diff options
Diffstat (limited to 'fs/btrfs/inode.c')
| -rw-r--r-- | fs/btrfs/inode.c | 974 |
1 files changed, 562 insertions, 412 deletions
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 5618ca02934a..fe2c810335ff 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -393,35 +393,14 @@ void btrfs_inode_unlock(struct btrfs_inode *inode, unsigned int ilock_flags) * extent (btrfs_finish_ordered_io()). */ static inline void btrfs_cleanup_ordered_extents(struct btrfs_inode *inode, - struct folio *locked_folio, u64 offset, u64 bytes) { unsigned long index = offset >> PAGE_SHIFT; unsigned long end_index = (offset + bytes - 1) >> PAGE_SHIFT; - u64 page_start = 0, page_end = 0; struct folio *folio; - if (locked_folio) { - page_start = folio_pos(locked_folio); - page_end = page_start + folio_size(locked_folio) - 1; - } - while (index <= end_index) { - /* - * For locked page, we will call btrfs_mark_ordered_io_finished - * through btrfs_mark_ordered_io_finished() on it - * in run_delalloc_range() for the error handling, which will - * clear page Ordered and run the ordered extent accounting. - * - * Here we can't just clear the Ordered bit, or - * btrfs_mark_ordered_io_finished() would skip the accounting - * for the page range, and the ordered extent will never finish. - */ - if (locked_folio && index == (page_start >> PAGE_SHIFT)) { - index++; - continue; - } - folio = __filemap_get_folio(inode->vfs_inode.i_mapping, index, 0, 0); + folio = filemap_get_folio(inode->vfs_inode.i_mapping, index); index++; if (IS_ERR(folio)) continue; @@ -436,23 +415,6 @@ static inline void btrfs_cleanup_ordered_extents(struct btrfs_inode *inode, folio_put(folio); } - if (locked_folio) { - /* The locked page covers the full range, nothing needs to be done */ - if (bytes + offset <= page_start + folio_size(locked_folio)) - return; - /* - * In case this page belongs to the delalloc range being - * instantiated then skip it, since the first page of a range is - * going to be properly cleaned up by the caller of - * run_delalloc_range - */ - if (page_start >= offset && page_end <= (offset + bytes - 1)) { - bytes = offset + bytes - folio_pos(locked_folio) - - folio_size(locked_folio); - offset = folio_pos(locked_folio) + folio_size(locked_folio); - } - } - return btrfs_mark_ordered_io_finished(inode, NULL, offset, bytes, false); } @@ -556,8 +518,7 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans, } else { struct folio *folio; - folio = __filemap_get_folio(inode->vfs_inode.i_mapping, - 0, 0, 0); + folio = filemap_get_folio(inode->vfs_inode.i_mapping, 0); ASSERT(!IS_ERR(folio)); btrfs_set_file_extent_compression(leaf, ei, 0); kaddr = kmap_local_folio(folio, 0); @@ -565,7 +526,6 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans, kunmap_local(kaddr); folio_put(folio); } - btrfs_mark_buffer_dirty(trans, leaf); btrfs_release_path(path); /* @@ -646,7 +606,7 @@ static bool can_cow_file_range_inline(struct btrfs_inode *inode, * If being used directly, you must have already checked we're allowed to cow * the range by getting true from can_cow_file_range_inline(). */ -static noinline int __cow_file_range_inline(struct btrfs_inode *inode, u64 offset, +static noinline int __cow_file_range_inline(struct btrfs_inode *inode, u64 size, size_t compressed_size, int compress_type, struct folio *compressed_folio, @@ -736,7 +696,7 @@ static noinline int cow_file_range_inline(struct btrfs_inode *inode, return 1; lock_extent(&inode->io_tree, offset, end, &cached); - ret = __cow_file_range_inline(inode, offset, size, compressed_size, + ret = __cow_file_range_inline(inode, size, compressed_size, compress_type, compressed_folio, update_i_size); if (ret > 0) { @@ -832,32 +792,16 @@ static inline int inode_need_compress(struct btrfs_inode *inode, u64 start, return 0; } /* - * Special check for subpage. - * - * We lock the full page then run each delalloc range in the page, thus - * for the following case, we will hit some subpage specific corner case: + * Only enable sector perfect compression for experimental builds. * - * 0 32K 64K - * | |///////| |///////| - * \- A \- B + * This is a big feature change for subpage cases, and can hit + * different corner cases, so only limit this feature for + * experimental build for now. * - * In above case, both range A and range B will try to unlock the full - * page [0, 64K), causing the one finished later will have page - * unlocked already, triggering various page lock requirement BUG_ON()s. - * - * So here we add an artificial limit that subpage compression can only - * if the range is fully page aligned. - * - * In theory we only need to ensure the first page is fully covered, but - * the tailing partial page will be locked until the full compression - * finishes, delaying the write of other range. - * - * TODO: Make btrfs_run_delalloc_range() to lock all delalloc range - * first to prevent any submitted async extent to unlock the full page. - * By this, we can ensure for subpage case that only the last async_cow - * will unlock the full page. + * ETA for moving this out of experimental builds is 6.15. */ - if (fs_info->sectorsize < PAGE_SIZE) { + if (fs_info->sectorsize < PAGE_SIZE && + !IS_ENABLED(CONFIG_BTRFS_EXPERIMENTAL)) { if (!PAGE_ALIGNED(start) || !PAGE_ALIGNED(end + 1)) return 0; @@ -896,13 +840,14 @@ static int extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 e for (unsigned long index = start >> PAGE_SHIFT; index <= end_index; index++) { - folio = __filemap_get_folio(inode->i_mapping, index, 0, 0); + folio = filemap_get_folio(inode->i_mapping, index); if (IS_ERR(folio)) { if (!ret) ret = PTR_ERR(folio); continue; } - folio_clear_dirty_for_io(folio); + btrfs_folio_clamp_clear_dirty(inode_to_fs_info(inode), folio, start, + end + 1 - start); folio_put(folio); } return ret; @@ -1001,17 +946,6 @@ again: (start > 0 || end + 1 < inode->disk_i_size)) goto cleanup_and_bail_uncompressed; - /* - * For subpage case, we require full page alignment for the sector - * aligned range. - * Thus we must also check against @actual_end, not just @end. - */ - if (blocksize < PAGE_SIZE) { - if (!PAGE_ALIGNED(start) || - !PAGE_ALIGNED(round_up(actual_end, blocksize))) - goto cleanup_and_bail_uncompressed; - } - total_compressed = min_t(unsigned long, total_compressed, BTRFS_MAX_UNCOMPRESSED); total_in = 0; @@ -1156,19 +1090,14 @@ static void submit_uncompressed_range(struct btrfs_inode *inode, &wbc, false); wbc_detach_inode(&wbc); if (ret < 0) { - btrfs_cleanup_ordered_extents(inode, locked_folio, - start, end - start + 1); - if (locked_folio) { - const u64 page_start = folio_pos(locked_folio); - - folio_start_writeback(locked_folio); - folio_end_writeback(locked_folio); - btrfs_mark_ordered_io_finished(inode, locked_folio, - page_start, PAGE_SIZE, - !ret); - mapping_set_error(locked_folio->mapping, ret); - folio_unlock(locked_folio); - } + btrfs_cleanup_ordered_extents(inode, start, end - start + 1); + if (locked_folio) + btrfs_folio_end_lock(inode->root->fs_info, locked_folio, + start, async_extent->ram_size); + btrfs_err_rl(inode->root->fs_info, + "%s failed, root=%llu inode=%llu start=%llu len=%llu: %d", + __func__, btrfs_root_id(inode->root), + btrfs_ino(inode), start, async_extent->ram_size, ret); } } @@ -1359,7 +1288,6 @@ static noinline int cow_file_range(struct btrfs_inode *inode, u64 alloc_hint = 0; u64 orig_start = start; u64 num_bytes; - unsigned long ram_size; u64 cur_alloc_size = 0; u64 min_alloc_size; u64 blocksize = fs_info->sectorsize; @@ -1367,7 +1295,6 @@ static noinline int cow_file_range(struct btrfs_inode *inode, struct extent_map *em; unsigned clear_bits; unsigned long page_ops; - bool extent_reserved = false; int ret = 0; if (btrfs_is_free_space_inode(inode)) { @@ -1402,6 +1329,17 @@ static noinline int cow_file_range(struct btrfs_inode *inode, alloc_hint = btrfs_get_extent_allocation_hint(inode, start, num_bytes); /* + * We're not doing compressed IO, don't unlock the first page (which + * the caller expects to stay locked), don't clear any dirty bits and + * don't set any writeback bits. + * + * Do set the Ordered (Private2) bit so we know this page was properly + * setup for writepage. + */ + page_ops = (keep_locked ? 0 : PAGE_UNLOCK); + page_ops |= PAGE_SET_ORDERED; + + /* * Relocation relies on the relocated extents to have exactly the same * size as the original extents. Normally writeback for relocation data * extents follows a NOCOW path because relocation preallocates the @@ -1421,8 +1359,7 @@ static noinline int cow_file_range(struct btrfs_inode *inode, struct btrfs_ordered_extent *ordered; struct btrfs_file_extent file_extent; - cur_alloc_size = num_bytes; - ret = btrfs_reserve_extent(root, cur_alloc_size, cur_alloc_size, + ret = btrfs_reserve_extent(root, num_bytes, num_bytes, min_alloc_size, 0, alloc_hint, &ins, 1, 1); if (ret == -EAGAIN) { @@ -1453,9 +1390,7 @@ static noinline int cow_file_range(struct btrfs_inode *inode, if (ret < 0) goto out_unlock; cur_alloc_size = ins.offset; - extent_reserved = true; - ram_size = ins.offset; file_extent.disk_bytenr = ins.objectid; file_extent.disk_num_bytes = ins.offset; file_extent.num_bytes = ins.offset; @@ -1463,14 +1398,18 @@ static noinline int cow_file_range(struct btrfs_inode *inode, file_extent.offset = 0; file_extent.compression = BTRFS_COMPRESS_NONE; - lock_extent(&inode->io_tree, start, start + ram_size - 1, + /* + * Locked range will be released either during error clean up or + * after the whole range is finished. + */ + lock_extent(&inode->io_tree, start, start + cur_alloc_size - 1, &cached); em = btrfs_create_io_em(inode, start, &file_extent, BTRFS_ORDERED_REGULAR); if (IS_ERR(em)) { unlock_extent(&inode->io_tree, start, - start + ram_size - 1, &cached); + start + cur_alloc_size - 1, &cached); ret = PTR_ERR(em); goto out_reserve; } @@ -1480,7 +1419,7 @@ static noinline int cow_file_range(struct btrfs_inode *inode, 1 << BTRFS_ORDERED_REGULAR); if (IS_ERR(ordered)) { unlock_extent(&inode->io_tree, start, - start + ram_size - 1, &cached); + start + cur_alloc_size - 1, &cached); ret = PTR_ERR(ordered); goto out_drop_extent_cache; } @@ -1501,35 +1440,20 @@ static noinline int cow_file_range(struct btrfs_inode *inode, */ if (ret) btrfs_drop_extent_map_range(inode, start, - start + ram_size - 1, + start + cur_alloc_size - 1, false); } btrfs_put_ordered_extent(ordered); btrfs_dec_block_group_reservations(fs_info, ins.objectid); - /* - * We're not doing compressed IO, don't unlock the first page - * (which the caller expects to stay locked), don't clear any - * dirty bits and don't set any writeback bits - * - * Do set the Ordered (Private2) bit so we know this page was - * properly setup for writepage. - */ - page_ops = (keep_locked ? 0 : PAGE_UNLOCK); - page_ops |= PAGE_SET_ORDERED; - - extent_clear_unlock_delalloc(inode, start, start + ram_size - 1, - locked_folio, &cached, - EXTENT_LOCKED | EXTENT_DELALLOC, - page_ops); if (num_bytes < cur_alloc_size) num_bytes = 0; else num_bytes -= cur_alloc_size; alloc_hint = ins.objectid + ins.offset; start += cur_alloc_size; - extent_reserved = false; + cur_alloc_size = 0; /* * btrfs_reloc_clone_csums() error, since start is increased @@ -1539,13 +1463,15 @@ static noinline int cow_file_range(struct btrfs_inode *inode, if (ret) goto out_unlock; } + extent_clear_unlock_delalloc(inode, orig_start, end, locked_folio, &cached, + EXTENT_LOCKED | EXTENT_DELALLOC, page_ops); done: if (done_offset) *done_offset = end; return ret; out_drop_extent_cache: - btrfs_drop_extent_map_range(inode, start, start + ram_size - 1, false); + btrfs_drop_extent_map_range(inode, start, start + cur_alloc_size - 1, false); out_reserve: btrfs_dec_block_group_reservations(fs_info, ins.objectid); btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1); @@ -1559,35 +1485,30 @@ out_unlock: * We process each region below. */ - clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW | - EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV; - page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK; - /* * For the range (1). We have already instantiated the ordered extents * for this region. They are cleaned up by * btrfs_cleanup_ordered_extents() in e.g, - * btrfs_run_delalloc_range(). EXTENT_LOCKED | EXTENT_DELALLOC are - * already cleared in the above loop. And, EXTENT_DELALLOC_NEW | - * EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV are handled by the cleanup - * function. + * btrfs_run_delalloc_range(). + * EXTENT_DELALLOC_NEW | EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV + * are also handled by the cleanup function. * - * However, in case of @keep_locked, we still need to unlock the pages - * (except @locked_folio) to ensure all the pages are unlocked. + * So here we only clear EXTENT_LOCKED and EXTENT_DELALLOC flag, and + * finish the writeback of the involved folios, which will be never submitted. */ - if (keep_locked && orig_start < start) { + if (orig_start < start) { + clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC; + page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK; + if (!locked_folio) mapping_set_error(inode->vfs_inode.i_mapping, ret); extent_clear_unlock_delalloc(inode, orig_start, start - 1, - locked_folio, NULL, 0, page_ops); + locked_folio, NULL, clear_bits, page_ops); } - /* - * At this point we're unlocked, we want to make sure we're only - * clearing these flags under the extent lock, so lock the rest of the - * range and clear everything up. - */ - lock_extent(&inode->io_tree, start, end, NULL); + clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW | + EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV; + page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK; /* * For