); + btrfs_release_path(path); index = key.offset; } + btrfs_release_path(path); - di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino, - index, name, name_len, -1); - BUG_ON(!di || IS_ERR(di)); - - leaf = path->nodes[0]; - btrfs_dir_item_key_to_cpu(leaf, di, &key); - WARN_ON(key.type != BTRFS_ROOT_ITEM_KEY || key.objectid != objectid); - ret = btrfs_delete_one_dir_name(trans, root, path, di); - BUG_ON(ret); - btrfs_release_path(root, path); + ret = btrfs_delete_delayed_dir_index(trans, root, dir, index); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto out; + } btrfs_i_size_write(dir, dir->i_size - name_len * 2); + inode_inc_iversion(dir); dir->i_mtime = dir->i_ctime = CURRENT_TIME; - ret = btrfs_update_inode(trans, root, dir); - BUG_ON(ret); - + ret = btrfs_update_inode_fallback(trans, root, dir); + if (ret) + btrfs_abort_transaction(trans, root, ret); +out: btrfs_free_path(path); - return 0; + return ret; } static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) @@ -2975,19 +3938,17 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) int err = 0; struct btrfs_root *root = BTRFS_I(dir)->root; struct btrfs_trans_handle *trans; - unsigned long nr = 0; - if (inode->i_size > BTRFS_EMPTY_DIR_SIZE || - inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) + if (inode->i_size > BTRFS_EMPTY_DIR_SIZE) return -ENOTEMPTY; + if (btrfs_ino(inode) == BTRFS_FIRST_FREE_OBJECTID) + return -EPERM; - trans = __unlink_start_trans(dir, dentry); + trans = __unlink_start_trans(dir); if (IS_ERR(trans)) return PTR_ERR(trans); - btrfs_set_trans_block_group(trans, dir); - - if (unlikely(inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) { + if (unlikely(btrfs_ino(inode) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) { err = btrfs_unlink_subvol(trans, root, dir, BTRFS_I(inode)->location.objectid, dentry->d_name.name, @@ -3005,185 +3966,12 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) if (!err) btrfs_i_size_write(inode, 0); out: - nr = trans->blocks_used; - __unlink_end_trans(trans, root); - btrfs_btree_balance_dirty(root, nr); + btrfs_end_transaction(trans, root); + btrfs_btree_balance_dirty(root); return err; } -#if 0 -/* - * when truncating bytes in a file, it is possible to avoid reading - * the leaves that contain only checksum items. This can be the - * majority of the IO required to delete a large file, but it must - * be done carefully. - * - * The keys in the level just above the leaves are checked to make sure - * the lowest key in a given leaf is a csum key, and starts at an offset - * after the new size. - * - * Then the key for the next leaf is checked to make sure it also has - * a checksum item for the same file. If it does, we know our target leaf - * contains only checksum items, and it can be safely freed without reading - * it. - * - * This is just an optimization targeted at large files. It may do - * nothing. It will return 0 unless things went badly. - */ -static noinline int drop_csum_leaves(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_path *path, - struct inode *inode, u64 new_size) -{ - struct btrfs_key key; - int ret; - int nritems; - struct btrfs_key found_key; - struct btrfs_key other_key; - struct btrfs_leaf_ref *ref; - u64 leaf_gen; - u64 leaf_start; - - path->lowest_level = 1; - key.objectid = inode->i_ino; - key.type = BTRFS_CSUM_ITEM_KEY; - key.offset = new_size; -again: - ret = btrfs_search_slot(trans, root, &key, path, -1, 1); - if (ret < 0) - goto out; - - if (path->nodes[1] == NULL) { - ret = 0; - goto out; - } - ret = 0; - btrfs_node_key_to_cpu(path->nodes[1], &found_key, path->slots[1]); - nritems = btrfs_header_nritems(path->nodes[1]); - - if (!nritems) - goto out; - - if (path->slots[1] >= nritems) - goto next_node; - - /* did we find a key greater than anything we want to delete? */ - if (found_key.objectid > inode->i_ino || - (found_key.objectid == inode->i_ino && found_key.type > key.type)) - goto out; - - /* we check the next key in the node to make sure the leave contains - * only checksum items. This comparison doesn't work if our - * leaf is the last one in the node - */ - if (path->slots[1] + 1 >= nritems) { -next_node: - /* search forward from the last key in the node, this - * will bring us into the next node in the tree - */ - btrfs_node_key_to_cpu(path->nodes[1], &found_key, nritems - 1); - - /* unlikely, but we inc below, so check to be safe */ - if (found_key.offset == (u64)-1) - goto out; - - /* search_forward needs a path with locks held, do the - * search again for the original key. It is possible - * this will race with a balance and return a path that - * we could modify, but this drop is just an optimization - * and is allowed to miss some leaves. - */ - btrfs_release_path(root, path); - found_key.offset++; - - /* setup a max key for search_forward */ - other_key.offset = (u64)-1; - other_key.type = key.type; - other_key.objectid = key.objectid; - - path->keep_locks = 1; - ret = btrfs_search_forward(root, &found_key, &other_key, - path, 0, 0); - path->keep_locks = 0; - if (ret || found_key.objectid != key.objectid || - found_key.type != key.type) { - ret = 0; - goto out; - } - - key.offset = found_key.offset; - btrfs_release_path(root, path); - cond_resched(); - goto again; - } - - /* we know there's one more slot after us in the tree, - * read that key so we can verify it is also a checksum item - */ - btrfs_node_key_to_cpu(path->nodes[1], &other_key, path->slots[1] + 1); - - if (found_key.objectid < inode->i_ino) - goto next_key; - - if (found_key.type != key.type || found_key.offset < new_size) - goto next_key; - - /* - * if the key for the next leaf isn't a csum key from this objectid, - * we can't be sure there aren't good items inside this leaf. - * Bail out - */ - if (other_key.objectid != inode->i_ino || other_key.type != key.type) - goto out; - - leaf_start = btrfs_node_blockptr(path->nodes[1], path->slots[1]); - leaf_gen = btrfs_node_ptr_generation(path->nodes[1], path->slots[1]); - /* - * it is safe to delete this leaf, it contains only - * csum items from this inode at an offset >= new_size - */ - ret = btrfs_del_leaf(trans, root, path, leaf_start); - BUG_ON(ret); - - if (root->ref_cows && leaf_gen < trans->transid) { - ref = btrfs_alloc_leaf_ref(root, 0); - if (ref) { - ref->root_gen = root->root_key.offset; - ref->bytenr = leaf_start; - ref->owner = 0; - ref->generation = leaf_gen; - ref->nritems = 0; - - btrfs_sort_leaf_ref(ref); - - ret = btrfs_add_leaf_ref(root, ref, 0); - WARN_ON(ret); - btrfs_free_leaf_ref(root, ref); - } else { - WARN_ON(1); - } - } -next_key: - btrfs_release_path(root, path); - - if (other_key.objectid == inode->i_ino && - other_key.type == key.type && other_key.offset > key.offset) { - key.offset = other_key.offset; - cond_resched(); - goto again; - } - ret = 0; -out: - /* fixup any changes we've made to the path */ - path->lowest_level = 0; - path->keep_locks = 0; - btrfs_release_path(root, path); - return ret; -} - -#endif - /* * 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 @@ -3209,27 +3997,44 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, u64 extent_num_bytes = 0; u64 extent_offset = 0; u64 item_end = 0; - u64 mask = root->sectorsize - 1; + u64 last_size = (u64)-1; u32 found_type = (u8)-1; int found_extent; int del_item; int pending_del_nr = 0; int pending_del_slot = 0; int extent_type = -1; - int encoding; int ret; int err = 0; + u64 ino = btrfs_ino(inode); BUG_ON(new_size > 0 && min_type != BTRFS_EXTENT_DATA_KEY); - if (root->ref_cows || root == root->fs_info->tree_root) - btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0); - path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) + return -ENOMEM; path->reada = -1; - key.objectid = inode->i_ino; + /* + * We want to drop from the next block forward in case this new size is + * not block aligned since we will be keeping the last block of the + * extent just the way it is. + */ + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) || + root == root->fs_info->tree_root) + btrfs_drop_extent_cache(inode, ALIGN(new_size, + root->sectorsize), (u64)-1, 0); + + /* + * This function is also used to drop the items in the log tree before + * we relog the inode, so if root != BTRFS_I(inode)->root, it means + * it is used to drop the loged items. So we shouldn't kill the delayed + * items. + */ + if (min_type == 0 && root == BTRFS_I(inode)->root) + btrfs_kill_delayed_inode_items(inode); + + key.objectid = ino; key.offset = (u64)-1; key.type = (u8)-1; @@ -3255,9 +4060,8 @@ search_again: leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); found_type = btrfs_key_type(&found_key); - encoding = 0; - if (found_key.objectid != inode->i_ino) + if (found_key.objectid != ino) break; if (found_type < min_type) @@ -3268,16 +4072,12 @@ search_again: fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); extent_type = btrfs_file_extent_type(leaf, fi); - encoding = btrfs_file_extent_compression(leaf, fi); - encoding |= btrfs_file_extent_encryption(leaf, fi); - encoding |= btrfs_file_extent_other_encoding(leaf, fi); - if (extent_type != BTRFS_FILE_EXTENT_INLINE) { item_end += btrfs_file_extent_num_bytes(leaf, fi); } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { item_end += btrfs_file_extent_inline_len(leaf, - fi); + path->slots[0], fi); } item_end--; } @@ -3296,21 +4096,27 @@ search_again: if (found_type != BTRFS_EXTENT_DATA_KEY) goto delete; + if (del_item) + last_size = found_key.offset; + else + last_size = new_size; + if (extent_type != BTRFS_FILE_EXTENT_INLINE) { u64 num_dec; extent_start = btrfs_file_extent_disk_bytenr(leaf, fi); - if (!del_item && !encoding) { + if (!del_item) { u64 orig_num_bytes = btrfs_file_extent_num_bytes(leaf, fi); - extent_num_bytes = new_size - - found_key.offset + root->sectorsize - 1; - extent_num_bytes = extent_num_bytes & - ~((u64)root->sectorsize - 1); + extent_num_bytes = ALIGN(new_size - + found_key.offset, + root->sectorsize); btrfs_set_file_extent_num_bytes(leaf, fi, extent_num_bytes); num_dec = (orig_num_bytes - extent_num_bytes); - if (root->ref_cows && extent_start != 0) + if (test_bit(BTRFS_ROOT_REF_COWS, + &root->state) && + extent_start != 0) inode_sub_bytes(inode, num_dec); btrfs_mark_buffer_dirty(leaf); } else { @@ -3324,7 +4130,8 @@ search_again: num_dec = btrfs_file_extent_num_bytes(leaf, fi); if (extent_start != 0) { found_extent = 1; - if (root->ref_cows) + if (test_bit(BTRFS_ROOT_REF_COWS, + &root->state)) inode_sub_bytes(inode, num_dec); } } @@ -3339,16 +4146,20 @@ search_again: btrfs_file_extent_other_encoding(leaf, fi) == 0) { u32 size = new_size - found_key.offset; - if (root->ref_cows) { + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) inode_sub_bytes(inode, item_end + 1 - new_size); - } + + /* + * update the ram bytes to properly reflect + * the new size of our item + */ + btrfs_set_file_extent_ram_bytes(leaf, fi, size); size = btrfs_file_extent_calc_inline_size(size); - ret = btrfs_truncate_item(trans, root, path, - size, 1); - BUG_ON(ret); - } else if (root->ref_cows) { + btrfs_truncate_item(root, path, size, 1); + } else if (test_bit(BTRFS_ROOT_REF_COWS, + &root->state)) { inode_sub_bytes(inode, item_end + 1 - found_key.offset); } @@ -3370,13 +4181,14 @@ delete: } else { break; } - if (found_extent && (root->ref_cows || - root == root->fs_info->tree_root)) { + if (found_extent && + (test_bit(BTRFS_ROOT_REF_COWS, &root->state) || + root == root->fs_info->tree_root)) { btrfs_set_path_blocking(path); ret = btrfs_free_extent(trans, root, extent_start, extent_num_bytes, 0, btrfs_header_owner(leaf), - inode->i_ino, extent_offset); + ino, extent_offset, 0); BUG_ON(ret); } @@ -3385,18 +4197,18 @@ delete: if (path->slots[0] == 0 || path->slots[0] != pending_del_slot) { - if (root->ref_cows) { - err = -EAGAIN; - goto out; - } if (pending_del_nr) { ret = btrfs_del_items(trans, root, path, pending_del_slot, pending_del_nr); - BUG_ON(ret); + if (ret) { + btrfs_abort_transaction(trans, + root, ret); + goto error; + } pending_del_nr = 0; } - btrfs_release_path(root, path); + btrfs_release_path(path); goto search_again; } else { path->slots[0]--; @@ -3406,19 +4218,31 @@ out: if (pending_del_nr) { ret = btrfs_del_items(trans, root, path, pending_del_slot, pending_del_nr); - BUG_ON(ret); + if (ret) + btrfs_abort_transaction(trans, root, ret); } +error: + if (last_size != (u64)-1) + btrfs_ordered_update_i_size(inode, last_size, NULL); btrfs_free_path(path); return err; } /* - * taken from block_truncate_page, but does cow as it zeros out - * any bytes left in the last page in the file. + * btrfs_truncate_page - read, zero a chunk and write a page + * @inode - inode that we're zeroing + * @from - the offset to start zeroing + * @len - the length to zero, 0 to zero the entire range respective to the + * offset + * @front - zero up to the offset instead of from the offset on + * + * This will find the page for the "from" offset and cow the page and zero the + * part we want to zero. This is used with truncate and hole punching. */ -static int btrfs_truncate_page(struct address_space *mapping, loff_t from) +int btrfs_truncate_page(struct inode *inode, loff_t from, loff_t len, + int front) { - struct inode *inode = mapping->host; + struct address_space *mapping = inode->i_mapping; struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_ordered_extent *ordered; @@ -3428,21 +4252,23 @@ static int btrfs_truncate_page(struct address_space *mapping, loff_t from) pgoff_t index = from >> PAGE_CACHE_SHIFT; unsigned offset = from & (PAGE_CACHE_SIZE-1); struct page *page; + gfp_t mask = btrfs_alloc_write_mask(mapping); int ret = 0; u64 page_start; u64 page_end; - if ((offset & (blocksize - 1)) == 0) + if ((offset & (blocksize - 1)) == 0 && + (!len || ((len & (blocksize - 1)) == 0))) goto out; ret = btrfs_delalloc_reserve_space(inode, PAGE_CACHE_SIZE); if (ret) goto out; - ret = -ENOMEM; again: - page = grab_cache_page(mapping, index); + page = find_or_create_page(mapping, index, mask); if (!page) { btrfs_delalloc_release_space(inode, PAGE_CACHE_SIZE); + ret = -ENOMEM; goto out; } @@ -3464,8 +4290,7 @@ again: } wait_on_page_writeback(page); - lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state, - GFP_NOFS); + lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state); set_page_extent_mapped(page); ordered = btrfs_lookup_ordered_extent(inode, page_start); @@ -3480,7 +4305,8 @@ again: } clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end, - EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, + EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0, &cached_state, GFP_NOFS); ret = btrfs_set_extent_delalloc(inode, page_start, page_end, @@ -3491,10 +4317,14 @@ again: goto out_unlock; } - ret = 0; if (offset != PAGE_CACHE_SIZE) { + if (!len) + len = PAGE_CACHE_SIZE - offset; kaddr = kmap(page); - memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset); + if (front) + memset(kaddr, 0, offset); + else + memset(kaddr + offset, 0, len); flush_dcache_page(page); kunmap(page); } @@ -3512,35 +4342,93 @@ out: return ret; } -int btrfs_cont_expand(struct inode *inode, loff_t size) +static int maybe_insert_hole(struct btrfs_root *root, struct inode *inode, + u64 offset, u64 len) { struct btrfs_trans_handle *trans; + int ret; + + /* + * Still need to make sure the inode looks like it's been updated so + * that any holes get logged if we fsync. + */ + if (btrfs_fs_incompat(root->fs_info, NO_HOLES)) { + BTRFS_I(inode)->last_trans = root->fs_info->generation; + BTRFS_I(inode)->last_sub_trans = root->log_transid; + BTRFS_I(inode)->last_log_commit = root->last_log_commit; + return 0; + } + + /* + * 1 - for the one we're dropping + * 1 - for the one we're adding + * 1 - for updating the inode. + */ + trans = btrfs_start_transaction(root, 3); + if (IS_ERR(trans)) + return PTR_ERR(trans); + + ret = btrfs_drop_extents(trans, root, inode, offset, offset + len, 1); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + btrfs_end_transaction(trans, root); + return ret; + } + + ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode), offset, + 0, 0, len, 0, len, 0, 0, 0); + if (ret) + btrfs_abort_transaction(trans, root, ret); + else + btrfs_update_inode(trans, root, inode); + btrfs_end_transaction(trans, root); + return ret; +} + +/* + * This function puts in dummy file extents for the area we're creating a hole + * for. So if we are truncating this file to a larger size we need to insert + * these file extents so that btrfs_get_extent will return a EXTENT_MAP_HOLE for + * the range between oldsize and size + */ +int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size) +{ struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_map *em = NULL; struct extent_state *cached_state = NULL; - u64 mask = root->sectorsize - 1; - u64 hole_start = (inode->i_size + mask) & ~mask; - u64 block_end = (size + mask) & ~mask; + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + u64 hole_start = ALIGN(oldsize, root->sectorsize); + u64 block_end = ALIGN(size, root->sectorsize); u64 last_byte; u64 cur_offset; u64 hole_size; int err = 0; + /* + * If our size started in the middle of a page we need to zero out the + * rest of the page before we expand the i_size, otherwise we could + * expose stale data. + */ + err = btrfs_truncate_page(inode, oldsize, 0, 0); + if (err) + return err; + if (size <= hole_start) return 0; while (1) { struct btrfs_ordered_extent *ordered; - btrfs_wait_ordered_range(inode, hole_start, - block_end - hole_start); + lock_extent_bits(io_tree, hole_start, block_end - 1, 0, - &cached_state, GFP_NOFS); - ordered = btrfs_lookup_ordered_extent(inode, hole_start); + &cached_state); + ordered = btrfs_lookup_ordered_range(inode, hole_start, + block_end - hole_start); if (!ordered) break; unlock_extent_cached(io_tree, hole_start, block_end - 1, &cached_state, GFP_NOFS); + btrfs_start_ordered_extent(inode, ordered, 1); btrfs_put_ordered_extent(ordered); } @@ -3548,162 +4436,275 @@ int btrfs_cont_expand(struct inode *inode, loff_t size) while (1) { em = btrfs_get_extent(inode, NULL, 0, cur_offset, block_end - cur_offset, 0); - BUG_ON(IS_ERR(em) || !em); + if (IS_ERR(em)) { + err = PTR_ERR(em); + em = NULL; + break; + } last_byte = min(extent_map_end(em), block_end); - last_byte = (last_byte + mask) & ~mask; + last_byte = ALIGN(last_byte , root->sectorsize); if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) { - u64 hint_byte = 0; + struct extent_map *hole_em; hole_size = last_byte - cur_offset; - trans = btrfs_start_transaction(root, 2); - if (IS_ERR(trans)) { - err = PTR_ERR(trans); + err = maybe_insert_hole(root, inode, cur_offset, + hole_size); + if (err) break; + btrfs_drop_extent_cache(inode, cur_offset, + cur_offset + hole_size - 1, 0); + hole_em = alloc_extent_map(); + if (!hole_em) { + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + goto next; } - btrfs_set_trans_block_group(trans, inode); - - err = btrfs_drop_extents(trans, inode, cur_offset, - cur_offset + hole_size, - &hint_byte, 1); - BUG_ON(err); - - err = btrfs_insert_file_extent(trans, root, - inode->i_ino, cur_offset, 0, - 0, hole_size, 0, hole_size, - 0, 0, 0); - BUG_ON(err); + hole_em->start = cur_offset; + hole_em->len = hole_size; + hole_em->orig_start = cur_offset; + + hole_em->block_start = EXTENT_MAP_HOLE; + hole_em->block_len = 0; + hole_em->orig_block_len = 0; + hole_em->ram_bytes = hole_size; + hole_em->bdev = root->fs_info->fs_devices->latest_bdev; + hole_em->compress_type = BTRFS_COMPRESS_NONE; + hole_em->generation = root->fs_info->generation; - btrfs_drop_extent_cache(inode, hole_start, - last_byte - 1, 0); - - btrfs_end_transaction(trans, root); + while (1) { + write_lock(&em_tree->lock); + err = add_extent_mapping(em_tree, hole_em, 1); + write_unlock(&em_tree->lock); + if (err != -EEXIST) + break; + btrfs_drop_extent_cache(inode, cur_offset, + cur_offset + + hole_size - 1, 0); + } + free_extent_map(hole_em); } +next: free_extent_map(em); em = NULL; cur_offset = last_byte; if (cur_offset >= block_end) break; } - free_extent_map(em); unlock_extent_cached(io_tree, hole_start, block_end - 1, &cached_state, GFP_NOFS); return err; } -static int btrfs_setattr_size(struct inode *inode, struct iattr *attr) +static int btrfs_setsize(struct inode *inode, struct iattr *attr) { struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans; - unsigned long nr; + loff_t oldsize = i_size_read(inode); + loff_t newsize = attr->ia_size; + int mask = attr->ia_valid; int ret; - if (attr->ia_size == inode->i_size) - return 0; - - if (attr->ia_size > inode->i_size) { - unsigned long limit; - limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur; - if (attr->ia_size > inode->i_sb->s_maxbytes) - return -EFBIG; - if (limit != RLIM_INFINITY && attr->ia_size > limit) { - send_sig(SIGXFSZ, current, 0); - return -EFBIG; - } + /* + * The regular truncate() case without ATTR_CTIME and ATTR_MTIME is a + * special case where we need to update the times despite not having + * these flags set. For all other operations the VFS set these flags + * explicitly if it wants a timestamp update. + */ + if (newsize != oldsize) { + inode_inc_iversion(inode); + if (!