From 941b2ddf71987ef369389517a7e215dd505fe01e Mon Sep 17 00:00:00 2001 From: Keith Mannthey Date: Tue, 29 Nov 2011 17:44:12 -0800 Subject: btrfs: Sector Size check during Mount Gracefully fail when trying to mount a BTRFS file system that has a sectorsize smaller than PAGE_SIZE. On PPC it is possible to build a FS while using a 4k PAGE_SIZE kernel then boot into a 64K PAGE_SIZE kernel. Presently open_ctree fails in an endless loop and hangs the machine in this situation. My debugging has show this Sector size < Page size to be a non trivial situation and a graceful exit from the situation would be nice for the time being. Signed-off-by: Keith Mannthey --- fs/btrfs/disk-io.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 4c867112b4c8..58d0678dfcba 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2258,6 +2258,12 @@ struct btrfs_root *open_ctree(struct super_block *sb, goto fail_sb_buffer; } + if (sectorsize < PAGE_SIZE) { + printk(KERN_WARNING "btrfs: Incompatible sector size " + "found on %s\n", sb->s_id); + goto fail_sb_buffer; + } + mutex_lock(&fs_info->chunk_mutex); ret = btrfs_read_sys_array(tree_root); mutex_unlock(&fs_info->chunk_mutex); -- cgit v1.2.3 From a6b0d5c8dbfd428717fc4db4c36757783f391c7b Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 20 Feb 2012 20:53:43 -0500 Subject: Btrfs: make sure we update latest_bdev When we are setting up the mount, we close all the devices that were not actually part of the metadata we found. But, we don't make sure that one of those devices wasn't fs_devices->latest_bdev, which means we can do a use after free on the one we closed. This updates latest_bdev as it goes. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 58d0678dfcba..b801d29f3f10 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2305,6 +2305,12 @@ struct btrfs_root *open_ctree(struct super_block *sb, btrfs_close_extra_devices(fs_devices); + if (!fs_devices->latest_bdev) { + printk(KERN_CRIT "btrfs: failed to read devices on %s\n", + sb->s_id); + goto fail_tree_roots; + } + retry_root_backup: blocksize = btrfs_level_size(tree_root, btrfs_super_root_level(disk_super)); -- cgit v1.2.3 From d5c13f927fe77b11a67f79559808c68b26474c77 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Thu, 1 Mar 2012 14:56:27 +0100 Subject: btrfs: clean_tree_block should panic on observed memory corruption and return void The only error condition in clean_tree_block is an accounting bug. Returning without modifying dirty_metadata_bytes and as if the cleaning as been performed may cause problems later so it should panic instead. It should probably be a BUG_ON but we have btrfs_panic now. Signed-off-by: Jeff Mahoney --- fs/btrfs/disk-io.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 534266fe505f..d52ec1155204 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1109,8 +1109,8 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr, } -int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct extent_buffer *buf) +void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, + struct extent_buffer *buf) { struct inode *btree_inode = root->fs_info->btree_inode; if (btrfs_header_generation(buf) == @@ -1121,8 +1121,14 @@ int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, spin_lock(&root->fs_info->delalloc_lock); if (root->fs_info->dirty_metadata_bytes >= buf->len) root->fs_info->dirty_metadata_bytes -= buf->len; - else - WARN_ON(1); + else { + spin_unlock(&root->fs_info->delalloc_lock); + btrfs_panic(root->fs_info, -EOVERFLOW, + "Can't clear %lu bytes from " + " dirty_mdatadata_bytes (%lu)", + buf->len, + root->fs_info->dirty_metadata_bytes); + } spin_unlock(&root->fs_info->delalloc_lock); } @@ -1131,7 +1137,6 @@ int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, buf); } - return 0; } static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, -- cgit v1.2.3 From 200a5c17677e1ee8b78382046f3748c9f5816281 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Mon, 3 Oct 2011 23:22:43 -0400 Subject: btrfs: find_and_setup_root error push-up find_and_setup_root BUGs when it encounters an error from btrfs_find_last_root, which can occur if a path can't be allocated. This patch pushes it up to its callers where it is already handled. Signed-off-by: Jeff Mahoney --- fs/btrfs/disk-io.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index d52ec1155204..155c4e35e536 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1197,10 +1197,10 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, return 0; } -static int find_and_setup_root(struct btrfs_root *tree_root, - struct btrfs_fs_info *fs_info, - u64 objectid, - struct btrfs_root *root) +static int __must_check find_and_setup_root(struct btrfs_root *tree_root, + struct btrfs_fs_info *fs_info, + u64 objectid, + struct btrfs_root *root) { int ret; u32 blocksize; @@ -1213,7 +1213,8 @@ static int find_and_setup_root(struct btrfs_root *tree_root, &root->root_item, &root->root_key); if (ret > 0) return -ENOENT; - BUG_ON(ret); + else if (ret < 0) + return ret; generation = btrfs_root_generation(&root->root_item); blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item)); -- cgit v1.2.3 From 355808c296c6923db6705f43639969a80b16d15d Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Mon, 3 Oct 2011 23:23:14 -0400 Subject: btrfs: ->submit_bio_hook error push-up This pushes failures from the submit_bio_hook callbacks, btrfs_submit_bio_hook and btree_submit_bio_hook into the callers, including callers of submit_one_bio where it catches the failures with BUG_ON. It also pushes up through the ->readpage_io_failed_hook to end_bio_extent_writepage where the error is already caught with BUG_ON. Signed-off-by: Jeff Mahoney --- fs/btrfs/disk-io.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 155c4e35e536..2e4428bd60ec 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -847,9 +847,9 @@ static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, { int ret; - ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info, - bio, 1); - BUG_ON(ret); + ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info, bio, 1); + if (ret) + return ret; if (!(rw & REQ_WRITE)) { /* -- cgit v1.2.3 From 143bede527b054a271053f41bfaca2b57baa9408 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Thu, 1 Mar 2012 14:56:26 +0100 Subject: btrfs: return void in functions without error conditions Signed-off-by: Jeff Mahoney --- fs/btrfs/disk-io.c | 39 ++++++++++++++------------------------- 1 file changed, 14 insertions(+), 25 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 2e4428bd60ec..73ccadce90bc 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -50,12 +50,12 @@ static void end_workqueue_fn(struct btrfs_work *work); static void free_fs_root(struct btrfs_root *root); static void btrfs_check_super_valid(struct btrfs_fs_info *fs_info, int read_only); -static int btrfs_destroy_ordered_operations(struct btrfs_root *root); -static int btrfs_destroy_ordered_extents(struct btrfs_root *root); +static void btrfs_destroy_ordered_operations(struct btrfs_root *root); +static void btrfs_destroy_ordered_extents(struct btrfs_root *root); static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, struct btrfs_root *root); -static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t); -static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root); +static void btrfs_destroy_pending_snapshots(struct btrfs_transaction *t); +static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root); static int btrfs_destroy_marked_extents(struct btrfs_root *root, struct extent_io_tree *dirty_pages, int mark); @@ -1139,10 +1139,10 @@ void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, } } -static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, - u32 stripesize, struct btrfs_root *root, - struct btrfs_fs_info *fs_info, - u64 objectid) +static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, + u32 stripesize, struct btrfs_root *root, + struct