X-Git-Url: http://ftp.safe.ca/?a=blobdiff_plain;f=fs%2Fbtrfs%2Ffile.c;h=51466dcd44e0dc26cf96453a02f66171ab0294f2;hb=2e18e047981ae04be9bd0d9760057f7c1a7b3785;hp=00b118a2db6925ba0810a461bd484dfa7a855391;hpb=6702ed490ca0bb44e17131818a5a18b773957c5a;p=safe%2Fjmp%2Flinux-2.6 diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 00b118a..9d08096 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -16,32 +16,36 @@ * Boston, MA 021110-1307, USA. */ -#include #include #include #include #include #include #include -#include #include #include #include #include #include #include -#include #include "ctree.h" #include "disk-io.h" #include "transaction.h" #include "btrfs_inode.h" #include "ioctl.h" #include "print-tree.h" +#include "tree-log.h" +#include "locking.h" +#include "compat.h" -static int btrfs_copy_from_user(loff_t pos, int num_pages, int write_bytes, - struct page **prepared_pages, - const char __user * buf) +/* simple helper to fault in pages and copy. This should go away + * and be replaced with calls into generic code. + */ +static noinline int btrfs_copy_from_user(loff_t pos, int num_pages, + int write_bytes, + struct page **prepared_pages, + const char __user *buf) { long page_fault = 0; int i; @@ -69,119 +73,184 @@ static int btrfs_copy_from_user(loff_t pos, int num_pages, int write_bytes, return page_fault ? -EFAULT : 0; } -static void btrfs_drop_pages(struct page **pages, size_t num_pages) +/* + * unlocks pages after btrfs_file_write is done with them + */ +static noinline void btrfs_drop_pages(struct page **pages, size_t num_pages) { size_t i; for (i = 0; i < num_pages; i++) { if (!pages[i]) break; + /* page checked is some magic around finding pages that + * have been modified without going through btrfs_set_page_dirty + * clear it here + */ + ClearPageChecked(pages[i]); unlock_page(pages[i]); mark_page_accessed(pages[i]); page_cache_release(pages[i]); } } -static int insert_inline_extent(struct btrfs_root *root, struct inode *inode, - u64 offset, ssize_t size, - struct buffer_head *bh) +/* + * after copy_from_user, pages need to be dirtied and we need to make + * sure holes are created between the current EOF and the start of + * any next extents (if required). + * + * this also makes the decision about creating an inline extent vs + * doing real data extents, marking pages dirty and delalloc as required. + */ +static noinline int dirty_and_release_pages(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct file *file, + struct page **pages, + size_t num_pages, + loff_t pos, + size_t write_bytes) { - struct btrfs_key key; - struct btrfs_path *path; - char *ptr, *kaddr; - struct btrfs_trans_handle *trans; - struct btrfs_file_extent_item *ei; - u32 datasize; int err = 0; - int ret; + int i; + struct inode *inode = fdentry(file)->d_inode; + u64 num_bytes; + u64 start_pos; + u64 end_of_last_block; + u64 end_pos = pos + write_bytes; + loff_t isize = i_size_read(inode); - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; + start_pos = pos & ~((u64)root->sectorsize - 1); + num_bytes = (write_bytes + pos - start_pos + + root->sectorsize - 1) & ~((u64)root->sectorsize - 1); - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - btrfs_set_trans_block_group(trans, inode); + end_of_last_block = start_pos + num_bytes - 1; + err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block); + if (err) + return err; - key.objectid = inode->i_ino; - key.offset = offset; - key.flags = 0; - btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY); - BUG_ON(size >= PAGE_CACHE_SIZE); - datasize = btrfs_file_extent_calc_inline_size(size); - - ret = btrfs_insert_empty_item(trans, root, path, &key, - datasize); - if (ret) { - err = ret; - goto fail; + for (i = 0; i < num_pages; i++) { + struct page *p = pages[i]; + SetPageUptodate(p); + ClearPageChecked(p); + set_page_dirty(p); + } + if (end_pos > isize) { + i_size_write(inode, end_pos); + /* we've only changed i_size in ram, and we haven't updated + * the disk i_size. There is no need to log the inode + * at this time. + */ } - ei = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]), - path->slots[0], struct btrfs_file_extent_item); - btrfs_set_file_extent_generation(ei, trans->transid); - btrfs_set_file_extent_type(ei, - BTRFS_FILE_EXTENT_INLINE); - ptr = btrfs_file_extent_inline_start(ei); - - kaddr = kmap_atomic(bh->b_page, KM_USER0); - btrfs_memcpy(root, path->nodes[0]->b_data, - ptr, kaddr + bh_offset(bh), - size); - kunmap_atomic(kaddr, KM_USER0); - btrfs_mark_buffer_dirty(path->nodes[0]); -fail: - btrfs_free_path(path); - ret = btrfs_end_transaction(trans, root); - if (ret && !err) - err = ret; - mutex_unlock(&root->fs_info->fs_mutex); return err; } -static int dirty_and_release_pages(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct file *file, - struct page **pages, - size_t num_pages, - loff_t pos, - size_t write_bytes) +/* + * this drops all the extents in the cache that intersect the range + * [start, end]. Existing extents are split as required. + */ +int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end, + int skip_pinned) { - int i; - int