(mask & (ATTR_CTIME | ATTR_MTIME))) + inode->i_ctime = inode->i_mtime = + current_fs_time(inode->i_sb); } - trans = btrfs_start_transaction(root, 5); - if (IS_ERR(trans)) - return PTR_ERR(trans); + if (newsize > oldsize) { + truncate_pagecache(inode, newsize); + ret = btrfs_cont_expand(inode, oldsize, newsize); + if (ret) + return ret; - btrfs_set_trans_block_group(trans, inode); + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) + return PTR_ERR(trans); - ret = btrfs_orphan_add(trans, inode); - BUG_ON(ret); + i_size_write(inode, newsize); + btrfs_ordered_update_i_size(inode, i_size_read(inode), NULL); + ret = btrfs_update_inode(trans, root, inode); + btrfs_end_transaction(trans, root); + } else { - nr = trans->blocks_used; - btrfs_end_transaction(trans, root); - btrfs_btree_balance_dirty(root, nr); + /* + * We're truncating a file that used to have good data down to + * zero. Make sure it gets into the ordered flush list so that + * any new writes get down to disk quickly. + */ + if (newsize == 0) + set_bit(BTRFS_INODE_ORDERED_DATA_CLOSE, + &BTRFS_I(inode)->runtime_flags); - if (attr->ia_size > inode->i_size) { - ret = btrfs_cont_expand(inode, attr->ia_size); - if (ret) { - btrfs_truncate(inode); + /* + * 1 for the orphan item we're going to add + * 1 for the orphan item deletion. + */ + trans = btrfs_start_transaction(root, 2); + if (IS_ERR(trans)) + return PTR_ERR(trans); + + /* + * We need to do this in case we fail at _any_ point during the + * actual truncate. Once we do the truncate_setsize we could + * invalidate pages which forces any outstanding ordered io to + * be instantly completed which will give us extents that need + * to be truncated. If we fail to get an orphan inode down we + * could have left over extents that were never meant to live, + * so we need to garuntee from this point on that everything + * will be consistent. + */ + ret = btrfs_orphan_add(trans, inode); + btrfs_end_transaction(trans, root); + if (ret) return ret; - } - i_size_write(inode, attr->ia_size); - btrfs_ordered_update_i_size(inode, inode->i_size, NULL); + /* we don't support swapfiles, so vmtruncate shouldn't fail */ + truncate_setsize(inode, newsize); - trans = btrfs_start_transaction(root, 0); - BUG_ON(IS_ERR(trans)); - btrfs_set_trans_block_group(trans, inode); - trans->block_rsv = root->orphan_block_rsv; - BUG_ON(!trans->block_rsv); + /* Disable nonlocked read DIO to avoid the end less truncate */ + btrfs_inode_block_unlocked_dio(inode); + inode_dio_wait(inode); + btrfs_inode_resume_unlocked_dio(inode); - ret = btrfs_update_inode(trans, root, inode); - BUG_ON(ret); - if (inode->i_nlink > 0) { - ret = btrfs_orphan_del(trans, inode); - BUG_ON(ret); + ret = btrfs_truncate(inode); + if (ret && inode->i_nlink) { + int err; + + /* + * failed to truncate, disk_i_size is only adjusted down + * as we remove extents, so it should represent the true + * size of the inode, so reset the in memory size and + * delete our orphan entry. + */ + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) { + btrfs_orphan_del(NULL, inode); + return ret; + } + i_size_write(inode, BTRFS_I(inode)->disk_i_size); + err = btrfs_orphan_del(trans, inode); + if (err) + btrfs_abort_transaction(trans, root, err); + btrfs_end_transaction(trans, root); } - nr = trans->blocks_used; - btrfs_end_transaction(trans, root); - btrfs_btree_balance_dirty(root, nr); - return 0; } - /* - * We're truncating a file that used to have good data down to - * zero. Make sure it gets into the ordered flush list so that - * any new writes get down to disk quickly. - */ - if (attr->ia_size == 0) - BTRFS_I(inode)->ordered_data_close = 1; - - /* we don't support swapfiles, so vmtruncate shouldn't fail */ - ret = vmtruncate(inode, attr->ia_size); - BUG_ON(ret); - - return 0; + return ret; } static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) { struct inode *inode = dentry->d_inode; + struct btrfs_root *root = BTRFS_I(inode)->root; int err; + if (btrfs_root_readonly(root)) + return -EROFS; + err = inode_change_ok(inode, attr); if (err) return err; if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) { - err = btrfs_setattr_size(inode, attr); + err = btrfs_setsize(inode, attr); if (err) return err; } if (attr->ia_valid) { setattr_copy(inode, attr); - mark_inode_dirty(inode); + inode_inc_iversion(inode); + err = btrfs_dirty_inode(inode); - if (attr->ia_valid & ATTR_MODE) - err = btrfs_acl_chmod(inode); + if (!err && attr->ia_valid & ATTR_MODE) + err = posix_acl_chmod(inode, inode->i_mode); } return err; } +/* + * While truncating the inode pages during eviction, we get the VFS calling + * btrfs_invalidatepage() against each page of the inode. This is slow because + * the calls to btrfs_invalidatepage() result in a huge amount of calls to + * lock_extent_bits() and clear_extent_bit(), which keep merging and splitting + * extent_state structures over and over, wasting lots of time. + * + * Therefore if the inode is being evicted, let btrfs_invalidatepage() skip all + * those expensive operations on a per page basis and do only the ordered io + * finishing, while we release here the extent_map and extent_state structures, + * without the excessive merging and splitting. + */ +static void evict_inode_truncate_pages(struct inode *inode) +{ + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; + struct extent_map_tree *map_tree = &BTRFS_I(inode)->extent_tree; + struct rb_node *node; + + ASSERT(inode->i_state & I_FREEING); + truncate_inode_pages_final(&inode->i_data); + + write_lock(&map_tree->lock); + while (!RB_EMPTY_ROOT(&map_tree->map)) { + struct extent_map *em; + + node = rb_first(&map_tree->map); + em = rb_entry(node, struct extent_map, rb_node); + clear_bit(EXTENT_FLAG_PINNED, &em->flags); + clear_bit(EXTENT_FLAG_LOGGING, &em->flags); + remove_extent_mapping(map_tree, em); + free_extent_map(em); + } + write_unlock(&map_tree->lock); + + spin_lock(&io_tree->lock); + while (!RB_EMPTY_ROOT(&io_tree->state)) { + struct extent_state *state; + struct extent_state *cached_state = NULL; + + node = rb_first(&io_tree->state); + state = rb_entry(node, struct extent_state, rb_node); + atomic_inc(&state->refs); + spin_unlock(&io_tree->lock); + + lock_extent_bits(io_tree, state->start, state->end, + 0, &cached_state); + clear_extent_bit(io_tree, state->start, state->end, + EXTENT_LOCKED | EXTENT_DIRTY | + EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | + EXTENT_DEFRAG, 1, 1, + &cached_state, GFP_NOFS); + free_extent_state(state); + + spin_lock(&io_tree->lock); + } + spin_unlock(&io_tree->lock); +} + void btrfs_evict_inode(struct inode *inode) { struct btrfs_trans_handle *trans; struct btrfs_root *root = BTRFS_I(inode)->root; - unsigned long nr; + struct btrfs_block_rsv *rsv, *global_rsv; + u64 min_size = btrfs_calc_trunc_metadata_size(root, 1); int ret; - truncate_inode_pages(&inode->i_data, 0); - if (inode->i_nlink && (btrfs_root_refs(&root->root_item) != 0 || - root == root->fs_info->tree_root)) + trace_btrfs_inode_evict(inode); + + evict_inode_truncate_pages(inode); + + if (inode->i_nlink && + ((btrfs_root_refs(&root->root_item) != 0 && + root->root_key.objectid != BTRFS_ROOT_TREE_OBJECTID) || + btrfs_is_free_space_inode(inode))) goto no_delete; if (is_bad_inode(inode)) { @@ -3714,53 +4715,103 @@ void btrfs_evict_inode(struct inode *inode) btrfs_wait_ordered_range(inode, 0, (u64)-1); if (root->fs_info->log_root_recovering) { - BUG_ON(!list_empty(&BTRFS_I(inode)->i_orphan)); + BUG_ON(test_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags)); goto no_delete; } if (inode->i_nlink > 0) { - BUG_ON(btrfs_root_refs(&root->root_item) != 0); + BUG_ON(btrfs_root_refs(&root->root_item) != 0 && + root->root_key.objectid != BTRFS_ROOT_TREE_OBJECTID); goto no_delete; } + ret = btrfs_commit_inode_delayed_inode(inode); + if (ret) { + btrfs_orphan_del(NULL, inode); + goto no_delete; + } + + rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP); + if (!rsv) { + btrfs_orphan_del(NULL, inode); + goto no_delete; + } + rsv->size = min_size; + rsv->failfast = 1; + global_rsv = &root->fs_info->global_block_rsv; + btrfs_i_size_write(inode, 0); + /* + * This is a bit simpler than btrfs_truncate since we've already + * reserved our space for our orphan item in the unlink, so we just + * need to reserve some slack space in case we add bytes and update + * inode item when doing the truncate. + */ while (1) { - trans = btrfs_start_transaction(root, 0); - BUG_ON(IS_ERR(trans)); - btrfs_set_trans_block_group(trans, inode); - trans->block_rsv = root->orphan_block_rsv; + ret = btrfs_block_rsv_refill(root, rsv, min_size, + BTRFS_RESERVE_FLUSH_LIMIT); + + /* + * Try and steal from the global reserve since we will + * likely not use this space anyway, we want to try as + * hard as possible to get this to work. + */ + if (ret) + ret = btrfs_block_rsv_migrate(global_rsv, rsv, min_size); - ret = btrfs_block_rsv_check(trans, root, - root->orphan_block_rsv, 0, 5); if (ret) { - BUG_ON(ret != -EAGAIN); - ret = btrfs_commit_transaction(trans, root); - BUG_ON(ret); - continue; + btrfs_warn(root->fs_info, + "Could not get space for a delete, will truncate on mount %d", + ret); + btrfs_orphan_del(NULL, inode); + btrfs_free_block_rsv(root, rsv); + goto no_delete; + } + + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) { + btrfs_orphan_del(NULL, inode); + btrfs_free_block_rsv(root, rsv); + goto no_delete; } + trans->block_rsv = rsv; + ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0); - if (ret != -EAGAIN) + if (ret != -ENOSPC) break; - nr = trans->blocks_used; + trans->block_rsv = &root->fs_info->trans_block_rsv; btrfs_end_transaction(trans, root); trans = NULL; - btrfs_btree_balance_dirty(root, nr); - + btrfs_btree_balance_dirty(root); } + btrfs_free_block_rsv(root, rsv); + + /* + * Errors here aren't a big deal, it just means we leave orphan items + * in the tree. They will be cleaned up on the next mount. + */ if (ret == 0) { - ret = btrfs_orphan_del(trans, inode); - BUG_ON(ret); + trans->block_rsv = root->orphan_block_rsv; + btrfs_orphan_del(trans, inode); + } else { + btrfs_orphan_del(NULL, inode); } - nr = trans->blocks_used; + trans->block_rsv = &root->fs_info->trans_block_rsv; + if (!(root == root->fs_info->tree_root || + root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)) + btrfs_return_ino(root, btrfs_ino(inode)); + btrfs_end_transaction(trans, root); - btrfs_btree_balance_dirty(root, nr); + btrfs_btree_balance_dirty(root); no_delete: - end_writeback(inode); + btrfs_remove_delayed_node(inode); + clear_inode(inode); return; } @@ -3779,14 +4830,15 @@ static int btrfs_inode_by_name(struct inode *dir, struct dentry *dentry, int ret = 0; path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) + return -ENOMEM; - di = btrfs_lookup_dir_item(NULL, root, path, dir->i_ino, name, + di = btrfs_lookup_dir_item(NULL, root, path, btrfs_ino(dir), name, namelen, 0); if (IS_ERR(di)) ret = PTR_ERR(di); - if (!di || IS_ERR(di)) + if (IS_ERR_OR_NULL(di)) goto out_err; btrfs_dir_item_key_to_cpu(path->nodes[0], di, location); @@ -3823,9 +4875,9 @@ static int fixup_tree_root_location(struct btrfs_root *root, } err = -ENOENT; - ret = btrfs_find_root_ref(root->fs_info->tree_root, path, - BTRFS_I(dir)->root->root_key.objectid, - location->objectid); + ret = btrfs_find_item(root->fs_info->tree_root, path, + BTRFS_I(dir)->root->root_key.objectid, + location->objectid, BTRFS_ROOT_REF_KEY, NULL); if (ret) { if (ret < 0) err = ret; @@ -3834,7 +4886,7 @@ static int fixup_tree_root_location(struct btrfs_root *root, leaf = path->nodes[0]; ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_root_ref); - if (btrfs_root_ref_dirid(leaf, ref) != dir->i_ino || + if (btrfs_root_ref_dirid(leaf, ref) != btrfs_ino(dir) || btrfs_root_ref_name_len(leaf, ref) != dentry->d_name.len) goto out; @@ -3844,7 +4896,7 @@ static int fixup_tree_root_location(struct btrfs_root *root, if (ret) goto out; - btrfs_release_path(root->fs_info->tree_root, path); + btrfs_release_path(path); new_root = btrfs_read_fs_root_no_name(root->fs_info, location); if (IS_ERR(new_root)) { @@ -3852,11 +4904,6 @@ static int fixup_tree_root_location(struct btrfs_root *root, goto out; } - if (btrfs_root_refs(&new_root->root_item) == 0) { - err = -ENOENT; - goto out; - } - *sub_root = new_root; location->objectid = btrfs_root_dirid(&new_root->root_item); location->type = BTRFS_INODE_ITEM_KEY; @@ -3873,33 +4920,33 @@ static void inode_tree_add(struct inode *inode) struct btrfs_inode *entry; struct rb_node **p; struct rb_node *parent; -again: - p = &root->inode_tree.rb_node; - parent = NULL; + struct rb_node *new = &BTRFS_I(inode)->rb_node; + u64 ino = btrfs_ino(inode); if (inode_unhashed(inode)) return; - + parent = NULL; spin_lock(&root->inode_lock); + p = &root->inode_tree.rb_node; while (*p) { parent = *p; entry = rb_entry(parent, struct btrfs_inode, rb_node); - if (inode->i_ino < entry->vfs_inode.i_ino) + if (ino < btrfs_ino(&entry->vfs_inode)) p = &parent->rb_left; - else if (inode->i_ino > entry->vfs_inode.i_ino) + else if (ino > btrfs_ino(&entry->vfs_inode)) p = &parent->rb_right; else { WARN_ON(!(entry->vfs_inode.i_state & (I_WILL_FREE | I_FREEING))); - rb_erase(parent, &root->inode_tree); + rb_replace_node(parent, new, &root->inode_tree); RB_CLEAR_NODE(parent); spin_unlock(&root->inode_lock); - goto again; + return; } } - rb_link_node(&BTRFS_I(inode)->rb_node, parent, p); - rb_insert_color(&BTRFS_I(inode)->rb_node, &root->inode_tree); + rb_link_node(new, parent, p); + rb_insert_color(new, &root->inode_tree); spin_unlock(&root->inode_lock); } @@ -3916,14 +4963,7 @@ static void inode_tree_del(struct inode *inode) } spin_unlock(&root->inode_lock); - /* - * Free space cache has inodes in the tree root, but the tree root has a - * root_refs of 0, so this could end up dropping the tree root as a - * snapshot, so we need the extra !root->fs_info->tree_root check to - * make sure we don't drop it. - */ - if (empty && btrfs_root_refs(&root->root_item) == 0 && - root != root->fs_info->tree_root) { + if (empty && btrfs_root_refs(&root->root_item) == 0) { synchronize_srcu(&root->fs_info->subvol_srcu); spin_lock(&root->inode_lock); empty = RB_EMPTY_ROOT(&root->inode_tree); @@ -3933,7 +4973,7 @@ static void inode_tree_del(struct inode *inode) } } -int btrfs_invalidate_inodes(struct btrfs_root *root) +void btrfs_invalidate_inodes(struct btrfs_root *root) { struct rb_node *node; struct rb_node *prev; @@ -3941,7 +4981,8 @@ int btrfs_invalidate_inodes(struct btrfs_root *root) struct inode *inode; u64 objectid = 0; - WARN_ON(btrfs_root_refs(&root->root_item) != 0); + if (!test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) + WARN_ON(btrfs_root_refs(&root->root_item) != 0); spin_lock(&root->inode_lock); again: @@ -3951,9 +4992,9 @@ again: prev = node; entry = rb_entry(node, struct btrfs_inode, rb_node); - if (objectid < entry->vfs_inode.i_ino) + if (objectid < btrfs_ino(&entry->vfs_inode)) node = node->rb_left; - else if (objectid > entry->vfs_inode.i_ino) + else if (objectid > btrfs_ino(&entry->vfs_inode)) node = node->rb_right; else break; @@ -3961,7 +5002,7 @@ again: if (!node) { while (prev) { entry = rb_entry(prev, struct btrfs_inode, rb_node); - if (objectid <= entry->vfs_inode.i_ino) { + if (objectid <= btrfs_ino(&entry->vfs_inode)) { node = prev; break; } @@ -3970,7 +5011,7 @@ again: } while (node) { entry = rb_entry(node, struct btrfs_inode, rb_node); - objectid = entry->vfs_inode.i_ino + 1; + objectid = btrfs_ino(&entry->vfs_inode) + 1; inode = igrab(&entry->vfs_inode); if (inode) { spin_unlock(&root->inode_lock); @@ -3993,35 +5034,37 @@ again: node = rb_next(node); } spin_unlock(&root->inode_lock); - return 0; } static int btrfs_init_locked_inode(struct inode *inode, void *p) { struct btrfs_iget_args *args = p; - inode->i_ino = args->ino; + inode->i_ino = args->location->objectid; + memcpy(&BTRFS_I(inode)->location, args->location, + sizeof(*args->location)); BTRFS_I(inode)->root = args->root; - btrfs_set_inode_space_info(args->root, inode); return 0; } static int btrfs_find_actor(struct inode *inode, void *opaque) { struct btrfs_iget_args *args = opaque; - return args->ino == inode->i_ino && + return args->location->objectid == BTRFS_I(inode)->location.objectid && args->root == BTRFS_I(inode)->root; } static struct inode *btrfs_iget_locked(struct super_block *s, - u64 objectid, + struct btrfs_key *location, struct btrfs_root *root) { struct inode *inode; struct btrfs_iget_args args; - args.ino = objectid; + unsigned long hashval = btrfs_inode_hash(location->objectid, root); + + args.location = location; args.root = root; - inode = iget5_locked(s, objectid, btrfs_find_actor, + inode = iget5_locked(s, hashval, btrfs_find_actor, btrfs_init_locked_inode, (void *)&args); return inode; @@ -4035,19 +5078,22 @@ struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location, { struct inode *inode; - inode = btrfs_iget_locked(s, location->objectid, root); + inode = btrfs_iget_locked(s, location, root); if (!inode) return ERR_PTR(-ENOMEM); if (inode->i_state & I_NEW) { - BTRFS_I(inode)->root = root; - memcpy(&BTRFS_I(inode)->location, location, sizeof(*location)); btrfs_read_locked_inode(inode); - - inode_tree_add(inode); - unlock_new_inode(inode); - if (new) - *new = 1; + if (!is_bad_inode(inode)) { + inode_tree_add(inode); + unlock_new_inode(inode); + if (new) + *new = 1; + } else { + unlock_new_inode(inode); + iput(inode); + inode = ERR_PTR(-ESTALE); + } } return inode; @@ -4064,10 +5110,10 @@ static struct inode *new_simple_dir(struct super_block *s, BTRFS_I(inode)->root = root; memcpy(&BTRFS_I(inode)->location, key, sizeof(*key)); - BTRFS_I(inode)->dummy_inode = 1; + set_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags); inode->i_ino = BTRFS_EMPTY_SUBVOL_DIR_OBJECTID; - inode->i_op = &simple_dir_inode_operations; + inode->i_op = &btrfs_dir_ro_inode_operations; inode->i_fop = &simple_dir_operations; inode->i_mode = S_IFDIR | S_IRUGO | S_IWUSR | S_IXUGO; inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME; @@ -4082,20 +5128,17 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry) struct btrfs_root *sub_root = root; struct btrfs_key location; int index; - int ret; - - dentry->d_op = &btrfs_dentry_operations; + int ret = 0; if (dentry->d_name.len > BTRFS_NAME_LEN) return ERR_PTR(-ENAMETOOLONG); ret = btrfs_inode_by_name(dir, dentry, &location); - if (ret < 0) return ERR_PTR(ret); if (location.objectid == 0) - return NULL; + return ERR_PTR(-ENOENT); if (location.type == BTRFS_INODE_ITEM_KEY) { inode = btrfs_iget(dir->i_sb, &location, root, NULL); @@ -4117,62 +5160,78 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry) } srcu_read_unlock(&root->fs_info->subvol_srcu, index); - if (root != sub_root) { + if (!IS_ERR(inode) && root != sub_root) { down_read(&root->fs_info->cleanup_work_sem); if (!