From: Chris Mason Date: Thu, 24 Jan 2008 21:13:08 +0000 (-0500) Subject: Btrfs: Split the extent_map code into two parts X-Git-Url: https://git.stricted.de/?a=commitdiff_plain;h=d1310b2e0cd98eb1348553e69b73827b436dca7b;p=GitHub%2Fmoto-9609%2Fandroid_kernel_motorola_exynos9610.git Btrfs: Split the extent_map code into two parts There is now extent_map for mapping offsets in the file to disk and extent_io for state tracking, IO submission and extent_bufers. The new extent_map code shifts from [start,end] pairs to [start,len], and pushes the locking out into the caller. This allows a few performance optimizations and is easier to use. A number of extent_map usage bugs were fixed, mostly with failing to remove extent_map entries when changing the file. Signed-off-by: Chris Mason --- diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile index 72cc3e16741f..9c61609f3714 100644 --- a/fs/btrfs/Makefile +++ b/fs/btrfs/Makefile @@ -5,7 +5,8 @@ obj-m := btrfs.o btrfs-y := super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \ hash.o file-item.o inode-item.o inode-map.o disk-io.o \ transaction.o bit-radix.o inode.o file.o tree-defrag.o \ - extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o + extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o \ + extent_io.o ifeq ($(CONFIG_FS_POSIX_ACL),y) btrfs-y += acl.o diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h index 9fc99883c70a..9a0647dc5a0b 100644 --- a/fs/btrfs/btrfs_inode.h +++ b/fs/btrfs/btrfs_inode.h @@ -20,6 +20,7 @@ #define __BTRFS_I__ #include "extent_map.h" +#include "extent_io.h" /* in memory btrfs inode */ struct btrfs_inode { @@ -27,6 +28,7 @@ struct btrfs_inode { struct btrfs_block_group_cache *block_group; struct btrfs_key location; struct extent_map_tree extent_tree; + struct extent_io_tree io_tree; struct inode vfs_inode; u64 ordered_trans; diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index b9f2975b55c9..6c65473e0fe3 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -27,6 +27,7 @@ #include #include #include "bit-radix.h" +#include "extent_io.h" #include "extent_map.h" struct btrfs_trans_handle; @@ -314,11 +315,11 @@ struct btrfs_fs_info { struct btrfs_root *tree_root; struct radix_tree_root fs_roots_radix; - struct extent_map_tree free_space_cache; - struct extent_map_tree block_group_cache; - struct extent_map_tree pinned_extents; - struct extent_map_tree pending_del; - struct extent_map_tree extent_ins; + struct extent_io_tree free_space_cache; + struct extent_io_tree block_group_cache; + struct extent_io_tree pinned_extents; + struct extent_io_tree pending_del; + struct extent_io_tree extent_ins; u64 generation; u64 last_trans_committed; @@ -956,7 +957,7 @@ u32 btrfs_count_snapshots_in_path(struct btrfs_root *root, u64 first_extent); int btrfs_extent_post_op(struct btrfs_trans_handle *trans, struct btrfs_root *root); -int btrfs_copy_pinned(struct btrfs_root *root, struct extent_map_tree *copy); +int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy); struct btrfs_block_group_cache *btrfs_lookup_block_group(struct btrfs_fs_info *info, u64 bytenr); @@ -1001,7 +1002,7 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root u64 owner_objectid, u64 owner_offset, int pin); int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct extent_map_tree *unpin); + struct extent_io_tree *unpin); int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 5d1f9bca2712..4c4ebea0b2a9 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -43,14 +43,14 @@ static int check_tree_block(struct btrfs_root *root, struct extent_buffer *buf) } #endif -static struct extent_map_ops btree_extent_map_ops; +static struct extent_io_ops btree_extent_io_ops; struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize) { struct inode *btree_inode = root->fs_info->btree_inode; struct extent_buffer *eb; - eb = find_extent_buffer(&BTRFS_I(btree_inode)->extent_tree, + eb = find_extent_buffer(&BTRFS_I(btree_inode)->io_tree, bytenr, blocksize, GFP_NOFS); return eb; } @@ -61,13 +61,13 @@ struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root, struct inode *btree_inode = root->fs_info->btree_inode; struct extent_buffer *eb; - eb = alloc_extent_buffer(&BTRFS_I(btree_inode)->extent_tree, + eb = alloc_extent_buffer(&BTRFS_I(btree_inode)->io_tree, bytenr, blocksize, NULL, GFP_NOFS); return eb; } struct extent_map *btree_get_extent(struct inode *inode, struct page *page, - size_t page_offset, u64 start, u64 end, + size_t page_offset, u64 start, u64 len, int create) { struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; @@ -75,7 +75,9 @@ struct extent_map *btree_get_extent(struct inode *inode, struct page *page, int ret; again: - em = lookup_extent_mapping(em_tree, start, end); + spin_lock(&em_tree->lock); + em = lookup_extent_mapping(em_tree, start, len); + spin_unlock(&em_tree->lock); if (em) { goto out; } @@ -85,11 +87,14 @@ again: goto out; } em->start = 0; - em->end = (i_size_read(inode) & ~((u64)PAGE_CACHE_SIZE -1)) - 1; + em->len = i_size_read(inode); em->block_start = 0; - em->block_end = em->end; em->bdev = inode->i_sb->s_bdev; + + spin_lock(&em_tree->lock); ret = add_extent_mapping(em_tree, em); + spin_unlock(&em_tree->lock); + if (ret == -EEXIST) { free_extent_map(em); em = NULL; @@ -175,13 +180,13 @@ static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf, int csum_dirty_buffer(struct btrfs_root *root, struct page *page) { - struct extent_map_tree *tree; + struct extent_io_tree *tree; u64 start = (u64)page->index << PAGE_CACHE_SHIFT; u64 found_start; int found_level; unsigned long len; struct extent_buffer *eb; - tree = &BTRFS_I(page->mapping->host)->extent_tree; + tree = &BTRFS_I(page->mapping->host)->io_tree; if (page->private == EXTENT_PAGE_PRIVATE) goto out; @@ -230,16 +235,16 @@ static int btree_writepage_io_hook(struct page *page, u64 start, u64 end) static int btree_writepage(struct page *page, struct writeback_control *wbc) { - struct extent_map_tree *tree; - tree = &BTRFS_I(page->mapping->host)->extent_tree; + struct extent_io_tree *tree; + tree = &BTRFS_I(page->mapping->host)->io_tree; return extent_write_full_page(tree, page, btree_get_extent, wbc); } static int btree_writepages(struct address_space *mapping, struct writeback_control *wbc) { - struct extent_map_tree *tree; - tree = &BTRFS_I(mapping->host)->extent_tree; + struct extent_io_tree *tree; + tree = &BTRFS_I(mapping->host)->io_tree; if (wbc->sync_mode == WB_SYNC_NONE) { u64 num_dirty; u64 start = 0; @@ -264,18 +269,20 @@ static int btree_writepages(struct address_space *mapping, int btree_readpage(struct file *file, struct page *page) { - struct extent_map_tree *tree; - tree = &BTRFS_I(page->mapping->host)->extent_tree; + struct extent_io_tree *tree; + tree = &BTRFS_I(page->mapping->host)->io_tree; return extent_read_full_page(tree, page, btree_get_extent); } static int btree_releasepage(struct page *page, gfp_t unused_gfp_flags) { - struct extent_map_tree *tree; + struct extent_io_tree *tree; + struct extent_map_tree *map; int ret; - tree = &BTRFS_I(page->mapping->host)->extent_tree; - ret = try_release_extent_mapping(tree, page); + tree = &BTRFS_I(page->mapping->host)->io_tree; + map = &BTRFS_I(page->mapping->host)->extent_tree; + ret = try_release_extent_mapping(map, tree, page); if (ret == 1) { ClearPagePrivate(page); set_page_private(page, 0); @@ -286,8 +293,8 @@ static int btree_releasepage(struct page *page, gfp_t unused_gfp_flags) static void btree_invalidatepage(struct page *page, unsigned long offset) { - struct extent_map_tree *tree; - tree = &BTRFS_I(page->mapping->host)->extent_tree; + struct extent_io_tree *tree; + tree = &BTRFS_I(page->mapping->host)->io_tree; extent_invalidatepage(tree, page, offset); btree_releasepage(page, GFP_NOFS); } @@ -331,7 +338,7 @@ int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize) buf = btrfs_find_create_tree_block(root, bytenr, blocksize); if (!buf) return 0; - read_extent_buffer_pages(&BTRFS_I(btree_inode)->extent_tree, + read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree, buf, 0, 0); free_extent_buffer(buf); return ret; @@ -342,40 +349,39 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr, { struct extent_buffer *buf = NULL; struct inode *btree_inode = root->fs_info->btree_inode; - struct extent_map_tree *extent_tree; + struct extent_io_tree *io_tree; u64 end; int ret; - extent_tree = &BTRFS_I(btree_inode)->extent_tree; + io_tree = &BTRFS_I(btree_inode)->io_tree; buf = btrfs_find_create_tree_block(root, bytenr, blocksize); if (!buf) return NULL; - read_extent_buffer_pages(&BTRFS_I(btree_inode)->extent_tree, - buf, 0, 1); + read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree, buf, 0, 1); if (buf->flags & EXTENT_CSUM) return buf; end = buf->start + PAGE_CACHE_SIZE - 1; - if (test_range_bit(extent_tree, buf->start, end, EXTENT_CSUM, 1)) { + if (test_range_bit(io_tree, buf->start, end, EXTENT_CSUM, 1)) { buf->flags |= EXTENT_CSUM; return buf; } - lock_extent(extent_tree, buf->start, end, GFP_NOFS); + lock_extent(io_tree, buf->start, end, GFP_NOFS); - if (test_range_bit(extent_tree, buf->start, end, EXTENT_CSUM, 1)) { + if (test_range_bit(io_tree, buf->start, end, EXTENT_CSUM, 1)) { buf->flags |= EXTENT_CSUM; goto out_unlock; } ret = csum_tree_block(root, buf, 1); - set_extent_bits(extent_tree, buf->start, end, EXTENT_CSUM, GFP_NOFS); + set_extent_bits(io_tree, buf->start, end, EXTENT_CSUM, GFP_NOFS); buf->flags |= EXTENT_CSUM; out_unlock: - unlock_extent(extent_tree, buf->start, end, GFP_NOFS); + unlock_extent(io_tree, buf->start, end, GFP_NOFS); return buf; } @@ -385,7 +391,7 @@ int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *btree_inode = root->fs_info->btree_inode; if (btrfs_header_generation(buf) == root->fs_info->running_transaction->transid) - clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->extent_tree, + clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, buf); return 0; } @@ -394,7 +400,7 @@ int wait_on_tree_block_writeback(struct btrfs_root *root, struct extent_buffer *buf) { struct inode *btree_inode = root->fs_info->btree_inode; - wait_on_extent_buffer_writeback(&BTRFS_I(btree_inode)->extent_tree, + wait_on_extent_buffer_writeback(&BTRFS_I(btree_inode)->io_tree, buf); return 0; } @@ -659,20 +665,23 @@ struct btrfs_root *open_ctree(struct super_block *sb) fs_info->btree_inode->i_nlink = 1; fs_info->btree_inode->i_size = sb->s_bdev->bd_inode->i_size; fs_info->btree_inode->i_mapping->a_ops = &btree_aops; - extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree, + extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree, fs_info->btree_inode->i_mapping, GFP_NOFS); - BTRFS_I(fs_info->btree_inode)->extent_tree.ops = &btree_extent_map_ops; + extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree, + GFP_NOFS); + + BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops; - extent_map_tree_init(&fs_info->free_space_cache, + extent_io_tree_init(&fs_info->free_space_cache, fs_info->btree_inode->i_mapping, GFP_NOFS); - extent_map_tree_init(&fs_info->block_group_cache, + extent_io_tree_init(&fs_info->block_group_cache, fs_info->btree_inode->i_mapping, GFP_NOFS); - extent_map_tree_init(&fs_info->pinned_extents, + extent_io_tree_init(&fs_info->pinned_extents, fs_info->btree_inode->i_mapping, GFP_NOFS); - extent_map_tree_init(&fs_info->pending_del, + extent_io_tree_init(&fs_info->pending_del, fs_info->btree_inode->i_mapping, GFP_NOFS); - extent_map_tree_init(&fs_info->extent_ins, + extent_io_tree_init(&fs_info->extent_ins, fs_info->btree_inode->i_mapping, GFP_NOFS); fs_info->do_barriers = 1; fs_info->closing = 0; @@ -787,7 +796,7 @@ int write_ctree_super(struct btrfs_trans_handle *trans, struct btrfs_root if (!btrfs_test_opt(root, NOBARRIER)) blkdev_issue_flush(sb->s_bdev, NULL); - set_extent_buffer_dirty(&BTRFS_I(btree_inode)->extent_tree, super); + set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, super); ret = sync_page_range_nolock(btree_inode, btree_inode->i_mapping, super->start, super->len); if (!btrfs_test_opt(root, NOBARRIER)) @@ -864,12 +873,12 @@ int close_ctree(struct btrfs_root *root) filemap_write_and_wait(fs_info->btree_inode->i_mapping); - extent_map_tree_empty_lru(&fs_info->free_space_cache); - extent_map_tree_empty_lru(&fs_info->block_group_cache); - extent_map_tree_empty_lru(&fs_info->pinned_extents); - extent_map_tree_empty_lru(&fs_info->pending_del); - extent_map_tree_empty_lru(&fs_info->extent_ins); - extent_map_tree_empty_lru(&BTRFS_I(fs_info->btree_inode)->extent_tree); + extent_io_tree_empty_lru(&fs_info->free_space_cache); + extent_io_tree_empty_lru(&fs_info->block_group_cache); + extent_io_tree_empty_lru(&fs_info->pinned_extents); + extent_io_tree_empty_lru(&fs_info->pending_del); + extent_io_tree_empty_lru(&fs_info->extent_ins); + extent_io_tree_empty_lru(&BTRFS_I(fs_info->btree_inode)->io_tree); truncate_inode_pages(fs_info->btree_inode->i_mapping, 0); @@ -892,13 +901,13 @@ int close_ctree(struct btrfs_root *root) int btrfs_buffer_uptodate(struct extent_buffer *buf) { struct inode *btree_inode = buf->first_page->mapping->host; - return extent_buffer_uptodate(&BTRFS_I(btree_inode)->extent_tree, buf); + return extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf); } int btrfs_set_buffer_uptodate(struct extent_buffer *buf) { struct inode *btree_inode = buf->first_page->mapping->host; - return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->extent_tree, + return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf); } @@ -914,7 +923,7 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf) transid, root->fs_info->generation); WARN_ON(1); } - set_extent_buffer_dirty(&BTRFS_I(btree_inode)->extent_tree, buf); + set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, buf); } void btrfs_throttle(struct btrfs_root *root) @@ -941,7 +950,7 @@ void btrfs_set_buffer_defrag(struct extent_buffer *buf) { struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; struct inode *btree_inode = root->fs_info->btree_inode; - set_extent_bits(&BTRFS_I(btree_inode)->extent_tree, buf->start, + set_extent_bits(&BTRFS_I(btree_inode)->io_tree, buf->start, buf->start + buf->len - 1, EXTENT_DEFRAG, GFP_NOFS); } @@ -949,7 +958,7 @@ void btrfs_set_buffer_defrag_done(struct extent_buffer *buf) { struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; struct inode *btree_inode = root->fs_info->btree_inode; - set_extent_bits(&BTRFS_I(btree_inode)->extent_tree, buf->start, + set_extent_bits(&BTRFS_I(btree_inode)->io_tree, buf->start, buf->start + buf->len - 1, EXTENT_DEFRAG_DONE, GFP_NOFS); } @@ -958,7 +967,7 @@ int btrfs_buffer_defrag(struct extent_buffer *buf) { struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; struct inode *btree_inode = root->fs_info->btree_inode; - return test_range_bit(&BTRFS_I(btree_inode)->extent_tree, + return test_range_bit(&BTRFS_I(btree_inode)->io_tree, buf->start, buf->start + buf->len - 1, EXTENT_DEFRAG, 0); } @@ -966,7 +975,7 @@ int btrfs_buffer_defrag_done(struct extent_buffer *buf) { struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; struct inode *btree_inode = root->fs_info->btree_inode; - return test_range_bit(&BTRFS_I(btree_inode)->extent_tree, + return test_range_bit(&BTRFS_I(btree_inode)->io_tree, buf->start, buf->start + buf->len - 1, EXTENT_DEFRAG_DONE, 0); } @@ -975,7 +984,7 @@ int btrfs_clear_buffer_defrag_done(struct extent_buffer *buf) { struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; struct inode *btree_inode = root->fs_info->btree_inode; - return clear_extent_bits(&BTRFS_I(btree_inode)->extent_tree, + return clear_extent_bits(&BTRFS_I(btree_inode)->io_tree, buf->start, buf->start + buf->len - 1, EXTENT_DEFRAG_DONE, GFP_NOFS); } @@ -984,7 +993,7 @@ int btrfs_clear_buffer_defrag(struct extent_buffer *buf) { struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; struct inode *btree_inode = root->fs_info->btree_inode; - return clear_extent_bits(&BTRFS_I(btree_inode)->extent_tree, + return clear_extent_bits(&BTRFS_I(btree_inode)->io_tree, buf->start, buf->start + buf->len - 1, EXTENT_DEFRAG, GFP_NOFS); } @@ -993,10 +1002,10 @@ int btrfs_read_buffer(struct extent_buffer *buf) { struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; struct inode *btree_inode = root->fs_info->btree_inode; - return read_extent_buffer_pages(&BTRFS_I(btree_inode)->extent_tree, + return read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree, buf, 0, 1); } -static struct extent_map_ops btree_extent_map_ops = { +static struct extent_io_ops btree_extent_io_ops = { .writepage_io_hook = btree_writepage_io_hook, }; diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index b69a46691a96..1cf125ab7822 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -63,7 +63,7 @@ static int cache_block_group(struct btrfs_root *root, int ret; struct btrfs_key key; struct extent_buffer *leaf; - struct extent_map_tree *free_space_cache; + struct extent_io_tree *free_space_cache; int slot; u64 last = 0; u64 hole_size; @@ -158,7 +158,7 @@ struct btrfs_block_group_cache *btrfs_lookup_block_group(struct btrfs_fs_info *info, u64 bytenr) { - struct extent_map_tree *block_group_cache; + struct extent_io_tree *block_group_cache; struct btrfs_block_group_cache *block_group = NULL; u64 ptr; u64 start; @@ -281,7 +281,7 @@ struct btrfs_block_group_cache *btrfs_find_block_group(struct btrfs_root *root, int data, int owner) { struct btrfs_block_group_cache *cache; - struct extent_map_tree *block_group_cache; + struct extent_io_tree *block_group_cache; struct btrfs_block_group_cache *found_group = NULL; struct btrfs_fs_info *info = root->fs_info; u64 used; @@ -951,7 +951,7 @@ fail: int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans, struct btrfs_root *root) { - struct extent_map_tree *block_group_cache; + struct extent_io_tree *block_group_cache; struct btrfs_block_group_cache *cache; int ret; int err = 0; @@ -1107,12 +1107,12 @@ static int update_pinned_extents(struct btrfs_root *root, return 0; } -int btrfs_copy_pinned(struct btrfs_root *root, struct extent_map_tree *copy) +int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy) { u64 last = 0; u64 start; u64 end; - struct extent_map_tree *pinned_extents = &root->fs_info->pinned_extents; + struct extent_io_tree *pinned_extents = &root->fs_info->pinned_extents; int ret; while(1) { @@ -1128,12 +1128,12 @@ int btrfs_copy_pinned(struct btrfs_root *root, struct extent_map_tree *copy) int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct extent_map_tree *unpin) + struct extent_io_tree *unpin) { u64 start; u64 end; int ret; - struct extent_map_tree *free_space_cache; + struct extent_io_tree *free_space_cache; free_space_cache = &root->fs_info->free_space_cache; while(1) { @@ -1329,8 +1329,8 @@ static int del_pending_extents(struct btrfs_trans_handle *trans, struct int err = 0; u64 start; u64 end; - struct extent_map_tree *pending_del; - struct extent_map_tree *pinned_extents; + struct extent_io_tree *pending_del; + struct extent_io_tree *pinned_extents; pending_del = &extent_root->fs_info->pending_del; pinned_extents = &extent_root->fs_info->pinned_extents; @@ -1802,7 +1802,7 @@ struct extent_buffer *__btrfs_alloc_free_block(struct btrfs_trans_handle *trans, set_extent_dirty(&trans->transaction->dirty_pages, buf->start, buf->start + buf->len - 1, GFP_NOFS); - set_extent_bits(&BTRFS_I(root->fs_info->btree_inode)->extent_tree, + set_extent_bits(&BTRFS_I(root->fs_info->btree_inode)->io_tree, buf->start, buf->start + buf->len - 1, EXTENT_CSUM, GFP_NOFS); buf->flags |= EXTENT_CSUM; @@ -2166,7 +2166,7 @@ static int noinline relocate_inode_pages(struct inode *inode, u64 start, unsigned long i; struct page *page; 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 = &BTRFS_I(inode)->io_tree; struct file_ra_state *ra; ra = kzalloc(sizeof(*ra), GFP_NOFS); @@ -2195,15 +2195,14 @@ static int noinline relocate_inode_pages(struct inode *inode, u64 start, page_start = (u64)page->index << PAGE_CACHE_SHIFT; page_end = page_start + PAGE_CACHE_SIZE - 1; - lock_extent(em_tree, page_start, page_end, GFP_NOFS); + lock_extent(io_tree, page_start, page_end, GFP_NOFS); delalloc_start = page_start; - existing_delalloc = - count_range_bits(&BTRFS_I(inode)->extent_tree, - &delalloc_start, page_end, - PAGE_CACHE_SIZE, EXTENT_DELALLOC); + existing_delalloc = count_range_bits(io_tree, + &delalloc_start, page_end, + PAGE_CACHE_SIZE, EXTENT_DELALLOC); - set_extent_delalloc(em_tree, page_start, + set_extent_delalloc(io_tree, page_start, page_end, GFP_NOFS); spin_lock(&root->fs_info->delalloc_lock); @@ -2211,7 +2210,7 @@ static int noinline relocate_inode_pages(struct inode *inode, u64 start, existing_delalloc; spin_unlock(&root->fs_info->delalloc_lock); - unlock_extent(em_tree, page_start, page_end, GFP_NOFS); + unlock_extent(io_tree, page_start, page_end, GFP_NOFS); set_page_dirty(page); unlock_page(page); page_cache_release(page); @@ -2379,7 +2378,7 @@ int btrfs_shrink_extent_tree(struct btrfs_root *root, u64 new_size) u64 cur_byte; u64 total_found; struct btrfs_fs_info *info = root->fs_info; - struct extent_map_tree *block_group_cache; + struct extent_io_tree *block_group_cache; struct btrfs_key key; struct btrfs_key found_key; struct extent_buffer *leaf; @@ -2561,7 +2560,7 @@ int btrfs_grow_extent_tree(struct btrfs_trans_handle *trans, struct btrfs_block_group_cache *cache; struct btrfs_block_group_item *item; struct btrfs_fs_info *info = root->fs_info; - struct extent_map_tree *block_group_cache; + struct extent_io_tree *block_group_cache; struct btrfs_key key; struct extent_buffer *leaf; int ret; @@ -2645,7 +2644,7 @@ int btrfs_read_block_groups(struct btrfs_root *root) int bit; struct btrfs_block_group_cache *cache; struct btrfs_fs_info *info = root->fs_info; - struct extent_map_tree *block_group_cache; + struct extent_io_tree *block_group_cache; struct btrfs_key key; struct btrfs_key found_key; struct extent_buffer *leaf; diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c new file mode 100644 index 000000000000..15cc158a0498 --- /dev/null +++ b/fs/btrfs/extent_io.c @@ -0,0 +1,3089 @@ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "extent_io.h" +#include "extent_map.h" + +/* temporary define until extent_map moves out of btrfs */ +struct kmem_cache *btrfs_cache_create(const char *name, size_t size, + unsigned long extra_flags, + void (*ctor)(void *, struct kmem_cache *, + unsigned long)); + +static struct kmem_cache *extent_state_cache; +static struct kmem_cache *extent_buffer_cache; + +static LIST_HEAD(buffers); +static LIST_HEAD(states); + +static spinlock_t state_lock = SPIN_LOCK_UNLOCKED; +#define BUFFER_LRU_MAX 64 + +struct tree_entry { + u64 start; + u64 end; + int in_tree; + struct rb_node rb_node; +}; + +struct extent_page_data { + struct bio *bio; + struct extent_io_tree *tree; + get_extent_t *get_extent; +}; + +int __init extent_io_init(void) +{ + extent_state_cache = btrfs_cache_create("extent_state", + sizeof(struct extent_state), 0, + NULL); + if (!extent_state_cache) + return -ENOMEM; + + extent_buffer_cache = btrfs_cache_create("extent_buffers", + sizeof(struct extent_buffer), 0, + NULL); + if (!extent_buffer_cache) + goto free_state_cache; + return 0; + +free_state_cache: + kmem_cache_destroy(extent_state_cache); + return -ENOMEM; +} + +void extent_io_exit(void) +{ + struct extent_state *state; + + while (!list_empty(&states)) { + state = list_entry(states.next, struct extent_state, list); + printk("state leak: start %Lu end %Lu state %lu in tree %d refs %d\n", state->start, state->end, state->state, state->in_tree, atomic_read(&state->refs)); + list_del(&state->list); + kmem_cache_free(extent_state_cache, state); + + } + + if (extent_state_cache) + kmem_cache_destroy(extent_state_cache); + if (extent_buffer_cache) + kmem_cache_destroy(extent_buffer_cache); +} + +void extent_io_tree_init(struct extent_io_tree *tree, + struct address_space *mapping, gfp_t mask) +{ + tree->state.rb_node = NULL; + tree->ops = NULL; + tree->dirty_bytes = 0; + rwlock_init(&tree->lock); + spin_lock_init(&tree->lru_lock); + tree->mapping = mapping; + INIT_LIST_HEAD(&tree->buffer_lru); + tree->lru_size = 0; +} +EXPORT_SYMBOL(extent_io_tree_init); + +void extent_io_tree_empty_lru(struct extent_io_tree *tree) +{ + struct extent_buffer *eb; + while(!list_empty(&tree->buffer_lru)) { + eb = list_entry(tree->buffer_lru.next, struct extent_buffer, + lru); + list_del_init(&eb->lru); + free_extent_buffer(eb); + } +} +EXPORT_SYMBOL(extent_io_tree_empty_lru); + +struct extent_state *alloc_extent_state(gfp_t mask) +{ + struct extent_state *state; + unsigned long flags; + + state = kmem_cache_alloc(extent_state_cache, mask); + if (!state || IS_ERR(state)) + return state; + state->state = 0; + state->in_tree = 0; + state->private = 0; + + spin_lock_irqsave(&state_lock, flags); + list_add(&state->list, &states); + spin_unlock_irqrestore(&state_lock, flags); + + atomic_set(&state->refs, 1); + init_waitqueue_head(&state->wq); + return state; +} +EXPORT_SYMBOL(alloc_extent_state); + +void free_extent_state(struct extent_state *state) +{ + unsigned long flags; + if (!state) + return; + if (atomic_dec_and_test(&state->refs)) { + WARN_ON(state->in_tree); + spin_lock_irqsave(&state_lock, flags); + list_del(&state->list); + spin_unlock_irqrestore(&state_lock, flags); + kmem_cache_free(extent_state_cache, state); + } +} +EXPORT_SYMBOL(free_extent_state); + +static struct rb_node *tree_insert(struct rb_root *root, u64 offset, + struct rb_node *node) +{ + struct rb_node ** p = &root->rb_node; + struct rb_node * parent = NULL; + struct tree_entry *entry; + + while(*p) { + parent = *p; + entry = rb_entry(parent, struct tree_entry, rb_node); + + if (offset < entry->start) + p = &(*p)->rb_left; + else if (offset > entry->end) + p = &(*p)->rb_right; + else + return parent; + } + + entry = rb_entry(node, struct tree_entry, rb_node); + entry->in_tree = 1; + rb_link_node(node, parent, p); + rb_insert_color(node, root); + return NULL; +} + +static struct rb_node *__tree_search(struct rb_root *root, u64 offset, + struct rb_node **prev_ret, + struct rb_node **next_ret) +{ + struct rb_node * n = root->rb_node; + struct rb_node *prev = NULL; + struct rb_node *orig_prev = NULL; + struct tree_entry *entry; + struct tree_entry *prev_entry = NULL; + + while(n) { + entry = rb_entry(n, struct tree_entry, rb_node); + prev = n; + prev_entry = entry; + + if (offset < entry->start) + n = n->rb_left; + else if (offset > entry->end) + n = n->rb_right; + else + return n; + } + + if (prev_ret) { + orig_prev = prev; + while(prev && offset > prev_entry->end) { + prev = rb_next(prev); + prev_entry = rb_entry(prev, struct tree_entry, rb_node); + } + *prev_ret = prev; + prev = orig_prev; + } + + if (next_ret) { + prev_entry = rb_entry(prev, struct tree_entry, rb_node); + while(prev && offset < prev_entry->start) { + prev = rb_prev(prev); + prev_entry = rb_entry(prev, struct tree_entry, rb_node); + } + *next_ret = prev; + } + return NULL; +} + +static inline struct rb_node *tree_search(struct rb_root *root, u64 offset) +{ + struct rb_node *prev; + struct rb_node *ret; + ret = __tree_search(root, offset, &prev, NULL); + if (!ret) + return prev; + return ret; +} + +/* + * utility function to look for merge candidates inside a given range. + * Any extents with matching state are merged together into a single + * extent in the tree. Extents with EXTENT_IO in their state field + * are not merged because the end_io handlers need to be able to do + * operations on them without sleeping (or doing allocations/splits). + * + * This should be called with the tree lock held. + */ +static int merge_state(struct extent_io_tree *tree, + struct extent_state *state) +{ + struct extent_state *other; + struct rb_node *other_node; + + if (state->state & EXTENT_IOBITS) + return 0; + + other_node = rb_prev(&state->rb_node); + if (other_node) { + other = rb_entry(other_node, struct extent_state, rb_node); + if (other->end == state->start - 1 && + other->state == state->state) { + state->start = other->start; + other->in_tree = 0; + rb_erase(&other->rb_node, &tree->state); + free_extent_state(other); + } + } + other_node = rb_next(&state->rb_node); + if (other_node) { + other = rb_entry(other_node, struct extent_state, rb_node); + if (other->start == state->end + 1 && + other->state == state->state) { + other->start = state->start; + state->in_tree = 0; + rb_erase(&state->rb_node, &tree->state); + free_extent_state(state); + } + } + return 0; +} + +/* + * insert an extent_state struct into the tree. 