X-Git-Url: http://ftp.safe.ca/?a=blobdiff_plain;f=fs%2Fbtrfs%2Finode.c;h=ec5423790bbbec9a39bdc4528a069d3b54738456;hb=c2ec175c39f62949438354f603f4aa170846aabb;hp=33b990878d84f40a2095786c4b9b04f620431867;hpb=db69e0ebae944690de89851315404f483e6464e0;p=safe%2Fjmp%2Flinux-2.6 diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 33b9908..ec54237 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -34,9 +34,10 @@ #include #include #include -#include #include #include +#include +#include "compat.h" #include "ctree.h" #include "disk-io.h" #include "transaction.h" @@ -45,6 +46,11 @@ #include "print-tree.h" #include "volumes.h" #include "ordered-data.h" +#include "xattr.h" +#include "tree-log.h" +#include "ref-cache.h" +#include "compression.h" +#include "locking.h" struct btrfs_iget_args { u64 ino; @@ -79,41 +85,612 @@ static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = { }; static void btrfs_truncate(struct inode *inode); +static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end); +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); -int btrfs_check_free_space(struct btrfs_root *root, u64 num_required, - int for_del) +static int btrfs_init_inode_security(struct inode *inode, struct inode *dir) { - u64 total; - u64 used; - u64 thresh; - unsigned long flags; + int err; + + err = btrfs_init_acl(inode, dir); + if (!err) + err = btrfs_xattr_security_init(inode, dir); + return err; +} + +/* + * this does all the hard work for inserting an inline extent into + * 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, + struct btrfs_root *root, struct inode *inode, + u64 start, size_t size, size_t compressed_size, + struct page **compressed_pages) +{ + struct btrfs_key key; + struct btrfs_path *path; + struct extent_buffer *leaf; + struct page *page = NULL; + char *kaddr; + unsigned long ptr; + struct btrfs_file_extent_item *ei; + 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; + cur_size = compressed_size; + } + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + btrfs_set_trans_block_group(trans, inode); + + 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); + + inode_add_bytes(inode, size); + ret = btrfs_insert_empty_item(trans, root, path, &key, + datasize); + BUG_ON(ret); + if (ret) { + err = ret; + goto fail; + } + leaf = path->nodes[0]; + ei = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_generation(leaf, ei, trans->transid); + btrfs_set_file_extent_type(leaf, ei, BTRFS_FILE_EXTENT_INLINE); + btrfs_set_file_extent_encryption(leaf, ei, 0); + btrfs_set_file_extent_other_encoding(leaf, ei, 0); + btrfs_set_file_extent_ram_bytes(leaf, ei, size); + ptr = btrfs_file_extent_inline_start(ei); + + if (use_compress) { + struct page *cpage; + int i = 0; + while (compressed_size > 0) { + cpage = compressed_pages[i]; + cur_size = min_t(unsigned long, compressed_size, + PAGE_CACHE_SIZE); + + kaddr = kmap(cpage); + write_extent_buffer(leaf, kaddr, ptr, cur_size); + kunmap(cpage); + + i++; + ptr += cur_size; + compressed_size -= cur_size; + } + btrfs_set_file_extent_compression(leaf, ei, + BTRFS_COMPRESS_ZLIB); + } 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); + offset = start & (PAGE_CACHE_SIZE - 1); + write_extent_buffer(leaf, kaddr + offset, ptr, size); + kunmap_atomic(kaddr, KM_USER0); + page_cache_release(page); + } + btrfs_mark_buffer_dirty(leaf); + btrfs_free_path(path); + + BTRFS_I(inode)->disk_i_size = inode->i_size; + btrfs_update_inode(trans, root, inode); + return 0; +fail: + btrfs_free_path(path); + return err; +} + + +/* + * conditionally insert an inline extent into the file. This + * does the checks required to make sure the data is small enough + * to fit as an inline extent. + */ +static 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) +{ + 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 data_len = inline_len; + int ret; + + if (compressed_size) + data_len = compressed_size; + + if (start > 0 || + actual_end >= PAGE_CACHE_SIZE || + data_len >= BTRFS_MAX_INLINE_DATA_SIZE(root) || + (!compressed_size && + (actual_end & (root->sectorsize - 1)) == 0) || + end + 1 < isize || + data_len > root->fs_info->max_inline) { + return 1; + } + + ret = btrfs_drop_extents(trans, root, inode, start, + aligned_end, start, &hint_byte); + BUG_ON(ret); + + if (isize > actual_end) + inline_len = min_t(u64, isize, actual_end); + ret = insert_inline_extent(trans, root, inode, start, + inline_len, compressed_size, + compressed_pages); + BUG_ON(ret); + btrfs_drop_extent_cache(inode, start, aligned_end, 0); + return 0; +} + +struct async_extent { + u64 start; + u64 ram_size; + u64 compressed_size; + struct page **pages; + unsigned long nr_pages; + struct list_head list; +}; + +struct async_cow { + struct inode *inode; + struct btrfs_root *root; + struct page *locked_page; + u64 start; + u64 end; + struct list_head extents; + struct btrfs_work work; +}; + +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) +{ + struct async_extent *async_extent; + + async_extent = kmalloc(sizeof(*async_extent), GFP_NOFS); + 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; + list_add_tail(&async_extent->list, &cow->extents); + return 0; +} + +/* + * we create compressed extents in two phases. The first + * phase compresses a range of pages that have already been + * locked (both pages and state bits are locked). + * + * This is done inside an ordered work queue, and the compression + * is spread across many cpus. The actual IO submission is step + * two, and the ordered work queue takes care of making sure that + * happens in the same order things were put onto the queue by + * writepages and friends. + * + * 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. + */ +static noinline int compress_file_range(struct inode *inode, + struct page *locked_page, + u64 start, u64 end, + struct async_cow *async_cow, + int *num_added) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_trans_handle *trans; + u64 num_bytes; + u64 orig_start; + u64 disk_num_bytes; + u64 blocksize = root->sectorsize; + u64 actual_end; + u64 isize = i_size_read(inode); int ret = 0; + struct page **pages = NULL; + unsigned long nr_pages; + unsigned long nr_pages_ret = 0; + unsigned long total_compressed = 0; + unsigned long total_in = 0; + unsigned long max_compressed = 128 * 1024; + unsigned long max_uncompressed = 128 * 1024; + int i; + int will_compress; + + orig_start = start; + + actual_end = min_t(u64, isize, end + 1); +again: + will_compress = 0; + nr_pages = (end >> PAGE_CACHE_SHIFT) - (start >> PAGE_CACHE_SHIFT) + 1; + nr_pages = min(nr_pages, (128 * 1024UL) / PAGE_CACHE_SIZE); - spin_lock_irqsave(&root->fs_info->delalloc_lock, flags); - total = btrfs_super_total_bytes(&root->fs_info->super_copy); - used = btrfs_super_bytes_used(&root->fs_info->super_copy); - if (for_del) - thresh = total * 90; - else - thresh = total * 85; + /* + * we don't want to send crud past the end of i_size through + * compression, that's just a waste of CPU time. So, if the + * end of the file is before the start of our current + * requested range of bytes, we bail out to the uncompressed + * cleanup code that can deal with all of this. + * + * It isn't really the fastest way to fix things, but this is a + * very uncommon corner. + */ + if (actual_end <= start) + goto cleanup_and_bail_uncompressed; + + total_compressed = actual_end - start; + + /* we want to make sure that amount of ram required to uncompress + * an extent is reasonable, so we limit the total size in ram + * of a compressed extent to 128k. This is a crucial number + * because it also controls how easily we can spread reads across + * cpus for decompression. + * + * We also want to make sure the amount of IO required to do + * a random read is reasonably small, so we limit the size of + * a compressed extent to 128k. + */ + total_compressed = min(total_compressed, max_uncompressed); + num_bytes = (end - start + blocksize) & ~(blocksize - 1); + num_bytes = max(blocksize, num_bytes); + disk_num_bytes = num_bytes; + total_in = 0; + ret = 0; + + /* + * we do compression for mount -o compress and when the + * inode has not been flagged as nocompress. This flag can + * change at any time if we discover bad compression ratios. + */ + if (!btrfs_test_flag(inode, NOCOMPRESS) && + btrfs_test_opt(root, COMPRESS)) { + WARN_ON(pages); + pages = kzalloc(sizeof(struct page *) * nr_pages, GFP_NOFS); + + ret = btrfs_zlib_compress_pages(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 & + (PAGE_CACHE_SIZE - 1); + struct page *page = pages[nr_pages_ret - 1]; + char *kaddr; + + /* zero the tail end of the last page, we might be + * sending it down to disk + */ + if (offset) { + kaddr = kmap_atomic(page, KM_USER0); + memset(kaddr + offset, 0, + PAGE_CACHE_SIZE - offset); + kunmap_atomic(kaddr, KM_USER0); + } + will_compress = 1; + } + } + if (start == 0) { + trans = btrfs_join_transaction(root, 1); + BUG_ON(!trans); + btrfs_set_trans_block_group(trans, inode); - do_div(thresh, 100); + /* 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); + } else { + /* try making a compressed inline extent */ + ret = cow_file_range_inline(trans, root, inode, + start, end, + total_compressed, pages); + } + btrfs_end_transaction(trans, root); + if (ret == 0) { + /* + * inline extent creation worked, 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, 1, 0, + 0, 1, 1, 1); + ret = 0; + goto free_pages_out; + } + } - if (used + root->fs_info->delalloc_bytes + num_required > thresh) - ret = -ENOSPC; - spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags); - return ret; + if (will_compress) { + /* + * we aren't doing an inline extent round the compressed size + * up to a block size boundary so the allocator does sane + * things + */ + total_compressed = (total_compressed + blocksize - 1) & + ~(blocksize - 1); + + /* + * 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); + if (total_compressed >= total_in) { + will_compress = 0; + } else { + disk_num_bytes = total_compressed; + num_bytes = total_in; + } + } + if (!will_compress && pages) { + /* + * the compression code ran but failed to make things smaller, + * free any pages it allocated and our page pointer array + */ + for (i = 0; i < nr_pages_ret; i++) { + WARN_ON(pages[i]->mapping); + page_cache_release(pages[i]); + } + kfree(pages); + pages = NULL; + total_compressed = 0; + nr_pages_ret = 0; + + /* flag the file so we don't compress in the future */ + btrfs_set_flag(inode, NOCOMPRESS); + } + if (will_compress) { + *num_added += 1; + + /* the async work queues will take care of doing actual + * allocation on disk for these compressed pages, + * and will submit them to the elevator. + */ + add_async_extent(async_cow, start, num_bytes, + total_compressed, pages, nr_pages_ret); + + if (start + num_bytes < end && start + num_bytes < actual_end) { + start += num_bytes; + pages = NULL; + cond_resched(); + goto again; + } + } else { +cleanup_and_bail_uncompressed: + /* + * No compression, but we still need to write the pages in + * the file we've been given so far. redirty the locked + * page if it corresponds to our extent and set things up + * for the async work queue to run cow_file_range to do + * the normal delalloc dance + */ + if (page_offset(locked_page) >= start && + page_offset(locked_page) <= end) { + __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); + *num_added += 1; + } + +out: + return 0; + +free_pages_out: + for (i = 0; i < nr_pages_ret; i++) { + WARN_ON(pages[i]->mapping); + page_cache_release(pages[i]); + } + kfree(pages); + + goto out; +} + +/* + * phase two of compressed writeback. This is the ordered portion + * of the code, which only gets called in the order the work was + * queued. We walk all the async extents created by compress_file_range + * and send them down to the disk. + */ +static noinline int submit_compressed_extents(struct inode *inode, + struct async_cow *async_cow) +{ + 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; + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_io_tree *io_tree; + int ret; + + if (list_empty(&async_cow->extents)) + return 0; + + trans = btrfs_join_transaction(root, 1); + + while (!list_empty(&async_cow->extents)) { + async_extent = list_entry(async_cow->extents.next, + struct async_extent, list); + list_del(&async_extent->list); + + io_tree = &BTRFS_I(inode)->io_tree; + + /* did the compression code fall back to uncompressed IO? */ + if (!async_extent->pages) { + int page_started = 0; + unsigned long nr_written = 0; + + lock_extent(io_tree, async_extent->start, + async_extent->start + + async_extent->ram_size - 1, GFP_NOFS); + + /* allocate blocks */ + cow_file_range(inode, async_cow->locked_page, + async_extent->start, + async_extent->start + + async_extent->ram_size - 1, + &page_started, &nr_written, 0); + + /* + * if page_started, cow_file_range inserted an + * inline extent and took care of all the unlocking + * and IO for us. Otherwise, we need to submit + * all those pages down to the drive. + */ + if (!page_started) + extent_write_locked_range(io_tree, + inode, async_extent->start, + async_extent->start + + async_extent->ram_size - 1, + btrfs_get_extent, + WB_SYNC_ALL); + kfree(async_extent); + cond_resched(); + continue; + } + + lock_extent(io_tree, async_extent->start, + async_extent->start + async_extent->ram_size - 1, + GFP_NOFS); + /* + * here we're doing allocation and writeback of the + * compressed pages + */ + btrfs_drop_extent_cache(inode, async_extent->start, + async_extent->start + + async_extent->ram_size - 1, 0); + + ret = btrfs_reserve_extent(trans, root, + async_extent->compressed_size, + async_extent->compressed_size, + 0, alloc_hint, + (u64)-1, &ins, 1); + BUG_ON(ret); + em = alloc_extent_map(GFP_NOFS); + em->start = async_extent->start; + em->len = async_extent->ram_size; + em->orig_start = em->start; + + 