From c3b9a62c8f932f32a733d6b628f61f3f28345727 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 18 Aug 2010 05:29:12 -0400 Subject: btrfs: replace barriers with explicit flush / FUA usage Switch to the WRITE_FLUSH_FUA flag for log writes, remove the EOPNOTSUPP detection for barriers and stop setting the barrier flag for discards. Signed-off-by: Christoph Hellwig Acked-by: Chris Mason Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- fs/btrfs/disk-io.c | 19 ++++--------------- 1 file changed, 4 insertions(+), 15 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 64f10082f048..5e789f4a3ed0 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2063,7 +2063,7 @@ static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate) if (uptodate) { set_buffer_uptodate(bh); } else { - if (!buffer_eopnotsupp(bh) && printk_ratelimit()) { + if (printk_ratelimit()) { printk(KERN_WARNING "lost page write due to " "I/O error on %s\n", bdevname(bh->b_bdev, b)); @@ -2200,21 +2200,10 @@ static int write_dev_supers(struct btrfs_device *device, bh->b_end_io = btrfs_end_buffer_write_sync; } - if (i == last_barrier && do_barriers && device->barriers) { - ret = submit_bh(WRITE_BARRIER, bh); - if (ret == -EOPNOTSUPP) { - printk("btrfs: disabling barriers on dev %s\n", - device->name); - set_buffer_uptodate(bh); - device->barriers = 0; - /* one reference for submit_bh */ - get_bh(bh); - lock_buffer(bh); - ret = submit_bh(WRITE_SYNC, bh); - } - } else { + if (i == last_barrier && do_barriers) + ret = submit_bh(WRITE_FLUSH_FUA, bh); + else ret = submit_bh(WRITE_SYNC, bh); - } if (ret) errors++; -- cgit v1.2.2 From 0af3d00bad38d3bb9912a60928ad0669f17bdb76 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 21 Jun 2010 14:48:16 -0400 Subject: Btrfs: create special free space cache inode In order to save free space cache, we need an inode to hold the data, and we need a special item to point at the right inode for the right block group. So first, create a special item that will point to the right inode, and the number of extent entries we will have and the number of bitmaps we will have. We truncate and pre-allocate space everytime to make sure it's uptodate. This feature will be turned on as soon as you mount with -o space_cache, however it is safe to boot into old kernels, they will just generate the cache the old fashion way. When you boot back into a newer kernel we will notice that we modified and not the cache and automatically discard the cache. Signed-off-by: Josef Bacik --- fs/btrfs/disk-io.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 64f10082f048..45cf64fc1e3e 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1685,7 +1685,6 @@ struct btrfs_root *open_ctree(struct super_block *sb, __setup_root(4096, 4096, 4096, 4096, tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID); - bh = btrfs_read_dev_super(fs_devices->latest_bdev); if (!bh) goto fail_iput; @@ -1993,6 +1992,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, if (!(sb->s_flags & MS_RDONLY)) { down_read(&fs_info->cleanup_work_sem); btrfs_orphan_cleanup(fs_info->fs_root); + btrfs_orphan_cleanup(fs_info->tree_root); up_read(&fs_info->cleanup_work_sem); } @@ -2421,6 +2421,7 @@ int close_ctree(struct btrfs_root *root) fs_info->closing = 1; smp_mb(); + btrfs_put_block_group_cache(fs_info); if (!(fs_info->sb->s_flags & MS_RDONLY)) { ret = btrfs_commit_super(root); if (ret) -- cgit v1.2.2 From 0cb59c9953171e9adf6da8142a5c85ceb77bb60d Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 2 Jul 2010 12:14:14 -0400 Subject: Btrfs: write out free space cache This is a simple bit, just dump the free space cache out to our preallocated inode when we're writing out dirty block groups. There are a bunch of changes in inode.c in order to account for special cases. Mostly when we're doing the writeout we're holding trans_mutex, so we need to use the nolock transacation functions. Also we can't do asynchronous completions since the async thread could be blocked on already completed IO waiting for the transaction lock. This has been tested with xfstests and btrfs filesystem balance, as well as my ENOSPC tests. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/disk-io.