(inode->i_sb->s_flags & MS_RDONLY)) - btrfs_orphan_cleanup(sub_root); + ret = btrfs_orphan_cleanup(sub_root); up_read(&root->fs_info->cleanup_work_sem); + if (ret) { + iput(inode); + inode = ERR_PTR(ret); + } } return inode; } -static int btrfs_dentry_delete(struct dentry *dentry) +static int btrfs_dentry_delete(const struct dentry *dentry) { struct btrfs_root *root; + struct inode *inode = dentry->d_inode; - if (!dentry->d_inode && !IS_ROOT(dentry)) - dentry = dentry->d_parent; + if (!inode && !IS_ROOT(dentry)) + inode = dentry->d_parent->d_inode; - if (dentry->d_inode) { - root = BTRFS_I(dentry->d_inode)->root; + if (inode) { + root = BTRFS_I(inode)->root; if (btrfs_root_refs(&root->root_item) == 0) return 1; + + if (btrfs_ino(inode) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID) + return 1; } return 0; } +static void btrfs_dentry_release(struct dentry *dentry) +{ + kfree(dentry->d_fsdata); +} + static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry, - struct nameidata *nd) + unsigned int flags) { struct inode *inode; inode = btrfs_lookup_dentry(dir, dentry); - if (IS_ERR(inode)) - return ERR_CAST(inode); + if (IS_ERR(inode)) { + if (PTR_ERR(inode) == -ENOENT) + inode = NULL; + else + return ERR_CAST(inode); + } - return d_splice_alias(inode, dentry); + return d_materialise_unique(dentry, inode); } -static unsigned char btrfs_filetype_table[] = { +unsigned char btrfs_filetype_table[] = { DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK }; -static int btrfs_real_readdir(struct file *filp, void *dirent, - filldir_t filldir) +static int btrfs_real_readdir(struct file *file, struct dir_context *ctx) { - struct inode *inode = filp->f_dentry->d_inode; + struct inode *inode = file_inode(file); struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_item *item; struct btrfs_dir_item *di; struct btrfs_key key; struct btrfs_key found_key; struct btrfs_path *path; + struct list_head ins_list; + struct list_head del_list; int ret; - u32 nritems; struct extent_buffer *leaf; int slot; - int advance; unsigned char d_type; int over = 0; u32 di_cur; @@ -4182,71 +5241,63 @@ static int btrfs_real_readdir(struct file *filp, void *dirent, char tmp_name[32]; char *name_ptr; int name_len; + int is_curr = 0; /* ctx->pos points to the current index? */ /* FIXME, use a real flag for deciding about the key type */ if (root->fs_info->tree_root == root) key_type = BTRFS_DIR_ITEM_KEY; - /* special case for "." */ - if (filp->f_pos == 0) { - over = filldir(dirent, ".", 1, - 1, inode->i_ino, - DT_DIR); - if (over) - return 0; - filp->f_pos = 1; - } - /* special case for .., just use the back ref */ - if (filp->f_pos == 1) { - u64 pino = parent_ino(filp->f_path.dentry); - over = filldir(dirent, "..", 2, - 2, pino, DT_DIR); - if (over) - return 0; - filp->f_pos = 2; - } + if (!dir_emit_dots(file, ctx)) + return 0; + path = btrfs_alloc_path(); - path->reada = 2; + if (!path) + return -ENOMEM; + + path->reada = 1; + + if (key_type == BTRFS_DIR_INDEX_KEY) { + INIT_LIST_HEAD(&ins_list); + INIT_LIST_HEAD(&del_list); + btrfs_get_delayed_items(inode, &ins_list, &del_list); + } btrfs_set_key_type(&key, key_type); - key.offset = filp->f_pos; - key.objectid = inode->i_ino; + key.offset = ctx->pos; + key.objectid = btrfs_ino(inode); ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); if (ret < 0) goto err; - advance = 0; while (1) { leaf = path->nodes[0]; - nritems = btrfs_header_nritems(leaf); slot = path->slots[0]; - if (advance || slot >= nritems) { - if (slot >= nritems - 1) { - ret = btrfs_next_leaf(root, path); - if (ret) - break; - leaf = path->nodes[0]; - nritems = btrfs_header_nritems(leaf); - slot = path->slots[0]; - } else { - slot++; - path->slots[0]++; - } + if (slot >= btrfs_header_nritems(leaf)) { + ret = btrfs_next_leaf(root, path); + if (ret < 0) + goto err; + else if (ret > 0) + break; + continue; } - advance = 1; - item = btrfs_item_nr(leaf, slot); + item = btrfs_item_nr(slot); btrfs_item_key_to_cpu(leaf, &found_key, slot); if (found_key.objectid != key.objectid) break; if (btrfs_key_type(&found_key) != key_type) break; - if (found_key.offset < filp->f_pos) - continue; + if (found_key.offset < ctx->pos) + goto next; + if (key_type == BTRFS_DIR_INDEX_KEY && + btrfs_should_delete_dir_index(&del_list, + found_key.offset)) + goto next; - filp->f_pos = found_key.offset; + ctx->pos = found_key.offset; + is_curr = 1; di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item); di_cur = 0; @@ -4255,6 +5306,9 @@ static int btrfs_real_readdir(struct file *filp, void *dirent, while (di_cur < di_total) { struct btrfs_key location; + if (verify_dir_item(root, leaf, di)) + break; + name_len = btrfs_dir_name_len(leaf, di); if (name_len <= sizeof(tmp_name)) { name_ptr = tmp_name; @@ -4271,17 +5325,23 @@ static int btrfs_real_readdir(struct file *filp, void *dirent, d_type = btrfs_filetype_table[btrfs_dir_type(leaf, di)]; btrfs_dir_item_key_to_cpu(leaf, di, &location); + /* is this a reference to our own snapshot? If so - * skip it + * skip it. + * + * In contrast to old kernels, we insert the snapshot's + * dir item and dir index after it has been created, so + * we won't find a reference to our own snapshot. We + * still keep the following code for backward + * compatibility. */ if (location.type == BTRFS_ROOT_ITEM_KEY && location.objectid == root->root_key.objectid) { over = 0; goto skip; } - over = filldir(dirent, name_ptr, name_len, - found_key.offset, location.objectid, - d_type); + over = !dir_emit(ctx, name_ptr, name_len, + location.objectid, d_type); skip: if (name_ptr != tmp_name) @@ -4294,20 +5354,49 @@ skip: di_cur += di_len; di = (struct btrfs_dir_item *)((char *)di + di_len); } +next: + path->slots[0]++; + } + + if (key_type == BTRFS_DIR_INDEX_KEY) { + if (is_curr) + ctx->pos++; + ret = btrfs_readdir_delayed_dir_index(ctx, &ins_list); + if (ret) + goto nopos; } /* Reached end of directory/root. Bump pos past the last item. */ - if (key_type == BTRFS_DIR_INDEX_KEY) - /* - * 32-bit glibc will use getdents64, but then strtol - - * so the last number we can serve is this. - */ - filp->f_pos = 0x7fffffff; - else - filp->f_pos++; + ctx->pos++; + + /* + * Stop new entries from being returned after we return the last + * entry. + * + * New directory entries are assigned a strictly increasing + * offset. This means that new entries created during readdir + * are *guaranteed* to be seen in the future by that readdir. + * This has broken buggy programs which operate on names as + * they're returned by readdir. Until we re-use freed offsets + * we have this hack to stop new entries from being returned + * under the assumption that they'll never reach this huge + * offset. + * + * This is being careful not to overflow 32bit loff_t unless the + * last entry requires it because doing so has broken 32bit apps + * in the past. + */ + if (key_type == BTRFS_DIR_INDEX_KEY) { + if (ctx->pos >= INT_MAX) + ctx->pos = LLONG_MAX; + else + ctx->pos = INT_MAX; + } nopos: ret = 0; err: + if (key_type == BTRFS_DIR_INDEX_KEY) + btrfs_put_delayed_items(&ins_list, &del_list); btrfs_free_path(path); return ret; } @@ -4319,22 +5408,20 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc) int ret = 0; bool nolock = false; - if (BTRFS_I(inode)->dummy_inode) + if (test_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags)) return 0; - smp_mb(); - nolock = (root->fs_info->closing && root == root->fs_info->tree_root); + if (btrfs_fs_closing(root->fs_info) && btrfs_is_free_space_inode(inode)) + nolock = true; if (wbc->sync_mode == WB_SYNC_ALL) { if (nolock) - trans = btrfs_join_transaction_nolock(root, 1); + trans = btrfs_join_transaction_nolock(root); else - trans = btrfs_join_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - if (nolock) - ret = btrfs_end_transaction_nolock(trans, root); - else - ret = btrfs_commit_transaction(trans, root); + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) + return PTR_ERR(trans); + ret = btrfs_commit_transaction(trans, root); } return ret; } @@ -4345,43 +5432,57 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc) * FIXME, needs more benchmarking...there are no reasons other than performance * to keep or drop this code. */ -void btrfs_dirty_inode(struct inode *inode) +static int btrfs_dirty_inode(struct inode *inode) { struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans; int ret; - if (BTRFS_I(inode)->dummy_inode) - return; + if (test_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags)) + return 0; - trans = btrfs_join_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) + return PTR_ERR(trans); ret = btrfs_update_inode(trans, root, inode); if (ret && ret == -ENOSPC) { /* whoops, lets try again with the full transaction */ btrfs_end_transaction(trans, root); trans = btrfs_start_transaction(root, 1); - if (IS_ERR(trans)) { - if (printk_ratelimit()) { - printk(KERN_ERR "btrfs: fail to " - "dirty inode %lu error %ld\n", - inode->i_ino, PTR_ERR(trans)); - } - return; - } - btrfs_set_trans_block_group(trans, inode); + if (IS_ERR(trans)) + return PTR_ERR(trans); ret = btrfs_update_inode(trans, root, inode); - if (ret) { - if (printk_ratelimit()) { - printk(KERN_ERR "btrfs: fail to " - "dirty inode %lu error %d\n", - inode->i_ino, ret); - } - } } btrfs_end_transaction(trans, root); + if (BTRFS_I(inode)->delayed_node) + btrfs_balance_delayed_items(root); + + return ret; +} + +/* + * This is a copy of file_update_time. We need this so we can return error on + * ENOSPC for updating the inode in the case of file write and mmap writes. + */ +static int btrfs_update_time(struct inode *inode, struct timespec *now, + int flags) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + + if (btrfs_root_readonly(root)) + return -EROFS; + + if (flags & S_VERSION) + inode_inc_iversion(inode); + if (flags & S_CTIME) + inode->i_ctime = *now; + if (flags & S_MTIME) + inode->i_mtime = *now; + if (flags & S_ATIME) + inode->i_atime = *now; + return btrfs_dirty_inode(inode); } /* @@ -4397,7 +5498,7 @@ static int btrfs_set_inode_index_count(struct inode *inode) struct extent_buffer *leaf; int ret; - key.objectid = inode->i_ino; + key.objectid = btrfs_ino(inode); btrfs_set_key_type(&key, BTRFS_DIR_INDEX_KEY); key.offset = (u64)-1; @@ -4429,7 +5530,7 @@ static int btrfs_set_inode_index_count(struct inode *inode) leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - if (found_key.objectid != inode->i_ino || + if (found_key.objectid != btrfs_ino(inode) || btrfs_key_type(&found_key) != BTRFS_DIR_INDEX_KEY) { BTRFS_I(inode)->index_cnt = 2; goto out; @@ -4450,9 +5551,12 @@ int btrfs_set_inode_index(struct inode *dir, u64 *index) int ret = 0; if (BTRFS_I(dir)->index_cnt == (u64)-1) { - ret = btrfs_set_inode_index_count(dir); - if (ret) - return ret; + ret = btrfs_inode_delayed_dir_index_count(dir); + if (ret) { + ret = btrfs_set_inode_index_count(dir); + if (ret) + return ret; + } } *index = BTRFS_I(dir)->index_cnt; @@ -4466,7 +5570,7 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, struct inode *dir, const char *name, int name_len, u64 ref_objectid, u64 objectid, - u64 alloc_hint, int mode, u64 *index) + umode_t mode, u64 *index) { struct inode *inode; struct btrfs_inode_item *inode_item; @@ -4475,23 +5579,37 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, struct btrfs_inode_ref *ref; struct btrfs_key key[2]; u32 sizes[2]; + int nitems = name ? 2 : 1; unsigned long ptr; int ret; - int owner; path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) + return ERR_PTR(-ENOMEM); inode = new_inode(root->fs_info->sb); - if (!inode) + if (!inode) { + btrfs_free_path(path); return ERR_PTR(-ENOMEM); + } + + /* + * we have to initialize this early, so we can reclaim the inode + * number if we fail afterwards in this function. + */ + inode->i_ino = objectid; + + if (dir && name) { + trace_btrfs_inode_request(dir); - if (dir) { ret = btrfs_set_inode_index(dir, index); if (ret) { + btrfs_free_path(path); iput(inode); return ERR_PTR(ret); } + } else if (dir) { + *index = 0; } /* * index_cnt is ignored for everything but a dir, @@ -4499,47 +5617,61 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, * number */ BTRFS_I(inode)->index_cnt = 2; + BTRFS_I(inode)->dir_index = *index; BTRFS_I(inode)->root = root; BTRFS_I(inode)->generation = trans->transid; - btrfs_set_inode_space_info(root, inode); + inode->i_generation = BTRFS_I(inode)->generation; - if (mode & S_IFDIR) - owner = 0; - else - owner = 1; - BTRFS_I(inode)->block_group = - btrfs_find_block_group(root, 0, alloc_hint, owner); + /* + * We could have gotten an inode number from somebody who was fsynced + * and then removed in this same transaction, so let's just set full + * sync since it will be a full sync anyway and this will blow away the + * old info in the log. + */ + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags); key[0].objectid = objectid; btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY); key[0].offset = 0; - key[1].objectid = objectid; - btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY); - key[1].offset = ref_objectid; - sizes[0] = sizeof(struct btrfs_inode_item); - sizes[1] = name_len + sizeof(*ref); + + if (name) { + /* + * Start new inodes with an inode_ref. This is slightly more + * efficient for small numbers of hard links since they will + * be packed into one item. Extended refs will kick in if we + * add more hard links than can fit in the ref item. + */ + key[1].objectid = objectid; + btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY); + key[1].offset = ref_objectid; + + sizes[1] = name_len + sizeof(*ref); + } path->leave_spinning = 1; - ret = btrfs_insert_empty_items(trans, root, path, key, sizes, 2); + ret = btrfs_insert_empty_items(trans, root, path, key, sizes, nitems); if (ret != 0) goto fail; inode_init_owner(inode, dir, mode); - inode->i_ino = objectid; inode_set_bytes(inode, 0); inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME; inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0], struct btrfs_inode_item); + memset_extent_buffer(path->nodes[0], 0, (unsigned long)inode_item, + sizeof(*inode_item)); fill_inode_item(trans, path->nodes[0], inode_item, inode); - ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1, - struct btrfs_inode_ref); - btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len); - btrfs_set_inode_ref_index(path->nodes[0], ref, *index); - ptr = (unsigned long)(ref + 1); - write_extent_buffer(path->nodes[0], name, ptr, name_len); + if (name) { + ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1, + struct btrfs_inode_ref); + btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len); + btrfs_set_inode_ref_index(path->nodes[0], ref, *index); + ptr = (unsigned long)(ref + 1); + write_extent_buffer(path->nodes[0], name, ptr, name_len); + } btrfs_mark_buffer_dirty(path->nodes[0]); btrfs_free_path(path); @@ -4551,18 +5683,31 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, btrfs_inherit_iflags(inode, dir); - if ((mode & S_IFREG)) { + if (S_ISREG(mode)) { if (btrfs_test_opt(root, NODATASUM)) BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM; if (btrfs_test_opt(root, NODATACOW)) - BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW; + BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW | + BTRFS_INODE_NODATASUM; } - insert_inode_hash(inode); + btrfs_insert_inode_hash(inode); inode_tree_add(inode); + + trace_btrfs_inode_new(inode); + btrfs_set_inode_last_trans(trans, inode); + + btrfs_update_root_times(trans, root); + + ret = btrfs_inode_inherit_props(trans, inode, dir); + if (ret) + btrfs_err(root->fs_info, + "error inheriting props for ino %llu (root %llu): %d", + btrfs_ino(inode), root->root_key.objectid, ret); + return inode; fail: - if (dir) + if (dir && name) BTRFS_I(dir)->index_cnt--; btrfs_free_path(path); iput(inode); @@ -4587,58 +5732,81 @@ int btrfs_add_link(struct btrfs_trans_handle *trans, int ret = 0; struct btrfs_key key; struct btrfs_root *root = BTRFS_I(parent_inode)->root; + u64 ino = btrfs_ino(inode); + u64 parent_ino = btrfs_ino(parent_inode); - if (unlikely(inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) { + if (unlikely(ino == BTRFS_FIRST_FREE_OBJECTID)) { memcpy(&key, &BTRFS_I(inode)->root->root_key, sizeof(key)); } else { - key.objectid = inode->i_ino; + key.objectid = ino; btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); key.offset = 0; } - if (unlikely(inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) { + if (unlikely(ino == BTRFS_FIRST_FREE_OBJECTID)) { ret = btrfs_add_root_ref(trans, root->fs_info->tree_root, key.objectid, root->root_key.objectid, - parent_inode->i_ino, - index, name, name_len); + parent_ino, index, name, name_len); } else if (add_backref) { - ret = btrfs_insert_inode_ref(trans, root, - name, name_len, inode->i_ino, - parent_inode->i_ino, index); + ret = btrfs_insert_inode_ref(trans, root, name, name_len, ino, + parent_ino, index); } - if (ret == 0) { - ret = btrfs_insert_dir_item(trans, root, name, name_len, - parent_inode->i_ino, &key, - btrfs_inode_type(inode), index); - BUG_ON(ret); + /* Nothing to clean up yet */ + if (ret) + return ret; + + ret = btrfs_insert_dir_item(trans, root, name, name_len, + parent_inode, &key, + btrfs_inode_type(inode), index); + if (ret == -EEXIST || ret == -EOVERFLOW) + goto fail_dir_item; + else if (ret) { + btrfs_abort_transaction(trans, root, ret); + return ret; + } + + btrfs_i_size_write(parent_inode, parent_inode->i_size + + name_len * 2); + inode_inc_iversion(parent_inode); + parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME; + ret = btrfs_update_inode(trans, root, parent_inode); + if (ret) + btrfs_abort_transaction(trans, root, ret); + return ret; + +fail_dir_item: + if (unlikely(ino == BTRFS_FIRST_FREE_OBJECTID)) { + u64 local_index; + int err; + err = btrfs_del_root_ref(trans, root->fs_info->tree_root, + key.objectid, root->root_key.objectid, + parent_ino, &local_index, name, name_len); + + } else if (add_backref) { + u64 local_index; + int err; - btrfs_i_size_write(parent_inode, parent_inode->i_size + - name_len * 2); - parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME; - ret = btrfs_update_inode(trans, root, parent_inode); + err = btrfs_del_inode_ref(trans, root, name, name_len, + ino, parent_ino, &local_index); } return ret; } static int btrfs_add_nondir(struct btrfs_trans_handle *trans, - struct dentry *dentry, struct inode *inode, - int backref, u64 index) + struct inode *dir, struct dentry *dentry, + struct inode *inode, int backref, u64 index) { - int err = btrfs_add_link(trans, dentry->d_parent->d_inode, - inode, dentry->d_name.name, - dentry->d_name.len, backref, index); - if (!err) { - d_instantiate(dentry, inode); - return 0; - } + int err = btrfs_add_link(trans, dir, inode, + dentry->d_name.name, dentry->d_name.len, + backref, index); if (err > 0) err = -EEXIST; return err; } static int btrfs_mknod(struct inode *dir, struct dentry *dentry, - int mode, dev_t rdev) + umode_t mode, dev_t rdev) { struct btrfs_trans_handle *trans; struct btrfs_root *root = BTRFS_I(dir)->root; @@ -4646,16 +5814,11 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry, int err; int drop_inode = 0; u64 objectid; - unsigned long nr = 0; u64 index = 0; if (!new_valid_dev(rdev)) return -EINVAL; - err = btrfs_find_free_objectid(NULL, root, dir->i_ino, &objectid); - if (err) - return err; - /* * 2 for inode item and ref * 2 for dir items @@ -4665,37 +5828,44 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry, if (IS_ERR(trans)) return PTR_ERR(trans); - btrfs_set_trans_block_group(trans, dir); + err = btrfs_find_free_ino(root, &objectid); + if (err) + goto out_unlock; inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name, - dentry->d_name.len, - dentry->d_parent->d_inode->i_ino, objectid, - BTRFS_I(dir)->block_group, mode, &index); - err = PTR_ERR(inode); - if (IS_ERR(inode)) + dentry->d_name.len, btrfs_ino(dir), objectid, + mode, &index); + if (IS_ERR(inode)) { + err = PTR_ERR(inode); goto out_unlock; + } - err = btrfs_init_inode_security(trans, inode, dir); + err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name); if (err) { drop_inode = 1; goto out_unlock; } - btrfs_set_trans_block_group(trans, inode); - err = btrfs_add_nondir(trans, dentry, inode, 0, index); + /* + * If the active LSM wants to access the inode during + * d_instantiate it needs these. Smack checks to see + * if the filesystem supports xattrs by looking at the + * ops vector. + */ + + inode->i_op = &btrfs_special_inode_operations; + err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index); if (err) drop_inode = 1; else { - inode->i_op = &btrfs_special_inode_operations; init_special_inode(inode, inode->i_mode, rdev); btrfs_update_inode(trans, root, inode); + d_instantiate(dentry, inode); } - btrfs_update_inode_block_group(trans, inode); - btrfs_update_inode_block_group(trans, dir); out_unlock: - nr = trans->blocks_used; - btrfs_end_transaction_throttle(trans, root); - btrfs_btree_balance_dirty(root, nr); + btrfs_end_transaction(trans, root); + btrfs_balance_delayed_items(root); + btrfs_btree_balance_dirty(root); if (drop_inode) { inode_dec_link_count(inode); iput(inode); @@ -4704,20 +5874,16 @@ out_unlock: } static int btrfs_create(struct inode *dir, struct dentry *dentry, - int mode, struct nameidata *nd) + umode_t mode, bool excl) { struct btrfs_trans_handle *trans; struct btrfs_root *root = BTRFS_I(dir)->root; struct inode *inode = NULL; - int drop_inode = 0; + int drop_inode_on_err = 0; int err; - unsigned long nr = 0; u64 objectid; u64 index = 0; - err = btrfs_find_free_objectid(NULL, root, dir->i_ino, &objectid); - if (err) - return err; /* * 2 for inode item and ref * 2 for dir items @@ -4727,44 +5893,53 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry, if (IS_ERR(trans)) return PTR_ERR(trans); - btrfs_set_trans_block_group(trans, dir); + err = btrfs_find_free_ino(root, &objectid); + if (err) + goto out_unlock; inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name, - dentry->d_name.len, - dentry->d_parent->d_inode->i_ino, - objectid, BTRFS_I(dir)->block_group, mode, - &index); - err = PTR_ERR(inode); - if (IS_ERR(inode)) + dentry->d_name.len, btrfs_ino(dir), objectid, + mode, &index); + if (IS_ERR(inode)) { + err = PTR_ERR(inode); goto out_unlock; + } + drop_inode_on_err = 1; - err = btrfs_init_inode_security(trans, inode, dir); - if (err) { - drop_inode = 1; + err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name); + if (err) goto out_unlock; - } - btrfs_set_trans_block_group(trans, inode); - err = btrfs_add_nondir(trans, dentry, inode, 0, index); + err = btrfs_update_inode(trans, root, inode); if (err) - drop_inode = 1; - else { - inode->i_mapping->a_ops = &btrfs_aops; - inode->i_mapping->backing_dev_info = &root->fs_info->bdi; - inode->i_fop = &btrfs_file_operations; - inode->i_op = &btrfs_file_inode_operations; - BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; - } - btrfs_update_inode_block_group(trans, inode); - btrfs_update_inode_block_group(trans, dir); + goto out_unlock; + + /* + * If the active LSM wants to access the inode during + * d_instantiate it needs these. Smack checks to see + * if the filesystem supports xattrs by looking at the + * ops vector. + */ + inode->i_fop = &btrfs_file_operations; + inode->i_op = &btrfs_file_inode_operations; + + err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index); + if (err) + goto out_unlock; + + inode->i_mapping->a_ops = &btrfs_aops; + inode->i_mapping->backing_dev_info = &root->fs_info->bdi; + BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; + d_instantiate(dentry, inode); + out_unlock: - nr = trans->blocks_used; - btrfs_end_transaction_throttle(trans, root); - if (drop_inode) { + btrfs_end_transaction(trans, root); + if (err && drop_inode_on_err) { inode_dec_link_count(inode); iput(inode); } - btrfs_btree_balance_dirty(root, nr); + btrfs_balance_delayed_items(root); + btrfs_btree_balance_dirty(root); return err; } @@ -4775,59 +5950,73 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir, struct btrfs_root *root = BTRFS_I(dir)->root; struct inode *inode = old_dentry->d_inode; u64 index; - unsigned long nr = 0; int err; int drop_inode = 0; - if (inode->i_nlink == 0) - return -ENOENT; - /* do not allow sys_link's with other subvols of the same device */ if (root->objectid != BTRFS_I(inode)->root->objectid) - return -EPERM; + return -EXDEV; - btrfs_inc_nlink(inode); + if (inode->i_nlink >= BTRFS_LINK_MAX) + return -EMLINK; err = btrfs_set_inode_index(dir, &index); if (err) goto fail; /* - * 1 item for inode ref + * 2 items for inode and inode ref * 2 items for dir items + * 1 item for parent inode */ - trans = btrfs_start_transaction(root, 3); + trans = btrfs_start_transaction(root, 5); if (IS_ERR(trans)) { err = PTR_ERR(trans); goto fail; } - btrfs_set_trans_block_group(trans, dir); + /* There are several dir indexes for this inode, clear the cache. */ + BTRFS_I(inode)->dir_index = 0ULL; + inc_nlink(inode); + inode_inc_iversion(inode); + inode->i_ctime = CURRENT_TIME; ihold(inode); + set_bit(BTRFS_INODE_COPY_EVERYTHING, &BTRFS_I(inode)->runtime_flags); - err = btrfs_add_nondir(trans, dentry, inode, 1, index); + err = btrfs_add_nondir(trans, dir, dentry, inode, 1, index); if (err) { drop_inode = 1; } else { - btrfs_update_inode_block_group(trans, dir); + struct dentry *parent = dentry->d_parent; err = btrfs_update_inode(trans, root, inode); - BUG_ON(err); - btrfs_log_new_name(trans, inode, NULL, dentry->d_parent); + if (err) + goto fail; + if (inode->i_nlink == 1) { + /* + * If new hard link count is 1, it's a file created + * with open(2) O_TMPFILE flag. + */ + err = btrfs_orphan_del(trans, inode); + if (err) + goto fail; + } + d_instantiate(dentry, inode); + btrfs_log_new_name(trans, inode, NULL, parent); } - nr = trans->blocks_used; - btrfs_end_transaction_throttle(trans, root); + btrfs_end_transaction(trans, root); + btrfs_balance_delayed_items(root); fail: if (drop_inode) { inode_dec_link_count(inode); iput(inode); } - btrfs_btree_balance_dirty(root, nr); + btrfs_btree_balance_dirty(root); return err; } -static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) +static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode) { struct inode *inode = NULL; struct btrfs_trans_handle *trans; @@ -4836,11 +6025,6 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) int drop_on_err = 0; u64 objectid = 0; u64 index = 0; - unsigned long nr = 1; - - err = btrfs_find_free_objectid(NULL, root, dir->i_ino, &objectid); - if (err) - return err; /* * 2 items for inode and ref @@ -4850,13 +6034,14 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) trans = btrfs_start_transaction(root, 5); if (IS_ERR(trans)) return PTR_ERR(trans); - btrfs_set_trans_block_group(trans, dir); + + err = btrfs_find_free_ino(root, &objectid); + if (err) + goto out_fail; inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name, - dentry->d_name.len, - dentry->d_parent->d_inode->i_ino, objectid, - BTRFS_I(dir)->block_group, S_IFDIR | mode, - &index); + dentry->d_name.len, btrfs_ino(dir), objectid, + S_IFDIR | mode, &index); if (IS_ERR(inode)) { err = PTR_ERR(inode); goto out_fail; @@ -4864,36 +6049,32 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) drop_on_err = 1; - err = btrfs_init_inode_security(trans, inode, dir); + err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name); if (err) goto out_fail; inode->i_op = &btrfs_dir_inode_operations; inode->i_fop = &btrfs_dir_file_operations; - btrfs_set_trans_block_group(trans, inode); btrfs_i_size_write(inode, 0); err = btrfs_update_inode(trans, root, inode); if (err) goto out_fail; - err = btrfs_add_link(trans, dentry->d_parent->d_inode, - inode, dentry->d_name.name, - dentry->d_name.len, 0, index); + err = btrfs_add_link(trans, dir, inode, dentry->d_name.name, + dentry->d_name.len, 0, index); if (err) goto out_fail; d_instantiate(dentry, inode); drop_on_err = 0; - btrfs_update_inode_block_group(trans, inode); - btrfs_update_inode_block_group(trans, dir); out_fail: - nr = trans->blocks_used; - btrfs_end_transaction_throttle(trans, root); + btrfs_end_transaction(trans, root); if (drop_on_err) iput(inode); - btrfs_btree_balance_dirty(root, nr); + btrfs_balance_delayed_items(root); + btrfs_btree_balance_dirty(root); return err; } @@ -4917,7 +6098,7 @@ static int merge_extent_mapping(struct extent_map_tree *em_tree, em->block_start += start_diff; em->block_len -= start_diff; } - return add_extent_mapping(em_tree, em); + return add_extent_mapping(em_tree, em, 0); } static noinline int uncompress_inline(struct btrfs_path *path, @@ -4931,29 +6112,25 @@ static noinline int uncompress_inline(struct btrfs_path *path, size_t max_size; unsigned long inline_size; unsigned long ptr; + int compress_type; WARN_ON(pg_offset != 0); + compress_type = btrfs_file_extent_compression(leaf, item); max_size = btrfs_file_extent_ram_bytes(leaf, item); inline_size = btrfs_file_extent_inline_item_len(leaf, - btrfs_item_nr(leaf, path->slots[0])); + btrfs_item_nr(path->slots[0])); tmp = kmalloc(inline_size, GFP_NOFS); + if (!tmp) + return -ENOMEM; ptr = btrfs_file_extent_inline_start(item); read_extent_buffer(leaf, tmp, ptr, inline_size); max_size = min_t(unsigned long, PAGE_CACHE_SIZE, max_size); - ret = btrfs_zlib_decompress(tmp, page, extent_offset, - inline_size, max_size); - if (ret) { - char *kaddr = kmap_atomic(page, KM_USER0); - unsigned long copy_size = min_t(u64, - PAGE_CACHE_SIZE - pg_offset, - max_size - extent_offset); - memset(kaddr + pg_offset, 0, copy_size); - kunmap_atomic(kaddr, KM_USER0); - } + ret = btrfs_decompress(compress_type, tmp, page, + extent_offset, inline_size, max_size); kfree(tmp); - return 0; + return ret; } /* @@ -4971,10 +6148,9 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, { int ret; int err = 0; - u64 bytenr; u64 extent_start = 0; u64 extent_end = 0; - u64 objectid = inode->i_ino; + u64 objectid = btrfs_ino(inode); u32 found_type; struct btrfs_path *path = NULL; struct btrfs_root *root = BTRFS_I(inode)->root; @@ -4985,7 +6161,7 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_trans_handle *trans = NULL; - int compressed; + const bool new_inline = !page || create; again: read_lock(&em_tree->lock); @@ -5002,7 +6178,7 @@ again: else goto out; } - em = alloc_extent_map(GFP_NOFS); + em = alloc_extent_map(); if (!em) { err = -ENOMEM; goto out; @@ -5015,7 +6191,15 @@ again: if (!path) { path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) { + err = -ENOMEM; + goto out; + } + /* + * Chances are we'll be called again, so go ahead and do + * readahead + */ + path->reada = 1; } ret = btrfs_lookup_file_extent(trans, root, path, @@ -5039,23 +6223,28 @@ again: found_type = btrfs_key_type(&found_key); if (found_key.objectid != objectid || found_type != BTRFS_EXTENT_DATA_KEY) { - goto not_found; + /* + * If we backup past the first extent we want to move forward + * and see if there is an extent in front of us, otherwise we'll + * say there is a hole for our whole search range which can + * cause problems. + */ + extent_end = start; + goto next; } found_type = btrfs_file_extent_type(leaf, item); extent_start = found_key.offset; - compressed = btrfs_file_extent_compression(leaf, item); if (found_type == BTRFS_FILE_EXTENT_REG || found_type == BTRFS_FILE_EXTENT_PREALLOC) { extent_end = extent_start + btrfs_file_extent_num_bytes(leaf, item); } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { size_t size; - size = btrfs_file_extent_inline_len(leaf, item); - extent_end = (extent_start + size + root->sectorsize - 1) & - ~((u64)root->sectorsize - 1); + size = btrfs_file_extent_inline_len(leaf, path->slots[0], item); + extent_end = ALIGN(extent_start + size, root->sectorsize); } - +next: if (start >= extent_end) { path->slots[0]++; if (path->slots[0] >= btrfs_header_nritems(leaf)) { @@ -5075,33 +6264,15 @@ again: if (start + len <= found_key.offset) goto not_found; em->start = start; + em->orig_start = start; em->len = found_key.offset - start; goto not_found_em; } + btrfs_extent_item_to_extent_map(inode, path, item, new_inline, em); + if (found_type == BTRFS_FILE_EXTENT_REG || found_type == BTRFS_FILE_EXTENT_PREALLOC) { - em->start = extent_start; - em->len = extent_end - extent_start; - em->orig_start = extent_start - - btrfs_file_extent_offset(leaf, item); - bytenr = btrfs_file_extent_disk_bytenr(leaf, item); - if (bytenr == 0) { - em->block_start = EXTENT_MAP_HOLE; - goto insert; - } - if (compressed) { - set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); - em->block_start = bytenr; - em->block_len = btrfs_file_extent_disk_num_bytes(leaf, - item); - } else { - bytenr += btrfs_file_extent_offset(leaf, item); - em->block_start = bytenr; - em->block_len = em->len; - if (found_type == BTRFS_FILE_EXTENT_PREALLOC) - set_bit(EXTENT_FLAG_PREALLOC, &em->flags); - } goto insert; } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { unsigned long ptr; @@ -5110,31 +6281,28 @@ again: size_t extent_offset; size_t copy_size; - em->block_start = EXTENT_MAP_INLINE; - if (!page || create) { - em->start = extent_start; - em->len = extent_end - extent_start; + if (new_inline) goto out; - } - size = btrfs_file_extent_inline_len(leaf, item); + size = btrfs_file_extent_inline_len(leaf, path->slots[0], item); extent_offset = page_offset(page) + pg_offset - extent_start; copy_size = min_t(u64, PAGE_CACHE_SIZE - pg_offset, size - extent_offset); em->start = extent_start + extent_offset; - em->len = (copy_size + root->sectorsize - 1) & - ~((u64)root->sectorsize - 1); - em->orig_start = EXTENT_MAP_INLINE; - if (compressed) - set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); + em->len = ALIGN(copy_size, root->sectorsize); + em->orig_block_len = em->len; + em->orig_start = em->start; ptr = btrfs_file_extent_inline_start(item) + extent_offset; if (create == 0 && !PageUptodate(page)) { - if (btrfs_file_extent_compression(leaf, item) == - BTRFS_COMPRESS_ZLIB) { + if (btrfs_file_extent_compression(leaf, item) != + BTRFS_COMPRESS_NONE) { ret = uncompress_inline(path, inode, page, pg_offset, extent_offset, item); - BUG_ON(ret); + if (ret) { + err = ret; + goto out; + } } else { map = kmap(page); read_extent_buffer(leaf, map + pg_offset, ptr, @@ -5148,13 +6316,17 @@ again: } flush_dcache_page(page); } else if (create && PageUptodate(page)) { - WARN_ON(1); + BUG(); if (!trans) { kunmap(page); free_extent_map(em); em = NULL; - btrfs_release_path(root, path); - trans = btrfs_join_transaction(root, 1); + + btrfs_release_path(path); + trans = btrfs_join_transaction(root); + + if (IS_ERR(trans)) + return ERR_CAST(trans); goto again; } map = kmap(page); @@ -5164,33 +6336,28 @@ again: btrfs_mark_buffer_dirty(leaf); } set_extent_uptodate(io_tree, em->start, - extent_map_end(em) - 1, GFP_NOFS); + extent_map_end(em) - 1, NULL, GFP_NOFS); goto insert; - } else { - printk(KERN_ERR "btrfs unknown found_type %d\n", found_type); - WARN_ON(1); } not_found: em->start = start; + em->orig_start = start; em->len = len; not_found_em: em->block_start = EXTENT_MAP_HOLE; set_bit(EXTENT_FLAG_VACANCY, &em->flags); insert: - btrfs_release_path(root, path); + btrfs_release_path(path); if (em->start > start || extent_map_end(em) <= start) { - printk(KERN_ERR "Btrfs: bad extent! em: [%llu %llu] passed " - "[%llu %llu]\n", (unsigned long long)em->start, - (unsigned long long)em->len, - (unsigned long long)start, - (unsigned long long)len); + btrfs_err(root->fs_info, "bad extent! em: [%llu %llu] passed [%llu %llu]", + em->start, em->len, start, len); err = -EIO; goto out; } err = 0; write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, em); + ret = add_extent_mapping(em_tree, em, 0); /* it is possible that someone inserted the extent into the tree * while we had the lock dropped. It is also possible that * an overlapping map exists in the tree @@ -5231,6 +6398,9 @@ insert: } write_unlock(&em_tree->lock); out: + + trace_btrfs_get_extent(root, em); + if (path) btrfs_free_path(path); if (trans) { @@ -5242,68 +6412,167 @@ out: free_extent_map(em); return ERR_PTR(err); } + BUG_ON(!em); /* Error is always set */ return em; } -static struct extent_map *btrfs_new_extent_direct(struct inode *inode, - u64 start, u64 len) +struct extent_map *btrfs_get_extent_fiemap(struct inode *inode, struct page *page, + size_t pg_offset, u64 start, u64 len, + int create) { - struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_trans_handle *trans; struct extent_map *em; - struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; - struct btrfs_key ins; - u64 alloc_hint; - int ret; + struct extent_map *hole_em = NULL; + u64 range_start = start; + u64 end; + u64 found; + u64 found_end; + int err = 0; - btrfs_drop_extent_cache(inode, start, start + len - 1, 0); + em = btrfs_get_extent(inode, page, pg_offset, start, len, create); + if (IS_ERR(em)) + return em; + if (em) { + /* + * if our em maps to + * - a hole or + * - a pre-alloc extent, + * there might actually be delalloc bytes behind it. + */ + if (em->block_start != EXTENT_MAP_HOLE && + !test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) + return em; + else + hole_em = em; + } - trans = btrfs_join_transaction(root, 0); - if (!trans) - return ERR_PTR(-ENOMEM); + /* check to see if we've wrapped (len == -1 or similar) */ + end = start + len; + if (end < start) + end = (u64)-1; + else + end -= 1; - trans->block_rsv = &root->fs_info->delalloc_block_rsv; + em = NULL; - alloc_hint = get_extent_allocation_hint(inode, start, len); - ret = btrfs_reserve_extent(trans, root, len, root->sectorsize, 0, - alloc_hint, (u64)-1, &ins, 1); - if (ret) { - em = ERR_PTR(ret); + /* ok, we didn't find anything, lets look for delalloc */ + found = count_range_bits(&BTRFS_I(inode)->io_tree, &range_start, + end, len, EXTENT_DELALLOC, 1); + found_end = range_start + found; + if (found_end < range_start) + found_end = (u64)-1; + + /* + * we didn't find anything useful, return + * the original results from get_extent() + */ + if (range_start > end || found_end <= start) { + em = hole_em; + hole_em = NULL; goto out; } - em = alloc_extent_map(GFP_NOFS); - if (!em) { - em = ERR_PTR(-ENOMEM); - goto out; + /* adjust the range_start to make sure it doesn't + * go backwards from the start they passed in + */ + range_start = max(start, range_start); + found = found_end - range_start; + + if (found > 0) { + u64 hole_start = start; + u64 hole_len = len; + + em = alloc_extent_map(); + if (!em) { + err = -ENOMEM; + goto out; + } + /* + * when btrfs_get_extent can't find anything it + * returns one huge hole + * + * make sure what it found really fits our range, and + * adjust to make sure it is based on the start from + * the caller + */ + if (hole_em) { + u64 calc_end = extent_map_end(hole_em); + + if (calc_end <= start || (hole_em->start > end)) { + free_extent_map(hole_em); + hole_em = NULL; + } else { + hole_start = max(hole_em->start, start); + hole_len = calc_end - hole_start; + } + } + em->bdev = NULL; + if (hole_em && range_start > hole_start) { + /* our hole starts before our delalloc, so we + * have to return just the parts of the hole + * that go until the delalloc starts + */ + em->len = min(hole_len, + range_start - hole_start); + em->start = hole_start; + em->orig_start = hole_start; + /* + * don't adjust block start at all, + * it is fixed at EXTENT_MAP_HOLE + */ + em->block_start = hole_em->block_start; + em->block_len = hole_len; + if (test_bit(EXTENT_FLAG_PREALLOC, &hole_em->flags)) + set_bit(EXTENT_FLAG_PREALLOC, &em->flags); + } else { + em->start = range_start; + em->len = found; + em->orig_start = range_start; + em->block_start = EXTENT_MAP_DELALLOC; + em->block_len = found; + } + } else if (hole_em) { + return hole_em; } +out: - em->start = start; - em->orig_start = em->start; - em->len = ins.offset; + free_extent_map(hole_em); + if (err) { + free_extent_map(em); + return ERR_PTR(err); + } + return em; +} - em->block_start = ins.objectid; - em->block_len = ins.offset; - em->bdev = root->fs_info->fs_devices->latest_bdev; - set_bit(EXTENT_FLAG_PINNED, &em->flags); +static struct extent_map *btrfs_new_extent_direct(struct inode *inode, + u64 start, u64 len) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_map *em; + struct btrfs_key ins; + u64 alloc_hint; + int ret; - while (1) { - write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, em); - write_unlock(&em_tree->lock); - if (ret != -EEXIST) - break; - btrfs_drop_extent_cache(inode, start, start + em->len - 1, 0); + alloc_hint = get_extent_allocation_hint(inode, start, len); + ret = btrfs_reserve_extent(root, len, root->sectorsize, 0, + alloc_hint, &ins, 1, 1); + if (ret) + return ERR_PTR(ret); + + em = create_pinned_em(inode, start, ins.offset, start, ins.objectid, + ins.offset, ins.offset, ins.offset, 0); + if (IS_ERR(em)) { + btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1); + return em; } ret = btrfs_add_ordered_extent_dio(inode, start, ins.objectid, ins.offset, ins.offset, 0); if (ret) { - btrfs_free_reserved_extent(root, ins.objectid, ins.offset); - em = ERR_PTR(ret); + btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1); + free_extent_map(em); + return ERR_PTR(ret); } -out: - btrfs_end_transaction(trans, root); + return em; } @@ -5311,13 +6580,16 @@ out: * returns 1 when the nocow is safe, < 1 on error, 0 if the * block must be cow'd */ -static noinline int can_nocow_odirect(struct