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); + set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); + + while (1) { + spin_lock(&em_tree->lock); + ret = add_extent_mapping(em_tree, em); + spin_unlock(&em_tree->lock); + if (ret != -EEXIST) { + free_extent_map(em); + break; + } + btrfs_drop_extent_cache(inode, async_extent->start, + async_extent->start + + 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); + + btrfs_end_transaction(trans, root); + + /* + * clear dirty, set writeback and unlock the pages. + */ + extent_clear_unlock_delalloc(inode, + &BTRFS_I(inode)->io_tree, + async_extent->start, + async_extent->start + + async_extent->ram_size - 1, + NULL, 1, 1, 0, 1, 1, 0); + + 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); + trans = btrfs_join_transaction(root, 1); + alloc_hint = ins.objectid + ins.offset; + kfree(async_extent); + cond_resched(); + } + + btrfs_end_transaction(trans, root); + return 0; } -static int cow_file_range(struct inode *inode, u64 start, u64 end) +/* + * when extent_io.c finds a delayed allocation range in the file, + * the call backs end up in this code. The basic idea is to + * allocate extents on disk for the range, and create ordered data structs + * in ram to track those extents. + * + * locked_page is the page that writepage had locked already. We use + * it to make sure we don't do extra locks or unlocks. + * + * *page_started is set to one if we unlock locked_page and do everything + * required to start IO on it. It may be clean and already done with + * IO when we return. + */ +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) { struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans; u64 alloc_hint = 0; u64 num_bytes; + unsigned long ram_size; + u64 disk_num_bytes; u64 cur_alloc_size; u64 blocksize = root->sectorsize; - u64 orig_num_bytes; + u64 actual_end; + u64 isize = i_size_read(inode); struct btrfs_key ins; struct extent_map *em; struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; @@ -123,35 +700,55 @@ static int cow_file_range(struct inode *inode, u64 start, u64 end) BUG_ON(!trans); btrfs_set_trans_block_group(trans, inode); + actual_end = min_t(u64, isize, end + 1); + num_bytes = (end - start + blocksize) & ~(blocksize - 1); num_bytes = max(blocksize, num_bytes); - orig_num_bytes = num_bytes; + disk_num_bytes = num_bytes; + ret = 0; - if (alloc_hint == EXTENT_MAP_INLINE) - goto out; + if (start == 0) { + /* lets try to make an inline extent */ + ret = cow_file_range_inline(trans, root, inode, + start, end, 0, NULL); + if (ret == 0) { + extent_clear_unlock_delalloc(inode, + &BTRFS_I(inode)->io_tree, + start, end, NULL, 1, 1, + 1, 1, 1, 1); + *nr_written = *nr_written + + (end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE; + *page_started = 1; + ret = 0; + goto out; + } + } - BUG_ON(num_bytes > btrfs_super_total_bytes(&root->fs_info->super_copy)); - mutex_lock(&BTRFS_I(inode)->extent_mutex); - btrfs_drop_extent_cache(inode, start, start + num_bytes - 1); - mutex_unlock(&BTRFS_I(inode)->extent_mutex); + BUG_ON(disk_num_bytes > + btrfs_super_total_bytes(&root->fs_info->super_copy)); - while(num_bytes > 0) { - cur_alloc_size = min(num_bytes, root->fs_info->max_extent); + btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0); + + while (disk_num_bytes > 0) { + cur_alloc_size = min(disk_num_bytes, root->fs_info->max_extent); ret = btrfs_reserve_extent(trans, root, cur_alloc_size, - root->sectorsize, 0, 0, + root->sectorsize, 0, alloc_hint, (u64)-1, &ins, 1); - if (ret) { - WARN_ON(1); - goto out; - } + BUG_ON(ret); + em = alloc_extent_map(GFP_NOFS); em->start = start; + em->orig_start = em->start; + + ram_size = ins.offset; em->len = ins.offset; + em->block_start = ins.objectid; + em->block_len = ins.offset; em->bdev = root->fs_info->fs_devices->latest_bdev; - mutex_lock(&BTRFS_I(inode)->extent_mutex); set_bit(EXTENT_FLAG_PINNED, &em->flags); - while(1) { + + while (1) { spin_lock(&em_tree->lock); ret = add_extent_mapping(em_tree, em); spin_unlock(&em_tree->lock); @@ -160,180 +757,451 @@ static int cow_file_range(struct inode *inode, u64 start, u64 end) break; } btrfs_drop_extent_cache(inode, start, - start + ins.offset - 1); + start + ram_size - 1, 0); } - mutex_unlock(&BTRFS_I(inode)->extent_mutex); cur_alloc_size = ins.offset; ret = btrfs_add_ordered_extent(inode, start, ins.objectid, - ins.offset, 0); + ram_size, cur_alloc_size, 0); BUG_ON(ret); - if (num_bytes < cur_alloc_size) { - printk("num_bytes %Lu cur_alloc %Lu\n", num_bytes, - cur_alloc_size); - break; + + if (root->root_key.objectid == + BTRFS_DATA_RELOC_TREE_OBJECTID) { + ret = btrfs_reloc_clone_csums(inode, start, + cur_alloc_size); + BUG_ON(ret); } + + if (disk_num_bytes < cur_alloc_size) + break; + + /* 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 + */ + extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree, + start, start + ram_size - 1, + locked_page, unlock, 1, + 1, 0, 0, 0); + 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; } -static int run_delalloc_nocow(struct inode *inode, u64 start, u64 end) +/* + * work queue call back to started compression on a file and pages + */ +static noinline void async_cow_start(struct btrfs_work *work) +{ + struct async_cow *async_cow; + int num_added = 0; + async_cow = container_of(work, struct async_cow, 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) + async_cow->inode = NULL; +} + +/* + * work queue call back to submit previously compressed pages + */ +static noinline void async_cow_submit(struct btrfs_work *work) +{ + struct async_cow *async_cow; + struct btrfs_root *root; + unsigned long nr_pages; + + async_cow = container_of(work, struct async_cow, work); + + root = async_cow->root; + 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 && + waitqueue_active(&root->fs_info->async_submit_wait)) + wake_up(&root->fs_info->async_submit_wait); + + if (async_cow->inode) + submit_compressed_extents(async_cow->inode, async_cow); +} + +static noinline void async_cow_free(struct btrfs_work *work) +{ + struct async_cow *async_cow; + async_cow = container_of(work, struct async_cow, work); + kfree(async_cow); +} + +static int cow_file_range_async(struct inode *inode, struct page *locked_page, + u64 start, u64 end, int *page_started, + unsigned long *nr_written) +{ + struct async_cow *async_cow; + struct btrfs_root *root = BTRFS_I(inode)->root; + unsigned long nr_pages; + u64 cur_end; + int limit = 10 * 1024 * 1042; + + if (!btrfs_test_opt(root, COMPRESS)) { + return cow_file_range(inode, locked_page, start, end, + page_started, nr_written, 1); + } + + clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED | + EXTENT_DELALLOC, 1, 0, GFP_NOFS); + while (start < end) { + async_cow = kmalloc(sizeof(*async_cow), GFP_NOFS); + async_cow->inode = inode; + async_cow->root = root; + async_cow->locked_page = locked_page; + async_cow->start = start; + + if (btrfs_test_flag(inode, NOCOMPRESS)) + cur_end = end; + else + cur_end = min(end, start + 512 * 1024 - 1); + + 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; + + 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); + + if (atomic_read(&root->fs_info->async_delalloc_pages) > limit) { + wait_event(root->fs_info->async_submit_wait, + (atomic_read(&root->fs_info->async_delalloc_pages) < + limit)); + } + + while (atomic_read(&root->fs_info->async_submit_draining) && + atomic_read(&root->fs_info->async_delalloc_pages)) { + wait_event(root->fs_info->async_submit_wait, + (atomic_read(&root->fs_info->async_delalloc_pages) == + 0)); + } + + *nr_written += nr_pages; + start = cur_end + 1; + } + *page_started = 1; + return 0; +} + +static noinline int csum_exist_in_range(struct btrfs_root *root, + u64 bytenr, u64 num_bytes) +{ + int ret; + struct btrfs_ordered_sum *sums; + LIST_HEAD(list); + + ret = btrfs_lookup_csums_range(root->fs_info->csum_root, bytenr, + bytenr + num_bytes - 1, &list); + if (ret == 0 && list_empty(&list)) + return 0; + + while (!list_empty(&list)) { + sums = list_entry(list.next, struct btrfs_ordered_sum, list); + list_del(&sums->list); + kfree(sums); + } + return 1; +} + +/* + * 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. + * + * If no cow copies or snapshots exist, we write directly to the existing + * blocks on disk + */ +static int run_delalloc_nocow(struct inode *inode, struct page *locked_page, + u64 start, u64 end, int *page_started, int force, + unsigned long *nr_written) { - u64 extent_start; - u64 extent_end; - u64 bytenr; - u64 loops = 0; - u64 total_fs_bytes; struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_block_group_cache *block_group; struct btrfs_trans_handle *trans; struct extent_buffer *leaf; - int found_type; struct btrfs_path *path; - struct btrfs_file_extent_item *item; - int ret; - int err = 0; + struct btrfs_file_extent_item *fi; struct btrfs_key found_key; + u64 cow_start; + u64 cur_offset; + u64 extent_end; + u64 disk_bytenr; + u64 num_bytes; + int extent_type; + int ret; + int type; + int nocow; + int check_prev = 1; - total_fs_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy); path = btrfs_alloc_path(); BUG_ON(!path); trans = btrfs_join_transaction(root, 1); BUG_ON(!trans); -again: - ret = btrfs_lookup_file_extent(NULL, root, path, - inode->i_ino, start, 0); - if (ret < 0) { - err = ret; - goto out; - } - if (ret != 0) { - if (path->slots[0] == 0) - goto not_found; - path->slots[0]--; - } + cow_start = (u64)-1; + cur_offset = start; + while (1) { + ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino, + cur_offset, 0); + BUG_ON(ret < 0); + 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 && + found_key.type == BTRFS_EXTENT_DATA_KEY) + path->slots[0]--; + } + check_prev = 0; +next_slot: + leaf = path->nodes[0]; + if (path->slots[0] >= btrfs_header_nritems(leaf)) { + ret = btrfs_next_leaf(root, path); + if (ret < 0) + BUG_ON(1); + if (ret > 0) + break; + leaf = path->nodes[0]; + } - leaf = path->nodes[0]; - item = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_file_extent_item); + nocow = 0; + disk_bytenr = 0; + num_bytes = 0; + btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - /* are we inside the extent that was found? */ - btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - found_type = btrfs_key_type(&found_key); - if (found_key.objectid != inode->i_ino || - found_type != BTRFS_EXTENT_DATA_KEY) - goto not_found; + if (found_key.objectid > inode->i_ino || + found_key.type > BTRFS_EXTENT_DATA_KEY || + found_key.offset > end) + break; - found_type = btrfs_file_extent_type(leaf, item); - extent_start = found_key.offset; - if (found_type == BTRFS_FILE_EXTENT_REG) { - u64 extent_num_bytes; + if (found_key.offset > cur_offset) { + extent_end = found_key.offset; + goto out_check; + } - extent_num_bytes = btrfs_file_extent_num_bytes(leaf, item); - extent_end = extent_start + extent_num_bytes; - err = 0; + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + extent_type = btrfs_file_extent_type(leaf, fi); - if (loops && start != extent_start) - goto not_found; + if (extent_type == BTRFS_FILE_EXTENT_REG || + extent_type == BTRFS_FILE_EXTENT_PREALLOC) { + disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); + extent_end = found_key.offset + + btrfs_file_extent_num_bytes(leaf, fi); + if (extent_end <= start) { + path->slots[0]++; + goto next_slot; + } + if (disk_bytenr == 0) + goto out_check; + if (btrfs_file_extent_compression(leaf, fi) || + btrfs_file_extent_encryption(leaf, fi) || + btrfs_file_extent_other_encoding(leaf, fi)) + goto out_check; + if (extent_type == BTRFS_FILE_EXTENT_REG && !force) + goto out_check; + if (btrfs_extent_readonly(root, disk_bytenr)) + goto out_check; + if (btrfs_cross_ref_exist(trans, root, inode->i_ino, + disk_bytenr)) + goto out_check; + disk_bytenr += btrfs_file_extent_offset(leaf, fi); + disk_bytenr += cur_offset - found_key.offset; + num_bytes = min(end + 1, extent_end) - cur_offset; + /* + * force cow if csum exists in the range. + * this ensure that csum for a given extent are + * either valid or do not exist. + */ + if (csum_exist_in_range(root, disk_bytenr, num_bytes)) + goto out_check; + nocow = 1; + } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { + extent_end = found_key.offset + + btrfs_file_extent_inline_len(leaf, fi); + extent_end = ALIGN(extent_end, root->sectorsize); + } else { + BUG_ON(1); + } +out_check: + if (extent_end <= start) { + path->slots[0]++; + goto next_slot; + } + if (!nocow) { + if (cow_start == (u64)-1) + cow_start = cur_offset; + cur_offset = extent_end; + if (cur_offset > end) + break; + path->slots[0]++; + goto next_slot; + } - if (start < extent_start || start >= extent_end) - goto not_found; + btrfs_release_path(root, 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); + cow_start = (u64)-1; + } - bytenr = btrfs_file_extent_disk_bytenr(leaf, item); - if (bytenr == 0) - goto not_found; + if (extent_type == BTRFS_FILE_EXTENT_PREALLOC) { + struct extent_map *em; + struct extent_map_tree *em_tree; + em_tree = &BTRFS_I(inode)->extent_tree; + em = alloc_extent_map(GFP_NOFS); + em->start = cur_offset; + em->orig_start = em->start; + em->len = num_bytes; + em->block_len = num_bytes; + em->block_start = disk_bytenr; + em->bdev = root->fs_info->fs_devices->latest_bdev; + set_bit(EXTENT_FLAG_PINNED, &em->flags); + while (1) { + spin_lock(&em_tree->lock); + ret = add_extent_mapping(em_tree, em); + spin_unlock(&em_tree->lock); + if (ret != -EEXIST) { + free_extent_map(em); + break; + } + btrfs_drop_extent_cache(inode, em->start, + em->start + em->len - 1, 0); + } + type = BTRFS_ORDERED_PREALLOC; + } else { + type = BTRFS_ORDERED_NOCOW; + } - if (btrfs_cross_ref_exists(trans, root, &found_key, bytenr)) - goto not_found; - /* - * we may be called by the resizer, make sure we're inside - * the limits of the FS - */ - block_group = btrfs_lookup_block_group(root->fs_info, - bytenr); - if (!block_group || block_group->ro) - goto not_found; + ret = btrfs_add_ordered_extent(inode, cur_offset, disk_bytenr, + num_bytes, num_bytes, type); + BUG_ON(ret); - bytenr += btrfs_file_extent_offset(leaf, item); - extent_num_bytes = min(end + 1, extent_end) - start; - ret = btrfs_add_ordered_extent(inode, start, bytenr, - extent_num_bytes, 1); - if (ret) { - err = ret; - goto out; - } + extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree, + cur_offset, cur_offset + num_bytes - 1, + locked_page, 1, 1, 1, 0, 0, 0); + cur_offset = extent_end; + if (cur_offset > end) + break; + } + btrfs_release_path(root, path); - btrfs_release_path(root, path); - start = extent_end; - if (start <= end) { - loops++; - goto again; - } - } else { -not_found: - btrfs_end_transaction(trans, root); - btrfs_free_path(path); - return cow_file_range(inode, start, end); + if (cur_offset <= end && cow_start == (u64)-1) + cow_start = cur_offset; + if (cow_start != (u64)-1) { + ret = cow_file_range(inode, locked_page, cow_start, end, + page_started, nr_written, 1); + BUG_ON(ret); } -out: - WARN_ON(err); - btrfs_end_transaction(trans, root); + + ret = btrfs_end_transaction(trans, root); + BUG_ON(ret); btrfs_free_path(path); - return err; + return 0; } -static int run_delalloc_range(struct inode *inode, u64 start, u64 end) +/* + * extent_io.c call back to do delayed allocation processing + */ +static int run_delalloc_range(struct inode *inode, struct page *locked_page, + u64 start, u64 end, int *page_started, + unsigned long *nr_written) { - struct btrfs_root *root = BTRFS_I(inode)->root; int ret; - if (btrfs_test_opt(root, NODATACOW) || - btrfs_test_flag(inode, NODATACOW)) - ret = run_delalloc_nocow(inode, start, end); + if (btrfs_test_flag(inode, NODATACOW)) + ret = run_delalloc_nocow(inode, locked_page, start, end, + page_started, 1, nr_written); + else if (btrfs_test_flag(inode, PREALLOC)) + ret = run_delalloc_nocow(inode, locked_page, start, end, + page_started, 0, nr_written); else - ret = cow_file_range(inode, start, end); + ret = cow_file_range_async(inode, locked_page, start, end, + page_started, nr_written); return ret; } -int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end, +/* + * extent_io.c set_bit_hook, used to track delayed allocation + * 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, u64 start, u64 end, unsigned long old, unsigned long bits) { - unsigned long flags; + /* + * set_bit and clear bit hooks normally require _irqsave/restore + * but in this case, we are only testeing for the DELALLOC + * bit, which is only set or cleared with irqs on + */ if (!(old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) { struct btrfs_root *root = BTRFS_I(inode)->root; - spin_lock_irqsave(&root->fs_info->delalloc_lock, flags); + btrfs_delalloc_reserve_space(root, inode, end - start + 1); + spin_lock(&root->fs_info->delalloc_lock); BTRFS_I(inode)->delalloc_bytes += end - start + 1; root->fs_info->delalloc_bytes += end - start + 1; if (list_empty(&BTRFS_I(inode)->delalloc_inodes)) { list_add_tail(&BTRFS_I(inode)->delalloc_inodes, &root->fs_info->delalloc_inodes); } - spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags); + spin_unlock(&root->fs_info->delalloc_lock); } return 0; } -int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end, +/* + * extent_io.c clear_bit_hook, see set_bit_hook for why + */ +static int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end, unsigned long old, unsigned long bits) { + /* + * set_bit and clear bit hooks normally require _irqsave/restore + * but in this case, we are only testeing for the DELALLOC + * bit, which is only set or cleared with irqs on + */ if ((old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) { struct btrfs_root *root = BTRFS_I(inode)->root; - unsigned long flags; - spin_lock_irqsave(&root->fs_info->delalloc_lock, flags); + spin_lock(&root->fs_info->delalloc_lock); if (end - start + 1 > root->fs_info->delalloc_bytes) { - printk("warning: delalloc account %Lu %Lu\n", - end - start + 1, root->fs_info->delalloc_bytes); + printk(KERN_INFO "btrfs warning: delalloc account " + "%llu %llu\n", + (unsigned long long)end - start + 1, + (unsigned long long) + root->fs_info->delalloc_bytes); + btrfs_delalloc_free_space(root, inode, (u64)-1); root->fs_info->delalloc_bytes = 0; BTRFS_I(inode)->delalloc_bytes = 0; } else { + btrfs_delalloc_free_space(root, inode, + end - start + 1); root->fs_info->delalloc_bytes -= end - start + 1; BTRFS_I(inode)->delalloc_bytes -= end - start + 1; } @@ -341,101 +1209,147 @@ int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end, !list_empty(&BTRFS_I(inode)->delalloc_inodes)) { list_del_init(&BTRFS_I(inode)->delalloc_inodes); } - spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags); + spin_unlock(&root->fs_info->delalloc_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, - size_t size, struct bio *bio) + 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 = bio->bi_sector << 9; + u64 logical = (u64)bio->bi_sector << 9; u64 length = 0; u64 map_length; int ret; + if (bio_flags & EXTENT_BIO_COMPRESSED) + return 0; + length = bio->bi_size; map_tree = &root->fs_info->mapping_tree; map_length = length; ret = btrfs_map_block(map_tree, READ, logical, &map_length, NULL, 0); - if (map_length < length + size) { + if (map_length < length + size) return 1; - } return 0; } -int __btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, - int mirror_num) +/* + * in order to insert checksums into the metadata in large chunks, + * we wait until bio submission time. All the pages in the bio are + * checksummed and sums are attached onto the ordered extent record. + * + * At IO completion time the cums attached on the ordered extent record + * are inserted into the btree + */ +static int __btrfs_submit_bio_start(struct inode *inode, int rw, + struct bio *bio, int mirror_num, + unsigned long bio_flags) { struct btrfs_root *root = BTRFS_I(inode)->root; int ret = 0; - ret = btrfs_csum_one_bio(root, inode, bio); + ret = btrfs_csum_one_bio(root, inode, bio, 0, 0); BUG_ON(ret); + return 0; +} +/* + * in order to insert checksums into the metadata in large chunks, + * we wait until bio submission time. All the pages in the bio are + * checksummed and sums are attached onto the ordered extent record. + * + * At IO completion time the cums attached on the ordered extent record + * are inserted into the btree + */ +static int __btrfs_submit_bio_done(struct inode *inode, int rw, struct bio *bio, + int mirror_num, unsigned long bio_flags) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; return btrfs_map_bio(root, rw, bio, mirror_num, 1); } -int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, - int mirror_num) +/* + * extent_io.c submission hook. This does the right thing for csum calculation + * on write, or reading the csums from the tree before a read + */ +static int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, + int mirror_num, unsigned long bio_flags) { struct btrfs_root *root = BTRFS_I(inode)->root; int ret = 0; + int skip_sum; + + skip_sum = btrfs_test_flag(inode, NODATASUM); ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0); BUG_ON(ret); if (!(rw & (1 << BIO_RW))) { + 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); goto mapit; + } else if (!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, + inode, rw, bio, mirror_num, + bio_flags, __btrfs_submit_bio_start, + __btrfs_submit_bio_done); } - if (btrfs_test_opt(root, NODATASUM) || - btrfs_test_flag(inode, NODATASUM)) { - goto mapit; - } - - return btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info, - inode, rw, bio, mirror_num, - __btrfs_submit_bio_hook); mapit: return btrfs_map_bio(root, rw, bio, mirror_num, 0); } +/* + * given a list of ordered sums record them in the inode. This happens + * at IO completion time based on sums calculated at bio submission time. + */ static noinline int add_pending_csums(struct btrfs_trans_handle *trans, struct inode *inode, u64 file_offset, struct list_head *list) { - struct list_head *cur; struct btrfs_ordered_sum *sum; btrfs_set_trans_block_group(trans, inode); - list_for_each(cur, list) { - sum = list_entry(cur, struct btrfs_ordered_sum, list); - mutex_lock(&BTRFS_I(inode)->csum_mutex); - btrfs_csum_file_blocks(trans, BTRFS_I(inode)->root, - inode, sum); - mutex_unlock(&BTRFS_I(inode)->csum_mutex); + + list_for_each_entry(sum, list, list) { + btrfs_csum_file_blocks(trans, + BTRFS_I(inode)->root->fs_info->csum_root, sum); } return 0; } int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end) { + if ((end & (PAGE_CACHE_SIZE - 1)) == 0) + WARN_ON(1); return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS); } +/* see btrfs_writepage_start_hook for details on why this is required */ struct btrfs_writepage_fixup { struct page *page; struct btrfs_work work; }; -/* see btrfs_writepage_start_hook for details on why this is required */ -void btrfs_writepage_fixup_worker(struct btrfs_work *work) +static void btrfs_writepage_fixup_worker(struct btrfs_work *work) { struct btrfs_writepage_fixup *fixup; struct btrfs_ordered_extent *ordered; @@ -489,12 +1403,12 @@ out_page: * good idea. This causes problems because we want to make sure COW * properly happens and the data=ordered rules are followed. * - * In our case any range that doesn't have the EXTENT_ORDERED bit set + * In our case any range that doesn't have the ORDERED bit set * hasn't been properly setup for IO. We kick off an async process * to fix it up. The async helper will wait for ordered extents, set * the delalloc bit and make it safe to write the page. */ -int btrfs_writepage_start_hook(struct page *page, u64 start, u64 end) +static int btrfs_writepage_start_hook(struct page *page, u64 start, u64 end) { struct inode *inode = page->mapping->host; struct btrfs_writepage_fixup *fixup; @@ -521,15 +1435,74 @@ int btrfs_writepage_start_hook(struct page *page, u64 start, u64 end) return -EAGAIN; } +static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, + struct inode *inode, u64 file_pos, + u64 disk_bytenr, u64 disk_num_bytes, + u64 num_bytes, u64 ram_bytes, + u8 compression, u8 encryption, + u16 other_encoding, int extent_type) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_file_extent_item *fi; + struct btrfs_path *path; + struct extent_buffer *leaf; + struct btrfs_key ins; + u64 hint; + int ret; + + path = btrfs_alloc_path(); + BUG_ON(!path); + + ret = btrfs_drop_extents(trans, root, inode, file_pos, + file_pos + num_bytes, file_pos, &hint); + BUG_ON(ret); + + 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); + leaf = path->nodes[0]; + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_generation(leaf, fi, trans->transid); + btrfs_set_file_extent_type(leaf, fi, extent_type); + btrfs_set_file_extent_disk_bytenr(leaf, fi, disk_bytenr); + btrfs_set_file_extent_disk_num_bytes(leaf, fi, disk_num_bytes); + btrfs_set_file_extent_offset(leaf, fi, 0); + btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes); + btrfs_set_file_extent_ram_bytes(leaf, fi, ram_bytes); + btrfs_set_file_extent_compression(leaf, fi, compression); + btrfs_set_file_extent_encryption(leaf, fi, encryption); + btrfs_set_file_extent_other_encoding(leaf, fi, other_encoding); + btrfs_mark_buffer_dirty(leaf); + + inode_add_bytes(inode, num_bytes); + btrfs_drop_extent_cache(inode, file_pos, file_pos + num_bytes - 1, 0); + + ins.objectid = disk_bytenr; + ins.offset = disk_num_bytes; + ins.type = BTRFS_EXTENT_ITEM_KEY; + ret = btrfs_alloc_reserved_extent(trans, root, leaf->start, + root->root_key.objectid, + trans->transid, inode->i_ino, &ins); + BUG_ON(ret); + + btrfs_free_path(path); + return 0; +} + +/* 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) { struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans; struct btrfs_ordered_extent *ordered_extent; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; - u64 alloc_hint = 0; - struct list_head list; - struct btrfs_key ins; + int compressed = 0; int ret; ret = btrfs_dec_test_ordered_pending(inode, start, end - start + 1); @@ -547,38 +1520,26 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) ordered_extent->file_offset + ordered_extent->len - 1, GFP_NOFS); - INIT_LIST_HEAD(&list); - - ins.objectid = ordered_extent->start; - ins.offset = ordered_extent->len; - ins.type = BTRFS_EXTENT_ITEM_KEY; - - ret = btrfs_alloc_reserved_extent(trans, root, root->root_key.objectid, - trans->transid, inode->i_ino, - ordered_extent->file_offset, &ins); - BUG_ON(ret); - - mutex_lock(&BTRFS_I(inode)->extent_mutex); - - ret = btrfs_drop_extents(trans, root, inode, - ordered_extent->file_offset, - ordered_extent->file_offset + - ordered_extent->len, - ordered_extent->file_offset, &alloc_hint); - BUG_ON(ret); - ret = btrfs_insert_file_extent(trans, root, inode->i_ino, - ordered_extent->file_offset, - ordered_extent->start, - ordered_extent->len, - ordered_extent->len, 0); - BUG_ON(ret); - - btrfs_drop_extent_cache(inode, ordered_extent->file_offset, - ordered_extent->file_offset + - ordered_extent->len - 1); - mutex_unlock(&BTRFS_I(inode)->extent_mutex); - - inode->i_blocks += ordered_extent->len >> 9; + if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags)) + compressed = 1; + if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) { + BUG_ON(compressed); + ret = btrfs_mark_extent_written(trans, root, inode, + ordered_extent->file_offset, + ordered_extent->file_offset + + ordered_extent->len); + BUG_ON(ret); + } else { + 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, + BTRFS_FILE_EXTENT_REG); + BUG_ON(ret); + } unlock_extent(io_tree, ordered_extent->file_offset, ordered_extent->file_offset + ordered_extent->len - 1, GFP_NOFS); @@ -586,86 +1547,45 @@ nocow: add_pending_csums(trans, inode, ordered_extent->file_offset, &ordered_extent->list); + mutex_lock(&BTRFS_I(inode)->extent_mutex); btrfs_ordered_update_i_size(inode, ordered_extent); + btrfs_update_inode(trans, root, inode); btrfs_remove_ordered_extent(inode, ordered_extent); + mutex_unlock(&BTRFS_I(inode)->extent_mutex); /* once