the range (2). If we reserved an extent for our delalloc range @@ -1599,13 +1520,12 @@ out_unlock: * to decrement again the data space_info's bytes_may_use counter, * therefore we do not pass it the flag EXTENT_CLEAR_DATA_RESV. */ - if (extent_reserved) { + if (cur_alloc_size) { extent_clear_unlock_delalloc(inode, start, start + cur_alloc_size - 1, locked_folio, &cached, clear_bits, page_ops); btrfs_qgroup_free_data(inode, NULL, start, cur_alloc_size, NULL); - start += cur_alloc_size; } /* @@ -1614,12 +1534,18 @@ out_unlock: * space_info's bytes_may_use counter, reserved in * btrfs_check_data_free_space(). */ - if (start < end) { + if (start + cur_alloc_size < end) { clear_bits |= EXTENT_CLEAR_DATA_RESV; - extent_clear_unlock_delalloc(inode, start, end, locked_folio, + extent_clear_unlock_delalloc(inode, start + cur_alloc_size, + end, locked_folio, &cached, clear_bits, page_ops); - btrfs_qgroup_free_data(inode, NULL, start, cur_alloc_size, NULL); + btrfs_qgroup_free_data(inode, NULL, start + cur_alloc_size, + end - start - cur_alloc_size + 1, NULL); } + btrfs_err_rl(fs_info, + "%s failed, root=%llu inode=%llu start=%llu len=%llu: %d", + __func__, btrfs_root_id(inode->root), + btrfs_ino(inode), orig_start, end + 1 - orig_start, ret); return ret; } @@ -1729,7 +1655,7 @@ static bool run_delalloc_compressed(struct btrfs_inode *inode, * need full accuracy. Just account the whole thing * against the first page. */ - wbc_account_cgroup_owner(wbc, &locked_folio->page, + wbc_account_cgroup_owner(wbc, locked_folio, cur_end - start); async_chunk[i].locked_folio = locked_folio; locked_folio = NULL; @@ -1840,7 +1766,7 @@ static int fallback_to_cow(struct btrfs_inode *inode, bytes = range_bytes; spin_lock(&sinfo->lock); - btrfs_space_info_update_bytes_may_use(fs_info, sinfo, bytes); + btrfs_space_info_update_bytes_may_use(sinfo, bytes); spin_unlock(&sinfo->lock); if (count > 0) @@ -1868,7 +1794,6 @@ struct can_nocow_file_extent_args { /* End file offset (inclusive) of the range we want to NOCOW. */ u64 end; bool writeback_path; - bool strict; /* * Free the path passed to can_nocow_file_extent() once it's not needed * anymore. @@ -1923,8 +1848,7 @@ static int can_nocow_file_extent(struct btrfs_path *path, * for its subvolume was created, then this implies the extent is shared, * hence we must COW. */ - if (!args->strict && - btrfs_file_extent_generation(leaf, fi) <= + if (btrfs_file_extent_generation(leaf, fi) <= btrfs_root_last_snapshot(&root->root_item)) goto out; @@ -1953,9 +1877,8 @@ static int can_nocow_file_extent(struct btrfs_path *path, */ btrfs_release_path(path); - ret = btrfs_cross_ref_exist(root, btrfs_ino(inode), - key->offset - args->file_extent.offset, - args->file_extent.disk_bytenr, args->strict, path); + ret = btrfs_cross_ref_exist(inode, key->offset - args->file_extent.offset, + args->file_extent.disk_bytenr, path); WARN_ON_ONCE(ret > 0 && is_freespace_inode); if (ret != 0) goto out; @@ -2002,6 +1925,53 @@ static int can_nocow_file_extent(struct btrfs_path *path, } /* + * Cleanup the dirty folios which will never be submitted due to error. + * + * When running a delalloc range, we may need to split the ranges (due to + * fragmentation or NOCOW). If we hit an error in the later part, we will error + * out and previously successfully executed range will never be submitted, thus + * we have to cleanup those folios by clearing their dirty flag, starting and + * finishing the writeback. + */ +static void cleanup_dirty_folios(struct btrfs_inode *inode, + struct folio *locked_folio, + u64 start, u64 end, int error) +{ + struct btrfs_fs_info *fs_info = inode->root->fs_info; + struct address_space *mapping = inode->vfs_inode.i_mapping; + pgoff_t start_index = start >> PAGE_SHIFT; + pgoff_t end_index = end >> PAGE_SHIFT; + u32 len; + + ASSERT(end + 1 - start < U32_MAX); + ASSERT(IS_ALIGNED(start, fs_info->sectorsize) && + IS_ALIGNED(end + 1, fs_info->sectorsize)); + len = end + 1 - start; + + /* + * Handle the locked folio first. + * The btrfs_folio_clamp_*() helpers can handle range out of the folio case. + */ + btrfs_folio_clamp_finish_io(fs_info, locked_folio, start, len); + + for (pgoff_t index = start_index; index <= end_index; index++) { + struct folio *folio; + + /* Already handled at the beginning. */ + if (index == locked_folio->index) + continue; + folio = __filemap_get_folio(mapping, index, FGP_LOCK, GFP_NOFS); + /* Cache already dropped, no need to do any cleanup. */ + if (IS_ERR(folio)) + continue; + btrfs_folio_clamp_finish_io(fs_info, locked_folio, start, len); + folio_unlock(folio); + folio_put(folio); + } + mapping_set_error(mapping, error); +} + +/* * when nowcow writeback call back. This checks for snapshots or COW copies * of the extents that exist in the file, and COWs the file as required. * @@ -2016,6 +1986,11 @@ static noinline int run_delalloc_nocow(struct btrfs_inode *inode, struct btrfs_root *root = inode->root; struct btrfs_path *path; u64 cow_start = (u64)-1; + /* + * If not 0, represents the inclusive end of the last fallback_to_cow() + * range. Only for error handling. + */ + u64 cow_end = 0; u64 cur_offset = start; int ret; bool check_prev = true; @@ -2176,6 +2151,7 @@ must_cow: found_key.offset - 1); cow_start = (u64)-1; if (ret) { + cow_end = found_key.offset - 1; btrfs_dec_nocow_writers(nocow_bg); goto error; } @@ -2249,11 +2225,12 @@ must_cow: cow_start = cur_offset; if (cow_start != (u64)-1) { - cur_offset = end; ret = fallback_to_cow(inode, locked_folio, cow_start, end); cow_start = (u64)-1; - if (ret) + if (ret) { + cow_end = end; goto error; + } } btrfs_free_path(path); @@ -2261,12 +2238,41 @@ must_cow: error: /* + * There are several error cases: + * + * 1) Failed without falling back to COW + * start cur_offset end + * |/////////////| | + * + * For range [start, cur_offset) the folios are already unlocked (except + * @locked_folio), EXTENT_DELALLOC already removed. + * Only need to clear the dirty flag as they will never be submitted. + * Ordered extent and extent maps are handled by + * btrfs_mark_ordered_io_finished() inside run_delalloc_range(). + * + * 2) Failed with error from fallback_to_cow() + * start cur_offset cow_end end + * |/////////////|-----------| | + * + * For range [start, cur_offset) it's the same as case 1). + * But for range [cur_offset, cow_end), the folios have dirty flag + * cleared and unlocked, EXTENT_DEALLLOC cleared by cow_file_range(). + * + * Thus we should not call extent_clear_unlock_delalloc() on range + * [cur_offset, cow_end), as the folios are already unlocked. + * + * So clear the folio dirty flags for [start, cur_offset) first. + */ + if (cur_offset > start) + cleanup_dirty_folios(inode, locked_folio, start, cur_offset - 1, ret); + + /* * If an error happened while a COW region is outstanding, cur_offset - * needs to be reset