From 5581a51a59a1f5f51ac3d4bacafb738d35e0350b Mon Sep 17 00:00:00 2001 From: Jan Schmidt Date: Wed, 16 May 2012 17:04:52 +0200 Subject: Btrfs: don't set for_cow parameter for tree block functions Three callers of btrfs_free_tree_block or btrfs_alloc_tree_block passed parameter for_cow = 1. In fact, these two functions should never mark their tree modification operations as for_cow, because they can change the number of blocks referenced by a tree. Hence, we remove the extra for_cow parameter from these functions and make them pass a zero down. Signed-off-by: Jan Schmidt --- 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 a7ffc88a7db..f4330749251 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1252,7 +1252,7 @@ static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans, leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0, BTRFS_TREE_LOG_OBJECTID, NULL, - 0, 0, 0, 0); + 0, 0, 0); if (IS_ERR(leaf)) { kfree(root); return ERR_CAST(leaf); -- cgit v1.2.3-70-g09d2 From f29021b29a85701c08afadfd51d87163fb078059 Mon Sep 17 00:00:00 2001 From: Jan Schmidt Date: Wed, 16 May 2012 17:55:38 +0200 Subject: Btrfs: add tree mod log to fs_info Signed-off-by: Jan Schmidt --- fs/btrfs/ctree.h | 9 +++++++++ fs/btrfs/disk-io.c | 5 +++++ 2 files changed, 14 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index e0da6dbeb57..01639e10004 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1129,6 +1129,15 @@ struct btrfs_fs_info { spinlock_t delayed_iput_lock; struct list_head delayed_iputs; + /* this protects tree_mod_seq_list */ + spinlock_t tree_mod_seq_lock; + atomic_t tree_mod_seq; + struct list_head tree_mod_seq_list; + + /* this protects tree_mod_log */ + rwlock_t tree_mod_log_lock; + struct rb_root tree_mod_log; + atomic_t nr_async_submits; atomic_t async_submit_draining; atomic_t nr_async_bios; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index f4330749251..f51ad8477f1 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1914,11 +1914,14 @@ int open_ctree(struct super_block *sb, spin_lock_init(&fs_info->delayed_iput_lock); spin_lock_init(&fs_info->defrag_inodes_lock); spin_lock_init(&fs_info->free_chunk_lock); + spin_lock_init(&fs_info->tree_mod_seq_lock); + rwlock_init(&fs_info->tree_mod_log_lock); mutex_init(&fs_info->reloc_mutex); init_completion(&fs_info->kobj_unregister); INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots); INIT_LIST_HEAD(&fs_info->space_info); + INIT_LIST_HEAD(&fs_info->tree_mod_seq_list); btrfs_mapping_init(&fs_info->mapping_tree); btrfs_init_block_rsv(&fs_info->global_block_rsv); btrfs_init_block_rsv(&fs_info->delalloc_block_rsv); @@ -1931,12 +1934,14 @@ int open_ctree(struct super_block *sb, atomic_set(&fs_info->async_submit_draining, 0); atomic_set(&fs_info->nr_async_bios, 0); atomic_set(&fs_info->defrag_running, 0); + atomic_set(&fs_info->tree_mod_seq, 0); fs_info->sb = sb; fs_info->max_inline = 8192 * 1024; fs_info->metadata_ratio = 0; fs_info->defrag_inodes = RB_ROOT; fs_info->trans_no_join = 0; fs_info->free_chunk_space = 0; + fs_info->tree_mod_log = RB_ROOT; /* readahead state */ INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_WAIT); -- cgit v1.2.3-70-g09d2 From 5fd02043553b02867b29de1ac9fff2ec16b84def Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 2 May 2012 14:00:54 -0400 Subject: Btrfs: finish ordered extents in their own thread We noticed that the ordered extent completion doesn't really rely on having a page and that it could be done independantly of ending the writeback on a page. This patch makes us not do the threaded endio stuff for normal buffered writes and direct writes so we can end page writeback as soon as possible (in irq context) and only start threads to do the ordered work when it is actually done. Compression needs to be reworked some to take advantage of this as well, but atm it has to do a find_get_page in its endio handler so it must be done in its own thread. This makes direct writes quite a bit faster. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/disk-io.c | 12 --- fs/btrfs/extent_io.c | 15 +--- fs/btrfs/extent_io.h | 5 +- fs/btrfs/free-space-cache.c | 4 +- fs/btrfs/inode.c | 177 +++++++++++++++++++------------------------- fs/btrfs/ordered-data.c | 129 ++++++++++++++++++-------------- fs/btrfs/ordered-data.h | 13 +++- 7 files changed, 164 insertions(+), 191 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index a7ffc88a7db..19f5b450f40 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -3671,17 +3671,6 @@ 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, @@ -3689,5 +3678,4 @@ 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, }; diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 836fc37a437..7af93435cee 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -1172,9 +1172,8 @@ int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, cached_state, mask); } -static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start, - u64 end, struct extent_state **cached_state, - gfp_t mask) +int clear_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, + struct extent_state **cached_state, gfp_t mask) { return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0, cached_state, mask); @@ -2221,17 +2220,7 @@ int end_extent_writepage(struct page *page, int err, u64 start, u64 end) uptodate = 0; } - if (!uptodate && tree->ops && - tree->ops->writepage_io_failed_hook) { - ret = tree->ops->writepage_io_failed_hook(NULL, page, - start, end, NULL); - /* Writeback already completed */ - if (ret == 0) - return 1; - } - if (!uptodate) { - clear_extent_uptodate(tree, start, end, NULL, GFP_NOFS); ClearPageUptodate(page); SetPageError(page); } diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h index b516c3b8dec..4d8124b6457 100644 --- a/fs/btrfs/extent_io.h +++ b/fs/btrfs/extent_io.h @@ -75,9 +75,6 @@ struct extent_io_ops { unsigned long bio_flags); int (*readpage_io_hook)(struct page *page, u64 start, u64 end); int (*readpage_io_failed_hook)(struct page *page, int failed_mirror); - int (*writepage_io_failed_hook)(struct bio *bio, struct page *page, - u64 start, u64 end, - struct extent_state *state); int (*readpage_end_io_hook)(struct page *page, u64 start, u64 end, struct extent_state *state, int mirror); int (*writepage_end_io_hook)(struct page *page, u64 start, u64 end, @@ -225,6 +222,8 @@ int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, struct extent_state **cached_state, gfp_t mask); int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, struct extent_state **cached_state, gfp_t mask); +int clear_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, + struct extent_state **cached_state, gfp_t mask); int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 202008ec367..cecf8df6248 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -972,9 +972,7 @@ int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, goto out; - ret = filemap_write_and_wait(inode->i_mapping); - if (ret) - goto out; + btrfs_wait_ordered_range(inode, 0, (u64)-1); key.objectid = BTRFS_FREE_SPACE_OBJECTID; key.offset = offset; diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 41a62e6954c..9a1b96fd672 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -89,7 +89,7 @@ static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = { static int btrfs_setsize(struct inode *inode, loff_t newsize); static int btrfs_truncate(struct inode *inode); -static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end); +static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent); static noinline int cow_file_range(struct inode *inode, struct page *locked_page, u64 start, u64 end, int *page_started, @@ -1572,11 +1572,11 @@ static int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, if (btrfs_is_free_space_inode(root, inode)) metadata = 2; - ret = btrfs_bio_wq_end_io(root->fs_info, bio, metadata); - if (ret) - return ret; - if (!(rw & REQ_WRITE)) { + ret = btrfs_bio_wq_end_io(root->fs_info, bio, metadata); + if (ret) + return ret; + if (bio_flags & EXTENT_BIO_COMPRESSED) { return btrfs_submit_compressed_read(inode, bio, mirror_num, bio_flags); @@ -1815,25 +1815,24 @@ out: * an ordered extent if the range of bytes in the file it covers are * fully written. */ -static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) +static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent) { + struct inode *inode = ordered_extent->inode; struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_trans_handle *trans = NULL; - struct btrfs_ordered_extent *ordered_extent = NULL; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_state *cached_state = NULL; int compress_type = 0; int ret; bool nolock; - ret = btrfs_dec_test_ordered_pending(inode, &ordered_extent, start, - end - start + 1); - if (!ret) - return 0; - BUG_ON(!ordered_extent); /* Logic error */ - nolock = btrfs_is_free_space_inode(root, inode); + if (test_bit(BTRFS_ORDERED_IOERR, &ordered_extent->flags)) { + ret = -EIO; + goto out; + } + if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) { BUG_ON(!list_empty(&ordered_extent->list)); /* Logic error */ ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent); @@ -1889,12 +1888,10 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) ordered_extent->file_offset, ordered_extent->len); } - unlock_extent_cached(io_tree, ordered_extent->file_offset, - ordered_extent->file_offset + - ordered_extent->len - 1, &cached_state, GFP_NOFS); + if (ret < 0) { btrfs_abort_transaction(trans, root, ret); - goto out; + goto out_unlock; } add_pending_csums(trans, inode, ordered_extent->file_offset, @@ -1905,10 +1902,14 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) ret = btrfs_update_inode_fallback(trans, root, inode); if (ret) { /* -ENOMEM or corruption */ btrfs_abort_transaction(trans, root, ret); - goto out; + goto out_unlock; } } ret = 0; +out_unlock: + unlock_extent_cached(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + + ordered_extent->len - 1, &cached_state, GFP_NOFS); out: if (root != root->fs_info->tree_root) btrfs_delalloc_release_metadata(inode, ordered_extent->len); @@ -1919,26 +1920,57 @@ out: btrfs_end_transaction(trans, root); } + if (ret) + clear_extent_uptodate(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + + ordered_extent->len - 1, NULL, GFP_NOFS); + + /* + * This needs to be dont to make sure anybody waiting knows we are done + * upating everything for this ordered extent. + */ + btrfs_remove_ordered_extent(inode, ordered_extent); + /* once for us */ btrfs_put_ordered_extent(ordered_extent); /* once for the tree */ btrfs_put_ordered_extent(ordered_extent); - return 0; -out_unlock: - unlock_extent_cached(io_tree, ordered_extent->file_offset, - ordered_extent->file_offset + - ordered_extent->len - 1, &cached_state, GFP_NOFS); - goto out; + return ret; +} + +static void finish_ordered_fn(struct btrfs_work *work) +{ + struct btrfs_ordered_extent *ordered_extent; + ordered_extent = container_of(work, struct btrfs_ordered_extent, work); + btrfs_finish_ordered_io(ordered_extent); } static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end, struct extent_state *state, int uptodate) { + struct inode *inode = page->mapping->host; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_ordered_extent *ordered_extent = NULL; + struct btrfs_workers *workers; + trace_btrfs_writepage_end_io_hook(page, start, end, uptodate); ClearPagePrivate2(page); - return btrfs_finish_ordered_io(page->mapping->host, start, end); + if (!btrfs_dec_test_ordered_pending(inode, &ordered_extent, start, + end - start + 1, uptodate)) + return 0; + + ordered_extent->work.func = finish_ordered_fn; + ordered_extent->work.flags = 0; + + if (btrfs_is_free_space_inode(root, inode)) + workers = &root->fs_info->endio_freespace_worker; + else + workers = &root->fs_info->endio_write_workers; + btrfs_queue_worker(workers, &ordered_extent->work); + + return 0; } /* @@ -5909,9 +5941,7 @@ static void btrfs_endio_direct_write(struct bio *bio, int err) struct btrfs_dio_private *dip = bio->bi_private; struct inode *inode = dip->inode; struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_trans_handle *trans; struct btrfs_ordered_extent *ordered = NULL; - struct extent_state *cached_state = NULL; u64 ordered_offset = dip->logical_offset; u64 ordered_bytes = dip->bytes; int ret; @@ -5921,73 +5951,14 @@ static void btrfs_endio_direct_write(struct bio *bio, int err) again: ret = btrfs_dec_test_first_ordered_pending(inode, &ordered, &ordered_offset, - ordered_bytes); + ordered_bytes, !err); if (!ret) goto out_test; - BUG_ON(!ordered); - - trans = btrfs_join_transaction(root); - if (IS_ERR(trans)) { - err = -ENOMEM; - goto out; - } - trans->block_rsv = &root->fs_info->delalloc_block_rsv; - - if (test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags)) { - ret = btrfs_ordered_update_i_size(inode, 0, ordered); - if (!ret) - err = btrfs_update_inode_fallback(trans, root, inode); - goto out; - } - - lock_extent_bits(&BTRFS_I(inode)->io_tree, ordered->file_offset, - ordered->file_offset + ordered->len - 1, 0, - &cached_state); - - if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags)) { - ret = btrfs_mark_extent_written(trans, inode, - ordered->file_offset, - ordered->file_offset + - ordered->len); - if (ret) { - err = ret; - goto out_unlock; - } - } else { - ret = insert_reserved_file_extent(trans, inode, - ordered->file_offset, - ordered->start, - ordered->disk_len, - ordered->len, - ordered->len, - 0, 0, 0, - BTRFS_FILE_EXTENT_REG); - unpin_extent_cache(&BTRFS_I(inode)->extent_tree, - ordered->file_offset, ordered->len); - if (ret) { - err = ret; - WARN_ON(1); - goto out_unlock; - } - } - - add_pending_csums(trans, inode, ordered->file_offset, &ordered->list); - ret = btrfs_ordered_update_i_size(inode, 0, ordered); - if (!ret || !test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags)) - btrfs_update_inode_fallback(trans, root, inode); - ret = 0; -out_unlock: - unlock_extent_cached(&BTRFS_I(inode)->io_tree, ordered->file_offset, - ordered->file_offset + ordered->len - 1, - &cached_state, GFP_NOFS); -out: - btrfs_delalloc_release_metadata(inode, ordered->len); - btrfs_end_transaction(trans, root); - ordered_offset = ordered->file_offset + ordered->len; - btrfs_put_ordered_extent(ordered); - btrfs_put_ordered_extent(ordered); - + ordered->work.func = finish_ordered_fn; + ordered->work.flags = 0; + btrfs_queue_worker(&root->fs_info->endio_write_workers, + &ordered->work); out_test: /* * our bio might span multiple ordered extents. If we haven't @@ -5996,12 +5967,12 @@ out_test: if (ordered_offset < dip->logical_offset + dip->bytes) { ordered_bytes = dip->logical_offset + dip->bytes - ordered_offset; + ordered = NULL; goto again; } out_done: bio->bi_private = dip->private; - kfree(dip->csums); kfree(dip); /* If we had an error make sure to clear the uptodate flag */ @@ -6069,9 +6040,12 @@ static inline int __btrfs_submit_dio_bio(struct bio *bio, struct inode *inode, int ret; bio_get(bio); - ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0); - if (ret) - goto err; + + if (!write) { + ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0); + if (ret) + goto err; + } if (skip_sum) goto map; @@ -6491,13 +6465,13 @@ static int btrfs_releasepage(struct page *page, gfp_t gfp_flags) static void btrfs_invalidatepage(struct page *page, unsigned long offset) { + struct inode *inode = page->mapping->host; struct extent_io_tree *tree; struct btrfs_ordered_extent *ordered; struct extent_state *cached_state = NULL; u64 page_start = page_offset(page); u64 page_end = page_start + PAGE_CACHE_SIZE - 1; - /* * we have the page locked, so new writeback can't start, * and the dirty bit won't be cleared while we are here. @@ -6507,13 +6481,13 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) */ wait_on_page_writeback(page); - tree = &BTRFS_I(page->mapping->host)->io_tree; + tree = &BTRFS_I(inode)->io_tree; if (offset) { btrfs_releasepage(page, GFP_NOFS); return; } lock_extent_bits(tree, page_start, page_end, 0, &cached_state); - ordered = btrfs_lookup_ordered_extent(page->mapping->host, + ordered = btrfs_lookup_ordered_extent(inode, page_offset(page)); if (ordered) { /* @@ -6528,9 +6502,10 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) * whoever cleared the private bit is responsible * for the finish_ordered_io */ - if (TestClearPagePrivate2(page)) { - btrfs_finish_ordered_io(page->mapping->host, - page_start, page_end); + if (TestClearPagePrivate2(page) && + btrfs_dec_test_ordered_pending(inode, &ordered, page_start, + PAGE_CACHE_SIZE, 1)) { + btrfs_finish_ordered_io(ordered); } btrfs_put_ordered_extent(ordered); cached_state = NULL; diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c index 9565c028916..9e138cdc36c 100644 --- a/fs/btrfs/ordered-data.c +++ b/fs/btrfs/ordered-data.c @@ -196,7 +196,7 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, entry->len = len; entry->disk_len = disk_len; entry->bytes_left = len; - entry->inode = inode; + entry->inode = igrab(inode); entry->compress_type = compress_type; if (type != BTRFS_ORDERED_IO_DONE && type != BTRFS_ORDERED_COMPLETE) set_bit(type, &entry->flags); @@ -212,12 +212,12 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, trace_btrfs_ordered_extent_add(inode, entry); - spin_lock(&tree->lock); + spin_lock_irq(&tree->lock); node = tree_insert(&tree->tree, file_offset, &entry->rb_node); if (node) ordered_data_tree_panic(inode, -EEXIST, file_offset); - spin_unlock(&tree->lock); + spin_unlock_irq(&tree->lock); spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock); list_add_tail(&entry->root_extent_list, @@ -264,9 +264,9 @@ void btrfs_add_ordered_sum(struct inode *inode, struct btrfs_ordered_inode_tree *tree; tree = &BTRFS_I(inode)->ordered_tree; - spin_lock(&tree->lock); + spin_lock_irq(&tree->lock); list_add_tail(&sum->list, &entry->list); - spin_unlock(&tree->lock); + spin_unlock_irq(&tree->lock); } /* @@ -283,18 +283,19 @@ void btrfs_add_ordered_sum(struct inode *inode, */ int btrfs_dec_test_first_ordered_pending(struct inode *inode, struct btrfs_ordered_extent **cached, - u64 *file_offset, u64 io_size) + u64 *file_offset, u64 io_size, int uptodate) { struct btrfs_ordered_inode_tree *tree; struct rb_node *node; struct btrfs_ordered_extent *entry = NULL; int ret; + unsigned long flags; u64 dec_end; u64 dec_start; u64 to_dec; tree = &BTRFS_I(inode)->ordered_tree; - spin_lock(&tree->lock); + spin_lock_irqsave(&tree->lock, flags); node = tree_search(tree, *file_offset); if (!node) { ret = 1; @@ -323,6 +324,9 @@ int btrfs_dec_test_first_ordered_pending(struct inode *inode, (unsigned long long)to_dec); } entry->bytes_left -= to_dec; + if (!uptodate) + set_bit(BTRFS_ORDERED_IOERR, &entry->flags); + if (entry->bytes_left == 0) ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags); else @@ -332,7 +336,7 @@ out: *cached = entry; atomic_inc(&entry->refs); } - spin_unlock(&tree->lock); + spin_unlock_irqrestore(&tree->lock, flags); return ret == 0; } @@ -347,15 +351,21 @@ out: */ int btrfs_dec_test_ordered_pending(struct inode *inode, struct btrfs_ordered_extent **cached, - u64 file_offset, u64 io_size) + u64 file_offset, u64 io_size, int uptodate) { struct btrfs_ordered_inode_tree *tree; struct rb_node *node; struct btrfs_ordered_extent *entry = NULL; + unsigned long flags; int ret; tree = &BTRFS_I(inode)->ordered_tree; - spin_lock(&tree->lock); + spin_lock_irqsave(&tree->lock, flags); + if (cached && *cached) { + entry = *cached; + goto have_entry; + } + node = tree_search(tree, file_offset); if (!node) { ret = 1; @@ -363,6 +373,7 @@ int btrfs_dec_test_ordered_pending(struct inode *inode, } entry = rb_entry(node, struct btrfs_ordered_extent, rb_node); +have_entry: if (!offset_in_entry(entry, file_offset)) { ret = 1; goto out; @@ -374,6 +385,9 @@ int btrfs_dec_test_ordered_pending(struct inode *inode, (unsigned long long)io_size); } entry->bytes_left -= io_size; + if (!uptodate) + set_bit(BTRFS_ORDERED_IOERR, &entry->flags); + if (entry->bytes_left == 0) ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags); else @@ -383,7 +397,7 @@ out: *cached = entry; atomic_inc(&entry->refs); } - spin_unlock(&tree->lock); + spin_unlock_irqrestore(&tree->lock, flags); return ret == 0; } @@ -399,6 +413,8 @@ void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry) trace_btrfs_ordered_extent_put(entry->inode, entry); if (atomic_dec_and_test(&entry->refs)) { + if (entry->inode) + btrfs_add_delayed_iput(entry->inode); while (!list_empty(&entry->list)) { cur = entry->list.next; sum = list_entry(cur, struct btrfs_ordered_sum, list); @@ -411,21 +427,22 @@ void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry) /* * remove an ordered extent from the tree. No references are dropped - * and you must wake_up entry->wait. You must hold the tree lock - * while you call this function. + * and waiters are woken up. */ -static void __btrfs_remove_ordered_extent(struct inode *inode, - struct btrfs_ordered_extent *entry) +void btrfs_remove_ordered_extent(struct inode *inode, + struct btrfs_ordered_extent *entry) { struct btrfs_ordered_inode_tree *tree; struct btrfs_root *root = BTRFS_I(inode)->root; struct rb_node *node; tree = &BTRFS_I(inode)->ordered_tree; + spin_lock_irq(&tree->lock); node = &entry->rb_node; rb_erase(node, &tree->tree); tree->last = NULL; set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags); + spin_unlock_irq(&tree->lock); spin_lock(&root->fs_info->ordered_extent_lock); list_del_init(&entry->root_extent_list); @@ -442,21 +459,6 @@ static void __btrfs_remove_ordered_extent(struct inode *inode, list_del_init(&BTRFS_I(inode)->ordered_operations); } spin_unlock(&root->fs_info->ordered_extent_lock); -} - -/* - * remove an ordered extent from the tree. No references are dropped - * but any waiters are woken. - */ -void btrfs_remove_ordered_extent(struct inode *inode, - struct btrfs_ordered_extent *entry) -{ - struct btrfs_ordered_inode_tree *tree; - - tree = &BTRFS_I(inode)->ordered_tree; - spin_lock(&tree->lock); - __btrfs_remove_ordered_extent(inode, entry); - spin_unlock(&tree->lock); wake_up(&entry->wait); } @@ -663,7 +665,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode, struct btrfs_ordered_extent *entry = NULL; tree = &BTRFS_I(inode)->ordered_tree; - spin_lock(&tree->lock); + spin_lock_irq(&tree->lock); node = tree_search(tree, file_offset); if (!node) goto out; @@ -674,7 +676,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode, if (entry) atomic_inc(&entry->refs); out: - spin_unlock(&tree->lock); + spin_unlock_irq(&tree->lock); return entry; } @@ -690,7 +692,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode, struct btrfs_ordered_extent *entry = NULL; tree = &BTRFS_I(inode)->ordered_tree; - spin_lock(&tree->lock); + spin_lock_irq(&tree->lock); node = tree_search(tree, file_offset); if (!node) { node = tree_search(tree, file_offset + len); @@ -715,7 +717,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode, out: if (entry) atomic_inc(&entry->refs); - spin_unlock(&tree->lock); + spin_unlock_irq(&tree->lock); return entry; } @@ -731,7 +733,7 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset) struct btrfs_ordered_extent *entry = NULL; tree = &BTRFS_I(inode)->ordered_tree; - spin_lock(&tree->lock); + spin_lock_irq(&tree->lock); node = tree_search(tree, file_offset); if (!node) goto out; @@ -739,7 +741,7 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset) entry = rb_entry(node, struct btrfs_ordered_extent, rb_node); atomic_inc(&entry->refs); out: - spin_unlock(&tree->lock); + spin_unlock_irq(&tree->lock); return entry; } @@ -765,7 +767,7 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 offset, else offset = ALIGN(offset, BTRFS_I(inode)->root->sectorsize); - spin_lock(&tree->lock); + spin_lock_irq(&tree->lock); disk_i_size = BTRFS_I(inode)->disk_i_size; /* truncate file */ @@ -803,15 +805,18 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 offset, } node = prev; } - while (node) { + for (; node; node = rb_prev(node)) { test = rb_entry(node, struct btrfs_ordered_extent, rb_node); + + /* We treat this entry as if it doesnt exist */ + if (test_bit(BTRFS_ORDERED_UPDATED_ISIZE, &test->flags)) + continue; if (test->file_offset + test->len <= disk_i_size) break; if (test->file_offset >= i_size) break; if (test->file_offset >= disk_i_size) goto out; - node = rb_prev(node); } new_i_size = min_t(u64, offset, i_size); @@ -829,17 +834,27 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 offset, else node = rb_first(&tree->tree); } - i_size_test = 0; - if (node) { - /* - * do we have an area where IO might have finished - * between our ordered extent and the next one. - */ + + /* + * We are looking for an area between our current extent and the next + * ordered extent to update the i_size to. There are 3 cases here + * + * 1) We don't actually have anything and we can update to i_size. + * 2) We have stuff but they already did their i_size update so again we + * can just update to i_size. + * 3) We have an outstanding ordered extent so the most we can update + * our disk_i_size to is the start of the next offset. + */ + i_size_test = i_size; + for (; node; node = rb_next(node)) { test = rb_entry(node, struct btrfs_ordered_extent, rb_node); - if (test->file_offset > offset) + + if (test_bit(BTRFS_ORDERED_UPDATED_ISIZE, &test->flags)) + continue; + if (test->file_offset > offset) { i_size_test = test->file_offset; - } else { - i_size_test = i_size; + break; + } } /* @@ -853,15 +868,15 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 offset, ret = 0; out: /* - * we need to remove the ordered extent with the tree lock held - * so that other people calling this function don't find our fully - * processed ordered entry and skip updating the i_size + * We need to do this because we can't remove ordered extents until + * after the i_disk_size has been updated and then the inode has been + * updated to reflect the change, so we need to tell anybody who finds + * this ordered extent that we've already done all the real work, we + * just haven't completed all the other work. */ if (ordered) - __btrfs_remove_ordered_extent(inode, ordered); - spin_unlock(&tree->lock); - if (ordered) - wake_up(&ordered->wait); + set_bit(BTRFS_ORDERED_UPDATED_ISIZE, &ordered->flags); + spin_unlock_irq(&tree->lock); return ret; } @@ -886,7 +901,7 @@ int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, if (!ordered) return 1; - spin_lock(&tree->lock); + spin_lock_irq(&tree->lock); list_for_each_entry_reverse(ordered_sum, &ordered->list, list) { if (disk_bytenr >= ordered_sum->bytenr) { num_sectors = ordered_sum->len / sectorsize; @@ -901,7 +916,7 @@ int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, } } out: - spin_unlock(&tree->lock); + spin_unlock_irq(&tree->lock); btrfs_put_ordered_extent(ordered); return ret; } diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h index c355ad4dc1a..e03c560d299 100644 --- a/fs/btrfs/ordered-data.h +++ b/fs/btrfs/ordered-data.h @@ -74,6 +74,12 @@ struct btrfs_ordered_sum { #define BTRFS_ORDERED_DIRECT 5 /* set when we're doing DIO with this extent */ +#define BTRFS_ORDERED_IOERR 6 /* We had an io error when writing this out */ + +#define BTRFS_ORDERED_UPDATED_ISIZE 7 /* indicates wether this ordered extent + * has done its due diligence in updating + * the isize. */ + struct btrfs_ordered_extent { /* logical offset in the file */ u64 file_offset; @@ -113,6 +119,8 @@ struct btrfs_ordered_extent { /* a per root list of all the pending ordered extents */ struct list_head root_extent_list; + + struct btrfs_work work; }; @@ -143,10 +151,11 @@ void btrfs_remove_ordered_extent(struct inode *inode, struct btrfs_ordered_extent *entry); int btrfs_dec_test_ordered_pending(struct inode *inode, struct btrfs_ordered_extent **cached, - u64 file_offset, u64 io_size); + u64 file_offset, u64 io_size, int uptodate); int btrfs_dec_test_first_ordered_pending(struct inode *inode, struct btrfs_ordered_extent **cached, - u64 *file_offset, u64 io_size); + u64 *file_offset, u64 io_size, + int uptodate); int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, u64 start, u64 len, u64 disk_len, int type); int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset, -- cgit v1.2.3-70-g09d2 From 72ac3c0d7921f943d92d1ef42a549fb52e56817d Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 23 May 2012 14:13:11 -0400 Subject: Btrfs: convert the inode bit field to use the actual bit operations Miao pointed this out while I was working on an orphan problem that messing with a bitfield where different ranges are protected by different locks doesn't work out right. Turns out we've been doing this forever where we have different parts of the bit field protected by either no lock at all or different locks which could cause all sorts of weird problems including the issue I was hitting. So instead make a runtime_flags thing that we use the normal bit operations on that are all atomic so we can keep having our no/different locking for the different flags and then make force_compress it's own thing so it can be treated normally. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/btrfs_inode.h | 30 ++++++++++++++++-------------- fs/btrfs/delayed-inode.c | 4 ++-- fs/btrfs/disk-io.c | 3 ++- fs/btrfs/extent-tree.c | 11 ++++++----- fs/btrfs/file.c | 12 ++++++------ fs/btrfs/inode.c | 28 ++++++++++++---------------- 6 files changed, 44 insertions(+), 44 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h index 3771b8543a7..6265edb219e 100644 --- a/fs/btrfs/btrfs_inode.h +++ b/fs/btrfs/btrfs_inode.h @@ -24,6 +24,19 @@ #include "ordered-data.h" #include "delayed-inode.h" +/* + * ordered_data_close is set by truncate when a file that used + * to have good data has been truncated to zero. When it is set + * the btrfs file release call will add this inode to the + * ordered operations list so that we make sure to flush out any + * new data the application may have written before commit. + */ +#define BTRFS_INODE_ORDERED_DATA_CLOSE 0 +#define BTRFS_INODE_ORPHAN_META_RESERVED 1 +#define BTRFS_INODE_DUMMY 2 +#define BTRFS_INODE_IN_DEFRAG 3 +#define BTRFS_INODE_DELALLOC_META_RESERVED 4 + /* in memory btrfs inode */ struct btrfs_inode { /* which subvolume this inode belongs to */ @@ -78,6 +91,8 @@ struct btrfs_inode { /* the space_info for where this inode's data allocations are done */ struct btrfs_space_info *space_info; + unsigned long runtime_flags; + /* full 64 bit generation number, struct vfs_inode doesn't have a big * enough field for this. */ @@ -141,23 +156,10 @@ struct btrfs_inode { unsigned outstanding_extents; unsigned reserved_extents; - /* - * ordered_data_close is set by truncate when a file that used - * to have good data has been truncated to zero. When it is set - * the btrfs file release call will add this inode to the - * ordered operations list so that we make sure to flush out any - * new data the application may have written before commit. - */ - unsigned ordered_data_close:1; - unsigned orphan_meta_reserved:1; - unsigned dummy_inode:1; - unsigned in_defrag:1; - unsigned delalloc_meta_reserved:1; - /* * always compress this one file */ - unsigned force_compress:4; + unsigned force_compress; struct btrfs_delayed_node *delayed_node; diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c index bcd40c7109f..c18d0442ae6 100644 --- a/fs/btrfs/delayed-inode.c +++ b/fs/btrfs/delayed-inode.c @@ -669,8 +669,8 @@ static int btrfs_delayed_inode_reserve_metadata( return ret; } else if (src_rsv == &root->fs_info->delalloc_block_rsv) { spin_lock(&BTRFS_I(inode)->lock); - if (BTRFS_I(inode)->delalloc_meta_reserved) { - BTRFS_I(inode)->delalloc_meta_reserved = 0; + if (test_and_clear_bit(BTRFS_INODE_DELALLOC_META_RESERVED, + &BTRFS_I(inode)->runtime_flags)) { spin_unlock(&BTRFS_I(inode)->lock); release = true; goto migrate; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 19f5b450f40..0cf8ef2b5b1 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2001,7 +2001,8 @@ int open_ctree(struct super_block *sb, BTRFS_I(fs_info->btree_inode)->root = tree_root; memset(&BTRFS_I(fs_info->btree_inode)->location, 0, sizeof(struct btrfs_key)); - BTRFS_I(fs_info->btree_inode)->dummy_inode = 1; + set_bit(BTRFS_INODE_DUMMY, + &BTRFS_I(fs_info->btree_inode)->runtime_flags); insert_inode_hash(fs_info->btree_inode); spin_lock_init(&fs_info->block_group_cache_lock); diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 59ae191d4f9..1902726fa70 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4355,10 +4355,9 @@ static unsigned drop_outstanding_extent(struct inode *inode) BTRFS_I(inode)->outstanding_extents--; if (BTRFS_I(inode)->outstanding_extents == 0 && - BTRFS_I(inode)->delalloc_meta_reserved) { + test_and_clear_bit(BTRFS_INODE_DELALLOC_META_RESERVED, + &BTRFS_I(inode)->runtime_flags)) drop_inode_space = 1; - BTRFS_I(inode)->delalloc_meta_reserved = 0; - } /* * If we have more or the same amount of outsanding extents than we have @@ -4465,7 +4464,8 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) * Add an item to reserve for updating the inode when we complete the * delalloc io. */ - if (!BTRFS_I(inode)->delalloc_meta_reserved) { + if (!test_bit(BTRFS_INODE_DELALLOC_META_RESERVED, + &BTRFS_I(inode)->runtime_flags)) { nr_extents++; extra_reserve = 1; } @@ -4511,7 +4511,8 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) spin_lock(&BTRFS_I(inode)->lock); if (extra_reserve) { - BTRFS_I(inode)->delalloc_meta_reserved = 1; + set_bit(BTRFS_INODE_DELALLOC_META_RESERVED, + &BTRFS_I(inode)->runtime_flags); nr_extents--; } BTRFS_I(inode)->reserved_extents += nr_extents; diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index cfc0ab915d0..c9005f21697 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -103,7 +103,7 @@ static void __btrfs_add_inode_defrag(struct inode *inode, goto exists; } } - BTRFS_I(inode)->in_defrag = 1; + set_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags); rb_link_node(&defrag->rb_node, parent, p); rb_insert_color(&defrag->rb_node, &root->fs_info->defrag_inodes); return; @@ -131,7 +131,7 @@ int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans, if (btrfs_fs_closing(root->fs_info)) return 0; - if (BTRFS_I(inode)->in_defrag) + if (test_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags)) return 0; if (trans) @@ -148,7 +148,7 @@ int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans, defrag->root = root->root_key.objectid; spin_lock(&root->fs_info->defrag_inodes_lock); - if (!BTRFS_I(inode)->in_defrag) + if (!test_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags)) __btrfs_add_inode_defrag(inode, defrag); else kfree(defrag); @@ -252,7 +252,7 @@ int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info) goto next; /* do a chunk of defrag */ - BTRFS_I(inode)->in_defrag = 0; + clear_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags); range.start = defrag->last_offset; num_defrag = btrfs_defrag_file(inode, NULL, &range, defrag->transid, defrag_batch); @@ -1465,8 +1465,8 @@ int btrfs_release_file(struct inode *inode, struct file *filp) * 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; + if (test_and_clear_bit(BTRFS_INODE_ORDERED_DATA_CLOSE, + &BTRFS_I(inode)->runtime_flags)) { btrfs_add_ordered_operation(NULL, BTRFS_I(inode)->root, inode); if (inode->i_size > BTRFS_ORDERED_OPERATIONS_FLUSH_LIMIT) filemap_flush(inode->i_mapping); diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 9a1b96fd672..91ad6390175 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -2182,10 +2182,9 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) insert = 1; } - if (!BTRFS_I(inode)->orphan_meta_reserved) { - BTRFS_I(inode)->orphan_meta_reserved = 1; + if (!test_and_set_bit(BTRFS_INODE_ORPHAN_META_RESERVED, + &BTRFS_I(inode)->runtime_flags)) reserve = 1; - } spin_unlock(&root->orphan_lock); /* grab metadata reservation from transaction handle */ @@ -2233,10 +2232,9 @@ int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode) delete_item = 1; } - if (BTRFS_I(inode)->orphan_meta_reserved) { - BTRFS_I(inode)->orphan_meta_reserved = 0; + if (test_and_clear_bit(BTRFS_INODE_ORPHAN_META_RESERVED, + &BTRFS_I(inode)->runtime_flags)) release_rsv = 1; - } spin_unlock(&root->orphan_lock); if (trans && delete_item) { @@ -3642,7 +3640,8 @@ static int btrfs_setsize(struct inode *inode, loff_t newsize) * any new writes get down to disk quickly. */ if (newsize == 0) - BTRFS_I(inode)->ordered_data_close = 1; + set_bit(BTRFS_INODE_ORDERED_DATA_CLOSE, + &BTRFS_I(inode)->runtime_flags); /* we don't support swapfiles, so vmtruncate shouldn't fail */ truncate_setsize(inode, newsize); @@ -4102,7 +4101,7 @@ static struct inode *new_simple_dir(struct super_block *s, BTRFS_I(inode)->root = root; memcpy(&BTRFS_I(inode)->location, key, sizeof(*key)); - BTRFS_I(inode)->dummy_inode = 1; + set_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags); inode->i_ino = BTRFS_EMPTY_SUBVOL_DIR_OBJECTID; inode->i_op = &btrfs_dir_ro_inode_operations; @@ -4406,7 +4405,7 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc) int ret = 0; bool nolock = false; - if (BTRFS_I(inode)->dummy_inode) + if (test_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags)) return 0; if (btrfs_fs_closing(root->fs_info) && btrfs_is_free_space_inode(root, inode)) @@ -4439,7 +4438,7 @@ int btrfs_dirty_inode(struct inode *inode) struct btrfs_trans_handle *trans; int ret; - if (BTRFS_I(inode)->dummy_inode) + if (test_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags)) return 0; trans = btrfs_join_transaction(root); @@ -6752,7 +6751,8 @@ static int btrfs_truncate(struct inode *inode) * using truncate to replace the contents of the file will * end up with a zero length file after a crash. */ - if (inode->i_size == 0 && BTRFS_I(inode)->ordered_data_close) + if (inode->i_size == 0 && test_bit(BTRFS_INODE_ORDERED_DATA_CLOSE, + &BTRFS_I(inode)->runtime_flags)) btrfs_add_ordered_operation(trans, root, inode); while (1) { @@ -6889,11 +6889,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb) ei->outstanding_extents = 0; ei->reserved_extents = 0; - ei->ordered_data_close = 0; - ei->orphan_meta_reserved = 0; - ei->dummy_inode = 0; - ei->in_defrag = 0; - ei->delalloc_meta_reserved = 0; + ei->runtime_flags = 0; ei->force_compress = BTRFS_COMPRESS_NONE; ei->delayed_node = NULL; -- cgit v1.2.3-70-g09d2 From 8a35d95ff4680a456d3ce47df9638f33d4f54f20 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 23 May 2012 14:26:42 -0400 Subject: Btrfs: fix how we deal with the orphan block rsv Ceph was hitting this race where we would remove an inode from the per-root orphan list before we would release the space we had reserved for the inode. We actually don't need a list or anything, we just need to make sure the root doesn't try to free up the orphan reserve until after the inodes have released their reservations. So use an atomic counter instead of a list on the root and only decrement the counter after we've released our reservation. I've tested this as well as several others and we no longer see the warnings that you would see while running ceph. Thanks, Btrfs: fix how we deal with the orphan block rsv Ceph was hitting this race where we would remove an inode from the per-root orphan list before we would release the space we had reserved for the inode. We actually don't need a list or anything, we just need to make sure the root doesn't try to free up the orphan reserve until after the inodes have released their reservations. So use an atomic counter instead of a list on the root and only decrement the counter after we've released our reservation. I've tested this as well as several others and we no longer see the warnings that you would see while running ceph. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/btrfs_inode.h | 4 +--- fs/btrfs/ctree.h | 2 +- fs/btrfs/disk-io.c | 2 +- fs/btrfs/inode.c | 38 +++++++++++++++++++++----------------- 4 files changed, 24 insertions(+), 22 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h index 6265edb219e..ce2c9d60031 100644 --- a/fs/btrfs/btrfs_inode.h +++ b/fs/btrfs/btrfs_inode.h @@ -36,6 +36,7 @@ #define BTRFS_INODE_DUMMY 2 #define BTRFS_INODE_IN_DEFRAG 3 #define BTRFS_INODE_DELALLOC_META_RESERVED 4 +#define BTRFS_INODE_HAS_ORPHAN_ITEM 5 /* in memory btrfs inode */ struct btrfs_inode { @@ -70,9 +71,6 @@ struct btrfs_inode { /* used to order data wrt metadata */ struct btrfs_ordered_inode_tree ordered_tree; - /* for keeping track of orphaned inodes */ - struct list_head i_orphan; - /* list of all the delalloc inodes in the FS. There are times we need * to write all the delalloc pages to disk, and this list is used * to walk them all. diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 8fd72331d60..aad2600718a 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1375,7 +1375,7 @@ struct btrfs_root { struct list_head root_list; spinlock_t orphan_lock; - struct list_head orphan_list; + atomic_t orphan_inodes; struct btrfs_block_rsv *orphan_block_rsv; int orphan_item_inserted; int orphan_cleanup_state; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 0cf8ef2b5b1..297e5a8ed93 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1153,7 +1153,6 @@ static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->orphan_block_rsv = NULL; INIT_LIST_HEAD(&root->dirty_list); - INIT_LIST_HEAD(&root->orphan_list); INIT_LIST_HEAD(&root->root_list); spin_lock_init(&root->orphan_lock); spin_lock_init(&root->inode_lock); @@ -1166,6 +1165,7 @@ static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, atomic_set(&root->log_commit[0], 0); atomic_set(&root->log_commit[1], 0); atomic_set(&root->log_writers, 0); + atomic_set(&root->orphan_inodes, 0); root->log_batch = 0; root->log_transid = 0; root->last_log_commit = 0; diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 91ad6390175..029892887fc 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -2104,12 +2104,12 @@ void btrfs_orphan_commit_root(struct btrfs_trans_handle *trans, struct btrfs_block_rsv *block_rsv; int ret; - if (!list_empty(&root->orphan_list) || + if (atomic_read(&root->orphan_inodes) || root->orphan_cleanup_state != ORPHAN_CLEANUP_DONE) return; spin_lock(&root->orphan_lock); - if (!list_empty(&root->orphan_list)) { + if (atomic_read(&root->orphan_inodes)) { spin_unlock(&root->orphan_lock); return; } @@ -2166,8 +2166,8 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) block_rsv = NULL; } - if (list_empty(&BTRFS_I(inode)->i_orphan)) { - list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list); + if (!test_and_set_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags)) { #if 0 /* * For proper ENOSPC handling, we should do orphan @@ -2180,6 +2180,7 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) insert = 1; #endif insert = 1; + atomic_dec(&root->orphan_inodes); } if (!test_and_set_bit(BTRFS_INODE_ORPHAN_META_RESERVED, @@ -2197,6 +2198,8 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) if (insert >= 1) { ret = btrfs_insert_orphan_item(trans, root, btrfs_ino(inode)); if (ret && ret != -EEXIST) { + clear_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags); btrfs_abort_transaction(trans, root, ret); return ret; } @@ -2227,10 +2230,9 @@ int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode) int ret = 0; spin_lock(&root->orphan_lock); - if (!list_empty(&BTRFS_I(inode)->i_orphan)) { - list_del_init(&BTRFS_I(inode)->i_orphan); + if (test_and_clear_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags)) delete_item = 1; - } if (test_and_clear_bit(BTRFS_INODE_ORPHAN_META_RESERVED, &BTRFS_I(inode)->runtime_flags)) @@ -2242,8 +2244,10 @@ int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode) BUG_ON(ret); /* -ENOMEM or corruption (JDM: Recheck) */ } - if (release_rsv) + if (release_rsv) { btrfs_orphan_release_metadata(inode); + atomic_dec(&root->orphan_inodes); + } return 0; } @@ -2371,6 +2375,8 @@ int btrfs_orphan_cleanup(struct btrfs_root *root) ret = PTR_ERR(trans); goto out; } + printk(KERN_ERR "auto deleting %Lu\n", + found_key.objectid); ret = btrfs_del_orphan_item(trans, root, found_key.objectid); BUG_ON(ret); /* -ENOMEM or corruption (JDM: Recheck) */ @@ -2382,9 +2388,8 @@ int btrfs_orphan_cleanup(struct btrfs_root *root) * add this inode to the orphan list so btrfs_orphan_del does * the proper thing when we hit it */ - spin_lock(&root->orphan_lock); - list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list); - spin_unlock(&root->orphan_lock); + set_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags); /* if we have links, this was a truncate, lets do that */ if (inode->i_nlink) { @@ -3706,7 +3711,8 @@ void btrfs_evict_inode(struct inode *inode) btrfs_wait_ordered_range(inode, 0, (u64)-1); if (root->fs_info->log_root_recovering) { - BUG_ON(!list_empty(&BTRFS_I(inode)->i_orphan)); + BUG_ON(!test_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags)); goto no_delete; } @@ -6903,7 +6909,6 @@ struct inode *btrfs_alloc_inode(struct super_block *sb) mutex_init(&ei->log_mutex); mutex_init(&ei->delalloc_mutex); btrfs_ordered_inode_tree_init(&ei->ordered_tree); - INIT_LIST_HEAD(&ei->i_orphan); INIT_LIST_HEAD(&ei->delalloc_inodes); INIT_LIST_HEAD(&ei->ordered_operations); RB_CLEAR_NODE(&ei->rb_node); @@ -6948,13 +6953,12 @@ void btrfs_destroy_inode(struct inode *inode) spin_unlock(&root->fs_info->ordered_extent_lock); } - spin_lock(&root->orphan_lock); - if (!list_empty(&BTRFS_I(inode)->i_orphan)) { + if (test_bit(BTRFS_INODE_HAS_ORPHAN_ITEM, + &BTRFS_I(inode)->runtime_flags)) { printk(KERN_INFO "BTRFS: inode %llu still on the orphan list\n", (unsigned long long)btrfs_ino(inode)); - list_del_init(&BTRFS_I(inode)->i_orphan); + atomic_dec(&root->orphan_inodes); } - spin_unlock(&root->orphan_lock); while (1) { ordered = btrfs_lookup_first_ordered_extent(inode, (u64)-1); -- cgit v1.2.3-70-g09d2 From d07eb9117050c9ed3f78296ebcc06128b52693be Mon Sep 17 00:00:00 2001 From: Asias He Date: Fri, 25 May 2012 11:10:21 +0800 Subject: btrfs: Drop unused function btrfs_abort_devices() 1) This function is not used anywhere. 