for us */ btrfs_put_ordered_extent(ordered_extent); /* once for the tree */ btrfs_put_ordered_extent(ordered_extent); - btrfs_update_inode(trans, root, inode); btrfs_end_transaction(trans, root); return 0; } -int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end, +static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end, struct extent_state *state, int uptodate) { return btrfs_finish_ordered_io(page->mapping->host, start, end); } -int btrfs_readpage_io_hook(struct page *page, u64 start, u64 end) -{ - int ret = 0; - struct inode *inode = page->mapping->host; - struct btrfs_root *root = BTRFS_I(inode)->root; - struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; - struct btrfs_csum_item *item; - struct btrfs_path *path = NULL; - u32 csum; - - if (btrfs_test_opt(root, NODATASUM) || - btrfs_test_flag(inode, NODATASUM)) - return 0; - - /* - * It is possible there is an ordered extent that has - * not yet finished for this range in the file. If so, - * that extent will have a csum cached, and it will insert - * the sum after all the blocks in the extent are fully - * on disk. So, look for an ordered extent and use the - * sum if found. We have to do this before looking in the - * btree because csum items are pre-inserted based on - * the file size. btrfs_lookup_csum might find an item - * that still hasn't been fully filled. - */ - ret = btrfs_find_ordered_sum(inode, start, &csum); - if (ret == 0) - goto found; - - ret = 0; - path = btrfs_alloc_path(); - item = btrfs_lookup_csum(NULL, root, path, inode->i_ino, start, 0); - if (IS_ERR(item)) { - ret = PTR_ERR(item); - /* a csum that isn't present is a preallocated region. */ - if (ret == -ENOENT || ret == -EFBIG) - ret = 0; - csum = 0; - printk("no csum found for inode %lu start %Lu\n", inode->i_ino, - start); - goto out; - } - read_extent_buffer(path->nodes[0], &csum, (unsigned long)item, - BTRFS_CRC32_SIZE); -found: - set_state_private(io_tree, start, csum); -out: - if (path) - btrfs_free_path(path); - return ret; -} - +/* + * 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; }; -int btrfs_io_failed_hook(struct bio *failed_bio, +static int btrfs_io_failed_hook(struct bio *failed_bio, struct page *page, u64 start, u64 end, struct extent_state *state) { @@ -689,6 +1609,7 @@ int btrfs_io_failed_hook(struct bio *failed_bio, failrec->start = start; failrec->len = end - start + 1; failrec->last_mirror = 0; + failrec->bio_flags = 0; spin_lock(&em_tree->lock); em = lookup_extent_mapping(em_tree, start, failrec->len); @@ -704,6 +1625,10 @@ int btrfs_io_failed_hook(struct bio *failed_bio, } 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 | @@ -718,13 +1643,13 @@ int btrfs_io_failed_hook(struct bio *failed_bio, failrec->logical, failrec->len); failrec->last_mirror++; if (!state) { - spin_lock_irq(&BTRFS_I(inode)->io_tree.lock); + 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_irq(&BTRFS_I(inode)->io_tree.lock); + spin_unlock(&BTRFS_I(inode)->io_tree.lock); } if (!state || failrec->last_mirror > num_copies) { set_state_private(failure_tree, failrec->start, 0); @@ -740,6 +1665,7 @@ int btrfs_io_failed_hook(struct bio *failed_bio, bio->bi_sector = failrec->logical >> 9; bio->bi_bdev = failed_bio->bi_bdev; bio->bi_size = 0; + bio_add_page(bio, page, failrec->len, start - page_offset(page)); if (failed_bio->bi_rw & (1 << BIO_RW)) rw = WRITE; @@ -747,11 +1673,16 @@ int btrfs_io_failed_hook(struct bio *failed_bio, rw = READ; BTRFS_I(inode)->io_tree.ops->submit_bio_hook(inode, rw, bio, - failrec->last_mirror); + failrec->last_mirror, + failrec->bio_flags); return 0; } -int btrfs_clean_io_failures(struct inode *inode, u64 start) +/* + * 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; @@ -779,7 +1710,12 @@ int btrfs_clean_io_failures(struct inode *inode, u64 start) return 0; } -int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end, +/* + * 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 + */ +static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end, struct extent_state *state) { size_t offset = start - ((u64)page->index << PAGE_CACHE_SHIFT); @@ -790,30 +1726,38 @@ int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end, int ret; struct btrfs_root *root = BTRFS_I(inode)->root; u32 csum = ~(u32)0; - unsigned long flags; - if (btrfs_test_opt(root, NODATASUM) || - btrfs_test_flag(inode, NODATASUM)) + if (PageChecked(page)) { + ClearPageChecked(page); + goto good; + } + if (btrfs_test_flag(inode, NODATASUM)) + return 0; + + if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID && + test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1)) { + clear_extent_bits(io_tree, start, end, EXTENT_NODATASUM, + GFP_NOFS); return 0; + } + if (state && state->start == start) { private = state->private; ret = 0; } else { ret = get_state_private(io_tree, start, &private); } - local_irq_save(flags); - kaddr = kmap_atomic(page, KM_IRQ0); - if (ret) { + kaddr = kmap_atomic(page, KM_USER0); + if (ret) goto zeroit; - } + csum = btrfs_csum_data(root, kaddr + offset, csum, end - start + 1); btrfs_csum_final(csum, (char *)&csum); - if (csum != private) { + if (csum != private) goto zeroit; - } - kunmap_atomic(kaddr, KM_IRQ0); - local_irq_restore(flags); + kunmap_atomic(kaddr, KM_USER0); +good: /* if the io failure tree for this inode is non-empty, * check to see if we've recovered from a failed IO */ @@ -821,13 +1765,13 @@ int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end, return 0; zeroit: - printk("btrfs csum failed ino %lu off %llu csum %u private %Lu\n", - page->mapping->host->i_ino, (unsigned long long)start, csum, - private); + 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); memset(kaddr + offset, 1, end - start + 1); flush_dcache_page(page); - kunmap_atomic(kaddr, KM_IRQ0); - local_irq_restore(flags); + kunmap_atomic(kaddr, KM_USER0); if (private == 0) return 0; return -EIO; @@ -905,10 +1849,6 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) struct inode *inode; int ret = 0, nr_unlink = 0, nr_truncate = 0; - /* don't do orphan cleanup if the fs is readonly. */ - if (root->inode->i_sb->s_flags & MS_RDONLY) - return; - path = btrfs_alloc_path(); if (!path) return; @@ -918,8 +1858,6 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY); key.offset = (u64)-1; - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, root->inode); while (1) { ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); @@ -959,7 +1897,7 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) * crossing root thing. we store the inode number in the * offset of the orphan item. */ - inode = btrfs_iget_locked(root->inode->i_sb, + inode = btrfs_iget_locked(root->fs_info->sb, found_key.offset, root); if (!inode) break; @@ -991,7 +1929,9 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) * do a destroy_inode */ if (is_bad_inode(inode)) { + trans = btrfs_start_transaction(root, 1); btrfs_orphan_del(trans, inode); + btrfs_end_transaction(trans, root); iput(inode); continue; } @@ -1014,9 +1954,11 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) printk(KERN_INFO "btrfs: truncated %d orphans\n", nr_truncate); btrfs_free_path(path); - btrfs_end_transaction(trans, root); } +/* + * read an inode from the btree into the in-memory inode + */ void btrfs_read_locked_inode(struct inode *inode) { struct btrfs_path *path; @@ -1059,22 +2001,20 @@ void btrfs_read_locked_inode(struct inode *inode) inode->i_ctime.tv_sec = btrfs_timespec_sec(leaf, tspec); inode->i_ctime.tv_nsec = btrfs_timespec_nsec(leaf, tspec); - inode->i_blocks = btrfs_inode_nblocks(leaf, inode_item); - inode->i_generation = btrfs_inode_generation(leaf, inode_item); + 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); + inode->i_generation = BTRFS_I(inode)->generation; inode->i_rdev = 0; rdev = btrfs_inode_rdev(leaf, inode_item); 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); - BTRFS_I(inode)->block_group = btrfs_lookup_block_group(root->fs_info, - alloc_group_block); - BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item); - if (!BTRFS_I(inode)->block_group) { - BTRFS_I(inode)->block_group = btrfs_find_block_group(root, - NULL, 0, - BTRFS_BLOCK_GROUP_METADATA, 0); - } + + BTRFS_I(inode)->block_group = btrfs_find_block_group(root, 0, + alloc_group_block, 0); btrfs_free_path(path); inode_item = NULL; @@ -1099,6 +2039,7 @@ void btrfs_read_locked_inode(struct inode *inode) inode->i_mapping->backing_dev_info = &root->fs_info->bdi; break; default: + inode->i_op = &btrfs_special_inode_operations; init_special_inode(inode, inode->i_mode, rdev); break; } @@ -1109,7 +2050,11 @@ make_bad: make_bad_inode(inode); } -static void fill_inode_item(struct extent_buffer *leaf, +/* + * given a leaf and an inode, copy the inode fields into the leaf + */ +static void fill_inode_item(struct btrfs_trans_handle *trans, + struct extent_buffer *leaf, struct btrfs_inode_item *item, struct inode *inode) { @@ -1134,17 +2079,20 @@ static void fill_inode_item(struct extent_buffer *leaf, btrfs_set_timespec_nsec(leaf, btrfs_inode_ctime(item), inode->i_ctime.tv_nsec); - btrfs_set_inode_nblocks(leaf, item, inode->i_blocks); - btrfs_set_inode_generation(leaf, item, inode->i_generation); + 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->key.objectid); + btrfs_set_inode_block_group(leaf, item, BTRFS_I(inode)->block_group); } -int noinline btrfs_update_inode(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode) +/* + * 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) { struct btrfs_inode_item *inode_item; struct btrfs_path *path; @@ -1161,11 +2109,12 @@ int noinline btrfs_update_inode(struct btrfs_trans_handle *trans, 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); - fill_inode_item(leaf, inode_item, inode); + fill_inode_item(trans, leaf, inode_item, inode); btrfs_mark_buffer_dirty(leaf); btrfs_set_inode_last_trans(trans, inode); ret = 0; @@ -1175,14 +2124,17 @@ failed: } -static int btrfs_unlink_trans(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *dir, - struct dentry *dentry) +/* + * 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) { struct btrfs_path *path; - const char *name = dentry->d_name.name; - int name_len = dentry->d_name.len; int ret = 0; struct extent_buffer *leaf; struct btrfs_dir_item *di; @@ -1213,13 +2165,12 @@ static int btrfs_unlink_trans(struct btrfs_trans_handle *trans, btrfs_release_path(root, path); ret = btrfs_del_inode_ref(trans, root, name, name_len, - dentry->d_inode->i_ino, - dentry->d_parent->d_inode->i_ino, &index); + inode->i_ino, + dir->i_ino, &index); if (ret) { - printk("failed to delete reference to %.*s, " + printk(KERN_INFO "btrfs failed to delete reference to %.*s, " "inode %lu parent %lu\n", name_len, name, - dentry->d_inode->i_ino, - dentry->d_parent->d_inode->i_ino); + inode->i_ino, dir->i_ino); goto err; } @@ -1236,21 +2187,27 @@ static int btrfs_unlink_trans(struct btrfs_trans_handle *trans, ret = btrfs_delete_one_dir_name(trans, root, path, di); btrfs_release_path(root, path); - dentry->d_inode->i_ctime = dir->i_ctime; + ret = btrfs_del_inode_ref_in_log(trans, root, name, name_len, + inode, dir->i_ino); + BUG_ON(ret != 0 && ret != -ENOENT); + if (ret != -ENOENT) + BTRFS_I(dir)->log_dirty_trans = trans->transid; + + ret = btrfs_del_dir_entries_in_log(trans, root, name, name_len, + dir, index); + BUG_ON(ret); err: btrfs_free_path(path); - if (!ret) { - btrfs_i_size_write(dir, dir->i_size - name_len * 2); - dir->i_mtime = dir->i_ctime = CURRENT_TIME; - btrfs_update_inode(trans, root, dir); -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18) - dentry->d_inode->i_nlink--; -#else - drop_nlink(dentry->d_inode); -#endif - ret = btrfs_update_inode(trans, root, dentry->d_inode); - dir->i_sb->s_dirt = 1; - } + if (ret) + goto out; + + btrfs_i_size_write(dir, dir->i_size - name_len * 2); + 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); + dir->i_sb->s_dirt = 1; +out: return ret; } @@ -1264,14 +2221,11 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry) root = BTRFS_I(dir)->root; - ret = btrfs_check_free_space(root, 1, 1); - if (ret) - goto fail; - trans = btrfs_start_transaction(root, 1); btrfs_set_trans_block_group(trans, dir); - ret = btrfs_unlink_trans(trans, root, dir, dentry); + ret = btrfs_unlink_inode(trans, root, dir, dentry->d_inode, + dentry->d_name.name, dentry->d_name.len); if (inode->i_nlink == 0) ret = btrfs_orphan_add(trans, inode); @@ -1279,7 +2233,6 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry) nr = trans->blocks_used; btrfs_end_transaction_throttle(trans, root); -fail: btrfs_btree_balance_dirty(root, nr); return ret; } @@ -1293,14 +2246,15 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) struct btrfs_trans_handle *trans; unsigned long nr = 0; - if (inode->i_size > BTRFS_EMPTY_DIR_SIZE) { + /* + * the FIRST_FREE_OBJECTID check makes sure we don't try to rmdir + * the root of a subvolume or snapshot + */ + if (inode->i_size > BTRFS_EMPTY_DIR_SIZE || + inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) { return -ENOTEMPTY; } - ret = btrfs_check_free_space(root, 1, 1); - if (ret) - goto fail; - trans = btrfs_start_transaction(root, 1); btrfs_set_trans_block_group(trans, dir); @@ -1309,15 +2263,14 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry) goto fail_trans; /* now the directory is empty */ - err = btrfs_unlink_trans(trans, root, dir, dentry); - if (!err) { + err = btrfs_unlink_inode(trans, root, dir, dentry->d_inode, + dentry->d_name.name, dentry->d_name.len); + if (!err) btrfs_i_size_write(inode, 0); - } fail_trans: nr = trans->blocks_used; ret = btrfs_end_transaction_throttle(trans, root); -fail: btrfs_btree_balance_dirty(root, nr); if (ret && !err) @@ -1325,10 +2278,182 @@ fail: 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 - * any higher than i_size. + * any higher than new_size * * csum items