btrfs_fs_info *fs_info, + u64 objectid) { root->node = NULL; root->commit_root = NULL; @@ -1194,7 +1194,6 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->defrag_running = 0; root->root_key.objectid = objectid; root->anon_dev = 0; - return 0; } static int __must_check find_and_setup_root(struct btrfs_root *tree_root, @@ -2897,7 +2896,7 @@ int write_ctree_super(struct btrfs_trans_handle *trans, return ret; } -int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root) +void btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root) { spin_lock(&fs_info->fs_roots_radix_lock); radix_tree_delete(&fs_info->fs_roots_radix, @@ -2910,7 +2909,6 @@ int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root) __btrfs_remove_free_space_cache(root->free_ino_pinned); __btrfs_remove_free_space_cache(root->free_ino_ctl); free_fs_root(root); - return 0; } static void free_fs_root(struct btrfs_root *root) @@ -2927,7 +2925,7 @@ static void free_fs_root(struct btrfs_root *root) kfree(root); } -static int del_fs_roots(struct btrfs_fs_info *fs_info) +static void del_fs_roots(struct btrfs_fs_info *fs_info) { int ret; struct btrfs_root *gang[8]; @@ -2956,7 +2954,6 @@ static int del_fs_roots(struct btrfs_fs_info *fs_info) for (i = 0; i < ret; i++) btrfs_free_fs_root(fs_info, gang[i]); } - return 0; } int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info) @@ -3299,7 +3296,7 @@ int btrfs_error_commit_super(struct btrfs_root *root) return ret; } -static int btrfs_destroy_ordered_operations(struct btrfs_root *root) +static void btrfs_destroy_ordered_operations(struct btrfs_root *root) { struct btrfs_inode *btrfs_inode; struct list_head splice; @@ -3321,11 +3318,9 @@ static int btrfs_destroy_ordered_operations(struct btrfs_root *root) spin_unlock(&root->fs_info->ordered_extent_lock); mutex_unlock(&root->fs_info->ordered_operations_mutex); - - return 0; } -static int btrfs_destroy_ordered_extents(struct btrfs_root *root) +static void btrfs_destroy_ordered_extents(struct btrfs_root *root) { struct list_head splice; struct btrfs_ordered_extent *ordered; @@ -3357,8 +3352,6 @@ static int btrfs_destroy_ordered_extents(struct btrfs_root *root) } spin_unlock(&root->fs_info->ordered_extent_lock); - - return 0; } static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, @@ -3413,7 +3406,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, return ret; } -static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t) +static void btrfs_destroy_pending_snapshots(struct btrfs_transaction *t) { struct btrfs_pending_snapshot *snapshot; struct list_head splice; @@ -3431,11 +3424,9 @@ static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t) kfree(snapshot); } - - return 0; } -static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root) +static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root) { struct btrfs_inode *btrfs_inode; struct list_head splice; @@ -3455,8 +3446,6 @@ static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root) } spin_unlock(&root->fs_info->delalloc_lock); - - return 0; } static int btrfs_destroy_marked_extents(struct btrfs_root *root, -- cgit v1.2.3 From d0082371cf086e0ba2bbd0367b2c9920532df24f Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Thu, 1 Mar 2012 14:57:19 +0100 Subject: btrfs: drop gfp_t from lock_extent lock_extent and unlock_extent are always called with GFP_NOFS, drop the argument and use GFP_NOFS consistently. Signed-off-by: Jeff Mahoney --- fs/btrfs/disk-io.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 73ccadce90bc..69ef456b32fa 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -332,7 +332,7 @@ static int verify_parent_transid(struct extent_io_tree *io_tree, return 0; lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1, - 0, &cached_state, GFP_NOFS); + 0, &cached_state); if (extent_buffer_uptodate(io_tree, eb, cached_state) && btrfs_header_generation(eb) == parent_transid) { ret = 0; -- cgit v1.2.3 From 49b25e0540904be0bf558b84475c69d72e4de66e Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Thu, 1 Mar 2012 17:24:58 +0100 Subject: btrfs: enhance transaction abort infrastructure Signed-off-by: Jeff Mahoney --- fs/btrfs/disk-io.c | 50 ++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 46 insertions(+), 4 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 69ef456b32fa..6297a030ac50 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -61,7 +61,6 @@ static int btrfs_destroy_marked_extents(struct btrfs_root *root, int mark); static int btrfs_destroy_pinned_extent(struct btrfs_root *root, struct extent_io_tree *pinned_extents); -static int btrfs_cleanup_transaction(struct btrfs_root *root); /* * end_io_wq structs are used to do processing in task context when an IO is @@ -2896,6 +2895,19 @@ int write_ctree_super(struct btrfs_trans_handle *trans, return ret; } +/* Kill all outstanding I/O */ +void btrfs_abort_devices(struct btrfs_root *root) +{ + struct list_head *head; + struct btrfs_device *dev; + mutex_lock(&root->fs_info->fs_devices->device_list_mutex); + head = &root->fs_info->fs_devices->devices; + list_for_each_entry_rcu(dev, head, dev_list) { + blk_abort_queue(dev->bdev->bd_disk->queue); + } + mutex_unlock(&root->fs_info->fs_devices->device_list_mutex); +} + void btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root) { spin_lock(&fs_info->fs_roots_radix_lock); @@ -3536,13 +3548,43 @@ static int btrfs_destroy_pinned_extent(struct btrfs_root *root, return 0; } -static int btrfs_cleanup_transaction(struct btrfs_root *root) +void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans, + struct btrfs_root *root) +{ + btrfs_destroy_delayed_refs(cur_trans, root); + btrfs_block_rsv_release(root, &root->fs_info->trans_block_rsv, + cur_trans->dirty_pages.dirty_bytes); + + /* FIXME: cleanup wait for commit */ + cur_trans->in_commit = 1; + cur_trans->blocked = 1; + if (waitqueue_active(&root->fs_info->transaction_blocked_wait)) + wake_up(&root->fs_info->transaction_blocked_wait); + + cur_trans->blocked = 0; + if (waitqueue_active(&root->fs_info->transaction_wait)) + wake_up(&root->fs_info->transaction_wait); + + cur_trans->commit_done = 1; + if (waitqueue_active(&cur_trans->commit_wait)) + wake_up(&cur_trans->commit_wait); + + btrfs_destroy_pending_snapshots(cur_trans); + + btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages, + EXTENT_DIRTY); + + /* + memset(cur_trans, 0, sizeof(*cur_trans)); + kmem_cache_free(btrfs_transaction_cachep, cur_trans); + */ +} + +int btrfs_cleanup_transaction(struct btrfs_root *root) { struct btrfs_transaction *t; LIST_HEAD(list); - WARN_ON(1); - mutex_lock(&root->fs_info->transaction_kthread_mutex); spin_lock(&root->fs_info->trans_lock); -- cgit v1.2.3 From 79787eaab46121d4713ed03c8fc63b9ec3eaec76 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Mon, 12 Mar 2012 16:03:00 +0100 Subject: btrfs: replace many BUG_ONs with proper error handling btrfs currently handles most errors with BUG_ON. This patch is a work-in- progress but aims to handle most errors other than internal logic errors and ENOMEM more gracefully. This iteration prevents most crashes but can run into lockups with the page lock on occasion when the timing "works out." Signed-off-by: Jeff Mahoney --- fs/btrfs/disk-io.c | 116 +++++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 86 insertions(+), 30 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 6297a030ac50..16a0cada26c2 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -98,6 +98,7 @@ struct async_submit_bio { */ u64 bio_offset; struct btrfs_work work; + int error; }; /* @@ -405,7 +406,7 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) u64 found_start; unsigned long len; struct extent_buffer *eb; - int ret; + int ret = -EIO; tree = &BTRFS_I(page->mapping->host)->io_tree; @@ -423,13 +424,20 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) eb = alloc_extent_buffer(tree, start, len, page); if (eb == NULL) { WARN_ON(1); + ret = -ENOMEM; goto