'bits' are set on the + * struct before it is inserted. + * + * This may return -EEXIST if the extent is already there, in which case the + * state struct is freed. + * + * The tree lock is not taken internally. This is a utility function and + * probably isn't what you want to call (see set/clear_extent_bit). + */ +static int insert_state(struct extent_io_tree *tree, + struct extent_state *state, u64 start, u64 end, + int bits) +{ + struct rb_node *node; + + if (end < start) { + printk("end < start %Lu %Lu\n", end, start); + WARN_ON(1); + } + if (bits & EXTENT_DIRTY) + tree->dirty_bytes += end - start + 1; + state->state |= bits; + state->start = start; + state->end = end; + node = tree_insert(&tree->state, end, &state->rb_node); + if (node) { + struct extent_state *found; + found = rb_entry(node, struct extent_state, rb_node); + printk("found node %Lu %Lu on insert of %Lu %Lu\n", found->start, found->end, start, end); + free_extent_state(state); + return -EEXIST; + } + merge_state(tree, state); + return 0; +} + +/* + * split a given extent state struct in two, inserting the preallocated + * struct 'prealloc' as the newly created second half. 'split' indicates an + * offset inside 'orig' where it should be split. + * + * Before calling, + * the tree has 'orig' at [orig->start, orig->end]. After calling, there + * are two extent state structs in the tree: + * prealloc: [orig->start, split - 1] + * orig: [ split, orig->end ] + * + * The tree locks are not taken by this function. They need to be held + * by the caller. + */ +static int split_state(struct extent_io_tree *tree, struct extent_state *orig, + struct extent_state *prealloc, u64 split) +{ + struct rb_node *node; + prealloc->start = orig->start; + prealloc->end = split - 1; + prealloc->state = orig->state; + orig->start = split; + + node = tree_insert(&tree->state, prealloc->end, &prealloc->rb_node); + if (node) { + struct extent_state *found; + found = rb_entry(node, struct extent_state, rb_node); + printk("found node %Lu %Lu on insert of %Lu %Lu\n", found->start, found->end, prealloc->start, prealloc->end); + free_extent_state(prealloc); + return -EEXIST; + } + return 0; +} + +/* + * utility function to clear some bits in an extent state struct. + * it will optionally wake up any one waiting on this state (wake == 1), or + * forcibly remove the state from the tree (delete == 1). + * + * If no bits are set on the state struct after clearing things, the + * struct is freed and removed from the tree + */ +static int clear_state_bit(struct extent_io_tree *tree, + struct extent_state *state, int bits, int wake, + int delete) +{ + int ret = state->state & bits; + + if ((bits & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) { + u64 range = state->end - state->start + 1; + WARN_ON(range > tree->dirty_bytes); + tree->dirty_bytes -= range; + } + state->state &= ~bits; + if (wake) + wake_up(&state->wq); + if (delete || state->state == 0) { + if (state->in_tree) { + rb_erase(&state->rb_node, &tree->state); + state->in_tree = 0; + free_extent_state(state); + } else { + WARN_ON(1); + } + } else { + merge_state(tree, state); + } + return ret; +} + +/* + * clear some bits on a range in the tree. This may require splitting + * or inserting elements in the tree, so the gfp mask is used to + * indicate which allocations or sleeping are allowed. + * + * pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove + * the given range from the tree regardless of state (ie for truncate). + * + * the range [start, end] is inclusive. + * + * This takes the tree lock, and returns < 0 on error, > 0 if any of the + * bits were already set, or zero if none of the bits were already set. + */ +int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, + int bits, int wake, int delete, gfp_t mask) +{ + struct extent_state *state; + struct extent_state *prealloc = NULL; + struct rb_node *node; + unsigned long flags; + int err; + int set = 0; + +again: + if (!prealloc && (mask & __GFP_WAIT)) { + prealloc = alloc_extent_state(mask); + if (!prealloc) + return -ENOMEM; + } + + write_lock_irqsave(&tree->lock, flags); + /* + * this search will find the extents that end after + * our range starts + */ + node = tree_search(&tree->state, start); + if (!node) + goto out; + state = rb_entry(node, struct extent_state, rb_node); + if (state->start > end) + goto out; + WARN_ON(state->end < start); + + /* + * | ---- desired range ---- | + * | state | or + * | ------------- state -------------- | + * + * We need to split the extent we found, and may flip + * bits on second half. + * + * If the extent we found extends past our range, we + * just split and search again. It'll get split again + * the next time though. + * + * If the extent we found is inside our range, we clear + * the desired bit on it. + */ + + if (state->start < start) { + err = split_state(tree, state, prealloc, start); + BUG_ON(err == -EEXIST); + prealloc = NULL; + if (err) + goto out; + if (state->end <= end) { + start = state->end + 1; + set |= clear_state_bit(tree, state, bits, + wake, delete); + } else { + start = state->start; + } + goto search_again; + } + /* + * | ---- desired range ---- | + * | state | + * We need to split the extent, and clear the bit + * on the first half + */ + if (state->start <= end && state->end > end) { + err = split_state(tree, state, prealloc, end + 1); + BUG_ON(err == -EEXIST); + + if (wake) + wake_up(&state->wq); + set |= clear_state_bit(tree, prealloc, bits, + wake, delete); + prealloc = NULL; + goto out; + } + + start = state->end + 1; + set |= clear_state_bit(tree, state, bits, wake, delete); + goto search_again; + +out: + write_unlock_irqrestore(&tree->lock, flags); + if (prealloc) + free_extent_state(prealloc); + + return set; + +search_again: + if (start > end) + goto out; + write_unlock_irqrestore(&tree->lock, flags); + if (mask & __GFP_WAIT) + cond_resched(); + goto again; +} +EXPORT_SYMBOL(clear_extent_bit); + +static int wait_on_state(struct extent_io_tree *tree, + struct extent_state *state) +{ + DEFINE_WAIT(wait); + prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE); + read_unlock_irq(&tree->lock); + schedule(); + read_lock_irq(&tree->lock); + finish_wait(&state->wq, &wait); + return 0; +} + +/* + * waits for one or more bits to clear on a range in the state tree. + * The range [start, end] is inclusive. + * The tree lock is taken by this function + */ +int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits) +{ + struct extent_state *state; + struct rb_node *node; + + read_lock_irq(&tree->lock); +again: + while (1) { + /* + * this search will find all the extents that end after + * our range starts + */ + node = tree_search(&tree->state, start); + if (!node) + break; + + state = rb_entry(node, struct extent_state, rb_node); + + if (state->start > end) + goto out; + + if (state->state & bits) { + start = state->start; + atomic_inc(&state->refs); + wait_on_state(tree, state); + free_extent_state(state); + goto again; + } + start = state->end + 1; + + if (start > end) + break; + + if (need_resched()) { + read_unlock_irq(&tree->lock); + cond_resched(); + read_lock_irq(&tree->lock); + } + } +out: + read_unlock_irq(&tree->lock); + return 0; +} +EXPORT_SYMBOL(wait_extent_bit); + +static void set_state_bits(struct extent_io_tree *tree, + struct extent_state *state, + int bits) +{ + if ((bits & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) { + u64 range = state->end - state->start + 1; + tree->dirty_bytes += range; + } + state->state |= bits; +} + +/* + * set some bits on a range in the tree. This may require allocations + * or sleeping, so the gfp mask is used to indicate what is allowed. + * + * If 'exclusive' == 1, this will fail with -EEXIST if some part of the + * range already has the desired bits set. The start of the existing + * range is returned in failed_start in this case. + * + * [start, end] is inclusive + * This takes the tree lock. + */ +int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits, + int exclusive, u64 *failed_start, gfp_t mask) +{ + struct extent_state *state; + struct extent_state *prealloc = NULL; + struct rb_node *node; + unsigned long flags; + int err = 0; + int set; + u64 last_start; + u64 last_end; +again: + if (!prealloc && (mask & __GFP_WAIT)) { + prealloc = alloc_extent_state(mask); + if (!prealloc) + return -ENOMEM; + } + + write_lock_irqsave(&tree->lock, flags); + /* + * this search will find all the extents that end after + * our range starts. + */ + node = tree_search(&tree->state, start); + if (!node) { + err = insert_state(tree, prealloc, start, end, bits); + prealloc = NULL; + BUG_ON(err == -EEXIST); + goto out; + } + + state = rb_entry(node, struct extent_state, rb_node); + last_start = state->start; + last_end = state->end; + + /* + * | ---- desired range ---- | + * | state | + * + * Just lock what we found and keep going + */ + if (state->start == start && state->end <= end) { + set = state->state & bits; + if (set && exclusive) { + *failed_start = state->start; + err = -EEXIST; + goto out; + } + set_state_bits(tree, state, bits); + start = state->end + 1; + merge_state(tree, state); + goto search_again; + } + + /* + * | ---- desired range ---- | + * | state | + * or + * | ------------- state -------------- | + * + * We need to split the extent we found, and may flip bits on + * second half. + * + * If the extent we found extends past our + * range, we just split and search again. It'll get split + * again the next time though. + * + * If the extent we found is inside our range, we set the + * desired bit on it. + */ + if (state->start < start) { + set = state->state & bits; + if (exclusive && set) { + *failed_start = start; + err = -EEXIST; + goto out; + } + err = split_state(tree, state, prealloc, start); + BUG_ON(err == -EEXIST); + prealloc = NULL; + if (err) + goto out; + if (state->end <= end) { + set_state_bits(tree, state, bits); + start = state->end + 1; + merge_state(tree, state); + } else { + start = state->start; + } + goto search_again; + } + /* + * | ---- desired range ---- | + * | state | or | state | + * + * There's a hole, we need to insert something in it and + * ignore the extent we found. + */ + if (state->start > start) { + u64 this_end; + if (end < last_start) + this_end = end; + else + this_end = last_start -1; + err = insert_state(tree, prealloc, start, this_end, + bits); + prealloc = NULL; + BUG_ON(err == -EEXIST); + if (err) + goto out; + start = this_end + 1; + goto search_again; + } + /* + * | ---- desired range ---- | + * | state | + * We need to split the extent, and set the bit + * on the first half + */ + if (state->start <= end && state->end > end) { + set = state->state & bits; + if (exclusive && set) { + *failed_start = start; + err = -EEXIST; + goto out; + } + err = split_state(tree, state, prealloc, end + 1); + BUG_ON(err == -EEXIST); + + set_state_bits(tree, prealloc, bits); + merge_state(tree, prealloc); + prealloc = NULL; + goto out; + } + + goto search_again; + +out: + write_unlock_irqrestore(&tree->lock, flags); + if (prealloc) + free_extent_state(prealloc); + + return err; + +search_again: + if (start > end) + goto out; + write_unlock_irqrestore(&tree->lock, flags); + if (mask & __GFP_WAIT) + cond_resched(); + goto again; +} +EXPORT_SYMBOL(set_extent_bit); + +/* wrappers around set/clear extent bit */ +int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL, + mask); +} +EXPORT_SYMBOL(set_extent_dirty); + +int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, + int bits, gfp_t mask) +{ + return set_extent_bit(tree, start, end, bits, 0, NULL, + mask); +} +EXPORT_SYMBOL(set_extent_bits); + +int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, + int bits, gfp_t mask) +{ + return clear_extent_bit(tree, start, end, bits, 0, 0, mask); +} +EXPORT_SYMBOL(clear_extent_bits); + +int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return set_extent_bit(tree, start, end, + EXTENT_DELALLOC | EXTENT_DIRTY, 0, NULL, + mask); +} +EXPORT_SYMBOL(set_extent_delalloc); + +int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return clear_extent_bit(tree, start, end, + EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, mask); +} +EXPORT_SYMBOL(clear_extent_dirty); + +int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL, + mask); +} +EXPORT_SYMBOL(set_extent_new); + +int clear_extent_new(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return clear_extent_bit(tree, start, end, EXTENT_NEW, 0, 0, mask); +} +EXPORT_SYMBOL(clear_extent_new); + +int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, NULL, + mask); +} +EXPORT_SYMBOL(set_extent_uptodate); + +int clear_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0, mask); +} +EXPORT_SYMBOL(clear_extent_uptodate); + +int set_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return set_extent_bit(tree, start, end, EXTENT_WRITEBACK, + 0, NULL, mask); +} +EXPORT_SYMBOL(set_extent_writeback); + +int clear_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return clear_extent_bit(tree, start, end, EXTENT_WRITEBACK, 1, 0, mask); +} +EXPORT_SYMBOL(clear_extent_writeback); + +int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end) +{ + return wait_extent_bit(tree, start, end, EXTENT_WRITEBACK); +} +EXPORT_SYMBOL(wait_on_extent_writeback); + +/* + * locks a range in ascending order, waiting for any locked regions + * it hits on the way. [start,end] are inclusive, and this will sleep. + */ +int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask) +{ + int err; + u64 failed_start; + while (1) { + err = set_extent_bit(tree, start, end, EXTENT_LOCKED, 1, + &failed_start, mask); + if (err == -EEXIST && (mask & __GFP_WAIT)) { + wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED); + start = failed_start; + } else { + break; + } + WARN_ON(start > end); + } + return err; +} +EXPORT_SYMBOL(lock_extent); + +int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask) +{ + return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, mask); +} +EXPORT_SYMBOL(unlock_extent); + +/* + * helper function to set pages and extents in the tree dirty + */ +int set_range_dirty(struct extent_io_tree *tree, u64 start, u64 end) +{ + unsigned long index = start >> PAGE_CACHE_SHIFT; + unsigned long end_index = end >> PAGE_CACHE_SHIFT; + struct page *page; + + while (index <= end_index) { + page = find_get_page(tree->mapping, index); + BUG_ON(!page); + __set_page_dirty_nobuffers(page); + page_cache_release(page); + index++; + } + set_extent_dirty(tree, start, end, GFP_NOFS); + return 0; +} +EXPORT_SYMBOL(set_range_dirty); + +/* + * helper function to set both pages and extents in the tree writeback + */ +int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end) +{ + unsigned long index = start >> PAGE_CACHE_SHIFT; + unsigned long end_index = end >> PAGE_CACHE_SHIFT; + struct page *page; + + while (index <= end_index) { + page = find_get_page(tree->mapping, index); + BUG_ON(!page); + set_page_writeback(page); + page_cache_release(page); + index++; + } + set_extent_writeback(tree, start, end, GFP_NOFS); + return 0; +} +EXPORT_SYMBOL(set_range_writeback); + +int find_first_extent_bit(struct extent_io_tree *tree, u64 start, + u64 *start_ret, u64 *end_ret, int bits) +{ + struct rb_node *node; + struct extent_state *state; + int ret = 1; + + read_lock_irq(&tree->lock); + /* + * this search will find all the extents that end after + * our range starts. + */ + node = tree_search(&tree->state, start); + if (!node || IS_ERR(node)) { + goto out; + } + + while(1) { + state = rb_entry(node, struct extent_state, rb_node); + if (state->end >= start && (state->state & bits)) { + *start_ret = state->start; + *end_ret = state->end; + ret = 0; + break; + } + node = rb_next(node); + if (!node) + break; + } +out: + read_unlock_irq(&tree->lock); + return ret; +} +EXPORT_SYMBOL(find_first_extent_bit); + +u64 find_lock_delalloc_range(struct extent_io_tree *tree, + u64 *start, u64 *end, u64 max_bytes) +{ + struct rb_node *node; + struct extent_state *state; + u64 cur_start = *start; + u64 found = 0; + u64 total_bytes = 0; + + write_lock_irq(&tree->lock); + /* + * this search will find all the extents that end after + * our range starts. + */ +search_again: + node = tree_search(&tree->state, cur_start); + if (!node || IS_ERR(node)) { + *end = (u64)-1; + goto out; + } + + while(1) { + state = rb_entry(node, struct extent_state, rb_node); + if (found && state->start != cur_start) { + goto out; + } + if (!(state->state & EXTENT_DELALLOC)) { + if (!found) + *end = state->end; + goto out; + } + if (!found) { + struct extent_state *prev_state; + struct rb_node *prev_node = node; + while(1) { + prev_node = rb_prev(prev_node); + if (!prev_node) + break; + prev_state = rb_entry(prev_node, + struct extent_state, + rb_node); + if (!(prev_state->state & EXTENT_DELALLOC)) + break; + state = prev_state; + node = prev_node; + } + } + if (state->state & EXTENT_LOCKED) { + DEFINE_WAIT(wait); + atomic_inc(&state->refs); + prepare_to_wait(&state->wq, &wait, + TASK_UNINTERRUPTIBLE); + write_unlock_irq(&tree->lock); + schedule(); + write_lock_irq(&tree->lock); + finish_wait(&state->wq, &wait); + free_extent_state(state); + goto search_again; + } + state->state |= EXTENT_LOCKED; + if (!found) + *start = state->start; + found++; + *end = state->end; + cur_start = state->end + 1; + node = rb_next(node); + if (!node) + break; + total_bytes += state->end - state->start + 1; + if (total_bytes >= max_bytes) + break; + } +out: + write_unlock_irq(&tree->lock); + return found; +} + +u64 count_range_bits(struct extent_io_tree *tree, + u64 *start, u64 search_end, u64 max_bytes, + unsigned long bits) +{ + struct rb_node *node; + struct extent_state *state; + u64 cur_start = *start; + u64 total_bytes = 0; + int found = 0; + + if (search_end <= cur_start) { + printk("search_end %Lu start %Lu\n", search_end, cur_start); + WARN_ON(1); + return 0; + } + + write_lock_irq(&tree->lock); + if (cur_start == 0 && bits == EXTENT_DIRTY) { + total_bytes = tree->dirty_bytes; + goto out; + } + /* + * this search will find all the extents that end after + * our range starts. + */ + node = tree_search(&tree->state, cur_start); + if (!node || IS_ERR(node)) { + goto out; + } + + while(1) { + state = rb_entry(node, struct extent_state, rb_node); + if (state->start > search_end) + break; + if (state->end >= cur_start && (state->state & bits)) { + total_bytes += min(search_end, state->end) + 1 - + max(cur_start, state->start); + if (total_bytes >= max_bytes) + break; + if (!found) { + *start = state->start; + found = 1; + } + } + node = rb_next(node); + if (!node) + break; + } +out: + write_unlock_irq(&tree->lock); + return total_bytes; +} +/* + * helper function to lock both pages and extents in the tree. + * pages must be locked first. + */ +int lock_range(struct extent_io_tree *tree, u64 start, u64 end) +{ + unsigned long index = start >> PAGE_CACHE_SHIFT; + unsigned long end_index = end >> PAGE_CACHE_SHIFT; + struct page *page; + int err; + + while (index <= end_index) { + page = grab_cache_page(tree->mapping, index); + if (!page) { + err = -ENOMEM; + goto failed; + } + if (IS_ERR(page)) { + err = PTR_ERR(page); + goto failed; + } + index++; + } + lock_extent(tree, start, end, GFP_NOFS); + return 0; + +failed: + /* + * we failed above in getting the page at 'index', so we undo here + * up to but not including the page at 'index' + */ + end_index = index; + index = start >> PAGE_CACHE_SHIFT; + while (index < end_index) { + page = find_get_page(tree->mapping, index); + unlock_page(page); + page_cache_release(page); + index++; + } + return err; +} +EXPORT_SYMBOL(lock_range); + +/* + * helper function to unlock both pages and extents in the tree. + */ +int unlock_range(struct extent_io_tree *tree, u64 start, u64 end) +{ + unsigned long index = start >> PAGE_CACHE_SHIFT; + unsigned long end_index = end >> PAGE_CACHE_SHIFT; + struct page *page; + + while (index <= end_index) { + page = find_get_page(tree->mapping, index); + unlock_page(page); + page_cache_release(page); + index++; + } + unlock_extent(tree, start, end, GFP_NOFS); + return 0; +} +EXPORT_SYMBOL(unlock_range); + +int set_state_private(struct extent_io_tree *tree, u64 start, u64 private) +{ + struct rb_node *node; + struct extent_state *state; + int ret = 0; + + write_lock_irq(&tree->lock); + /* + * this search will find all the extents that end after + * our range starts. + */ + node = tree_search(&tree->state, start); + if (!node || IS_ERR(node)) { + ret = -ENOENT; + goto out; + } + state = rb_entry(node, struct extent_state, rb_node); + if (state->start != start) { + ret = -ENOENT; + goto out; + } + state->private = private; +out: + write_unlock_irq(&tree->lock); + return ret; +} + +int get_state_private(struct extent_io_tree *tree, u64 start, u64 *private) +{ + struct rb_node *node; + struct extent_state *state; + int ret = 0; + + read_lock_irq(&tree->lock); + /* + * this search will find all the extents that end after + * our range starts. + */ + node = tree_search(&tree->state, start); + if (!node || IS_ERR(node)) { + ret = -ENOENT; + goto out; + } + state = rb_entry(node, struct extent_state, rb_node); + if (state->start != start) { + ret = -ENOENT; + goto out; + } + *private = state->private; +out: + read_unlock_irq(&tree->lock); + return ret; +} + +/* + * searches a range in the state tree for a given mask. + * If 'filled' == 1, this returns 1 only if ever extent in the tree + * has the bits set. Otherwise, 1 is returned if any bit in the + * range is found set. + */ +int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end, + int bits, int filled) +{ + struct extent_state *state = NULL; + struct rb_node *node; + int bitset = 0; + unsigned long flags; + + read_lock_irqsave(&tree->lock, flags); + node = tree_search(&tree->state, start); + while (node && start <= end) { + state = rb_entry(node, struct extent_state, rb_node); + + if (filled && state->start > start) { + bitset = 0; + break; + } + + if (state->start > end) + break; + + if (state->state & bits) { + bitset = 1; + if (!filled) + break; + } else if (filled) { + bitset = 0; + break; + } + start = state->end + 1; + if (start > end) + break; + node = rb_next(node); + if (!node) { + if (filled) + bitset = 0; + break; + } + } + read_unlock_irqrestore(&tree->lock, flags); + return bitset; +} +EXPORT_SYMBOL(test_range_bit); + +/* + * helper function to set a given page up to date if all the + * extents in the tree for that page are up to date + */ +static int check_page_uptodate(struct extent_io_tree *tree, + struct page *page) +{ + u64 start = (u64)page->index << PAGE_CACHE_SHIFT; + u64 end = start + PAGE_CACHE_SIZE - 1; + if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1)) + SetPageUptodate(page); + return 0; +} + +/* + * helper function to unlock a page if all the extents in the tree + * for that page are unlocked + */ +static int check_page_locked(struct extent_io_tree *tree, + struct page *page) +{ + u64 start = (u64)page->index << PAGE_CACHE_SHIFT; + u64 end = start + PAGE_CACHE_SIZE - 1; + if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0)) + unlock_page(page); + return 0; +} + +/* + * helper function to end page writeback if all the extents + * in the tree for that page are done with writeback + */ +static int check_page_writeback(struct extent_io_tree *tree, + struct page *page) +{ + u64 start = (u64)page->index << PAGE_CACHE_SHIFT; + u64 end = start + PAGE_CACHE_SIZE - 1; + if (!test_range_bit(tree, start, end, EXTENT_WRITEBACK, 0)) + end_page_writeback(page); + return 0; +} + +/* lots and lots of room for performance fixes in the end_bio funcs */ + +/* + * after a writepage IO is done, we need to: + * clear the uptodate bits on error + * clear the writeback bits in the extent tree for this IO + * end_page_writeback if the page has no more pending IO + * + * Scheduling is not allowed, so the extent state tree is expected + * to have one and only one object corresponding to this IO. + */ +#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23) +static void end_bio_extent_writepage(struct bio *bio, int err) +#else +static int end_bio_extent_writepage(struct bio *bio, + unsigned int bytes_done, int err) +#endif +{ + const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); + struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; + struct extent_io_tree *tree = bio->bi_private; + u64 start; + u64 end; + int whole_page; + +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) + if (bio->bi_size) + return 1; +#endif + + do { + struct page *page = bvec->bv_page; + start = ((u64)page->index << PAGE_CACHE_SHIFT) + + bvec->bv_offset; + end = start + bvec->bv_len - 1; + + if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE) + whole_page = 1; + else + whole_page = 0; + + if (--bvec >= bio->bi_io_vec) + prefetchw(&bvec->bv_page->flags); + + if (!uptodate) { + clear_extent_uptodate(tree, start, end, GFP_ATOMIC); + ClearPageUptodate(page); + SetPageError(page); + } + clear_extent_writeback(tree, start, end, GFP_ATOMIC); + + if (whole_page) + end_page_writeback(page); + else + check_page_writeback(tree, page); + if (tree->ops && tree->ops->writepage_end_io_hook) + tree->ops->writepage_end_io_hook(page, start, end); + } while (bvec >= bio->bi_io_vec); + + bio_put(bio); +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) + return 0; +#endif +} + +/* + * after a readpage IO is done, we need to: + * clear the uptodate bits on error + * set the uptodate bits if things worked + * set the page up to date if all extents in the tree are uptodate + * clear the lock bit in the extent tree + * unlock the page if there are no other extents locked for it + * + * Scheduling is not allowed, so the extent state tree is expected + * to have one and only one object corresponding to this IO. + */ +#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23) +static void end_bio_extent_readpage(struct bio *bio, int err) +#else +static int end_bio_extent_readpage(struct bio *bio, + unsigned int bytes_done, int err) +#endif +{ + int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); + struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; + struct extent_io_tree *tree = bio->bi_private; + u64 start; + u64 end; + int whole_page; + int ret; + +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) + if (bio->bi_size) + return 1; +#endif + + do { + struct page *page = bvec->bv_page; + start = ((u64)page->index << PAGE_CACHE_SHIFT) + + bvec->bv_offset; + end = start + bvec->bv_len - 1; + + if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE) + whole_page = 1; + else + whole_page = 0; + + if (--bvec >= bio->bi_io_vec) + prefetchw(&bvec->bv_page->flags); + + if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) { + ret = tree->ops->readpage_end_io_hook(page, start, end); + if (ret) + uptodate = 0; + } + if (uptodate) { + set_extent_uptodate(tree, start, end, GFP_ATOMIC); + if (whole_page) + SetPageUptodate(page); + else + check_page_uptodate(tree, page); + } else { + ClearPageUptodate(page); + SetPageError(page); + } + + unlock_extent(tree, start, end, GFP_ATOMIC); + + if (whole_page) + unlock_page(page); + else + check_page_locked(tree, page); + } while (bvec >= bio->bi_io_vec); + + bio_put(bio); +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) + return 0; +#endif +} + +/* + * IO done from prepare_write is pretty simple, we just unlock + * the structs in the extent tree when done, and set the uptodate bits + * as appropriate. + */ +#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23) +static void end_bio_extent_preparewrite(struct bio *bio, int err) +#else +static int end_bio_extent_preparewrite(struct bio *bio, + unsigned int bytes_done, int err) +#endif +{ + const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); + struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; + struct extent_io_tree *tree = bio->bi_private; + u64 start; + u64 end; + +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) + if (bio->bi_size) + return 1; +#endif + + do { + struct page *page = bvec->bv_page; + start = ((u64)page->index << PAGE_CACHE_SHIFT) + + bvec->bv_offset; + end = start + bvec->bv_len - 1; + + if (--bvec >= bio->bi_io_vec) + prefetchw(&bvec->bv_page->flags); + + if (uptodate) { + set_extent_uptodate(tree, start, end, GFP_ATOMIC); + } else { + ClearPageUptodate(page); + SetPageError(page); + } + + unlock_extent(tree, start, end, GFP_ATOMIC); + + } while (bvec >= bio->bi_io_vec); + + bio_put(bio); +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) + return 0; +#endif +} + +static struct bio * +extent_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs, + gfp_t gfp_flags) +{ + struct bio *bio; + + bio = bio_alloc(gfp_flags, nr_vecs); + + if (bio == NULL && (current->flags & PF_MEMALLOC)) { + while (!bio && (nr_vecs /= 2)) + bio = bio_alloc(gfp_flags, nr_vecs); + } + + if (bio) { + bio->bi_bdev = bdev; + bio->bi_sector = first_sector; + } + return bio; +} + +static int submit_one_bio(int rw, struct bio *bio) +{ + u64 maxsector; + int ret = 0; + + bio_get(bio); + + maxsector = bio->bi_bdev->bd_inode->i_size >> 9; + if (maxsector < bio->bi_sector) { + printk("sector too large max %Lu got %llu\n", maxsector, + (unsigned long long)bio->bi_sector); + WARN_ON(1); + } + + submit_bio(rw, bio); + if (bio_flagged(bio, BIO_EOPNOTSUPP)) + ret = -EOPNOTSUPP; + bio_put(bio); + return ret; +} + +static int submit_extent_page(int rw, struct extent_io_tree *tree, + struct page *page, sector_t sector, + size_t size, unsigned long offset, + struct block_device *bdev, + struct bio **bio_ret, + unsigned long max_pages, + bio_end_io_t end_io_func) +{ + int ret = 0; + struct bio *bio; + int nr; + + if (bio_ret && *bio_ret) { + bio = *bio_ret; + if (bio->bi_sector + (bio->bi_size >> 9) != sector || + bio_add_page(bio, page, size, offset) < size) { + ret = submit_one_bio(rw, bio); + bio = NULL; + } else { + return 0; + } + } + nr = min_t(int, max_pages, bio_get_nr_vecs(bdev)); + bio = extent_bio_alloc(bdev, sector, nr, GFP_NOFS | __GFP_HIGH); + if (!bio) { + printk("failed to allocate bio nr %d\n", nr); + } + bio_add_page(bio, page, size, offset); + bio->bi_end_io = end_io_func; + bio->bi_private = tree; + if (bio_ret) { + *bio_ret = bio; + } else { + ret = submit_one_bio(rw, bio); + } + + return ret; +} + +void set_page_extent_mapped(struct page *page) +{ + if (!PagePrivate(page)) { + SetPagePrivate(page); + WARN_ON(!page->mapping->a_ops->invalidatepage); + set_page_private(page, EXTENT_PAGE_PRIVATE); + page_cache_get(page); + } +} + +void set_page_extent_head(struct page *page, unsigned long len) +{ + set_page_private(page, EXTENT_PAGE_PRIVATE_FIRST_PAGE | len << 2); +} + +/* + * basic readpage implementation. Locked extent state structs are inserted + * into the tree that are removed when the IO is done (by the end_io + * handlers) + */ +static int __extent_read_full_page(struct extent_io_tree *tree, + struct page *page, + get_extent_t *get_extent, + struct bio **bio) +{ + struct inode *inode = page->mapping->host; + u64 start = (u64)page->index << PAGE_CACHE_SHIFT; + u64 page_end = start + PAGE_CACHE_SIZE - 1; + u64 end; + u64 cur = start; + u64 extent_offset; + u64 last_byte = i_size_read(inode); + u64 block_start; + u64 cur_end; + sector_t sector; + struct extent_map *em; + struct block_device *bdev; + int ret; + int nr = 0; + size_t page_offset = 0; + size_t iosize; + size_t blocksize = inode->i_sb->s_blocksize; + + set_page_extent_mapped(page); + + end = page_end; + lock_extent(tree, start, end, GFP_NOFS); + + while (cur <= end) { + if (cur >= last_byte) { + char *userpage; + iosize = PAGE_CACHE_SIZE - page_offset; + userpage = kmap_atomic(page, KM_USER0); + memset(userpage + page_offset, 0, iosize); + flush_dcache_page(page); + kunmap_atomic(userpage, KM_USER0); + set_extent_uptodate(tree, cur, cur + iosize - 1, + GFP_NOFS); + unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); + break; + } + em = get_extent(inode, page, page_offset, cur, + end - cur + 1, 0); + if (IS_ERR(em) || !em) { + SetPageError(page); + unlock_extent(tree, cur, end, GFP_NOFS); + break; + } + + extent_offset = cur - em->start; + BUG_ON(extent_map_end(em) <= cur); + BUG_ON(end < cur); + + iosize = min(extent_map_end(em) - cur, end - cur + 1); + cur_end = min(extent_map_end(em) - 1, end); + iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1); + sector = (em->block_start + extent_offset) >> 9; + bdev = em->bdev; + block_start = em->block_start; + free_extent_map(em); + em = NULL; + + /* we've found a hole, just zero and go on */ + if (block_start == EXTENT_MAP_HOLE) { + char *userpage; + userpage = kmap_atomic(page, KM_USER0); + memset(userpage + page_offset, 0, iosize); + flush_dcache_page(page); + kunmap_atomic(userpage, KM_USER0); + + set_extent_uptodate(tree, cur, cur + iosize - 1, + GFP_NOFS); + unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); + cur = cur + iosize; + page_offset += iosize; + continue; + } + /* the get_extent function already copied into the page */ + if (test_range_bit(tree, cur, cur_end, EXTENT_UPTODATE, 1)) { + unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); + cur = cur + iosize; + page_offset += iosize; + continue; + } + + ret = 0; + if (tree->ops && tree->ops->readpage_io_hook) { + ret = tree->ops->readpage_io_hook(page, cur, + cur + iosize - 1); + } + if (!ret) { + unsigned long nr = (last_byte >> PAGE_CACHE_SHIFT) + 1; + nr -= page->index; + ret = submit_extent_page(READ, tree, page, + sector, iosize, page_offset, + bdev, bio, nr, + end_bio_extent_readpage); + } + if (ret) + SetPageError(page); + cur = cur + iosize; + page_offset += iosize; + nr++; + } + if (!nr) { + if (!PageError(page)) + SetPageUptodate(page); + unlock_page(page); + } + return 0; +} + +int extent_read_full_page(struct extent_io_tree *tree, struct page *page, + get_extent_t *get_extent) +{ + struct bio *bio = NULL; + int ret; + + ret = __extent_read_full_page(tree, page, get_extent, &bio); + if (bio) + submit_one_bio(READ, bio); + return ret; +} +EXPORT_SYMBOL(extent_read_full_page); + +/* + * the writepage semantics are similar to regular writepage. extent + * records are inserted to lock ranges in the tree, and as dirty areas + * are found, they are marked writeback. Then the lock bits are removed + * and the end_io handler clears the writeback ranges + */ +static int __extent_writepage(struct page *page, struct writeback_control *wbc, + void *data) +{ + struct inode *inode = page->mapping->host; + struct extent_page_data *epd = data; + struct extent_io_tree *tree = epd->tree; + u64 start = (u64)page->index << PAGE_CACHE_SHIFT; + u64 delalloc_start; + u64 page_end = start + PAGE_CACHE_SIZE - 1; + u64 end; + u64 cur = start; + u64 extent_offset; + u64 last_byte = i_size_read(inode); + u64 block_start; + u64 iosize; + sector_t sector; + struct extent_map *em; + struct block_device *bdev; + int ret; + int nr = 0; + size_t page_offset = 0; + size_t blocksize; + loff_t i_size = i_size_read(inode); + unsigned long end_index = i_size >> PAGE_CACHE_SHIFT; + u64 nr_delalloc; + u64 delalloc_end; + + WARN_ON(!PageLocked(page)); + if (page->index > end_index) { + clear_extent_dirty(tree, start, page_end, GFP_NOFS); + unlock_page(page); + return 0; + } + + if (page->index == end_index) { + char *userpage; + + size_t offset = i_size & (PAGE_CACHE_SIZE - 1); + + userpage = kmap_atomic(page, KM_USER0); + memset(userpage + offset, 0, PAGE_CACHE_SIZE - offset); + flush_dcache_page(page); + kunmap_atomic(userpage, KM_USER0); + } + + set_page_extent_mapped(page); + + delalloc_start = start; + delalloc_end = 0; + while(delalloc_end < page_end) { + nr_delalloc = find_lock_delalloc_range(tree, &delalloc_start, + &delalloc_end, + 128 * 1024 * 1024); + if (nr_delalloc == 0) { + delalloc_start = delalloc_end + 1; + continue; + } + tree->ops->fill_delalloc(inode, delalloc_start, + delalloc_end); + clear_extent_bit(tree, delalloc_start, + delalloc_end, + EXTENT_LOCKED | EXTENT_DELALLOC, + 1, 0, GFP_NOFS); + delalloc_start = delalloc_end + 1; + } + lock_extent(tree, start, page_end, GFP_NOFS); + + end = page_end; + if (test_range_bit(tree, start, page_end, EXTENT_DELALLOC, 0)) { + printk("found delalloc bits after lock_extent\n"); + } + + if (last_byte <= start) { + clear_extent_dirty(tree, start, page_end, GFP_NOFS); + goto done; + } + + set_extent_uptodate(tree, start, page_end, GFP_NOFS); + blocksize = inode->i_sb->s_blocksize; + + while (cur <= end) { + if (cur >= last_byte) { + clear_extent_dirty(tree, cur, page_end, GFP_NOFS); + break; + } + em = epd->get_extent(inode, page, page_offset, cur, + end - cur + 1, 1); + if (IS_ERR(em) || !em) { + SetPageError(page); + break; + } + + extent_offset = cur - em->start; + BUG_ON(extent_map_end(em) <= cur); + BUG_ON(end < cur); + iosize = min(extent_map_end(em) - cur, end - cur + 1); + iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1); + sector = (em->block_start + extent_offset) >> 9; + bdev = em->bdev; + block_start = em->block_start; + free_extent_map(em); + em = NULL; + + if (block_start == EXTENT_MAP_HOLE || + block_start == EXTENT_MAP_INLINE) { + clear_extent_dirty(tree, cur, + cur + iosize - 1, GFP_NOFS); + cur = cur + iosize; + page_offset += iosize; + continue; + } + + /* leave this out until we have a page_mkwrite call */ + if (0 && !test_range_bit(tree, cur, cur + iosize - 1, + EXTENT_DIRTY, 0)) { + cur = cur + iosize; + page_offset += iosize; + continue; + } + clear_extent_dirty(tree, cur, cur + iosize - 1, GFP_NOFS); + if (tree->ops && tree->ops->writepage_io_hook) { + ret = tree->ops->writepage_io_hook(page, cur, + cur + iosize - 1); + } else { + ret = 0; + } + if (ret) + SetPageError(page); + else { + unsigned long max_nr = end_index + 1; + set_range_writeback(tree, cur, cur + iosize - 1); + if (!PageWriteback(page)) { + printk("warning page %lu not writeback, " + "cur %llu end %llu\n", page->index, + (unsigned long long)cur, + (unsigned long long)end); + } + + ret = submit_extent_page(WRITE, tree, page, sector, + iosize, page_offset, bdev, + &epd->bio, max_nr, + end_bio_extent_writepage); + if (ret) + SetPageError(page); + } + cur = cur + iosize; + page_offset += iosize; + nr++; + } +done: + if (nr == 0) { + /* make sure the mapping tag for page dirty gets cleared */ + set_page_writeback(page); + end_page_writeback(page); + } + unlock_extent(tree, start, page_end, GFP_NOFS); + unlock_page(page); + return 0; +} + +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18) + +/* Taken directly from 2.6.23 for 2.6.18 back port */ +typedef int (*writepage_t)(struct page *page, struct writeback_control *wbc, + void *data); + +/** + * write_cache_pages - walk the list of dirty pages of the given address space + * and write all of them. + * @mapping: address space structure to write + * @wbc: subtract the number of written pages from *@wbc->nr_to_write + * @writepage: function called for each page + * @data: data passed to writepage function + * + * If a page is already under I/O, write_cache_pages() skips it, even + * if it's dirty. This is desirable behaviour for memory-cleaning writeback, + * but it is INCORRECT for data-integrity system calls such as fsync(). fsync() + * and msync() need to guarantee that all the data which was dirty at the time + * the call was made get new I/O started against them. If wbc->sync_mode is + * WB_SYNC_ALL then we were called for data integrity and we must wait for + * existing IO to complete. + */ +static int write_cache_pages(struct address_space *mapping, + struct writeback_control *wbc, writepage_t writepage, + void *data) +{ + struct backing_dev_info *bdi = mapping->backing_dev_info; + int ret = 0; + int done = 0; + struct pagevec pvec; + int nr_pages; + pgoff_t index; + pgoff_t end; /* Inclusive */ + int scanned = 0; + int range_whole = 0; + + if (wbc->nonblocking && bdi_write_congested(bdi)) { + wbc->encountered_congestion = 1; + return 0; + } + + pagevec_init(&pvec, 0); + if (wbc->range_cyclic) { + index = mapping->writeback_index; /* Start from prev offset */ + end = -1; + } else { + index = wbc->range_start >> PAGE_CACHE_SHIFT; + end = wbc->range_end >> PAGE_CACHE_SHIFT; + if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX) + range_whole = 1; + scanned = 1; + } +retry: + while (!done && (index <= end) && + (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, + PAGECACHE_TAG_DIRTY, + min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) { + unsigned i; + + scanned = 1; + for (i = 0; i < nr_pages; i++) { + struct page *page = pvec.pages[i]; + + /* + * At this point we hold neither mapping->tree_lock nor + * lock on the page itself: the page may be truncated or + * invalidated (changing page->mapping to NULL), or even + * swizzled back from swapper_space to tmpfs file + * mapping + */ + lock_page(page); + + if (unlikely(page->mapping != mapping)) { + unlock_page(page); + continue; + } + + if (!wbc->range_cyclic && page->index > end) { + done = 1; + unlock_page(page); + continue; + } + + if (wbc->sync_mode != WB_SYNC_NONE) + wait_on_page_writeback(page); + + if (PageWriteback(page) || + !clear_page_dirty_for_io(page)) { + unlock_page(page); + continue; + } + + ret = (*writepage)(page, wbc, data); + + if (unlikely(ret == AOP_WRITEPAGE_ACTIVATE)) { + unlock_page(page); + ret = 0; + } + if (ret || (--(wbc->nr_to_write) <= 0)) + done = 1; + if (wbc->nonblocking && bdi_write_congested(bdi)) { + wbc->encountered_congestion = 1; + done = 1; + } + } + pagevec_release(&pvec); + cond_resched(); + } + if (!scanned && !done) { + /* + * We hit the last page and there is more work to be done: wrap + * back to the start of the file + */ + scanned = 1; + index = 0; + goto retry; + } + if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0)) + mapping->writeback_index = index; + return ret; +} +#endif + +int extent_write_full_page(struct extent_io_tree *tree, struct page *page, + get_extent_t *get_extent, + struct writeback_control *wbc) +{ + int ret; + struct address_space *mapping = page->mapping; + struct extent_page_data epd = { + .bio = NULL, + .tree = tree, + .get_extent = get_extent, + }; + struct writeback_control wbc_writepages = { + .bdi = wbc->bdi, + .sync_mode = WB_SYNC_NONE, + .older_than_this = NULL, + .nr_to_write = 64, + .range_start = page_offset(page) + PAGE_CACHE_SIZE, + .range_end = (loff_t)-1, + }; + + + ret = __extent_writepage(page, wbc, &epd); + + write_cache_pages(mapping, &wbc_writepages, __extent_writepage, &epd); + if (epd.bio) { + submit_one_bio(WRITE, epd.bio); + } + return ret; +} +EXPORT_SYMBOL(extent_write_full_page); + + +int extent_writepages(struct extent_io_tree *tree, + struct address_space *mapping, + get_extent_t *get_extent, + struct writeback_control *wbc) +{ + int ret = 0; + struct extent_page_data epd = { + .bio = NULL, + .tree = tree, + .get_extent = get_extent, + }; + + ret = write_cache_pages(mapping, wbc, __extent_writepage, &epd); + if (epd.bio) { + submit_one_bio(WRITE, epd.bio); + } + return ret; +} +EXPORT_SYMBOL(extent_writepages); + +int extent_readpages(struct extent_io_tree *tree, + struct address_space *mapping, + struct list_head *pages, unsigned nr_pages, + get_extent_t get_extent) +{ + struct bio *bio = NULL; + unsigned page_idx; + struct pagevec pvec; + + pagevec_init(&pvec, 0); + for (page_idx = 0; page_idx < nr_pages; page_idx++) { + struct page *page = list_entry(pages->prev, struct page, lru); + + prefetchw(&page->flags); + list_del(&page->lru); + /* + * what we want to do here is call add_to_page_cache_lru, + * but that isn't exported, so we reproduce it here + */ + if (!add_to_page_cache(page, mapping, + page->index, GFP_KERNEL)) { + + /* open coding of lru_cache_add, also not exported */ + page_cache_get(page); + if (!pagevec_add(&pvec, page)) + __pagevec_lru_add(&pvec); + __extent_read_full_page(tree, page, get_extent, &bio); + } + page_cache_release(page); + } + if (pagevec_count(&pvec)) + __pagevec_lru_add(&pvec); + BUG_ON(!list_empty(pages)); + if (bio) + submit_one_bio(READ, bio); + return 0; +} +EXPORT_SYMBOL(extent_readpages); + +/* + * basic invalidatepage code, this waits on any locked or writeback + * ranges corresponding to the page, and then deletes any extent state + * records from the tree + */ +int extent_invalidatepage(struct extent_io_tree *tree, + struct page *page, unsigned long offset) +{ + u64 start = ((u64)page->index << PAGE_CACHE_SHIFT); + u64 end = start + PAGE_CACHE_SIZE - 1; + size_t blocksize = page->mapping->host->i_sb->s_blocksize; + + start += (offset + blocksize -1) & ~(blocksize - 1); + if (start > end) + return 0; + + lock_extent(tree, start, end, GFP_NOFS); + wait_on_extent_writeback(tree, start, end); + clear_extent_bit(tree, start, end, + EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC, + 1, 1, GFP_NOFS); + return 0; +} +EXPORT_SYMBOL(extent_invalidatepage); + +/* + * simple commit_write call, set_range_dirty is used to mark both + * the pages and the extent records as dirty + */ +int extent_commit_write(struct extent_io_tree *tree, + struct inode *inode, struct page *page, + unsigned from, unsigned to) +{ + loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to; + + set_page_extent_mapped(page); + set_page_dirty(page); + + if (pos > inode->i_size) { + i_size_write(inode, pos); + mark_inode_dirty(inode); + } + return 0; +} +EXPORT_SYMBOL(extent_commit_write); + +int extent_prepare_write(struct extent_io_tree *tree, + struct inode *inode, struct page *page, + unsigned from, unsigned to, get_extent_t *get_extent) +{ + u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT; + u64 page_end = page_start + PAGE_CACHE_SIZE - 1; + u64 block_start; + u64 orig_block_start; + u64 block_end; + u64 cur_end; + struct extent_map *em; + unsigned blocksize = 1 << inode->i_blkbits; + size_t page_offset = 0; + size_t block_off_start; + size_t block_off_end; + int err = 0; + int iocount = 0; + int ret = 0; + int isnew; + + set_page_extent_mapped(page); + + block_start = (page_start + from) & ~((u64)blocksize - 1); + block_end = (page_start + to - 1) | (blocksize - 1); + orig_block_start = block_start; + + lock_extent(tree, page_start, page_end, GFP_NOFS); + while(block_start <= block_end) { + em = get_extent(inode, page, page_offset, block_start, + block_end - block_start + 1, 1); + if (IS_ERR(em) || !em) { + goto err; + } + cur_end = min(block_end, extent_map_end(em) - 1); + block_off_start = block_start & (PAGE_CACHE_SIZE - 1); + block_off_end = block_off_start + blocksize; + isnew = clear_extent_new(tree, block_start, cur_end, GFP_NOFS); + + if (!PageUptodate(page) && isnew && + (block_off_end > to || block_off_start < from)) { + void *kaddr; + + kaddr = kmap_atomic(page, KM_USER0); + if (block_off_end > to) + memset(kaddr + to, 0, block_off_end - to); + if (block_off_start < from) + memset(kaddr + block_off_start, 0, + from - block_off_start); + flush_dcache_page(page); + kunmap_atomic(kaddr, KM_USER0); + } + if ((em->block_start != EXTENT_MAP_HOLE && + em->block_start != EXTENT_MAP_INLINE) && + !isnew && !PageUptodate(page) && + (block_off_end > to || block_off_start < from) && + !test_range_bit(tree, block_start, cur_end, + EXTENT_UPTODATE, 1)) { + u64 sector; + u64 extent_offset = block_start - em->start; + size_t iosize; + sector = (em->block_start + extent_offset) >> 9; + iosize = (cur_end - block_start + blocksize) & + ~((u64)blocksize - 1); + /* + * we've already got the extent locked, but we + * need to split the state such that our end_bio + * handler can clear the lock. + */ + set_extent_bit(tree, block_start, + block_start + iosize - 1, + EXTENT_LOCKED, 0, NULL, GFP_NOFS); + ret = submit_extent_page(READ, tree, page, + sector, iosize, page_offset, em->bdev, + NULL, 1, + end_bio_extent_preparewrite); + iocount++; + block_start = block_start + iosize; + } else { + set_extent_uptodate(tree, block_start, cur_end, + GFP_NOFS); + unlock_extent(tree, block_start, cur_end, GFP_NOFS); + block_start = cur_end + 1; + } + page_offset = block_start & (PAGE_CACHE_SIZE - 1); + free_extent_map(em); + } + if (iocount) { + wait_extent_bit(tree, orig_block_start, + block_end, EXTENT_LOCKED); + } + check_page_uptodate(tree, page); +err: + /* FIXME, zero out newly allocated blocks on error */ + return err; +} +EXPORT_SYMBOL(extent_prepare_write); + +/* + * a helper for releasepage. As long as there are no locked extents + * in the range corresponding to the page, both state records and extent + * map records are removed + */ +int try_release_extent_mapping(struct extent_map_tree *map, + struct extent_io_tree *tree, struct page *page) +{ + struct extent_map *em; + u64 start = (u64)page->index << PAGE_CACHE_SHIFT; + u64 end = start + PAGE_CACHE_SIZE - 1; + u64 orig_start = start; + int ret = 1; + + while (start <= end) { + spin_lock(&map->lock); + em = lookup_extent_mapping(map, start, end); + if (!em || IS_ERR(em)) { + spin_unlock(&map->lock); + break; + } + if (!test_range_bit(tree, em->start, extent_map_end(em) - 1, + EXTENT_LOCKED, 0)) { + remove_extent_mapping(map, em); + /* once for the rb tree */ + free_extent_map(em); + } + start = extent_map_end(em); + spin_unlock(&map->lock); + + /* once for us */ + free_extent_map(em); + } + if (test_range_bit(tree, orig_start, end, EXTENT_LOCKED, 0)) + ret = 0; + else + clear_extent_bit(tree, orig_start, end, EXTENT_UPTODATE, + 1, 1, GFP_NOFS); + return ret; +} +EXPORT_SYMBOL(try_release_extent_mapping); + +sector_t extent_bmap(struct address_space *mapping, sector_t iblock, + get_extent_t *get_extent) +{ + struct inode *inode = mapping->host; + u64 start = iblock << inode->i_blkbits; + sector_t sector = 0; + struct extent_map *em; + + em = get_extent(inode, NULL, 0, start, (1 << inode->i_blkbits), 0); + if (!em || IS_ERR(em)) + return 0; + + if (em->block_start == EXTENT_MAP_INLINE || + em->block_start == EXTENT_MAP_HOLE) + goto out; + + sector = (em->block_start + start - em->start) >> inode->i_blkbits; +printk("bmap finds %Lu %Lu block %Lu\n", em->start, em->len, em->block_start); +out: + free_extent_map(em); + return sector; +} + +static int add_lru(struct extent_io_tree *tree, struct extent_buffer *eb) +{ + if (list_empty(&eb->lru)) { + extent_buffer_get(eb); + list_add(&eb->lru, &tree->buffer_lru); + tree->lru_size++; + if (tree->lru_size >= BUFFER_LRU_MAX) { + struct extent_buffer *rm; + rm = list_entry(tree->buffer_lru.prev, + struct extent_buffer, lru); + tree->lru_size--; + list_del_init(&rm->lru); + free_extent_buffer(rm); + } + } else + list_move(&eb->lru, &tree->buffer_lru); + return 0; +} +static struct extent_buffer *find_lru(struct extent_io_tree *tree, + u64 start, unsigned long len) +{ + struct list_head *lru = &tree->buffer_lru; + struct list_head *cur = lru->next; + struct extent_buffer *eb; + + if (list_empty(lru)) + return NULL; + + do { + eb = list_entry(cur, struct extent_buffer, lru); + if (eb->start == start && eb->len == len) { + extent_buffer_get(eb); + return eb; + } + cur = cur->next; + } while (cur != lru); + return NULL; +} + +static inline unsigned long num_extent_pages(u64 start, u64 len) +{ + return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) - + (start >> PAGE_CACHE_SHIFT); +} + +static inline struct page *extent_buffer_page(struct extent_buffer *eb, + unsigned long i) +{ + struct page *p; + struct address_space *mapping; + + if (i == 0) + return eb->first_page; + i += eb->start >> PAGE_CACHE_SHIFT; + mapping = eb->first_page->mapping; + read_lock_irq(&mapping->tree_lock); + p = radix_tree_lookup(&mapping->page_tree, i); + read_unlock_irq(&mapping->tree_lock); + return p; +} + +static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree, + u64 start, + unsigned long len, + gfp_t mask) +{ + struct extent_buffer *eb = NULL; + + spin_lock(&tree->lru_lock); + eb = find_lru(tree, start, len); + spin_unlock(&tree->lru_lock); + if (eb) { + return eb; + } + + eb = kmem_cache_zalloc(extent_buffer_cache, mask); + INIT_LIST_HEAD(&eb->lru); + eb->start = start; + eb->len = len; + atomic_set(&eb->refs, 1); + + return eb; +} + +static void __free_extent_buffer(struct extent_buffer *eb) +{ + kmem_cache_free(extent_buffer_cache, eb); +} + +struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree, + u64 start, unsigned long len, + struct page *page0, + gfp_t mask) +{ + unsigned long num_pages = num_extent_pages(start, len); + unsigned long i; + unsigned long index = start >> PAGE_CACHE_SHIFT; + struct extent_buffer *eb; + struct page *p; + struct address_space *mapping = tree->mapping; + int uptodate = 1; + + eb = __alloc_extent_buffer(tree, start, len, mask); + if (!eb || IS_ERR(eb)) + return NULL; + + if (eb->flags & EXTENT_BUFFER_FILLED) + goto lru_add; + + if (page0) { + eb->first_page = page0; + i = 1; + index++; + page_cache_get(page0); + mark_page_accessed(page0); + set_page_extent_mapped(page0); + WARN_ON(!PageUptodate(page0)); + set_page_extent_head(page0, len); + } else { + i = 0; + } + for (; i < num_pages; i++, index++) { + p = find_or_create_page(mapping, index, mask | __GFP_HIGHMEM); + if (!p) { + WARN_ON(1); + goto fail; + } + set_page_extent_mapped(p); + mark_page_accessed(p); + if (i == 0) { + eb->first_page = p; + set_page_extent_head(p, len); + } else { + set_page_private(p, EXTENT_PAGE_PRIVATE); + } + if (!PageUptodate(p)) + uptodate = 0; + unlock_page(p); + } + if (uptodate) + eb->flags |= EXTENT_UPTODATE; + eb->flags |= EXTENT_BUFFER_FILLED; + +lru_add: + spin_lock(&tree->lru_lock); + add_lru(tree, eb); + spin_unlock(&tree->lru_lock); + return eb; + +fail: + spin_lock(&tree->lru_lock); + list_del_init(&eb->lru); + spin_unlock(&tree->lru_lock); + if (!atomic_dec_and_test(&eb->refs)) + return NULL; + for (index = 1; index < i; index++) { + page_cache_release(extent_buffer_page(eb, index)); + } + if (i > 