btrfs_trans_handle *trans, - struct inode *inode, u64 offset, u64 len) +noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len, + u64 *orig_start, u64 *orig_block_len, + u64 *ram_bytes) { + struct btrfs_trans_handle *trans; struct btrfs_path *path; int ret; struct extent_buffer *leaf; struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_file_extent_item *fi; struct btrfs_key key; u64 disk_bytenr; @@ -5326,12 +6598,13 @@ static noinline int can_nocow_odirect(struct btrfs_trans_handle *trans, u64 num_bytes; int slot; int found_type; + bool nocow = (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW); path = btrfs_alloc_path(); if (!path) return -ENOMEM; - ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino, + ret = btrfs_lookup_file_extent(NULL, root, path, btrfs_ino(inode), offset, 0); if (ret < 0) goto out; @@ -5348,7 +6621,7 @@ static noinline int can_nocow_odirect(struct btrfs_trans_handle *trans, ret = 0; leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &key, slot); - if (key.objectid != inode->i_ino || + if (key.objectid != btrfs_ino(inode) || key.type != BTRFS_EXTENT_DATA_KEY) { /* not our file or wrong item type, must cow */ goto out; @@ -5366,25 +6639,66 @@ static noinline int can_nocow_odirect(struct btrfs_trans_handle *trans, /* not a regular extent, must cow */ goto out; } - disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); - backref_offset = btrfs_file_extent_offset(leaf, fi); + + if (!nocow && found_type == BTRFS_FILE_EXTENT_REG) + goto out; extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); - if (extent_end < offset + len) { - /* extent doesn't include our full range, must cow */ + if (extent_end <= offset) + goto out; + + disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); + if (disk_bytenr == 0) goto out; + + if (btrfs_file_extent_compression(leaf, fi) || + btrfs_file_extent_encryption(leaf, fi) || + btrfs_file_extent_other_encoding(leaf, fi)) + goto out; + + backref_offset = btrfs_file_extent_offset(leaf, fi); + + if (orig_start) { + *orig_start = key.offset - backref_offset; + *orig_block_len = btrfs_file_extent_disk_num_bytes(leaf, fi); + *ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi); } if (btrfs_extent_readonly(root, disk_bytenr)) goto out; + num_bytes = min(offset + *len, extent_end) - offset; + if (!nocow && found_type == BTRFS_FILE_EXTENT_PREALLOC) { + u64 range_end; + + range_end = round_up(offset + num_bytes, root->sectorsize) - 1; + ret = test_range_bit(io_tree, offset, range_end, + EXTENT_DELALLOC, 0, NULL); + if (ret) { + ret = -EAGAIN; + goto out; + } + } + + btrfs_release_path(path); + /* * look for other files referencing this extent, if we * find any we must cow */ - if (btrfs_cross_ref_exist(trans, root, inode->i_ino, - key.offset - backref_offset, disk_bytenr)) + trans = btrfs_join_transaction(root); + if (IS_ERR(trans)) { + ret = 0; goto out; + } + + ret = btrfs_cross_ref_exist(trans, root, btrfs_ino(inode), + key.offset - backref_offset, disk_bytenr); + btrfs_end_transaction(trans, root); + if (ret) { + ret = 0; + goto out; + } /* * adjust disk_bytenr and num_bytes to cover just the bytes @@ -5394,31 +6708,229 @@ static noinline int can_nocow_odirect(struct btrfs_trans_handle *trans, */ disk_bytenr += backref_offset; disk_bytenr += offset - key.offset; - num_bytes = min(offset + len, extent_end) - offset; if (csum_exist_in_range(root, disk_bytenr, num_bytes)) goto out; /* * all of the above have passed, it is safe to overwrite this extent * without cow */ + *len = num_bytes; ret = 1; out: btrfs_free_path(path); return ret; } +bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end) +{ + struct radix_tree_root *root = &inode->i_mapping->page_tree; + int found = false; + void **pagep = NULL; + struct page *page = NULL; + int start_idx; + int end_idx; + + start_idx = start >> PAGE_CACHE_SHIFT; + + /* + * end is the last byte in the last page. end == start is legal + */ + end_idx = end >> PAGE_CACHE_SHIFT; + + rcu_read_lock(); + + /* Most of the code in this while loop is lifted from + * find_get_page. It's been modified to begin searching from a + * page and return just the first page found in that range. If the + * found idx is less than or equal to the end idx then we know that + * a page exists. If no pages are found or if those pages are + * outside of the range then we're fine (yay!) */ + while (page == NULL && + radix_tree_gang_lookup_slot(root, &pagep, NULL, start_idx, 1)) { + page = radix_tree_deref_slot(pagep); + if (unlikely(!page)) + break; + + if (radix_tree_exception(page)) { + if (radix_tree_deref_retry(page)) { + page = NULL; + continue; + } + /* + * Otherwise, shmem/tmpfs must be storing a swap entry + * here as an exceptional entry: so return it without + * attempting to raise page count. + */ + page = NULL; + break; /* TODO: Is this relevant for this use case? */ + } + + if (!page_cache_get_speculative(page)) { + page = NULL; + continue; + } + + /* + * Has the page moved? + * This is part of the lockless pagecache protocol. See + * include/linux/pagemap.h for details. + */ + if (unlikely(page != *pagep)) { + page_cache_release(page); + page = NULL; + } + } + + if (page) { + if (page->index <= end_idx) + found = true; + page_cache_release(page); + } + + rcu_read_unlock(); + return found; +} + +static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend, + struct extent_state **cached_state, int writing) +{ + struct btrfs_ordered_extent *ordered; + int ret = 0; + + while (1) { + lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, + 0, cached_state); + /* + * We're concerned with the entire range that we're going to be + * doing DIO to, so we need to make sure theres no ordered + * extents in this range. + */ + ordered = btrfs_lookup_ordered_range(inode, lockstart, + lockend - lockstart + 1); + + /* + * We need to make sure there are no buffered pages in this + * range either, we could have raced between the invalidate in + * generic_file_direct_write and locking the extent. The + * invalidate needs to happen so that reads after a write do not + * get stale data. + */ + if (!ordered && + (!writing || + !btrfs_page_exists_in_range(inode, lockstart, lockend))) + break; + + unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, + cached_state, GFP_NOFS); + + if (ordered) { + btrfs_start_ordered_extent(inode, ordered, 1); + btrfs_put_ordered_extent(ordered); + } else { + /* Screw you mmap */ + ret = filemap_write_and_wait_range(inode->i_mapping, + lockstart, + lockend); + if (ret) + break; + + /* + * If we found a page that couldn't be invalidated just + * fall back to buffered. + */ + ret = invalidate_inode_pages2_range(inode->i_mapping, + lockstart >> PAGE_CACHE_SHIFT, + lockend >> PAGE_CACHE_SHIFT); + if (ret) + break; + } + + cond_resched(); + } + + return ret; +} + +static struct extent_map *create_pinned_em(struct inode *inode, u64 start, + u64 len, u64 orig_start, + u64 block_start, u64 block_len, + u64 orig_block_len, u64 ram_bytes, + int type) +{ + struct extent_map_tree *em_tree; + struct extent_map *em; + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + + em_tree = &BTRFS_I(inode)->extent_tree; + em = alloc_extent_map(); + if (!em) + return ERR_PTR(-ENOMEM); + + em->start = start; + em->orig_start = orig_start; + em->mod_start = start; + em->mod_len = len; + em->len = len; + em->block_len = block_len; + em->block_start = block_start; + em->bdev = root->fs_info->fs_devices->latest_bdev; + em->orig_block_len = orig_block_len; + em->ram_bytes = ram_bytes; + em->generation = -1; + set_bit(EXTENT_FLAG_PINNED, &em->flags); + if (type == BTRFS_ORDERED_PREALLOC) + set_bit(EXTENT_FLAG_FILLING, &em->flags); + + do { + btrfs_drop_extent_cache(inode, em->start, + em->start + em->len - 1, 0); + write_lock(&em_tree->lock); + ret = add_extent_mapping(em_tree, em, 1); + write_unlock(&em_tree->lock); + } while (ret == -EEXIST); + + if (ret) { + free_extent_map(em); + return ERR_PTR(ret); + } + + return em; +} + + static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, struct buffer_head *bh_result, int create) { struct extent_map *em; struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_state *cached_state = NULL; u64 start = iblock << inode->i_blkbits; + u64 lockstart, lockend; u64 len = bh_result->b_size; - struct btrfs_trans_handle *trans; + int unlock_bits = EXTENT_LOCKED; + int ret = 0; + + if (create) + unlock_bits |= EXTENT_DELALLOC | EXTENT_DIRTY; + else + len = min_t(u64, len, root->sectorsize); + + lockstart = start; + lockend = start + len - 1; + + /* + * If this errors out it's because we couldn't invalidate pagecache for + * this range and we need to fallback to buffered. + */ + if (lock_extent_direct(inode, lockstart, lockend, &cached_state, create)) + return -ENOTBLK; em = btrfs_get_extent(inode, NULL, 0, start, len, 0); - if (IS_ERR(em)) - return PTR_ERR(em); + if (IS_ERR(em)) { + ret = PTR_ERR(em); + goto unlock_err; + } /* * Ok for INLINE and COMPRESSED extents we need to fallback on buffered @@ -5437,17 +6949,15 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags) || em->block_start == EXTENT_MAP_INLINE) { free_extent_map(em); - return -ENOTBLK; + ret = -ENOTBLK; + goto unlock_err; } /* Just a good old fashioned hole, return */ if (!create && (em->block_start == EXTENT_MAP_HOLE || test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { free_extent_map(em); - /* DIO will do one hole at a time, so just unlock a sector */ - unlock_extent(&BTRFS_I(inode)->io_tree, start, - start + root->sectorsize - 1, GFP_NOFS); - return 0; + goto unlock_err; } /* @@ -5460,8 +6970,9 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, * */ if (!create) { - len = em->len - (start - em->start); - goto map; + len = min(len, em->len - (start - em->start)); + lockstart = start + len; + goto unlock; } if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags) || @@ -5469,7 +6980,7 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, em->block_start != EXTENT_MAP_HOLE)) { int type; int ret; - u64 block_start; + u64 block_start, orig_start, orig_block_len, ram_bytes; if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) type = BTRFS_ORDERED_PREALLOC; @@ -5478,28 +6989,29 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, len = min(len, em->len - (start - em->start)); block_start = em->block_start + (start - em->start); - /* - * we're not going to log anything, but we do need - * to make sure the current transaction stays open - * while we look for nocow cross refs - */ - trans = btrfs_join_transaction(root, 0); - if (!trans) - goto must_cow; + if (can_nocow_extent(inode, start, &len, &orig_start, + &orig_block_len, &ram_bytes) == 1) { + if (type == BTRFS_ORDERED_PREALLOC) { + free_extent_map(em); + em = create_pinned_em(inode, start, len, + orig_start, + block_start, len, + orig_block_len, + ram_bytes, type); + if (IS_ERR(em)) + goto unlock_err; + } - if (can_nocow_odirect(trans, inode, start, len) == 1) { ret = btrfs_add_ordered_extent_dio(inode, start, block_start, len, len, type); - btrfs_end_transaction(trans, root); if (ret) { free_extent_map(em); - return ret; + goto unlock_err; } goto unlock; } - btrfs_end_transaction(trans, root); } -must_cow: + /* * this will cow the extent, reset the len in case we changed * it above @@ -5507,48 +7019,74 @@ must_cow: len = bh_result->b_size; free_extent_map(em); em = btrfs_new_extent_direct(inode, start, len); - if (IS_ERR(em)) - return PTR_ERR(em); + if (IS_ERR(em)) { + ret = PTR_ERR(em); + goto unlock_err; + } len = min(len, em->len - (start - em->start)); unlock: - clear_extent_bit(&BTRFS_I(inode)->io_tree, start, start + len - 1, - EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DIRTY, 1, - 0, NULL, GFP_NOFS); -map: bh_result->b_blocknr = (em->block_start + (start - em->start)) >> inode->i_blkbits; bh_result->b_size = len; bh_result->b_bdev = em->bdev; set_buffer_mapped(bh_result); - if (create && !test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) - set_buffer_new(bh_result); + if (create) { + if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) + set_buffer_new(bh_result); + + /* + * Need to update the i_size under the extent lock so buffered + * readers will get the updated i_size when we unlock. + */ + if (start + len > i_size_read(inode)) + i_size_write(inode, start + len); + + spin_lock(&BTRFS_I(inode)->lock); + BTRFS_I(inode)->outstanding_extents++; + spin_unlock(&BTRFS_I(inode)->lock); + + ret = set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, + lockstart + len - 1, EXTENT_DELALLOC, NULL, + &cached_state, GFP_NOFS); + BUG_ON(ret); + } + + /* + * In the case of write we need to clear and unlock the entire range, + * in the case of read we need to unlock only the end area that we + * aren't using if there is any left over space. + */ + if (lockstart < lockend) { + clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, + lockend, unlock_bits, 1, 0, + &cached_state, GFP_NOFS); + } else { + free_extent_state(cached_state); + } free_extent_map(em); return 0; -} -struct btrfs_dio_private { - struct inode *inode; - u64 logical_offset; - u64 disk_bytenr; - u64 bytes; - u32 *csums; - void *private; -}; +unlock_err: + clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend, + unlock_bits, 1, 0, &cached_state, GFP_NOFS); + return ret; +} static void btrfs_endio_direct_read(struct bio *bio, int err) { - struct bio_vec *bvec_end = bio->bi_io_vec + bio->bi_vcnt - 1; - struct bio_vec *bvec = bio->bi_io_vec; struct btrfs_dio_private *dip = bio->bi_private; + struct bio_vec *bvec; struct inode *inode = dip->inode; struct btrfs_root *root = BTRFS_I(inode)->root; + struct bio *dio_bio; + u32 *csums = (u32 *)dip->csum; u64 start; - u32 *private = dip->csums; + int i; start = dip->logical_offset; - do { + bio_for_each_segment_all(bvec, bio, i) { if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) { struct page *page = bvec->bv_page; char *kaddr; @@ -5556,35 +7094,36 @@ static void btrfs_endio_direct_read(struct bio *bio, int err) unsigned long flags; local_irq_save(flags); - kaddr = kmap_atomic(page, KM_IRQ0); - csum = btrfs_csum_data(root, kaddr + bvec->bv_offset, + kaddr = kmap_atomic(page); + csum = btrfs_csum_data(kaddr + bvec->bv_offset, csum, bvec->bv_len); btrfs_csum_final(csum, (char *)&csum); - kunmap_atomic(kaddr, KM_IRQ0); + kunmap_atomic(kaddr); local_irq_restore(flags); flush_dcache_page(bvec->bv_page); - if (csum != *private) { - printk(KERN_ERR "btrfs csum failed ino %lu off" - " %llu csum %u private %u\n", - inode->i_ino, (unsigned long long)start, - csum, *private); + if (csum != csums[i]) { + btrfs_err(root->fs_info, "csum failed ino %llu off %llu csum %u expected csum %u", + btrfs_ino(inode), start, csum, + csums[i]); err = -EIO; } } start += bvec->bv_len; - private++; - bvec++; - } while (bvec <= bvec_end); + } unlock_extent(&BTRFS_I(inode)->io_tree, dip->logical_offset, - dip->logical_offset + dip->bytes - 1, GFP_NOFS); - bio->bi_private = dip->private; + dip->logical_offset + dip->bytes - 1); + dio_bio = dip->dio_bio; - kfree(dip->csums); kfree(dip); - dio_end_io(bio, err); + + /* If we had a csum failure make sure to clear the uptodate flag */ + if (err) + clear_bit(BIO_UPTODATE, &dio_bio->bi_flags); + dio_end_io(dio_bio, err); + bio_put(bio); } static void btrfs_endio_direct_write(struct bio *bio, int err) @@ -5592,167 +7131,320 @@ static void btrfs_endio_direct_write(struct bio *bio, int err) struct btrfs_dio_private *dip = bio->bi_private; struct inode *inode = dip->inode; struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_trans_handle *trans; struct btrfs_ordered_extent *ordered = NULL; - struct extent_state *cached_state = NULL; + u64 ordered_offset = dip->logical_offset; + u64 ordered_bytes = dip->bytes; + struct bio *dio_bio; int ret; if (err) goto out_done; - - ret = btrfs_dec_test_ordered_pending(inode, &ordered, - dip->logical_offset, dip->bytes); +again: + ret = btrfs_dec_test_first_ordered_pending(inode, &ordered, + &ordered_offset, + ordered_bytes, !err); if (!ret) - goto out_done; + goto out_test; - BUG_ON(!ordered); - - trans = btrfs_join_transaction(root, 1); - if (!trans) { - err = -ENOMEM; - goto out; + btrfs_init_work(&ordered->work, finish_ordered_fn, NULL, NULL); + btrfs_queue_work(root->fs_info->endio_write_workers, + &ordered->work); +out_test: + /* + * our bio might span multiple ordered extents. If we haven't + * completed the accounting for the whole dio, go back and try again + */ + if (ordered_offset < dip->logical_offset + dip->bytes) { + ordered_bytes = dip->logical_offset + dip->bytes - + ordered_offset; + ordered = NULL; + goto again; } - trans->block_rsv = &root->fs_info->delalloc_block_rsv; +out_done: + dio_bio = dip->dio_bio; - if (test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags)) { - ret = btrfs_ordered_update_i_size(inode, 0, ordered); - if (!ret) - ret = btrfs_update_inode(trans, root, inode); - err = ret; - goto out; + kfree(dip); + + /* If we had an error make sure to clear the uptodate flag */ + if (err) + clear_bit(BIO_UPTODATE, &dio_bio->bi_flags); + dio_end_io(dio_bio, err); + bio_put(bio); +} + +static int __btrfs_submit_bio_start_direct_io(struct inode *inode, int rw, + struct bio *bio, int mirror_num, + unsigned long bio_flags, u64 offset) +{ + int ret; + struct btrfs_root *root = BTRFS_I(inode)->root; + ret = btrfs_csum_one_bio(root, inode, bio, offset, 1); + BUG_ON(ret); /* -ENOMEM */ + return 0; +} + +static void btrfs_end_dio_bio(struct bio *bio, int err) +{ + struct btrfs_dio_private *dip = bio->bi_private; + + if (err) { + btrfs_err(BTRFS_I(dip->inode)->root->fs_info, + "direct IO failed ino %llu rw %lu sector %#Lx len %u err no %d", + btrfs_ino(dip->inode), bio->bi_rw, + (unsigned long long)bio->bi_iter.bi_sector, + bio->bi_iter.bi_size, err); + dip->errors = 1; + + /* + * before atomic variable goto zero, we must make sure + * dip->errors is perceived to be set. + */ + smp_mb__before_atomic(); } - lock_extent_bits(&BTRFS_I(inode)->io_tree, ordered->file_offset, - ordered->file_offset + ordered->len - 1, 0, - &cached_state, GFP_NOFS); + /* if there are more bios still pending for this dio, just exit */ + if (!atomic_dec_and_test(&dip->pending_bios)) + goto out; - if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags)) { - ret = btrfs_mark_extent_written(trans, inode, - ordered->file_offset, - ordered->file_offset + - ordered->len); - if (ret) { - err = ret; - goto out_unlock; - } + if (dip->errors) { + bio_io_error(dip->orig_bio); } else { - ret = insert_reserved_file_extent(trans, inode, - ordered->file_offset, - ordered->start, - ordered->disk_len, - ordered->len, - ordered->len, - 0, 0, 0, - BTRFS_FILE_EXTENT_REG); - unpin_extent_cache(&BTRFS_I(inode)->extent_tree, - ordered->file_offset, ordered->len); - if (ret) { - err = ret; - WARN_ON(1); - goto out_unlock; - } + set_bit(BIO_UPTODATE, &dip->dio_bio->bi_flags); + bio_endio(dip->orig_bio, 0); } - - add_pending_csums(trans, inode, ordered->file_offset, &ordered->list); - btrfs_ordered_update_i_size(inode, 0, ordered); - btrfs_update_inode(trans, root, inode); -out_unlock: - unlock_extent_cached(&BTRFS_I(inode)->io_tree, ordered->file_offset, - ordered->file_offset + ordered->len - 1, - &cached_state, GFP_NOFS); out: - btrfs_delalloc_release_metadata(inode, ordered->len); - btrfs_end_transaction(trans, root); - btrfs_put_ordered_extent(ordered); - btrfs_put_ordered_extent(ordered); -out_done: - bio->bi_private = dip->private; + bio_put(bio); +} - kfree(dip->csums); - kfree(dip); - dio_end_io(bio, err); +static struct bio *btrfs_dio_bio_alloc(struct block_device *bdev, + u64 first_sector, gfp_t gfp_flags) +{ + int nr_vecs = bio_get_nr_vecs(bdev); + return btrfs_bio_alloc(bdev, first_sector, nr_vecs, gfp_flags); } -static int __btrfs_submit_bio_start_direct_io(struct inode *inode, int rw, - struct bio *bio, int mirror_num, - unsigned long bio_flags, u64 offset) +static inline int __btrfs_submit_dio_bio(struct bio *bio, struct inode *inode, + int rw, u64 file_offset, int skip_sum, + int async_submit) { + struct btrfs_dio_private *dip = bio->bi_private; + int write = rw & REQ_WRITE; + struct btrfs_root *root = BTRFS_I(inode)->root; int ret; + + if (async_submit) + async_submit = !atomic_read(&BTRFS_I(inode)->sync_writers); + + bio_get(bio); + + if (!write) { + ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0); + if (ret) + goto err; + } + + if (skip_sum) + goto map; + + if (write && async_submit) { + ret = btrfs_wq_submit_bio(root->fs_info, + inode, rw, bio, 0, 0, + file_offset, + __btrfs_submit_bio_start_direct_io, + __btrfs_submit_bio_done); + goto err; + } else if (write) { + /* + * If we aren't doing async submit, calculate the csum of the + * bio now. + */ + ret = btrfs_csum_one_bio(root, inode, bio, file_offset, 1); + if (ret) + goto err; + } else if (!skip_sum) { + ret = btrfs_lookup_bio_sums_dio(root, inode, dip, bio, + file_offset); + if (ret) + goto err; + } + +map: + ret = btrfs_map_bio(root, rw, bio, 0, async_submit); +err: + bio_put(bio); + return ret; +} + +static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip, + int skip_sum) +{ + struct inode *inode = dip->inode; struct btrfs_root *root = BTRFS_I(inode)->root; - ret = btrfs_csum_one_bio(root, inode, bio, offset, 1); - BUG_ON(ret); + struct bio *bio; + struct bio *orig_bio = dip->orig_bio; + struct bio_vec *bvec = orig_bio->bi_io_vec; + u64 start_sector = orig_bio->bi_iter.bi_sector; + u64 file_offset = dip->logical_offset; + u64 submit_len = 0; + u64 map_length; + int nr_pages = 0; + int ret = 0; + int async_submit = 0; + + map_length = orig_bio->bi_iter.bi_size; + ret = btrfs_map_block(root->fs_info, rw, start_sector << 9, + &map_length, NULL, 0); + if (ret) { + bio_put(orig_bio); + return -EIO; + } + + if (map_length >= orig_bio->bi_iter.bi_size) { + bio = orig_bio; + goto submit; + } + + /* async crcs make it difficult to collect full stripe writes. */ + if (btrfs_get_alloc_profile(root, 1) & + (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) + async_submit = 0; + else + async_submit = 1; + + bio = btrfs_dio_bio_alloc(orig_bio->bi_bdev, start_sector, GFP_NOFS); + if (!bio) + return -ENOMEM; + bio->bi_private = dip; + bio->bi_end_io = btrfs_end_dio_bio; + atomic_inc(&dip->pending_bios); + + while (bvec <= (orig_bio->bi_io_vec + orig_bio->bi_vcnt - 1)) { + if (unlikely(map_length < submit_len + bvec->bv_len || + bio_add_page(bio, bvec->bv_page, bvec->bv_len, + bvec->bv_offset) < bvec->bv_len)) { + /* + * inc the count before we submit the bio so + * we know the end IO handler won't happen before + * we inc the count. Otherwise, the dip might get freed + * before we're done setting it up + */ + atomic_inc(&dip->pending_bios); + ret = __btrfs_submit_dio_bio(bio, inode, rw, + file_offset, skip_sum, + async_submit); + if (ret) { + bio_put(bio); + atomic_dec(&dip->pending_bios); + goto out_err; + } + + start_sector += submit_len >> 9; + file_offset += submit_len; + + submit_len = 0; + nr_pages = 0; + + bio = btrfs_dio_bio_alloc(orig_bio->bi_bdev, + start_sector, GFP_NOFS); + if (!bio) + goto out_err; + bio->bi_private = dip; + bio->bi_end_io = btrfs_end_dio_bio; + + map_length = orig_bio->bi_iter.bi_size; + ret = btrfs_map_block(root->fs_info, rw, + start_sector << 9, + &map_length, NULL, 0); + if (ret) { + bio_put(bio); + goto out_err; + } + } else { + submit_len += bvec->bv_len; + nr_pages++; + bvec++; + } + } + +submit: + ret = __btrfs_submit_dio_bio(bio, inode, rw, file_offset, skip_sum, + async_submit); + if (!ret) + return 0; + + bio_put(bio); +out_err: + dip->errors = 1; + /* + * before atomic variable goto zero, we must + * make sure dip->errors is perceived to be set. + */ + smp_mb__before_atomic(); + if (atomic_dec_and_test(&dip->pending_bios)) + bio_io_error(dip->orig_bio); + + /* bio_end_io() will handle error, so we needn't return it */ return 0; } -static void btrfs_submit_direct(int rw, struct bio *bio, struct inode *inode, - loff_t file_offset) +static void btrfs_submit_direct(int rw, struct bio *dio_bio, + struct inode *inode, loff_t file_offset) { struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_dio_private *dip; - struct bio_vec *bvec = bio->bi_io_vec; + struct bio *io_bio; int skip_sum; + int sum_len; int write = rw & REQ_WRITE; int ret = 0; + u16 csum_size; skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM; - dip = kmalloc(sizeof(*dip), GFP_NOFS); - if (!dip) { + io_bio = btrfs_bio_clone(dio_bio, GFP_NOFS); + if (!io_bio) { ret = -ENOMEM; goto free_ordered; } - dip->csums = NULL; - if (!skip_sum) { - dip->csums = kmalloc(sizeof(u32) * bio->bi_vcnt, GFP_NOFS); - if (!dip->csums) { - ret = -ENOMEM; - goto free_ordered; - } + if (!skip_sum && !write) { + csum_size = btrfs_super_csum_size(root->fs_info->super_copy); + sum_len = dio_bio->bi_iter.bi_size >> + inode->i_sb->s_blocksize_bits; + sum_len *= csum_size; + } else { + sum_len = 0; + } + + dip = kmalloc(sizeof(*dip) + sum_len, GFP_NOFS); + if (!dip) { + ret = -ENOMEM; + goto free_io_bio; } - dip->private = bio->bi_private; + dip->private = dio_bio->bi_private; dip->inode = inode; dip->logical_offset = file_offset; - - dip->bytes = 0; - do { - dip->bytes += bvec->bv_len; - bvec++; - } while (bvec <= (bio->bi_io_vec + bio->bi_vcnt - 1)); - - dip->disk_bytenr = (u64)bio->bi_sector << 9; - bio->bi_private = dip; + dip->bytes = dio_bio->bi_iter.bi_size; + dip->disk_bytenr = (u64)dio_bio->bi_iter.bi_sector << 9; + io_bio->bi_private = dip; + dip->errors = 0; + dip->orig_bio = io_bio; + dip->dio_bio = dio_bio; + atomic_set(&dip->pending_bios, 0); if (write) - bio->bi_end_io = btrfs_endio_direct_write; + io_bio->bi_end_io = btrfs_endio_direct_write; else - bio->bi_end_io = btrfs_endio_direct_read; - - ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0); - if (ret) - goto out_err; + io_bio->bi_end_io = btrfs_endio_direct_read; - if (write && !skip_sum) { - ret = btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info, - inode, rw, bio, 0, 0, - dip->logical_offset, - __btrfs_submit_bio_start_direct_io, - __btrfs_submit_bio_done); - if (ret) - goto out_err; + ret = btrfs_submit_direct_hook(rw, dip, skip_sum); + if (!ret) return; - } else if (!skip_sum) - btrfs_lookup_bio_sums_dio(root, inode, bio, - dip->logical_offset, dip->csums); - ret = btrfs_map_bio(root, rw, bio, 0, 1); - if (ret) - goto out_err; - return; -out_err: - kfree(dip->csums); - kfree(dip); +free_io_bio: + bio_put(io_bio); + free_ordered: /* * If this is a write, we need to clean up the reserved space and kill @@ -5760,147 +7452,139 @@ free_ordered: */ if (write) { struct btrfs_ordered_extent *ordered; - ordered = btrfs_lookup_ordered_extent(inode, - dip->logical_offset); + ordered = btrfs_lookup_ordered_extent(inode, file_offset); if (!test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags) && !test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags)) btrfs_free_reserved_extent(root, ordered->start, - ordered->disk_len); + ordered->disk_len, 1); btrfs_put_ordered_extent(ordered); btrfs_put_ordered_extent(ordered); } - bio_endio(bio, ret); + bio_endio(dio_bio, ret); } static ssize_t check_direct_IO(struct btrfs_root *root, int rw, struct kiocb *iocb, - const struct iovec *iov, loff_t offset, - unsigned long nr_segs) + const struct iov_iter *iter, loff_t offset) { int seg; - size_t size; - unsigned long addr; + int i; unsigned blocksize_mask = root->sectorsize - 1; ssize_t retval = -EINVAL; - loff_t end = offset; if (offset & blocksize_mask) goto out; - /* Check the memory alignment. Blocks cannot straddle pages */ - for (seg = 0; seg < nr_segs; seg++) { - addr = (unsigned long)iov[seg].iov_base; - size = iov[seg].iov_len; - end += size; - if ((addr & blocksize_mask) || (size & blocksize_mask)) - goto out; + if (iov_iter_alignment(iter) & blocksize_mask) + goto out; + + /* If this is a write we don't need to check anymore */ + if (rw & WRITE) + return 0; + /* + * Check to make sure we don't have duplicate iov_base's in this + * iovec, if so return EINVAL, otherwise we'll get csum errors + * when reading back. + */ + for (seg = 0; seg < iter->nr_segs; seg++) { + for (i = seg + 1; i < iter->nr_segs; i++) { + if (iter->iov[seg].iov_base == iter->iov[i].iov_base) + goto out; + } } retval = 0; out: return retval; } + static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb, - const struct iovec *iov, loff_t offset, - unsigned long nr_segs) + struct iov_iter *iter, loff_t offset) { struct file *file = iocb->ki_filp; struct inode *inode = file->f_mapping->host; - struct btrfs_ordered_extent *ordered; - struct extent_state *cached_state = NULL; - u64 lockstart, lockend; + size_t count = 0; + int flags = 0; + bool wakeup = true; + bool relock = false; ssize_t ret; - int writing = rw & WRITE; - int write_bits = 0; - size_t count = iov_length(iov, nr_segs); - if (check_direct_IO(BTRFS_I(inode)->root, rw, iocb, iov, - offset, nr_segs)) { + if (check_direct_IO(BTRFS_I(inode)->root, rw, iocb, iter, offset)) return 0; - } - lockstart = offset; - lockend = offset + count - 1; + atomic_inc(&inode->i_dio_count); + smp_mb__after_atomic(); - if (writing) { - ret = btrfs_delalloc_reserve_space(inode, count); - if (ret) - goto out; - } + /* + * The generic stuff only does filemap_write_and_wait_range, which + * isn't enough if we've written compressed pages to this area, so + * we need to flush the dirty pages again to make absolutely sure + * that any outstanding dirty pages are on disk. + */ + count = iov_iter_count(iter); + if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, + &BTRFS_I(inode)->runtime_flags)) + filemap_fdatawrite_range(inode->i_mapping, offset, count); - while (1) { - lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, - 0, &cached_state, GFP_NOFS); + if (rw & WRITE) { /* - * We're concerned with the entire range that we're going to be - * doing DIO to, so we need to make sure theres no ordered - * extents in this range. + * If the write DIO is beyond the EOF, we need update + * the isize, but it is protected by i_mutex. So we can + * not unlock the i_mutex at this case. */ - ordered = btrfs_lookup_ordered_range(inode, lockstart, - lockend - lockstart + 1); - if (!ordered) - break; - unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, - &cached_state, GFP_NOFS); - btrfs_start_ordered_extent(inode, ordered, 1); - btrfs_put_ordered_extent(ordered); - cond_resched(); - } - - /* - * we don't use btrfs_set_extent_delalloc because we don't want - * the dirty or uptodate bits - */ - if (writing) { - write_bits = EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING; - ret = set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend, - EXTENT_DELALLOC, 0, NULL, &cached_state, - GFP_NOFS); - if (ret) { - clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, - lockend, EXTENT_LOCKED | write_bits, - 1, 0, &cached_state, GFP_NOFS); - goto out; + if (offset + count <= inode->i_size) { + mutex_unlock(&inode->i_mutex); + relock = true; } + ret = btrfs_delalloc_reserve_space(inode, count); + if (ret) + goto out; + } else if (unlikely(test_bit(BTRFS_INODE_READDIO_NEED_LOCK, + &BTRFS_I(inode)->runtime_flags))) { + inode_dio_done(inode); + flags = DIO_LOCKING | DIO_SKIP_HOLES; + wakeup = false; } - free_extent_state(cached_state); - cached_state = NULL; - ret = __blockdev_direct_IO(rw, iocb, inode, - BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev, - iov, offset, nr_segs, btrfs_get_blocks_direct, NULL, - btrfs_submit_direct, 0); - - if (ret < 0 && ret != -EIOCBQUEUED) { - clear_extent_bit(&BTRFS_I(inode)->io_tree, offset, - offset + iov_length(iov, nr_segs) - 1, - EXTENT_LOCKED | write_bits, 1, 0, - &cached_state, GFP_NOFS); - } else if (ret >= 0 && ret < iov_length(iov, nr_segs)) { - /* - * We're falling back to buffered, unlock the section we didn't - * do IO on. - */ - clear_extent_bit(&BTRFS_I(inode)->io_tree, offset + ret, - offset + iov_length(iov, nr_segs) - 1, - EXTENT_LOCKED | write_bits, 1, 0, - &cached_state, GFP_NOFS); + BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev, + iter, offset, btrfs_get_blocks_direct, NULL, + btrfs_submit_direct, flags); + if (rw & WRITE) { + if (ret < 0 && ret != -EIOCBQUEUED) + btrfs_delalloc_release_space(inode, count); + else if (ret >= 0 && (size_t)ret < count) + btrfs_delalloc_release_space(inode, + count - (size_t)ret); + else + btrfs_delalloc_release_metadata(inode, 0); } out: - free_extent_state(cached_state); + if (wakeup) + inode_dio_done(inode); + if (relock) + mutex_lock(&inode->i_mutex); + return ret; } +#define BTRFS_FIEMAP_FLAGS (FIEMAP_FLAG_SYNC) + static int btrfs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, __u64 start, __u64 len) { - return extent_fiemap(inode, fieinfo, start, len, btrfs_get_extent); + int ret; + + ret = fiemap_check_flags(fieinfo, BTRFS_FIEMAP_FLAGS); + if (ret) + return ret; + + return extent_fiemap(inode, fieinfo, start, len, btrfs_get_extent_fiemap); } int btrfs_readpage(struct file *file, struct page *page) { struct extent_io_tree *tree; tree = &BTRFS_I(page->mapping->host)->io_tree; - return extent_read_full_page(tree, page, btrfs_get_extent); + return extent_read_full_page(tree, page, btrfs_get_extent, 0); } static int btrfs_writepage(struct page *page, struct writeback_control *wbc) @@ -5917,8 +7601,8 @@ static int btrfs_writepage(struct page *page, struct writeback_control *wbc) return extent_write_full_page(tree, page, btrfs_get_extent, wbc); } -int btrfs_writepages(struct address_space *mapping, - struct writeback_control *wbc) +static int btrfs_writepages(struct address_space *mapping, + struct writeback_control *wbc) { struct extent_io_tree *tree; @@ -5959,14 +7643,16 @@ static int btrfs_releasepage(struct page *page, gfp_t gfp_flags) return __btrfs_releasepage(page, gfp_flags & GFP_NOFS); } -static void btrfs_invalidatepage(struct page *page, unsigned long offset) +static void btrfs_invalidatepage(struct page *page, unsigned int offset, + unsigned int length) { + struct inode *inode = page->mapping->host; struct extent_io_tree *tree; struct btrfs_ordered_extent *ordered; struct extent_state *cached_state = NULL; u64 page_start = page_offset(page); u64 page_end = page_start + PAGE_CACHE_SIZE - 1; - + int inode_evicting = inode->i_state & I_FREEING; /* * we have the page locked, so new writeback can't start, @@ -5977,41 +7663,65 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) */ wait_on_page_writeback(page); - tree = &BTRFS_I(page->mapping->host)->io_tree; + tree = &BTRFS_I(inode)->io_tree; if (offset) { btrfs_releasepage(page, GFP_NOFS); return; } - lock_extent_bits(tree, page_start, page_end, 0, &cached_state, - GFP_NOFS); - ordered = btrfs_lookup_ordered_extent(page->mapping->host, - page_offset(page)); + + if (!inode_evicting) + lock_extent_bits(tree, page_start, page_end, 0, &cached_state); + ordered = btrfs_lookup_ordered_extent(inode, page_start); if (ordered) { /* * IO on this page will never be started, so we need * to account for any ordered extents now */ - clear_extent_bit(tree, page_start, page_end, - EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_LOCKED | EXTENT_DO_ACCOUNTING, 1, 0, - &cached_state, GFP_NOFS); + if (!inode_evicting) + clear_extent_bit(tree, page_start, page_end, + EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_LOCKED | EXTENT_DO_ACCOUNTING | + EXTENT_DEFRAG, 1, 0, &cached_state, + GFP_NOFS); /* * whoever cleared the private bit is responsible * for the finish_ordered_io */ if (TestClearPagePrivate2(page)) { - btrfs_finish_ordered_io(page->mapping->host, - page_start, page_end); + struct btrfs_ordered_inode_tree *tree; + u64 new_len; + + tree = &BTRFS_I(inode)->ordered_tree; + + spin_lock_irq(&tree->lock); + set_bit(BTRFS_ORDERED_TRUNCATED, &ordered->flags); + new_len = page_start - ordered->file_offset; + if (new_len < ordered->truncated_len) + ordered->truncated_len = new_len; + spin_unlock_irq(&tree->lock); + + if (btrfs_dec_test_ordered_pending(inode, &ordered, + page_start, + PAGE_CACHE_SIZE, 1)) + btrfs_finish_ordered_io(ordered); } btrfs_put_ordered_extent(ordered); - cached_state = NULL; - lock_extent_bits(tree, page_start, page_end, 0, &cached_state, - GFP_NOFS); + if (!inode_evicting) { + cached_state = NULL; + lock_extent_bits(tree, page_start, page_end, 0, + &cached_state); + } + } + + if (!inode_evicting) { + clear_extent_bit(tree, page_start, page_end, + EXTENT_LOCKED | EXTENT_DIRTY | + EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | + EXTENT_DEFRAG, 1, 1, + &cached_state, GFP_NOFS); + + __btrfs_releasepage(page, GFP_NOFS); } - clear_extent_bit(tree, page_start, page_end, - EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_DO_ACCOUNTING, 1, 1, &cached_state, GFP_NOFS); - __btrfs_releasepage(page, GFP_NOFS); ClearPageChecked(page); if (PagePrivate(page)) { @@ -6039,7 +7749,7 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) { struct page *page = vmf->page; - struct inode *inode = fdentry(vma->vm_file)->d_inode; + struct inode *inode = file_inode(vma->vm_file); struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_ordered_extent *ordered; @@ -6048,16 +7758,24 @@ int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) unsigned long zero_start; loff_t size; int ret; + int reserved = 0; u64 page_start; u64 page_end; + sb_start_pagefault(inode->i_sb); ret = btrfs_delalloc_reserve_space(inode, PAGE_CACHE_SIZE); + if (!ret) { + ret = file_update_time(vma->vm_file); + reserved = 1; + } if (ret) { if (ret == -ENOMEM) ret = VM_FAULT_OOM; else /* -ENOSPC, -EIO, etc */ ret = VM_FAULT_SIGBUS; - goto out; + if (reserved) + goto out; + goto out_noreserve; } ret = VM_FAULT_NOPAGE; /* make the VM retry the fault */ @@ -6074,8 +7792,7 @@ again: } wait_on_page_writeback(page); - lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state, - GFP_NOFS); + lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state); set_page_extent_mapped(page); /* @@ -6100,7 +7817,8 @@ again: * prepare_pages in the normal write path. */ clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end, - EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, + EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0, &cached_state, GFP_NOFS); ret = btrfs_set_extent_delalloc(inode, page_start, page_end, @@ -6131,42 +7849,94 @@ again: BTRFS_I(inode)->last_trans = root->fs_info->generation; BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid; + BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->root->last_log_commit; unlock_extent_cached(io_tree, page_start, page_end, &cached_state, GFP_NOFS); out_unlock: - if (!ret) + if (!ret) { + sb_end_pagefault(inode->i_sb); return VM_FAULT_LOCKED; + } unlock_page(page); - btrfs_delalloc_release_space(inode, PAGE_CACHE_SIZE); out: + btrfs_delalloc_release_space(inode, PAGE_CACHE_SIZE); +out_noreserve: + sb_end_pagefault(inode->i_sb); return ret; } -static void btrfs_truncate(struct inode *inode) +static int btrfs_truncate(struct inode *inode) { struct btrfs_root *root = BTRFS_I(inode)->root; - int ret; + struct btrfs_block_rsv *rsv; + int ret = 0; + int err = 0; struct btrfs_trans_handle *trans; - unsigned long nr; u64 mask = root->sectorsize - 1; + u64 min_size = btrfs_calc_trunc_metadata_size(root, 1); - if (!S_ISREG(inode->i_mode)) { - WARN_ON(1); - return; - } - - ret = btrfs_truncate_page(inode->i_mapping, inode->i_size); + ret = btrfs_wait_ordered_range(inode, inode->i_size & (~mask), + (u64)-1); if (ret) - return; + return ret; + + /* + * Yes ladies and gentelment, this is indeed ugly. The fact is we have + * 3 things going on here + * + * 1) We need to reserve space for our orphan item and the space to + * delete our orphan item. Lord knows we don't want to have a dangling + * orphan item because we didn't reserve space to remove it. + * + * 2) We need to reserve space to update our inode. + * + * 3) We need to have something to cache all the space that is going to + * be free'd up by the truncate operation, but also have some slack + * space reserved in case it uses space during the truncate (thank you + * very much snapshotting). + * + * And we need these to all be seperate. The fact is we can use alot of + * space doing the truncate, and we have no earthly idea how much space + * we will use, so we need the truncate reservation to be seperate so it + * doesn't end up using space reserved for updating the inode or + * removing the orphan item. We also need to be able to stop the + * transaction and start a new one, which means we need to be able to + * update the inode several times, and we have no idea of knowing how + * many times that will be, so we can't just reserve 1 item for the + * entirety of the opration, so that has to be done seperately as well. + * Then there