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 45cf64fc1e3e..77e5dabfd45a 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -481,9 +481,12 @@ static void end_workqueue_bio(struct bio *bio, int err) end_io_wq->work.flags = 0; if (bio->bi_rw & REQ_WRITE) { - if (end_io_wq->metadata) + if (end_io_wq->metadata == 1) btrfs_queue_worker(&fs_info->endio_meta_write_workers, &end_io_wq->work); + else if (end_io_wq->metadata == 2) + btrfs_queue_worker(&fs_info->endio_freespace_worker, + &end_io_wq->work); else btrfs_queue_worker(&fs_info->endio_write_workers, &end_io_wq->work); @@ -497,6 +500,13 @@ static void end_workqueue_bio(struct bio *bio, int err) } } +/* + * For the metadata arg you want + * + * 0 - if data + * 1 - if normal metadta + * 2 - if writing to the free space cache area + */ int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio, int metadata) { @@ -1774,6 +1784,8 @@ struct btrfs_root *open_ctree(struct super_block *sb, btrfs_init_workers(&fs_info->endio_write_workers, "endio-write", fs_info->thread_pool_size, &fs_info->generic_worker); + btrfs_init_workers(&fs_info->endio_freespace_worker, "freespace-write", + 1, &fs_info->generic_worker); /* * endios are largely parallel and should have a very @@ -1794,6 +1806,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, btrfs_start_workers(&fs_info->endio_meta_workers, 1); btrfs_start_workers(&fs_info->endio_meta_write_workers, 1); btrfs_start_workers(&fs_info->endio_write_workers, 1); + btrfs_start_workers(&fs_info->endio_freespace_worker, 1); fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super); fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages, @@ -2035,6 +2048,7 @@ fail_sb_buffer: btrfs_stop_workers(&fs_info->endio_meta_workers); btrfs_stop_workers(&fs_info->endio_meta_write_workers); btrfs_stop_workers(&fs_info->endio_write_workers); + btrfs_stop_workers(&fs_info->endio_freespace_worker); btrfs_stop_workers(&fs_info->submit_workers); fail_iput: invalidate_inode_pages2(fs_info->btree_inode->i_mapping); @@ -2468,6 +2482,7 @@ int close_ctree(struct btrfs_root *root) btrfs_stop_workers(&fs_info->endio_meta_workers); btrfs_stop_workers(&fs_info->endio_meta_write_workers); btrfs_stop_workers(&fs_info->endio_write_workers); + btrfs_stop_workers(&fs_info->endio_freespace_worker); btrfs_stop_workers(&fs_info->submit_workers); btrfs_close_devices(fs_info->fs_devices); -- cgit v1.2.2 From 559af8211433b8c0b20e6c43c61409cb9c9c2996 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Fri, 29 Oct 2010 15:14:37 -0400 Subject: Btrfs: cleanup warnings from gcc 4.6 (nonbugs) These are all the cases where a variable is set, but not read which are not bugs as far as I can see, but simply leftovers. Still needs more review. Found by gcc 4.6's new warnings Signed-off-by: Andi Kleen Cc: Chris Mason Signed-off-by: Andrew Morton Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 11 ----------- 1 file changed, 11 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 77e5dabfd45a..e163424c7fce 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -338,7 +338,6 @@ 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; - int found_level; unsigned long len; struct extent_buffer *eb; int ret; @@ -369,8 +368,6 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) WARN_ON(1); goto err; } - found_level = btrfs_header_level(eb); - csum_tree_block(root, eb, 0); err: free_extent_buffer(eb); @@ -543,11 +540,9 @@ int btrfs_congested_async(struct btrfs_fs_info *info, int iodone) static void run_one_async_start(struct btrfs_work *work) { - struct btrfs_fs_info *fs_info; struct async_submit_bio *async; async = container_of(work, struct async_submit_bio, work); - fs_info = BTRFS_I(async->inode)->root->fs_info; async->submit_bio_start(async->inode, async->rw, async->bio, async->mirror_num, async->bio_flags, async->bio_offset); @@ -860,12 +855,8 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize, u64 parent_transid) { struct extent_buffer *buf = NULL; - struct inode *btree_inode = root->fs_info->btree_inode; - struct extent_io_tree *io_tree; int ret; - io_tree = &BTRFS_I(btree_inode)->io_tree; - buf = btrfs_find_create_tree_block(root, bytenr, blocksize); if (!buf) return NULL; @@ -1387,7 +1378,6 @@ static int bio_ready_for_csum(struct bio *bio) u64 start = 0; struct page *page; struct extent_io_tree *io_tree = NULL; - struct btrfs_fs_info *info = NULL; struct bio_vec *bvec; int i; int ret; @@ -1406,7 +1396,6 @@ static int bio_ready_for_csum(struct bio *bio) buf_len = page->private >> 2; start = page_offset(page) + bvec->bv_offset; io_tree = &BTRFS_I(page->mapping->host)->io_tree; - info = BTRFS_I(page->mapping->host)->root->fs_info; } /* are we fully contained in this bio? */ if (buf_len <= length) -- cgit v1.2.2 From bb9c12c945cbd1b0eaa1589546dde772ccabeeba Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Fri, 29 Oct 2010 15:37:34 -0400 Subject: Btrfs: async transaction commit Add support for an async transaction commit that is ordered such that any subsequent operations will join the following transaction, but does not wait until the current commit is fully on disk. This avoids much of the latency associated with the btrfs_commit_transaction for callers concerned with serialization and not safety. The wait_for_unblock flag controls whether we wait for the 'middle' portion of commit_transaction to complete, which is necessary if the caller expects some of the modifications contained in the commit to be available (this is the case for subvol/snapshot