out; } ret = btree_read_extent_buffer_pages(root, eb, start + PAGE_CACHE_SIZE, btrfs_header_generation(eb)); - BUG_ON(ret); + if (ret) { + btrfs_printk(root->fs_info, KERN_WARNING + "Failed to checksum dirty buffer @ %llu[%lu]\n", + start, len); + goto err; + } WARN_ON(!btrfs_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN)); + ret = -EIO; found_start = btrfs_header_bytenr(eb); if (found_start != start) { WARN_ON(1); @@ -444,10 +452,11 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) goto err; } csum_tree_block(root, eb, 0); + ret = 0; err: free_extent_buffer(eb); out: - return 0; + return ret; } static int check_tree_block_fsid(struct btrfs_root *root, @@ -718,11 +727,14 @@ unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info) static void run_one_async_start(struct btrfs_work *work) { struct async_submit_bio *async; + int ret; async = container_of(work, struct async_submit_bio, work); - async->submit_bio_start(async->inode, async->rw, async->bio, - async->mirror_num, async->bio_flags, - async->bio_offset); + ret = async->submit_bio_start(async->inode, async->rw, async->bio, + async->mirror_num, async->bio_flags, + async->bio_offset); + if (ret) + async->error = ret; } static void run_one_async_done(struct btrfs_work *work) @@ -743,6 +755,12 @@ static void run_one_async_done(struct btrfs_work *work) waitqueue_active(&fs_info->async_submit_wait)) wake_up(&fs_info->async_submit_wait); + /* If an error occured we just want to clean up the bio and move on */ + if (async->error) { + bio_endio(async->bio, async->error); + return; + } + async->submit_bio_done(async->inode, async->rw, async->bio, async->mirror_num, async->bio_flags, async->bio_offset); @@ -784,6 +802,8 @@ int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode, async->bio_flags = bio_flags; async->bio_offset = bio_offset; + async->error = 0; + atomic_inc(&fs_info->nr_async_submits); if (rw & REQ_SYNC) @@ -805,15 +825,18 @@ static int btree_csum_one_bio(struct bio *bio) struct bio_vec *bvec = bio->bi_io_vec; int bio_index = 0; struct btrfs_root *root; + int ret = 0; WARN_ON(bio->bi_vcnt <= 0); while (bio_index < bio->bi_vcnt) { root = BTRFS_I(bvec->bv_page->mapping->host)->root; - csum_dirty_buffer(root, bvec->bv_page); + ret = csum_dirty_buffer(root, bvec->bv_page); + if (ret) + break; bio_index++; bvec++; } - return 0; + return ret; } static int __btree_submit_bio_start(struct inode *inode, int rw, @@ -825,8 +848,7 @@ static int __btree_submit_bio_start(struct inode *inode, int rw, * when we're called for a write, we're already in the async * submission context. Just jump into btrfs_map_bio */ - btree_csum_one_bio(bio); - return 0; + return btree_csum_one_bio(bio); } static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio, @@ -1381,7 +1403,7 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root, root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item), blocksize, generation); root->commit_root = btrfs_root_node(root); - BUG_ON(!root->node); + BUG_ON(!root->node); /* -ENOMEM */ out: if (location->objectid != BTRFS_TREE_LOG_OBJECTID) { root->ref_cows = 1; @@ -1618,7 +1640,6 @@ static int transaction_kthread(void *arg) u64 transid; unsigned long now; unsigned long delay; - int ret; do { delay = HZ * 30; @@ -1642,11 +1663,12 @@ static int transaction_kthread(void *arg) transid = cur->transid; spin_unlock(&root->fs_info->trans_lock); + /* If the file system is aborted, this will always fail. */ trans = btrfs_join_transaction(root); - BUG_ON(IS_ERR(trans)); + if (IS_ERR(trans)) + goto sleep; if (transid == trans->transid) { - ret = btrfs_commit_transaction(trans, root); - BUG_ON(ret); + btrfs_commit_transaction(trans, root); } else { btrfs_end_transaction(trans, root); } @@ -2289,7 +2311,7 @@ int open_ctree(struct super_block *sb, chunk_root->node = read_tree_block(chunk_root, btrfs_super_chunk_root(disk_super), blocksize, generation); - BUG_ON(!chunk_root->node); + BUG_ON(!chunk_root->node); /* -ENOMEM */ if (!test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) { printk(KERN_WARNING "btrfs: failed to read chunk root on %s\n", sb->s_id); @@ -2429,21 +2451,31 @@ retry_root_backup: log_tree_root->node = read_tree_block(tree_root, bytenr, blocksize, generation + 1); + /* returns with log_tree_root freed on success */ ret = btrfs_recover_log_trees(log_tree_root); - BUG_ON(ret); + if (ret) { + btrfs_error(tree_root->fs_info, ret, + "Failed to recover log tree"); + free_extent_buffer(log_tree_root->node); + kfree(log_tree_root); + goto fail_trans_kthread; + } if (sb->s_flags & MS_RDONLY) { - ret = btrfs_commit_super(tree_root); - BUG_ON(ret); + ret = btrfs_commit_super(tree_root); + if (ret) + goto fail_trans_kthread; } } ret = btrfs_find_orphan_roots(tree_root); - BUG_ON(ret); + if (ret) + goto fail_trans_kthread; if (!(sb->s_flags & MS_RDONLY)) { ret = btrfs_cleanup_fs_roots(fs_info); - BUG_ON(ret); + if (ret) { + } ret = btrfs_recover_relocation(tree_root); if (ret < 0) { @@ -2863,6 +2895,8 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors) if (total_errors > max_errors) { printk(KERN_ERR "btrfs: %d errors while writing supers\n", total_errors); + + /* This shouldn't happen. FUA is masked off if unsupported */ BUG(); } @@ -2879,9 +2913,9 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors) } mutex_unlock(&root->fs_info->fs_devices->device_list_mutex); if (total_errors > max_errors) { - printk(KERN_ERR "btrfs: %d errors while writing supers\n", - total_errors); - BUG(); + btrfs_error(root->fs_info, -EIO, + "%d errors while writing supers", total_errors); + return -EIO; } return 0; } @@ -3014,14 +3048,21 @@ int btrfs_commit_super(struct btrfs_root *root) if (IS_ERR(trans)) return PTR_ERR(trans); ret = btrfs_commit_transaction(trans, root); - BUG_ON(ret); + if (ret) + return ret; /* run commit again to drop the original snapshot */ trans = btrfs_join_transaction(root); if (IS_ERR(trans)) return PTR_ERR(trans); - btrfs_commit_transaction(trans, root); + ret = btrfs_commit_transaction(trans, root); + if (ret) + return ret; ret = btrfs_write_and_wait_transaction(NULL, root); - BUG_ON(ret); + if (ret) { + btrfs_error(root->fs_info, ret, + "Failed to sync btree inode to disk."); + return ret; + } ret = write_ctree_super(NULL, root, 0); return ret; @@ -3366,8 +3407,8 @@ static void btrfs_destroy_ordered_extents(struct btrfs_root *root) spin_unlock(&root->fs_info->ordered_extent_lock); } -static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, - struct btrfs_root *root) +int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, + struct btrfs_root *root) { struct rb_node *node; struct btrfs_delayed_ref_root *delayed_refs; @@ -3376,6 +3417,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, delayed_refs = &trans->delayed_refs; +again: spin_lock(&delayed_refs->lock); if (delayed_refs->num_entries == 0) { spin_unlock(&delayed_refs->lock); @@ -3397,6 +3439,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, struct btrfs_delayed_ref_head *head; head = btrfs_delayed_node_to_head(ref); + spin_unlock(&delayed_refs->lock); mutex_lock(&head->mutex); kfree(head->extent_op); delayed_refs->num_heads--; @@ -3404,8 +3447,9 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, delayed_refs->num_heads_ready--; list_del_init(&head->cluster); mutex_unlock(&head->mutex); + btrfs_put_delayed_ref(ref); + goto again; } - spin_unlock(&delayed_refs->lock); btrfs_put_delayed_ref(ref); @@ -3649,6 +3693,17 @@ int btrfs_cleanup_transaction(struct btrfs_root *root) return 0; } +static int btree_writepage_io_failed_hook(struct bio *bio, struct page *page, + u64 start, u64 end, + struct extent_state *state) +{ + struct super_block *sb = page->mapping->host->i_sb; + struct btrfs_fs_info *fs_info = btrfs_sb(sb); + btrfs_error(fs_info, -EIO, + "Error occured while writing out btree at %llu", start); + return -EIO; +} + static struct extent_io_ops btree_extent_io_ops = { .write_cache_pages_lock_hook = btree_lock_page_hook, .readpage_end_io_hook = btree_readpage_end_io_hook, @@ -3656,4 +3711,5 @@ static struct extent_io_ops btree_extent_io_ops = { .submit_bio_hook = btree_submit_bio_hook, /* note we're sharing with inode.c for the merge bio hook */ .merge_bio_hook = btrfs_merge_bio_hook, + .writepage_io_failed_hook = btree_writepage_io_failed_hook, }; -- cgit v1.2.3 From 914b20070b413ca10f832c45a58b2894990f065f Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Mon, 12 Mar 2012 16:05:50 +0100 Subject: btrfs: Fix busyloop in transaction_kthread() When a filesystem got aborted due do error, transaction_kthread() will busyloop. Fix it by going to sleep in that case as well. Maybe we should just stop transaction_kthread() when filesystem is aborted but that would be more complex. Signed-off-by: Jan Kara --- fs/btrfs/disk-io.