0) + page_cache_release(extent_buffer_page(eb, 0)); + __free_extent_buffer(eb); + return NULL; +} +EXPORT_SYMBOL(alloc_extent_buffer); + +struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree, + u64 start, unsigned long len, + gfp_t mask) +{ + unsigned long num_pages = num_extent_pages(start, len); + unsigned long i; + unsigned long index = start >> PAGE_CACHE_SHIFT; + struct extent_buffer *eb; + struct page *p; + struct address_space *mapping = tree->mapping; + int uptodate = 1; + + eb = __alloc_extent_buffer(tree, start, len, mask); + if (!eb || IS_ERR(eb)) + return NULL; + + if (eb->flags & EXTENT_BUFFER_FILLED) + goto lru_add; + + for (i = 0; i < num_pages; i++, index++) { + p = find_lock_page(mapping, index); + if (!p) { + goto fail; + } + set_page_extent_mapped(p); + mark_page_accessed(p); + + if (i == 0) { + eb->first_page = p; + set_page_extent_head(p, len); + } else { + set_page_private(p, EXTENT_PAGE_PRIVATE); + } + + if (!PageUptodate(p)) + uptodate = 0; + unlock_page(p); + } + if (uptodate) + eb->flags |= EXTENT_UPTODATE; + eb->flags |= EXTENT_BUFFER_FILLED; + +lru_add: + spin_lock(&tree->lru_lock); + add_lru(tree, eb); + spin_unlock(&tree->lru_lock); + return eb; +fail: + spin_lock(&tree->lru_lock); + list_del_init(&eb->lru); + spin_unlock(&tree->lru_lock); + if (!atomic_dec_and_test(&eb->refs)) + return NULL; + for (index = 1; index < i; index++) { + page_cache_release(extent_buffer_page(eb, index)); + } + if (i > 0) + page_cache_release(extent_buffer_page(eb, 0)); + __free_extent_buffer(eb); + return NULL; +} +EXPORT_SYMBOL(find_extent_buffer); + +void free_extent_buffer(struct extent_buffer *eb) +{ + unsigned long i; + unsigned long num_pages; + + if (!eb) + return; + + if (!atomic_dec_and_test(&eb->refs)) + return; + + WARN_ON(!list_empty(&eb->lru)); + num_pages = num_extent_pages(eb->start, eb->len); + + for (i = 1; i < num_pages; i++) { + page_cache_release(extent_buffer_page(eb, i)); + } + page_cache_release(extent_buffer_page(eb, 0)); + __free_extent_buffer(eb); +} +EXPORT_SYMBOL(free_extent_buffer); + +int clear_extent_buffer_dirty(struct extent_io_tree *tree, + struct extent_buffer *eb) +{ + int set; + unsigned long i; + unsigned long num_pages; + struct page *page; + + u64 start = eb->start; + u64 end = start + eb->len - 1; + + set = clear_extent_dirty(tree, start, end, GFP_NOFS); + num_pages = num_extent_pages(eb->start, eb->len); + + for (i = 0; i < num_pages; i++) { + page = extent_buffer_page(eb, i); + lock_page(page); + if (i == 0) + set_page_extent_head(page, eb->len); + else + set_page_private(page, EXTENT_PAGE_PRIVATE); + + /* + * if we're on the last page or the first page and the + * block isn't aligned on a page boundary, do extra checks + * to make sure we don't clean page that is partially dirty + */ + if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) || + ((i == num_pages - 1) && + ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) { + start = (u64)page->index << PAGE_CACHE_SHIFT; + end = start + PAGE_CACHE_SIZE - 1; + if (test_range_bit(tree, start, end, + EXTENT_DIRTY, 0)) { + unlock_page(page); + continue; + } + } + clear_page_dirty_for_io(page); + write_lock_irq(&page->mapping->tree_lock); + if (!PageDirty(page)) { + radix_tree_tag_clear(&page->mapping->page_tree, + page_index(page), + PAGECACHE_TAG_DIRTY); + } + write_unlock_irq(&page->mapping->tree_lock); + unlock_page(page); + } + return 0; +} +EXPORT_SYMBOL(clear_extent_buffer_dirty); + +int wait_on_extent_buffer_writeback(struct extent_io_tree *tree, + struct extent_buffer *eb) +{ + return wait_on_extent_writeback(tree, eb->start, + eb->start + eb->len - 1); +} +EXPORT_SYMBOL(wait_on_extent_buffer_writeback); + +int set_extent_buffer_dirty(struct extent_io_tree *tree, + struct extent_buffer *eb) +{ + unsigned long i; + unsigned long num_pages; + + num_pages = num_extent_pages(eb->start, eb->len); + for (i = 0; i < num_pages; i++) { + struct page *page = extent_buffer_page(eb, i); + /* writepage may need to do something special for the + * first page, we have to make sure page->private is + * properly set. releasepage may drop page->private + * on us if the page isn't already dirty. + */ + if (i == 0) { + lock_page(page); + set_page_extent_head(page, eb->len); + } else if (PagePrivate(page) && + page->private != EXTENT_PAGE_PRIVATE) { + lock_page(page); + set_page_extent_mapped(page); + unlock_page(page); + } + __set_page_dirty_nobuffers(extent_buffer_page(eb, i)); + if (i == 0) + unlock_page(page); + } + return set_extent_dirty(tree, eb->start, + eb->start + eb->len - 1, GFP_NOFS); +} +EXPORT_SYMBOL(set_extent_buffer_dirty); + +int set_extent_buffer_uptodate(struct extent_io_tree *tree, + struct extent_buffer *eb) +{ + unsigned long i; + struct page *page; + unsigned long num_pages; + + num_pages = num_extent_pages(eb->start, eb->len); + + set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1, + GFP_NOFS); + for (i = 0; i < num_pages; i++) { + page = extent_buffer_page(eb, i); + if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) || + ((i == num_pages - 1) && + ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) { + check_page_uptodate(tree, page); + continue; + } + SetPageUptodate(page); + } + return 0; +} +EXPORT_SYMBOL(set_extent_buffer_uptodate); + +int extent_buffer_uptodate(struct extent_io_tree *tree, + struct extent_buffer *eb) +{ + if (eb->flags & EXTENT_UPTODATE) + return 1; + return test_range_bit(tree, eb->start, eb->start + eb->len - 1, + EXTENT_UPTODATE, 1); +} +EXPORT_SYMBOL(extent_buffer_uptodate); + +int read_extent_buffer_pages(struct extent_io_tree *tree, + struct extent_buffer *eb, + u64 start, + int wait) +{ + unsigned long i; + unsigned long start_i; + struct page *page; + int err; + int ret = 0; + unsigned long num_pages; + + if (eb->flags & EXTENT_UPTODATE) + return 0; + + if (0 && test_range_bit(tree, eb->start, eb->start + eb->len - 1, + EXTENT_UPTODATE, 1)) { + return 0; + } + + if (start) { + WARN_ON(start < eb->start); + start_i = (start >> PAGE_CACHE_SHIFT) - + (eb->start >> PAGE_CACHE_SHIFT); + } else { + start_i = 0; + } + + num_pages = num_extent_pages(eb->start, eb->len); + for (i = start_i; i < num_pages; i++) { + page = extent_buffer_page(eb, i); + if (PageUptodate(page)) { + continue; + } + if (!wait) { + if (TestSetPageLocked(page)) { + continue; + } + } else { + lock_page(page); + } + if (!PageUptodate(page)) { + err = page->mapping->a_ops->readpage(NULL, page); + if (err) { + ret = err; + } + } else { + unlock_page(page); + } + } + + if (ret || !wait) { + return ret; + } + + for (i = start_i; i < num_pages; i++) { + page = extent_buffer_page(eb, i); + wait_on_page_locked(page); + if (!PageUptodate(page)) { + ret = -EIO; + } + } + if (!ret) + eb->flags |= EXTENT_UPTODATE; + return ret; +} +EXPORT_SYMBOL(read_extent_buffer_pages); + +void read_extent_buffer(struct extent_buffer *eb, void *dstv, + unsigned long start, + unsigned long len) +{ + size_t cur; + size_t offset; + struct page *page; + char *kaddr; + char *dst = (char *)dstv; + size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); + unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; + unsigned long num_pages = num_extent_pages(eb->start, eb->len); + + WARN_ON(start > eb->len); + WARN_ON(start + len > eb->start + eb->len); + + offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); + + while(len > 0) { + page = extent_buffer_page(eb, i); + if (!PageUptodate(page)) { + printk("page %lu not up to date i %lu, total %lu, len %lu\n", page->index, i, num_pages, eb->len); + WARN_ON(1); + } + WARN_ON(!PageUptodate(page)); + + cur = min(len, (PAGE_CACHE_SIZE - offset)); + kaddr = kmap_atomic(page, KM_USER1); + memcpy(dst, kaddr + offset, cur); + kunmap_atomic(kaddr, KM_USER1); + + dst += cur; + len -= cur; + offset = 0; + i++; + } +} +EXPORT_SYMBOL(read_extent_buffer); + +int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start, + unsigned long min_len, char **token, char **map, + unsigned long *map_start, + unsigned long *map_len, int km) +{ + size_t offset = start & (PAGE_CACHE_SIZE - 1); + char *kaddr; + struct page *p; + size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); + unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; + unsigned long end_i = (start_offset + start + min_len - 1) >> + PAGE_CACHE_SHIFT; + + if (i != end_i) + return -EINVAL; + + if (i == 0) { + offset = start_offset; + *map_start = 0; + } else { + offset = 0; + *map_start = ((u64)i << PAGE_CACHE_SHIFT) - start_offset; + } + if (start + min_len > eb->len) { +printk("bad mapping eb start %Lu len %lu, wanted %lu %lu\n", eb->start, eb->len, start, min_len); + WARN_ON(1); + } + + p = extent_buffer_page(eb, i); + WARN_ON(!PageUptodate(p)); + kaddr = kmap_atomic(p, km); + *token = kaddr; + *map = kaddr + offset; + *map_len = PAGE_CACHE_SIZE - offset; + return 0; +} +EXPORT_SYMBOL(map_private_extent_buffer); + +int map_extent_buffer(struct extent_buffer *eb, unsigned long start, + unsigned long min_len, + char **token, char **map, + unsigned long *map_start, + unsigned long *map_len, int km) +{ + int err; + int save = 0; + if (eb->map_token) { + unmap_extent_buffer(eb, eb->map_token, km); + eb->map_token = NULL; + save = 1; + } + err = map_private_extent_buffer(eb, start, min_len, token, map, + map_start, map_len, km); + if (!err && save) { + eb->map_token = *token; + eb->kaddr = *map; + eb->map_start = *map_start; + eb->map_len = *map_len; + } + return err; +} +EXPORT_SYMBOL(map_extent_buffer); + +void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km) +{ + kunmap_atomic(token, km); +} +EXPORT_SYMBOL(unmap_extent_buffer); + +int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv, + unsigned long start, + unsigned long len) +{ + size_t cur; + size_t offset; + struct page *page; + char *kaddr; + char *ptr = (char *)ptrv; + size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); + unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; + int ret = 0; + + WARN_ON(start > eb->len); + WARN_ON(start + len > eb->start + eb->len); + + offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); + + while(len > 0) { + page = extent_buffer_page(eb, i); + WARN_ON(!PageUptodate(page)); + + cur = min(len, (PAGE_CACHE_SIZE - offset)); + + kaddr = kmap_atomic(page, KM_USER0); + ret = memcmp(ptr, kaddr + offset, cur); + kunmap_atomic(kaddr, KM_USER0); + if (ret) + break; + + ptr += cur; + len -= cur; + offset = 0; + i++; + } + return ret; +} +EXPORT_SYMBOL(memcmp_extent_buffer); + +void write_extent_buffer(struct extent_buffer *eb, const void *srcv, + unsigned long start, unsigned long len) +{ + size_t cur; + size_t offset; + struct page *page; + char *kaddr; + char *src = (char *)srcv; + size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); + unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; + + WARN_ON(start > eb->len); + WARN_ON(start + len > eb->start + eb->len); + + offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); + + while(len > 0) { + page = extent_buffer_page(eb, i); + WARN_ON(!PageUptodate(page)); + + cur = min(len, PAGE_CACHE_SIZE - offset); + kaddr = kmap_atomic(page, KM_USER1); + memcpy(kaddr + offset, src, cur); + kunmap_atomic(kaddr, KM_USER1); + + src += cur; + len -= cur; + offset = 0; + i++; + } +} +EXPORT_SYMBOL(write_extent_buffer); + +void memset_extent_buffer(struct extent_buffer *eb, char c, + unsigned long start, unsigned long len) +{ + size_t cur; + size_t offset; + struct page *page; + char *kaddr; + size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); + unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; + + WARN_ON(start > eb->len); + WARN_ON(start + len > eb->start + eb->len); + + offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); + + while(len > 0) { + page = extent_buffer_page(eb, i); + WARN_ON(!PageUptodate(page)); + + cur = min(len, PAGE_CACHE_SIZE - offset); + kaddr = kmap_atomic(page, KM_USER0); + memset(kaddr + offset, c, cur); + kunmap_atomic(kaddr, KM_USER0); + + len -= cur; + offset = 0; + i++; + } +} +EXPORT_SYMBOL(memset_extent_buffer); + +void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src, + unsigned long dst_offset, unsigned long src_offset, + unsigned long len) +{ + u64 dst_len = dst->len; + size_t cur; + size_t offset; + struct page *page; + char *kaddr; + size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); + unsigned long i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT; + + WARN_ON(src->len != dst_len); + + offset = (start_offset + dst_offset) & + ((unsigned long)PAGE_CACHE_SIZE - 1); + + while(len > 0) { + page = extent_buffer_page(dst, i); + WARN_ON(!PageUptodate(page)); + + cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - offset)); + + kaddr = kmap_atomic(page, KM_USER0); + read_extent_buffer(src, kaddr + offset, src_offset, cur); + kunmap_atomic(kaddr, KM_USER0); + + src_offset += cur; + len -= cur; + offset = 0; + i++; + } +} +EXPORT_SYMBOL(copy_extent_buffer); + +static void move_pages(struct page *dst_page, struct page *src_page, + unsigned long dst_off, unsigned long src_off, + unsigned long len) +{ + char *dst_kaddr = kmap_atomic(dst_page, KM_USER0); + if (dst_page == src_page) { + memmove(dst_kaddr + dst_off, dst_kaddr + src_off, len); + } else { + char *src_kaddr = kmap_atomic(src_page, KM_USER1); + char *p = dst_kaddr + dst_off + len; + char *s = src_kaddr + src_off + len; + + while (len--) + *--p = *--s; + + kunmap_atomic(src_kaddr, KM_USER1); + } + kunmap_atomic(dst_kaddr, KM_USER0); +} + +static void copy_pages(struct page *dst_page, struct page *src_page, + unsigned long dst_off, unsigned long src_off, + unsigned long len) +{ + char *dst_kaddr = kmap_atomic(dst_page, KM_USER0); + char *src_kaddr; + + if (dst_page != src_page) + src_kaddr = kmap_atomic(src_page, KM_USER1); + else + src_kaddr = dst_kaddr; + + memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len); + kunmap_atomic(dst_kaddr, KM_USER0); + if (dst_page != src_page) + kunmap_atomic(src_kaddr, KM_USER1); +} + +void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, + unsigned long src_offset, unsigned long len) +{ + size_t cur; + size_t dst_off_in_page; + size_t src_off_in_page; + size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); + unsigned long dst_i; + unsigned long src_i; + + if (src_offset + len > dst->len) { + printk("memmove bogus src_offset %lu move len %lu len %lu\n", + src_offset, len, dst->len); + BUG_ON(1); + } + if (dst_offset + len > dst->len) { + printk("memmove bogus dst_offset %lu move len %lu len %lu\n", + dst_offset, len, dst->len); + BUG_ON(1); + } + + while(len > 0) { + dst_off_in_page = (start_offset + dst_offset) & + ((unsigned long)PAGE_CACHE_SIZE - 1); + src_off_in_page = (start_offset + src_offset) & + ((unsigned long)PAGE_CACHE_SIZE - 1); + + dst_i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT; + src_i = (start_offset + src_offset) >> PAGE_CACHE_SHIFT; + + cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - + src_off_in_page)); + cur = min_t(unsigned long, cur, + (unsigned long)(PAGE_CACHE_SIZE - dst_off_in_page)); + + copy_pages(extent_buffer_page(dst, dst_i), + extent_buffer_page(dst, src_i), + dst_off_in_page, src_off_in_page, cur); + + src_offset += cur; + dst_offset += cur; + len -= cur; + } +} +EXPORT_SYMBOL(memcpy_extent_buffer); + +void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, + unsigned long src_offset, unsigned long len) +{ + size_t cur; + size_t dst_off_in_page; + size_t src_off_in_page; + unsigned long dst_end = dst_offset + len - 1; + unsigned long src_end = src_offset + len - 1; + size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); + unsigned long dst_i; + unsigned long src_i; + + if (src_offset + len > dst->len) { + printk("memmove bogus src_offset %lu move len %lu len %lu\n", + src_offset, len, dst->len); + BUG_ON(1); + } + if (dst_offset + len > dst->len) { + printk("memmove bogus dst_offset %lu move len %lu len %lu\n", + dst_offset, len, dst->len); + BUG_ON(1); + } + if (dst_offset < src_offset) { + memcpy_extent_buffer(dst, dst_offset, src_offset, len); + return; + } + while(len > 0) { + dst_i = (start_offset + dst_end) >> PAGE_CACHE_SHIFT; + src_i = (start_offset + src_end) >> PAGE_CACHE_SHIFT; + + dst_off_in_page = (start_offset + dst_end) & + ((unsigned long)PAGE_CACHE_SIZE - 1); + src_off_in_page = (start_offset + src_end) & + ((unsigned long)PAGE_CACHE_SIZE - 1); + + cur = min_t(unsigned long, len, src_off_in_page + 1); + cur = min(cur, dst_off_in_page + 1); + move_pages(extent_buffer_page(dst, dst_i), + extent_buffer_page(dst, src_i), + dst_off_in_page - cur + 1, + src_off_in_page - cur + 1, cur); + + dst_end -= cur; + src_end -= cur; + len -= cur; + } +} +EXPORT_SYMBOL(memmove_extent_buffer); diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h new file mode 100644 index 000000000000..06be1fe84b29 --- /dev/null +++ b/fs/btrfs/extent_io.h @@ -0,0 +1,193 @@ +#ifndef __EXTENTIO__ +#define __EXTENTIO__ + +#include + +/* bits for the extent state */ +#define EXTENT_DIRTY 1 +#define EXTENT_WRITEBACK (1 << 1) +#define EXTENT_UPTODATE (1 << 2) +#define EXTENT_LOCKED (1 << 3) +#define EXTENT_NEW (1 << 4) +#define EXTENT_DELALLOC (1 << 5) +#define EXTENT_DEFRAG (1 << 6) +#define EXTENT_DEFRAG_DONE (1 << 7) +#define EXTENT_BUFFER_FILLED (1 << 8) +#define EXTENT_CSUM (1 << 9) +#define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK) + +/* + * page->private values. Every page that is controlled by the extent + * map has page->private set to one. + */ +#define EXTENT_PAGE_PRIVATE 1 +#define EXTENT_PAGE_PRIVATE_FIRST_PAGE 3 + +struct extent_io_ops { + int (*fill_delalloc)(struct inode *inode, u64 start, u64 end); + int (*writepage_io_hook)(struct page *page, u64 start, u64 end); + int (*readpage_io_hook)(struct page *page, u64 start, u64 end); + int (*readpage_end_io_hook)(struct page *page, u64 start, u64 end); + void (*writepage_end_io_hook)(struct page *page, u64 start, u64 end); +}; + +struct extent_io_tree { + struct rb_root state; + struct address_space *mapping; + u64 dirty_bytes; + rwlock_t lock; + struct extent_io_ops *ops; + spinlock_t lru_lock; + struct list_head buffer_lru; + int lru_size; +}; + +struct extent_state { + u64 start; + u64 end; /* inclusive */ + int in_tree; + struct rb_node rb_node; + wait_queue_head_t wq; + atomic_t refs; + unsigned long state; + + /* for use by the FS */ + u64 private; + + struct list_head list; +}; + +struct extent_buffer { + u64 start; + unsigned long len; + char *map_token; + char *kaddr; + unsigned long map_start; + unsigned long map_len; + struct page *first_page; + struct list_head lru; + atomic_t refs; + int flags; +}; + +struct extent_map_tree; + +typedef struct extent_map *(get_extent_t)(struct inode *inode, + struct page *page, + size_t page_offset, + u64 start, u64 len, + int create); + +void extent_io_tree_init(struct extent_io_tree *tree, + struct address_space *mapping, gfp_t mask); +void extent_io_tree_empty_lru(struct extent_io_tree *tree); +int try_release_extent_mapping(struct extent_map_tree *map, + struct extent_io_tree *tree, struct page *page); +int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); +int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); +int extent_read_full_page(struct extent_io_tree *tree, struct page *page, + get_extent_t *get_extent); +int __init extent_io_init(void); +void extent_io_exit(void); + +u64 count_range_bits(struct extent_io_tree *tree, + u64 *start, u64 search_end, + u64 max_bytes, unsigned long bits); + +int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end, + int bits, int filled); +int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, + int bits, gfp_t mask); +int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, + int bits, gfp_t mask); +int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask); +int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask); +int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask); +int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask); +int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end, + gfp_t mask); +int find_first_extent_bit(struct extent_io_tree *tree, u64 start, + u64 *start_ret, u64 *end_ret, int bits); +int extent_invalidatepage(struct extent_io_tree *tree, + struct page *page, unsigned long offset); +int extent_write_full_page(struct extent_io_tree *tree, struct page *page, + get_extent_t *get_extent, + struct writeback_control *wbc); +int extent_writepages(struct extent_io_tree *tree, + struct address_space *mapping, + get_extent_t *get_extent, + struct writeback_control *wbc); +int extent_readpages(struct extent_io_tree *tree, + struct address_space *mapping, + struct list_head *pages, unsigned nr_pages, + get_extent_t get_extent); +int extent_prepare_write(struct extent_io_tree *tree, + struct inode *inode, struct page *page, + unsigned from, unsigned to, get_extent_t *get_extent); +int extent_commit_write(struct extent_io_tree *tree, + struct inode *inode, struct page *page, + unsigned from, unsigned to); +sector_t extent_bmap(struct address_space *mapping, sector_t iblock, + get_extent_t *get_extent); +int set_range_dirty(struct extent_io_tree *tree, u64 start, u64 end); +int set_state_private(struct extent_io_tree *tree, u64 start, u64 private); +int get_state_private(struct extent_io_tree *tree, u64 start, u64 *private); +void set_page_extent_mapped(struct page *page); + +struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree, + u64 start, unsigned long len, + struct page *page0, + gfp_t mask); +struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree, + u64 start, unsigned long len, + gfp_t mask); +void free_extent_buffer(struct extent_buffer *eb); +int read_extent_buffer_pages(struct extent_io_tree *tree, + struct extent_buffer *eb, u64 start, int wait); + +static inline void extent_buffer_get(struct extent_buffer *eb) +{ + atomic_inc(&eb->refs); +} + +int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv, + unsigned long start, + unsigned long len); +void read_extent_buffer(struct extent_buffer *eb, void *dst, + unsigned long start, + unsigned long len); +void write_extent_buffer(struct extent_buffer *eb, const void *src, + unsigned long start, unsigned long len); +void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src, + unsigned long dst_offset, unsigned long src_offset, + unsigned long len); +void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, + unsigned long src_offset, unsigned long len); +void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, + unsigned long src_offset, unsigned long len); +void memset_extent_buffer(struct extent_buffer *eb, char c, + unsigned long start, unsigned long len); +int wait_on_extent_buffer_writeback(struct extent_io_tree *tree, + struct extent_buffer *eb); +int clear_extent_buffer_dirty(struct extent_io_tree *tree, + struct extent_buffer *eb); +int set_extent_buffer_dirty(struct extent_io_tree *tree, + struct extent_buffer *eb); +int set_extent_buffer_uptodate(struct extent_io_tree *tree, + struct extent_buffer *eb); +int extent_buffer_uptodate(struct extent_io_tree *tree, + struct extent_buffer *eb); +int map_extent_buffer(struct extent_buffer *eb, unsigned long offset, + unsigned long min_len, char **token, char **map, + unsigned long *map_start, + unsigned long *map_len, int km); +int map_private_extent_buffer(struct extent_buffer *eb, unsigned long offset, + unsigned long min_len, char **token, char **map, + unsigned long *map_start, + unsigned long *map_len, int km); +void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km); +#endif diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c index 010a287fbd71..268ad8facf6e 100644 --- a/fs/btrfs/extent_map.c +++ b/fs/btrfs/extent_map.c @@ -1,17 +1,10 @@ -#include -#include -#include -#include +#include #include -#include -#include +#include #include #include -#include -#include #include -#include -#include +#include #include "extent_map.h" /* temporary define until extent_map moves out of btrfs */ @@ -21,27 +14,6 @@ struct kmem_cache *btrfs_cache_create(const char *name, size_t size, unsigned long)); static struct kmem_cache *extent_map_cache; -static struct kmem_cache *extent_state_cache; -static struct kmem_cache *extent_buffer_cache; - -static LIST_HEAD(buffers); -static LIST_HEAD(states); - -static spinlock_t state_lock = SPIN_LOCK_UNLOCKED; -#define BUFFER_LRU_MAX 64 - -struct tree_entry { - u64 start; - u64 end; - int in_tree; - struct rb_node rb_node; -}; - -struct extent_page_data { - struct bio *bio; - struct extent_map_tree *tree; - get_extent_t *get_extent; -}; int __init extent_map_init(void) { @@ -50,72 +22,23 @@ int __init extent_map_init(void) NULL); if (!extent_map_cache) return -ENOMEM; - extent_state_cache = btrfs_cache_create("extent_state", - sizeof(struct extent_state), 0, - NULL); - if (!extent_state_cache) - goto free_map_cache; - extent_buffer_cache = btrfs_cache_create("extent_buffers", - sizeof(struct extent_buffer), 0, - NULL); - if (!extent_buffer_cache) - goto free_state_cache; return 0; - -free_state_cache: - kmem_cache_destroy(extent_state_cache); -free_map_cache: - kmem_cache_destroy(extent_map_cache); - return -ENOMEM; } void extent_map_exit(void) { - struct extent_state *state; - - while (!list_empty(&states)) { - state = list_entry(states.next, struct extent_state, list); - printk("state leak: start %Lu end %Lu state %lu in tree %d refs %d\n", state->start, state->end, state->state, state->in_tree, atomic_read(&state->refs)); - list_del(&state->list); - kmem_cache_free(extent_state_cache, state); - - } - if (extent_map_cache) kmem_cache_destroy(extent_map_cache); - if (extent_state_cache) - kmem_cache_destroy(extent_state_cache); - if (extent_buffer_cache) - kmem_cache_destroy(extent_buffer_cache); } -void extent_map_tree_init(struct extent_map_tree *tree, - struct address_space *mapping, gfp_t mask) +void extent_map_tree_init(struct extent_map_tree *tree, gfp_t mask) { tree->map.rb_node = NULL; - tree->state.rb_node = NULL; - tree->ops = NULL; - tree->dirty_bytes = 0; - rwlock_init(&tree->lock); - spin_lock_init(&tree->lru_lock); - tree->mapping = mapping; - INIT_LIST_HEAD(&tree->buffer_lru); - tree->lru_size = 0; + tree->last = NULL; + spin_lock_init(&tree->lock); } EXPORT_SYMBOL(extent_map_tree_init); -void extent_map_tree_empty_lru(struct extent_map_tree *tree) -{ - struct extent_buffer *eb; - while(!list_empty(&tree->buffer_lru)) { - eb = list_entry(tree->buffer_lru.next, struct extent_buffer, - lru); - list_del_init(&eb->lru); - free_extent_buffer(eb); - } -} -EXPORT_SYMBOL(extent_map_tree_empty_lru); - struct extent_map *alloc_extent_map(gfp_t mask) { struct extent_map *em; @@ -123,6 +46,7 @@ struct extent_map *alloc_extent_map(gfp_t mask) if (!em || IS_ERR(em)) return em; em->in_tree = 0; + em->flags = 0; atomic_set(&em->refs, 1); return em; } @@ -132,6 +56,7 @@ void free_extent_map(struct extent_map *em) { if (!em) return; + WARN_ON(atomic_read(&em->refs) == 0); if (atomic_dec_and_test(&em->refs)) { WARN_ON(em->in_tree); kmem_cache_free(extent_map_cache, em); @@ -139,64 +64,28 @@ void free_extent_map(struct extent_map *em) } EXPORT_SYMBOL(free_extent_map); - -struct extent_state *alloc_extent_state(gfp_t mask) -{ - struct extent_state *state; - unsigned long flags; - - state = kmem_cache_alloc(extent_state_cache, mask); - if (!state || IS_ERR(state)) - return state; - state->state = 0; - state->in_tree = 0; - state->private = 