offset; - int err = 0; + struct extent_map *em; + struct extent_map *split = NULL; + struct extent_map *split2 = NULL; + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + u64 len = end - start + 1; int ret; - int this_write; - struct inode *inode = file->f_path.dentry->d_inode; - struct buffer_head *bh; + int testend = 1; + unsigned long flags; + int compressed = 0; - for (i = 0; i < num_pages; i++) { - offset = pos & (PAGE_CACHE_SIZE -1); - this_write = min((size_t)PAGE_CACHE_SIZE - offset, write_bytes); - - /* FIXME, one block at a time */ - bh = page_buffers(pages[i]); - - if (buffer_mapped(bh) && bh->b_blocknr == 0) { - ret = insert_inline_extent(root, inode, - pages[i]->index << PAGE_CACHE_SHIFT, - offset + this_write, bh); - if (ret) { - err = ret; - goto failed; + WARN_ON(end < start); + if (end == (u64)-1) { + len = (u64)-1; + testend = 0; + } + while (1) { + if (!split) + split = alloc_extent_map(GFP_NOFS); + if (!split2) + split2 = alloc_extent_map(GFP_NOFS); + + write_lock(&em_tree->lock); + em = lookup_extent_mapping(em_tree, start, len); + if (!em) { + write_unlock(&em_tree->lock); + break; + } + flags = em->flags; + if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) { + if (testend && em->start + em->len >= start + len) { + free_extent_map(em); + write_unlock(&em_tree->lock); + break; } + start = em->start + em->len; + if (testend) + len = start + len - (em->start + em->len); + free_extent_map(em); + write_unlock(&em_tree->lock); + continue; } + compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags); + clear_bit(EXTENT_FLAG_PINNED, &em->flags); + remove_extent_mapping(em_tree, em); - ret = btrfs_commit_write(file, pages[i], offset, - offset + this_write); - pos += this_write; - if (ret) { - err = ret; - goto failed; + if (em->block_start < EXTENT_MAP_LAST_BYTE && + em->start < start) { + split->start = em->start; + split->len = start - em->start; + split->orig_start = em->orig_start; + split->block_start = em->block_start; + + if (compressed) + split->block_len = em->block_len; + else + split->block_len = split->len; + + split->bdev = em->bdev; + split->flags = flags; + ret = add_extent_mapping(em_tree, split); + BUG_ON(ret); + free_extent_map(split); + split = split2; + split2 = NULL; + } + if (em->block_start < EXTENT_MAP_LAST_BYTE && + testend && em->start + em->len > start + len) { + u64 diff = start + len - em->start; + + split->start = start + len; + split->len = em->start + em->len - (start + len); + split->bdev = em->bdev; + split->flags = flags; + + if (compressed) { + split->block_len = em->block_len; + split->block_start = em->block_start; + split->orig_start = em->orig_start; + } else { + split->block_len = split->len; + split->block_start = em->block_start + diff; + split->orig_start = split->start; + } + + ret = add_extent_mapping(em_tree, split); + BUG_ON(ret); + free_extent_map(split); + split = NULL; } - WARN_ON(this_write > write_bytes); - write_bytes -= this_write; + write_unlock(&em_tree->lock); + + /* once for us */ + free_extent_map(em); + /* once for the tree*/ + free_extent_map(em); } -failed: - return err; + if (split) + free_extent_map(split); + if (split2) + free_extent_map(split2); + return 0; } /* @@ -193,422 +262,618 @@ failed: * it is either truncated or split. Anything entirely inside the range * is deleted from the tree. */ -int btrfs_drop_extents(struct btrfs_trans_handle *trans, - struct btrfs_root *root, struct inode *inode, - u64 start, u64 end, u64 *hint_block) +int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode, + u64 start, u64 end, u64 *hint_byte, int drop_cache) { - int ret; - struct btrfs_key key; - struct btrfs_leaf *leaf; - int slot; - struct btrfs_file_extent_item *extent; - u64 extent_end = 0; - int keep; - struct btrfs_file_extent_item old; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_buffer *leaf; + struct btrfs_file_extent_item *fi; struct btrfs_path *path; + struct btrfs_key key; + struct btrfs_key new_key; u64 search_start = start; - int bookend; - int found_type; - int found_extent; - int found_inline; + u64 disk_bytenr = 0; + u64 num_bytes = 0; + u64 extent_offset = 0; + u64 extent_end = 0; + int del_nr = 0; + int del_slot = 0; + int extent_type; int recow; + int ret; + + if (drop_cache) + btrfs_drop_extent_cache(inode, start, end - 1, 0); path = btrfs_alloc_path(); if (!path) return -ENOMEM; - while(1) { + + while (1) { recow = 0; - btrfs_release_path(root, path); ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino, search_start, -1); if (ret < 0) - goto out; - if (ret > 0) { - if (path->slots[0] == 0) { + break; + if (ret > 0 && path->slots[0] > 0 && search_start == start) { + leaf = path->nodes[0]; + btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1); + if (key.objectid == inode->i_ino && + key.type == BTRFS_EXTENT_DATA_KEY) + path->slots[0]--; + } + ret = 0; +next_slot: + leaf = path->nodes[0]; + if (path->slots[0] >= btrfs_header_nritems(leaf)) { + BUG_ON(del_nr > 0); + ret = btrfs_next_leaf(root, path); + if (ret < 0) + break; + if (ret > 0) { ret = 0; - goto out; + break; } - path->slots[0]--; + leaf = path->nodes[0]; + recow = 1; } -next_slot: - keep = 0; - bookend = 0; - found_extent = 0; - found_inline = 0; - extent = NULL; - leaf = btrfs_buffer_leaf(path->nodes[0]); - slot = path->slots[0]; - ret = 0; - btrfs_disk_key_to_cpu(&key, &leaf->items[slot].key); - if (key.offset >= end || key.objectid != inode->i_ino) { - goto out; + + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); + if (key.objectid > inode->i_ino || + key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end) + break; + + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + extent_type = btrfs_file_extent_type(leaf, fi); + + if (extent_type == BTRFS_FILE_EXTENT_REG || + extent_type == BTRFS_FILE_EXTENT_PREALLOC) { + disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); + num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); + extent_offset = btrfs_file_extent_offset(leaf, fi); + extent_end = key.offset + + btrfs_file_extent_num_bytes(leaf, fi); + } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { + extent_end = key.offset + + btrfs_file_extent_inline_len(leaf, fi); + } else { + WARN_ON(1); + extent_end = search_start; } - if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY) { - goto out; + + if (extent_end <= search_start) { + path->slots[0]++; + goto next_slot; } + + search_start = max(key.offset, start); if (recow) { - search_start = key.offset; + btrfs_release_path(root, path); continue; } - if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) { - extent = btrfs_item_ptr(leaf, slot, - struct btrfs_file_extent_item); - found_type = btrfs_file_extent_type(extent); - if (found_type == BTRFS_FILE_EXTENT_REG) { - extent_end = key.offset + - (btrfs_file_extent_num_blocks(extent) << - inode->i_blkbits); - found_extent = 1; - } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { - found_inline = 1; - extent_end = key.offset + - btrfs_file_extent_inline_len(leaf->items + - slot); + + /* + * | - range to drop - | + * | -------- extent -------- | + */ + if (start > key.offset && end < extent_end) { + BUG_ON(del_nr > 0); + BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); + + memcpy(&new_key, &key, sizeof(new_key)); + new_key.offset = start; + ret = btrfs_duplicate_item(trans, root, path, + &new_key); + if (ret == -EAGAIN) { + btrfs_release_path(root, path); + continue; } - } else { - extent_end = search_start; - } + if (ret < 0) + break; - /* we found nothing we can drop */ - if ((!found_extent && !found_inline) || - search_start >= extent_end) { - int nextret; - u32 nritems; - nritems = btrfs_header_nritems( - btrfs_buffer_header(path->nodes[0])); - if (slot >= nritems - 1) { - nextret = btrfs_next_leaf(root, path); - if (nextret) - goto out; - recow = 1; - } else { - path->slots[0]++; + leaf = path->nodes[0]; + fi = btrfs_item_ptr(leaf, path->slots[0] - 1, + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + start - key.offset); + + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + + extent_offset += start - key.offset; + btrfs_set_file_extent_offset(leaf, fi, extent_offset); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - start); + btrfs_mark_buffer_dirty(leaf); + + if (disk_bytenr > 0) { + ret = btrfs_inc_extent_ref(trans, root, + disk_bytenr, num_bytes, 0, + root->root_key.objectid, + new_key.objectid, + start - extent_offset); + BUG_ON(ret); + *hint_byte = disk_bytenr; } - goto next_slot; + key.offset = start; } + /* + * | ---- range to drop ----- | + * | -------- extent -------- | + */ + if (start <= key.offset && end < extent_end) { + BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); - /* FIXME, there's only one inline extent allowed right now */ - if (found_inline) { - u64 mask = root->blocksize - 1; - search_start = (extent_end + mask) & ~mask; - } else - search_start = extent_end; - - if (end < extent_end && end >= key.offset) { - if (found_extent) { - u64 disk_blocknr = - btrfs_file_extent_disk_blocknr(extent); - u64 disk_num_blocks = - btrfs_file_extent_disk_num_blocks(extent); - memcpy(&old, extent, sizeof(old)); - if (disk_blocknr != 0) { - ret = btrfs_inc_extent_ref(trans, root, - disk_blocknr, disk_num_blocks); - BUG_ON(ret); - } + memcpy(&new_key, &key, sizeof(new_key)); + new_key.offset = end; + btrfs_set_item_key_safe(trans, root, path, &new_key); + + extent_offset += end - key.offset; + btrfs_set_file_extent_offset(leaf, fi, extent_offset); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - end); + btrfs_mark_buffer_dirty(leaf); + if (disk_bytenr > 0) { + inode_sub_bytes(inode, end - key.offset); + *hint_byte = disk_bytenr; } - WARN_ON(found_inline); - bookend = 1; + break; } - /* truncate existing extent */ - if (start > key.offset) { - u64 new_num; - u64 old_num; - keep = 1; - WARN_ON(start & (root->blocksize - 1)); - if (found_extent) { - new_num = (start - key.offset) >> - inode->i_blkbits; - old_num = btrfs_file_extent_num_blocks(extent); - *hint_block = - btrfs_file_extent_disk_blocknr(extent); - if (btrfs_file_extent_disk_blocknr(extent)) { - inode->i_blocks -= - (old_num - new_num) << 3; - } - btrfs_set_file_extent_num_blocks(extent, - new_num); - btrfs_mark_buffer_dirty(path->nodes[0]); - } else { - WARN_ON(1); + + search_start = extent_end; + /* + * | ---- range to drop ----- | + * | -------- extent -------- | + */ + if (start > key.offset && end >= extent_end) { + BUG_ON(del_nr > 0); + BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); + + btrfs_set_file_extent_num_bytes(leaf, fi, + start - key.offset); + btrfs_mark_buffer_dirty(leaf); + if (disk_bytenr > 0) { + inode_sub_bytes(inode, extent_end - start); + *hint_byte = disk_bytenr; } + if (end == extent_end) + break; + + path->slots[0]++; + goto next_slot; } - /* delete the entire extent */ - if (!keep) { - u64 disk_blocknr = 0; - u64 disk_num_blocks = 0; - u64 extent_num_blocks = 0; - if (found_extent) { - disk_blocknr = - btrfs_file_extent_disk_blocknr(extent); - disk_num_blocks = - btrfs_file_extent_disk_num_blocks(extent); - extent_num_blocks = - btrfs_file_extent_num_blocks(extent); - *hint_block = - btrfs_file_extent_disk_blocknr(extent); + + /* + * | ---- range to drop ----- | + * | ------ extent ------ | + */ + if (start <= key.offset && end >= extent_end) { + if (del_nr == 0) { + del_slot = path->slots[0]; + del_nr = 1; + } else { + BUG_ON(del_slot + del_nr != path->slots[0]); + del_nr++; } - ret = btrfs_del_item(trans, root, path); - /* TODO update progress marker and return */ - BUG_ON(ret); - btrfs_release_path(root, path); - extent = NULL; - if (found_extent && disk_blocknr != 0) { - inode->i_blocks -= extent_num_blocks << 3; + + if (extent_type == BTRFS_FILE_EXTENT_INLINE) { + inode_sub_bytes(inode, + extent_end - key.offset); + extent_end = ALIGN(extent_end, + root->sectorsize); + } else if (disk_bytenr > 0) { ret = btrfs_free_extent(trans, root, - disk_blocknr, - disk_num_blocks, 0); + disk_bytenr, num_bytes, 0, + root->root_key.objectid, + key.objectid, key.offset - + extent_offset); + BUG_ON(ret); + inode_sub_bytes(inode, + extent_end - key.offset); + *hint_byte = disk_bytenr; } - BUG_ON(ret); - if (!bookend && search_start >= end) { - ret = 0; - goto out; + if (end == extent_end) + break; + + if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) { + path->slots[0]++; + goto next_slot; } - if (!bookend) - continue; + + ret = btrfs_del_items(trans, root, path, del_slot, + del_nr); + BUG_ON(ret); + + del_nr = 0; + del_slot = 0; + + btrfs_release_path(root, path); + continue; } - /* create bookend, splitting the extent in two */ - if (bookend && found_extent) { - struct btrfs_key ins; - ins.objectid = inode->i_ino; - ins.offset = end; - ins.flags = 0; - btrfs_set_key_type(&ins, BTRFS_EXTENT_DATA_KEY); + + BUG_ON(1); + } + + if (del_nr > 0) { + ret = btrfs_del_items(trans, root, path, del_slot, del_nr); + BUG_ON(ret); + } + + btrfs_free_path(path); + return ret; +} + +static int extent_mergeable(struct extent_buffer *leaf, int slot, + u64 objectid, u64 bytenr, u64 orig_offset, + u64 *start, u64 *end) +{ + struct btrfs_file_extent_item *fi; + struct btrfs_key key; + u64 extent_end; + + if (slot < 0 || slot >= btrfs_header_nritems(leaf)) + return 0; + + btrfs_item_key_to_cpu(leaf, &key, slot); + if (key.objectid != objectid || key.type != BTRFS_EXTENT_DATA_KEY) + return 0; + + fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); + if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG || + btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr || + btrfs_file_extent_offset(leaf, fi) != key.offset - orig_offset || + btrfs_file_extent_compression(leaf, fi) || + btrfs_file_extent_encryption(leaf, fi) || + btrfs_file_extent_other_encoding(leaf, fi)) + return 0; + + extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); + if ((*start && *start != key.offset) || (*end && *end != extent_end)) + return 0; + + *start = key.offset; + *end = extent_end; + return 1; +} + +/* + * Mark extent in the range start - end as written. + * + * This changes extent type from 'pre-allocated' to 'regular'. If only + * part of extent is marked as written, the extent will be split into + * two or three. + */ +int btrfs_mark_extent_written(struct btrfs_trans_handle *trans, + struct inode *inode, u64 start, u64 end) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_buffer *leaf; + struct btrfs_path *path; + struct btrfs_file_extent_item *fi; + struct btrfs_key key; + struct btrfs_key new_key; + u64 bytenr; + u64 num_bytes; + u64 extent_end; + u64 orig_offset; + u64 other_start; + u64 other_end; + u64 split; + int del_nr = 0; + int del_slot = 0; + int recow; + int ret; + + btrfs_drop_extent_cache(inode, start, end - 1, 0); + + path = btrfs_alloc_path(); + BUG_ON(!path); +again: + recow = 0; + split = start; + key.objectid = inode->i_ino; + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = split; + + ret = btrfs_search_slot(trans, root, &key, path, -1, 1); + if (ret > 0 && path->slots[0] > 0) + path->slots[0]--; + + leaf = path->nodes[0]; + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); + BUG_ON(key.objectid != inode->i_ino || + key.type != BTRFS_EXTENT_DATA_KEY); + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + BUG_ON(btrfs_file_extent_type(leaf, fi) != + BTRFS_FILE_EXTENT_PREALLOC); + extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); + BUG_ON(key.offset > start || extent_end < end); + + bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); + num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); + orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi); + memcpy(&new_key, &key, sizeof(new_key)); + + if (start == key.offset && end < extent_end) { + other_start = 0; + other_end = start; + if (extent_mergeable(leaf, path->slots[0] - 1, + inode->i_ino, bytenr, orig_offset, + &other_start, &other_end)) { + new_key.offset = end; + btrfs_set_item_key_safe(trans, root, path, &new_key); + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - end); + btrfs_set_file_extent_offset(leaf, fi, + end - orig_offset); + fi = btrfs_item_ptr(leaf, path->slots[0] - 1, + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + end - other_start); + btrfs_mark_buffer_dirty(leaf); + goto out; + } + } + + if (start > key.offset && end == extent_end) { + other_start = end; + other_end = 0; + if (extent_mergeable(leaf, path->slots[0] + 1, + inode->i_ino, bytenr, orig_offset, + &other_start, &other_end)) { + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + start - key.offset); + path->slots[0]++; + new_key.offset = start; + btrfs_set_item_key_safe(trans, root, path, &new_key); + + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_num_bytes(leaf, fi, + other_end - start); + btrfs_set_file_extent_offset(leaf, fi, + start - orig_offset); + btrfs_mark_buffer_dirty(leaf); + goto out; + } + } + + while (start > key.offset || end < extent_end) { + if (key.offset == start) + split = end; + + new_key.offset = split; + ret = btrfs_duplicate_item(trans, root, path, &new_key); + if (ret == -EAGAIN) { btrfs_release_path(root, path); - ret = btrfs_insert_empty_item(trans, root, path, &ins, - sizeof(*extent)); + goto again; + } + BUG_ON(ret < 0); - if (ret) { - btrfs_print_leaf(root, btrfs_buffer_leaf(path->nodes[0])); - printk("got %d on inserting %Lu %u %Lu start %Lu end %Lu found %Lu %Lu\n", ret , ins.objectid, ins.flags, ins.offset, start, end, key.offset, extent_end); - } - BUG_ON(ret); - extent = btrfs_item_ptr( - btrfs_buffer_leaf(path->nodes[0]), - path->slots[0], + leaf = path->nodes[0]; + fi = btrfs_item_ptr(leaf, path->slots[0] - 1, struct btrfs_file_extent_item); - btrfs_set_file_extent_disk_blocknr(extent, - btrfs_file_extent_disk_blocknr(&old)); - btrfs_set_file_extent_disk_num_blocks(extent, - btrfs_file_extent_disk_num_blocks(&old)); - - btrfs_set_file_extent_offset(extent, - btrfs_file_extent_offset(&old) + - ((end - key.offset) >> inode->i_blkbits)); - WARN_ON(btrfs_file_extent_num_blocks(&old) < - (extent_end - end) >> inode->i_blkbits); - btrfs_set_file_extent_num_blocks(extent, - (extent_end - end) >> inode->i_blkbits); - - btrfs_set_file_extent_type(extent, - BTRFS_FILE_EXTENT_REG); - btrfs_set_file_extent_generation(extent, - btrfs_file_extent_generation(&old)); - btrfs_mark_buffer_dirty(path->nodes[0]); - if (btrfs_file_extent_disk_blocknr(&old) != 0) { - inode->i_blocks += - btrfs_file_extent_num_blocks(extent) << 3; - } - ret = 0; - goto out; + btrfs_set_file_extent_num_bytes(leaf, fi, + split - key.offset); + + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + + btrfs_set_file_extent_offset(leaf, fi, split - orig_offset); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - split); + btrfs_mark_buffer_dirty(leaf); + + ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0, + root->root_key.objectid, + inode->i_ino, orig_offset); + BUG_ON(ret); + + if (split == start) { + key.offset = start; + } else { + BUG_ON(start != key.offset); + path->slots[0]--; + extent_end = end; } + recow = 1; + } + + other_start = end; + other_end = 0; + if (extent_mergeable(leaf, path->slots[0] + 1, + inode->i_ino, bytenr, orig_offset, + &other_start, &other_end)) { + if (recow) { + btrfs_release_path(root, path); + goto again; + } + extent_end = other_end; + del_slot = path->slots[0] + 1; + del_nr++; + ret = btrfs_free_extent(trans, root, bytenr, num_bytes, + 0, root->root_key.objectid, + inode->i_ino, orig_offset); + BUG_ON(ret); + } + other_start = 0; + other_end = start; + if (extent_mergeable(leaf, path->slots[0] - 1, + inode->i_ino, bytenr, orig_offset, + &other_start, &other_end)) { + if (recow) { + btrfs_release_path(root, path); + goto again; + } + key.offset = other_start; + del_slot = path->slots[0]; + del_nr++; + ret = btrfs_free_extent(trans, root, bytenr, num_bytes, + 0, root->root_key.objectid, + inode->i_ino, orig_offset); + BUG_ON(ret); + } + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + if (del_nr == 0) { + btrfs_set_file_extent_type(leaf, fi, + BTRFS_FILE_EXTENT_REG); + btrfs_mark_buffer_dirty(leaf); + } else { + btrfs_set_file_extent_type(leaf, fi, + BTRFS_FILE_EXTENT_REG); + btrfs_set_file_extent_num_bytes(leaf, fi, + extent_end - key.offset); + btrfs_mark_buffer_dirty(leaf); + + ret = btrfs_del_items(trans, root, path, del_slot, del_nr); + BUG_ON(ret); } out: btrfs_free_path(path); - return ret; + return 0; } /* - * this gets pages into the page cache and locks them