is the orphan item, which does indeed need to be held on + * to for the whole operation, and we need nobody to touch this reserved + * space except the orphan code. + * + * So that leaves us with + * + * 1) root->orphan_block_rsv - for the orphan deletion. + * 2) rsv - for the truncate reservation, which we will steal from the + * transaction reservation. + * 3) fs_info->trans_block_rsv - this will have 1 items worth left for + * updating the inode. + */ + rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP); + if (!rsv) + return -ENOMEM; + rsv->size = min_size; + rsv->failfast = 1; - btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1); - btrfs_ordered_update_i_size(inode, inode->i_size, NULL); + /* + * 1 for the truncate slack space + * 1 for updating the inode. + */ + trans = btrfs_start_transaction(root, 2); + if (IS_ERR(trans)) { + err = PTR_ERR(trans); + goto out; + } - trans = btrfs_start_transaction(root, 0); - BUG_ON(IS_ERR(trans)); - btrfs_set_trans_block_group(trans, inode); - trans->block_rsv = root->orphan_block_rsv; + /* Migrate the slack space for the truncate to our reserve */ + ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, rsv, + min_size); + BUG_ON(ret); /* * setattr is responsible for setting the ordered_data_close flag, @@ -6185,54 +7955,76 @@ static void btrfs_truncate(struct inode *inode) * using truncate to replace the contents of the file will * end up with a zero length file after a crash. */ - if (inode->i_size == 0 && BTRFS_I(inode)->ordered_data_close) + if (inode->i_size == 0 && test_bit(BTRFS_INODE_ORDERED_DATA_CLOSE, + &BTRFS_I(inode)->runtime_flags)) btrfs_add_ordered_operation(trans, root, inode); - while (1) { - if (!trans) { - trans = btrfs_start_transaction(root, 0); - BUG_ON(IS_ERR(trans)); - btrfs_set_trans_block_group(trans, inode); - trans->block_rsv = root->orphan_block_rsv; - } - - ret = btrfs_block_rsv_check(trans, root, - root->orphan_block_rsv, 0, 5); - if (ret) { - BUG_ON(ret != -EAGAIN); - ret = btrfs_commit_transaction(trans, root); - BUG_ON(ret); - trans = NULL; - continue; - } + /* + * So if we truncate and then write and fsync we normally would just + * write the extents that changed, which is a problem if we need to + * first truncate that entire inode. So set this flag so we write out + * all of the extents in the inode to the sync log so we're completely + * safe. + */ + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags); + trans->block_rsv = rsv; + while (1) { ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, BTRFS_EXTENT_DATA_KEY); - if (ret != -EAGAIN) + if (ret != -ENOSPC) { + err = ret; break; + } + trans->block_rsv = &root->fs_info->trans_block_rsv; ret = btrfs_update_inode(trans, root, inode); - BUG_ON(ret); + if (ret) { + err = ret; + break; + } - nr = trans->blocks_used; btrfs_end_transaction(trans, root); - trans = NULL; - btrfs_btree_balance_dirty(root, nr); + btrfs_btree_balance_dirty(root); + + trans = btrfs_start_transaction(root, 2); + if (IS_ERR(trans)) { + ret = err = PTR_ERR(trans); + trans = NULL; + break; + } + + ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, + rsv, min_size); + BUG_ON(ret); /* shouldn't happen */ + trans->block_rsv = rsv; } if (ret == 0 && inode->i_nlink > 0) { + trans->block_rsv = root->orphan_block_rsv; ret = btrfs_orphan_del(trans, inode); - BUG_ON(ret); + if (ret) + err = ret; } - ret = btrfs_update_inode(trans, root, inode); - BUG_ON(ret); + if (trans) { + trans->block_rsv = &root->fs_info->trans_block_rsv; + ret = btrfs_update_inode(trans, root, inode); + if (ret && !err) + err = ret; - nr = trans->blocks_used; - ret = btrfs_end_transaction_throttle(trans, root); - BUG_ON(ret); - btrfs_btree_balance_dirty(root, nr); + ret = btrfs_end_transaction(trans, root); + btrfs_btree_balance_dirty(root); + } + +out: + btrfs_free_block_rsv(root, rsv); + + if (ret && !err) + err = ret; + + return err; } /* @@ -6240,40 +8032,35 @@ static void btrfs_truncate(struct inode *inode) */ int btrfs_create_subvol_root(struct btrfs_trans_handle *trans, struct btrfs_root *new_root, - u64 new_dirid, u64 alloc_hint) + struct btrfs_root *parent_root, + u64 new_dirid) { struct inode *inode; int err; u64 index = 0; - inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, new_dirid, - new_dirid, alloc_hint, S_IFDIR | 0700, &index); + inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, + new_dirid, new_dirid, + S_IFDIR | (~current_umask() & S_IRWXUGO), + &index); if (IS_ERR(inode)) return PTR_ERR(inode); inode->i_op = &btrfs_dir_inode_operations; inode->i_fop = &btrfs_dir_file_operations; - inode->i_nlink = 1; + set_nlink(inode, 1); btrfs_i_size_write(inode, 0); + err = btrfs_subvol_inherit_props(trans, new_root, parent_root); + if (err) + btrfs_err(new_root->fs_info, + "error inheriting subvolume %llu properties: %d", + new_root->root_key.objectid, err); + err = btrfs_update_inode(trans, new_root, inode); - BUG_ON(err); iput(inode); - return 0; -} - -/* helper function for file defrag and space balancing. This - * forces readahead on a given range of bytes in an inode - */ -unsigned long btrfs_force_ra(struct address_space *mapping, - struct file_ra_state *ra, struct file *file, - pgoff_t offset, pgoff_t last_index) -{ - pgoff_t req_size = last_index - offset + 1; - - page_cache_sync_readahead(mapping, ra, file, offset, req_size); - return offset + req_size; + return err; } struct inode *btrfs_alloc_inode(struct super_block *sb) @@ -6286,35 +8073,38 @@ struct inode *btrfs_alloc_inode(struct super_block *sb) return NULL; ei->root = NULL; - ei->space_info = NULL; ei->generation = 0; - ei->sequence = 0; ei->last_trans = 0; ei->last_sub_trans = 0; ei->logged_trans = 0; ei->delalloc_bytes = 0; - ei->reserved_bytes = 0; ei->disk_i_size = 0; ei->flags = 0; + ei->csum_bytes = 0; ei->index_cnt = (u64)-1; + ei->dir_index = 0; ei->last_unlink_trans = 0; + ei->last_log_commit = 0; - spin_lock_init(&ei->accounting_lock); - atomic_set(&ei->outstanding_extents, 0); + spin_lock_init(&ei->lock); + ei->outstanding_extents = 0; ei->reserved_extents = 0; - ei->ordered_data_close = 0; - ei->orphan_meta_reserved = 0; - ei->dummy_inode = 0; - ei->force_compress = 0; + ei->runtime_flags = 0; + ei->force_compress = BTRFS_COMPRESS_NONE; + + ei->delayed_node = NULL; inode = &ei->vfs_inode; - extent_map_tree_init(&ei->extent_tree, GFP_NOFS); - extent_io_tree_init(&ei->io_tree, &inode->i_data, GFP_NOFS); - extent_io_tree_init(&ei->io_failure_tree, &inode->i_data, GFP_NOFS); + extent_map_tree_init(&ei->extent_tree); + extent_io_tree_init(&ei->io_tree, &inode->i_data); + extent_io_tree_init(&ei->io_failure_tree, &inode->i_data); + ei->io_tree.track_uptodate = 1; + ei->io_failure_tree.track_uptodate = 1; + atomic_set(&ei->sync_writers, 0); mutex_init(&ei->log_mutex); + mutex_init(&ei->delalloc_mutex); btrfs_ordered_inode_tree_init(&ei->ordered_tree); - INIT_LIST_HEAD(&ei->i_orphan); INIT_LIST_HEAD(&ei->delalloc_inodes); INIT_LIST_HEAD(&ei->ordered_operations); RB_CLEAR_NODE(&ei->rb_node); @@ -6322,15 +8112,31 @@ struct inode *btrfs_alloc_inode(struct super_block *sb) return inode; } +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS +void btrfs_test_destroy_inode(struct inode *inode) +{ + btrfs_drop_extent_cache(inode, 0, (u64)-1, 0); + kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode)); +} +#endif + +static void btrfs_i_callback(struct rcu_head *head) +{ + struct inode *inode = container_of(head, struct inode, i_rcu); + kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode)); +} + void btrfs_destroy_inode(struct inode *inode) { struct btrfs_ordered_extent *ordered; struct btrfs_root *root = BTRFS_I(inode)->root; - WARN_ON(!list_empty(&inode->i_dentry)); + WARN_ON(!hlist_empty(&inode->i_dentry)); WARN_ON(inode->i_data.nrpages); - WARN_ON(atomic_read(&BTRFS_I(inode)->outstanding_extents)); + WARN_ON(BTRFS_I(inode)->outstanding_extents); WARN_ON(BTRFS_I(inode)->reserved_extents); + WARN_ON(BTRFS_I(inode)->delalloc_bytes); + WARN_ON(BTRFS_I(inode)->csum_bytes); /* * This can happen where we create an inode, but somebody else also @@ -6346,43 +8152,25 @@ void btrfs_destroy_inode(struct inode *inode) */ smp_mb(); if (!list_empty(&BTRFS_I(inode)->ordered_operations)) { - spin_lock(&root->fs_info->ordered_extent_lock); + spin_lock(&root->fs_info->ordered_root_lock); list_del_init(&BTRFS_I(inode)->ordered_operations); - spin_unlock(&root->fs_info->ordered_extent_lock); - } - - if (root == root->fs_info->tree_root) { - struct btrfs_block_group_cache *block_group; - - block_group = btrfs_lookup_block_group(root->fs_info, - BTRFS_I(inode)->block_group); - if (block_group && block_group->inode == inode) { - spin_lock(&block_group->lock); - block_group->inode = NULL; - spin_unlock(&block_group->lock); - btrfs_put_block_group(block_group); - } else if (block_group) { - btrfs_put_block_group(block_group); - } + spin_unlock(&root->fs_info->ordered_root_lock); } - spin_lock(&root->orphan_lock); - if (!list_empty(&BTRFS_I(inode)->i_orphan)) { - printk(KERN_INFO "BTRFS: inode %lu still on the orphan list\n", - inode->i_ino); - list_del_init(&BTRFS_I(inode)->i_orphan); + if (test_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags)) { + btrfs_info(root->fs_info, "inode %llu still on the orphan list", + btrfs_ino(inode)); + atomic_dec(&root->orphan_inodes); } - spin_unlock(&root->orphan_lock); while (1) { ordered = btrfs_lookup_first_ordered_extent(inode, (u64)-1); if (!ordered) break; else { - printk(KERN_ERR "btrfs found ordered " - "extent %llu %llu on inode cleanup\n", - (unsigned long long)ordered->file_offset, - (unsigned long long)ordered->len); + btrfs_err(root->fs_info, "found ordered extent %llu %llu on inode cleanup", + ordered->file_offset, ordered->len); btrfs_remove_ordered_extent(inode, ordered); btrfs_put_ordered_extent(ordered); btrfs_put_ordered_extent(ordered); @@ -6391,15 +8179,18 @@ void btrfs_destroy_inode(struct inode *inode) inode_tree_del(inode); btrfs_drop_extent_cache(inode, 0, (u64)-1, 0); free: - kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode)); + call_rcu(&inode->i_rcu, btrfs_i_callback); } int btrfs_drop_inode(struct inode *inode) { struct btrfs_root *root = BTRFS_I(inode)->root; - if (btrfs_root_refs(&root->root_item) == 0 && - root != root->fs_info->tree_root) + if (root == NULL) + return 1; + + /* the snap/subvol tree is on deleting */ + if (btrfs_root_refs(&root->root_item) == 0) return 1; else return generic_drop_inode(inode); @@ -6414,6 +8205,11 @@ static void init_once(void *foo) void btrfs_destroy_cachep(void) { + /* + * Make sure all delayed rcu free inodes are flushed before we + * destroy cache. + */ + rcu_barrier(); if (btrfs_inode_cachep) kmem_cache_destroy(btrfs_inode_cachep); if (btrfs_trans_handle_cachep) @@ -6422,34 +8218,51 @@ void btrfs_destroy_cachep(void) kmem_cache_destroy(btrfs_transaction_cachep); if (btrfs_path_cachep) kmem_cache_destroy(btrfs_path_cachep); + if (btrfs_free_space_cachep) + kmem_cache_destroy(btrfs_free_space_cachep); + if (btrfs_delalloc_work_cachep) + kmem_cache_destroy(btrfs_delalloc_work_cachep); } int btrfs_init_cachep(void) { - btrfs_inode_cachep = kmem_cache_create("btrfs_inode_cache", + btrfs_inode_cachep = kmem_cache_create("btrfs_inode", sizeof(struct btrfs_inode), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, init_once); if (!btrfs_inode_cachep) goto fail; - btrfs_trans_handle_cachep = kmem_cache_create("btrfs_trans_handle_cache", + btrfs_trans_handle_cachep = kmem_cache_create("btrfs_trans_handle", sizeof(struct btrfs_trans_handle), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); if (!btrfs_trans_handle_cachep) goto fail; - btrfs_transaction_cachep = kmem_cache_create("btrfs_transaction_cache", + btrfs_transaction_cachep = kmem_cache_create("btrfs_transaction", sizeof(struct btrfs_transaction), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); if (!btrfs_transaction_cachep) goto fail; - btrfs_path_cachep = kmem_cache_create("btrfs_path_cache", + btrfs_path_cachep = kmem_cache_create("btrfs_path", sizeof(struct btrfs_path), 0, SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); if (!btrfs_path_cachep) goto fail; + btrfs_free_space_cachep = kmem_cache_create("btrfs_free_space", + sizeof(struct btrfs_free_space), 0, + SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); + if (!btrfs_free_space_cachep) + goto fail; + + btrfs_delalloc_work_cachep = kmem_cache_create("btrfs_delalloc_work", + sizeof(struct btrfs_delalloc_work), 0, + SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, + NULL); + if (!btrfs_delalloc_work_cachep) + goto fail; + return 0; fail: btrfs_destroy_cachep(); @@ -6459,12 +8272,19 @@ fail: static int btrfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat) { + u64 delalloc_bytes; struct inode *inode = dentry->d_inode; + u32 blocksize = inode->i_sb->s_blocksize; + generic_fillattr(inode, stat); - stat->dev = BTRFS_I(inode)->root->anon_super.s_dev; + stat->dev = BTRFS_I(inode)->root->anon_dev; stat->blksize = PAGE_CACHE_SIZE; - stat->blocks = (inode_get_bytes(inode) + - BTRFS_I(inode)->delalloc_bytes) >> 9; + + spin_lock(&BTRFS_I(inode)->lock); + delalloc_bytes = BTRFS_I(inode)->delalloc_bytes; + spin_unlock(&BTRFS_I(inode)->lock); + stat->blocks = (ALIGN(inode_get_bytes(inode), blocksize) + + ALIGN(delalloc_bytes, blocksize)) >> 9; return 0; } @@ -6480,21 +8300,43 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, u64 index = 0; u64 root_objectid; int ret; + u64 old_ino = btrfs_ino(old_inode); - if (new_dir->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID) + if (btrfs_ino(new_dir) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID) return -EPERM; /* we only allow rename subvolume link between subvolumes */ - if (old_inode->i_ino != BTRFS_FIRST_FREE_OBJECTID && root != dest) + if (old_ino != BTRFS_FIRST_FREE_OBJECTID && root != dest) return -EXDEV; - if (old_inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID || - (new_inode && new_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) + if (old_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID || + (new_inode && btrfs_ino(new_inode) == BTRFS_FIRST_FREE_OBJECTID)) return -ENOTEMPTY; if (S_ISDIR(old_inode->i_mode) && new_inode && new_inode->i_size > BTRFS_EMPTY_DIR_SIZE) return -ENOTEMPTY; + + + /* check for collisions, even if the name isn't there */ + ret = btrfs_check_dir_item_collision(dest, new_dir->i_ino, + new_dentry->d_name.name, + new_dentry->d_name.len); + + if (ret) { + if (ret == -EEXIST) { + /* we shouldn't get + * eexist without a new_inode */ + if (WARN_ON(!new_inode)) { + return ret; + } + } else { + /* maybe -EOVERFLOW */ + return ret; + } + } + ret = 0; + /* * we're using rename to replace one file with another. * and the replacement file is large. Start IO on it now so @@ -6505,7 +8347,7 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, filemap_flush(old_inode->i_mapping); /* close the racy window with snapshot create/destroy ioctl */ - if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) + if (old_ino == BTRFS_FIRST_FREE_OBJECTID) down_read(&root->fs_info->subvol_sem); /* * We want to reserve the absolute worst case amount of items. So if @@ -6515,11 +8357,11 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, * inodes. So 5 * 2 is 10, plus 1 for the new link, so 11 total items * should cover the worst case number of items we'll modify. */ - trans = btrfs_start_transaction(root, 20); - if (IS_ERR(trans)) - return PTR_ERR(trans); - - btrfs_set_trans_block_group(trans, new_dir); + trans = btrfs_start_transaction(root, 11); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + goto out_notrans; + } if (dest != root) btrfs_record_root_in_trans(trans, dest); @@ -6528,15 +8370,16 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, if (ret) goto out_fail; - if (unlikely(old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) { + BTRFS_I(old_inode)->dir_index = 0ULL; + if (unlikely(old_ino == BTRFS_FIRST_FREE_OBJECTID)) { /* force full log commit if subvolume involved. */ - root->fs_info->last_trans_log_full_commit = trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); } else { ret = btrfs_insert_inode_ref(trans, dest, new_dentry->d_name.name, new_dentry->d_name.len, - old_inode->i_ino, - new_dir->i_ino, index); + old_ino, + btrfs_ino(new_dir), index); if (ret) goto out_fail; /* @@ -6552,11 +8395,12 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, * make sure the inode gets flushed if it is replacing * something. */ - if (new_inode && new_inode->i_size && - old_inode && S_ISREG(old_inode->i_mode)) { + if (new_inode && new_inode->i_size && S_ISREG(old_inode->i_mode)) btrfs_add_ordered_operation(trans, root, old_inode); - } + inode_inc_iversion(old_dir); + inode_inc_iversion(new_dir); + inode_inc_iversion(old_inode); old_dir->i_ctime = old_dir->i_mtime = ctime; new_dir->i_ctime = new_dir->i_mtime = ctime; old_inode->i_ctime = ctime; @@ -6564,23 +8408,28 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, if (old_dentry->d_parent != new_dentry->d_parent) btrfs_record_unlink_dir(trans, old_dir, old_inode, 1); - if (unlikely(old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)) { + if (unlikely(old_ino == BTRFS_FIRST_FREE_OBJECTID)) { root_objectid = BTRFS_I(old_inode)->root->root_key.objectid; ret = btrfs_unlink_subvol(trans, root, old_dir, root_objectid, old_dentry->d_name.name, old_dentry->d_name.len); } else { - btrfs_inc_nlink(old_dentry->d_inode); - ret = btrfs_unlink_inode(trans, root, old_dir, - old_dentry->d_inode, - old_dentry->d_name.name, - old_dentry->d_name.len); + ret = __btrfs_unlink_inode(trans, root, old_dir, + old_dentry->d_inode, + old_dentry->d_name.name, + old_dentry->d_name.len); + if (!ret) + ret = btrfs_update_inode(trans, root, old_inode); + } + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto out_fail; } - BUG_ON(ret); if (new_inode) { + inode_inc_iversion(new_inode); new_inode->i_ctime = CURRENT_TIME; - if (unlikely(new_inode->i_ino == + if (unlikely(btrfs_ino(new_inode) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) { root_objectid = BTRFS_I(new_inode)->location.objectid; ret = btrfs_unlink_subvol(trans, dest, new_dir, @@ -6594,66 +8443,168 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, new_dentry->d_name.name, new_dentry->d_name.len); } - BUG_ON(ret); - if (new_inode->i_nlink == 0) { + if (!ret && new_inode->i_nlink == 0) ret = btrfs_orphan_add(trans, new_dentry->d_inode); - BUG_ON(ret); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto out_fail; } } ret = btrfs_add_link(trans, new_dir, old_inode, new_dentry->d_name.name, new_dentry->d_name.len, 0, index); - BUG_ON(ret); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto out_fail; + } + + if (old_inode->i_nlink == 1) + BTRFS_I(old_inode)->dir_index = index; - if (old_inode->i_ino != BTRFS_FIRST_FREE_OBJECTID) { - btrfs_log_new_name(trans, old_inode, old_dir, - new_dentry->d_parent); + if (old_ino != BTRFS_FIRST_FREE_OBJECTID) { + struct dentry *parent = new_dentry->d_parent; + btrfs_log_new_name(trans, old_inode, old_dir, parent); btrfs_end_log_trans(root); } out_fail: - btrfs_end_transaction_throttle(trans, root); - - if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) + btrfs_end_transaction(trans, root); +out_notrans: + if (old_ino == BTRFS_FIRST_FREE_OBJECTID) up_read(&root->fs_info->subvol_sem); return ret; } +static void btrfs_run_delalloc_work(struct btrfs_work *work) +{ + struct btrfs_delalloc_work *delalloc_work; + struct inode *inode; + + delalloc_work = container_of(work, struct btrfs_delalloc_work, + work); + inode = delalloc_work->inode; + if (delalloc_work->wait) { + btrfs_wait_ordered_range(inode, 0, (u64)-1); + } else { + filemap_flush(inode->i_mapping); + if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, + &BTRFS_I(inode)->runtime_flags)) + filemap_flush(inode->i_mapping); + } + + if (delalloc_work->delay_iput) + btrfs_add_delayed_iput(inode); + else + iput(inode); + complete(&delalloc_work->completion); +} + +struct btrfs_delalloc_work *btrfs_alloc_delalloc_work(struct inode *inode, + int wait, int delay_iput) +{ + struct btrfs_delalloc_work *work; + + work = kmem_cache_zalloc(btrfs_delalloc_work_cachep, GFP_NOFS); + if (!work) + return NULL; + + init_completion(&work->completion); + INIT_LIST_HEAD(&work->list); + work->inode = inode; + work->wait = wait; + work->delay_iput = delay_iput; + btrfs_init_work(&work->work, btrfs_run_delalloc_work, NULL, NULL); + + return work; +} + +void btrfs_wait_and_free_delalloc_work(struct btrfs_delalloc_work *work) +{ + wait_for_completion(&work->completion); + kmem_cache_free(btrfs_delalloc_work_cachep, work); +} + /* * some fairly slow code that needs optimization. This walks the list * of all the inodes with pending delalloc and forces them to disk. */ -int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput) +static int __start_delalloc_inodes(struct btrfs_root *root, int delay_iput, + int nr) { - struct list_head *head = &root->fs_info->delalloc_inodes; struct btrfs_inode *binode; struct inode *inode; + struct btrfs_delalloc_work *work, *next; + struct list_head works; + struct list_head splice; + int ret = 0; - if (root->fs_info->sb->s_flags & MS_RDONLY) - return -EROFS; + INIT_LIST_HEAD(&works); + INIT_LIST_HEAD(&splice); - spin_lock(&root->fs_info->delalloc_lock); - while (!list_empty(head)) { - binode = list_entry(head->next, struct btrfs_inode, + mutex_lock(&root->delalloc_mutex); + spin_lock(&root->delalloc_lock); + list_splice_init(&root->delalloc_inodes, &splice); + while (!list_empty(&splice)) { + binode = list_entry(splice.next, struct btrfs_inode, delalloc_inodes); + + list_move_tail(&binode->delalloc_inodes, + &root->delalloc_inodes); inode = igrab(&binode->vfs_inode); - if (!inode) - list_del_init(&binode->delalloc_inodes); - spin_unlock(&root->fs_info->delalloc_lock); - if (inode) { - filemap_flush(inode->i_mapping); + if (!inode) { + cond_resched_lock(&root->delalloc_lock); + continue; + } + spin_unlock(&root->delalloc_lock); + + work = btrfs_alloc_delalloc_work(inode, 0, delay_iput); + if (unlikely(!work)) { if (delay_iput) btrfs_add_delayed_iput(inode); else iput(inode); + ret = -ENOMEM; + goto out; } + list_add_tail(&work->list, &works); + btrfs_queue_work(root->fs_info->flush_workers, + &work->work); + ret++; + if (nr != -1 && ret >= nr) + goto out; cond_resched(); - spin_lock(&root->fs_info->delalloc_lock); + spin_lock(&root->delalloc_lock); } - spin_unlock(&root->fs_info->delalloc_lock); + spin_unlock(&root->delalloc_lock); - /* the filemap_flush will queue IO into the worker threads, but +out: + list_for_each_entry_safe(work, next, &works, list) { + list_del_init(&work->list); + btrfs_wait_and_free_delalloc_work(work); + } + + if (!list_empty_careful(&splice)) { + spin_lock(&root->delalloc_lock); + list_splice_tail(&splice, &root->delalloc_inodes); + spin_unlock(&root->delalloc_lock); + } + mutex_unlock(&root->delalloc_mutex); + return ret; +} + +int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput) +{ + int ret; + + if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) + return -EROFS; + + ret = __start_delalloc_inodes(root, delay_iput, -1); + if (ret > 0) + ret = 0; + /* + * the filemap_flush will queue IO into the worker threads, but * we have to make sure the IO is actually started and that * ordered extents get created before we return */ @@ -6665,59 +8616,63 @@ int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput) atomic_read(&root->fs_info->async_delalloc_pages) == 0)); } atomic_dec(&root->fs_info->async_submit_draining); - return 0; + return ret; } -int btrfs_start_one_delalloc_inode(struct btrfs_root *root, int delay_iput, - int sync) +int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, int delay_iput, + int nr) { - struct btrfs_inode *binode; - struct inode *inode = NULL; + struct btrfs_root *root; + struct list_head splice; + int ret; - spin_lock(&root->fs_info->delalloc_lock); - while (!list_empty(&root->fs_info->delalloc_inodes)) { - binode = list_entry(root->fs_info->delalloc_inodes.next, - struct btrfs_inode, delalloc_inodes); - inode = igrab(&binode->vfs_inode); - if (inode) { - list_move_tail(&binode->delalloc_inodes, - &root->fs_info->delalloc_inodes); - break; - } + if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) + return -EROFS; - list_del_init(&binode->delalloc_inodes); - cond_resched_lock(&root->fs_info->delalloc_lock); - } - spin_unlock(&root->fs_info->delalloc_lock); + INIT_LIST_HEAD(&splice); + + mutex_lock(&fs_info->delalloc_root_mutex); + spin_lock(&fs_info->delalloc_root_lock); + list_splice_init(&fs_info->delalloc_roots, &splice); + while (!list_empty(&splice) && nr) { + root = list_first_entry(&splice, struct btrfs_root, + delalloc_root); + root = btrfs_grab_fs_root(root); + BUG_ON(!root); + list_move_tail(&root->delalloc_root, + &fs_info->delalloc_roots); + spin_unlock(&fs_info->delalloc_root_lock); + + ret = __start_delalloc_inodes(root, delay_iput, nr); + btrfs_put_fs_root(root); + if (ret < 0) + goto out; - if (inode) { - if (sync) { - filemap_write_and_wait(inode->i_mapping); - /* - * We have to do this because compression doesn't - * actually set PG_writeback until it submits the pages - * for IO, which happens in an async thread, so we could - * race and not actually wait for any writeback pages - * because they've not been submitted yet. Technically - * this could still be the case for the ordered stuff - * since the async thread may not have started to do its - * work yet. If this becomes the case then we need to - * figure out a way to make sure that in writepage we - * wait for any async pages to be submitted before - * returning so that fdatawait does what its supposed to - * do. - */ - btrfs_wait_ordered_range(inode, 0, (u64)-1); - } else { - filemap_flush(inode->i_mapping); + if (nr != -1) { + nr -= ret; + WARN_ON(nr < 0); } - if (delay_iput) - btrfs_add_delayed_iput(inode); - else - iput(inode); - return 1; + spin_lock(&fs_info->delalloc_root_lock); } - return 0; + spin_unlock(&fs_info->delalloc_root_lock); + + ret = 0; + atomic_inc(&fs_info->async_submit_draining); + while (atomic_read(&fs_info->nr_async_submits) || + atomic_read(&fs_info->async_delalloc_pages)) { + wait_event(fs_info->async_submit_wait, + (atomic_read(&fs_info->nr_async_submits) == 0 && + atomic_read(&fs_info->async_delalloc_pages) == 0)); + } + atomic_dec(&fs_info->async_submit_draining); +out: + if (!list_empty_careful(&splice)) { + spin_lock(&fs_info->delalloc_root_lock); + list_splice_tail(&splice, &fs_info->delalloc_roots); + spin_unlock(&fs_info->delalloc_root_lock); + } + mutex_unlock(&fs_info->delalloc_root_mutex); + return ret; } static int btrfs_symlink(struct inode *dir, struct dentry *dentry, @@ -6731,21 +8686,17 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, int err; int drop_inode = 0; u64 objectid; - u64 index = 0 ; + u64 index = 0; int name_len; int datasize; unsigned long ptr; struct btrfs_file_extent_item *ei; struct extent_buffer *leaf; - unsigned long nr = 0; - name_len = strlen(symname) + 1; + name_len = strlen(symname); if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root)) return -ENAMETOOLONG; - err = btrfs_find_free_objectid(NULL, root, dir->i_ino, &objectid); - if (err) - return err; /* * 2 items for inode item and ref * 2 items for dir items @@ -6755,42 +8706,51 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, if (IS_ERR(trans)) return PTR_ERR(trans); - btrfs_set_trans_block_group(trans, dir); + err = btrfs_find_free_ino(root, &objectid); + if (err) + goto out_unlock; inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name, - dentry->d_name.len, - dentry->d_parent->d_inode->i_ino, objectid, - BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO, - &index); - err = PTR_ERR(inode); - if (IS_ERR(inode)) + dentry->d_name.len, btrfs_ino(dir), objectid, + S_IFLNK|S_IRWXUGO, &index); + if (IS_ERR(inode)) { + err = PTR_ERR(inode); goto out_unlock; + } - err = btrfs_init_inode_security(trans, inode, dir); + err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name); if (err) { drop_inode = 1; goto out_unlock; } - btrfs_set_trans_block_group(trans, inode); - err = btrfs_add_nondir(trans, dentry, inode, 0, index); + /* + * If the active LSM wants to access the inode during + * d_instantiate it needs these. Smack checks to see + * if the filesystem supports xattrs by looking at the + * ops vector. + */ + inode->i_fop = &btrfs_file_operations; + inode->i_op = &btrfs_file_inode_operations; + + err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index); if (err) drop_inode = 1; else { inode->i_mapping->a_ops = &btrfs_aops; inode->i_mapping->backing_dev_info = &root->fs_info->bdi; - inode->i_fop = &btrfs_file_operations; - inode->i_op = &btrfs_file_inode_operations; BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; } - 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; + if (!path) { + err = -ENOMEM; + drop_inode = 1; + goto out_unlock; + } + key.objectid = btrfs_ino(inode); key.offset = 0; btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY); datasize = btrfs_file_extent_calc_inline_size(name_len); @@ -6798,6 +8758,7 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, datasize); if (err) { drop_inode = 1; + btrfs_free_path(path); goto out_unlock; } leaf = path->nodes[0]; @@ -6820,19 +8781,20 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, inode->i_mapping->a_ops = &btrfs_symlink_aops; inode->i_mapping->backing_dev_info = &root->fs_info->bdi; inode_set_bytes(inode, name_len); - btrfs_i_size_write(inode, name_len - 1); + btrfs_i_size_write(inode, name_len); err = btrfs_update_inode(trans, root, inode); if (err) drop_inode = 1; out_unlock: - nr = trans->blocks_used; - btrfs_end_transaction_throttle(trans, root); + if (!err) + d_instantiate(dentry, inode); + btrfs_end_transaction(trans, root); if (drop_inode) { inode_dec_link_count(inode); iput(inode); } - btrfs_btree_balance_dirty(root, nr); + btrfs_btree_balance_dirty(root); return err; } @@ -6841,9 +8803,13 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode, loff_t actual_len, u64 *alloc_hint, struct btrfs_trans_handle *trans) { + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_map *em; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_key ins; u64 cur_offset = start; + u64 i_size; + u64 cur_bytes; int ret = 0; bool own_trans = true; @@ -6858,8 +8824,10 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode, } } - ret = btrfs_reserve_extent(trans, root, num_bytes, min_size, - 0, *alloc_hint, (u64)-1, &ins, 1); + cur_bytes = min(num_bytes, 256ULL * 1024 * 1024); + cur_bytes = max(cur_bytes, min_size); + ret = btrfs_reserve_extent(root, cur_bytes, min_size, 0, + *alloc_hint, &ins, 1, 0); if (ret) { if (own_trans) btrfs_end_transaction(trans, root); @@ -6871,29 +8839,73 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode, ins.offset, ins.offset, ins.offset, 0, 0, 0, BTRFS_FILE_EXTENT_PREALLOC); - BUG_ON(ret); + if (ret) { + btrfs_free_reserved_extent(root, ins.objectid, + ins.offset, 0); + btrfs_abort_transaction(trans, root, ret); + if (own_trans) + btrfs_end_transaction(trans, root); + break; + } btrfs_drop_extent_cache(inode, cur_offset, cur_offset + ins.offset -1, 0); + em = alloc_extent_map(); + if (!em) { + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + goto next; + } + + em->start = cur_offset; + em->orig_start = cur_offset; + em->len = ins.offset; + em->block_start = ins.objectid; + em->block_len = ins.offset; + em->orig_block_len = ins.offset; + em->ram_bytes = ins.offset; + em->bdev = root->fs_info->fs_devices->latest_bdev; + set_bit(EXTENT_FLAG_PREALLOC, &em->flags); + em->generation = trans->transid; + + while (1) { + write_lock(&em_tree->lock); + ret = add_extent_mapping(em_tree, em, 1); + write_unlock(&em_tree->lock); + if (ret != -EEXIST) + break; + btrfs_drop_extent_cache(inode, cur_offset, + cur_offset + ins.offset - 1, + 0); + } + free_extent_map(em); +next: num_bytes -= ins.offset; cur_offset += ins.offset; *alloc_hint = ins.objectid + ins.offset; + inode_inc_iversion(inode); inode->i_ctime = CURRENT_TIME; BTRFS_I(inode)->flags |= BTRFS_INODE_PREALLOC; if (!(mode & FALLOC_FL_KEEP_SIZE) && (actual_len > inode->i_size) && (cur_offset > inode->i_size)) { if (cur_offset > actual_len) - i_size_write(inode, actual_len); + i_size = actual_len; else - i_size_write(inode, cur_offset); - i_size_write(inode, cur_offset); - btrfs_ordered_update_i_size(inode, cur_offset, NULL); + i_size = cur_offset; + i_size_write(inode, i_size); + btrfs_ordered_update_i_size(inode, i_size, NULL); } ret = btrfs_update_inode(trans, root, inode); - BUG_ON(ret); + + if (ret) { + btrfs_abort_transaction(trans, root, ret); + if (own_trans) + btrfs_end_transaction(trans, root); + break; + } if (own_trans) btrfs_end_transaction(trans, root); @@ -6919,118 +8931,84 @@ int btrfs_prealloc_file_range_trans(struct inode *inode, min_size, actual_len, alloc_hint, trans); } -static long btrfs_fallocate(struct inode *inode, int mode, - loff_t offset, loff_t len) +static int btrfs_set_page_dirty(struct page *page) { - struct extent_state *cached_state = NULL; - u64 cur_offset; - u64 last_byte; - u64 alloc_start; - u64 alloc_end; - u64 alloc_hint = 0; - u64 locked_end; - u64 mask = BTRFS_I(inode)->root->sectorsize - 1; - struct extent_map *em; - int ret; + return __set_page_dirty_nobuffers(page); +} + +static int btrfs_permission(struct inode *inode, int mask) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + umode_t mode = inode->i_mode; - alloc_start = offset & ~mask; - alloc_end = (offset + len + mask) & ~mask; + if (mask & MAY_WRITE && + (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode))) { + if (btrfs_root_readonly(root)) + return -EROFS; + if (BTRFS_I(inode)->flags & BTRFS_INODE_READONLY) + return -EACCES; + } + return generic_permission(inode, mask); +} + +static int btrfs_tmpfile(struct inode *dir, struct dentry *dentry, umode_t mode) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(dir)->root; + struct inode *inode = NULL; + u64 objectid; + u64 index; + int ret = 0; /* - * wait for ordered IO before we have any locks. We'll loop again - * below with the locks held. + * 5 units required for adding orphan entry */ - btrfs_wait_ordered_range(inode, alloc_start, alloc_end - alloc_start); + trans = btrfs_start_transaction(root, 5); + if (IS_ERR(trans)) + return PTR_ERR(trans); - mutex_lock(&inode->i_mutex); - if (alloc_start > inode->i_size) { - ret = btrfs_cont_expand(inode, alloc_start); - if (ret) - goto out; + ret = btrfs_find_free_ino(root, &objectid); + if (ret) + goto out; + + inode = btrfs_new_inode(trans, root, dir, NULL, 0, + btrfs_ino(dir), objectid, mode, &index); + if (IS_ERR(inode)) { + ret = PTR_ERR(inode); + inode = NULL; + goto out; } - ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start); + ret = btrfs_init_inode_security(trans, inode, dir, NULL); if (ret) goto out; - locked_end = alloc_end - 1; - while (1) { - struct btrfs_ordered_extent *ordered; + ret = btrfs_update_inode(trans, root, inode); + if (ret) + goto out; - /* the extent lock is ordered inside the running - * transaction - */ - lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start, - locked_end, 0, &cached_state, GFP_NOFS); - ordered = btrfs_lookup_first_ordered_extent(inode, - alloc_end - 1); - if (ordered && - ordered->file_offset + ordered->len > alloc_start && - ordered->file_offset < alloc_end) { - btrfs_put_ordered_extent(ordered); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, - alloc_start, locked_end, - &cached_state, GFP_NOFS); - /* - * we can't wait on the range with the transaction - * running or with the extent lock held - */ - btrfs_wait_ordered_range(inode, alloc_start, - alloc_end - alloc_start); - } else { - if (ordered) - btrfs_put_ordered_extent(ordered); - break; - } - } + inode->i_fop = &btrfs_file_operations; + inode->i_op = &btrfs_file_inode_operations; - cur_offset = alloc_start; - while (1) { - em = btrfs_get_extent(inode, NULL, 0, cur_offset, - alloc_end - cur_offset, 0); - BUG_ON(IS_ERR(em) || !em); - last_byte = min(extent_map_end(em), alloc_end); - last_byte = (last_byte + mask) & ~mask; - if (em->block_start == EXTENT_MAP_HOLE || - (cur_offset >= inode->i_size && - !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { - ret = btrfs_prealloc_file_range(inode, mode, cur_offset, - last_byte - cur_offset, - 1 << inode->i_blkbits, - offset + len, - &alloc_hint); - if (ret < 0) { - free_extent_map(em); - break; - } - } - free_extent_map(em); + inode->i_mapping->a_ops = &btrfs_aops; + inode->i_mapping->backing_dev_info = &root->fs_info->bdi; + BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; - cur_offset = last_byte; - if (cur_offset >= alloc_end) { - ret = 0; - break; - } - } - unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, - &cached_state, GFP_NOFS); + ret = btrfs_orphan_add(trans, inode); + if (ret) + goto out; - btrfs_free_reserved_data_space(inode, alloc_end - alloc_start); -out: - mutex_unlock(&inode->i_mutex); - return ret; -} + d_tmpfile(dentry, inode); + mark_inode_dirty(inode); -static int btrfs_set_page_dirty(struct page *page) -{ - return __set_page_dirty_nobuffers(page); -} +out: + btrfs_end_transaction(trans, root); + if (ret) + iput(inode); + btrfs_balance_delayed_items(root); + btrfs_btree_balance_dirty(root); -static int btrfs_permission(struct inode *inode, int mask) -{ - if ((BTRFS_I(inode)->flags & BTRFS_INODE_READONLY) && (mask & MAY_WRITE)) - return -EACCES; - return generic_permission(inode, mask, btrfs_check_acl); + return ret; } static const struct inode_operations btrfs_dir_inode_operations = { @@ -7050,16 +9028,23 @@ static const struct inode_operations btrfs_dir_inode_operations = { .listxattr = btrfs_listxattr, .removexattr = btrfs_removexattr, .permission = btrfs_permission, + .get_acl = btrfs_get_acl, + .set_acl = btrfs_set_acl, + .update_time = btrfs_update_time, + .tmpfile = btrfs_tmpfile, }; static const struct inode_operations btrfs_dir_ro_inode_operations = { .lookup = btrfs_lookup, .permission = btrfs_permission, + .get_acl = btrfs_get_acl, + .set_acl = btrfs_set_acl, + .update_time = btrfs_update_time, }; static const struct file_operations btrfs_dir_file_operations = { .llseek = generic_file_llseek, .read = generic_read_dir, - .readdir = btrfs_real_readdir, + .iterate = btrfs_real_readdir, .unlocked_ioctl = btrfs_ioctl, #ifdef CONFIG_COMPAT .compat_ioctl = btrfs_ioctl, @@ -7075,7 +9060,6 @@ static struct extent_io_ops btrfs_extent_io_ops = { .readpage_end_io_hook = btrfs_readpage_end_io_hook, .writepage_end_io_hook = btrfs_writepage_end_io_hook, .writepage_start_hook = btrfs_writepage_start_hook, - .readpage_io_failed_hook = btrfs_io_failed_hook, .set_bit_hook = btrfs_set_bit_hook, .clear_bit_hook = btrfs_clear_bit_hook, .merge_extent_hook = btrfs_merge_extent_hook, @@ -7099,7 +9083,6 @@ static const struct address_space_operations btrfs_aops = { .writepage = btrfs_writepage, .writepages = btrfs_writepages, .readpages = btrfs_readpages, - .sync_page = block_sync_page, .direct_IO = btrfs_direct_IO, .invalidatepage = btrfs_invalidatepage, .releasepage = btrfs_releasepage, @@ -7115,7 +9098,6 @@ static const struct address_space_operations btrfs_symlink_aops = { }; static const struct inode_operations btrfs_file_inode_operations = { - .truncate = btrfs_truncate, .getattr = btrfs_getattr, .setattr = btrfs_setattr, .setxattr = btrfs_setxattr, @@ -7123,8 +9105,10 @@ static const struct inode_operations btrfs_file_inode_operations = { .listxattr = btrfs_listxattr, .removexattr = btrfs_removexattr, .permission = btrfs_permission, - .fallocate = btrfs_fallocate, .fiemap = btrfs_fiemap, + .get_acl = btrfs_get_acl, + .set_acl = btrfs_set_acl, + .update_time = btrfs_update_time, }; static const struct inode_operations btrfs_special_inode_operations = { .getattr = btrfs_getattr, @@ -7134,18 +9118,25 @@ static const struct inode_operations btrfs_special_inode_operations = { .getxattr = btrfs_getxattr, .listxattr = btrfs_listxattr, .removexattr = btrfs_removexattr, + .get_acl = btrfs_get_acl, + .set_acl = btrfs_set_acl, + .update_time = btrfs_update_time, }; static const struct inode_operations btrfs_symlink_inode_operations = { .readlink = generic_readlink, .follow_link = page_follow_link_light, .put_link = page_put_link, + .getattr = btrfs_getattr, + .setattr = btrfs_setattr, .permission = btrfs_permission, .setxattr = btrfs_setxattr, .getxattr = btrfs_getxattr, .listxattr = btrfs_listxattr, .removexattr = btrfs_removexattr, + .update_time = btrfs_update_time, }; const struct dentry_operations btrfs_dentry_operations = { .d_delete = btrfs_dentry_delete, + .d_release = btrfs_dentry_release, }; |