to cow_start to ensure the COW region is unlocked - * as well. + * needs to be reset to @cow_end + 1 to skip the COW range, as + * cow_file_range() will do the proper cleanup at error. */ - if (cow_start != (u64)-1) - cur_offset = cow_start; + if (cow_end) + cur_offset = cow_end + 1; /* * We need to lock the extent here because we're clearing DELALLOC and @@ -2286,6 +2292,10 @@ error: btrfs_qgroup_free_data(inode, NULL, cur_offset, end - cur_offset + 1, NULL); } btrfs_free_path(path); + btrfs_err_rl(fs_info, + "%s failed, root=%llu inode=%llu start=%llu len=%llu: %d", + __func__, btrfs_root_id(inode->root), + btrfs_ino(inode), start, end + 1 - start, ret); return ret; } @@ -2336,8 +2346,7 @@ int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct folio *locked_fol out: if (ret < 0) - btrfs_cleanup_ordered_extents(inode, locked_folio, start, - end - start + 1); + btrfs_cleanup_ordered_extents(inode, start, end - start + 1); return ret; } @@ -2952,7 +2961,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, btrfs_item_ptr_offset(leaf, path->slots[0]), sizeof(struct btrfs_file_extent_item)); - btrfs_mark_buffer_dirty(trans, leaf); btrfs_release_path(path); /* @@ -3094,34 +3102,19 @@ int btrfs_finish_one_ordered(struct btrfs_ordered_extent *ordered_extent) goto out; } - if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) { - BUG_ON(!list_empty(&ordered_extent->list)); /* Logic error */ - - btrfs_inode_safe_disk_i_size_write(inode, 0); - if (freespace_inode) - trans = btrfs_join_transaction_spacecache(root); - else - trans = btrfs_join_transaction(root); - if (IS_ERR(trans)) { - ret = PTR_ERR(trans); - trans = NULL; - goto out; - } - trans->block_rsv = &inode->block_rsv; - ret = btrfs_update_inode_fallback(trans, inode); - if (ret) /* -ENOMEM or corruption */ - btrfs_abort_transaction(trans, ret); - - ret = btrfs_insert_raid_extent(trans, ordered_extent); - if (ret) - btrfs_abort_transaction(trans, ret); - - goto out; + /* + * If it's a COW write we need to lock the extent range as we will be + * inserting/replacing file extent items and unpinning an extent map. + * This must be taken before joining a transaction, as it's a higher + * level lock (like the inode's VFS lock), otherwise we can run into an + * ABBA deadlock with other tasks (transactions work like a lock, + * depending on their current state). + */ + if (!test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) { + clear_bits |= EXTENT_LOCKED; + lock_extent(io_tree, start, end, &cached_state); } - clear_bits |= EXTENT_LOCKED; - lock_extent(io_tree, start, end, &cached_state); - if (freespace_inode) trans = btrfs_join_transaction_spacecache(root); else @@ -3135,8 +3128,28 @@ int btrfs_finish_one_ordered(struct btrfs_ordered_extent *ordered_extent) trans->block_rsv = &inode->block_rsv; ret = btrfs_insert_raid_extent(trans, ordered_extent); - if (ret) + if (ret) { + btrfs_abort_transaction(trans, ret); goto out; + } + + if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) { + /* Logic error */ + ASSERT(list_empty(&ordered_extent->list)); + if (!list_empty(&ordered_extent->list)) { + ret = -EINVAL; + btrfs_abort_transaction(trans, ret); + goto out; + } + + btrfs_inode_safe_disk_i_size_write(inode, 0); + ret = btrfs_update_inode_fallback(trans, inode); + if (ret) { + /* -ENOMEM or corruption */ + btrfs_abort_transaction(trans, ret); + } + goto out; + } if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags)) compress_type = ordered_extent->compress_type; @@ -3791,14 +3804,45 @@ static int btrfs_init_file_extent_tree(struct btrfs_inode *inode) return 0; } +static int btrfs_add_inode_to_root(struct btrfs_inode *inode, bool prealloc) +{ + struct btrfs_root *root = inode->root; + struct btrfs_inode *existing; + const u64 ino = btrfs_ino(inode); + int ret; + + if (inode_unhashed(&inode->vfs_inode)) + return 0; + + if (prealloc) { + ret = xa_reserve(&root->inodes, ino, GFP_NOFS); + if (ret) + return ret; + } + + existing = xa_store(&root->inodes, ino, inode, GFP_ATOMIC); + + if (xa_is_err(existing)) { + ret = xa_err(existing); + ASSERT(ret != -EINVAL); + ASSERT(ret != -ENOMEM); + return ret; + } else if (existing) { + WARN_ON(!(existing->vfs_inode.i_state & (I_WILL_FREE | I_FREEING))); + } + + return 0; +} + /* - * read an inode from the btree into the in-memory inode + * Read a locked inode from the btree into the in-memory inode and add it to + * its root list/tree. + * + * On failure clean up the inode. */ -static int btrfs_read_locked_inode(struct inode *inode, - struct btrfs_path *in_path) +static int btrfs_read_locked_inode(struct inode *inode, struct btrfs_path *path) { struct btrfs_fs_info *fs_info = inode_to_fs_info(inode); - struct btrfs_path *path = in_path; struct extent_buffer *leaf; struct btrfs_inode_item *inode_item; struct btrfs_root *root = BTRFS_I(inode)->root; @@ -3812,25 +3856,25 @@ static int btrfs_read_locked_inode(struct inode *inode, ret = btrfs_init_file_extent_tree(BTRFS_I(inode)); if (ret) - return ret; + goto out; ret = btrfs_fill_inode(inode, &rdev); if (!ret) filled = true; - if (!path) { - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; - } + ASSERT(path); btrfs_get_inode_key(BTRFS_I(inode), &location); ret = btrfs_lookup_inode(NULL, root, path, &location, 0); if (ret) { - if (path != in_path) - btrfs_free_path(path); - return ret; + /* + * ret > 0 can come from btrfs_search_slot called by + * btrfs_lookup_inode(), this means the inode was not found. + */ + if (ret > 0) + ret = -ENOENT; + goto out; } leaf = path->nodes[0]; @@ -3965,8 +4009,6 @@ cache_acl: btrfs_ino(BTRFS_I(inode)), btrfs_root_id(root), ret); } - if (path != in_path) - btrfs_free_path(path); if (!maybe_acls) cache_no_acl(inode); @@ -3993,7 +4035,15 @@ cache_acl: } btrfs_sync_inode_flags_to_i_flags(inode); + + ret = btrfs_add_inode_to_root(BTRFS_I(inode), true); + if (ret) + goto out; + return 0; +out: + iget_failed(inode); + return ret; } /* @@ -4074,7 +4124,6 @@ static noinline int btrfs_update_inode_item(struct btrfs_trans_handle *trans, struct btrfs_inode_item); fill_inode_item(trans, leaf, inode_item, &inode->vfs_inode); - btrfs_mark_buffer_dirty(trans, leaf); btrfs_set_inode_last_trans(trans, inode); ret = 0; failed: @@ -4368,11 +4417,8 @@ static int btrfs_unlink_subvol(struct btrfs_trans_handle *trans, */ if (btrfs_ino(inode) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID) { di = btrfs_search_dir_index_item(root, path, dir_ino, &fname.disk_name); - if (IS_ERR_OR_NULL(di)) { - if (!di) - ret = -ENOENT; - else - ret = PTR_ERR(di); + if (IS_ERR(di)) { + ret = PTR_ERR(di); btrfs_abort_transaction(trans, ret); goto out; } @@ -5505,35 +5551,7 @@ out: return err; } -static int btrfs_add_inode_to_root(struct btrfs_inode *inode, bool prealloc) -{ - struct btrfs_root *root = inode->root; - struct btrfs_inode *existing; - const u64 ino = btrfs_ino(inode); - int ret; - - if (inode_unhashed(&inode->vfs_inode)) - return 0; - - if (prealloc) { - ret = xa_reserve(&root->inodes, ino, GFP_NOFS); - if (ret) - return ret; - } - - existing = xa_store(&root->inodes, ino, inode, GFP_ATOMIC); - - if (xa_is_err(existing)) { - ret = xa_err(existing); - ASSERT(ret != -EINVAL); - ASSERT(ret != -ENOMEM); - return ret; - } else if (existing) { - WARN_ON(!