0; - - spin_lock_irqsave(&state_lock, flags); - list_add(&state->list, &states); - spin_unlock_irqrestore(&state_lock, flags); - - atomic_set(&state->refs, 1); - init_waitqueue_head(&state->wq); - return state; -} -EXPORT_SYMBOL(alloc_extent_state); - -void free_extent_state(struct extent_state *state) -{ - unsigned long flags; - if (!state) - return; - if (atomic_dec_and_test(&state->refs)) { - WARN_ON(state->in_tree); - spin_lock_irqsave(&state_lock, flags); - list_del(&state->list); - spin_unlock_irqrestore(&state_lock, flags); - kmem_cache_free(extent_state_cache, state); - } -} -EXPORT_SYMBOL(free_extent_state); - static struct rb_node *tree_insert(struct rb_root *root, u64 offset, struct rb_node *node) { struct rb_node ** p = &root->rb_node; struct rb_node * parent = NULL; - struct tree_entry *entry; + struct extent_map *entry; while(*p) { parent = *p; - entry = rb_entry(parent, struct tree_entry, rb_node); + entry = rb_entry(parent, struct extent_map, rb_node); + + WARN_ON(!entry->in_tree); if (offset < entry->start) p = &(*p)->rb_left; - else if (offset > entry->end) + else if (offset >= extent_map_end(entry)) p = &(*p)->rb_right; else return parent; } - entry = rb_entry(node, struct tree_entry, rb_node); + entry = rb_entry(node, struct extent_map, rb_node); entry->in_tree = 1; rb_link_node(node, parent, p); rb_insert_color(node, root); @@ -210,17 +99,19 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 offset, struct rb_node * n = root->rb_node; struct rb_node *prev = NULL; struct rb_node *orig_prev = NULL; - struct tree_entry *entry; - struct tree_entry *prev_entry = NULL; + struct extent_map *entry; + struct extent_map *prev_entry = NULL; while(n) { - entry = rb_entry(n, struct tree_entry, rb_node); + entry = rb_entry(n, struct extent_map, rb_node); prev = n; prev_entry = entry; + WARN_ON(!entry->in_tree); + if (offset < entry->start) n = n->rb_left; - else if (offset > entry->end) + else if (offset >= extent_map_end(entry)) n = n->rb_right; else return n; @@ -228,19 +119,19 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 offset, if (prev_ret) { orig_prev = prev; - while(prev && offset > prev_entry->end) { + while(prev && offset >= extent_map_end(prev_entry)) { prev = rb_next(prev); - prev_entry = rb_entry(prev, struct tree_entry, rb_node); + prev_entry = rb_entry(prev, struct extent_map, rb_node); } *prev_ret = prev; prev = orig_prev; } if (next_ret) { - prev_entry = rb_entry(prev, struct tree_entry, rb_node); + prev_entry = rb_entry(prev, struct extent_map, rb_node); while(prev && offset < prev_entry->start) { prev = rb_prev(prev); - prev_entry = rb_entry(prev, struct tree_entry, rb_node); + prev_entry = rb_entry(prev, struct extent_map, rb_node); } *next_ret = prev; } @@ -257,22 +148,26 @@ static inline struct rb_node *tree_search(struct rb_root *root, u64 offset) return ret; } -static int tree_delete(struct rb_root *root, u64 offset) +static int mergable_maps(struct extent_map *prev, struct extent_map *next) { - struct rb_node *node; - struct tree_entry *entry; - - node = __tree_search(root, offset, NULL, NULL); - if (!node) - return -ENOENT; - entry = rb_entry(node, struct tree_entry, rb_node); - entry->in_tree = 0; - rb_erase(node, root); + if (extent_map_end(prev) == next->start && + prev->flags == next->flags && + prev->bdev == next->bdev && + ((next->block_start == EXTENT_MAP_HOLE && + prev->block_start == EXTENT_MAP_HOLE) || + (next->block_start == EXTENT_MAP_INLINE && + prev->block_start == EXTENT_MAP_INLINE) || + (next->block_start == EXTENT_MAP_DELALLOC && + prev->block_start == EXTENT_MAP_DELALLOC) || + (next->block_start < EXTENT_MAP_LAST_BYTE - 1 && + next->block_start == extent_map_block_end(prev)))) { + return 1; + } return 0; } /* - * add_extent_mapping tries a simple backward merge with existing + * add_extent_mapping tries a simple forward/backward merge with existing * mappings. The extent_map struct passed in will be inserted into * the tree directly (no copies made, just a reference taken). */ @@ -280,13 +175,12 @@ int add_extent_mapping(struct extent_map_tree *tree, struct extent_map *em) { int ret = 0; - struct extent_map *prev = NULL; + struct extent_map *merge = NULL; struct rb_node *rb; - write_lock_irq(&tree->lock); - rb = tree_insert(&tree->map, em->end, &em->rb_node); + rb = tree_insert(&tree->map, em->start, &em->rb_node); if (rb) { - prev = rb_entry(rb, struct extent_map, rb_node); + merge = rb_entry(rb, struct extent_map, rb_node); ret = -EEXIST; goto out; } @@ -294,53 +188,60 @@ int add_extent_mapping(struct extent_map_tree *tree, if (em->start != 0) { rb = rb_prev(&em->rb_node); if (rb) - prev = rb_entry(rb, struct extent_map, rb_node); - if (prev && prev->end + 1 == em->start && - ((em->block_start == EXTENT_MAP_HOLE && - prev->block_start == EXTENT_MAP_HOLE) || - (em->block_start == EXTENT_MAP_INLINE && - prev->block_start == EXTENT_MAP_INLINE) || - (em->block_start == EXTENT_MAP_DELALLOC && - prev->block_start == EXTENT_MAP_DELALLOC) || - (em->block_start < EXTENT_MAP_DELALLOC - 1 && - em->block_start == prev->block_end + 1))) { - em->start = prev->start; - em->block_start = prev->block_start; - rb_erase(&prev->rb_node, &tree->map); - prev->in_tree = 0; - free_extent_map(prev); + merge = rb_entry(rb, struct extent_map, rb_node); + if (rb && mergable_maps(merge, em)) { + em->start = merge->start; + em->len += merge->len; + em->block_start = merge->block_start; + merge->in_tree = 0; + rb_erase(&merge->rb_node, &tree->map); + free_extent_map(merge); } } + rb = rb_next(&em->rb_node); + if (rb) + merge = rb_entry(rb, struct extent_map, rb_node); + if (rb && mergable_maps(em, merge)) { + em->len += merge->len; + rb_erase(&merge->rb_node, &tree->map); + merge->in_tree = 0; + free_extent_map(merge); + } + tree->last = em; out: - write_unlock_irq(&tree->lock); return ret; } EXPORT_SYMBOL(add_extent_mapping); +static u64 range_end(u64 start, u64 len) +{ + if (start + len < start) + return (u64)-1; + return start + len; +} + /* * lookup_extent_mapping returns the first extent_map struct in the - * tree that intersects the [start, end] (inclusive) range. There may + * tree that intersects the [start, len] range. There may * be additional objects in the tree that intersect, so check the object * returned carefully to make sure you don't need additional lookups. */ struct extent_map *lookup_extent_mapping(struct extent_map_tree *tree, - u64 start, u64 end) + u64 start, u64 len) { struct extent_map *em; struct rb_node *rb_node; - struct rb_node *prev = NULL; - struct rb_node *next = NULL; + struct rb_node *prev = NULL; struct rb_node *next = NULL; u64 end = range_end(start, len); em = tree->last; if (em && end > em->start && start < extent_map_end(em)) goto found; - read_lock_irq(&tree->lock); rb_node = __tree_search(&tree->map, start, &prev, &next); if (!rb_node && prev) { em = rb_entry(prev, struct extent_map, rb_node); - if (em->start <= end && em->end >= start) + if (end > em->start && start < extent_map_end(em)) goto found; } if (!rb_node && next) { em = rb_entry(next, struct extent_map, rb_node); - if (em->start <= end && em->end >= start) + if (end > em->start && start < extent_map_end(em)) goto found; } if (!rb_node) { @@ -352,14 +253,16 @@ struct extent_map *lookup_extent_mapping(struct extent_map_tree *tree, goto out; } em = rb_entry(rb_node, struct extent_map, rb_node); - if (em->end < start || em->start > end) { - em = NULL; - goto out; - } + if (end > em->start && start < extent_map_end(em)) + goto found; + + em = NULL; + goto out; + found: atomic_inc(&em->refs); + tree->last = em; out: - read_unlock_irq(&tree->lock); return em; } EXPORT_SYMBOL(lookup_extent_mapping); @@ -370,2866 +273,12 @@ EXPORT_SYMBOL(lookup_extent_mapping); */ int remove_extent_mapping(struct extent_map_tree *tree, struct extent_map *em) { - int ret; + int ret = 0; - write_lock_irq(&tree->lock); - ret = tree_delete(&tree->map, em->end); - write_unlock_irq(&tree->lock); + rb_erase(&em->rb_node, &tree->map); + em->in_tree = 0; + if (tree->last == em) + tree->last = NULL; return ret; } EXPORT_SYMBOL(remove_extent_mapping); - -/* - * utility function to look for merge candidates inside a given range. - * Any extents with matching state are merged together into a single - * extent in the tree. Extents with EXTENT_IO in their state field - * are not merged because the end_io handlers need to be able to do - * operations on them without sleeping (or doing allocations/splits). - * - * This should be called with the tree lock held. - */ -static int merge_state(struct extent_map_tree *tree, - struct extent_state *state) -{ - struct extent_state *other; - struct rb_node *other_node; - - if (state->state & EXTENT_IOBITS) - return 0; - - other_node = rb_prev(&state->rb_node); - if (other_node) { - other = rb_entry(other_node, struct extent_state, rb_node); - if (other->end == state->start - 1 && - other->state == state->state) { - state->start = other->start; - other->in_tree = 0; - rb_erase(&other->rb_node, &tree->state); - free_extent_state(other); - } - } - other_node = rb_next(&state->rb_node); - if (other_node) { - other = rb_entry(other_node, struct extent_state, rb_node); - if (other->start == state->end + 1 && - other->state == state->state) { - other->start = state->start; - state->in_tree = 0; - rb_erase(&state->rb_node, &tree->state); - free_extent_state(state); - } - } - return 0; -} - -/* - * insert an extent_state struct into the tree. 'bits' are set on the - * struct before it is inserted. - * - * This may return -EEXIST if the extent is already there, in which case the - * state struct is freed. - * - * The tree lock is not taken internally. This is a utility function and - * probably isn't what you want to call (see set/clear_extent_bit). - */ -static int insert_state(struct extent_map_tree *tree, - struct extent_state *state, u64 start, u64 end, - int bits) -{ - struct rb_node *node; - - if (end < start) { - printk("end < start %Lu %Lu\n", end, start); - WARN_ON(1); - } - if (bits & EXTENT_DIRTY) - tree->dirty_bytes += end - start + 1; - state->state |= bits; - state->start = start; - state->end = end; - node = tree_insert(&tree->state, end, &state->rb_node); - if (node) { - struct extent_state *found; - found = rb_entry(node, struct extent_state, rb_node); - printk("found node %Lu %Lu on insert of %Lu %Lu\n", found->start, found->end, start, end); - free_extent_state(state); - return -EEXIST; - } - merge_state(tree, state); - return 0; -} - -/* - * split a given extent state struct in two, inserting the preallocated - * struct 'prealloc' as the newly created second half. 'split' indicates an - * offset inside 'orig' where it should be split. - * - * Before calling, - * the tree has 'orig' at [orig->start, orig->end]. After calling, there - * are two extent state structs in the tree: - * prealloc: [orig->start, split - 1] - * orig: [ split, orig->end ] - * - * The tree locks are not taken by this function. They need to be held - * by the caller. - */ -static int split_state(struct extent_map_tree *tree, struct extent_state *orig, - struct extent_state *prealloc, u64 split) -{ - struct rb_node *node; - prealloc->start = orig->start; - prealloc->end = split - 1; - prealloc->state = orig->state; - orig->start = split; - - node = tree_insert(&tree->state, prealloc->end, &prealloc->rb_node); - if (node) { - struct extent_state *found; - found = rb_entry(node, struct extent_state, rb_node); - printk("found node %Lu %Lu on insert of %Lu %Lu\n", found->start, found->end, prealloc->start, prealloc->end); - free_extent_state(prealloc); - return -EEXIST; - } - return 0; -} - -/* - * utility function to clear some bits in an extent state struct. - * it will optionally wake up any one waiting on this state (wake == 1), or - * forcibly remove the state from the tree (delete == 1). - * - * If no bits are set on the state struct after clearing things, the - * struct is freed and removed from the tree - */ -static int clear_state_bit(struct extent_map_tree *tree, - struct extent_state *state, int bits, int wake, - int delete) -{ - int ret = state->state & bits; - - if ((bits & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) { - u64 range = state->end - state->start + 1; - WARN_ON(range > tree->dirty_bytes); - tree->dirty_bytes -= range; - } - state->state &= ~bits; - if (wake) - wake_up(&state->wq); - if (delete || state->state == 0) { - if (state->in_tree) { - rb_erase(&state->rb_node, &tree->state); - state->in_tree = 0; - free_extent_state(state); - } else { - WARN_ON(1); - } - } else { - merge_state(tree, state); - } - return ret; -} - -/* - * clear some bits on a range in the tree. This may require splitting - * or inserting elements in the tree, so the gfp mask is used to - * indicate which allocations or sleeping are allowed. - * - * pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove - * the given range from the tree regardless of state (ie for truncate). - * - * the range [start, end] is inclusive. - * - * This takes the tree lock, and returns < 0 on error, > 0 if any of the - * bits were already set, or zero if none of the bits were already set. - */ -int clear_extent_bit(struct extent_map_tree *tree, u64 start, u64 end, - int bits, int wake, int delete, gfp_t mask) -{ - struct extent_state *state; - struct extent_state *prealloc = NULL; - struct rb_node *node; - unsigned long flags; - int err; - int set = 0; - -again: - if (!prealloc && (mask & __GFP_WAIT)) { - prealloc = alloc_extent_state(mask); - if (!prealloc) - return -ENOMEM; - } - - write_lock_irqsave(&tree->lock, flags); - /* - * this search will find the extents that end after - * our range starts - */ - node = tree_search(&tree->state, start); - if (!node) - goto out; - state = rb_entry(node, struct extent_state, rb_node); - if (state->start > end) - goto out; - WARN_ON(state->end < start); - - /* - * | ---- desired range ---- | - * | state | or - * | ------------- state -------------- | - * - * We need to split the extent we found, and may flip - * bits on second half. - * - * If the extent we found extends past our range, we - * just split and search again. It'll get split again - * the next time though. - * - * If the extent we found is inside our range, we clear - * the desired bit on it. - */ - - if (state->start < start) { - err = split_state(tree, state, prealloc, start); - BUG_ON(err == -EEXIST); - prealloc = NULL; - if (err) - goto out; - if (state->end <= end) { - start = state->end + 1; - set |= clear_state_bit(tree, state, bits, - wake, delete); - } else { - start = state->start; - } - goto search_again; - } - /* - * | ---- desired range ---- | - * | state | - * We need to split the extent, and clear the bit - * on the first half - */ - if (state->start <= end && state->end > end) { - err = split_state(tree, state, prealloc, end + 1); - BUG_ON(err == -EEXIST); - - if (wake) - wake_up(&state->wq); - set |= clear_state_bit(tree, prealloc, bits, - wake, delete); - prealloc = NULL; - goto out; - } - - start = state->end + 1; - set |= clear_state_bit(tree, state, bits, wake, delete); - goto search_again; - -out: - write_unlock_irqrestore(&tree->lock, flags); - if (prealloc) - free_extent_state(prealloc); - - return set; - -search_again: - if (start > end) - goto out; - write_unlock_irqrestore(&tree->lock, flags); - if (mask & __GFP_WAIT) - cond_resched(); - goto again; -} -EXPORT_SYMBOL(clear_extent_bit); - -static int wait_on_state(struct extent_map_tree *tree, - struct extent_state *state) -{ - DEFINE_WAIT(wait); - prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE); - read_unlock_irq(&tree->lock); - schedule(); - read_lock_irq(&tree->lock); - finish_wait(&state->wq, &wait); - return 0; -} - -/* - * waits for one or more bits to clear on a range in the state tree. - * The range [start, end] is inclusive. - * The tree lock is taken by this function - */ -int wait_extent_bit(struct extent_map_tree *tree, u64 start, u64 end, int bits) -{ - struct extent_state *state; - struct rb_node *node; - - read_lock_irq(&tree->lock); -again: - while (1) { - /* - * this search will find all the extents that end after - * our range starts - */ - node = tree_search(&tree->state, start); - if (!node) - break; - - state = rb_entry(node, struct extent_state, rb_node); - - if (state->start > end) - goto out; - - if (state->state & bits) { - start = state->start; - atomic_inc(&state->refs); - wait_on_state(tree, state); - free_extent_state(state); - goto again; - } - start = state->end + 1; - - if (start > end) - break; - - if (need_resched()) { - read_unlock_irq(&tree->lock); - cond_resched(); - read_lock_irq(&tree->lock); - } - } -out: - read_unlock_irq(&tree->lock); - return 0; -} -EXPORT_SYMBOL(wait_extent_bit); - -static void set_state_bits(struct extent_map_tree *tree, - struct extent_state *state, - int bits) -{ - if ((bits & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) { - u64 range = state->end - state->start + 1; - tree->dirty_bytes += range; - } - state->state |= bits; -} - -/* - * set some bits on a range in the tree. This may require allocations - * or sleeping, so the gfp mask is used to indicate what is allowed. - * - * If 'exclusive' == 1, this will fail with -EEXIST if some part of the - * range already has the desired bits set. The start of the existing - * range is returned in failed_start in this case. - * - * [start, end] is inclusive - * This takes the tree lock. - */ -int set_extent_bit(struct extent_map_tree *tree, u64 start, u64 end, int bits, - int exclusive, u64 *failed_start, gfp_t mask) -{ - struct extent_state *state; - struct extent_state *prealloc = NULL; - struct rb_node *node; - unsigned long flags; - int err = 0; - int set; - u64 last_start; - u64 last_end; -again: - if (!prealloc && (mask & __GFP_WAIT)) { - prealloc = alloc_extent_state(mask); - if (!prealloc) - return -ENOMEM; - } - - write_lock_irqsave(&tree->lock, flags); - /* - * this search will find all the extents that end after - * our range starts. - */ - node = tree_search(&tree->state, start); - if (!node) { - err = insert_state(tree, prealloc, start, end, bits); - prealloc = NULL; - BUG_ON(err == -EEXIST); - goto out; - } - - state = rb_entry(node, struct extent_state, rb_node); - last_start = state->start; - last_end = state->end; - - /* - * | ---- desired range ---- | - * | state | - * - * Just lock what we found and keep going - */ - if (state->start == start && state->end <= end) { - set = state->state & bits; - if (set && exclusive) { - *failed_start = state->start; - err = -EEXIST; - goto out; - } - set_state_bits(tree, state, bits); - start = state->end + 1; - merge_state(tree, state); - goto search_again; - } - - /* - * | ---- desired range ---- | - * | state | - * or - * | ------------- state -------------- | - * - * We need to split the extent we found, and may flip bits on - * second half. - * - * If the extent we found extends past our - * range, we just split and search again. It'll get split - * again the next time though. - * - * If the extent we found is inside our range, we set the - * desired bit on it. - */ - if (state->start < start) { - set = state->state & bits; - if (exclusive && set) { - *failed_start = start; - err = -EEXIST; - goto out; - } - err = split_state(tree, state, prealloc, start); - BUG_ON(err == -EEXIST); - prealloc = NULL; - if (err) - goto out; - if (state->end <= end) { - set_state_bits(tree, state, bits); - start = state->end + 1; - merge_state(tree, state); - } else { - start = state->start; - } - goto search_again; - } - /* - * | ---- desired range ---- | - * | state | or | state | - * - * There's a hole, we need to insert something in it and - * ignore the extent we found. - */ - if (state->start > start) { - u64 this_end; - if (end < last_start) - this_end = end; - else - this_end = last_start -1; - err = insert_state(tree, prealloc, start, this_end, - bits); - prealloc = NULL; - BUG_ON(err == -EEXIST); - if (err) - goto out; - start = this_end + 1; - goto search_again; - } - /* - * | ---- desired range ---- | - * | state | - * We need to split the extent, and set the bit - * on the first half - */ - if (state->start <= end && state->end > end) { - set = state->state & bits; - if (exclusive && set) { - *failed_start = start; - err = -EEXIST; - goto out; - } - err = split_state(tree, state, prealloc, end + 1); - BUG_ON(err == -EEXIST); - - set_state_bits(tree, prealloc, bits); - merge_state(tree, prealloc); - prealloc = NULL; - goto out; - } - - goto search_again; - -out: - write_unlock_irqrestore(&tree->lock, flags); - if (prealloc) - free_extent_state(prealloc); - - return err; - -search_again: - if (start > end) - goto out; - write_unlock_irqrestore(&tree->lock, flags); - if (mask & __GFP_WAIT) - cond_resched(); - goto again; -} -EXPORT_SYMBOL(set_extent_bit); - -/* wrappers around set/clear extent bit */ -int set_extent_dirty(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL, - mask); -} -EXPORT_SYMBOL(set_extent_dirty); - -int set_extent_bits(struct extent_map_tree *tree, u64 start, u64 end, - int bits, gfp_t mask) -{ - return set_extent_bit(tree, start, end, bits, 0, NULL, - mask); -} -EXPORT_SYMBOL(set_extent_bits); - -int clear_extent_bits(struct extent_map_tree *tree, u64 start, u64 end, - int bits, gfp_t mask) -{ - return clear_extent_bit(tree, start, end, bits, 0, 0, mask); -} -EXPORT_SYMBOL(clear_extent_bits); - -int set_extent_delalloc(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return set_extent_bit(tree, start, end, - EXTENT_DELALLOC | EXTENT_DIRTY, 0, NULL, - mask); -} -EXPORT_SYMBOL(set_extent_delalloc); - -int clear_extent_dirty(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return clear_extent_bit(tree, start, end, - EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, mask); -} -EXPORT_SYMBOL(clear_extent_dirty); - -int set_extent_new(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL, - mask); -} -EXPORT_SYMBOL(set_extent_new); - -int clear_extent_new(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return clear_extent_bit(tree, start, end, EXTENT_NEW, 0, 0, mask); -} -EXPORT_SYMBOL(clear_extent_new); - -int set_extent_uptodate(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, NULL, - mask); -} -EXPORT_SYMBOL(set_extent_uptodate); - -int clear_extent_uptodate(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0, mask); -} -EXPORT_SYMBOL(clear_extent_uptodate); - -int set_extent_writeback(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return set_extent_bit(tree, start, end, EXTENT_WRITEBACK, - 0, NULL, mask); -} -EXPORT_SYMBOL(set_extent_writeback); - -int clear_extent_writeback(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return clear_extent_bit(tree, start, end, EXTENT_WRITEBACK, 1, 0, mask); -} -EXPORT_SYMBOL(clear_extent_writeback); - -int wait_on_extent_writeback(struct extent_map_tree *tree, u64 start, u64 end) -{ - return wait_extent_bit(tree, start, end, EXTENT_WRITEBACK); -} -EXPORT_SYMBOL(wait_on_extent_writeback); - -/* - * locks a range in ascending order, waiting for any locked regions - * it hits on the way. [start,end] are inclusive, and this will sleep. - */ -int lock_extent(struct extent_map_tree *tree, u64 start, u64 end, gfp_t mask) -{ - int err; - u64 failed_start; - while (1) { - err = set_extent_bit(tree, start, end, EXTENT_LOCKED, 1, - &failed_start, mask); - if (err == -EEXIST && (mask & __GFP_WAIT)) { - wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED); - start = failed_start; - } else { - break; - } - WARN_ON(start > end); - } - return err; -} -EXPORT_SYMBOL(lock_extent); - -int unlock_extent(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask) -{ - return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, mask); -} -EXPORT_SYMBOL(unlock_extent); - -/* - * helper function to set pages and extents in the tree dirty - */ -int set_range_dirty(struct extent_map_tree *tree, u64 start, u64 end) -{ - unsigned long index = start >> PAGE_CACHE_SHIFT; - unsigned long end_index = end >> PAGE_CACHE_SHIFT; - struct page *page; - - while (index <= end_index) { - page = find_get_page(tree->mapping, index); - BUG_ON(!page); - __set_page_dirty_nobuffers(page); - page_cache_release(page); - index++; - } - set_extent_dirty(tree, start, end, GFP_NOFS); - return 0; -} -EXPORT_SYMBOL(set_range_dirty); - -/* - * helper function to set both pages and extents in the tree writeback - */ -int set_range_writeback(struct extent_map_tree *tree, u64 start, u64 end) -{ - unsigned long index = start >> PAGE_CACHE_SHIFT; - unsigned long end_index = end >> PAGE_CACHE_SHIFT; - struct page *page; - - while (index <= end_index) { - page = find_get_page(tree->mapping, index); - BUG_ON(!page); - set_page_writeback(page); - page_cache_release(page); - index++; - } - set_extent_writeback(tree, start, end, GFP_NOFS); - return 0; -} -EXPORT_SYMBOL(set_range_writeback); - -int find_first_extent_bit(struct extent_map_tree *tree, u64 start, - u64 *start_ret, u64 *end_ret, int bits) -{ - struct rb_node *node; - struct extent_state *state; - int ret = 1; - - read_lock_irq(&tree->lock); - /* - * this search will find all the extents that end after - * our range starts. - */ - node = tree_search(&tree->state, start); - if (!node || IS_ERR(node)) { - goto out; - } - - while(1) { - state = rb_entry(node, struct extent_state, rb_node); - if (state->end >= start && (state->state & bits)) { - *start_ret = state->start; - *end_ret = state->end; - ret = 0; - break; - } - node = rb_next(node); - if (!node) - break; - } -out: - read_unlock_irq(&tree->lock); - return ret; -} -EXPORT_SYMBOL(find_first_extent_bit); - -u64 find_lock_delalloc_range(struct extent_map_tree *tree, - u64 *start, u64 *end, u64 max_bytes) -{ - struct rb_node *node; - struct extent_state *state; - u64 cur_start = *start; - u64 found = 0; - u64 total_bytes = 0; - - write_lock_irq(&tree->lock); - /* - * this search will find all the extents that end after - * our range starts. - */ -search_again: - node = tree_search(&tree->state, cur_start); - if (!node || IS_ERR(node)) { - *end = (u64)-1; - goto out; - } - - while(1) { - state = rb_entry(node, struct extent_state, rb_node); - if (found && state->start != cur_start) { - goto out; - } - if (!(state->state & EXTENT_DELALLOC)) { - if (!found) - *end = state->end; - goto out; - } - if (!found) { - struct extent_state *prev_state; - struct rb_node *prev_node = node; - while(1) { - prev_node = rb_prev(prev_node); - if (!prev_node) - break; - prev_state = rb_entry(prev_node, - struct extent_state, - rb_node); - if (!