down + * this gets pages into the page cache and locks them down, it also properly + * waits for data=ordered extents to finish before allowing the pages to be + * modified. */ -static int prepare_pages(struct btrfs_root *root, - struct file *file, - struct page **pages, - size_t num_pages, - loff_t pos, - unsigned long first_index, - unsigned long last_index, - size_t write_bytes) +static noinline int prepare_pages(struct btrfs_root *root, struct file *file, + struct page **pages, size_t num_pages, + loff_t pos, unsigned long first_index, + unsigned long last_index, size_t write_bytes) { int i; unsigned long index = pos >> PAGE_CACHE_SHIFT; - struct inode *inode = file->f_path.dentry->d_inode; - int offset; + struct inode *inode = fdentry(file)->d_inode; int err = 0; - int this_write; - struct buffer_head *bh; - struct buffer_head *head; - loff_t isize = i_size_read(inode); - struct btrfs_trans_handle *trans; - u64 hint_block; - u64 num_blocks; - u64 alloc_extent_start; u64 start_pos; - struct btrfs_key ins; + u64 last_pos; - start_pos = pos & ~((u64)PAGE_CACHE_SIZE - 1); - num_blocks = (write_bytes + pos - start_pos + root->blocksize - 1) >> - inode->i_blkbits; + start_pos = pos & ~((u64)root->sectorsize - 1); + last_pos = ((u64)index + num_pages) << PAGE_CACHE_SHIFT; - memset(pages, 0, num_pages * sizeof(struct page *)); + if (start_pos > inode->i_size) { + err = btrfs_cont_expand(inode, start_pos); + if (err) + return err; + } + memset(pages, 0, num_pages * sizeof(struct page *)); +again: for (i = 0; i < num_pages; i++) { pages[i] = grab_cache_page(inode->i_mapping, index + i); if (!pages[i]) { err = -ENOMEM; - goto failed_release; + BUG_ON(1); } - cancel_dirty_page(pages[i], PAGE_CACHE_SIZE); wait_on_page_writeback(pages[i]); } - - mutex_lock(&root->fs_info->fs_mutex); - trans = btrfs_start_transaction(root, 1); - if (!trans) { - err = -ENOMEM; - mutex_unlock(&root->fs_info->fs_mutex); - goto out_unlock; - } - btrfs_set_trans_block_group(trans, inode); - /* FIXME blocksize != 4096 */ - inode->i_blocks += num_blocks << 3; - hint_block = 0; - - /* FIXME...EIEIO, ENOSPC and more */ - - /* step one, delete the existing extents in this range */ - /* FIXME blocksize != pagesize */ if (start_pos < inode->i_size) { - err = btrfs_drop_extents(trans, root, inode, - start_pos, (pos + write_bytes + root->blocksize -1) & - ~((u64)root->blocksize - 1), &hint_block); - if (err) - goto failed_release; - } - - /* insert any holes we need to create */ - if (inode->i_size < start_pos) { - u64 last_pos_in_file; - u64 hole_size; - u64 mask = root->blocksize - 1; - last_pos_in_file = (isize + mask) & ~mask; - hole_size = (start_pos - last_pos_in_file + mask) & ~mask; - hole_size >>= inode->i_blkbits; - if (last_pos_in_file < start_pos) { - err = btrfs_insert_file_extent(trans, root, - inode->i_ino, - last_pos_in_file, - 0, 0, hole_size); + struct btrfs_ordered_extent *ordered; + lock_extent(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, GFP_NOFS); + ordered = btrfs_lookup_first_ordered_extent(inode, + last_pos - 1); + if (ordered && + ordered->file_offset + ordered->len > start_pos && + ordered->file_offset < last_pos) { + btrfs_put_ordered_extent(ordered); + unlock_extent(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, GFP_NOFS); + for (i = 0; i < num_pages; i++) { + unlock_page(pages[i]); + page_cache_release(pages[i]); + } + btrfs_wait_ordered_range(inode, start_pos, + last_pos - start_pos); + goto again; } - if (err) - goto failed_release; - } + if (ordered) + btrfs_put_ordered_extent(ordered); - /* - * either allocate an extent for the new bytes or setup the key - * to show we are doing inline data in the extent - */ - if (isize >= PAGE_CACHE_SIZE || pos + write_bytes < inode->i_size || - pos + write_bytes - start_pos > BTRFS_MAX_INLINE_DATA_SIZE(root)) { - err = btrfs_alloc_extent(trans, root, inode->i_ino, - num_blocks, 0, hint_block, (u64)-1, - &ins, 1); - if (err) - goto failed_truncate; - err = btrfs_insert_file_extent(trans, root, inode->i_ino, - start_pos, ins.objectid, ins.offset, - ins.offset); - if (err) - goto failed_truncate; - } else { - ins.offset = 0; - ins.objectid = 0; + clear_extent_bits(&BTRFS_I(inode)->io_tree, start_pos, + last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING, + GFP_NOFS); + unlock_extent(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, GFP_NOFS); } - BUG_ON(err); - alloc_extent_start = ins.objectid; - err = btrfs_end_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); - for (i = 0; i < num_pages; i++) { - offset = pos & (PAGE_CACHE_SIZE -1); - this_write = min((size_t)PAGE_CACHE_SIZE - offset, write_bytes); - if (!page_has_buffers(pages[i])) { - create_empty_buffers(pages[i], - root->fs_info->sb->s_blocksize, - (1 << BH_Uptodate)); - } - head = page_buffers(pages[i]); - bh = head; - do { - err = btrfs_map_bh_to_logical(root, bh, - alloc_extent_start); - BUG_ON(err); - if (err) - goto failed_truncate; - bh = bh->b_this_page; - if (alloc_extent_start) - alloc_extent_start++; - } while (bh != head); - pos += this_write; - WARN_ON(this_write > write_bytes); - write_bytes -= this_write; + clear_page_dirty_for_io(pages[i]); + set_page_extent_mapped(pages[i]); + WARN_ON(!PageLocked(pages[i])); } return 0; - -failed_release: - btrfs_drop_pages(pages, num_pages); - return err; - -failed_truncate: - btrfs_drop_pages(pages, num_pages); - if (pos > isize) - vmtruncate(inode, isize); - return err; - -out_unlock: - mutex_unlock(&root->fs_info->fs_mutex); - goto failed_release; - } static ssize_t btrfs_file_write(struct file *file, const char __user *buf, size_t count, loff_t *ppos) { loff_t pos; - size_t num_written = 0; - int err = 0; + loff_t start_pos; + ssize_t num_written = 0; + ssize_t err = 0; int ret = 0; - struct inode *inode = file->f_path.dentry->d_inode; + struct inode *inode = fdentry(file)->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; struct page **pages = NULL; int nrptrs; struct page *pinned[2]; unsigned long first_index; unsigned long last_index; + int will_write; + + will_write = ((file->f_flags & O_DSYNC) || IS_SYNC(inode) || + (file->f_flags & O_DIRECT)); nrptrs = min((count + PAGE_CACHE_SIZE - 1) / PAGE_CACHE_SIZE, PAGE_CACHE_SIZE / (sizeof(struct page *))); pinned[0] = NULL; pinned[1] = NULL; - if (file->f_flags & O_DIRECT) - return -EINVAL; + pos = *ppos; + start_pos = pos; + vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE); + + /* do the reserve before the mutex lock in case we have to do some + * flushing. We wouldn't deadlock, but this is more polite. + */ + err = btrfs_reserve_metadata_for_delalloc(root, inode, 1); + if (err) + goto out_nolock; + + mutex_lock(&inode->i_mutex); + current->backing_dev_info = inode->i_mapping->backing_dev_info; err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode)); if (err) goto out; + if (count == 0) goto out; - err = remove_suid(file->f_path.dentry); + + err = file_remove_suid(file); if (err) goto out; + file_update_time(file); pages = kmalloc(nrptrs * sizeof(struct page *), GFP_KERNEL); - mutex_lock(&inode->i_mutex); + /* generic_write_checks can change our pos */ + start_pos = pos; + + BTRFS_I(inode)->sequence++; first_index = pos >> PAGE_CACHE_SHIFT; last_index = (pos + count) >> PAGE_CACHE_SHIFT; @@ -638,7 +903,7 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf, } } - while(count > 0) { + while (count > 0) { size_t offset = pos & (PAGE_CACHE_SIZE - 1); size_t write_bytes = min(count, nrptrs * (size_t)PAGE_CACHE_SIZE - @@ -647,16 +912,26 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf, PAGE_CACHE_SHIFT; WARN_ON(num_pages > nrptrs); - memset(pages, 0, sizeof(pages)); + memset(pages, 0, sizeof(struct page *) * nrptrs); + + ret = btrfs_check_data_free_space(root, inode, write_bytes); + if (ret) + goto out; + ret = prepare_pages(root, file, pages, num_pages, pos, first_index, last_index, write_bytes); - if (ret) + if (ret) { + btrfs_free_reserved_data_space(root, inode, + write_bytes); goto out; + } ret = btrfs_copy_from_user(pos, num_pages, write_bytes, pages, buf); if (ret) { + btrfs_free_reserved_data_space(root, inode, + write_bytes); btrfs_drop_pages(pages, num_pages); goto out; } @@ -664,62 +939,205 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf, ret = dirty_and_release_pages(NULL, root, file, pages, num_pages, pos, write_bytes); btrfs_drop_pages(pages, num_pages); - if (ret) + if (ret) { + btrfs_free_reserved_data_space(root, inode, + write_bytes); goto out; + } + + if (will_write) { + filemap_fdatawrite_range(inode->i_mapping, pos, + pos + write_bytes - 1); + } else { + balance_dirty_pages_ratelimited_nr(inode->i_mapping, + num_pages); + if (num_pages < + (root->leafsize >> PAGE_CACHE_SHIFT) + 1) + btrfs_btree_balance_dirty(root, 1); + btrfs_throttle(root); + } buf += write_bytes; count -= write_bytes; pos += write_bytes; num_written += write_bytes; - balance_dirty_pages_ratelimited_nr(inode->i_mapping, num_pages); - btrfs_btree_balance_dirty(root); cond_resched(); } - mutex_unlock(&inode->i_mutex); out: + mutex_unlock(&inode->i_mutex); + if (ret) + err = ret; + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); + +out_nolock: kfree(pages); if (pinned[0]) page_cache_release(pinned[0]); if (pinned[1]) page_cache_release(pinned[1]); *ppos = pos; + + /* + * we want to make sure fsync finds this change + * but we haven't joined a transaction running right now. + * + * Later on, someone is sure to update the inode and get the + * real transid recorded. + * + * We set last_trans now to the fs_info generation + 1, + * this will either be one more than the running transaction + * or the generation used for the next transaction if there isn't + * one running right now. + */ + BTRFS_I(inode)->last_trans = root->fs_info->generation + 1; + + if (num_written > 0 && will_write) { + struct btrfs_trans_handle *trans; + + err = btrfs_wait_ordered_range(inode, start_pos, num_written); + if (err) + num_written = err; + + if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) { + trans = btrfs_start_transaction(root, 1); + ret = btrfs_log_dentry_safe(trans, root, + file->f_dentry); + if (ret == 0) { + ret = btrfs_sync_log(trans, root); + if (ret == 0) + btrfs_end_transaction(trans, root); + else + btrfs_commit_transaction(trans, root); + } else if (ret != BTRFS_NO_LOG_SYNC) { + btrfs_commit_transaction(trans, root); + } else { + btrfs_end_transaction(trans, root); + } + } + if (file->f_flags & O_DIRECT) { + invalidate_mapping_pages(inode->i_mapping, + start_pos >> PAGE_CACHE_SHIFT, + (start_pos + num_written - 1) >> PAGE_CACHE_SHIFT); + } + } current->backing_dev_info = NULL; - mark_inode_dirty(inode); return num_written ? num_written : err; } -static int btrfs_sync_file(struct file *file, - struct dentry *dentry, int datasync) +int btrfs_release_file(struct inode *inode, struct file *filp) +{ + /* + * ordered_data_close is set by settattr when we are about to truncate + * a file from a non-zero size to a zero size. This tries to + * flush down new bytes that may have been written if the + * application were using truncate to replace a file in place. + */ + if (BTRFS_I(inode)->ordered_data_close) { + BTRFS_I(inode)->ordered_data_close = 0; + btrfs_add_ordered_operation(NULL, BTRFS_I(inode)->root, inode); + if (inode->i_size > BTRFS_ORDERED_OPERATIONS_FLUSH_LIMIT) + filemap_flush(inode->i_mapping); + } + if (filp->private_data) + btrfs_ioctl_trans_end(filp); + return 0; +} + +/* + * fsync call for both files and directories. This logs the inode into + * the tree log instead of forcing full commits whenever possible. + * + * It needs to call filemap_fdatawait so that all ordered extent updates are + * in the metadata btree are up to date for copying to the log. + * + * It drops the inode mutex before doing the tree log commit. This is an + * important optimization for directories because holding the mutex prevents + * new operations on the dir while we write to disk. + */ +int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync) { struct inode *inode = dentry->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; - int ret; + int ret = 0; struct btrfs_trans_handle *trans; + + /* we wait first, since the writeback may change the inode */ + root->log_batch++; + /* the VFS called filemap_fdatawrite for us */ + btrfs_wait_ordered_range(inode, 0, (u64)-1); + root->log_batch++; + + /* + * check the transaction that last modified this inode + * and see if its already been committed + */ + if (!BTRFS_I(inode)->last_trans) + goto out; + + /* + * if the last transaction that changed this file was before + * the current transaction, we can bail out now without any + * syncing + */ + mutex_lock(&root->fs_info->trans_mutex); + if (BTRFS_I(inode)->last_trans <= + root->fs_info->last_trans_committed) { + BTRFS_I(inode)->last_trans = 0; + mutex_unlock(&root->fs_info->trans_mutex); + goto out; + } + mutex_unlock(&root->fs_info->trans_mutex); + /* - * FIXME, use inode generation number to check if we can skip the - * commit + * ok we haven't committed the transaction yet, lets do a commit */ - mutex_lock(&root->fs_info->fs_mutex); + if (file && file->private_data) + btrfs_ioctl_trans_end(file); + trans = btrfs_start_transaction(root, 1); if (!trans) { ret = -ENOMEM; goto out; } - ret = btrfs_commit_transaction(trans, root); - mutex_unlock(&root->fs_info->fs_mutex); + + ret = btrfs_log_dentry_safe(trans, root, dentry); + if (ret < 0) + goto out; + + /* we've logged all the items and now have a consistent + * version of the file in the log. It is possible that + * someone will come in and modify the file, but that's + * fine because the log is consistent on disk, and we + * have references to all of the file's extents + * + * It is possible that someone will come in and log the + * file again, but that will end up using the synchronization + * inside btrfs_sync_log to keep things safe. + */ + mutex_unlock(&dentry->d_inode->i_mutex); + + if (ret != BTRFS_NO_LOG_SYNC) { + if (ret > 0) { + ret = btrfs_commit_transaction(trans, root); + } else { + ret = btrfs_sync_log(trans, root); + if (ret == 0) + ret = btrfs_end_transaction(trans, root); + else + ret = btrfs_commit_transaction(trans, root); + } + } else { + ret = btrfs_end_transaction(trans, root); + } + mutex_lock(&dentry->d_inode->i_mutex); out: - return ret > 0 ? EIO : ret; + return ret > 0 ? -EIO : ret; } -static struct vm_operations_struct btrfs_file_vm_ops = { -#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23) - .nopage = filemap_nopage, - .populate = filemap_populate, -#else +static const struct vm_operations_struct btrfs_file_vm_ops = { .fault = filemap_fault, -#endif .page_mkwrite = btrfs_page_mkwrite, }; @@ -730,17 +1148,18 @@ static int btrfs_file_mmap(struct file *filp, struct vm_area_struct *vma) return 0; } -struct file_operations btrfs_file_operations = { +const struct file_operations btrfs_file_operations = { .llseek = generic_file_llseek, .read = do_sync_read, .aio_read = generic_file_aio_read, + .splice_read = generic_file_splice_read, .write = btrfs_file_write, .mmap = btrfs_file_mmap, .open = generic_file_open, - .ioctl = btrfs_ioctl, + .release = btrfs_release_file, .fsync = btrfs_sync_file, + .unlocked_ioctl = btrfs_ioctl, #ifdef CONFIG_COMPAT - .compat_ioctl = btrfs_compat_ioctl, + .compat_ioctl = btrfs_ioctl, #endif }; -