2) Using the blk_abort_queue() to abort the queue seems not correct. blk_abort_queue() is used for timeout handling (block/blk-timeout.c). Cc: Chris Mason Cc: linux-btrfs@vger.kernel.org Cc: Jens Axboe Cc: linux-kernel@vger.kernel.org Signed-off-by: Asias He --- fs/btrfs/disk-io.c | 13 ------------- fs/btrfs/disk-io.h | 1 - 2 files changed, 14 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 297e5a8ed93..0f788c05906 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2903,19 +2903,6 @@ 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); diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h index ab1830aaf0e..05b3fab39f7 100644 --- a/fs/btrfs/disk-io.h +++ b/fs/btrfs/disk-io.h @@ -89,7 +89,6 @@ int btrfs_add_log_tree(struct btrfs_trans_handle *trans, int btrfs_cleanup_transaction(struct btrfs_root *root); void btrfs_cleanup_one_transaction(struct btrfs_transaction *trans, struct btrfs_root *root); -void btrfs_abort_devices(struct btrfs_root *root); #ifdef CONFIG_DEBUG_LOCK_ALLOC void btrfs_init_lockdep(void); -- cgit v1.2.3-70-g09d2 From 442a4f6308e694e0fa6025708bd5e4e424bbf51c Mon Sep 17 00:00:00 2001 From: Stefan Behrens Date: Fri, 25 May 2012 16:06:08 +0200 Subject: Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens --- fs/btrfs/disk-io.c | 13 +++++--- fs/btrfs/extent_io.c | 18 ++++++++-- fs/btrfs/ioctl.h | 19 +++++++++++ fs/btrfs/scrub.c | 65 +++++++++++++++++++++++++++--------- fs/btrfs/volumes.c | 94 ++++++++++++++++++++++++++++++++++++++++++++++++++-- fs/btrfs/volumes.h | 45 +++++++++++++++++++++++++ 6 files changed, 230 insertions(+), 24 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 0f788c05906..46d474e74aa 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2557,18 +2557,19 @@ recovery_tree_root: static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate) { - char b[BDEVNAME_SIZE]; - if (uptodate) { set_buffer_uptodate(bh); } else { + struct btrfs_device *device = (struct btrfs_device *) + bh->b_private; + printk_ratelimited(KERN_WARNING "lost page write due to " - "I/O error on %s\n", - bdevname(bh->b_bdev, b)); + "I/O error on %s\n", device->name); /* note, we dont' set_buffer_write_io_error because we have * our own ways of dealing with the IO errors */ clear_buffer_uptodate(bh); + btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS); } unlock_buffer(bh); put_bh(bh); @@ -2683,6 +2684,7 @@ static int write_dev_supers(struct btrfs_device *device, set_buffer_uptodate(bh); lock_buffer(bh); bh->b_end_io = btrfs_end_buffer_write_sync; + bh->b_private = device; } /* @@ -2741,6 +2743,9 @@ static int write_dev_flush(struct btrfs_device *device, int wait) } if (!bio_flagged(bio, BIO_UPTODATE)) { ret = -EIO; + if (!bio_flagged(bio, BIO_EOPNOTSUPP)) + btrfs_dev_stat_inc_and_print(device, + BTRFS_DEV_STAT_FLUSH_ERRS); } /* drop the reference from the wait == 0 run */ diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 69a527c7a0b..b3692c1373a 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -1913,6 +1913,7 @@ int repair_io_failure(struct btrfs_mapping_tree *map_tree, u64 start, if (!test_bit(BIO_UPTODATE, &bio->bi_flags)) { /* try to remap that extent elsewhere? */ bio_put(bio); + btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_WRITE_ERRS); return -EIO; } @@ -2327,10 +2328,23 @@ static void end_bio_extent_readpage(struct bio *bio, int err) if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) { ret = tree->ops->readpage_end_io_hook(page, start, end, state, mirror); - if (ret) + if (ret) { + /* no IO indicated but software detected errors + * in the block, either checksum errors or + * issues with the contents */ + struct btrfs_root *root = + BTRFS_I(page->mapping->host)->root; + struct btrfs_device *device; + uptodate = 0; - else + device = btrfs_find_device_for_logical( + root, start, mirror); + if (device) + btrfs_dev_stat_inc_and_print(device, + BTRFS_DEV_STAT_CORRUPTION_ERRS); + } else { clean_io_failure(start, page); + } } if (!uptodate && tree->ops && tree->ops->readpage_io_failed_hook) { diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h index 086e6bdae1c..5bf05e28b82 100644 --- a/fs/btrfs/ioctl.h +++ b/fs/btrfs/ioctl.h @@ -266,6 +266,25 @@ struct btrfs_ioctl_logical_ino_args { __u64 inodes; }; +enum btrfs_dev_stat_values { + /* disk I/O failure stats */ + BTRFS_DEV_STAT_WRITE_ERRS, /* EIO or EREMOTEIO from lower layers */ + BTRFS_DEV_STAT_READ_ERRS, /* EIO or EREMOTEIO from lower layers */ + BTRFS_DEV_STAT_FLUSH_ERRS, /* EIO or EREMOTEIO from lower layers */ + + /* stats for indirect indications for I/O failures */ + BTRFS_DEV_STAT_CORRUPTION_ERRS, /* checksum error, bytenr error or + * contents is illegal: this is an + * indication that the block was damaged + * during read or write, or written to + * wrong location or read from wrong + * location */ + BTRFS_DEV_STAT_GENERATION_ERRS, /* an indication that blocks have not + * been written */ + + BTRFS_DEV_STAT_VALUES_MAX +}; + #define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \ struct btrfs_ioctl_vol_args) #define BTRFS_IOC_DEFRAG _IOW(BTRFS_IOCTL_MAGIC, 2, \ diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index 2f3d6f917fb..a38cfa4f251 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -50,7 +50,7 @@ struct scrub_dev; struct scrub_page { struct scrub_block *sblock; struct page *page; - struct block_device *bdev; + struct btrfs_device *dev; u64 flags; /* extent flags */ u64 generation; u64 logical; @@ -86,6 +86,7 @@ struct scrub_block { unsigned int header_error:1; unsigned int checksum_error:1; unsigned int no_io_error_seen:1; + unsigned int generation_error:1; /* also sets header_error */ }; }; @@ -675,6 +676,8 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check) sdev->stat.read_errors++; sdev->stat.uncorrectable_errors++; spin_unlock(&sdev->stat_lock); + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_READ_ERRS); goto out; } @@ -686,6 +689,8 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check) sdev->stat.read_errors++; sdev->stat.uncorrectable_errors++; spin_unlock(&sdev->stat_lock); + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_READ_ERRS); goto out; } BUG_ON(failed_mirror_index >= BTRFS_MAX_MIRRORS); @@ -699,6 +704,8 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check) sdev->stat.read_errors++; sdev->stat.uncorrectable_errors++; spin_unlock(&sdev->stat_lock); + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_READ_ERRS); goto out; } @@ -725,12 +732,16 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check) spin_unlock(&sdev->stat_lock); if (__ratelimit(&_rs)) scrub_print_warning("i/o error", sblock_to_check); + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_READ_ERRS); } else if (sblock_bad->checksum_error) { spin_lock(&sdev->stat_lock); sdev->stat.csum_errors++; spin_unlock(&sdev->stat_lock); if (__ratelimit(&_rs)) scrub_print_warning("checksum error", sblock_to_check); + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_CORRUPTION_ERRS); } else if (sblock_bad->header_error) { spin_lock(&sdev->stat_lock); sdev->stat.verify_errors++; @@ -738,6 +749,12 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check) if (__ratelimit(&_rs)) scrub_print_warning("checksum/header error", sblock_to_check); + if (sblock_bad->generation_error) + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_GENERATION_ERRS); + else + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_CORRUPTION_ERRS); } if (sdev->readonly) @@ -998,8 +1015,8 @@ static int scrub_setup_recheck_block(struct scrub_dev *sdev, page = sblock->pagev + page_index; page->logical = logical; page->physical = bbio->stripes[mirror_index].physical; - /* for missing devices, bdev is NULL */ - page->bdev = bbio->stripes[mirror_index].dev->bdev; + /* for missing devices, dev->bdev is NULL */ + page->dev = bbio->stripes[mirror_index].dev; page->mirror_num = mirror_index + 1; page->page = alloc_page(GFP_NOFS); if (!page->page) { @@ -1043,7 +1060,7 @@ static int scrub_recheck_block(struct btrfs_fs_info *fs_info, struct scrub_page *page = sblock->pagev + page_num; DECLARE_COMPLETION_ONSTACK(complete); - if (page->bdev == NULL) { + if (page->dev->bdev == NULL) { page->io_error = 1; sblock->no_io_error_seen = 0; continue; @@ -1053,7 +1070,7 @@ static int scrub_recheck_block(struct btrfs_fs_info *fs_info, bio = bio_alloc(GFP_NOFS, 1); if (!bio) return -EIO; - bio->bi_bdev = page->bdev; + bio->bi_bdev = page->dev->bdev; bio->bi_sector = page->physical >> 9; bio->bi_end_io = scrub_complete_bio_end_io; bio->bi_private = &complete; @@ -1102,11 +1119,14 @@ static void scrub_recheck_block_checksum(struct btrfs_fs_info *fs_info, h = (struct btrfs_header *)mapped_buffer; if (sblock->pagev[0].logical != le64_to_cpu(h->bytenr) || - generation != le64_to_cpu(h->generation) || memcmp(h->fsid, fs_info->fsid, BTRFS_UUID_SIZE) || memcmp(h->chunk_tree_uuid, fs_info->chunk_tree_uuid, - BTRFS_UUID_SIZE)) + BTRFS_UUID_SIZE)) { sblock->header_error = 1; + } else if (generation != le64_to_cpu(h->generation)) { + sblock->header_error = 1; + sblock->generation_error = 1; + } csum = h->csum; } else { if (!have_csum) @@ -1182,7 +1202,7 @@ static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad, bio = bio_alloc(GFP_NOFS, 1); if (!bio) return -EIO; - bio->bi_bdev = page_bad->bdev; + bio->bi_bdev = page_bad->dev->bdev; bio->bi_sector = page_bad->physical >> 9; bio->bi_end_io = scrub_complete_bio_end_io; bio->bi_private = &complete; @@ -1196,6 +1216,12 @@ static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad, /* this will also unplug the queue */ wait_for_completion(&complete); + if (!bio_flagged(bio, BIO_UPTODATE)) { + btrfs_dev_stat_inc_and_print(page_bad->dev, + BTRFS_DEV_STAT_WRITE_ERRS); + bio_put(bio); + return -EIO; + } bio_put(bio); } @@ -1352,7 +1378,8 @@ static int scrub_checksum_super(struct scrub_block *sblock) u64 mapped_size; void *p; u32 crc = ~(u32)0; - int fail = 0; + int fail_gen = 0; + int fail_cor = 0; u64 len; int index; @@ -1363,13 +1390,13 @@ static int scrub_checksum_super(struct scrub_block *sblock) memcpy(on_disk_csum, s->csum, sdev->csum_size); if (sblock->pagev[0].logical != le64_to_cpu(s->bytenr)) - ++fail; + ++fail_cor; if (sblock->pagev[0].generation != le64_to_cpu(s->generation)) - ++fail; + ++fail_gen; if (memcmp(s->fsid, fs_info->fsid, BTRFS_UUID_SIZE)) - ++fail; + ++fail_cor; len = BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE; mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE; @@ -1394,9 +1421,9 @@ static int scrub_checksum_super(struct scrub_block *sblock) btrfs_csum_final(crc, calculated_csum); if (memcmp(calculated_csum, on_disk_csum, sdev->csum_size)) - ++fail; + ++fail_cor; - if (fail) { + if (fail_cor + fail_gen) { /* * if we find an error in a super block, we just report it. * They will get written with the next transaction commit @@ -1405,9 +1432,15 @@ static int scrub_checksum_super(struct scrub_block *sblock) spin_lock(&sdev->stat_lock); ++sdev->stat.super_errors; spin_unlock(&sdev->stat_lock); + if (fail_cor) + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_CORRUPTION_ERRS); + else + btrfs_dev_stat_inc_and_print(sdev->dev, + BTRFS_DEV_STAT_GENERATION_ERRS); } - return fail; + return fail_cor + fail_gen; } static void scrub_block_get(struct scrub_block *sblock) @@ -1551,7 +1584,7 @@ static int scrub_pages(struct scrub_dev *sdev, u64 logical, u64 len, return -ENOMEM; } spage->sblock = sblock; - spage->bdev = sdev->dev->bdev; + spage->dev = sdev->dev; spage->flags = flags; spage->generation = gen; spage->logical = logical; diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 48a06d1fc06..2915521f44e 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -23,6 +23,7 @@ #include #include #include +#include #include #include #include "compat.h" @@ -4001,13 +4002,58 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree, return 0; } +static void *merge_stripe_index_into_bio_private(void *bi_private, + unsigned int stripe_index) +{ + /* + * with single, dup, RAID0, RAID1 and RAID10, stripe_index is + * at most 1. + * The alternative solution (instead of stealing bits from the + * pointer) would be to allocate an intermediate structure + * that contains the old private pointer plus the stripe_index. + */ + BUG_ON((((uintptr_t)bi_private) & 3) != 0); + BUG_ON(stripe_index > 3); + return (void *)(((uintptr_t)bi_private) | stripe_index); +} + +static struct btrfs_bio *extract_bbio_from_bio_private(void *bi_private) +{ + return (struct btrfs_bio *)(((uintptr_t)bi_private) & ~((uintptr_t)3)); +} + +static unsigned int extract_stripe_index_from_bio_private(void *bi_private) +{ + return (unsigned int)((uintptr_t)bi_private) & 3; +} + static void btrfs_end_bio(struct bio *bio, int err) { - struct btrfs_bio *bbio = bio->bi_private; + struct btrfs_bio *bbio = extract_bbio_from_bio_private(bio->bi_private); int is_orig_bio = 0; - if (err) + if (err) { atomic_inc(&bbio->error); + if (err == -EIO || err == -EREMOTEIO) { + unsigned int stripe_index = + extract_stripe_index_from_bio_private( + bio->bi_private); + struct btrfs_device *dev; + + BUG_ON(stripe_index >= bbio->num_stripes); + dev = bbio->stripes[stripe_index].dev; + if (bio->bi_rw & WRITE) + btrfs_dev_stat_inc(dev, + BTRFS_DEV_STAT_WRITE_ERRS); + else + btrfs_dev_stat_inc(dev, + BTRFS_DEV_STAT_READ_ERRS); + if ((bio->bi_rw & WRITE_FLUSH) == WRITE_FLUSH) + btrfs_dev_stat_inc(dev, + BTRFS_DEV_STAT_FLUSH_ERRS); + btrfs_dev_stat_print_on_error(dev); + } + } if (bio == bbio->orig_bio) is_orig_bio = 1; @@ -4149,6 +4195,8 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio, bio = first_bio; } bio->bi_private = bbio; + bio->bi_private = merge_stripe_index_into_bio_private( + bio->bi_private, (unsigned int)dev_nr); bio->bi_end_io = btrfs_end_bio; bio->bi_sector = bbio->stripes[dev_nr].physical >> 9; dev = bbio->stripes[dev_nr].dev; @@ -4509,6 +4557,28 @@ int btrfs_read_sys_array(struct btrfs_root *root) return ret; } +struct btrfs_device *btrfs_find_device_for_logical(struct btrfs_root *root, + u64 logical, int mirror_num) +{ + struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree; + int ret; + u64 map_length = 0; + struct btrfs_bio *bbio = NULL; + struct btrfs_device *device; + + BUG_ON(mirror_num == 0); + ret = btrfs_map_block(map_tree, WRITE, logical, &map_length, &bbio, + mirror_num); + if (ret) { + BUG_ON(bbio != NULL); + return NULL; + } + BUG_ON(mirror_num != bbio->mirror_num); + device = bbio->stripes[mirror_num - 1].dev; + kfree(bbio); + return device; +} + int btrfs_read_chunk_tree(struct btrfs_root *root) { struct btrfs_path *path; @@ -4583,3 +4653,23 @@ error: btrfs_free_path(path); return ret; } + +void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index) +{ + btrfs_dev_stat_inc(dev, index); + btrfs_dev_stat_print_on_error(dev); +} + +void btrfs_dev_stat_print_on_error(struct btrfs_device *dev) +{ + printk_ratelimited(KERN_ERR + "btrfs: bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n", + dev->name, + btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_WRITE_ERRS), + btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_READ_ERRS), + btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_FLUSH_ERRS), + btrfs_dev_stat_read(dev, + BTRFS_DEV_STAT_CORRUPTION_ERRS), + btrfs_dev_stat_read(dev, + BTRFS_DEV_STAT_GENERATION_ERRS)); +} diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h index bb6b03f97aa..193b2835e6a 100644 --- a/fs/btrfs/volumes.h +++ b/fs/btrfs/volumes.h @@ -22,6 +22,7 @@ #include #include #include "async-thread.h" +#include "ioctl.h" #define BTRFS_STRIPE_LEN (64 * 1024) @@ -106,6 +107,10 @@ struct btrfs_device { struct completion flush_wait; int nobarriers; + /* disk I/O failure stats. For detailed description refer to + * enum btrfs_dev_stat_values in ioctl.h */ + int dev_stats_dirty; /* counters need to be written to disk */ + atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX]; }; struct btrfs_fs_devices { @@ -281,4 +286,44 @@ int btrfs_cancel_balance(struct btrfs_fs_info *fs_info); int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset); int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes, u64 *start, u64 *max_avail); +struct btrfs_device *btrfs_find_device_for_logical(struct btrfs_root *root, + u64 logical, int mirror_num); +void btrfs_dev_stat_print_on_error(struct btrfs_device *device); +void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index); + +static inline void btrfs_dev_stat_inc(struct btrfs_device *dev, + int index) +{ + atomic_inc(dev->dev_stat_values + index); + dev->dev_stats_dirty = 1; +} + +static inline int btrfs_dev_stat_read(struct btrfs_device *dev, + int index) +{ + return atomic_read(dev->dev_stat_values + index); +} + +static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev, + int index) +{ + int ret; + + ret = atomic_xchg(dev->dev_stat_values + index, 0); + dev->dev_stats_dirty = 1; + return ret; +} + +static inline void btrfs_dev_stat_set(struct btrfs_device *dev, + int index, unsigned long val) +{ + atomic_set(dev->dev_stat_values + index, val); + dev->dev_stats_dirty = 1; +} + +static inline void btrfs_dev_stat_reset(struct btrfs_device *dev, + int index) +{ + btrfs_dev_stat_set(dev, index, 0); +} #endif -- cgit v1.2.3-70-g09d2 From 733f4fbbc1083aa343da739f46ee839705d6cfe3 Mon Sep 17 00:00:00 2001 From: Stefan Behrens Date: Fri, 25 May 2012 16:06:10 +0200 Subject: Btrfs: read device stats on mount, write modified ones during commit The device statistics are written into the device tree with each transaction commit. Only modified statistics are written. When a filesystem is mounted, the device statistics for each involved device are read from the device tree and used to initialize the counters. Signed-off-by: Stefan Behrens --- fs/btrfs/ctree.h | 38 +++++++++++ fs/btrfs/disk-io.c | 7 ++ fs/btrfs/print-tree.c | 3 + fs/btrfs/transaction.c | 4 ++ fs/btrfs/volumes.c | 176 +++++++++++++++++++++++++++++++++++++++++++++++++ fs/btrfs/volumes.h | 4 ++ 6 files changed, 232 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index aad2600718a..e176f8c551f 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -823,6 +823,14 @@ struct btrfs_csum_item { u8 csum; } __attribute__ ((__packed__)); +struct btrfs_dev_stats_item { + /* + * grow this item struct at the end for future enhancements and keep + * the existing values unchanged + */ + __le64 values[BTRFS_DEV_STAT_VALUES_MAX]; +} __attribute__ ((__packed__)); + /* different types of block groups (and chunks) */ #define BTRFS_BLOCK_GROUP_DATA (1ULL << 0) #define BTRFS_BLOCK_GROUP_SYSTEM (1ULL << 1) @@ -1507,6 +1515,12 @@ struct btrfs_ioctl_defrag_range_args { #define BTRFS_BALANCE_ITEM_KEY 248 +/* + * Persistantly stores the io stats in the device tree. + * One key for all stats, (0, BTRFS_DEV_STATS_KEY, devid). + */ +#define BTRFS_DEV_STATS_KEY 249 + /* * string items are for debugging. They just store a short string of * data in the FS @@ -2415,6 +2429,30 @@ static inline u32 btrfs_file_extent_inline_item_len(struct extent_buffer *eb, return btrfs_item_size(eb, e) - offset; } +/* btrfs_dev_stats_item */ +static inline u64 btrfs_dev_stats_value(struct extent_buffer *eb, + struct btrfs_dev_stats_item *ptr, + int index) +{ + u64 val; + + read_extent_buffer(eb, &val, + offsetof(struct btrfs_dev_stats_item, values) + + ((unsigned long)ptr) + (index * sizeof(u64)), + sizeof(val)); + return val; +} + +static inline void btrfs_set_dev_stats_value(struct extent_buffer *eb, + struct btrfs_dev_stats_item *ptr, + int index, u64 val) +{ + write_extent_buffer(eb, &val, + offsetof(struct btrfs_dev_stats_item, values) + + ((unsigned long)ptr) + (index * sizeof(u64)), + sizeof(val)); +} + static inline struct btrfs_fs_info *btrfs_sb(struct super_block *sb) { return sb->s_fs_info; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 46d474e74aa..b0d49e21b0b 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2354,6 +2354,13 @@ retry_root_backup: fs_info->generation = generation; fs_info->last_trans_committed = generation; + ret = btrfs_init_dev_stats(fs_info); + if (ret) { + printk(KERN_ERR "btrfs: failed to init dev_stats: %d\n", + ret); + goto fail_block_groups; + } + ret = btrfs_init_space_info(fs_info); if (ret) { printk(KERN_ERR "Failed to initial space info: %d\n", ret); diff --git a/fs/btrfs/print-tree.c b/fs/btrfs/print-tree.c index f38e452486b..5e23684887e 100644 --- a/fs/btrfs/print-tree.c +++ b/fs/btrfs/print-tree.c @@ -294,6 +294,9 @@ void btrfs_print_leaf(struct btrfs_root *root, struct extent_buffer *l) btrfs_dev_extent_chunk_offset(l, dev_extent), (unsigned long long) btrfs_dev_extent_length(l, dev_extent)); + case BTRFS_DEV_STATS_KEY: + printk(KERN_INFO "\t\tdevice stats\n"); + break; }; } } diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 36422254ef6..82b03afcbd9 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -28,6 +28,7 @@ #include "locking.h" #include "tree-log.h" #include "inode-map.h" +#include "volumes.h" #define BTRFS_ROOT_TRANS_TAG 0 @@ -758,6 +759,9 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans, if (ret) return ret; + ret = btrfs_run_dev_stats(trans, root->fs_info); + BUG_ON(ret); + while (!list_empty(&fs_info->dirty_cowonly_roots)) { next = fs_info->dirty_cowonly_roots.next; list_del_init(next); diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index a112b758822..7782020996f 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -40,6 +40,8 @@ static int init_first_rw_device(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_device *device); static int btrfs_relocate_sys_chunks(struct btrfs_root *root); +static void __btrfs_reset_dev_stats(struct btrfs_device *dev); +static void btrfs_dev_stat_print_on_load(struct btrfs_device *device); static DEFINE_MUTEX(uuid_mutex); static LIST_HEAD(fs_uuids); @@ -362,6 +364,7 @@ static noinline int device_list_add(const char *path, return -ENOMEM; } device->devid = devid; + device->dev_stats_valid = 0; device->work.func = pending_bios_fn; memcpy(device->uuid, disk_super->dev_item.uuid, BTRFS_UUID_SIZE); @@ -4654,6 +4657,162 @@ error: return ret; } +static void __btrfs_reset_dev_stats(struct btrfs_device *dev) +{ + int i; + + for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) + btrfs_dev_stat_reset(dev, i); +} + +int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info) +{ + struct btrfs_key key; + struct btrfs_key found_key; + struct btrfs_root *dev_root = fs_info->dev_root; + struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; + struct extent_buffer *eb; + int slot; + int ret = 0; + struct btrfs_device *device; + struct btrfs_path *path = NULL; + int i; + + path = btrfs_alloc_path(); + if (!path) { + ret = -ENOMEM; + goto out; + } + + mutex_lock(&fs_devices->device_list_mutex); + list_for_each_entry(device, &fs_devices->devices, dev_list) { + int item_size; + struct btrfs_dev_stats_item *ptr; + + key.objectid = 0; + key.type = BTRFS_DEV_STATS_KEY; + key.offset = device->devid; + ret = btrfs_search_slot(NULL, dev_root, &key, path, 0, 0); + if (ret) { + printk(KERN_WARNING "btrfs: no dev_stats entry found for device %s (devid %llu) (OK on first mount after mkfs)\n", + device->name, (unsigned long long)device->devid); + __btrfs_reset_dev_stats(device); + device->dev_stats_valid = 1; + btrfs_release_path(path); + continue; + } + slot = path->slots[0]; + eb = path->nodes[0]; + btrfs_item_key_to_cpu(eb, &found_key, slot); + item_size = btrfs_item_size_nr(eb, slot); + + ptr = btrfs_item_ptr(eb, slot, + struct btrfs_dev_stats_item); + + for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) { + if (item_size >= (1 + i) * sizeof(__le64)) + btrfs_dev_stat_set(device, i, + btrfs_dev_stats_value(eb, ptr, i)); + else + btrfs_dev_stat_reset(device, i); + } + + device->dev_stats_valid = 1; + btrfs_dev_stat_print_on_load(device); + btrfs_release_path(path); + } + mutex_unlock(&fs_devices->device_list_mutex); + +out: + btrfs_free_path(path); + return ret < 0 ? ret : 0; +} + +static int update_dev_stat_item(struct btrfs_trans_handle *trans, + struct btrfs_root *dev_root, + struct btrfs_device *device) +{ + struct btrfs_path *path; + struct btrfs_key key; + struct extent_buffer *eb; + struct btrfs_dev_stats_item *ptr; + int ret; + int i; + + key.objectid = 0; + key.type = BTRFS_DEV_STATS_KEY; + key.offset = device->devid; + + path = btrfs_alloc_path(); + BUG_ON(!path); + ret = btrfs_search_slot(trans, dev_root, &key, path, -1, 1); + if (ret < 0) { + printk(KERN_WARNING "btrfs: error %d while searching for dev_stats item for device %s!\n", + ret, device->name); + goto out; + } + + if (ret == 0 && + btrfs_item_size_nr(path->nodes[0], path->slots[0]) < sizeof(*ptr)) { + /* need to delete old one and insert a new one */ + ret = btrfs_del_item(trans, dev_root, path); + if (ret != 0) { + printk(KERN_WARNING "btrfs: delete too small dev_stats item for device %s failed %d!\n", + device->name, ret); + goto out; + } + ret = 1; + } + + if (ret == 1) { + /* need to insert a new item */ + btrfs_release_path(path); + ret = btrfs_insert_empty_item(trans, dev_root, path, + &key, sizeof(*ptr)); + if (ret < 0) { + printk(KERN_WARNING "btrfs: insert dev_stats item for device %s failed %d!\n", + device->name, ret); + goto out; + } + } + + eb = path->nodes[0]; + ptr = btrfs_item_ptr(eb, path->slots[0], struct btrfs_dev_stats_item); + for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) + btrfs_set_dev_stats_value(eb, ptr, i, + btrfs_dev_stat_read(device, i)); + btrfs_mark_buffer_dirty(eb); + +out: + btrfs_free_path(path); + return ret; +} + +/* + * called from commit_transaction. Writes all changed device stats to disk. + */ +int btrfs_run_dev_stats(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info) +{ + struct btrfs_root *dev_root = fs_info->dev_root; + struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; + struct btrfs_device *device; + int ret = 0; + + mutex_lock(&fs_devices->device_list_mutex); + list_for_each_entry(device, &fs_devices->devices, dev_list) { + if (!device->dev_stats_valid || !device->dev_stats_dirty) + continue; + + ret = update_dev_stat_item(trans, dev_root, device); + if (!ret) + device->dev_stats_dirty = 0; + } + mutex_unlock(&fs_devices->device_list_mutex); + + return ret; +} + void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index) { btrfs_dev_stat_inc(dev, index); @@ -4662,6 +4821,8 @@ void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index) void btrfs_dev_stat_print_on_error(struct btrfs_device *dev) { + if (!dev->dev_stats_valid) + return; printk_ratelimited(KERN_ERR "btrfs: bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n", dev->name, @@ -4674,6 +4835,17 @@ void btrfs_dev_stat_print_on_error(struct btrfs_device *dev) BTRFS_DEV_STAT_GENERATION_ERRS)); } +static void btrfs_dev_stat_print_on_load(struct btrfs_device *dev) +{ + printk(KERN_INFO "btrfs: bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n", + dev->name, + btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_WRITE_ERRS), + btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_READ_ERRS), + btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_FLUSH_ERRS), + btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_CORRUPTION_ERRS), + btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_GENERATION_ERRS)); +} + int btrfs_get_dev_stats(struct btrfs_root *root, struct btrfs_ioctl_get_dev_stats *stats, int reset_after_read) @@ -4690,6 +4862,10 @@ int btrfs_get_dev_stats(struct btrfs_root *root, printk(KERN_WARNING "btrfs: get dev_stats failed, device not found\n"); return -ENODEV; + } else if (!dev->dev_stats_valid) { + printk(KERN_WARNING + "btrfs: get dev_stats failed, not yet valid\n"); + return -ENODEV; } else if (reset_after_read) { for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) { if (stats->nr_items > i) diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h index 6798f8674b1..3406a88ca83 100644 --- a/fs/btrfs/volumes.h +++ b/fs/btrfs/volumes.h @@ -109,6 +109,7 @@ struct btrfs_device { /* disk I/O failure stats. For detailed description refer to * enum btrfs_dev_stat_values in ioctl.h */ + int dev_stats_valid; int dev_stats_dirty; /* counters need to be written to disk */ atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX]; }; @@ -293,6 +294,9 @@ void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index); int btrfs_get_dev_stats(struct btrfs_root *root, struct btrfs_ioctl_get_dev_stats *stats, int reset_after_read); +int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info); +int btrfs_run_dev_stats(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info); static inline void btrfs_dev_stat_inc(struct btrfs_device *dev, int index) -- cgit v1.2.3-70-g09d2