(prev_state->state & EXTENT_DELALLOC)) - break; - state = prev_state; - node = prev_node; - } - } - if (state->state & EXTENT_LOCKED) { - DEFINE_WAIT(wait); - atomic_inc(&state->refs); - prepare_to_wait(&state->wq, &wait, - TASK_UNINTERRUPTIBLE); - write_unlock_irq(&tree->lock); - schedule(); - write_lock_irq(&tree->lock); - finish_wait(&state->wq, &wait); - free_extent_state(state); - goto search_again; - } - state->state |= EXTENT_LOCKED; - if (!found) - *start = state->start; - found++; - *end = state->end; - cur_start = state->end + 1; - node = rb_next(node); - if (!node) - break; - total_bytes += state->end - state->start + 1; - if (total_bytes >= max_bytes) - break; - } -out: - write_unlock_irq(&tree->lock); - return found; -} - -u64 count_range_bits(struct extent_map_tree *tree, - u64 *start, u64 search_end, u64 max_bytes, - unsigned long bits) -{ - struct rb_node *node; - struct extent_state *state; - u64 cur_start = *start; - u64 total_bytes = 0; - int found = 0; - - if (search_end <= cur_start) { - printk("search_end %Lu start %Lu\n", search_end, cur_start); - WARN_ON(1); - return 0; - } - - write_lock_irq(&tree->lock); - if (cur_start == 0 && bits == EXTENT_DIRTY) { - total_bytes = tree->dirty_bytes; - goto out; - } - /* - * this search will find all the extents that end after - * our range starts. - */ - node = tree_search(&tree->state, cur_start); - if (!node || IS_ERR(node)) { - goto out; - } - - while(1) { - state = rb_entry(node, struct extent_state, rb_node); - if (state->start > search_end) - break; - if (state->end >= cur_start && (state->state & bits)) { - total_bytes += min(search_end, state->end) + 1 - - max(cur_start, state->start); - if (total_bytes >= max_bytes) - break; - if (!found) { - *start = state->start; - found = 1; - } - } - node = rb_next(node); - if (!node) - break; - } -out: - write_unlock_irq(&tree->lock); - return total_bytes; -} -/* - * helper function to lock both pages and extents in the tree. - * pages must be locked first. - */ -int lock_range(struct extent_map_tree *tree, u64 start, u64 end) -{ - unsigned long index = start >> PAGE_CACHE_SHIFT; - unsigned long end_index = end >> PAGE_CACHE_SHIFT; - struct page *page; - int err; - - while (index <= end_index) { - page = grab_cache_page(tree->mapping, index); - if (!page) { - err = -ENOMEM; - goto failed; - } - if (IS_ERR(page)) { - err = PTR_ERR(page); - goto failed; - } - index++; - } - lock_extent(tree, start, end, GFP_NOFS); - return 0; - -failed: - /* - * we failed above in getting the page at 'index', so we undo here - * up to but not including the page at 'index' - */ - end_index = index; - index = start >> PAGE_CACHE_SHIFT; - while (index < end_index) { - page = find_get_page(tree->mapping, index); - unlock_page(page); - page_cache_release(page); - index++; - } - return err; -} -EXPORT_SYMBOL(lock_range); - -/* - * helper function to unlock both pages and extents in the tree. - */ -int unlock_range(struct extent_map_tree *tree, u64 start, u64 end) -{ - unsigned long index = start >> PAGE_CACHE_SHIFT; - unsigned long end_index = end >> PAGE_CACHE_SHIFT; - struct page *page; - - while (index <= end_index) { - page = find_get_page(tree->mapping, index); - unlock_page(page); - page_cache_release(page); - index++; - } - unlock_extent(tree, start, end, GFP_NOFS); - return 0; -} -EXPORT_SYMBOL(unlock_range); - -int set_state_private(struct extent_map_tree *tree, u64 start, u64 private) -{ - struct rb_node *node; - struct extent_state *state; - int ret = 0; - - write_lock_irq(&tree->lock); - /* - * this search will find all the extents that end after - * our range starts. - */ - node = tree_search(&tree->state, start); - if (!node || IS_ERR(node)) { - ret = -ENOENT; - goto out; - } - state = rb_entry(node, struct extent_state, rb_node); - if (state->start != start) { - ret = -ENOENT; - goto out; - } - state->private = private; -out: - write_unlock_irq(&tree->lock); - return ret; -} - -int get_state_private(struct extent_map_tree *tree, u64 start, u64 *private) -{ - struct rb_node *node; - struct extent_state *state; - int ret = 0; - - read_lock_irq(&tree->lock); - /* - * this search will find all the extents that end after - * our range starts. - */ - node = tree_search(&tree->state, start); - if (!node || IS_ERR(node)) { - ret = -ENOENT; - goto out; - } - state = rb_entry(node, struct extent_state, rb_node); - if (state->start != start) { - ret = -ENOENT; - goto out; - } - *private = state->private; -out: - read_unlock_irq(&tree->lock); - return ret; -} - -/* - * searches a range in the state tree for a given mask. - * If 'filled' == 1, this returns 1 only if ever extent in the tree - * has the bits set. Otherwise, 1 is returned if any bit in the - * range is found set. - */ -int test_range_bit(struct extent_map_tree *tree, u64 start, u64 end, - int bits, int filled) -{ - struct extent_state *state = NULL; - struct rb_node *node; - int bitset = 0; - - read_lock_irq(&tree->lock); - node = tree_search(&tree->state, start); - while (node && start <= end) { - state = rb_entry(node, struct extent_state, rb_node); - - if (filled && state->start > start) { - bitset = 0; - break; - } - - if (state->start > end) - break; - - if (state->state & bits) { - bitset = 1; - if (!filled) - break; - } else if (filled) { - bitset = 0; - break; - } - start = state->end + 1; - if (start > end) - break; - node = rb_next(node); - if (!node) { - if (filled) - bitset = 0; - break; - } - } - read_unlock_irq(&tree->lock); - return bitset; -} -EXPORT_SYMBOL(test_range_bit); - -/* - * helper function to set a given page up to date if all the - * extents in the tree for that page are up to date - */ -static int check_page_uptodate(struct extent_map_tree *tree, - struct page *page) -{ - u64 start = (u64)page->index << PAGE_CACHE_SHIFT; - u64 end = start + PAGE_CACHE_SIZE - 1; - if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1)) - SetPageUptodate(page); - return 0; -} - -/* - * helper function to unlock a page if all the extents in the tree - * for that page are unlocked - */ -static int check_page_locked(struct extent_map_tree *tree, - struct page *page) -{ - u64 start = (u64)page->index << PAGE_CACHE_SHIFT; - u64 end = start + PAGE_CACHE_SIZE - 1; - if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0)) - unlock_page(page); - return 0; -} - -/* - * helper function to end page writeback if all the extents - * in the tree for that page are done with writeback - */ -static int check_page_writeback(struct extent_map_tree *tree, - struct page *page) -{ - u64 start = (u64)page->index << PAGE_CACHE_SHIFT; - u64 end = start + PAGE_CACHE_SIZE - 1; - if (!test_range_bit(tree, start, end, EXTENT_WRITEBACK, 0)) - end_page_writeback(page); - return 0; -} - -/* lots and lots of room for performance fixes in the end_bio funcs */ - -/* - * after a writepage IO is done, we need to: - * clear the uptodate bits on error - * clear the writeback bits in the extent tree for this IO - * end_page_writeback if the page has no more pending IO - * - * Scheduling is not allowed, so the extent state tree is expected - * to have one and only one object corresponding to this IO. - */ -#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23) -static void end_bio_extent_writepage(struct bio *bio, int err) -#else -static int end_bio_extent_writepage(struct bio *bio, - unsigned int bytes_done, int err) -#endif -{ - const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); - struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; - struct extent_map_tree *tree = bio->bi_private; - u64 start; - u64 end; - int whole_page; - -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) - if (bio->bi_size) - return 1; -#endif - - do { - struct page *page = bvec->bv_page; - start = ((u64)page->index << PAGE_CACHE_SHIFT) + - bvec->bv_offset; - end = start + bvec->bv_len - 1; - - if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE) - whole_page = 1; - else - whole_page = 0; - - if (--bvec >= bio->bi_io_vec) - prefetchw(&bvec->bv_page->flags); - - if (!uptodate) { - clear_extent_uptodate(tree, start, end, GFP_ATOMIC); - ClearPageUptodate(page); - SetPageError(page); - } - clear_extent_writeback(tree, start, end, GFP_ATOMIC); - - if (whole_page) - end_page_writeback(page); - else - check_page_writeback(tree, page); - if (tree->ops && tree->ops->writepage_end_io_hook) - tree->ops->writepage_end_io_hook(page, start, end); - } while (bvec >= bio->bi_io_vec); - - bio_put(bio); -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) - return 0; -#endif -} - -/* - * after a readpage IO is done, we need to: - * clear the uptodate bits on error - * set the uptodate bits if things worked - * set the page up to date if all extents in the tree are uptodate - * clear the lock bit in the extent tree - * unlock the page if there are no other extents locked for it - * - * Scheduling is not allowed, so the extent state tree is expected - * to have one and only one object corresponding to this IO. - */ -#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23) -static void end_bio_extent_readpage(struct bio *bio, int err) -#else -static int end_bio_extent_readpage(struct bio *bio, - unsigned int bytes_done, int err) -#endif -{ - int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); - struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; - struct extent_map_tree *tree = bio->bi_private; - u64 start; - u64 end; - int whole_page; - int ret; - -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) - if (bio->bi_size) - return 1; -#endif - - do { - struct page *page = bvec->bv_page; - start = ((u64)page->index << PAGE_CACHE_SHIFT) + - bvec->bv_offset; - end = start + bvec->bv_len - 1; - - if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE) - whole_page = 1; - else - whole_page = 0; - - if (--bvec >= bio->bi_io_vec) - prefetchw(&bvec->bv_page->flags); - - if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) { - ret = tree->ops->readpage_end_io_hook(page, start, end); - if (ret) - uptodate = 0; - } - if (uptodate) { - set_extent_uptodate(tree, start, end, GFP_ATOMIC); - if (whole_page) - SetPageUptodate(page); - else - check_page_uptodate(tree, page); - } else { - ClearPageUptodate(page); - SetPageError(page); - } - - unlock_extent(tree, start, end, GFP_ATOMIC); - - if (whole_page) - unlock_page(page); - else - check_page_locked(tree, page); - } while (bvec >= bio->bi_io_vec); - - bio_put(bio); -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) - return 0; -#endif -} - -/* - * IO done from prepare_write is pretty simple, we just unlock - * the structs in the extent tree when done, and set the uptodate bits - * as appropriate. - */ -#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23) -static void end_bio_extent_preparewrite(struct bio *bio, int err) -#else -static int end_bio_extent_preparewrite(struct bio *bio, - unsigned int bytes_done, int err) -#endif -{ - const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); - struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; - struct extent_map_tree *tree = bio->bi_private; - u64 start; - u64 end; - -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) - if (bio->bi_size) - return 1; -#endif - - do { - struct page *page = bvec->bv_page; - start = ((u64)page->index << PAGE_CACHE_SHIFT) + - bvec->bv_offset; - end = start + bvec->bv_len - 1; - - if (--bvec >= bio->bi_io_vec) - prefetchw(&bvec->bv_page->flags); - - if (uptodate) { - set_extent_uptodate(tree, start, end, GFP_ATOMIC); - } else { - ClearPageUptodate(page); - SetPageError(page); - } - - unlock_extent(tree, start, end, GFP_ATOMIC); - - } while (bvec >= bio->bi_io_vec); - - bio_put(bio); -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23) - return 0; -#endif -} - -static struct bio * -extent_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs, - gfp_t gfp_flags) -{ - struct bio *bio; - - bio = bio_alloc(gfp_flags, nr_vecs); - - if (bio == NULL && (current->flags & PF_MEMALLOC)) { - while (!bio && (nr_vecs /= 2)) - bio = bio_alloc(gfp_flags, nr_vecs); - } - - if (bio) { - bio->bi_bdev = bdev; - bio->bi_sector = first_sector; - } - return bio; -} - -static int submit_one_bio(int rw, struct bio *bio) -{ - u64 maxsector; - int ret = 0; - - bio_get(bio); - - maxsector = bio->bi_bdev->bd_inode->i_size >> 9; - if (maxsector < bio->bi_sector) { - printk("sector too large max %Lu got %llu\n", maxsector, - (unsigned long long)bio->bi_sector); - WARN_ON(1); - } - - submit_bio(rw, bio); - if (bio_flagged(bio, BIO_EOPNOTSUPP)) - ret = -EOPNOTSUPP; - bio_put(bio); - return ret; -} - -static int submit_extent_page(int rw, struct extent_map_tree *tree, - struct page *page, sector_t sector, - size_t size, unsigned long offset, - struct block_device *bdev, - struct bio **bio_ret, - unsigned long max_pages, - bio_end_io_t end_io_func) -{ - int ret = 0; - struct bio *bio; - int nr; - - if (bio_ret && *bio_ret) { - bio = *bio_ret; - if (bio->bi_sector + (bio->bi_size >> 9) != sector || - bio_add_page(bio, page, size, offset) < size) { - ret = submit_one_bio(rw, bio); - bio = NULL; - } else { - return 0; - } - } - nr = min_t(int, max_pages, bio_get_nr_vecs(bdev)); - bio = extent_bio_alloc(bdev, sector, nr, GFP_NOFS | __GFP_HIGH); - if (!bio) { - printk("failed to allocate bio nr %d\n", nr); - } - bio_add_page(bio, page, size, offset); - bio->bi_end_io = end_io_func; - bio->bi_private = tree; - if (bio_ret) { - *bio_ret = bio; - } else { - ret = submit_one_bio(rw, bio); - } - - return ret; -} - -void set_page_extent_mapped(struct page *page) -{ - if (!PagePrivate(page)) { - SetPagePrivate(page); - WARN_ON(!page->mapping->a_ops->invalidatepage); - set_page_private(page, EXTENT_PAGE_PRIVATE); - page_cache_get(page); - } -} - -void set_page_extent_head(struct page *page, unsigned long len) -{ - set_page_private(page, EXTENT_PAGE_PRIVATE_FIRST_PAGE | len << 2); -} - -/* - * basic readpage implementation. Locked extent state structs are inserted - * into the tree that are removed when the IO is done (by the end_io - * handlers) - */ -static int __extent_read_full_page(struct extent_map_tree *tree, - struct page *page, - get_extent_t *get_extent, - struct bio **bio) -{ - struct inode *inode = page->mapping->host; - u64 start = (u64)page->index << PAGE_CACHE_SHIFT; - u64 page_end = start + PAGE_CACHE_SIZE - 1; - u64 end; - u64 cur = start; - u64 extent_offset; - u64 last_byte = i_size_read(inode); - u64 block_start; - u64 cur_end; - sector_t sector; - struct extent_map *em; - struct block_device *bdev; - int ret; - int nr = 0; - size_t page_offset = 0; - size_t iosize; - size_t blocksize = inode->i_sb->s_blocksize; - - set_page_extent_mapped(page); - - end = page_end; - lock_extent(tree, start, end, GFP_NOFS); - - while (cur <= end) { - if (cur >= last_byte) { - char *userpage; - iosize = PAGE_CACHE_SIZE - page_offset; - userpage = kmap_atomic(page, KM_USER0); - memset(userpage + page_offset, 0, iosize); - flush_dcache_page(page); - kunmap_atomic(userpage, KM_USER0); - set_extent_uptodate(tree, cur, cur + iosize - 1, - GFP_NOFS); - unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); - break; - } - em = get_extent(inode, page, page_offset, cur, end, 0); - if (IS_ERR(em) || !em) { - SetPageError(page); - unlock_extent(tree, cur, end, GFP_NOFS); - break; - } - - extent_offset = cur - em->start; - BUG_ON(em->end < cur); - BUG_ON(end < cur); - - iosize = min(em->end - cur, end - cur) + 1; - cur_end = min(em->end, end); - iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1); - sector = (em->block_start + extent_offset) >> 9; - bdev = em->bdev; - block_start = em->block_start; - free_extent_map(em); - em = NULL; - - /* we've found a hole, just zero and go on */ - if (block_start == EXTENT_MAP_HOLE) { - char *userpage; - userpage = kmap_atomic(page, KM_USER0); - memset(userpage + page_offset, 0, iosize); - flush_dcache_page(page); - kunmap_atomic(userpage, KM_USER0); - - set_extent_uptodate(tree, cur, cur + iosize - 1, - GFP_NOFS); - unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); - cur = cur + iosize; - page_offset += iosize; - continue; - } - /* the get_extent function already copied into the page */ - if (test_range_bit(tree, cur, cur_end, EXTENT_UPTODATE, 1)) { - unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); - cur = cur + iosize; - page_offset += iosize; - continue; - } - - ret = 0; - if (tree->ops && tree->ops->readpage_io_hook) { - ret = tree->ops->readpage_io_hook(page, cur, - cur + iosize - 1); - } - if (!ret) { - unsigned long nr = (last_byte >> PAGE_CACHE_SHIFT) + 1; - nr -= page->index; - ret = submit_extent_page(READ, tree, page, - sector, iosize, page_offset, - bdev, bio, nr, - end_bio_extent_readpage); - } - if (ret) - SetPageError(page); - cur = cur + iosize; - page_offset += iosize; - nr++; - } - if (!nr) { - if (!PageError(page)) - SetPageUptodate(page); - unlock_page(page); - } - return 0; -} - -int extent_read_full_page(struct extent_map_tree *tree, struct page *page, - get_extent_t *get_extent) -{ - struct bio *bio = NULL; - int ret; - - ret = __extent_read_full_page(tree, page, get_extent, &bio); - if (bio) - submit_one_bio(READ, bio); - return ret; -} -EXPORT_SYMBOL(extent_read_full_page); - -/* - * the writepage semantics are similar to regular writepage. extent - * records are inserted to lock ranges in the tree, and as dirty areas - * are found, they are marked writeback. Then the lock bits are removed - * and the end_io handler clears the writeback ranges - */ -static int __extent_writepage(struct page *page, struct writeback_control *wbc, - void *data) -{ - struct inode *inode = page->mapping->host; - struct extent_page_data *epd = data; - struct extent_map_tree *tree = epd->tree; - u64 start = (u64)page->index << PAGE_CACHE_SHIFT; - u64 delalloc_start; - u64 page_end = start + PAGE_CACHE_SIZE - 1; - u64 end; - u64 cur = start; - u64 extent_offset; - u64 last_byte = i_size_read(inode); - u64 block_start; - u64 iosize; - sector_t sector; - struct extent_map *em; - struct block_device *bdev; - int ret; - int nr = 0; - size_t page_offset = 0; - size_t blocksize; - loff_t i_size = i_size_read(inode); - unsigned long end_index = i_size >> PAGE_CACHE_SHIFT; - u64 nr_delalloc; - u64 delalloc_end; - - WARN_ON(!PageLocked(page)); - if (page->index > end_index) { - clear_extent_dirty(tree, start, page_end, GFP_NOFS); - unlock_page(page); - return 0; - } - - if (page->index == end_index) { - char *userpage; - - size_t offset = i_size & (PAGE_CACHE_SIZE - 1); - - userpage = kmap_atomic(page, KM_USER0); - memset(userpage + offset, 0, PAGE_CACHE_SIZE - offset); - flush_dcache_page(page); - kunmap_atomic(userpage, KM_USER0); - } - - set_page_extent_mapped(page); - - delalloc_start = start; - delalloc_end = 0; - while(delalloc_end < page_end) { - nr_delalloc = find_lock_delalloc_range(tree, &delalloc_start, - &delalloc_end, - 128 * 1024 * 1024); - if (nr_delalloc == 0) { - delalloc_start = delalloc_end + 1; - continue; - } - tree->ops->fill_delalloc(inode, delalloc_start, - delalloc_end); - clear_extent_bit(tree, delalloc_start, - delalloc_end, - EXTENT_LOCKED | EXTENT_DELALLOC, - 1, 0, GFP_NOFS); - delalloc_start = delalloc_end + 1; - } - lock_extent(tree, start, page_end, GFP_NOFS); - - end = page_end; - if (test_range_bit(tree, start, page_end, EXTENT_DELALLOC, 0)) { - printk("found delalloc bits after lock_extent\n"); - } - - if (last_byte <= start) { - clear_extent_dirty(tree, start, page_end, GFP_NOFS); - goto done; - } - - set_extent_uptodate(tree, start, page_end, GFP_NOFS); - blocksize = inode->i_sb->s_blocksize; - - while (cur <= end) { - if (cur >= last_byte) { - clear_extent_dirty(tree, cur, page_end, GFP_NOFS); - break; - } - em = epd->get_extent(inode, page, page_offset, cur, end, 1); - if (IS_ERR(em) || !em) { - SetPageError(page); - break; - } - - extent_offset = cur - em->start; - BUG_ON(em->end < cur); - BUG_ON(end < cur); - iosize = min(em->end - cur, end - cur) + 1; - iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1); - sector = (em->block_start + extent_offset) >> 9; - bdev = em->bdev; - block_start = em->block_start; - free_extent_map(em); - em = NULL; - - if (block_start == EXTENT_MAP_HOLE || - block_start == EXTENT_MAP_INLINE) { - clear_extent_dirty(tree, cur, - cur + iosize - 1, GFP_NOFS); - cur = cur + iosize; - page_offset += iosize; - continue; - } - - /* leave this out until we have a page_mkwrite call */ - if (0 && !test_range_bit(tree, cur, cur + iosize - 1, - EXTENT_DIRTY, 0)) { - cur = cur + iosize; - page_offset += iosize; - continue; - } - clear_extent_dirty(tree, cur, cur + iosize - 1, GFP_NOFS); - if (tree->ops && tree->ops->writepage_io_hook) { - ret = tree->ops->writepage_io_hook(page, cur, - cur + iosize - 1); - } else { - ret = 0; - } - if (ret) - SetPageError(page); - else { - unsigned long max_nr = end_index + 1; - set_range_writeback(tree, cur, cur + iosize - 1); - if (!PageWriteback(page)) { - printk("warning page %lu not writeback, " - "cur %llu end %llu\n", page->index, - (unsigned long long)cur, - (unsigned long long)end); - } - - ret = submit_extent_page(WRITE, tree, page, sector, - iosize, page_offset, bdev, - &epd->bio, max_nr, - end_bio_extent_writepage); - if (ret) - SetPageError(page); - } - cur = cur + iosize; - page_offset += iosize; - nr++; - } -done: - if (nr == 0) { - /* make sure the mapping tag for page dirty gets cleared */ - set_page_writeback(page); - end_page_writeback(page); - } - unlock_extent(tree, start, page_end, GFP_NOFS); - unlock_page(page); - return 0; -} - -#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18) - -/* Taken directly from 2.6.23 for 2.6.18 back port */ -typedef int (*writepage_t)(struct page *page, struct writeback_control *wbc, - void *data); - -/** - * write_cache_pages - walk the list of dirty pages of the given address space - * and write all of them. - * @mapping: address space structure to write - * @wbc: subtract the number of written pages from *@wbc->nr_to_write - * @writepage: function called for each page - * @data: data passed to writepage function - * - * If a page is already under I/O, write_cache_pages() skips it, even - * if it's dirty. This is desirable behaviour for memory-cleaning writeback, - * but it is INCORRECT for data-integrity system calls such as fsync(). fsync() - * and msync() need to guarantee that all the data which was dirty at the time - * the call was made get new I/O started against them. If wbc->sync_mode is - * WB_SYNC_ALL then we were called for data integrity and we must wait for - * existing IO to complete. - */ -static int write_cache_pages(struct address_space *mapping, - struct writeback_control *wbc, writepage_t writepage, - void *data) -{ - struct backing_dev_info *bdi = mapping->backing_dev_info; - int ret = 0; - int done = 0; - struct pagevec pvec; - int nr_pages; - pgoff_t index; - pgoff_t end; /* Inclusive */ - int scanned = 0; - int range_whole = 0; - - if (wbc->nonblocking && bdi_write_congested(bdi)) { - wbc->encountered_congestion = 1; - return 0; - } - - pagevec_init(&pvec, 0); - if (wbc->range_cyclic) { - index = mapping->writeback_index; /* Start from prev offset */ - end = -1; - } else { - index = wbc->range_start >> PAGE_CACHE_SHIFT; - end = wbc->range_end >> PAGE_CACHE_SHIFT; - if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX) - range_whole = 1; - scanned = 1; - } -retry: - while (!done && (index <= end) && - (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, - PAGECACHE_TAG_DIRTY, - min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) { - unsigned i; - - scanned = 1; - for (i = 0; i < nr_pages; i++) { - struct page *page = pvec.pages[i]; - - /* - * At this point we hold neither mapping->tree_lock nor - * lock on the page itself: the page may be truncated or - * invalidated (changing page->mapping to NULL), or even - * swizzled back from swapper_space to tmpfs file - * mapping - */ - lock_page(page); - - if (unlikely(page->mapping != mapping)) { - unlock_page(page); - continue; - } - - if (!wbc->range_cyclic && page->index > end) { - done = 1; - unlock_page(page); - continue; - } - - if (wbc->sync_mode != WB_SYNC_NONE) - wait_on_page_writeback(page); - - if (PageWriteback(page) || - !clear_page_dirty_for_io(page)) { - unlock_page(page); - continue; - } - - ret = (*writepage)(page, wbc, data); - - if (unlikely(ret == AOP_WRITEPAGE_ACTIVATE)) { - unlock_page(page); - ret = 0; - } - if (ret || (--(wbc->nr_to_write) <= 0)) - done = 1; - if (wbc->nonblocking && bdi_write_congested(bdi)) { - wbc->encountered_congestion = 1; - done = 1; - } - } - pagevec_release(&pvec); - cond_resched(); - } - if (!scanned && !done) { - /* - * We hit the last page and there is more work to be done: wrap - * back to the start of the file - */ - scanned = 1; - index = 0; - goto retry; - } - if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0)) - mapping->writeback_index = index; - return ret; -} -#endif - -int extent_write_full_page(struct extent_map_tree *tree, struct page *page, - get_extent_t *get_extent, - struct writeback_control *wbc) -{ - int ret; - struct address_space *mapping = page->mapping; - struct extent_page_data epd = { - .bio = NULL, - .tree = tree, - .get_extent = get_extent, - }; - struct writeback_control wbc_writepages = { - .bdi = wbc->bdi, - .sync_mode = WB_SYNC_NONE, - .older_than_this = NULL, - .nr_to_write = 64, - .range_start = page_offset(page) + PAGE_CACHE_SIZE, - .range_end = (loff_t)-1, - }; - - - ret = __extent_writepage(page, wbc, &epd); - - write_cache_pages(mapping, &wbc_writepages, __extent_writepage, &epd); - if (epd.bio) { - submit_one_bio(WRITE, epd.bio); - } - return ret; -} -EXPORT_SYMBOL(extent_write_full_page); - - -int extent_writepages(struct extent_map_tree *tree, - struct address_space *mapping, - get_extent_t *get_extent, - struct writeback_control *wbc) -{ - int ret = 0; - struct extent_page_data epd = { - .bio = NULL, - .tree = tree, - .get_extent = get_extent, - }; - - ret = write_cache_pages(mapping, wbc, __extent_writepage, &epd); - if (epd.bio) { - submit_one_bio(WRITE, epd.bio); - } - return ret; -} -EXPORT_SYMBOL(extent_writepages); - -int extent_readpages(struct extent_map_tree *tree, - struct address_space *mapping, - struct list_head *pages, unsigned nr_pages, - get_extent_t get_extent) -{ - struct bio *bio = NULL; - unsigned page_idx; - struct pagevec pvec; - - pagevec_init(&pvec, 0); - for (page_idx = 0; page_idx < nr_pages; page_idx++) { - struct page *page = list_entry(pages->prev, struct page, lru); - - prefetchw(&page->flags); - list_del(&page->lru); - /* - * what we want to do here is call add_to_page_cache_lru, - * but that isn't exported, so we reproduce it here - */ - if (!add_to_page_cache(page, mapping, - page->index, GFP_KERNEL)) { - - /* open coding of lru_cache_add, also not exported */ - page_cache_get(page); - if (!pagevec_add(&pvec, page)) - __pagevec_lru_add(&pvec); - __extent_read_full_page(tree, page, get_extent, &bio); - } - page_cache_release(page); - } - if (pagevec_count(&pvec)) - __pagevec_lru_add(&pvec); - BUG_ON(!list_empty(pages)); - if (bio) - submit_one_bio(READ, bio); - return 0; -} -EXPORT_SYMBOL(extent_readpages); - -/* - * basic invalidatepage code, this waits on any locked or writeback - * ranges corresponding to the page, and then deletes any extent state - * records from the tree - */ -int extent_invalidatepage(struct extent_map_tree *tree, - struct page *page, unsigned long offset) -{ - u64 start = ((u64)page->index << PAGE_CACHE_SHIFT); - u64 end = start + PAGE_CACHE_SIZE - 1; - size_t blocksize = page->mapping->host->i_sb->s_blocksize; - - start += (offset + blocksize -1) & ~(blocksize - 1); - if (start > end) - return 0; - - lock_extent(tree, start, end, GFP_NOFS); - wait_on_extent_writeback(tree, start, end); - clear_extent_bit(tree, start, end, - EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC, - 1, 1, GFP_NOFS); - return 0; -} -EXPORT_SYMBOL(extent_invalidatepage); - -/* - * simple commit_write call, set_range_dirty is used to mark both - * the pages and the extent records as dirty - */ -int extent_commit_write(struct extent_map_tree *tree, - struct inode *inode, struct page *page, - unsigned from, unsigned to) -{ - loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to; - - set_page_extent_mapped(page); - set_page_dirty(page); - - if (pos > inode->i_size) { - i_size_write(inode, pos); - mark_inode_dirty(inode); - } - return 0; -} -EXPORT_SYMBOL(extent_commit_write); - -int extent_prepare_write(struct extent_map_tree *tree, - struct inode *inode, struct page *page, - unsigned from, unsigned to, get_extent_t *get_extent) -{ - u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT; - u64 page_end = page_start + PAGE_CACHE_SIZE - 1; - u64 block_start; - u64 orig_block_start; - u64 block_end; - u64 cur_end; - struct extent_map *em; - unsigned blocksize = 1 << inode->i_blkbits; - size_t page_offset = 0; - size_t block_off_start; - size_t block_off_end; - int err = 0; - int iocount = 0; - int ret = 0; - int isnew; - - set_page_extent_mapped(page); - - block_start = (page_start + from) & ~((u64)blocksize - 1); - block_end = (page_start + to - 1) | (blocksize - 1); - orig_block_start = block_start; - - lock_extent(tree, page_start, page_end, GFP_NOFS); - while(block_start <= block_end) { - em = get_extent(inode, page, page_offset, block_start, - block_end, 1); - if (IS_ERR(em) || !em) { - goto err; - } - cur_end = min(block_end, em->end); - block_off_start = block_start & (PAGE_CACHE_SIZE - 1); - block_off_end = block_off_start + blocksize; - isnew = clear_extent_new(tree, block_start, cur_end, GFP_NOFS); - - if (!PageUptodate(page) && isnew && - (block_off_end > to || block_off_start < from)) { - void *kaddr; - - kaddr = kmap_atomic(page, KM_USER0); - if (block_off_end > to) - memset(kaddr + to, 0, block_off_end - to); - if (block_off_start < from) - memset(kaddr + block_off_start, 0, - from - block_off_start); - flush_dcache_page(page); - kunmap_atomic(kaddr, KM_USER0); - } - if ((em->block_start != EXTENT_MAP_HOLE && - em->block_start != EXTENT_MAP_INLINE) && - !isnew && !PageUptodate(page) && - (block_off_end > to || block_off_start < from) && - !test_range_bit(tree, block_start, cur_end, - EXTENT_UPTODATE, 1)) { - u64 sector; - u64 extent_offset = block_start - em->start; - size_t iosize; - sector = (em->block_start + extent_offset) >> 9; - iosize = (cur_end - block_start + blocksize) & - ~((u64)blocksize - 1); - /* - * we've already got the extent locked, but we - * need to split the state such that our end_bio - * handler can clear the lock. - */ - set_extent_bit(tree, block_start, - block_start + iosize - 1, - EXTENT_LOCKED, 0, NULL, GFP_NOFS); - ret = submit_extent_page(READ, tree, page, - sector, iosize, page_offset, em->bdev, - NULL, 1, - end_bio_extent_preparewrite); - iocount++; - block_start = block_start + iosize; - } else { - set_extent_uptodate(tree, block_start, cur_end, - GFP_NOFS); - unlock_extent(tree, block_start, cur_end, GFP_NOFS); - block_start = cur_end + 1; - } - page_offset = block_start & (PAGE_CACHE_SIZE - 1); - free_extent_map(em); - } - if (iocount) { - wait_extent_bit(tree, orig_block_start, - block_end, EXTENT_LOCKED); - } - check_page_uptodate(tree, page); -err: - /* FIXME, zero out newly allocated blocks on error */ - return err; -} -EXPORT_SYMBOL(extent_prepare_write); - -/* - * a helper for releasepage. As long as there are no locked extents - * in the range corresponding to the page, both state records and extent - * map records are removed - */ -int try_release_extent_mapping(struct extent_map_tree *tree, struct page *page) -{ - struct extent_map *em; - u64 start = (u64)page->index << PAGE_CACHE_SHIFT; - u64 end = start + PAGE_CACHE_SIZE - 1; - u64 orig_start = start; - int ret = 1; - - while (start <= end) { - em = lookup_extent_mapping(tree, start, end); - if (!em || IS_ERR(em)) - break; - if (!test_range_bit(tree, em->start, em->end, - EXTENT_LOCKED, 0)) { - remove_extent_mapping(tree, em); - /* once for the rb tree */ - free_extent_map(em); - } - start = em->end + 1; - /* once for us */ - free_extent_map(em); - } - if (test_range_bit(tree, orig_start, end, EXTENT_LOCKED, 0)) - ret = 0; - else - clear_extent_bit(tree, orig_start, end, EXTENT_UPTODATE, - 1, 1, GFP_NOFS); - return ret; -} -EXPORT_SYMBOL(try_release_extent_mapping); - -sector_t extent_bmap(struct address_space *mapping, sector_t iblock, - get_extent_t *get_extent) -{ - struct inode *inode = mapping->host; - u64 start = iblock << inode->i_blkbits; - u64 end = start + (1 << inode->i_blkbits) - 1; - sector_t sector = 0; - struct extent_map *em; - - em = get_extent(inode, NULL, 0, start, end, 0); - if (!em || IS_ERR(em)) - return 0; - - if (em->block_start == EXTENT_MAP_INLINE || - em->block_start == EXTENT_MAP_HOLE) - goto out; - - sector = (em->block_start + start - em->start) >> inode->i_blkbits; -out: - free_extent_map(em); - return sector; -} - -static int add_lru(struct extent_map_tree *tree, struct extent_buffer *eb) -{ - if (list_empty(&eb->lru)) { - extent_buffer_get(eb); - list_add(&eb->lru, &tree->buffer_lru); - tree->lru_size++; - if (tree->lru_size >= BUFFER_LRU_MAX) { - struct extent_buffer *rm; - rm = list_entry(tree->buffer_lru.prev, - struct extent_buffer, lru); - tree->lru_size--; - list_del_init(&rm->lru); - free_extent_buffer(rm); - } - } else - list_move(&eb->lru, &tree->buffer_lru); - return 0; -} -static struct extent_buffer *find_lru(struct extent_map_tree *tree, - u64 start, unsigned long len) -{ - struct list_head *lru = &tree->buffer_lru; - struct list_head *cur = lru->next; - struct extent_buffer *eb; - - if (list_empty(lru)) - return NULL; - - do { - eb = list_entry(cur, struct extent_buffer, lru); - if (eb->start == start && eb->len == len) { - extent_buffer_get(eb); - return eb; - } - cur = cur->next; - } while (cur != lru); - return NULL; -} - -static inline unsigned long num_extent_pages(u64 start, u64 len) -{ - return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) - - (start >> PAGE_CACHE_SHIFT); -} - -static inline struct page *extent_buffer_page(struct extent_buffer *eb, - unsigned long i) -{ - struct page *p; - struct address_space *mapping; - - if (i == 0) - return eb->first_page; - i += eb->start >> PAGE_CACHE_SHIFT; - mapping = eb->first_page->mapping; - read_lock_irq(&mapping->tree_lock); - p = radix_tree_lookup(&mapping->page_tree, i); - read_unlock_irq(&mapping->tree_lock); - return p; -} - -static struct extent_buffer *__alloc_extent_buffer(struct extent_map_tree *tree, - u64 start, - unsigned long len, - gfp_t mask) -{ - struct extent_buffer *eb = NULL; - - spin_lock(&tree->lru_lock); - eb = find_lru(tree, start, len); - spin_unlock(&tree->lru_lock); - if (eb) { - return eb; - } - - eb = kmem_cache_zalloc(extent_buffer_cache, mask); - INIT_LIST_HEAD(&eb->lru); - eb->start = start; - eb->len = len; - atomic_set(&eb->refs, 1); - - return eb; -} - -static void __free_extent_buffer(struct extent_buffer *eb) -{ - kmem_cache_free(extent_buffer_cache, eb); -} - -struct extent_buffer *alloc_extent_buffer(struct extent_map_tree *tree, - u64 start, unsigned long len, - struct page *page0, - gfp_t mask) -{ - unsigned long num_pages = num_extent_pages(start, len); - unsigned long i; - unsigned long index = start >> PAGE_CACHE_SHIFT; - struct extent_buffer *eb; - struct page *p; - struct address_space *mapping = tree->mapping; - int uptodate = 1; - - eb = __alloc_extent_buffer(tree, start, len, mask); - if (!eb || IS_ERR(eb)) - return NULL; - - if (eb->flags & EXTENT_BUFFER_FILLED) - goto lru_add; - - if (page0) { - eb->first_page = page0; - i = 1; - index++; - page_cache_get(page0); - mark_page_accessed(page0); - set_page_extent_mapped(page0); - WARN_ON(!PageUptodate(page0)); - set_page_extent_head(page0, len); - } else { - i = 0; - } - for (; i < num_pages; i++, index++) { - p = find_or_create_page(mapping, index, mask | __GFP_HIGHMEM); - if (!p) { - WARN_ON(1); - goto fail; - } - set_page_extent_mapped(p); - mark_page_accessed(p); - if (i == 0) { - eb->first_page = p; - set_page_extent_head(p, len); - } else { - set_page_private(p, EXTENT_PAGE_PRIVATE); - } - if (!PageUptodate(p)) - uptodate = 0; - unlock_page(p); - } - if (uptodate) - eb->flags |= EXTENT_UPTODATE; - eb->flags |= EXTENT_BUFFER_FILLED; - -lru_add: - spin_lock(&tree->lru_lock); - add_lru(tree, eb); - spin_unlock(&tree->lru_lock); - return eb; - -fail: - spin_lock(&tree->lru_lock); - list_del_init(&eb->lru); - spin_unlock(&tree->lru_lock); - if (!atomic_dec_and_test(&eb->refs)) - return NULL; - for (index = 1; index < i; index++) { - page_cache_release(extent_buffer_page(eb, index)); - } - if (i > 0) - page_cache_release(extent_buffer_page(eb, 0)); - __free_extent_buffer(eb); - return NULL; -} -EXPORT_SYMBOL(alloc_extent_buffer); - -struct extent_buffer *find_extent_buffer(struct extent_map_tree *tree, - u64 start, unsigned long len, - gfp_t mask) -{ - unsigned long num_pages = num_extent_pages(start, len); - unsigned long i; - unsigned long index = start >> PAGE_CACHE_SHIFT; - struct extent_buffer *eb; - struct page *p; - struct address_space *mapping = tree->mapping; - int uptodate = 1; - - eb = __alloc_extent_buffer(tree, start, len, mask); - if (!eb || IS_ERR(eb)) - return NULL; - - if (eb->flags & EXTENT_BUFFER_FILLED) - goto lru_add; - - for (i = 0; i < num_pages; i++, index++) { - p = find_lock_page(mapping, index); - if (!p) { - goto fail; - } - set_page_extent_mapped(p); - mark_page_accessed(p); - - if (i == 0) { - eb->first_page = p; - set_page_extent_head(p, len); - } else { - set_page_private(p, EXTENT_PAGE_PRIVATE); - } - - if (!PageUptodate(p)) - uptodate = 0; - unlock_page(p); - } - if (uptodate) - eb->flags |= EXTENT_UPTODATE; - eb->flags |= EXTENT_BUFFER_FILLED; - -lru_add: - spin_lock(&tree->lru_lock); - add_lru(tree, eb); - spin_unlock(&tree->lru_lock); - return eb; -fail: - spin_lock(&tree->lru_lock); - list_del_init(&eb->lru); - spin_unlock(&tree->lru_lock); - if (!atomic_dec_and_test(&eb->refs)) - return NULL; - for (index = 1; index < i; index++) { - page_cache_release(extent_buffer_page(eb, index)); - } - if (i > 0) - page_cache_release(extent_buffer_page(eb, 0)); - __free_extent_buffer(eb); - return NULL; -} -EXPORT_SYMBOL(find_extent_buffer); - -void free_extent_buffer(struct extent_buffer *eb) -{ - unsigned long i; - unsigned long num_pages; - - if (!eb) - return; - - if (!atomic_dec_and_test(&eb->refs)) - return; - - WARN_ON(!list_empty(&eb->lru)); - num_pages = num_extent_pages(eb->start, eb->len); - - for (i = 1; i < num_pages; i++) { - page_cache_release(extent_buffer_page(eb, i)); - } - page_cache_release(extent_buffer_page(eb, 0)); - __free_extent_buffer(eb); -} -EXPORT_SYMBOL(free_extent_buffer); - -int clear_extent_buffer_dirty(struct extent_map_tree *tree, - struct extent_buffer *eb) -{ - int set; - unsigned long i; - unsigned long num_pages; - struct page *page; - - u64 start = eb->start; - u64 end = start + eb->len - 1; - - set = clear_extent_dirty(tree, start, end, GFP_NOFS); - num_pages = num_extent_pages(eb->start, eb->len); - - for (i = 0; i < num_pages; i++) { - page = extent_buffer_page(eb, i); - lock_page(page); - if (i == 0) - set_page_extent_head(page, eb->len); - else - set_page_private(page, EXTENT_PAGE_PRIVATE); - - /* - * if we're on the last page or the first page and the - * block isn't aligned on a page boundary, do extra checks - * to make sure we don't clean page that is partially dirty - */ - if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) || - ((i == num_pages - 1) && - ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) { - start = (u64)page->index << PAGE_CACHE_SHIFT; - end = start + PAGE_CACHE_SIZE - 1; - if (test_range_bit(tree, start, end, - EXTENT_DIRTY, 0)) { - unlock_page(page); - continue; - } - } - clear_page_dirty_for_io(page); - write_lock_irq(&page->mapping->tree_lock); - if (!PageDirty(page)) { - radix_tree_tag_clear(&page->mapping->page_tree, - page_index(page), - PAGECACHE_TAG_DIRTY); - } - write_unlock_irq(&page->mapping->tree_lock); - unlock_page(page); - } - return 0; -} -EXPORT_SYMBOL(clear_extent_buffer_dirty); - -int wait_on_extent_buffer_writeback(struct extent_map_tree *tree, - struct extent_buffer *eb) -{ - return wait_on_extent_writeback(tree, eb->start, - eb->start + eb->len - 1); -} -EXPORT_SYMBOL(wait_on_extent_buffer_writeback); - -int set_extent_buffer_dirty(struct extent_map_tree *tree, - struct extent_buffer *eb) -{ - unsigned long i; - unsigned long num_pages; - - num_pages = num_extent_pages(eb->start, eb->len); - for (i = 0; i < num_pages; i++) { - struct page *page = extent_buffer_page(eb, i); - /* writepage may need to do something special for the - * first page, we have to make sure page->private is - * properly set. releasepage may drop page->private - * on us if the page isn't already dirty. - */ - if (i == 0) { - lock_page(page); - set_page_extent_head(page, eb->len); - } else if (PagePrivate(page) && - page->private != EXTENT_PAGE_PRIVATE) { - lock_page(page); - set_page_extent_mapped(page); - unlock_page(page); - } - __set_page_dirty_nobuffers(extent_buffer_page(eb, i)); - if (i == 0) - unlock_page(page); - } - return set_extent_dirty(tree, eb->start, - eb->start + eb->len - 1, GFP_NOFS); -} -EXPORT_SYMBOL(set_extent_buffer_dirty); - -int set_extent_buffer_uptodate(struct extent_map_tree *tree, - struct extent_buffer *eb) -{ - unsigned long i; - struct page *page; - unsigned long num_pages; - - num_pages = num_extent_pages(eb->start, eb->len); - - set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1, - GFP_NOFS); - for (i = 0; i < num_pages; i++) { - page = extent_buffer_page(eb, i); - if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) || - ((i == num_pages - 1) && - ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) { - check_page_uptodate(tree, page); - continue; - } - SetPageUptodate(page); - } - return 0; -} -EXPORT_SYMBOL(set_extent_buffer_uptodate); - -int extent_buffer_uptodate(struct extent_map_tree *tree, - struct extent_buffer *eb) -{ - if (eb->flags & EXTENT_UPTODATE) - return 1; - return test_range_bit(tree, eb->start, eb->start + eb->len - 1, - EXTENT_UPTODATE, 1); -} -EXPORT_SYMBOL(extent_buffer_uptodate); - -int read_extent_buffer_pages(struct extent_map_tree *tree, - struct extent_buffer *eb, - u64 start, - int wait) -{ - unsigned long i; - unsigned long start_i; - struct page *page; - int err; - int ret = 0; - unsigned long num_pages; - - if (eb->flags & EXTENT_UPTODATE) - return 0; - - if (0 && test_range_bit(tree, eb->start, eb->start + eb->len - 1, - EXTENT_UPTODATE, 1)) { - return 0; - } - - if (start) { - WARN_ON(start < eb->start); - start_i = (start >> PAGE_CACHE_SHIFT) - - (eb->start >> PAGE_CACHE_SHIFT); - } else { - start_i = 0; - } - - num_pages = num_extent_pages(eb->start, eb->len); - for (i = start_i; i < num_pages; i++) { - page = extent_buffer_page(eb, i); - if (PageUptodate(page)) { - continue; - } - if (!wait) { - if (TestSetPageLocked(page)) { - continue; - } - } else { - lock_page(page); - } - if (!PageUptodate(page)) { - err = page->mapping->a_ops->readpage(NULL, page); - if (err) { - ret = err; - } - } else { - unlock_page(page); - } - } - - if (ret || !wait) { - return ret; - } - - for (i = start_i; i < num_pages; i++) { - page = extent_buffer_page(eb, i); - wait_on_page_locked(page); - if (!PageUptodate(page)) { - ret = -EIO; - } - } - if (!ret) - eb->flags |= EXTENT_UPTODATE; - return ret; -} -EXPORT_SYMBOL(read_extent_buffer_pages); - -void read_extent_buffer(struct extent_buffer *eb, void *dstv, - unsigned long start, - unsigned long len) -{ - size_t cur; - size_t offset; - struct page *page; - char *kaddr; - char *dst = (char *)dstv; - size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); - unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; - unsigned long num_pages = num_extent_pages(eb->start, eb->len); - - WARN_ON(start > eb->len); - WARN_ON(start + len > eb->start + eb->len); - - offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); - - while(len > 0) { - page = extent_buffer_page(eb, i); - if (!PageUptodate(page)) { - printk("page %lu not up to date i %lu, total %lu, len %lu\n", page->index, i, num_pages, eb->len); - WARN_ON(1); - } - WARN_ON(!PageUptodate(page)); - - cur = min(len, (PAGE_CACHE_SIZE - offset)); - kaddr = kmap_atomic(page, KM_USER1); - memcpy(dst, kaddr + offset, cur); - kunmap_atomic(kaddr, KM_USER1); - - dst += cur; - len -= cur; - offset = 0; - i++; - } -} -EXPORT_SYMBOL(read_extent_buffer); - -int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start, - unsigned long min_len, char **token, char **map, - unsigned long *map_start, - unsigned long *map_len, int km) -{ - size_t offset = start & (PAGE_CACHE_SIZE - 1); - char *kaddr; - struct page *p; - size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); - unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; - unsigned long end_i = (start_offset + start + min_len - 1) >> - PAGE_CACHE_SHIFT; - - if (i != end_i) - return -EINVAL; - - if (i == 0) { - offset = start_offset; - *map_start = 0; - } else { - offset = 0; - *map_start = ((u64)i << PAGE_CACHE_SHIFT) - start_offset; - } - if (start + min_len > eb->len) { -printk("bad mapping eb start %Lu len %lu, wanted %lu %lu\n", eb->start, eb->len, start, min_len); - WARN_ON(1); - } - - p = extent_buffer_page(eb, i); - WARN_ON(!PageUptodate(p)); - kaddr = kmap_atomic(p, km); - *token = kaddr; - *map = kaddr + offset; - *map_len = PAGE_CACHE_SIZE - offset; - return 0; -} -EXPORT_SYMBOL(map_private_extent_buffer); - -int map_extent_buffer(struct extent_buffer *eb, unsigned long start, - unsigned long min_len, - char **token, char **map, - unsigned long *map_start, - unsigned long *map_len, int km) -{ - int err; - int save = 0; - if (eb->map_token) { - unmap_extent_buffer(eb, eb->map_token, km); - eb->map_token = NULL; - save = 1; - } - err = map_private_extent_buffer(eb, start, min_len, token, map, - map_start, map_len, km); - if (!err && save) { - eb->map_token = *token; - eb->kaddr = *map; - eb->map_start = *map_start; - eb->map_len = *map_len; - } - return err; -} -EXPORT_SYMBOL(map_extent_buffer); - -void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km) -{ - kunmap_atomic(token, km); -} -EXPORT_SYMBOL(unmap_extent_buffer); - -int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv, - unsigned long start, - unsigned long len) -{ - size_t cur; - size_t offset; - struct page *page; - char *kaddr; - char *ptr = (char *)ptrv; - size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); - unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; - int ret = 0; - - WARN_ON(start > eb->len); - WARN_ON(start + len > eb->start + eb->len); - - offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); - - while(len > 0) { - page = extent_buffer_page(eb, i); - WARN_ON(!PageUptodate(page)); - - cur = min(len, (PAGE_CACHE_SIZE - offset)); - - kaddr = kmap_atomic(page, KM_USER0); - ret = memcmp(ptr, kaddr + offset, cur); - kunmap_atomic(kaddr, KM_USER0); - if (ret) - break; - - ptr += cur; - len -= cur; - offset = 0; - i++; - } - return ret; -} -EXPORT_SYMBOL(memcmp_extent_buffer); - -void write_extent_buffer(struct extent_buffer *eb, const void *srcv, - unsigned long start, unsigned long len) -{ - size_t cur; - size_t offset; - struct page *page; - char *kaddr; - char *src = (char *)srcv; - size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); - unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; - - WARN_ON(start > eb->len); - WARN_ON(start + len > eb->start + eb->len); - - offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); - - while(len > 0) { - page = extent_buffer_page(eb, i); - WARN_ON(!PageUptodate(page)); - - cur = min(len, PAGE_CACHE_SIZE - offset); - kaddr = kmap_atomic(page, KM_USER1); - memcpy(kaddr + offset, src, cur); - kunmap_atomic(kaddr, KM_USER1); - - src += cur; - len -= cur; - offset = 0; - i++; - } -} -EXPORT_SYMBOL(write_extent_buffer); - -void memset_extent_buffer(struct extent_buffer *eb, char c, - unsigned long start, unsigned long len) -{ - size_t cur; - size_t offset; - struct page *page; - char *kaddr; - size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); - unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; - - WARN_ON(start > eb->len); - WARN_ON(start + len > eb->start + eb->len); - - offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); - - while(len > 0) { - page = extent_buffer_page(eb, i); - WARN_ON(!PageUptodate(page)); - - cur = min(len, PAGE_CACHE_SIZE - offset); - kaddr = kmap_atomic(page, KM_USER0); - memset(kaddr + offset, c, cur); - kunmap_atomic(kaddr, KM_USER0); - - len -= cur; - offset = 0; - i++; - } -} -EXPORT_SYMBOL(memset_extent_buffer); - -void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src, - unsigned long dst_offset, unsigned long src_offset, - unsigned long len) -{ - u64 dst_len = dst->len; - size_t cur; - size_t offset; - struct page *page; - char *kaddr; - size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); - unsigned long i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT; - - WARN_ON(src->len != dst_len); - - offset = (start_offset + dst_offset) & - ((unsigned long)PAGE_CACHE_SIZE - 1); - - while(len > 0) { - page = extent_buffer_page(dst, i); - WARN_ON(!PageUptodate(page)); - - cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - offset)); - - kaddr = kmap_atomic(page, KM_USER0); - read_extent_buffer(src, kaddr + offset, src_offset, cur); - kunmap_atomic(kaddr, KM_USER0); - - src_offset += cur; - len -= cur; - offset = 0; - i++; - } -} -EXPORT_SYMBOL(copy_extent_buffer); - -static void move_pages(struct page *dst_page, struct page *src_page, - unsigned long dst_off, unsigned long src_off, - unsigned long len) -{ - char *dst_kaddr = kmap_atomic(dst_page, KM_USER0); - if (dst_page == src_page) { - memmove(dst_kaddr + dst_off, dst_kaddr + src_off, len); - } else { - char *src_kaddr = kmap_atomic(src_page, KM_USER1); - char *p = dst_kaddr + dst_off + len; - char *s = src_kaddr + src_off + len; - - while (len--) - *--p = *--s; - - kunmap_atomic(src_kaddr, KM_USER1); - } - kunmap_atomic(dst_kaddr, KM_USER0); -} - -static void copy_pages(struct page *dst_page, struct page *src_page, - unsigned long dst_off, unsigned long src_off, - unsigned long len) -{ - char *dst_kaddr = kmap_atomic(dst_page, KM_USER0); - char *src_kaddr; - - if (dst_page != src_page) - src_kaddr = kmap_atomic(src_page, KM_USER1); - else - src_kaddr = dst_kaddr; - - memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len); - kunmap_atomic(dst_kaddr, KM_USER0); - if (dst_page != src_page) - kunmap_atomic(src_kaddr, KM_USER1); -} - -void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, - unsigned long src_offset, unsigned long len) -{ - size_t cur; - size_t dst_off_in_page; - size_t src_off_in_page; - size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); - unsigned long dst_i; - unsigned long src_i; - - if (src_offset + len > dst->len) { - printk("memmove bogus src_offset %lu move len %lu len %lu\n", - src_offset, len, dst->len); - BUG_ON(1); - } - if (dst_offset + len > dst->len) { - printk("memmove bogus dst_offset %lu move len %lu len %lu\n", - dst_offset, len, dst->len); - BUG_ON(1); - } - - while(len > 0) { - dst_off_in_page = (start_offset + dst_offset) & - ((unsigned long)PAGE_CACHE_SIZE - 1); - src_off_in_page = (start_offset + src_offset) & - ((unsigned long)PAGE_CACHE_SIZE - 1); - - dst_i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT; - src_i = (start_offset + src_offset) >> PAGE_CACHE_SHIFT; - - cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - - src_off_in_page)); - cur = min_t(unsigned long, cur, - (unsigned long)(PAGE_CACHE_SIZE - dst_off_in_page)); - - copy_pages(extent_buffer_page(dst, dst_i), - extent_buffer_page(dst, src_i), - dst_off_in_page, src_off_in_page, cur); - - src_offset += cur; - dst_offset += cur; - len -= cur; - } -} -EXPORT_SYMBOL(memcpy_extent_buffer); - -void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, - unsigned long src_offset, unsigned long len) -{ - size_t cur; - size_t dst_off_in_page; - size_t src_off_in_page; - unsigned long dst_end = dst_offset + len - 1; - unsigned long src_end = src_offset + len - 1; - size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); - unsigned long dst_i; - unsigned long src_i; - - if (src_offset + len > dst->len) { - printk("memmove bogus src_offset %lu move len %lu len %lu\n", - src_offset, len, dst->len); - BUG_ON(1); - } - if (dst_offset + len > dst->len) { - printk("memmove bogus dst_offset %lu move len %lu len %lu\n", - dst_offset, len, dst->len); - BUG_ON(1); - } - if (dst_offset < src_offset) { - memcpy_extent_buffer(dst, dst_offset, src_offset, len); - return; - } - while(len > 0) { - dst_i = (start_offset + dst_end) >> PAGE_CACHE_SHIFT; - src_i = (start_offset + src_end) >> PAGE_CACHE_SHIFT; - - dst_off_in_page = (start_offset + dst_end) & - ((unsigned long)PAGE_CACHE_SIZE - 1); - src_off_in_page = (start_offset + src_end) & - ((unsigned long)PAGE_CACHE_SIZE - 1); - - cur = min_t(unsigned long, len, src_off_in_page + 1); - cur = min(cur, dst_off_in_page + 1); - move_pages(extent_buffer_page(dst, dst_i), - extent_buffer_page(dst, src_i), - dst_off_in_page - cur + 1, - src_off_in_page - cur + 1, cur); - - dst_end -= cur; - src_end -= cur; - len -= cur; - } -} -EXPORT_SYMBOL(memmove_extent_buffer); diff --git a/fs/btrfs/extent_map.h b/fs/btrfs/extent_map.h index ea60f5447b5b..56314217cfc0 100644 --- a/fs/btrfs/extent_map.h +++ b/fs/btrfs/extent_map.h @@ -3,215 +3,53 @@ #include +#define EXTENT_MAP_LAST_BYTE (u64)-4 #define EXTENT_MAP_HOLE (u64)-3 #define EXTENT_MAP_INLINE (u64)-2 #define EXTENT_MAP_DELALLOC (u64)-1 -/* bits for the extent state */ -#define EXTENT_DIRTY 1 -#define EXTENT_WRITEBACK (1 << 1) -#define EXTENT_UPTODATE (1 << 2) -#define EXTENT_LOCKED (1 << 3) -#define EXTENT_NEW (1 << 4) -#define EXTENT_DELALLOC (1 << 5) -#define EXTENT_DEFRAG (1 << 6) -#define EXTENT_DEFRAG_DONE (1 << 7) -#define EXTENT_BUFFER_FILLED (1 << 8) -#define EXTENT_CSUM (1 << 9) -#define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK) - -/* - * page->private values. Every page that is controlled by the extent - * map has page->private set to one. - */ -#define EXTENT_PAGE_PRIVATE 1 -#define EXTENT_PAGE_PRIVATE_FIRST_PAGE 3 - - -struct extent_map_ops { - int (*fill_delalloc)(struct inode *inode, u64 start, u64 end); - int (*writepage_io_hook)(struct page *page, u64 start, u64 end); - int (*readpage_io_hook)(struct page *page, u64 start, u64 end); - int (*readpage_end_io_hook)(struct page *page, u64 start, u64 end); - void (*writepage_end_io_hook)(struct page *page, u64 start, u64 end); -}; - -struct extent_map_tree { - struct rb_root map; - struct rb_root state; - struct address_space *mapping; - u64 dirty_bytes; - rwlock_t lock; - struct extent_map_ops *ops; - spinlock_t lru_lock; - struct list_head buffer_lru; - int lru_size; -}; - -/* note, this must start with the same fields as fs/extent_map.c:tree_entry */ struct extent_map { - u64 start; - u64 end; /* inclusive */ - int in_tree; struct rb_node rb_node; - /* block_start and block_end are in bytes */ + + /* all of these are in bytes */ + u64 start; + u64 len; u64 block_start; - u64 block_end; /* inclusive */ + unsigned long flags; struct block_device *bdev; atomic_t refs; -}; - -/* note, this must start with the same fields as fs/extent_map.c:tree_entry */ -struct extent_state { - u64 start; - u64 end; /* inclusive */ int in_tree; - struct rb_node rb_node; - wait_queue_head_t wq; - atomic_t refs; - unsigned long state; - - /* for use by the FS */ - u64 private; - - struct list_head list; }; -struct extent_buffer { - u64 start; - unsigned long len; - char *map_token; - char *kaddr; - unsigned long map_start; - unsigned long map_len; - struct page *first_page; - struct list_head lru; - atomic_t refs; - int flags; +struct extent_map_tree { + struct rb_root map; + struct extent_map *last; + spinlock_t lock; }; -typedef struct extent_map *(get_extent_t)(struct inode *inode, - struct page *page, - size_t page_offset, - u64 start, u64 end, - int create); +static inline u64 extent_map_end(struct extent_map *em) +{ + if (em->start + em->len < em->start) + return (u64)-1; + return em->start + em->len; +} + +static inline u64 extent_map_block_end(struct extent_map *em) +{ + if (em->block_start + em->len < em->block_start) + return (u64)-1; + return em->block_start + em->len; +} -void extent_map_tree_init(struct extent_map_tree *tree, - struct address_space *mapping, gfp_t mask); -void extent_map_tree_empty_lru(struct extent_map_tree *tree); +void extent_map_tree_init(struct extent_map_tree *tree, gfp_t mask); struct extent_map *lookup_extent_mapping(struct extent_map_tree *tree, - u64 start, u64 end); + u64 start, u64 len); int add_extent_mapping(struct extent_map_tree *tree, struct extent_map *em); int remove_extent_mapping(struct extent_map_tree *tree, struct extent_map *em); -int try_release_extent_mapping(struct extent_map_tree *tree, struct page *page); -int lock_extent(struct extent_map_tree *tree, u64 start, u64 end, gfp_t mask); -int unlock_extent(struct extent_map_tree *tree, u64 start, u64 end, gfp_t mask); + struct extent_map *alloc_extent_map(gfp_t mask); void free_extent_map(struct extent_map *em); -int extent_read_full_page(struct extent_map_tree *tree, struct page *page, - get_extent_t *get_extent); int __init extent_map_init(void); void extent_map_exit(void); - -u64 count_range_bits(struct extent_map_tree *tree, - u64 *start, u64 search_end, - u64 max_bytes, unsigned long bits); - -int test_range_bit(struct extent_map_tree *tree, u64 start, u64 end, - int bits, int filled); -int clear_extent_bits(struct extent_map_tree *tree, u64 start, u64 end, - int bits, gfp_t mask); -int set_extent_bits(struct extent_map_tree *tree, u64 start, u64 end, - int bits, gfp_t mask); -int set_extent_uptodate(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask); -int set_extent_new(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask); -int set_extent_dirty(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask); -int clear_extent_dirty(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask); -int set_extent_delalloc(struct extent_map_tree *tree, u64 start, u64 end, - gfp_t mask); -int find_first_extent_bit(struct extent_map_tree *tree, u64 start, - u64 *start_ret, u64 *end_ret, int bits); -int extent_invalidatepage(struct extent_map_tree *tree, - struct page *page, unsigned long offset); -int extent_write_full_page(struct extent_map_tree *tree, struct page *page, - get_extent_t *get_extent, - struct writeback_control *wbc); -int extent_writepages(struct extent_map_tree *tree, - struct address_space *mapping, - get_extent_t *get_extent, - struct writeback_control *wbc); -int extent_readpages(struct extent_map_tree *tree, - struct address_space *mapping, - struct list_head *pages, unsigned nr_pages, - get_extent_t get_extent); -int extent_prepare_write(struct extent_map_tree *tree, - struct inode *inode, struct page *page, - unsigned from, unsigned to, get_extent_t *get_extent); -int extent_commit_write(struct extent_map_tree *tree, - struct inode *inode, struct page *page, - unsigned from, unsigned to); -sector_t extent_bmap(struct address_space *mapping, sector_t iblock, - get_extent_t *get_extent); -int set_range_dirty(struct extent_map_tree *tree, u64 start, u64 end); -int set_state_private(struct extent_map_tree *tree, u64 start, u64 private); -int get_state_private(struct extent_map_tree *tree, u64 start, u64 *private); -void set_page_extent_mapped(struct page *page); - -struct extent_buffer *alloc_extent_buffer(struct extent_map_tree *tree, - u64 start, unsigned long len, - struct page *page0, - gfp_t mask); -struct extent_buffer *find_extent_buffer(struct extent_map_tree *tree, - u64 start, unsigned long len, - gfp_t mask); -void free_extent_buffer(struct extent_buffer *eb); -int read_extent_buffer_pages(struct extent_map_tree *tree, - struct extent_buffer *eb, u64 start, int wait); - -static inline void extent_buffer_get(struct extent_buffer *eb) -{ - atomic_inc(&eb->refs); -} - -int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv, - unsigned long start, - unsigned long len); -void read_extent_buffer(struct extent_buffer *eb, void *dst, - unsigned long start, - unsigned long len); -void write_extent_buffer(struct extent_buffer *eb, const void *src, - unsigned long start, unsigned long len); -void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src, - unsigned long dst_offset, unsigned long src_offset, - unsigned long len); -void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, - unsigned long src_offset, unsigned long len); -void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, - unsigned long src_offset, unsigned long len); -void memset_extent_buffer(struct extent_buffer *eb, char c, - unsigned long start, unsigned long len); -int wait_on_extent_buffer_writeback(struct extent_map_tree *tree, - struct extent_buffer *eb); -int clear_extent_buffer_dirty(struct extent_map_tree *tree, - struct extent_buffer *eb); -int set_extent_buffer_dirty(struct extent_map_tree *tree, - struct extent_buffer *eb); -int set_extent_buffer_uptodate(struct extent_map_tree *tree, - struct extent_buffer *eb); -int extent_buffer_uptodate(struct extent_map_tree *tree, - struct extent_buffer *eb); -int map_extent_buffer(struct extent_buffer *eb, unsigned long offset, - unsigned long min_len, char **token, char **map, - unsigned long *map_start, - unsigned long *map_len, int km); -int map_private_extent_buffer(struct extent_buffer *eb, unsigned long offset, - unsigned long min_len, char **token, char **map, - unsigned long *map_start, - unsigned long *map_len, int km); -void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km); #endif diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 1cd8c908811e..c5bb00f92396 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -233,8 +233,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans, int err = 0; int i; struct inode *inode = fdentry(file)->d_inode; - struct extent_map *em; - struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; u64 hint_byte; u64 num_bytes; u64 start_pos; @@ -242,11 +241,6 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans, u64 end_pos = pos + write_bytes; u64 inline_size; loff_t isize = i_size_read(inode); - em = alloc_extent_map(GFP_NOFS); - if (!em) - return -ENOMEM; - - em->bdev = inode->i_sb->s_bdev; start_pos = pos & ~((u64)root->sectorsize - 1); num_bytes = (write_bytes + pos - start_pos + @@ -254,7 +248,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans, end_of_last_block = start_pos + num_bytes - 1; - lock_extent(em_tree, start_pos, end_of_last_block, GFP_NOFS); + lock_extent(io_tree, start_pos, end_of_last_block, GFP_NOFS); mutex_lock(&root->fs_info->fs_mutex); trans = btrfs_start_transaction(root, 1); if (!trans) { @@ -268,7 +262,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans, if ((end_of_last_block & 4095) == 0) { printk("strange end of last %Lu %zu %Lu\n", start_pos, write_bytes, end_of_last_block); } - set_extent_uptodate(em_tree, start_pos, end_of_last_block, GFP_NOFS); + set_extent_uptodate(io_tree, start_pos, end_of_last_block, GFP_NOFS); /* FIXME...EIEIO, ENOSPC and more */ @@ -293,6 +287,8 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans, inode->i_ino, last_pos_in_file, 0, 0, hole_size); + btrfs_drop_extent_cache(inode, last_pos_in_file, + last_pos_in_file + hole_size -1); btrfs_check_file(root, inode); } if (err) @@ -320,12 +316,12 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans, last_end += PAGE_CACHE_SIZE - 1; if (start_pos < isize) { u64 delalloc_start = start_pos; - existing_delalloc = count_range_bits(em_tree, + existing_delalloc = count_range_bits(io_tree, &delalloc_start, end_of_last_block, (u64)-1, EXTENT_DELALLOC); } - set_extent_delalloc(em_tree, start_pos, end_of_last_block, + set_extent_delalloc(io_tree, start_pos, end_of_last_block, GFP_NOFS); spin_lock(&root->fs_info->delalloc_lock); root->fs_info->delalloc_bytes += (end_of_last_block + 1 - @@ -346,6 +342,7 @@ static int noinline dirty_and_release_pages(struct btrfs_trans_handle *trans, inline_size -= start_pos; err = insert_inline_extent(trans, root, inode, start_pos, inline_size, pages, 0, num_pages); + btrfs_drop_extent_cache(inode, start_pos, aligned_end - 1); BUG_ON(err); } if (end_pos > isize) { @@ -356,8 +353,7 @@ failed: err = btrfs_end_transaction(trans, root); out_unlock: mutex_unlock(&root->fs_info->fs_mutex); - unlock_extent(em_tree, start_pos, end_of_last_block, GFP_NOFS); - free_extent_map(em); + unlock_extent(io_tree, start_pos, end_of_last_block, GFP_NOFS); return err; } @@ -367,10 +363,15 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end) struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; while(1) { + spin_lock(&em_tree->lock); em = lookup_extent_mapping(em_tree, start, end); - if (!em) + if (!em) { + spin_unlock(&em_tree->lock); break; + } remove_extent_mapping(em_tree, em); + spin_unlock(&em_tree->lock); + /* once for us */ free_extent_map(em); /* once for the tree*/ diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 67005480e139..16d3aef45d18 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -53,7 +53,7 @@ static struct inode_operations btrfs_file_inode_operations; static struct address_space_operations btrfs_aops; static struct address_space_operations btrfs_symlink_aops; static struct file_operations btrfs_dir_file_operations; -static struct extent_map_ops btrfs_extent_map_ops; +static struct extent_io_ops btrfs_extent_io_ops; static struct kmem_cache *btrfs_inode_cachep; struct kmem_cache *btrfs_trans_handle_cachep; @@ -104,6 +104,8 @@ static int cow_file_range(struct inode *inode, u64 start, u64 end) u64 num_bytes; u64 cur_alloc_size; u64 blocksize = root->sectorsize; + u64 orig_start = start; + u64 orig_num_bytes; struct btrfs_key ins; int ret; @@ -115,6 +117,7 @@ static int cow_file_range(struct inode *inode, u64 start, u64 end) num_bytes = max(blocksize, num_bytes); ret = btrfs_drop_extents(trans, root, inode, start, start + num_bytes, start, &alloc_hint); + orig_num_bytes = num_bytes; if (alloc_hint == EXTENT_MAP_INLINE) goto out; @@ -138,6 +141,8 @@ static int cow_file_range(struct inode *inode, u64 start, u64 end) alloc_hint = ins.objectid + ins.offset; start += cur_alloc_size; } + btrfs_drop_extent_cache(inode, orig_start, + orig_start + orig_num_bytes - 1); btrfs_add_ordered_inode(inode); out: btrfs_end_transaction(trans, root); @@ -297,7 +302,7 @@ 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_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_csum_item *item; struct btrfs_path *path = NULL; u32 csum; @@ -317,7 +322,7 @@ int btrfs_readpage_io_hook(struct page *page, u64 start, u64 end) } read_extent_buffer(path->nodes[0], &csum, (unsigned long)item, BTRFS_CRC32_SIZE); - set_state_private(em_tree, start, csum); + set_state_private(io_tree, start, csum); out: if (path) btrfs_free_path(path); @@ -329,17 +334,19 @@ int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end) { size_t offset = start - ((u64)page->index << PAGE_CACHE_SHIFT); struct inode *inode = page->mapping->host; - struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; char *kaddr; u64 private; 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)) return 0; - ret = get_state_private(em_tree, start, &private); + + ret = get_state_private(io_tree, start, &private); local_irq_save(flags); kaddr = kmap_atomic(page, KM_IRQ0); if (ret) { @@ -428,7 +435,7 @@ void btrfs_read_locked_inode(struct inode *inode) switch (inode->i_mode & S_IFMT) { case S_IFREG: inode->i_mapping->a_ops = &btrfs_aops; - BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops; + BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; inode->i_fop = &btrfs_file_operations; inode->i_op = &btrfs_file_inode_operations; break; @@ -873,7 +880,7 @@ static int btrfs_cow_one_page(struct inode *inode, struct page *page, size_t zero_start) { char *kaddr; - struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_root *root = BTRFS_I(inode)->root; u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT; u64 page_end = page_start + PAGE_CACHE_SIZE - 1; @@ -884,12 +891,12 @@ static int btrfs_cow_one_page(struct inode *inode, struct page *page, WARN_ON(!PageLocked(page)); set_page_extent_mapped(page); - lock_extent(em_tree, page_start, page_end, GFP_NOFS); + lock_extent(io_tree, page_start, page_end, GFP_NOFS); delalloc_start = page_start; - existing_delalloc = count_range_bits(&BTRFS_I(inode)->extent_tree, + existing_delalloc = count_range_bits(&BTRFS_I(inode)->io_tree, &delalloc_start, page_end, PAGE_CACHE_SIZE, EXTENT_DELALLOC); - set_extent_delalloc(&BTRFS_I(inode)->extent_tree, page_start, + set_extent_delalloc(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS); spin_lock(&root->fs_info->delalloc_lock); @@ -903,7 +910,7 @@ static int btrfs_cow_one_page(struct inode *inode, struct page *page, kunmap(page); } set_page_dirty(page); - unlock_extent(em_tree, page_start, page_end, GFP_NOFS); + unlock_extent(io_tree, page_start, page_end, GFP_NOFS); return ret; } @@ -961,7 +968,7 @@ static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) 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_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; u64 mask = root->sectorsize - 1; u64 pos = (inode->i_size + mask) & ~mask; @@ -986,7 +993,7 @@ static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) btrfs_truncate_page(inode->i_mapping, inode->i_size); - lock_extent(em_tree, pos, block_end, GFP_NOFS); + lock_extent(io_tree, pos, block_end, GFP_NOFS); hole_size = block_end - hole_start; mutex_lock(&root->fs_info->fs_mutex); @@ -1001,11 +1008,13 @@ static int btrfs_setattr(struct dentry *dentry, struct iattr *attr) inode->i_ino, hole_start, 0, 0, hole_size); + btrfs_drop_extent_cache(inode, hole_start, + hole_size - 1); btrfs_check_file(root, inode); } btrfs_end_transaction(trans, root); mutex_unlock(&root->fs_info->fs_mutex); - unlock_extent(em_tree, pos, block_end, GFP_NOFS); + unlock_extent(io_tree, pos, block_end, GFP_NOFS); if (err) return err; } @@ -1189,7 +1198,8 @@ static int btrfs_init_locked_inode(struct inode *inode, void *p) struct btrfs_iget_args *args = p; inode->i_ino = args->ino; BTRFS_I(inode)->root = args->root; - extent_map_tree_init(&BTRFS_I(inode)->extent_tree, + 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); return 0; } @@ -1485,7 +1495,8 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, if (!inode) return ERR_PTR(-ENOMEM); - extent_map_tree_init(&BTRFS_I(inode)->extent_tree, + 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); BTRFS_I(inode)->root = root; @@ -1672,9 +1683,10 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry, inode->i_mapping->a_ops = &btrfs_aops; inode->i_fop = &btrfs_file_operations; inode->i_op = &btrfs_file_inode_operations; - extent_map_tree_init(&BTRFS_I(inode)->extent_tree, + 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); - BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops; + BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; } dir->i_sb->s_dirt = 1; btrfs_update_inode_block_group(trans, inode); @@ -1816,7 +1828,7 @@ out_unlock: } struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, - size_t page_offset, u64 start, u64 end, + size_t page_offset, u64 start, u64 len, int create) { int ret; @@ -1826,7 +1838,6 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, u64 extent_end = 0; u64 objectid = inode->i_ino; u32 found_type; - int failed_insert = 0; struct btrfs_path *path; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_file_extent_item *item; @@ -1834,6 +1845,7 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, struct btrfs_key found_key; struct extent_map *em = NULL; 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; path = btrfs_alloc_path(); @@ -1841,24 +1853,26 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, mutex_lock(&root->fs_info->fs_mutex); again: - em = lookup_extent_mapping(em_tree, start, end); + spin_lock(&em_tree->lock); + em = lookup_extent_mapping(em_tree, start, len); + spin_unlock(&em_tree->lock); + if (em) { if (em->start > start) { - printk("get_extent start %Lu em start %Lu\n", - start, em->start); + printk("get_extent lookup [%Lu %Lu] em [%Lu %Lu]\n", + start, len, em->start, em->len); WARN_ON(1); } goto out; } + em = alloc_extent_map(GFP_NOFS); if (!em) { - em = alloc_extent_map(GFP_NOFS); - if (!em) { - err = -ENOMEM; - goto out; - } - em->start = EXTENT_MAP_HOLE; - em->end = EXTENT_MAP_HOLE; + err = -ENOMEM; + goto out; } + + em->start = EXTENT_MAP_HOLE; + em->len = (u64)-1; em->bdev = inode->i_sb->s_bdev; ret = btrfs_lookup_file_extent(trans, root, path, objectid, start, trans != NULL); @@ -1893,28 +1907,25 @@ again: if (start < extent_start || start >= extent_end) { em->start = start; if (start < extent_start) { - if (end < extent_start) + if (start + len <= extent_start) goto not_found; - em->end = extent_end - 1; + em->len = extent_end - extent_start; } else { - em->end = end; + em->len = len; } goto not_found_em; } bytenr = btrfs_file_extent_disk_bytenr(leaf, item); if (bytenr == 0) { em->start = extent_start; - em->end = extent_end - 1; + em->len = extent_end - extent_start; em->block_start = EXTENT_MAP_HOLE; - em->block_end = EXTENT_MAP_HOLE; goto insert; } bytenr += btrfs_file_extent_offset(leaf, item); em->block_start = bytenr; - em->block_end = em->block_start + - btrfs_file_extent_num_bytes(leaf, item) - 1; em->start = extent_start; - em->end = extent_end - 1; + em->len = extent_end - extent_start; goto insert; } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { unsigned long ptr; @@ -1925,25 +1936,24 @@ again: size = btrfs_file_extent_inline_len(leaf, btrfs_item_nr(leaf, path->slots[0])); - extent_end = (extent_start + size - 1) | - ((u64)root->sectorsize - 1); + 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 (end < extent_start) + if (start + len <= extent_start) goto not_found; - em->end = extent_end; + em->len = extent_end - extent_start; } else { - em->end = end; + em->len = len; } goto not_found_em; } em->block_start = EXTENT_MAP_INLINE; - em->block_end = EXTENT_MAP_INLINE; if (!page) { em->start = extent_start; - em->end = extent_start + size - 1; + em->len = size; goto out; } @@ -1952,8 +1962,7 @@ again: copy_size = min_t(u64, PAGE_CACHE_SIZE - page_offset, size - extent_offset); em->start = extent_start + extent_offset; - em->end = (em->start + copy_size -1) | - ((u64)root->sectorsize -1); + em->len = copy_size; map = kmap(page); ptr = btrfs_file_extent_inline_start(item) + extent_offset; if (create == 0 && !PageUptodate(page)) { @@ -1974,7 +1983,8 @@ again: btrfs_mark_buffer_dirty(leaf); } kunmap(page); - set_extent_uptodate(em_tree, em->start, em->end, GFP_NOFS); + 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); @@ -1982,33 +1992,29 @@ again: } not_found: em->start = start; - em->end = end; + em->len = len; not_found_em: em->block_start = EXTENT_MAP_HOLE; - em->block_end = EXTENT_MAP_HOLE; insert: btrfs_release_path(root, path); - if (em->start > start || em->end < start) { - printk("bad extent! em: [%Lu %Lu] passed [%Lu %Lu]\n", em->start, em->end, start, end); + 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); err = -EIO; goto out; } + + err = 0; + spin_lock(&em_tree->lock); ret = add_extent_mapping(em_tree, em); if (ret == -EEXIST) { free_extent_map(em); - em = NULL; - if (0 && failed_insert == 1) { - btrfs_drop_extent_cache(inode, start, end); - } - failed_insert++; - if (failed_insert > 5) { - printk("failing to insert %Lu %Lu\n", start, end); + em = lookup_extent_mapping(em_tree, start, len); + if (!em) { err = -EIO; - goto out; + printk("failing to insert %Lu %Lu\n", start, len); } - goto again; } - err = 0; + spin_unlock(&em_tree->lock); out: btrfs_free_path(path); if (trans) { @@ -2032,14 +2038,14 @@ static sector_t btrfs_bmap(struct address_space *mapping, sector_t iblock) int btrfs_readpage(struct file *file, struct page *page) { - struct extent_map_tree *tree; - tree = &BTRFS_I(page->mapping->host)->extent_tree; + struct extent_io_tree *tree; + tree = &BTRFS_I(page->mapping->host)->io_tree; return extent_read_full_page(tree, page, btrfs_get_extent); } static int btrfs_writepage(struct page *page, struct writeback_control *wbc) { - struct extent_map_tree *tree; + struct extent_io_tree *tree; if (current->flags & PF_MEMALLOC) { @@ -2047,15 +2053,15 @@ static int btrfs_writepage(struct page *page, struct writeback_control *wbc) unlock_page(page); return 0; } - tree = &BTRFS_I(page->mapping->host)->extent_tree; + tree = &BTRFS_I(page->mapping->host)->io_tree; return extent_write_full_page(tree, page, btrfs_get_extent, wbc); } static int btrfs_writepages(struct address_space *mapping, struct writeback_control *wbc) { - struct extent_map_tree *tree; - tree = &BTRFS_I(mapping->host)->extent_tree; + struct extent_io_tree *tree; + tree = &BTRFS_I(mapping->host)->io_tree; return extent_writepages(tree, mapping, btrfs_get_extent, wbc); } @@ -2063,19 +2069,21 @@ static int btrfs_readpages(struct file *file, struct address_space *mapping, struct list_head *pages, unsigned nr_pages) { - struct extent_map_tree *tree; - tree = &BTRFS_I(mapping->host)->extent_tree; + struct extent_io_tree *tree; + tree = &BTRFS_I(mapping->host)->io_tree; return extent_readpages(tree, mapping, pages, nr_pages, btrfs_get_extent); } static int btrfs_releasepage(struct page *page, gfp_t unused_gfp_flags) { - struct extent_map_tree *tree; + struct extent_io_tree *tree; + struct extent_map_tree *map; int ret; - tree = &BTRFS_I(page->mapping->host)->extent_tree; - ret = try_release_extent_mapping(tree, page); + tree = &BTRFS_I(page->mapping->host)->io_tree; + map = &BTRFS_I(page->mapping->host)->extent_tree; + ret = try_release_extent_mapping(map, tree, page); if (ret == 1) { ClearPagePrivate(page); set_page_private(page, 0); @@ -2086,9 +2094,9 @@ static int btrfs_releasepage(struct page *page, gfp_t unused_gfp_flags) static void btrfs_invalidatepage(struct page *page, unsigned long offset) { - struct extent_map_tree *tree; + struct extent_io_tree *tree; - tree = &BTRFS_I(page->mapping->host)->extent_tree; + tree = &BTRFS_I(page->mapping->host)->io_tree; extent_invalidatepage(tree, page, offset); btrfs_releasepage(page, GFP_NOFS); } @@ -2374,7 +2382,7 @@ unsigned long btrfs_force_ra(struct address_space *mapping, int btrfs_defrag_file(struct file *file) { struct inode *inode = fdentry(file)->d_inode; 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 = &BTRFS_I(inode)->io_tree; struct page *page; unsigned long last_index; unsigned long ra_index = 0; @@ -2414,13 +2422,13 @@ int btrfs_defrag_file(struct file *file) { page_start = (u64)page->index << PAGE_CACHE_SHIFT; page_end = page_start + PAGE_CACHE_SIZE - 1; - lock_extent(em_tree, page_start, page_end, GFP_NOFS); + lock_extent(io_tree, page_start, page_end, GFP_NOFS); delalloc_start = page_start; existing_delalloc = - count_range_bits(&BTRFS_I(inode)->extent_tree, + count_range_bits(&BTRFS_I(inode)->io_tree, &delalloc_start, page_end, PAGE_CACHE_SIZE, EXTENT_DELALLOC); - set_extent_delalloc(em_tree, page_start, + set_extent_delalloc(io_tree, page_start, page_end, GFP_NOFS); spin_lock(&root->fs_info->delalloc_lock); @@ -2428,7 +2436,7 @@ int btrfs_defrag_file(struct file *file) { existing_delalloc; spin_unlock(&root->fs_info->delalloc_lock); - unlock_extent(em_tree, page_start, page_end, GFP_NOFS); + unlock_extent(io_tree, page_start, page_end, GFP_NOFS); set_page_dirty(page); unlock_page(page); page_cache_release(page); @@ -2842,9 +2850,10 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry, inode->i_mapping->a_ops = &btrfs_aops; inode->i_fop = &btrfs_file_operations; inode->i_op = &btrfs_file_inode_operations; - extent_map_tree_init(&BTRFS_I(inode)->extent_tree, + 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); - BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops; + BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; } dir->i_sb->s_dirt = 1; btrfs_update_inode_block_group(trans, inode); @@ -2934,7 +2943,7 @@ static struct file_operations btrfs_dir_file_operations = { #endif }; -static struct extent_map_ops btrfs_extent_map_ops = { +static struct extent_io_ops btrfs_extent_io_ops = { .fill_delalloc = run_delalloc_range, .writepage_io_hook = btrfs_writepage_io_hook, .readpage_io_hook = btrfs_readpage_io_hook, diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 8b52c69fda2e..f8a1016600b1 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -468,10 +468,15 @@ static int __init init_btrfs_fs(void) err = btrfs_init_cachep(); if (err) goto free_transaction_sys; - err = extent_map_init(); + + err = extent_io_init(); if (err) goto free_cachep; + err = extent_map_init(); + if (err) + goto free_extent_io; + err = register_filesystem(&btrfs_fs_type); if (err) goto free_extent_map; @@ -479,6 +484,8 @@ static int __init init_btrfs_fs(void) free_extent_map: extent_map_exit(); +free_extent_io: + extent_io_exit(); free_cachep: btrfs_destroy_cachep(); free_transaction_sys: @@ -492,6 +499,7 @@ static void __exit exit_btrfs_fs(void) btrfs_exit_transaction_sys(); btrfs_destroy_cachep(); extent_map_exit(); + extent_io_exit(); unregister_filesystem(&btrfs_fs_type); btrfs_exit_sysfs(); } diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 163c01a24498..b4a1bc62a784 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -70,7 +70,7 @@ static int join_transaction(struct btrfs_root *root) INIT_LIST_HEAD(&cur_trans->pending_snapshots); list_add_tail(&cur_trans->list, &root->fs_info->trans_list); btrfs_ordered_inode_tree_init(&cur_trans->ordered_inode_tree); - extent_map_tree_init(&cur_trans->dirty_pages, + extent_io_tree_init(&cur_trans->dirty_pages, root->fs_info->btree_inode->i_mapping, GFP_NOFS); } else { @@ -153,7 +153,7 @@ int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans, int ret; int err; int werr = 0; - struct extent_map_tree *dirty_pages; + struct extent_io_tree *dirty_pages; struct page *page; struct inode *btree_inode = root->fs_info->btree_inode; u64 start; @@ -610,7 +610,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, struct btrfs_transaction *cur_trans; struct btrfs_transaction *prev_trans = NULL; struct list_head dirty_fs_roots; - struct extent_map_tree *pinned_copy; + struct extent_io_tree *pinned_copy; DEFINE_WAIT(wait); int ret; @@ -639,7 +639,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, if (!pinned_copy) return -ENOMEM; - extent_map_tree_init(pinned_copy, + extent_io_tree_init(pinned_copy, root->fs_info->btree_inode->i_mapping, GFP_NOFS); trans->transaction->in_commit = 1; diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h index fd52e9b23922..c3172ddb3321 100644 --- a/fs/btrfs/transaction.h +++ b/fs/btrfs/transaction.h @@ -29,7 +29,7 @@ struct btrfs_transaction { int use_count; int commit_done; struct list_head list; - struct extent_map_tree dirty_pages; + struct extent_io_tree dirty_pages; unsigned long start_time; struct btrfs_ordered_inode_tree ordered_inode_tree; wait_queue_head_t writer_wait;