that cross the new i_size are truncated to the new size * as well. @@ -1336,16 +2461,16 @@ fail: * min_type is the minimum key type to truncate down to. If set to 0, this * will kill all the items on this inode, including the INODE_ITEM_KEY. */ -static int btrfs_truncate_in_trans(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct inode *inode, - u32 min_type) +noinline int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode, + u64 new_size, u32 min_type) { int ret; struct btrfs_path *path; struct btrfs_key key; struct btrfs_key found_key; - u32 found_type; + u32 found_type = (u8)-1; struct extent_buffer *leaf; struct btrfs_file_extent_item *fi; u64 extent_start = 0; @@ -1358,9 +2483,11 @@ static int btrfs_truncate_in_trans(struct btrfs_trans_handle *trans, int pending_del_nr = 0; int pending_del_slot = 0; int extent_type = -1; + int encoding; u64 mask = root->sectorsize - 1; - btrfs_drop_extent_cache(inode, inode->i_size & (~mask), (u64)-1); + if (root->ref_cows) + btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0); path = btrfs_alloc_path(); path->reada = -1; BUG_ON(!path); @@ -1370,22 +2497,28 @@ static int btrfs_truncate_in_trans(struct btrfs_trans_handle *trans, key.offset = (u64)-1; key.type = (u8)-1; - btrfs_init_path(path); search_again: ret = btrfs_search_slot(trans, root, &key, path, -1, 1); - if (ret < 0) { + if (ret < 0) goto error; - } + if (ret > 0) { - BUG_ON(path->slots[0] == 0); + /* there are no items in the tree for us to truncate, we're + * done + */ + if (path->slots[0] == 0) { + ret = 0; + goto error; + } path->slots[0]--; } - while(1) { + while (1) { fi = NULL; 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) break; @@ -1398,40 +2531,36 @@ 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) { - struct btrfs_item *item = btrfs_item_nr(leaf, - path->slots[0]); item_end += btrfs_file_extent_inline_len(leaf, - item); + fi); } item_end--; } - if (found_type == BTRFS_CSUM_ITEM_KEY) { - ret = btrfs_csum_truncate(trans, root, path, - inode->i_size); - BUG_ON(ret); - } - if (item_end < inode->i_size) { - if (found_type == BTRFS_DIR_ITEM_KEY) { + if (item_end < new_size) { + if (found_type == BTRFS_DIR_ITEM_KEY) found_type = BTRFS_INODE_ITEM_KEY; - } else if (found_type == BTRFS_EXTENT_ITEM_KEY) { - found_type = BTRFS_CSUM_ITEM_KEY; - } else if (found_type == BTRFS_EXTENT_DATA_KEY) { + else if (found_type == BTRFS_EXTENT_ITEM_KEY) + found_type = BTRFS_EXTENT_DATA_KEY; + else if (found_type == BTRFS_EXTENT_DATA_KEY) found_type = BTRFS_XATTR_ITEM_KEY; - } else if (found_type == BTRFS_XATTR_ITEM_KEY) { + else if (found_type == BTRFS_XATTR_ITEM_KEY) found_type = BTRFS_INODE_REF_KEY; - } else if (found_type) { + else if (found_type) found_type--; - } else { + else break; - } btrfs_set_key_type(&key, found_type); goto next; } - if (found_key.offset >= inode->i_size) + if (found_key.offset >= new_size) del_item = 1; else del_item = 0; @@ -1444,10 +2573,10 @@ search_again: if (extent_type != BTRFS_FILE_EXTENT_INLINE) { u64 num_dec; extent_start = btrfs_file_extent_disk_bytenr(leaf, fi); - if (!del_item) { + if (!del_item && !encoding) { u64 orig_num_bytes = btrfs_file_extent_num_bytes(leaf, fi); - extent_num_bytes = inode->i_size - + extent_num_bytes = new_size - found_key.offset + root->sectorsize - 1; extent_num_bytes = extent_num_bytes & ~((u64)root->sectorsize - 1); @@ -1455,8 +2584,8 @@ search_again: extent_num_bytes); num_dec = (orig_num_bytes - extent_num_bytes); - if (extent_start != 0) - dec_i_blocks(inode, num_dec); + if (root->ref_cows && extent_start != 0) + inode_sub_bytes(inode, num_dec); btrfs_mark_buffer_dirty(leaf); } else { extent_num_bytes = @@ -1466,24 +2595,35 @@ search_again: num_dec = btrfs_file_extent_num_bytes(leaf, fi); if (extent_start != 0) { found_extent = 1; - dec_i_blocks(inode, num_dec); + if (root->ref_cows) + inode_sub_bytes(inode, num_dec); } root_gen = btrfs_header_generation(leaf); root_owner = btrfs_header_owner(leaf); } } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { - if (!del_item) { - u32 newsize = inode->i_size - found_key.offset; - dec_i_blocks(inode, item_end + 1 - - found_key.offset - newsize); - newsize = - btrfs_file_extent_calc_inline_size(newsize); + /* + * we can't truncate inline items that have had + * special encodings + */ + if (!del_item && + btrfs_file_extent_compression(leaf, fi) == 0 && + btrfs_file_extent_encryption(leaf, fi) == 0 && + btrfs_file_extent_other_encoding(leaf, fi) == 0) { + u32 size = new_size - found_key.offset; + + if (root->ref_cows) { + inode_sub_bytes(inode, item_end + 1 - + new_size); + } + size = + btrfs_file_extent_calc_inline_size(size); ret = btrfs_truncate_item(trans, root, path, - newsize, 1); + size, 1); BUG_ON(ret); - } else { - dec_i_blocks(inode, item_end + 1 - - found_key.offset); + } else if (root->ref_cows) { + inode_sub_bytes(inode, item_end + 1 - + found_key.offset); } } delete: @@ -1498,7 +2638,7 @@ delete: pending_del_nr++; pending_del_slot = path->slots[0]; } else { - printk("bad pending slot %d pending_del_nr %d pending_del_slot %d\n", path->slots[0], pending_del_nr, pending_del_slot); + BUG(); } } else { break; @@ -1506,9 +2646,8 @@ delete: if (found_extent) { ret = btrfs_free_extent(trans, root, extent_start, extent_num_bytes, - root_owner, - root_gen, inode->i_ino, - found_key.offset, 0); + leaf->start, root_owner, + root_gen, inode->i_ino, 0); BUG_ON(ret); } next: @@ -1516,6 +2655,8 @@ next: if (pending_del_nr) goto del_pending; btrfs_release_path(root, path); + if (found_type == BTRFS_INODE_ITEM_KEY) + break; goto search_again; } @@ -1532,6 +2673,8 @@ del_pending: BUG_ON(ret); pending_del_nr = 0; btrfs_release_path(root, path); + if (found_type == BTRFS_INODE_ITEM_KEY) + break; goto search_again; } } @@ -1624,68 +2767,98 @@ out: return ret; } -static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) +int btrfs_cont_expand(struct inode *inode, loff_t size) { - struct inode *inode = dentry->d_inode; + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; + struct extent_map *em; + u64 mask = root->sectorsize - 1; + u64 hole_start = (inode->i_size + mask) & ~mask; + u64 block_end = (size + mask) & ~mask; + u64 last_byte; + u64 cur_offset; + u64 hole_size; int err; - err = inode_change_ok(inode, attr); + if (size <= hole_start) + return 0; + + err = btrfs_check_metadata_free_space(root); if (err) return err; - if (S_ISREG(inode->i_mode) && - attr->ia_valid & ATTR_SIZE && attr->ia_size > inode->i_size) { - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(inode)->root; - struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; - - u64 mask = root->sectorsize - 1; - u64 hole_start = (inode->i_size + mask) & ~mask; - u64 block_end = (attr->ia_size + mask) & ~mask; - u64 hole_size; - u64 alloc_hint = 0; - - if (attr->ia_size <= hole_start) - goto out; - - err = btrfs_check_free_space(root, 1, 0); - if (err) - goto fail; - - btrfs_truncate_page(inode->i_mapping, inode->i_size); + btrfs_truncate_page(inode->i_mapping, inode->i_size); - hole_size = block_end - hole_start; - btrfs_wait_ordered_range(inode, hole_start, hole_size); + while (1) { + struct btrfs_ordered_extent *ordered; + btrfs_wait_ordered_range(inode, hole_start, + block_end - hole_start); lock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + ordered = btrfs_lookup_ordered_extent(inode, hole_start); + if (!ordered) + break; + unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + btrfs_put_ordered_extent(ordered); + } - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); - mutex_lock(&BTRFS_I(inode)->extent_mutex); - err = btrfs_drop_extents(trans, root, inode, - hole_start, block_end, hole_start, - &alloc_hint); + trans = btrfs_start_transaction(root, 1); + btrfs_set_trans_block_group(trans, inode); - if (alloc_hint != EXTENT_MAP_INLINE) { + cur_offset = hole_start; + while (1) { + em = btrfs_get_extent(inode, NULL, 0, cur_offset, + block_end - cur_offset, 0); + BUG_ON(IS_ERR(em) || !em); + last_byte = min(extent_map_end(em), block_end); + last_byte = (last_byte + mask) & ~mask; + if (test_bit(EXTENT_FLAG_VACANCY, &em->flags)) { + u64 hint_byte = 0; + hole_size = last_byte - cur_offset; + err = btrfs_drop_extents(trans, root, inode, + cur_offset, + cur_offset + hole_size, + cur_offset, &hint_byte); + if (err) + break; err = btrfs_insert_file_extent(trans, root, - inode->i_ino, - hole_start, 0, 0, - hole_size, 0); + inode->i_ino, cur_offset, 0, + 0, hole_size, 0, hole_size, + 0, 0, 0); btrfs_drop_extent_cache(inode, hole_start, - (u64)-1); - btrfs_check_file(root, inode); + last_byte - 1, 0); } - mutex_unlock(&BTRFS_I(inode)->extent_mutex); - btrfs_end_transaction(trans, root); - unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + free_extent_map(em); + cur_offset = last_byte; + if (err || cur_offset >= block_end) + break; + } + + btrfs_end_transaction(trans, root); + unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + return err; +} + +static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) +{ + struct inode *inode = dentry->d_inode; + int err; + + err = inode_change_ok(inode, attr); + if (err) + return err; + + if (S_ISREG(inode->i_mode) && + attr->ia_valid & ATTR_SIZE && attr->ia_size > inode->i_size) { + err = btrfs_cont_expand(inode, attr->ia_size); if (err) return err; } -out: + err = inode_setattr(inode, attr); if (!err && ((attr->ia_valid & ATTR_MODE))) err = btrfs_acl_chmod(inode); -fail: return err; } @@ -1704,10 +2877,10 @@ void btrfs_delete_inode(struct inode *inode) btrfs_wait_ordered_range(inode, 0, (u64)-1); btrfs_i_size_write(inode, 0); - trans = btrfs_start_transaction(root, 1); + trans = btrfs_join_transaction(root, 1); btrfs_set_trans_block_group(trans, inode); - ret = btrfs_truncate_in_trans(trans, root, inode, 0); + ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, 0); if (ret) { btrfs_orphan_del(NULL, inode); goto no_delete_lock; @@ -1744,49 +2917,17 @@ static int btrfs_inode_by_name(struct inode *dir, struct dentry *dentry, struct btrfs_root *root = BTRFS_I(dir)->root; int ret = 0; - if (namelen == 1 && strcmp(name, ".") == 0) { - location->objectid = dir->i_ino; - location->type = BTRFS_INODE_ITEM_KEY; - location->offset = 0; - return 0; - } path = btrfs_alloc_path(); BUG_ON(!path); - if (namelen == 2 && strcmp(name, "..") == 0) { - struct btrfs_key key; - struct extent_buffer *leaf; - int slot; - - key.objectid = dir->i_ino; - key.offset = (u64)-1; - btrfs_set_key_type(&key, BTRFS_INODE_REF_KEY); - if (ret < 0 || path->slots[0] == 0) - goto out_err; - ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); - BUG_ON(ret == 0); - ret = 0; - leaf = path->nodes[0]; - slot = path->slots[0] - 1; - - btrfs_item_key_to_cpu(leaf, &key, slot); - if (key.objectid != dir->i_ino || - key.type != BTRFS_INODE_REF_KEY) { - goto out_err; - } - location->objectid = key.offset; - location->type = BTRFS_INODE_ITEM_KEY; - location->offset = 0; - goto out; - } - di = btrfs_lookup_dir_item(NULL, root, path, dir->i_ino, name, namelen, 0); if (IS_ERR(di)) ret = PTR_ERR(di); - if (!di || IS_ERR(di)) { + + if (!di || IS_ERR(di)) goto out_err; - } + btrfs_dir_item_key_to_cpu(path->nodes[0], di, location); out: btrfs_free_path(path); @@ -1827,15 +2968,23 @@ static int fixup_tree_root_location(struct btrfs_root *root, return 0; } -static int btrfs_init_locked_inode(struct inode *inode, void *p) +static noinline void init_btrfs_i(struct inode *inode) { - struct btrfs_iget_args *args = p; - inode->i_ino = args->ino; - BTRFS_I(inode)->root = args->root; - BTRFS_I(inode)->delalloc_bytes = 0; - inode->i_mapping->writeback_index = 0; - BTRFS_I(inode)->disk_i_size = 0; - BTRFS_I(inode)->index_cnt = (u64)-1; + struct btrfs_inode *bi = BTRFS_I(inode); + + bi->i_acl = NULL; + bi->i_default_acl = NULL; + + bi->generation = 0; + bi->sequence = 0; + bi->last_trans = 0; + bi->logged_trans = 0; + bi->delalloc_bytes = 0; + bi->reserved_bytes = 0; + bi->disk_i_size = 0; + bi->flags = 0; + bi->index_cnt = (u64)-1; + bi->log_dirty_trans = 0; extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS); extent_io_tree_init(&BTRFS_I(inode)->io_tree, inode->i_mapping, GFP_NOFS); @@ -1843,29 +2992,43 @@ static int btrfs_init_locked_inode(struct inode *inode, void *p) inode->i_mapping, GFP_NOFS); INIT_LIST_HEAD(&BTRFS_I(inode)->delalloc_inodes); btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree); - mutex_init(&BTRFS_I(inode)->csum_mutex); mutex_init(&BTRFS_I(inode)->extent_mutex); + mutex_init(&BTRFS_I(inode)->log_mutex); +} + +static int btrfs_init_locked_inode(struct inode *inode, void *p) +{ + struct btrfs_iget_args *args = p; + inode->i_ino = args->ino; + init_btrfs_i(inode); + 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 && - args->root == BTRFS_I(inode)->root); + return args->ino == inode->i_ino && + args->root == BTRFS_I(inode)->root; } struct inode *btrfs_ilookup(struct super_block *s, u64 objectid, - u64 root_objectid) + struct btrfs_root *root, int wait) { + struct inode *inode; struct btrfs_iget_args args; args.ino = objectid; - args.root = btrfs_lookup_fs_root(btrfs_sb(s)->fs_info, root_objectid); - - if (!args.root) - return NULL; + args.root = root; - return ilookup5(s, objectid, btrfs_find_actor, (void *)&args); + if (wait) { + inode = ilookup5(s, objectid, btrfs_find_actor, + (void *)&args); + } else { + inode = ilookup5_nowait(s, objectid, btrfs_find_actor, + (void *)&args); + } + return inode; } struct inode *btrfs_iget_locked(struct super_block *s, u64 objectid, @@ -1876,21 +3039,47 @@ struct inode *btrfs_iget_locked(struct super_block *s, u64 objectid, args.ino = objectid; args.root = root; - inode = iget5_locked(s, objectid, btrfs_find_actor, - btrfs_init_locked_inode, - (void *)&args); + inode = iget5_locked(s, objectid, btrfs_find_actor, + btrfs_init_locked_inode, + (void *)&args); + return inode; +} + +/* Get an inode object given its location and corresponding root. + * Returns in *is_new if