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 16a0cada26c2..438993e3d832 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1640,8 +1640,10 @@ static int transaction_kthread(void *arg) u64 transid; unsigned long now; unsigned long delay; + bool cannot_commit; do { + cannot_commit = false; delay = HZ * 30; vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE); mutex_lock(&root->fs_info->transaction_kthread_mutex); @@ -1665,8 +1667,10 @@ static int transaction_kthread(void *arg) /* If the file system is aborted, this will always fail. */ trans = btrfs_join_transaction(root); - if (IS_ERR(trans)) + if (IS_ERR(trans)) { + cannot_commit = true; goto sleep; + } if (transid == trans->transid) { btrfs_commit_transaction(trans, root); } else { @@ -1679,7 +1683,8 @@ sleep: if (!try_to_freeze()) { set_current_state(TASK_INTERRUPTIBLE); if (!kthread_should_stop() && - !btrfs_transaction_blocked(root->fs_info)) + (!btrfs_transaction_blocked(root->fs_info) || + cannot_commit)) schedule_timeout(delay); __set_current_state(TASK_RUNNING); } -- cgit v1.2.3 From 727011e07cbdf87772fcc1999cccd15cc915eb62 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 6 Aug 2010 13:21:20 -0400 Subject: Btrfs: allow metadata blocks larger than the page size A few years ago the btrfs code to support blocks lager than the page size was disabled to fix a few corner cases in the page cache handling. This fixes the code to properly support large metadata blocks again. Since current kernels will crash early and often with larger metadata blocks, this adds an incompat bit so that older kernels can't mount it. This also does away with different blocksizes for nodes and leaves. You get a single block size for all tree blocks. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 196 ++++++++++++++++++++++++++--------------------------- 1 file changed, 95 insertions(+), 101 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 534266fe505f..68fc93e18db8 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -370,8 +370,7 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root, ret = read_extent_buffer_pages(io_tree, eb, start, WAIT_COMPLETE, btree_get_extent, mirror_num); - if (!ret && - !verify_parent_transid(io_tree, eb, parent_transid)) + if (!ret && !verify_parent_transid(io_tree, eb, parent_transid)) return ret; /* @@ -406,14 +405,11 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) u64 found_start; unsigned long len; struct extent_buffer *eb; - int ret; tree = &BTRFS_I(page->mapping->host)->io_tree; - if (page->private == EXTENT_PAGE_PRIVATE) { - WARN_ON(1); + if (page->private == EXTENT_PAGE_PRIVATE) goto out; - } if (!page->private) { WARN_ON(1); goto out; @@ -421,22 +417,14 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) len = page->private >> 2; WARN_ON(len == 0); - eb = alloc_extent_buffer(tree, start, len, page); - if (eb == NULL) { - WARN_ON(1); - goto out; - } - ret = btree_read_extent_buffer_pages(root, eb, start + PAGE_CACHE_SIZE, - btrfs_header_generation(eb)); - BUG_ON(ret); - WARN_ON(!btrfs_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN)); + eb = find_extent_buffer(tree, start, len); found_start = btrfs_header_bytenr(eb); if (found_start != start) { WARN_ON(1); goto err; } - if (eb->first_page != page) { + if (eb->pages[0] != page) { WARN_ON(1); goto err; } @@ -537,6 +525,41 @@ static noinline int check_leaf(struct btrfs_root *root, return 0; } +struct extent_buffer *find_eb_for_page(struct extent_io_tree *tree, + struct page *page, int max_walk) +{ + struct extent_buffer *eb; + u64 start = page_offset(page); + u64 target = start; + u64 min_start; + + if (start < max_walk) + min_start = 0; + else + min_start = start - max_walk; + + while (start >= min_start) { + eb = find_extent_buffer(tree, start, 0); + if (eb) { + /* + * we found an extent buffer and it contains our page + * horray! + */ + if (eb->start <= target && + eb->start + eb->len > target) + return eb; + + /* we found an extent buffer that wasn't for us */ + free_extent_buffer(eb); + return NULL; + } + if (start == 0) + break; + start -= PAGE_CACHE_SIZE; + } + return NULL; +} + static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, struct extent_state *state) { @@ -547,24 +570,25 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, struct extent_buffer *eb; struct btrfs_root *root = BTRFS_I(page->mapping->host)->root; int ret = 0; + int reads_done; - tree = &BTRFS_I(page->mapping->host)->io_tree; - if (page->private == EXTENT_PAGE_PRIVATE) - goto out; if (!page->private) goto out; + tree = &BTRFS_I(page->mapping->host)->io_tree; len = page->private >> 2; - WARN_ON(len == 0); - eb = alloc_extent_buffer(tree, start, len, page); - if (eb == NULL) { + eb = find_eb_for_page(tree, page, max(root->leafsize, root->nodesize)); + if (!eb) { ret = -EIO; goto out; } + reads_done = atomic_dec_and_test(&eb->pages_reading); + if (!reads_done) + goto err; found_start = btrfs_header_bytenr(eb); - if (found_start != start) { + if (found_start != eb->start) { printk_ratelimited(KERN_INFO "btrfs bad tree block start " "%llu %llu\n", (unsigned long long)found_start, @@ -572,13 +596,6 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, ret = -EIO; goto err; } - if (eb->first_page != page) { - printk(KERN_INFO "btrfs bad first page %lu %lu\n", - eb->first_page->index, page->index); - WARN_ON(1); - ret = -EIO; - goto err; - } if (check_tree_block_fsid(root, eb)) { printk_ratelimited(KERN_INFO "btrfs bad fsid on block %llu\n", (unsigned long long)eb->start); @@ -606,14 +623,14 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, ret = -EIO; } - end = min_t(u64, eb->len, PAGE_CACHE_SIZE); - end = eb->start + end - 1; err: if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) { clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags); btree_readahead_hook(root, eb, eb->start, ret); } + if (ret && eb) + clear_extent_buffer_uptodate(tree, eb, NULL); free_extent_buffer(eb); out: return ret; @@ -637,7 +654,7 @@ static int btree_io_failed_hook(struct bio *failed_bio, len = page->private >> 2; WARN_ON(len == 0); - eb = alloc_extent_buffer(tree, start, len, page); + eb = alloc_extent_buffer(tree, start, len); if (eb == NULL) goto out; @@ -896,28 +913,14 @@ static int btree_migratepage(struct address_space *mapping, static int btree_writepage(struct page *page, struct writeback_control *wbc) { struct extent_io_tree *tree; - struct btrfs_root *root = BTRFS_I(page->mapping->host)->root; - struct extent_buffer *eb; - int was_dirty; - tree = &BTRFS_I(page->mapping->host)->io_tree; + if (!