(existing->vfs_inode.i_state & (I_WILL_FREE | I_FREEING))); - } - return 0; -} static void btrfs_del_inode_from_root(struct btrfs_inode *inode) { @@ -5595,10 +5613,8 @@ static struct inode *btrfs_iget_locked(u64 ino, struct btrfs_root *root) } /* - * Get an inode object given its inode number and corresponding root. - * Path can be preallocated to prevent recursing back to iget through - * allocator. NULL is also valid but may require an additional allocation - * later. + * Get an inode object given its inode number and corresponding root. Path is + * preallocated to prevent recursing back to iget through allocator. */ struct inode *btrfs_iget_path(u64 ino, struct btrfs_root *root, struct btrfs_path *path) @@ -5614,30 +5630,40 @@ struct inode *btrfs_iget_path(u64 ino, struct btrfs_root *root, return inode; ret = btrfs_read_locked_inode(inode, path); - /* - * ret > 0 can come from btrfs_search_slot called by - * btrfs_read_locked_inode(), this means the inode item was not found. - */ - if (ret > 0) - ret = -ENOENT; - if (ret < 0) - goto error; - - ret = btrfs_add_inode_to_root(BTRFS_I(inode), true); - if (ret < 0) - goto error; + if (ret) + return ERR_PTR(ret); unlock_new_inode(inode); - return inode; -error: - iget_failed(inode); - return ERR_PTR(ret); } +/* + * Get an inode object given its inode number and corresponding root. + */ struct inode *btrfs_iget(u64 ino, struct btrfs_root *root) { - return btrfs_iget_path(ino, root, NULL); + struct inode *inode; + struct btrfs_path *path; + int ret; + + inode = btrfs_iget_locked(ino, root); + if (!inode) + return ERR_PTR(-ENOMEM); + + if (!(inode->i_state & I_NEW)) + return inode; + + path = btrfs_alloc_path(); + if (!path) + return ERR_PTR(-ENOMEM); + + ret = btrfs_read_locked_inode(inode, path); + btrfs_free_path(path); + if (ret) + return ERR_PTR(ret); + + unlock_new_inode(inode); + return inode; } static struct inode *new_simple_dir(struct inode *dir, @@ -6026,7 +6052,7 @@ again: * 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 + * they're returned by readdir. Until we reuse freed offsets * we have this hack to stop new entries from being returned * under the assumption that they'll never reach this huge * offset. @@ -6392,7 +6418,6 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans, } } - btrfs_mark_buffer_dirty(trans, path->nodes[0]); /* * We don't need the path anymore, plus inheriting properties, adding * ACLs, security xattrs, orphan item or adding the link, will result in @@ -6768,8 +6793,7 @@ static noinline int uncompress_inline(struct btrfs_path *path, return ret; } -static int read_inline_extent(struct btrfs_inode *inode, struct btrfs_path *path, - struct folio *folio) +static int read_inline_extent(struct btrfs_path *path, struct folio *folio) { struct btrfs_file_extent_item *fi; void *kaddr; @@ -6967,7 +6991,7 @@ next: ASSERT(em->disk_bytenr == EXTENT_MAP_INLINE); ASSERT(em->len == fs_info->sectorsize); - ret = read_inline_extent(inode, path, folio); + ret = read_inline_extent(path, folio); if (ret < 0) goto out; goto insert; @@ -7024,8 +7048,6 @@ static bool btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr) * @orig_start: (optional) Return the original file offset of the file extent * @orig_len: (optional) Return the original on-disk length of the file extent * @ram_bytes: (optional) Return the ram_bytes of the file extent - * @strict: if true, omit optimizations that might force us into unnecessary - * cow. e.g., don't trust generation number. * * Return: * >0 and update @len if we can do nocow write @@ -7037,7 +7059,7 @@ static bool btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr) */ noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len, struct btrfs_file_extent *file_extent, - bool nowait, bool strict) + bool nowait) { struct btrfs_fs_info *fs_info = inode_to_fs_info(inode); struct can_nocow_file_extent_args nocow_args = { 0 }; @@ -7090,7 +7112,6 @@ noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len, nocow_args.start = offset; nocow_args.end = offset + *len - 1; - nocow_args.strict = strict; nocow_args.free_path = true; ret = can_nocow_file_extent(path, &key, BTRFS_I(inode), &nocow_args); @@ -7297,7 +7318,7 @@ static void btrfs_invalidate_folio(struct folio *folio, size_t offset, * * But already submitted bio can still be finished on this folio. * Furthermore, endio function won't skip folio which has Ordered - * (Private2) already cleared, so it's possible for endio and + * already cleared, so it's possible for endio and * invalidate_folio to do the same ordered extent accounting twice * on one folio. * @@ -7363,7 +7384,7 @@ static void btrfs_invalidate_folio(struct folio *folio, size_t offset, range_len = range_end + 1 - cur; if (!btrfs_folio_test_ordered(fs_info, folio, cur, range_len)) { /* - * If Ordered (Private2) is cleared, it means endio has + * If Ordered is cleared, it means endio has * already been executed for the range. * We can't delete the extent states as * btrfs_finish_ordered_io() may still use some of them. @@ -7436,7 +7457,7 @@ next: } /* * We have iterated through all ordered extents of the page, the page - * should not have Ordered (Private2) anymore, or the above iteration + * should not have Ordered anymore, or the above iteration * did something wrong. */ ASSERT(!folio_test_ordered(folio)); @@ -8040,31 +8061,45 @@ static int btrfs_rename_exchange(struct inode *old_dir, /* src is a subvolume */ if (old_ino == BTRFS_FIRST_FREE_OBJECTID) { ret = btrfs_unlink_subvol(trans, BTRFS_I(old_dir), old_dentry); + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } } else { /* src is an inode */ ret = __btrfs_unlink_inode(trans, BTRFS_I(old_dir), BTRFS_I(old_dentry->d_inode), old_name, &old_rename_ctx); - if (!ret) - ret = btrfs_update_inode(trans, BTRFS_I(old_inode)); - } - if (ret) { - btrfs_abort_transaction(trans, ret); - goto out_fail; + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } + ret = btrfs_update_inode(trans, BTRFS_I(old_inode)); + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } } /* dest is a subvolume */ if (new_ino == BTRFS_FIRST_FREE_OBJECTID) { ret = btrfs_unlink_subvol(trans, BTRFS_I(new_dir), new_dentry); + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } } else { /* dest is an inode */ ret = __btrfs_unlink_inode(trans, BTRFS_I(new_dir), BTRFS_I(new_dentry->d_inode), new_name, &new_rename_ctx); - if (!ret) - ret = btrfs_update_inode(trans, BTRFS_I(new_inode)); - } - if (ret) { - btrfs_abort_transaction(trans, ret); - goto out_fail; + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } + ret = btrfs_update_inode(trans, BTRFS_I(new_inode)); + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } } ret = btrfs_add_link(trans, BTRFS_I(new_dir), BTRFS_I(old_inode), @@ -8300,16 +8335,23 @@ static int btrfs_rename(struct mnt_idmap *idmap, if (unlikely(old_ino == BTRFS_FIRST_FREE_OBJECTID)) { ret = btrfs_unlink_subvol(trans, BTRFS_I(old_dir), old_dentry); + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } } else { ret = __btrfs_unlink_inode(trans, BTRFS_I(old_dir), BTRFS_I(d_inode(old_dentry)), &old_fname.disk_name, &rename_ctx); - if (!ret) - ret = btrfs_update_inode(trans, BTRFS_I(old_inode)); - } - if (ret) { - btrfs_abort_transaction(trans, ret); - goto out_fail; + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } + ret = btrfs_update_inode(trans, BTRFS_I(old_inode)); + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } } if (new_inode) { @@ -8317,18 +8359,27 @@ static int btrfs_rename(struct mnt_idmap *idmap, if (unlikely(btrfs_ino(BTRFS_I(new_inode)) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) { ret = btrfs_unlink_subvol(trans, BTRFS_I(new_dir), new_dentry); + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } BUG_ON(new_inode->i_nlink == 0); } else { ret = btrfs_unlink_inode(trans, BTRFS_I(new_dir), BTRFS_I(d_inode(new_dentry)), &new_fname.disk_name); + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } } - if (!ret && new_inode->i_nlink == 0) + if (new_inode->i_nlink == 0) { ret = btrfs_orphan_add(trans, BTRFS_I(d_inode(new_dentry))); - if (ret) { - btrfs_abort_transaction(trans, ret); - goto out_fail; + if (ret) { + btrfs_abort_transaction(trans, ret); + goto out_fail; + } } } @@ -8668,7 +8719,6 @@ static int btrfs_symlink(struct mnt_idmap *idmap, struct inode *dir, ptr = btrfs_file_extent_inline_start(ei); write_extent_buffer(leaf, symname, ptr, name_len); - btrfs_mark_buffer_dirty(trans, leaf); btrfs_free_path(path); d_instantiate_new(dentry, inode); @@ -8975,28 +9025,6 @@ out_inode: return finish_open_simple(file, ret); } -void btrfs_set_range_writeback(struct btrfs_inode *inode, u64 start, u64 end) -{ - struct btrfs_fs_info *fs_info = inode->root->fs_info; - unsigned long index = start >> PAGE_SHIFT; - unsigned long end_index = end >> PAGE_SHIFT; - struct folio *folio; - u32 len; - - ASSERT(end + 1 - start <= U32_MAX); - len = end + 1 - start; - while (index <= end_index) { - folio = __filemap_get_folio(inode->vfs_inode.i_mapping, index, 0, 0); - ASSERT(!IS_ERR(folio)); /* folios should be in the extent_io_tree */ - - /* This is for data, which doesn't yet support larger folio. */ - ASSERT(folio_order(folio) == 0); - btrfs_folio_set_writeback(fs_info, folio, start, len); - folio_put(folio); - index++; - } -} - int btrfs_encoded_io_compression_from_extent(struct btrfs_fs_info *fs_info, int compress_type) { @@ -9041,12 +9069,16 @@ static ssize_t btrfs_encoded_read_inline( unsigned long ptr; void *tmp; ssize_t ret; + const bool nowait = (iocb->ki_flags & IOCB_NOWAIT); path = btrfs_alloc_path(); if (!path) { ret = -ENOMEM; goto out; } + + path->nowait = nowait; + ret = btrfs_lookup_file_extent(NULL, root, path, btrfs_ino(inode), extent_start, 0); if (ret) { @@ -9109,8 +9141,9 @@ out: } struct btrfs_encoded_read_private { - wait_queue_head_t wait; - atomic_t pending; + struct completion done; + void *uring_ctx; + refcount_t pending_refs; blk_status_t status; }; @@ -9129,26 +9162,40 @@ static void btrfs_encoded_read_endio(struct btrfs_bio *bbio) */ WRITE_ONCE(priv->status, bbio->bio.bi_status); } - if (!atomic_dec_return(&priv->pending)) - wake_up(&priv->wait); + if (refcount_dec_and_test(&priv->pending_refs)) { + int err = blk_status_to_errno(READ_ONCE(priv->status)); + + if (priv->uring_ctx) { + btrfs_uring_read_extent_endio(priv->uring_ctx, err); + kfree(priv); + } else { + complete(&priv->done); + } + } bio_put(&bbio->bio); } int btrfs_encoded_read_regular_fill_pages(struct btrfs_inode *inode, - u64 file_offset, u64 disk_bytenr, - u64 disk_io_size, struct page **pages) + u64 disk_bytenr, u64 disk_io_size, + struct page **pages, void *uring_ctx) { struct btrfs_fs_info *fs_info = inode->root->fs_info; - struct btrfs_encoded_read_private priv = { - .pending = ATOMIC_INIT(1), - }; + struct btrfs_encoded_read_private *priv; unsigned long i = 0; struct btrfs_bio *bbio; + int ret; + + priv = kmalloc(sizeof(struct btrfs_encoded_read_private), GFP_NOFS); + if (!priv) + return -ENOMEM; - init_waitqueue_head(&priv.wait); + init_completion(&priv->done); + refcount_set(&priv->pending_refs, 1); + priv->status = 0; + priv->uring_ctx = uring_ctx; bbio = btrfs_bio_alloc(BIO_MAX_VECS, REQ_OP_READ, fs_info, - btrfs_encoded_read_endio, &priv); + btrfs_encoded_read_endio, priv); bbio->bio.bi_iter.bi_sector = disk_bytenr >> SECTOR_SHIFT; bbio->inode = inode; @@ -9156,11 +9203,11 @@ int btrfs_encoded_read_regular_fill_pages(struct btrfs_inode *inode, size_t bytes = min_t(u64, disk_io_size, PAGE_SIZE); if (bio_add_page(&bbio->bio, pages[i], bytes, 0) < bytes) { - atomic_inc(&priv.pending); + refcount_inc(&priv->pending_refs); btrfs_submit_bbio(bbio, 0); bbio = btrfs_bio_alloc(BIO_MAX_VECS, REQ_OP_READ, fs_info, - btrfs_encoded_read_endio, &priv); + btrfs_encoded_read_endio, priv); bbio->bio.bi_iter.bi_sector = disk_bytenr >> SECTOR_SHIFT; bbio->inode = inode; continue; @@ -9171,22 +9218,33 @@ int btrfs_encoded_read_regular_fill_pages(struct btrfs_inode *inode, disk_io_size -= bytes; } while (disk_io_size); - atomic_inc(&priv.pending); + refcount_inc(&priv->pending_refs); btrfs_submit_bbio(bbio, 0); - if (atomic_dec_return(&priv.pending)) - io_wait_event(priv.wait, !atomic_read(&priv.pending)); - /* See btrfs_encoded_read_endio() for ordering. */ - return blk_status_to_errno(READ_ONCE(priv.status)); + if (uring_ctx) { + if (refcount_dec_and_test(&priv->pending_refs)) { + ret = blk_status_to_errno(READ_ONCE(priv->status)); + btrfs_uring_read_extent_endio(uring_ctx, ret); + kfree(priv); + return ret; + } + + return -EIOCBQUEUED; + } else { + if (!refcount_dec_and_test(&priv->pending_refs)) + wait_for_completion_io(&priv->done); + /* See btrfs_encoded_read_endio() for ordering. */ + ret = blk_status_to_errno(READ_ONCE(priv->status)); + kfree(priv); + return ret; + } } -static ssize_t btrfs_encoded_read_regular(struct kiocb *iocb, - struct iov_iter *iter, - u64 start, u64 lockend, - struct extent_state **cached_state, - u64 disk_bytenr, u64 disk_io_size, - size_t count, bool compressed, - bool *unlocked) +ssize_t btrfs_encoded_read_regular(struct kiocb *iocb, struct iov_iter *iter, + u64 start, u64 lockend, + struct extent_state **cached_state, + u64 disk_bytenr, u64 disk_io_size, + size_t count, bool compressed, bool *unlocked) { struct btrfs_inode *inode = BTRFS_I(file_inode(iocb->ki_filp)); struct extent_io_tree *io_tree = &inode->io_tree; @@ -9206,8 +9264,8 @@ static ssize_t btrfs_encoded_read_regular(struct kiocb *iocb, goto out; } - ret = btrfs_encoded_read_regular_fill_pages(inode, start, disk_bytenr, - disk_io_size, pages); + ret = btrfs_encoded_read_regular_fill_pages(inode, disk_bytenr, + disk_io_size, pages, NULL); if (ret) goto out; @@ -9247,21 +9305,26 @@ out: } ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter, - struct btrfs_ioctl_encoded_io_args *encoded) + struct btrfs_ioctl_encoded_io_args *encoded, + struct extent_state **cached_state, + u64 *disk_bytenr, u64 *disk_io_size) { struct btrfs_inode *inode = BTRFS_I(file_inode(iocb->ki_filp)); struct btrfs_fs_info *fs_info = inode->root->fs_info; struct extent_io_tree *io_tree = &inode->io_tree; ssize_t ret; size_t count = iov_iter_count(iter); - u64 start, lockend, disk_bytenr, disk_io_size; - struct extent_state *cached_state = NULL; + u64 start, lockend; struct extent_map *em; + const bool nowait = (iocb->ki_flags & IOCB_NOWAIT); bool unlocked = false; file_accessed(iocb->ki_filp); - btrfs_inode_lock(inode, BTRFS_ILOCK_SHARED); + ret = btrfs_inode_lock(inode, + BTRFS_ILOCK_SHARED | (nowait ? BTRFS_ILOCK_TRY : 0)); + if (ret) + return ret; if (iocb->ki_pos >= inode->vfs_inode.i_size) { btrfs_inode_unlock(inode, BTRFS_ILOCK_SHARED); @@ -9274,21 +9337,46 @@ ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter, */ lockend = start + BTRFS_MAX_UNCOMPRESSED - 1; - for (;;) { + if (nowait) { struct btrfs_ordered_extent *ordered; - ret = btrfs_wait_ordered_range(inode, start, - lockend - start + 1); - if (ret) + if (filemap_range_needs_writeback(inode->vfs_inode.i_mapping, + start, lockend)) { + ret = -EAGAIN; goto out_unlock_inode; - lock_extent(io_tree, start, lockend, &cached_state); + } + + if (!try_lock_extent(io_tree, start, lockend, cached_state)) { + ret = -EAGAIN; + goto out_unlock_inode; + } + ordered = btrfs_lookup_ordered_range(inode, start, lockend - start + 1); - if (!ordered) - break; - btrfs_put_ordered_extent(ordered); - unlock_extent(io_tree, start, lockend, &cached_state); - cond_resched(); + if (ordered) { + btrfs_put_ordered_extent(ordered); + unlock_extent(io_tree, start, lockend, cached_state); + ret = -EAGAIN; + goto out_unlock_inode; + } + } else { + for (;;) { + struct btrfs_ordered_extent *ordered; + + ret = btrfs_wait_ordered_range(inode, start, + lockend - start + 1); + if (ret) + goto out_unlock_inode; + + lock_extent(io_tree, start, lockend, cached_state); + ordered = btrfs_lookup_ordered_range(inode, start, + lockend - start + 1); + if (!ordered) + break; + btrfs_put_ordered_extent(ordered); + unlock_extent(io_tree, start, lockend, cached_state); + cond_resched(); + } } em = btrfs_get_extent(inode, NULL, start, lockend - start + 1); @@ -9307,9 +9395,9 @@ ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter, free_extent_map(em); em = NULL; ret = btrfs_encoded_read_inline(iocb, iter, start, lockend, - &cached_state, extent_start, + cached_state, extent_start, count, encoded, &unlocked); - goto out; + goto out_unlock_extent; } /* @@ -9320,12 +9408,12 @@ ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter, inode->vfs_inode.i_size) - iocb->ki_pos; if (em->disk_bytenr == EXTENT_MAP_HOLE || (em->flags & EXTENT_FLAG_PREALLOC)) { - disk_bytenr = EXTENT_MAP_HOLE; + *disk_bytenr = EXTENT_MAP_HOLE; count = min_t(u64, count, encoded->len); encoded->len = count; encoded->unencoded_len = count; } else if (extent_map_is_compressed(em)) { - disk_bytenr = em->disk_bytenr; + *disk_bytenr = em->disk_bytenr; /* * Bail if the buffer isn't large enough to return the whole * compressed extent. @@ -9334,7 +9422,7 @@ ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter, ret = -ENOBUFS; goto out_em; } - disk_io_size = em->disk_num_bytes; + *disk_io_size = em->disk_num_bytes; count = em->disk_num_bytes; encoded->unencoded_len = em->ram_bytes; encoded->unencoded_offset = iocb->ki_pos - (em->start - em->offset); @@ -9344,47 +9432,42 @@ ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter, goto out_em; encoded->compression = ret; } else { - disk_bytenr = extent_map_block_start(em) + (start - em->start); + *disk_bytenr = extent_map_block_start(em) + (start - em->start); if (encoded->len > count) encoded->len = count; /* * Don't read beyond what we locked. This also limits the page * allocations that we'll do. */ - disk_io_size = min(lockend + 1, iocb->ki_pos + encoded->len) - start; - count = start + disk_io_size - iocb->ki_pos; + *disk_io_size = min(lockend + 1, iocb->ki_pos + encoded->len) - start; + count = start + *disk_io_size - iocb->ki_pos; encoded->len = count; encoded->unencoded_len = count; - disk_io_size = ALIGN(disk_io_size, fs_info->sectorsize); + *disk_io_size = ALIGN(*disk_io_size, fs_info->sectorsize); } free_extent_map(em); em = NULL; - if (disk_bytenr == EXTENT_MAP_HOLE) { - unlock_extent(io_tree, start, lockend, &cached_state); + if (*disk_bytenr == EXTENT_MAP_HOLE) { + unlock_extent(io_tree, start, lockend, cached_state); btrfs_inode_unlock(inode, BTRFS_ILOCK_SHARED); unlocked = true; ret = iov_iter_zero(count, iter); if (ret != count) ret = -EFAULT; } else { - ret = btrfs_encoded_read_regular(iocb, iter, start, lockend, - &cached_state, disk_bytenr, - disk_io_size, count, - encoded->compression, - &unlocked); + ret = -EIOCBQUEUED; + goto out_unlock_extent; } -out: - if (ret >= 0) - iocb->ki_pos += encoded->len; out_em: free_extent_map(em); out_unlock_extent: - if (!unlocked) - unlock_extent(io_tree, start, lockend, &cached_state); + /* Leave inode and extent locked if we need to do a read. */ + if (!unlocked && ret != -EIOCBQUEUED) + unlock_extent(io_tree, start, lockend, cached_state); out_unlock_inode: - if (!unlocked) + if (!unlocked && ret != -EIOCBQUEUED) btrfs_inode_unlock(inode, BTRFS_ILOCK_SHARED); return ret; } @@ -9495,7 +9578,7 @@ ssize_t btrfs_do_encoded_write(struct kiocb *iocb, struct iov_iter *from, */ disk_num_bytes = ALIGN(orig_count, fs_info->sectorsize); nr_folios = DIV_ROUND_UP(disk_num_bytes, PAGE_SIZE); - folios = kvcalloc(nr_folios, sizeof(struct page *), GFP_KERNEL_ACCOUNT); + folios = kvcalloc(nr_folios, sizeof(struct folio *), GFP_KERNEL_ACCOUNT); if (!folios) return -ENOMEM; for (i = 0; i < nr_folios; i++) { @@ -9559,7 +9642,7 @@ ssize_t btrfs_do_encoded_write(struct kiocb *iocb, struct iov_iter *from, if (encoded->unencoded_len == encoded->len && encoded->unencoded_offset == 0 && can_cow_file_range_inline(inode, start, encoded->len, orig_count)) { - ret = __cow_file_range_inline(inode, start, encoded->len, + ret = __cow_file_range_inline(inode, encoded->len, orig_count, compression, folios[0], true); if (ret <= 0) { @@ -9779,15 +9862,25 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, struct btrfs_fs_info *fs_info = root->fs_info; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_state *cached_state = NULL; - struct extent_map *em = NULL; struct btrfs_chunk_map *map = NULL; struct btrfs_device *device = NULL; struct btrfs_swap_info bsi = { .lowest_ppage = (sector_t)-1ULL, }; + struct btrfs_backref_share_check_ctx *backref_ctx = NULL; + struct btrfs_path *path = NULL; int ret = 0; u64 isize; - u64 start; + u64 prev_extent_end = 0; + + /* + * Acquire the inode's mmap lock to prevent races with memory mapped + * writes, as they could happen after we flush delalloc below and before + * we lock the extent range further below. The inode was already locked + * up in the call chain. + */ + btrfs_assert_inode_locked(BTRFS_I(inode)); + down_write(&BTRFS_I(inode)->i_mmap_lock); /* * If the swap file was just created, make sure delalloc is done. If the @@ -9796,22 +9889,32 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, */ ret = btrfs_wait_ordered_range(BTRFS_I(inode), 0, (u64)-1); if (ret) - return ret; + goto out_unlock_mmap; /* * The inode is locked, so these flags won't change after we check them. */ if (BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS) { btrfs_warn(fs_info, "swapfile must not be compressed"); - return -EINVAL; + ret = -EINVAL; + goto out_unlock_mmap; } if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW)) { btrfs_warn(fs_info, "swapfile must not be copy-on-write"); - return -EINVAL; + ret = -EINVAL; + goto out_unlock_mmap; } if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) { btrfs_warn(fs_info, "swapfile must not be checksummed"); - return -EINVAL; + ret = -EINVAL; + goto out_unlock_mmap; + } + + path = btrfs_alloc_path(); + backref_ctx = btrfs_alloc_backref_share_check_ctx(); + if (!path || !backref_ctx) { + ret = -ENOMEM; + goto out_unlock_mmap; } /* @@ -9826,7 +9929,8 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_SWAP_ACTIVATE)) { btrfs_warn(fs_info, "cannot activate swapfile while exclusive operation is running"); - return -EBUSY; + ret = -EBUSY; + goto out_unlock_mmap; } /* @@ -9840,7 +9944,8 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, btrfs_exclop_finish(fs_info); btrfs_warn(fs_info, "cannot activate swapfile because snapshot creation is in progress"); - return -EINVAL; + ret = -EINVAL; + goto out_unlock_mmap; } /* * Snapshots can create extents which require COW even if NODATACOW is @@ -9856,11 +9961,13 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, if (btrfs_root_dead(root)) { spin_unlock(&root->root_item_lock); + btrfs_drew_write_unlock(&root->snapshot_lock); btrfs_exclop_finish(fs_info); btrfs_warn(fs_info, "cannot activate swapfile because subvolume %llu is being deleted", btrfs_root_id(root)); - return -EPERM; + ret = -EPERM; + goto out_unlock_mmap; } atomic_inc(&root->nr_swapfiles); spin_unlock(&root->root_item_lock); @@ -9868,24 +9975,39 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, isize = ALIGN_DOWN(inode->i_size, fs_info->sectorsize); lock_extent(io_tree, 0, isize - 1, &cached_state); - start = 0; - while (start < isize) { - u64 logical_block_start, physical_block_start; + while (prev_extent_end < isize) { + struct btrfs_key key; + struct extent_buffer *leaf; + struct btrfs_file_extent_item *ei; struct btrfs_block_group *bg; - u64 len = isize - start; + u64 logical_block_start; + u64 physical_block_start; + u64 extent_gen; + u64 disk_bytenr; + u64 len; - em = btrfs_get_extent(BTRFS_I(inode), NULL, start, len); - if (IS_ERR(em)) { - ret = PTR_ERR(em); + key.objectid = btrfs_ino(BTRFS_I(inode)); + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = prev_extent_end; + + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) goto out; - } - if (em->disk_bytenr == EXTENT_MAP_HOLE) { + /* + * If key not found it means we have an implicit hole (NO_HOLES + * is enabled). + */ + if (ret > 0) { btrfs_warn(fs_info, "swapfile must not have holes"); ret = -EINVAL; goto out; } - if (em->disk_bytenr == EXTENT_MAP_INLINE) { + + leaf = path->nodes[0]; + ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); + + if (btrfs_file_extent_type(leaf, ei) == BTRFS_FILE_EXTENT_INLINE) { /* * It's unlikely we'll ever actually find ourselves * here, as a file small enough to fit inline won't be @@ -9897,23 +10019,45 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, ret = -EINVAL; goto out; } - if (extent_map_is_compressed(em)) { + + if (btrfs_file_extent_compression(leaf, ei) != BTRFS_COMPRESS_NONE) { btrfs_warn(fs_info, "swapfile must not be compressed"); ret = -EINVAL; goto out; } - logical_block_start = extent_map_block_start(em) + (start - em->start); - len = min(len, em->len - (start - em->start)); - free_extent_map(em); - em = NULL; + disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, ei); + if (disk_bytenr == 0) { + btrfs_warn(fs_info, "swapfile must not have holes"); + ret = -EINVAL; + goto out; + } + + logical_block_start = disk_bytenr + btrfs_file_extent_offset(leaf, ei); + extent_gen = btrfs_file_extent_generation(leaf, ei); + prev_extent_end = btrfs_file_extent_end(path); + + if (prev_extent_end > isize) + len = isize - key.offset; + else + len = btrfs_file_extent_num_bytes(leaf, ei); - ret = can_nocow_extent(inode, start, &len, NULL, false, true); + backref_ctx->curr_leaf_bytenr = leaf->start; + + /* + * Don't need the path anymore, release to avoid deadlocks when + * calling btrfs_is_data_extent_shared() because when joining a + * transaction it can block waiting for the current one's commit + * which in turn may be trying to lock the same leaf to flush + * delayed items for example. + */ + btrfs_release_path(path); + + ret = btrfs_is_data_extent_shared(BTRFS_I(inode), disk_bytenr, + extent_gen, backref_ctx); if (ret < 0) { goto out; - } else if (ret) { - ret = 0; - } else { + } else if (ret > 0) { btrfs_warn(fs_info, "swapfile must not be copy-on-write"); ret = -EINVAL; @@ -9948,7 +10092,6 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, physical_block_start = (map->stripes[0].physical + (logical_block_start - map->start)); - len = min(len, map->chunk_len - (logical_block_start - map->start)); btrfs_free_chunk_map(map); map = NULL; @@ -9989,20 +10132,23 @@ static int btrfs_swap_activate(struct swap_info_struct *sis, struct file *file, if (ret) goto out; } - bsi.start = start; + bsi.start = key.offset; bsi.block_start = physical_block_start; bsi.block_len = len; } - start += len; + if (fatal_signal_pending(current)) { + ret = -EINTR; + goto out; + } + + cond_resched(); } if (bsi.block_len) ret = btrfs_add_swap_extent(sis, &bsi); out: - if (!IS_ERR_OR_NULL(em)) - free_extent_map(em); if (!IS_ERR_OR_NULL(map)) btrfs_free_chunk_map(map); @@ -10015,6 +10161,10 @@ out: btrfs_exclop_finish(fs_info); +out_unlock_mmap: + up_write(&BTRFS_I(inode)->i_mmap_lock); + btrfs_free_backref_share_ctx(backref_ctx); + btrfs_free_path(path); if (ret) return ret; |