the inode was read from disk + */ +struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location, + struct btrfs_root *root, int *is_new) +{ + struct inode *inode; + + inode = btrfs_iget_locked(s, location->objectid, root); + if (!inode) + return ERR_PTR(-EACCES); + + if (inode->i_state & I_NEW) { + BTRFS_I(inode)->root = root; + memcpy(&BTRFS_I(inode)->location, location, sizeof(*location)); + btrfs_read_locked_inode(inode); + unlock_new_inode(inode); + if (is_new) + *is_new = 1; + } else { + if (is_new) + *is_new = 0; + } + return inode; } -static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry, - struct nameidata *nd) +struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry) { - struct inode * inode; + struct inode *inode; struct btrfs_inode *bi = BTRFS_I(dir); struct btrfs_root *root = bi->root; struct btrfs_root *sub_root = root; struct btrfs_key location; - int ret, do_orphan = 0; + int ret, new; if (dentry->d_name.len > BTRFS_NAME_LEN) return ERR_PTR(-ENAMETOOLONG); @@ -1908,28 +3097,24 @@ static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry, return ERR_PTR(ret); if (ret > 0) return ERR_PTR(-ENOENT); - - inode = btrfs_iget_locked(dir->i_sb, location.objectid, - sub_root); - if (!inode) - return ERR_PTR(-EACCES); - if (inode->i_state & I_NEW) { - /* the inode and parent dir are two different roots */ - if (sub_root != root) { - igrab(inode); - sub_root->inode = inode; - do_orphan = 1; - } - BTRFS_I(inode)->root = sub_root; - memcpy(&BTRFS_I(inode)->location, &location, - sizeof(location)); - btrfs_read_locked_inode(inode); - unlock_new_inode(inode); - } + inode = btrfs_iget(dir->i_sb, &location, sub_root, &new); + if (IS_ERR(inode)) + return ERR_CAST(inode); } + return inode; +} - if (unlikely(do_orphan)) - btrfs_orphan_cleanup(sub_root); +static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry, + struct nameidata *nd) +{ + struct inode *inode; + + if (dentry->d_name.len > BTRFS_NAME_LEN) + return ERR_PTR(-ENAMETOOLONG); + + inode = btrfs_lookup_dentry(dir, dentry); + if (IS_ERR(inode)) + return ERR_CAST(inode); return d_splice_alias(inode, dentry); } @@ -1938,7 +3123,8 @@ static unsigned char btrfs_filetype_table[] = { DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK }; -static int btrfs_readdir(struct file *filp, void *dirent, filldir_t filldir) +static int btrfs_real_readdir(struct file *filp, void *dirent, + filldir_t filldir) { struct inode *inode = filp->f_dentry->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; @@ -1975,49 +3161,33 @@ static int btrfs_readdir(struct file *filp, void *dirent, filldir_t filldir) return 0; filp->f_pos = 1; } - - key.objectid = inode->i_ino; - path = btrfs_alloc_path(); - path->reada = 2; - /* special case for .., just use the back ref */ if (filp->f_pos == 1) { - btrfs_set_key_type(&key, BTRFS_INODE_REF_KEY); - key.offset = (u64)-1; - ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); - if (ret < 0 || path->slots[0] == 0) { - btrfs_release_path(root, path); - goto read_dir_items; - } - BUG_ON(ret == 0); - leaf = path->nodes[0]; - slot = path->slots[0] - 1; - btrfs_item_key_to_cpu(leaf, &found_key, slot); - btrfs_release_path(root, path); - if (found_key.objectid != key.objectid || - found_key.type != BTRFS_INODE_REF_KEY) - goto read_dir_items; + u64 pino = parent_ino(filp->f_path.dentry); over = filldir(dirent, "..", 2, - 2, found_key.offset, DT_DIR); + 2, pino, DT_DIR); if (over) - goto nopos; + return 0; filp->f_pos = 2; } + path = btrfs_alloc_path(); + path->reada = 2; -read_dir_items: btrfs_set_key_type(&key, key_type); key.offset = filp->f_pos; + key.objectid = inode->i_ino; ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); if (ret < 0) goto err; advance = 0; - while(1) { + + while (1) { leaf = path->nodes[0]; nritems = btrfs_header_nritems(leaf); slot = path->slots[0]; if (advance || slot >= nritems) { - if (slot >= nritems -1) { + if (slot >= nritems - 1) { ret = btrfs_next_leaf(root, path); if (ret) break; @@ -2029,6 +3199,7 @@ read_dir_items: path->slots[0]++; } } + advance = 1; item = btrfs_item_nr(leaf, slot); btrfs_item_key_to_cpu(leaf, &found_key, slot); @@ -2041,43 +3212,58 @@ read_dir_items: continue; filp->f_pos = found_key.offset; - advance = 1; + di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item); di_cur = 0; di_total = btrfs_item_size(leaf, item); - while(di_cur < di_total) { + + while (di_cur < di_total) { struct btrfs_key location; name_len = btrfs_dir_name_len(leaf, di); - if (name_len < 32) { + if (name_len <= sizeof(tmp_name)) { name_ptr = tmp_name; } else { name_ptr = kmalloc(name_len, GFP_NOFS); - BUG_ON(!name_ptr); + if (!name_ptr) { + ret = -ENOMEM; + goto err; + } } read_extent_buffer(leaf, name_ptr, (unsigned long)(di + 1), name_len); 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 + */ + 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, + found_key.offset, location.objectid, d_type); +skip: if (name_ptr != tmp_name) kfree(name_ptr); if (over) goto nopos; di_len = btrfs_dir_name_len(leaf, di) + - btrfs_dir_data_len(leaf, di) +sizeof(*di); + btrfs_dir_data_len(leaf, di) + sizeof(*di); di_cur += di_len; di = (struct btrfs_dir_item *)((char *)di + di_len); } } + + /* Reached end of directory/root. Bump pos past the last item. */ if (key_type == BTRFS_DIR_INDEX_KEY) - filp->f_pos = INT_LIMIT(typeof(filp->f_pos)); + filp->f_pos = INT_LIMIT(off_t); else filp->f_pos++; nopos: @@ -2093,7 +3279,7 @@ int btrfs_write_inode(struct inode *inode, int wait) struct btrfs_trans_handle *trans; int ret = 0; - if (root->fs_info->closing > 1) + if (root->fs_info->btree_inode == inode) return 0; if (wait) { @@ -2121,6 +3307,11 @@ void btrfs_dirty_inode(struct inode *inode) btrfs_end_transaction(trans, root); } +/* + * find the highest existing sequence number in a directory + * and then set the in-memory index_cnt variable to reflect + * free sequence numbers + */ static int btrfs_set_inode_index_count(struct inode *inode) { struct btrfs_root *root = BTRFS_I(inode)->root; @@ -2173,8 +3364,11 @@ out: return ret; } -static int btrfs_set_inode_index(struct inode *dir, struct inode *inode, - u64 *index) +/* + * helper to find a free sequence number in a given directory. This current + * code is very simple, later versions will do smarter things in the btree + */ +int btrfs_set_inode_index(struct inode *dir, u64 *index) { int ret = 0; @@ -2194,14 +3388,11 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *dir, const char *name, int name_len, - u64 ref_objectid, - u64 objectid, - struct btrfs_block_group_cache *group, - int mode, u64 *index) + u64 ref_objectid, u64 objectid, + u64 alloc_hint, int mode, u64 *index) { struct inode *inode; struct btrfs_inode_item *inode_item; - struct btrfs_block_group_cache *new_inode_group; struct btrfs_key *location; struct btrfs_path *path; struct btrfs_inode_ref *ref; @@ -2219,7 +3410,7 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, return ERR_PTR(-ENOMEM); if (dir) { - ret = btrfs_set_inode_index(dir, inode, index); + ret = btrfs_set_inode_index(dir, index); if (ret) return ERR_PTR(ret); } @@ -2228,34 +3419,24 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, * btrfs_get_inode_index_count has an explanation for the magic * number */ + init_btrfs_i(inode); BTRFS_I(inode)->index_cnt = 2; - - extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS); - extent_io_tree_init(&BTRFS_I(inode)->io_tree, - inode->i_mapping, GFP_NOFS); - extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree, - inode->i_mapping, GFP_NOFS); - btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree); - INIT_LIST_HEAD(&BTRFS_I(inode)->delalloc_inodes); - mutex_init(&BTRFS_I(inode)->csum_mutex); - mutex_init(&BTRFS_I(inode)->extent_mutex); - BTRFS_I(inode)->delalloc_bytes = 0; - inode->i_mapping->writeback_index = 0; - BTRFS_I(inode)->disk_i_size = 0; BTRFS_I(inode)->root = root; + BTRFS_I(inode)->generation = trans->transid; + btrfs_set_inode_space_info(root, inode); if (mode & S_IFDIR) owner = 0; else owner = 1; - new_inode_group = btrfs_find_block_group(root, group, 0, - BTRFS_BLOCK_GROUP_METADATA, owner); - if (!new_inode_group) { - printk("find_block group failed\n"); - new_inode_group = group; + BTRFS_I(inode)->block_group = + btrfs_find_block_group(root, 0, alloc_hint, owner); + if ((mode & S_IFREG)) { + if (btrfs_test_opt(root, NODATASUM)) + btrfs_set_flag(inode, NODATASUM); + if (btrfs_test_opt(root, NODATACOW)) + btrfs_set_flag(inode, NODATACOW); } - BTRFS_I(inode)->block_group = new_inode_group; - BTRFS_I(inode)->flags = 0; key[0].objectid = objectid; btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY); @@ -2275,15 +3456,22 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, if (objectid > root->highest_inode) root->highest_inode = objectid; - inode->i_uid = current->fsuid; - inode->i_gid = current->fsgid; + inode->i_uid = current_fsuid(); + + if (dir && (dir->i_mode & S_ISGID)) { + inode->i_gid = dir->i_gid; + if (S_ISDIR(mode)) + mode |= S_ISGID; + } else + inode->i_gid = current_fsgid(); + inode->i_mode = mode; inode->i_ino = objectid; - inode->i_blocks = 0; + 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); - fill_inode_item(path->nodes[0], inode_item, inode); + 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); @@ -2314,38 +3502,40 @@ static inline u8 btrfs_inode_type(struct inode *inode) return btrfs_type_by_mode[(inode->i_mode & S_IFMT) >> S_SHIFT]; } -static int btrfs_add_link(struct btrfs_trans_handle *trans, - struct dentry *dentry, struct inode *inode, - int add_backref, u64 index) +/* + * utility function to add 'inode' into 'parent_inode' with + * a give name and a given sequence number. + * if 'add_backref' is true, also insert a backref from the + * inode to the parent directory. + */ +int btrfs_add_link(struct btrfs_trans_handle *trans, + struct inode *parent_inode, struct inode *inode, + const char *name, int name_len, int add_backref, u64 index) { int ret; struct btrfs_key key; - struct btrfs_root *root = BTRFS_I(dentry->d_parent->d_inode)->root; - struct inode *parent_inode = dentry->d_parent->d_inode; + struct btrfs_root *root = BTRFS_I(parent_inode)->root; key.objectid = inode->i_ino; btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); key.offset = 0; - ret = btrfs_insert_dir_item(trans, root, - dentry->d_name.name, dentry->d_name.len, - dentry->d_parent->d_inode->i_ino, + ret = btrfs_insert_dir_item(trans, root, name, name_len, + parent_inode->i_ino, &key, btrfs_inode_type(inode), index); if (ret == 0) { if (add_backref) { ret = btrfs_insert_inode_ref(trans, root, - dentry->d_name.name, - dentry->d_name.len, - inode->i_ino, - parent_inode->i_ino, - index); + name, name_len, + inode->i_ino, + parent_inode->i_ino, + index); } btrfs_i_size_write(parent_inode, parent_inode->i_size + - dentry->d_name.len * 2); + name_len * 2); parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME; - ret = btrfs_update_inode(trans, root, - dentry->d_parent->d_inode); + ret = btrfs_update_inode(trans, root, parent_inode); } return ret; } @@ -2354,7 +3544,9 @@ static int btrfs_add_nondir(struct btrfs_trans_handle *trans, struct dentry *dentry, struct inode *inode, int backref, u64 index) { - int err = btrfs_add_link(trans, dentry, inode, backref, 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; @@ -2379,7 +3571,7 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry, if (!new_valid_dev(rdev)) return -EINVAL; - err = btrfs_check_free_space(root, 1, 0); + err = btrfs_check_metadata_free_space(root); if (err) goto fail; @@ -2400,7 +3592,7 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry, if (IS_ERR(inode)) goto out_unlock; - err = btrfs_init_acl(inode, dir); + err = btrfs_init_inode_security(inode, dir); if (err) { drop_inode = 1; goto out_unlock; @@ -2442,7 +3634,7 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry, u64 objectid; u64 index = 0; - err = btrfs_check_free_space(root, 1, 0); + err = btrfs_check_metadata_free_space(root); if (err) goto fail; trans = btrfs_start_transaction(root, 1); @@ -2463,7 +3655,7 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry, if (IS_ERR(inode)) goto out_unlock; - err = btrfs_init_acl(inode, dir); + err = btrfs_init_inode_security(inode, dir); if (err) { drop_inode = 1; goto out_unlock; @@ -2478,19 +3670,7 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry, inode->i_mapping->backing_dev_info = &root->fs_info->bdi; inode->i_fop = &btrfs_file_operations; inode->i_op = &btrfs_file_inode_operations; - extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS); - extent_io_tree_init(&BTRFS_I(inode)->io_tree, - inode->i_mapping, GFP_NOFS); - extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree, - inode->i_mapping, GFP_NOFS); - INIT_LIST_HEAD(&BTRFS_I(inode)->delalloc_inodes); - mutex_init(&BTRFS_I(inode)->csum_mutex); - mutex_init(&BTRFS_I(inode)->extent_mutex); - BTRFS_I(inode)->delalloc_bytes = 0; - BTRFS_I(inode)->disk_i_size = 0; - inode->i_mapping->writeback_index = 0; BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; - btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree); } dir->i_sb->s_dirt = 1; btrfs_update_inode_block_group(trans, inode); @@ -2521,15 +3701,11 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir, if (inode->i_nlink == 0) return -ENOENT; -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18) - inode->i_nlink++; -#else - inc_nlink(inode); -#endif - err = btrfs_check_free_space(root, 1, 0); + btrfs_inc_nlink(inode); + err = btrfs_check_metadata_free_space(root); if (err) goto fail; - err = btrfs_set_inode_index(dir, inode, &index); + err = btrfs_set_inode_index(dir, &index); if (err) goto fail; @@ -2572,7 +3748,7 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) u64 index = 0; unsigned long nr = 1; - err = btrfs_check_free_space(root, 1, 0); + err = btrfs_check_metadata_free_space(root); if (err) goto out_unlock; @@ -2602,7 +3778,7 