(current->flags & PF_MEMALLOC)) { return extent_write_full_page(tree, page, btree_get_extent, wbc); } redirty_page_for_writepage(wbc, page); - eb = btrfs_find_tree_block(root, page_offset(page), PAGE_CACHE_SIZE); - WARN_ON(!eb); - - was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags); - if (!was_dirty) { - spin_lock(&root->fs_info->delalloc_lock); - root->fs_info->dirty_metadata_bytes += PAGE_CACHE_SIZE; - spin_unlock(&root->fs_info->delalloc_lock); - } - free_extent_buffer(eb); - unlock_page(page); return 0; } @@ -954,6 +957,8 @@ static int btree_releasepage(struct page *page, gfp_t gfp_flags) { struct extent_io_tree *tree; struct extent_map_tree *map; + struct extent_buffer *eb; + struct btrfs_root *root; int ret; if (PageWriteback(page) || PageDirty(page)) @@ -962,6 +967,13 @@ static int btree_releasepage(struct page *page, gfp_t gfp_flags) tree = &BTRFS_I(page->mapping->host)->io_tree; map = &BTRFS_I(page->mapping->host)->extent_tree; + root = BTRFS_I(page->mapping->host)->root; + if (page->private == EXTENT_PAGE_PRIVATE) { + eb = find_eb_for_page(tree, page, max(root->leafsize, root->nodesize)); + free_extent_buffer(eb); + if (eb) + return 0; + } /* * We need to mask out eg. __GFP_HIGHMEM and __GFP_DMA32 as we're doing * slab allocation from alloc_extent_state down the callchain where @@ -1074,20 +1086,20 @@ struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root, struct extent_buffer *eb; eb = alloc_extent_buffer(&BTRFS_I(btree_inode)->io_tree, - bytenr, blocksize, NULL); + bytenr, blocksize); return eb; } int btrfs_write_tree_block(struct extent_buffer *buf) { - return filemap_fdatawrite_range(buf->first_page->mapping, buf->start, + return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start, buf->start + buf->len - 1); } int btrfs_wait_tree_block_writeback(struct extent_buffer *buf) { - return filemap_fdatawait_range(buf->first_page->mapping, + return filemap_fdatawait_range(buf->pages[0]->mapping, buf->start, buf->start + buf->len - 1); } @@ -1513,41 +1525,6 @@ static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi) return 0; } -static int bio_ready_for_csum(struct bio *bio) -{ - u64 length = 0; - u64 buf_len = 0; - u64 start = 0; - struct page *page; - struct extent_io_tree *io_tree = NULL; - struct bio_vec *bvec; - int i; - int ret; - - bio_for_each_segment(bvec, bio, i) { - page = bvec->bv_page; - if (page->private == EXTENT_PAGE_PRIVATE) { - length += bvec->bv_len; - continue; - } - if (!page->private) { - length += bvec->bv_len; - continue; - } - length = bvec->bv_len; - buf_len = page->private >> 2; - start = page_offset(page) + bvec->bv_offset; - io_tree = &BTRFS_I(page->mapping->host)->io_tree; - } - /* are we fully contained in this bio? */ - if (buf_len <= length) - return 1; - - ret = extent_range_uptodate(io_tree, start + length, - start + buf_len - 1); - return ret; -} - /* * called by the kthread helper functions to finally call the bio end_io * functions. This is where read checksum verification actually happens @@ -1563,17 +1540,6 @@ static void end_workqueue_fn(struct btrfs_work *work) bio = end_io_wq->bio; fs_info = end_io_wq->info; - /* metadata bio reads are special because the whole tree block must - * be checksummed at once. This makes sure the entire block is in - * ram and up to date before trying to verify things. For - * blocksize <= pagesize, it is basically a noop - */ - if (!(bio->bi_rw & REQ_WRITE) && end_io_wq->metadata && - !bio_ready_for_csum(bio)) { - btrfs_queue_worker(&fs_info->endio_meta_workers, - &end_io_wq->work); - return; - } error = end_io_wq->error; bio->bi_private = end_io_wq->private; bio->bi_end_io = end_io_wq->end_io; @@ -2135,10 +2101,38 @@ int open_ctree(struct super_block *sb, goto fail_alloc; } + if (btrfs_super_leafsize(disk_super) != + btrfs_super_nodesize(disk_super)) { + printk(KERN_ERR "BTRFS: couldn't mount because metadata " + "blocksizes don't match. node %d leaf %d\n", + btrfs_super_nodesize(disk_super), + btrfs_super_leafsize(disk_super)); + err = -EINVAL; + goto fail_alloc; + } + if (btrfs_super_leafsize(disk_super) > BTRFS_MAX_METADATA_BLOCKSIZE) { + printk(KERN_ERR "BTRFS: couldn't mount because metadata " + "blocksize (%d) was too large\n", + btrfs_super_leafsize(disk_super)); + err = -EINVAL; + goto fail_alloc; + } + features = btrfs_super_incompat_flags(disk_super); features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF; if (tree_root->fs_info->compress_type & BTRFS_COMPRESS_LZO) features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO; + + /* + * flag our filesystem as having big metadata blocks if + * they are bigger than the page size + */ + if (btrfs_super_leafsize(disk_super) > PAGE_CACHE_SIZE) { + if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA)) + printk(KERN_INFO "btrfs flagging fs with big metadata feature\n"); + features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA; + } + btrfs_set_super_incompat_flags(disk_super, features); features = btrfs_super_compat_ro_flags(disk_super) & @@ -3122,7 +3116,7 @@ int close_ctree(struct btrfs_root *root) int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid) { int ret; - struct inode *btree_inode = buf->first_page->mapping->host; + struct inode *btree_inode = buf->pages[0]->mapping->host; ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf, NULL); @@ -3136,14 +3130,14 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid) int btrfs_set_buffer_uptodate(struct extent_buffer *buf) { - struct inode *btree_inode = buf->first_page->mapping->host; + struct inode *btree_inode = buf->pages[0]->mapping->host; return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf); } void btrfs_mark_buffer_dirty(struct extent_buffer *buf) { - struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; + struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root; u64 transid = btrfs_header_generation(buf); struct inode *btree_inode = root->fs_info->btree_inode; int was_dirty; @@ -3212,7 +3206,7 @@ void __btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr) int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid) { - struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root; + struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root; int ret; ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid); if (ret == 0) -- cgit v1.2.3 From 4f2de97acee6532b36dd6e995b858343771ad126 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 7 Mar 2012 16:20:05 -0500 Subject: Btrfs: set page->private to the eb We spend a lot of time looking up extent buffers from pages when we could just store the pointer to the eb the page is associated with in page->private. This patch does just that, and it makes things a little simpler and reduces a bit of CPU overhead involved with doing metadata IO. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/disk-io.c | 91 ++++++++++++++---------------------------------------- 1 file changed, 23 insertions(+), 68 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 68fc93e18db8..bc88649cffb7 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -403,39 +403,28 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) struct extent_io_tree *tree; u64 start = (u64)page->index << PAGE_CACHE_SHIFT; u64 found_start; - unsigned long len; struct extent_buffer *eb; tree = &BTRFS_I(page->mapping->host)->io_tree; - if (page->private == EXTENT_PAGE_PRIVATE) - goto out; - if (!page->private) { - WARN_ON(1); - goto out; - } - len = page->private >> 2; - WARN_ON(len == 0); - - eb = find_extent_buffer(tree, start, len); + eb = (struct extent_buffer *)page->private; + if (page != eb->pages[0]) + return 0; found_start = btrfs_header_bytenr(eb); if (found_start != start) { WARN_ON(1); - goto err; + return 0; } if (eb->pages[0] != page) { WARN_ON(1); - goto err; + return 0; } if (!PageUptodate(page)) { WARN_ON(1); - goto err; + return 0; } csum_tree_block(root, eb, 0); -err: - free_extent_buffer(eb); -out: return 0; } @@ -566,7 +555,6 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, struct extent_io_tree *tree; u64 found_start; int found_level; - unsigned long len; struct extent_buffer *eb; struct btrfs_root *root = BTRFS_I(page->mapping->host)->root; int ret = 0; @@ -576,13 +564,8 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, goto out; tree = &BTRFS_I(page->mapping->host)->io_tree; - len = page->private >> 2; + eb = (struct extent_buffer *)page->private; - eb = find_eb_for_page(tree, page, max(root->leafsize, root->nodesize)); - if (!eb) { - ret = -EIO; - goto out; - } reads_done = atomic_dec_and_test(&eb->pages_reading); if (!reads_done) goto err; @@ -631,7 +614,6 @@ err: if (ret && eb) clear_extent_buffer_uptodate(tree, eb, NULL); - free_extent_buffer(eb); out: return ret; } @@ -640,31 +622,17 @@ static int btree_io_failed_hook(struct bio *failed_bio, struct page *page, u64 start, u64 end, int mirror_num, struct extent_state *state) { - struct extent_io_tree *tree; - unsigned long len; struct extent_buffer *eb; struct btrfs_root *root = BTRFS_I(page->mapping->host)->root; - tree = &BTRFS_I(page->mapping->host)->io_tree; - if (page->private == EXTENT_PAGE_PRIVATE) - goto out; - if (!page->private) - goto out; - - len = page->private >> 2; - WARN_ON(len == 0); - - eb = alloc_extent_buffer(tree, start, len); - if (eb == NULL) - goto out; + eb = (struct extent_buffer *)page->private; + if (page != eb->pages[0]) + return -EIO; if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) { clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags); btree_readahead_hook(root, eb, eb->start, -EIO); } - free_extent_buffer(eb); - -out: return -EIO; /* we fixed nothing */ } @@ -955,10 +923,8 @@ static int btree_readpage(struct file *file, struct page *page) static int btree_releasepage(struct page *page, gfp_t gfp_flags) { - struct extent_io_tree *tree; struct extent_map_tree *map; - struct extent_buffer *eb; - struct btrfs_root *root; + struct extent_io_tree *tree; int ret; if (PageWriteback(page) || PageDirty(page)) @@ -967,13 +933,6 @@ static int btree_releasepage(struct page *page, gfp_t gfp_flags) tree = &BTRFS_I(page->mapping->host)->io_tree; map = &BTRFS_I(page->mapping->host)->extent_tree; - root = BTRFS_I(page->mapping->host)->root; - if (page->private == EXTENT_PAGE_PRIVATE) { - eb = find_eb_for_page(tree, page, max(root->leafsize, root->nodesize)); - free_extent_buffer(eb); - if (eb) - return 0; - } /* * We need to mask out eg. __GFP_HIGHMEM and __GFP_DMA32 as we're doing * slab allocation from alloc_extent_state down the callchain where @@ -985,14 +944,7 @@ static int btree_releasepage(struct page *page, gfp_t gfp_flags) if (!ret) return 0; - ret = try_release_extent_buffer(tree, page); - if (ret == 1) { - ClearPagePrivate(page); - set_page_private(page, 0); - page_cache_release(page); - } - - return ret; + return try_release_extent_buffer(tree, page); } static void btree_invalidatepage(struct page *page, unsigned long offset) @@ -3219,17 +3171,21 @@ static int btree_lock_page_hook(struct page *page, void *data, { struct inode *inode = page->mapping->host; struct btrfs_root *root = BTRFS_I(inode)->root; - struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_buffer *eb; - unsigned long len; - u64 bytenr = page_offset(page); - if (page->private == EXTENT_PAGE_PRIVATE) + /* + * We culled this eb but the page is still hanging out on the mapping, + * carry on. + */ + if (!PagePrivate(page)) goto out; - len = page->private >> 2; - eb = find_extent_buffer(io_tree, bytenr, len); - if (!eb) + eb = (struct extent_buffer *)page->private; + if (!eb) { + WARN_ON(1); + goto out; + } + if (page != eb->pages[0]) goto out; if (!btrfs_try_tree_write_lock(eb)) { @@ -3248,7 +3204,6 @@ static int btree_lock_page_hook(struct page *page, void *data, } btrfs_tree_unlock(eb); - free_extent_buffer(eb); out: if (!trylock_page(page)) { flush_fn(data); -- cgit v1.2.3 From 3083ee2e18b701122a3b841db83448543a87a583 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 9 Mar 2012 16:01:49 -0500 Subject: Btrfs: introduce free_extent_buffer_stale Because btrfs cow's we can end up with extent buffers that are no longer necessary just sitting around in memory. So instead of evicting these pages, we could end up evicting things we actually care about. Thus we have free_extent_buffer_stale for use when we are freeing tree blocks. This will make it so that the ref for the eb being in the radix tree is dropped as soon as possible and then is freed when the refcount hits 0 instead of waiting to be released by releasepage. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/disk-io.c | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index bc88649cffb7..0ba055e03eb8 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -923,16 +923,8 @@ static int btree_readpage(struct file *file, struct page *page) static int btree_releasepage(struct page *page, gfp_t gfp_flags) { - struct extent_map_tree *map; - struct extent_io_tree *tree; - int ret; - if (PageWriteback(page) || PageDirty(page)) return 0; - - tree = &BTRFS_I(page->mapping->host)->io_tree; - map = &BTRFS_I(page->mapping->host)->extent_tree; - /* * We need to mask out eg. __GFP_HIGHMEM and __GFP_DMA32 as we're doing * slab allocation from alloc_extent_state down the callchain where @@ -940,11 +932,7 @@ static int btree_releasepage(struct page *page, gfp_t gfp_flags) */ gfp_flags &= ~GFP_SLAB_BUG_MASK; - ret = try_release_extent_state(map, tree, page, gfp_flags); - if (!ret) - return 0; - - return try_release_extent_buffer(tree, page); + return try_release_extent_buffer(page, gfp_flags); } static void btree_invalidatepage(struct page *page, unsigned long offset) -- cgit v1.2.3 From 0b32f4bbb423f02acee6d43cd442f5f0775db7e0 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 13 Mar 2012 09:38:00 -0400 Subject: Btrfs: ensure an entire eb is written at once This patch simplifies how we track our extent buffers. Previously we could exit writepages with only having written half of an extent buffer, which meant we had to track the state of the pages and the state of the extent buffers differently. Now we only read in entire extent buffers and write out entire extent buffers, this allows us to simply set bits in our bflags to indicate the state of the eb and we no longer have to do things like track uptodate with our iotree. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 76 +++++++++++++++++++++++++----------------------------- 1 file changed, 35 insertions(+), 41 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 0ba055e03eb8..c54aec87e89d 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -333,7 +333,7 @@ static int verify_parent_transid(struct extent_io_tree *io_tree, lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1, 0, &cached_state, GFP_NOFS); - if (extent_buffer_uptodate(io_tree, eb, cached_state) && + if (extent_buffer_uptodate(eb) && btrfs_header_generation(eb) == parent_transid) { ret = 0; goto out; @@ -344,7 +344,7 @@ static int verify_parent_transid(struct extent_io_tree *io_tree, (unsigned long long)parent_transid, (unsigned long long)btrfs_header_generation(eb)); ret = 1; - clear_extent_buffer_uptodate(io_tree, eb, &cached_state); + clear_extent_buffer_uptodate(eb); out: unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1, &cached_state, GFP_NOFS); @@ -566,7 +566,12 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, tree = &BTRFS_I(page->mapping->host)->io_tree; eb = (struct extent_buffer *)page->private; - reads_done = atomic_dec_and_test(&eb->pages_reading); + /* the pending IO might have been the only thing that kept this buffer + * in memory. Make sure we have a ref for all this other checks + */ + extent_buffer_get(eb); + + reads_done = atomic_dec_and_test(&eb->io_pages); if (!reads_done) goto err; @@ -606,14 +611,17 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, ret = -EIO; } + if (!ret) + set_extent_buffer_uptodate(eb); err: if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) { clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags); btree_readahead_hook(root, eb, eb->start, ret); } - if (ret && eb) - clear_extent_buffer_uptodate(tree, eb, NULL); + if (ret) + clear_extent_buffer_uptodate(eb); + free_extent_buffer(eb); out: return ret; } @@ -878,20 +886,6 @@ static int btree_migratepage(struct address_space *mapping, } #endif -static int btree_writepage(struct page *page, struct writeback_control *wbc) -{ - struct extent_io_tree *tree; - tree = &BTRFS_I(page->mapping->host)->io_tree; - - if (!