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) drop_on_err = 1; - err = btrfs_init_acl(inode, dir); + err = btrfs_init_inode_security(inode, dir); if (err) goto out_fail; @@ -2615,7 +3791,9 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) if (err) goto out_fail; - err = btrfs_add_link(trans, dentry, inode, 0, index); + err = btrfs_add_link(trans, dentry->d_parent->d_inode, + inode, dentry->d_name.name, + dentry->d_name.len, 0, index); if (err) goto out_fail; @@ -2636,6 +3814,10 @@ out_unlock: return err; } +/* helper for btfs_get_extent. Given an existing extent in the tree, + * and an extent that you want to insert, deal with overlap and insert + * the new extent into the tree. + */ static int merge_extent_mapping(struct extent_map_tree *em_tree, struct extent_map *existing, struct extent_map *em, @@ -2647,11 +3829,59 @@ static int merge_extent_mapping(struct extent_map_tree *em_tree, start_diff = map_start - em->start; em->start = map_start; em->len = map_len; - if (em->block_start < EXTENT_MAP_LAST_BYTE) + if (em->block_start < EXTENT_MAP_LAST_BYTE && + !test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) { em->block_start += start_diff; + em->block_len -= start_diff; + } return add_extent_mapping(em_tree, em); } +static noinline int uncompress_inline(struct btrfs_path *path, + struct inode *inode, struct page *page, + size_t pg_offset, u64 extent_offset, + struct btrfs_file_extent_item *item) +{ + int ret; + struct extent_buffer *leaf = path->nodes[0]; + char *tmp; + size_t max_size; + unsigned long inline_size; + unsigned long ptr; + + WARN_ON(pg_offset != 0); + 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])); + tmp = kmalloc(inline_size, GFP_NOFS); + 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); + } + kfree(tmp); + return 0; +} + +/* + * a bit scary, this does extent mapping from logical file offset to the disk. + * the ugly parts come from merging extents from the disk with the in-ram + * representation. This gets more complex because of the data=ordered code, + * where the in-ram extents might be locked pending data=ordered completion. + * + * This also copies inline extents directly into the page. + */ + struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, size_t pg_offset, u64 start, u64 len, int create) @@ -2672,6 +3902,7 @@ 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; again: spin_lock(&em_tree->lock); @@ -2695,7 +3926,9 @@ again: } em->bdev = root->fs_info->fs_devices->latest_bdev; em->start = EXTENT_MAP_HOLE; + em->orig_start = EXTENT_MAP_HOLE; em->len = (u64)-1; + em->block_len = (u64)-1; if (!path) { path = btrfs_alloc_path(); @@ -2728,76 +3961,103 @@ again: found_type = btrfs_file_extent_type(leaf, item); extent_start = found_key.offset; - if (found_type == BTRFS_FILE_EXTENT_REG) { + 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); - err = 0; - if (start < extent_start || start >= extent_end) { - em->start = start; - if (start < extent_start) { - if (start + len <= extent_start) - goto not_found; - em->len = extent_end - extent_start; - } else { - em->len = len; + } 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); + } + + if (start >= extent_end) { + path->slots[0]++; + if (path->slots[0] >= btrfs_header_nritems(leaf)) { + ret = btrfs_next_leaf(root, path); + if (ret < 0) { + err = ret; + goto out; } - goto not_found_em; + if (ret > 0) + goto not_found; + leaf = path->nodes[0]; } + btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); + if (found_key.objectid != objectid || + found_key.type != BTRFS_EXTENT_DATA_KEY) + goto not_found; + if (start + len <= found_key.offset) + goto not_found; + em->start = start; + em->len = found_key.offset - start; + goto not_found_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->start = extent_start; - em->len = extent_end - extent_start; em->block_start = EXTENT_MAP_HOLE; goto insert; } - bytenr += btrfs_file_extent_offset(leaf, item); - em->block_start = bytenr; - em->start = extent_start; - em->len = extent_end - extent_start; + 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) { - u64 page_start; unsigned long ptr; char *map; size_t size; size_t extent_offset; size_t copy_size; - size = btrfs_file_extent_inline_len(leaf, btrfs_item_nr(leaf, - path->slots[0])); - extent_end = (extent_start + size + root->sectorsize - 1) & - ~((u64)root->sectorsize - 1); - if (start < extent_start || start >= extent_end) { - em->start = start; - if (start < extent_start) { - if (start + len <= extent_start) - goto not_found; - em->len = extent_end - extent_start; - } else { - em->len = len; - } - goto not_found_em; - } em->block_start = EXTENT_MAP_INLINE; - - if (!page) { + if (!page || create) { em->start = extent_start; - em->len = size; + em->len = extent_end - extent_start; goto out; } - page_start = page_offset(page) + pg_offset; - extent_offset = page_start - extent_start; + size = btrfs_file_extent_inline_len(leaf, 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); - map = kmap(page); + em->orig_start = EXTENT_MAP_INLINE; + if (compressed) + set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); ptr = btrfs_file_extent_inline_start(item) + extent_offset; if (create == 0 && !PageUptodate(page)) { - read_extent_buffer(leaf, map + pg_offset, ptr, - copy_size); + if (btrfs_file_extent_compression(leaf, item) == + BTRFS_COMPRESS_ZLIB) { + ret = uncompress_inline(path, inode, page, + pg_offset, + extent_offset, item); + BUG_ON(ret); + } else { + map = kmap(page); + read_extent_buffer(leaf, map + pg_offset, ptr, + copy_size); + kunmap(page); + } flush_dcache_page(page); } else if (create && PageUptodate(page)) { if (!trans) { @@ -2808,16 +4068,17 @@ again: trans = btrfs_join_transaction(root, 1); goto again; } + map = kmap(page); write_extent_buffer(leaf, map + pg_offset, ptr, copy_size); + kunmap(page); btrfs_mark_buffer_dirty(leaf); } - kunmap(page); set_extent_uptodate(io_tree, em->start, extent_map_end(em) - 1, GFP_NOFS); goto insert; } else { - printk("unkknown found_type %d\n", found_type); + printk(KERN_ERR "btrfs unknown found_type %d\n", found_type); WARN_ON(1); } not_found: @@ -2825,10 +4086,15 @@ not_found: 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); if (em->start > start || extent_map_end(em) <= start) { - printk("bad extent! em: [%Lu %Lu] passed [%Lu %Lu]\n", em->start, em->len, start, len); + 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); err = -EIO; goto out; } @@ -2865,8 +4131,6 @@ insert: } } else { err = -EIO; - printk("failing to insert %Lu %Lu\n", - start, len); free_extent_map(em); em = NULL; } @@ -2882,9 +4146,8 @@ out: btrfs_free_path(path); if (trans) { ret = btrfs_end_transaction(trans, root); - if (!err) { + if (!err) err = ret; - } } if (err) { free_extent_map(em); @@ -2894,81 +4157,17 @@ out: return em; } -#if 0 /* waiting for O_DIRECT reads */ -static int btrfs_get_block(struct inode *inode, sector_t iblock, - struct buffer_head *bh_result, int create) -{ - struct extent_map *em; - u64 start = (u64)iblock << inode->i_blkbits; - struct btrfs_multi_bio *multi = NULL; - struct btrfs_root *root = BTRFS_I(inode)->root; - u64 len; - u64 logical; - u64 map_length; - int ret = 0; - - em = btrfs_get_extent(inode, NULL, 0, start, bh_result->b_size, 0); - - if (!em || IS_ERR(em)) - goto out; - - if (em->start > start || em->start + em->len <= start) { - goto out; - } - - if (em->block_start == EXTENT_MAP_INLINE) { - ret = -EINVAL; - goto out; - } - - len = em->start + em->len - start; - len = min_t(u64, len, INT_LIMIT(typeof(bh_result->b_size))); - - if (em->block_start == EXTENT_MAP_HOLE || - em->block_start == EXTENT_MAP_DELALLOC) { - bh_result->b_size = len; - goto out; - } - - logical = start - em->start; - logical = em->block_start + logical; - - map_length = len; - ret = btrfs_map_block(&root->fs_info->mapping_tree, READ, - logical, &map_length, &multi, 0); - BUG_ON(ret); - bh_result->b_blocknr = multi->stripes[0].physical >> inode->i_blkbits; - bh_result->b_size = min(map_length, len); - - bh_result->b_bdev = multi->stripes[0].dev->bdev; - set_buffer_mapped(bh_result); - kfree(multi); -out: - free_extent_map(em); - return ret; -} -#endif - static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov, loff_t offset, unsigned long nr_segs) { return -EINVAL; -#if 0 - struct file *file = iocb->ki_filp; - struct inode *inode = file->f_mapping->host; - - if (rw == WRITE) - return -EINVAL; - - return blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov, - offset, nr_segs, btrfs_get_block, NULL); -#endif } -static sector_t btrfs_bmap(struct address_space *mapping, sector_t iblock) +static int btrfs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, + __u64 start, __u64 len) { - return extent_bmap(mapping, iblock, btrfs_get_extent); + return extent_fiemap(inode, fieinfo, start, len, btrfs_get_extent); } int btrfs_readpage(struct file *file, struct page *page) @@ -2996,6 +4195,7 @@ int btrfs_writepages(struct address_space *mapping, struct writeback_control *wbc) { struct extent_io_tree *tree; + tree = &BTRFS_I(mapping->host)->io_tree; return extent_writepages(tree, mapping, btrfs_get_extent, wbc); } @@ -3028,7 +4228,9 @@ static int __btrfs_releasepage(struct page *page, gfp_t gfp_flags) static int btrfs_releasepage(struct page *page, gfp_t gfp_flags) { - return __btrfs_releasepage(page, gfp_flags); + if (PageWriteback(page) || PageDirty(page)) + return 0; + return __btrfs_releasepage(page, gfp_flags & GFP_NOFS); } static void btrfs_invalidatepage(struct page *page, unsigned long offset) @@ -3090,8 +4292,9 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) * beyond EOF, then the page is guaranteed safe against truncation until we * unlock the page. */ -int btrfs_page_mkwrite(struct vm_area_struct *vma, struct page *page) +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 btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; @@ -3103,7 +4306,7 @@ int btrfs_page_mkwrite(struct vm_area_struct *vma, struct page *page) u64 page_start; u64 page_end; - ret = btrfs_check_free_space(root, PAGE_CACHE_SIZE, 0); + ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE); if (ret) goto out; @@ -3116,6 +4319,7 @@ again: if ((page->mapping != inode->i_mapping) || (page_start >= size)) { + btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE); /* page got truncated out from underneath us */ goto out_unlock; } @@ -3159,6 +4363,8 @@ again: out_unlock: unlock_page(page); out: + if (ret) + ret = VM_FAULT_SIGBUS; return ret; } @@ -3186,7 +4392,7 @@ static void btrfs_truncate(struct inode *inode) if (ret) goto out; /* FIXME, add redo link to tree so we don't leak on crash */ - ret = btrfs_truncate_in_trans(trans, root, inode, + ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, BTRFS_EXTENT_DATA_KEY); btrfs_update_inode(trans, root, inode); @@ -3201,64 +4407,45 @@ out: } /* - * Invalidate a single dcache entry at the root of the filesystem. - * Needed after creation of snapshot or subvolume. + * create a new subvolume directory/inode (helper for the ioctl). */ -void btrfs_invalidate_dcache_root(struct btrfs_root *root, char *name, - int namelen) -{ - struct dentry *alias, *entry; - struct qstr qstr; - - alias = d_find_alias(root->fs_info->sb->s_root->d_inode); - if (alias) { - qstr.name = name; - qstr.len = namelen; - /* change me if btrfs ever gets a d_hash operation */ - qstr.hash = full_name_hash(qstr.name, qstr.len); - entry = d_lookup(alias, &qstr); - dput(alias); - if (entry) { - d_invalidate(entry); - dput(entry); - } - } -} - -int btrfs_create_subvol_root(struct btrfs_root *new_root, - struct btrfs_trans_handle *trans, u64 new_dirid, - struct btrfs_block_group_cache *block_group) +int btrfs_create_subvol_root(struct btrfs_trans_handle *trans, + struct btrfs_root *new_root, struct dentry *dentry, + u64 new_dirid, u64 alloc_hint) { struct inode *inode; + int error; u64 index = 0; inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, new_dirid, - new_dirid, block_group, S_IFDIR | 0700, &index); + new_dirid, alloc_hint, S_IFDIR | 0700, &index); if (IS_ERR(inode)) return PTR_ERR(inode); inode->i_op = &btrfs_dir_inode_operations; inode->i_fop = &btrfs_dir_file_operations; - new_root->inode = inode; inode->i_nlink = 1; btrfs_i_size_write(inode, 0); - return btrfs_update_inode(trans, new_root, inode); + error = btrfs_update_inode(trans, new_root, inode); + if (error) + return error; + + d_instantiate(dentry, 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; -#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23) - offset = page_cache_readahead(mapping, ra, file, offset, req_size); - return offset; -#else page_cache_sync_readahead(mapping, ra, file, offset, req_size); return offset + req_size; -#endif } struct inode *btrfs_alloc_inode(struct super_block *sb) @@ -3269,6 +4456,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb) if (!ei) return NULL; ei->last_trans = 0; + ei->logged_trans = 0; btrfs_ordered_inode_tree_init(&ei->ordered_tree); ei->i_acl = BTRFS_ACL_NOT_CACHED; ei->i_default_acl = BTRFS_ACL_NOT_CACHED; @@ -3297,30 +4485,25 @@ void btrfs_destroy_inode(struct inode *inode) } spin_unlock(&BTRFS_I(inode)->root->list_lock); - while(1) { + while (1) { ordered = btrfs_lookup_first_ordered_extent(inode, (u64)-1); if (!ordered) break; else { - printk("found ordered extent %Lu %Lu\n", - ordered->file_offset, ordered->len); + 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_remove_ordered_extent(inode, ordered); btrfs_put_ordered_extent(ordered); btrfs_put_ordered_extent(ordered); } } - btrfs_drop_extent_cache(inode, 0, (u64)-1); + btrfs_drop_extent_cache(inode, 0, (u64)-1, 0); kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode)); } -#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,26) static void init_once(void *foo) -#elif LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23) -static void init_once(struct kmem_cache * cachep, void *foo) -#else -static void init_once(void * foo, struct kmem_cache * cachep, - unsigned long flags) -#endif { struct btrfs_inode *ei = (struct btrfs_inode *) foo; @@ -3343,22 +4526,10 @@ void btrfs_destroy_cachep(void) struct kmem_cache *btrfs_cache_create(const char *name, size_t size, unsigned long extra_flags, -#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,26) - void (*ctor)(void *) -#elif LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23) - void (*ctor)(struct kmem_cache *, void *) -#else - void (*ctor)(void *, struct kmem_cache *, - unsigned long) -#endif - ) + void (*ctor)(void *)) { return kmem_cache_create(name, size, 0, (SLAB_RECLAIM_ACCOUNT | - SLAB_MEM_SPREAD | extra_flags), ctor -#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23) - ,NULL -#endif - ); + SLAB_MEM_SPREAD | extra_flags), ctor); } int btrfs_init_cachep(void) @@ -3399,13 +4570,15 @@ static int btrfs_getattr(struct vfsmount *mnt, { struct inode *inode = dentry->d_inode; generic_fillattr(inode, stat); + stat->dev = BTRFS_I(inode)->root->anon_super.s_dev; stat->blksize = PAGE_CACHE_SIZE; - stat->blocks = inode->i_blocks + (BTRFS_I(inode)->delalloc_bytes >> 9); + stat->blocks = (inode_get_bytes(inode) + + BTRFS_I(inode)->delalloc_bytes) >> 9; return 0; } -static int btrfs_rename(struct inode * old_dir, struct dentry *old_dentry, - struct inode * new_dir,struct dentry *new_dentry) +static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, + struct inode *new_dir, struct dentry *new_dentry) { struct btrfs_trans_handle *trans; struct btrfs_root *root = BTRFS_I(old_dir)->root; @@ -3415,12 +4588,23 @@ static int btrfs_rename(struct inode * old_dir, struct dentry *old_dentry, u64 index = 0; int ret; + /* we're not allowed to rename between subvolumes */ + if (BTRFS_I(old_inode)->root->root_key.objectid != + BTRFS_I(new_dir)->root->root_key.objectid) + return -EXDEV; + if (S_ISDIR(old_inode->i_mode) && new_inode && new_inode->i_size > BTRFS_EMPTY_DIR_SIZE) { return -ENOTEMPTY; } - ret = btrfs_check_free_space(root, 1, 0); + /* to rename a snapshot or subvolume, we need to juggle the + * backrefs. This isn't coded yet + */ + if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) + return -EXDEV; + + ret = btrfs_check_metadata_free_space(root); if (ret) goto out_unlock; @@ -3428,31 +4612,39 @@ static int btrfs_rename(struct inode * old_dir, struct dentry *old_dentry, btrfs_set_trans_block_group(trans, new_dir); - old_dentry->d_inode->i_nlink++; + btrfs_inc_nlink(old_dentry->d_inode); old_dir->i_ctime = old_dir->i_mtime = ctime; new_dir->i_ctime = new_dir->i_mtime = ctime; old_inode->i_ctime = ctime; - ret = btrfs_unlink_trans(trans, root, old_dir, old_dentry); + ret = btrfs_unlink_inode(trans, root, old_dir, old_dentry->d_inode, + old_dentry->d_name.name, + old_dentry->d_name.len); if (ret) goto out_fail; if (new_inode) { new_inode->i_ctime = CURRENT_TIME; - ret = btrfs_unlink_trans(trans, root, new_dir, new_dentry); + ret = btrfs_unlink_inode(trans, root, new_dir, + new_dentry->d_inode, + new_dentry->d_name.name, + new_dentry->d_name.len); if (ret) goto out_fail; if (new_inode->i_nlink == 0) { - ret = btrfs_orphan_add(trans, new_inode); + ret = btrfs_orphan_add(trans, new_dentry->d_inode); if (ret) goto out_fail; } + } - ret = btrfs_set_inode_index(new_dir, old_inode, &index); + ret = btrfs_set_inode_index(new_dir, &index); if (ret) goto out_fail; - ret = btrfs_add_link(trans, new_dentry, old_inode, 1, index); + ret = btrfs_add_link(trans, new_dentry->d_parent->d_inode, + old_inode, new_dentry->d_name.name, + new_dentry->d_name.len, 1, index); if (ret) goto out_fail; @@ -3462,23 +4654,48 @@ out_unlock: return ret; } +/* + * 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) { struct list_head *head = &root->fs_info->delalloc_inodes; struct btrfs_inode *binode; - unsigned long flags; + struct inode *inode; - spin_lock_irqsave(&root->fs_info->delalloc_lock, flags); - while(!list_empty(head)) { + if (root->fs_info->sb->s_flags & MS_RDONLY) + return -EROFS; + + spin_lock(&root->fs_info->delalloc_lock); + while (!list_empty(head)) { binode = list_entry(head->next, struct btrfs_inode, delalloc_inodes); - atomic_inc(&binode->vfs_inode.i_count); - spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags); - filemap_write_and_wait(binode->vfs_inode.i_mapping); - iput(&binode->vfs_inode); - spin_lock_irqsave(&root->fs_info->delalloc_lock, flags); + 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); + iput(inode); + } + cond_resched(); + spin_lock(&root->fs_info->delalloc_lock); + } + spin_unlock(&root->fs_info->delalloc_lock); + + /* 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 + */ + atomic_inc(&root->fs_info->async_submit_draining); + while (atomic_read(&root->fs_info->nr_async_submits) || + atomic_read(&root->fs_info->async_delalloc_pages)) { + wait_event(root->fs_info->async_submit_wait, + (atomic_read(&root->fs_info->nr_async_submits) == 0 && + atomic_read(&root->fs_info->async_delalloc_pages) == 0)); } - spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags); + atomic_dec(&root->fs_info->async_submit_draining); return 0; } @@ -3505,7 +4722,7 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root)) return -ENAMETOOLONG; - err = btrfs_check_free_space(root, 1, 0); + err = btrfs_check_metadata_free_space(root); if (err) goto out_fail; @@ -3527,7 +4744,7 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, if (IS_ERR(inode)) goto out_unlock; - err = btrfs_init_acl(inode, dir); + err = btrfs_init_inode_security(inode, dir); if (err) { drop_inode = 1; goto out_unlock; @@ -3542,19 +4759,7 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, inode->i_mapping->backing_dev_info = &root->fs_info->bdi; inode->i_fop = &btrfs_file_operations; inode->i_op = &btrfs_file_inode_operations; - extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS); - extent_io_tree_init(&BTRFS_I(inode)->io_tree, - inode->i_mapping, GFP_NOFS); - extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree, - inode->i_mapping, GFP_NOFS); - INIT_LIST_HEAD(&BTRFS_I(inode)->delalloc_inodes); - mutex_init(&BTRFS_I(inode)->csum_mutex); - mutex_init(&BTRFS_I(inode)->extent_mutex); - BTRFS_I(inode)->delalloc_bytes = 0; - BTRFS_I(inode)->disk_i_size = 0; - inode->i_mapping->writeback_index = 0; BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; - btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree); } dir->i_sb->s_dirt = 1; btrfs_update_inode_block_group(trans, inode); @@ -3580,6 +4785,11 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, btrfs_set_file_extent_generation(leaf, ei, trans->transid); btrfs_set_file_extent_type(leaf, ei, BTRFS_FILE_EXTENT_INLINE); + btrfs_set_file_extent_encryption(leaf, ei, 0); + btrfs_set_file_extent_compression(leaf, ei, 0); + btrfs_set_file_extent_other_encoding(leaf, ei, 0); + btrfs_set_file_extent_ram_bytes(leaf, ei, name_len); + ptr = btrfs_file_extent_inline_start(ei); write_extent_buffer(leaf, symname, ptr, name_len); btrfs_mark_buffer_dirty(leaf); @@ -3588,6 +4798,7 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, inode->i_op = &btrfs_symlink_inode_operations; 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); err = btrfs_update_inode(trans, root, inode); if (err) @@ -3605,17 +4816,136 @@ out_fail: return err; } +static int prealloc_file_range(struct inode *inode, u64 start, u64 end, + u64 alloc_hint, int mode) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_key ins; + u64 alloc_size; + u64 cur_offset = start; + u64 num_bytes = end - start; + int ret = 0; + + trans = btrfs_join_transaction(root, 1); + BUG_ON(!trans); + btrfs_set_trans_block_group(trans, inode); + + while (num_bytes > 0) { + alloc_size = min(num_bytes, root->fs_info->max_extent); + ret = btrfs_reserve_extent(trans, root, alloc_size, + root->sectorsize, 0, alloc_hint, + (u64)-1, &ins, 1); + if (ret) { + WARN_ON(1); + goto out; + } + ret = insert_reserved_file_extent(trans, inode, + cur_offset, ins.objectid, + ins.offset, ins.offset, + ins.offset, 0, 0, 0, + BTRFS_FILE_EXTENT_PREALLOC); + BUG_ON(ret); + num_bytes -= ins.offset; + cur_offset += ins.offset; + alloc_hint = ins.objectid + ins.offset; + } +out: + if (cur_offset > start) { + inode->i_ctime = CURRENT_TIME; + btrfs_set_flag(inode, PREALLOC); + if (!(mode & FALLOC_FL_KEEP_SIZE) && + cur_offset > i_size_read(inode)) + btrfs_i_size_write(inode, cur_offset); + ret = btrfs_update_inode(trans, root, inode); + BUG_ON(ret); + } + + btrfs_end_transaction(trans, root); + return ret; +} + +static long btrfs_fallocate(struct inode *inode, int mode, + loff_t offset, loff_t len) +{ + u64 cur_offset; + u64 last_byte; + u64 alloc_start; + u64 alloc_end; + u64 alloc_hint = 0; + u64 mask = BTRFS_I(inode)->root->sectorsize - 1; + struct extent_map *em; + int ret; + + alloc_start = offset & ~mask; + alloc_end = (offset + len + mask) & ~mask; + + mutex_lock(&inode->i_mutex); + if (alloc_start > inode->i_size) { + ret = btrfs_cont_expand(inode, alloc_start); + if (ret) + goto out; + } + + while (1) { + struct btrfs_ordered_extent *ordered; + lock_extent(&BTRFS_I(inode)->io_tree, alloc_start, + alloc_end - 1, 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(&BTRFS_I(inode)->io_tree, + alloc_start, alloc_end - 1, GFP_NOFS); + btrfs_wait_ordered_range(inode, alloc_start, + alloc_end - alloc_start); + } else { + if (ordered) + btrfs_put_ordered_extent(ordered); + break; + } + } + + 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) { + ret = prealloc_file_range(inode, cur_offset, + last_byte, alloc_hint, mode); + if (ret < 0) { + free_extent_map(em); + break; + } + } + if (em->block_start <= EXTENT_MAP_LAST_BYTE) + alloc_hint = em->block_start; + free_extent_map(em); + + cur_offset = last_byte; + if (cur_offset >= alloc_end) { + ret = 0; + break; + } + } + unlock_extent(&BTRFS_I(inode)->io_tree, alloc_start, alloc_end - 1, + GFP_NOFS); +out: + mutex_unlock(&inode->i_mutex); + return ret; +} + static int btrfs_set_page_dirty(struct page *page) { return __set_page_dirty_nobuffers(page); } -#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,26) static int btrfs_permission(struct inode *inode, int mask) -#else -static int btrfs_permission(struct inode *inode, int mask, - struct nameidata *nd) -#endif { if (btrfs_test_flag(inode, READONLY) && (mask & MAY_WRITE)) return -EACCES; @@ -3623,6 +4953,7 @@ static int btrfs_permission(struct inode *inode, int mask, } static struct inode_operations btrfs_dir_inode_operations = { + .getattr = btrfs_getattr, .lookup = btrfs_lookup, .create = btrfs_create, .unlink = btrfs_unlink, @@ -3633,10 +4964,10 @@ static struct inode_operations btrfs_dir_inode_operations = { .symlink = btrfs_symlink, .setattr = btrfs_setattr, .mknod = btrfs_mknod, - .setxattr = generic_setxattr, - .getxattr = generic_getxattr, + .setxattr = btrfs_setxattr, + .getxattr = btrfs_getxattr, .listxattr = btrfs_listxattr, - .removexattr = generic_removexattr, + .removexattr = btrfs_removexattr, .permission = btrfs_permission, }; static struct inode_operations btrfs_dir_ro_inode_operations = { @@ -3646,19 +4977,19 @@ static struct inode_operations btrfs_dir_ro_inode_operations = { static struct file_operations btrfs_dir_file_operations = { .llseek = generic_file_llseek, .read = generic_read_dir, - .readdir = btrfs_readdir, + .readdir = btrfs_real_readdir, .unlocked_ioctl = btrfs_ioctl, #ifdef CONFIG_COMPAT .compat_ioctl = btrfs_ioctl, #endif .release = btrfs_release_file, + .fsync = btrfs_sync_file, }; static struct extent_io_ops btrfs_extent_io_ops = { .fill_delalloc = run_delalloc_range, .submit_bio_hook = btrfs_submit_bio_hook, .merge_bio_hook = btrfs_merge_bio_hook, - .readpage_io_hook = btrfs_readpage_io_hook, .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, @@ -3667,13 +4998,24 @@ static struct extent_io_ops btrfs_extent_io_ops = { .clear_bit_hook = btrfs_clear_bit_hook, }; +/* + * btrfs doesn't support the bmap operation because swapfiles + * use bmap to make a mapping of extents in the file. They assume + * these extents won't change over the life of the file and they + * use the bmap result to do IO directly to the drive. + * + * the btrfs bmap call would return logical addresses that aren't + * suitable for IO and they also will change frequently as COW + * operations happen. So, swapfile + btrfs == corruption. + * + * For now we're avoiding this by dropping bmap. + */ static struct address_space_operations btrfs_aops = { .readpage = btrfs_readpage, .writepage = btrfs_writepage, .writepages = btrfs_writepages, .readpages = btrfs_readpages, .sync_page = block_sync_page, - .bmap = btrfs_bmap, .direct_IO = btrfs_direct_IO, .invalidatepage = btrfs_invalidatepage, .releasepage = btrfs_releasepage, @@ -3691,24 +5033,30 @@ static struct inode_operations btrfs_file_inode_operations = { .truncate = btrfs_truncate, .getattr = btrfs_getattr, .setattr = btrfs_setattr, - .setxattr = generic_setxattr, - .getxattr = generic_getxattr, + .setxattr = btrfs_setxattr, + .getxattr = btrfs_getxattr, .listxattr = btrfs_listxattr, - .removexattr = generic_removexattr, + .removexattr = btrfs_removexattr, .permission = btrfs_permission, + .fallocate = btrfs_fallocate, + .fiemap = btrfs_fiemap, }; static struct inode_operations btrfs_special_inode_operations = { .getattr = btrfs_getattr, .setattr = btrfs_setattr, .permission = btrfs_permission, - .setxattr = generic_setxattr, - .getxattr = generic_getxattr, + .setxattr = btrfs_setxattr, + .getxattr = btrfs_getxattr, .listxattr = btrfs_listxattr, - .removexattr = generic_removexattr, + .removexattr = btrfs_removexattr, }; static struct inode_operations btrfs_symlink_inode_operations = { .readlink = generic_readlink, .follow_link = page_follow_link_light, .put_link = page_put_link, .permission = btrfs_permission, + .setxattr = btrfs_setxattr, + .getxattr = btrfs_getxattr, + .listxattr = btrfs_listxattr, + .removexattr = btrfs_removexattr, };