(current->flags & PF_MEMALLOC)) { - return extent_write_full_page(tree, page, - btree_get_extent, wbc); - } - - redirty_page_for_writepage(wbc, page); - unlock_page(page); - return 0; -} static int btree_writepages(struct address_space *mapping, struct writeback_control *wbc) @@ -911,7 +905,7 @@ static int btree_writepages(struct address_space *mapping, if (num_dirty < thresh) return 0; } - return extent_writepages(tree, mapping, btree_get_extent, wbc); + return btree_write_cache_pages(mapping, wbc); } static int btree_readpage(struct file *file, struct page *page) @@ -950,15 +944,28 @@ static void btree_invalidatepage(struct page *page, unsigned long offset) } } +static int btree_set_page_dirty(struct page *page) +{ + struct extent_buffer *eb; + + BUG_ON(!PagePrivate(page)); + eb = (struct extent_buffer *)page->private; + BUG_ON(!eb); + BUG_ON(!test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)); + BUG_ON(!atomic_read(&eb->refs)); + btrfs_assert_tree_locked(eb); + return __set_page_dirty_nobuffers(page); +} + static const struct address_space_operations btree_aops = { .readpage = btree_readpage, - .writepage = btree_writepage, .writepages = btree_writepages, .releasepage = btree_releasepage, .invalidatepage = btree_invalidatepage, #ifdef CONFIG_MIGRATION .migratepage = btree_migratepage, #endif + .set_page_dirty = btree_set_page_dirty, }; int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize, @@ -1001,7 +1008,7 @@ int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr, u32 blocksize, if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) { free_extent_buffer(buf); return -EIO; - } else if (extent_buffer_uptodate(io_tree, buf, NULL)) { + } else if (extent_buffer_uptodate(buf)) { *eb = buf; } else { free_extent_buffer(buf); @@ -1054,9 +1061,6 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr, return NULL; ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid); - - if (ret == 0) - set_bit(EXTENT_BUFFER_UPTODATE, &buf->bflags); return buf; } @@ -1064,7 +1068,6 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr, int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf) { - struct inode *btree_inode = root->fs_info->btree_inode; if (btrfs_header_generation(buf) == root->fs_info->running_transaction->transid) { btrfs_assert_tree_locked(buf); @@ -1080,8 +1083,7 @@ int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, /* ugh, clear_extent_buffer_dirty needs to lock the page */ btrfs_set_lock_blocking(buf); - clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, - buf); + clear_extent_buffer_dirty(buf); } return 0; } @@ -1948,6 +1950,7 @@ int open_ctree(struct super_block *sb, RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node); extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree, fs_info->btree_inode->i_mapping); + BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0; extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree); BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops; @@ -3058,8 +3061,7 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid) int ret; struct inode *btree_inode = buf->pages[0]->mapping->host; - ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf, - NULL); + ret = extent_buffer_uptodate(buf); if (!ret) return ret; @@ -3070,16 +3072,13 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid) int btrfs_set_buffer_uptodate(struct extent_buffer *buf) { - struct inode *btree_inode = buf->pages[0]->mapping->host; - return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, - buf); + return set_extent_buffer_uptodate(buf); } void btrfs_mark_buffer_dirty(struct extent_buffer *buf) { struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root; u64 transid = btrfs_header_generation(buf); - struct inode *btree_inode = root->fs_info->btree_inode; int was_dirty; btrfs_assert_tree_locked(buf); @@ -3091,8 +3090,7 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf) (unsigned long long)root->fs_info->generation); WARN_ON(1); } - was_dirty = set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree, - buf); + was_dirty = set_extent_buffer_dirty(buf); if (!was_dirty) { spin_lock(&root->fs_info->delalloc_lock); root->fs_info->dirty_metadata_bytes += buf->len; @@ -3147,11 +3145,7 @@ void __btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr) int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid) { struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root; - int ret; - ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid); - if (ret == 0) - set_bit(EXTENT_BUFFER_UPTODATE, &buf->bflags); - return ret; + return btree_read_extent_buffer_pages(root, buf, 0, parent_transid); } static int btree_lock_page_hook(struct page *page, void *data, -- cgit v1.2.3 From f3f266ab1bfe4770375d24fa8e72a03278e9450a Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 23 Mar 2012 10:22:46 -0400 Subject: Btrfs: don't use threaded IO completion helpers for metadata writes The metadata write IO completion code is now simple enough that we don't need the threaded helpers anymore. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index c54aec87e89d..53c5ea702799 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -840,15 +840,15 @@ static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, { int ret; - ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info, - bio, 1); - BUG_ON(ret); - if (!(rw & REQ_WRITE)) { + /* * called for a read, do the setup so that checksum validation * can happen in the async kernel threads */ + ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info, + bio, 1); + BUG_ON(ret); return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 0); } -- cgit v1.2.3 From ea466794084f55d8fcc100711cf17923bf57e962 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 26 Mar 2012 21:57:36 -0400 Subject: Btrfs: deal with read errors on extent buffers differently Since we need to read and write extent buffers in their entirety we can't use the normal bio_readpage_error stuff since it only works on a per page basis. So instead make it so that if we see an io error in endio we just mark the eb as having an IO error and then in btree_read_extent_buffer_pages we will manually try other mirrors and then overwrite the bad mirror if we find a good copy. This works with larger than page size blocks. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 43 +++++++++++++++++++++++++++++-------------- 1 file changed, 29 insertions(+), 14 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 53c5ea702799..6107b6958413 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -360,9 +360,11 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root, u64 start, u64 parent_transid) { struct extent_io_tree *io_tree; + int failed = 0; int ret; int num_copies = 0; int mirror_num = 0; + int failed_mirror = 0; clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags); io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree; @@ -371,7 +373,7 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root, WAIT_COMPLETE, btree_get_extent, mirror_num); if (!ret && !verify_parent_transid(io_tree, eb, parent_transid)) - return ret; + break; /* * This buffer's crc is fine, but its contents are corrupted, so @@ -379,18 +381,31 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root, * any less wrong. */ if (test_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags)) - return ret; + break; + + if (!failed_mirror) { + failed = 1; + printk(KERN_ERR "failed mirror was %d\n", eb->failed_mirror); + failed_mirror = eb->failed_mirror; + } num_copies = btrfs_num_copies(&root->fs_info->mapping_tree, eb->start, eb->len); if (num_copies == 1) - return ret; + break; mirror_num++; + if (mirror_num == failed_mirror) + mirror_num++; + if (mirror_num > num_copies) - return ret; + break; } - return -EIO; + + if (failed && !ret) + repair_eb_io_failure(root, eb, failed_mirror); + + return ret; } /* @@ -575,6 +590,11 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, if (!reads_done) goto err; + if (test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) { + ret = -EIO; + goto err; + } + found_start = btrfs_header_bytenr(eb); if (found_start != eb->start) { printk_ratelimited(KERN_INFO "btrfs bad tree block start " @@ -626,21 +646,16 @@ out: return ret; } -static int btree_io_failed_hook(struct bio *failed_bio, - struct page *page, u64 start, u64 end, - int mirror_num, struct extent_state *state) +static int btree_io_failed_hook(struct page *page, int failed_mirror) { struct extent_buffer *eb; struct btrfs_root *root = BTRFS_I(page->mapping->host)->root; eb = (struct extent_buffer *)page->private; - if (page != eb->pages[0]) - return -EIO; - - if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) { - clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags); + set_bit(EXTENT_BUFFER_IOERR, &eb->bflags); + eb->failed_mirror = failed_mirror; + if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) btree_readahead_hook(root, eb, eb->start, -EIO); - } return -EIO; /* we fixed nothing */ } -- cgit v1.2.3 From fcd1f065daca593badb7f99d473639cf3b551795 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Tue, 6 Mar 2012 00:06:18 +0100 Subject: Btrfs: enhance superblock sanity checks Validate checksum algorithm during mount and prevent BUG_ON later in btrfs_super_csum_size. Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 23 ++++++++++++++++++----- 1 file changed, 18 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 438993e3d832..38c2ee1ca0d5 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -48,7 +48,7 @@ static struct extent_io_ops btree_extent_io_ops; static void end_workqueue_fn(struct btrfs_work *work); static void free_fs_root(struct btrfs_root *root); -static void btrfs_check_super_valid(struct btrfs_fs_info *fs_info, +static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info, int read_only); static void btrfs_destroy_ordered_operations(struct btrfs_root *root); static void btrfs_destroy_ordered_extents(struct btrfs_root *root); @@ -2135,7 +2135,12 @@ int open_ctree(struct super_block *sb, /* check FS state, whether FS is broken. */ fs_info->fs_state |= btrfs_super_flags(disk_super); - btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY); + ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY); + if (ret) { + printk(KERN_ERR "btrfs: superblock contains fatal errors\n"); + err = ret; + goto fail_alloc; + } /* * run through our array of backup supers and setup @@ -3324,15 +3329,23 @@ out: return 0; } -static void btrfs_check_super_valid(struct btrfs_fs_info *fs_info, +static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info, int read_only) { + if (btrfs_super_csum_type(fs_info->super_copy) >= ARRAY_SIZE(btrfs_csum_sizes)) { + printk(KERN_ERR "btrfs: unsupported checksum algorithm\n"); + return -EINVAL; + } + if (read_only) - return; + return 0; - if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) + if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { printk(KERN_WARNING "warning: mount fs with errors, " "running btrfsck is recommended\n"); + } + + return 0; } int btrfs_error_commit_super(struct btrfs_root *root) -- cgit v1.2.3 From 65139ed99234d8505948cdb7a835452eb5c191f9 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Fri, 17 Feb 2012 12:26:09 +0100 Subject: btrfs: disallow unequal data/metadata blocksize for mixed block groups With support for bigger metadata blocks, we must avoid mounting a filesystem with different block size for mixed block groups, this causes corruption (found by xfstests/083). Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 38c2ee1ca0d5..fe087847c8e7 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2302,6 +2302,14 @@ int open_ctree(struct super_block *sb, goto fail_sb_buffer; } + if ((features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) && + (leafsize != nodesize || sectorsize != nodesize)) { + printk(KERN_WARNING "btrfs: unequal leaf/node/sector sizes " + "are not allowed for mixed block groups on %s\n", + sb->s_id); + goto fail_sb_buffer; + } + mutex_lock(&fs_info->chunk_mutex); ret = btrfs_read_sys_array(tree_root); mutex_unlock(&fs_info->chunk_mutex); -- cgit v1.2.3 From 3c4bb26b213e618473e486776483a5bad15ba6da Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Tue, 27 Mar 2012 18:56:56 -0400 Subject: Btrfs: flush out and clean up any block device pages during mount Btrfs puts the filesystem metadata into its own address space, and somehow the block device address space isn't getting onto disk properly before a mount. The end result is that a loop of mkfs and mounting the filesystem will sometimes find stale or incorrect data. This commit should fix it by sprinkling fdatawrites and invalidate_bdev calls around. This is a short term measure to make sure it is fixed. The block devices really should be flushed and cleaned up higher in the stack. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 7b55eee15a51..fd43f6b2f438 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2031,6 +2031,7 @@ int open_ctree(struct super_block *sb, __setup_root(4096, 4096, 4096, 4096, tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID); + invalidate_bdev(fs_devices->latest_bdev); bh = btrfs_read_dev_super(fs_devices->latest_bdev); if (!bh) { err = -EINVAL; -- cgit v1.2.3 From bc3f116fec194f1d7329b160c266fe16b9266a1e Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 29 Mar 2012 17:02:47 -0400 Subject: Btrfs: update the checks for mixed block groups with big metadata blocks Dave Sterba had put in patches to look for mixed data/metadata groups with metadata bigger than 4KB. But these ended up in the wrong place and it wasn't testing the feature flag correctly. This updates the tests to make sure our sizes are matching Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 29 +++++++++++++++++------------ 1 file changed, 17 insertions(+), 12 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index fd43f6b2f438..20196f411206 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2120,6 +2120,23 @@ int open_ctree(struct super_block *sb, features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA; } + nodesize = btrfs_super_nodesize(disk_super); + leafsize = btrfs_super_leafsize(disk_super); + sectorsize = btrfs_super_sectorsize(disk_super); + stripesize = btrfs_super_stripesize(disk_super); + + /* + * mixed block groups end up with duplicate but slightly offset + * extent buffers for the same range. It leads to corruptions + */ + if ((features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) && + (sectorsize != leafsize)) { + printk(KERN_WARNING "btrfs: unequal leaf/node/sector sizes " + "are not allowed for mixed block groups on %s\n", + sb->s_id); + goto fail_alloc; + } + btrfs_set_super_incompat_flags(disk_super, features); features = btrfs_super_compat_ro_flags(disk_super) & @@ -2223,10 +2240,6 @@ int open_ctree(struct super_block *sb, fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages, 4 * 1024 * 1024 / PAGE_CACHE_SIZE); - nodesize = btrfs_super_nodesize(disk_super); - leafsize = btrfs_super_leafsize(disk_super); - sectorsize = btrfs_super_sectorsize(disk_super); - stripesize = btrfs_super_stripesize(disk_super); tree_root->nodesize = nodesize; tree_root->leafsize = leafsize; tree_root->sectorsize = sectorsize; @@ -2247,14 +2260,6 @@ int open_ctree(struct super_block *sb, goto fail_sb_buffer; } - if ((features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) && - (leafsize != nodesize || sectorsize != nodesize)) { - printk(KERN_WARNING "btrfs: unequal leaf/node/sector sizes " - "are not allowed for mixed block groups on %s\n", - sb->s_id); - goto fail_sb_buffer; - } - mutex_lock(&fs_info->chunk_mutex); ret = btrfs_read_sys_array(tree_root); mutex_unlock(&fs_info->chunk_mutex); -- cgit v1.2.3