creation). Signed-off-by: Sage Weil 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 e163424c7fce..b40dfe48017b 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1679,6 +1679,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, init_waitqueue_head(&fs_info->transaction_throttle); init_waitqueue_head(&fs_info->transaction_wait); + init_waitqueue_head(&fs_info->transaction_blocked_wait); init_waitqueue_head(&fs_info->async_submit_wait); __setup_root(4096, 4096, 4096, 4096, tree_root, -- cgit v1.2.2 From 784b4e29a26617589edd290dd2919735e190c06e Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Sun, 21 Nov 2010 22:20:49 -0500 Subject: Btrfs: add migrate page for metadata inode Migrate page will directly call the btrfs btree writepage function, which isn't actually allowed. Our writepage assumes that you have locked the extent_buffer and flagged the block as written. Without doing these steps, we can corrupt metadata blocks. A later commit will remove the btree writepage function since it is really only safely used internally by btrfs. We use writepages for everything else. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 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 b40dfe48017b..a67b98d58c2a 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -28,6 +28,7 @@ #include #include #include +#include #include "compat.h" #include "ctree.h" #include "disk-io.h" @@ -355,6 +356,8 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) 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)); + found_start = btrfs_header_bytenr(eb); if (found_start != start) { WARN_ON(1); @@ -693,6 +696,26 @@ static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, __btree_submit_bio_done); } +static int btree_migratepage(struct address_space *mapping, + struct page *newpage, struct page *page) +{ + /* + * we can't safely write a btree page from here, + * we haven't done the locking hook + */ + if (PageDirty(page)) + return -EAGAIN; + /* + * Buffers may be managed in a filesystem specific way. + * We must have no buffers or drop them. + */ + if (page_has_private(page) && + !try_to_release_page(page, GFP_KERNEL)) + return -EAGAIN; + + return migrate_page(mapping, newpage, page); +} + static int btree_writepage(struct page *page, struct writeback_control *wbc) { struct extent_io_tree *tree; @@ -707,8 +730,7 @@ static int btree_writepage(struct page *page, struct writeback_control *wbc) } redirty_page_for_writepage(wbc, page); - eb = btrfs_find_tree_block(root, page_offset(page), - PAGE_CACHE_SIZE); + 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); @@ -799,6 +821,7 @@ static const struct address_space_operations btree_aops = { .releasepage = btree_releasepage, .invalidatepage = btree_invalidatepage, .sync_page = block_sync_page, + .migratepage = btree_migratepage, }; int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize, -- cgit v1.2.2 From 450ba0ea06b6ed3612d27f2b7127a9de4160f285 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 19 Nov 2010 14:59:15 -0500 Subject: Btrfs: setup blank root and fs_info for mount time There is a problem with how we use sget, it searches through the list of supers attached to the fs_type looking for a super with the same fs_devices as what we're trying to mount. This depends on sb->s_fs_info being filled, but we don't fill that in until we get to btrfs_fill_super, so we could hit supers on the fs_type super list that have a null s_fs_info. In order to fix that we need to go ahead and setup a blank root with a blank fs_info to hold fs_devices, that way our test will work out right and then we can set s_fs_info in btrfs_set_super, and then open_ctree will simply use our pre-allocated root and fs_info when setting everything up. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 6 ++---- 1 file changed, 2 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 a67b98d58c2a..57c9d8eeb7dc 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1561,10 +1561,8 @@ struct btrfs_root *open_ctree(struct super_block *sb, GFP_NOFS); struct btrfs_root *csum_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS); - struct btrfs_root *tree_root = kzalloc(sizeof(struct btrfs_root), - GFP_NOFS); - struct btrfs_fs_info *fs_info = kzalloc(sizeof(*fs_info), - GFP_NOFS); + struct btrfs_root *tree_root = btrfs_sb(sb); + struct btrfs_fs_info *fs_info = tree_root->fs_info; struct btrfs_root *chunk_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS); struct btrfs_root *dev_root = kzalloc(sizeof(struct btrfs_root), -- cgit v1.2.2 From 5a92bc88cef279261d3f138e25850c122df67045 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 29 Nov 2010 09:49:11 -0500 Subject: Btrfs: don't use migrate page without CONFIG_MIGRATION Fixes compile error Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 57c9d8eeb7dc..33b6d459494c 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -712,8 +712,11 @@ static int btree_migratepage(struct address_space *mapping, if (page_has_private(page) && !try_to_release_page(page, GFP_KERNEL)) return -EAGAIN; - +#ifdef CONFIG_MIGRATION return migrate_page(mapping, newpage, page); +#else + return -ENOSYS; +#endif } static int btree_writepage(struct page *page, struct writeback_control *wbc) @@ -821,7 +824,9 @@ static const struct address_space_operations btree_aops = { .releasepage = btree_releasepage, .invalidatepage = btree_invalidatepage, .sync_page = block_sync_page, +#ifdef CONFIG_MIGRATION .migratepage = btree_migratepage, +#endif }; int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize, -- cgit v1.2.2 From 3dd1462e82bcab7625cec129952f26dae7a8b742 Mon Sep 17 00:00:00 2001 From: Jan Beulich Date: Tue, 7 Dec 2010 14:54:09 +0000 Subject: Btrfs: fix compiler warnings ... regarding an unused function when !MIGRATION, and regarding a printk() format string vs argument mismatch. Signed-off-by: Jan Beulich Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 6 ++---- 1 file changed, 2 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 33b6d459494c..b803c2667673 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -696,6 +696,7 @@ static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio, __btree_submit_bio_done); } +#ifdef CONFIG_MIGRATION static int btree_migratepage(struct address_space *mapping, struct page *newpage, struct page *page) { @@ -712,12 +713,9 @@ static int btree_migratepage(struct address_space *mapping, if (page_has_private(page) && !try_to_release_page(page, GFP_KERNEL)) return -EAGAIN; -#ifdef CONFIG_MIGRATION return migrate_page(mapping, newpage, page); -#else - return -ENOSYS; -#endif } +#endif static int btree_writepage(struct page *page, struct writeback_control *wbc) { -- cgit v1.2.2 From 68433b73b104bff388aac376631d32abbbd872b0 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 13 Dec 2010 14:47:58 -0500 Subject: Btrfs: EIO when we fail to read tree roots If we just get a plain IO error when we read tree roots, the code wasn't properly sending that error up the chain. This allowed mounts to continue when they should failed, and allowed operations on partially setup root structs. The end result was usually oopsen on spinlocks that hadn't been spun up correctly. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index b803c2667673..a5d2249e6da5 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1007,7 +1007,10 @@ static int find_and_setup_root(struct btrfs_root *tree_root, blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item)); root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item), blocksize, generation); - BUG_ON(!root->node); + if (!root->node || !btrfs_buffer_uptodate(root->node, generation)) { + free_extent_buffer(root->node); + return -EIO; + } root->commit_root = btrfs_root_node(root); return 0; } -- cgit v1.2.2 From a6fa6fae40ec336c7df6155255ae64ebef43a8bc Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Mon, 25 Oct 2010 15:12:26 +0800 Subject: btrfs: Add lzo compression support Lzo is a much faster compression algorithm than gzib, so would allow more users to enable transparent compression, and some users can choose from compression ratio and speed for different applications Usage: # mount -t btrfs -o compress[=] dev /mnt or # mount -t btrfs -o compress-force[=] dev /mnt "-o compress" without argument is still allowed for compatability. Compatibility: If we mount a filesystem with lzo compression, it will not be able be mounted in old kernels. One reason is, otherwise btrfs will directly dump compressed data, which sits in inline extent, to user. Performance: The test copied a linux source tarball (~400M) from an ext4 partition to the btrfs partition, and then extracted it. (time in second) lzo zlib nocompress copy: 10.6 21.7 14.9 extract: 70.1 94.4 66.6 (data size in MB) lzo zlib nocompress copy: 185.87 108.69 394.49 extract: 193.80 132.36 381.21 Changelog: v1 -> v2: - Select LZO_COMPRESS and LZO_DECOMPRESS in btrfs Kconfig. - Add incompability flag. - Fix error handling in compress code. Signed-off-by: Li Zefan --- 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 a5d2249e6da5..f88eb2ce7919 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1744,10 +1744,10 @@ struct btrfs_root *open_ctree(struct super_block *sb, } features = btrfs_super_incompat_flags(disk_super); - if (!(features & BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF)) { - features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF; - btrfs_set_super_incompat_flags(disk_super, features); - } + features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF; + if (tree_root->fs_info->compress_type & BTRFS_COMPRESS_LZO) + features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO; + btrfs_set_super_incompat_flags(disk_super, features); features = btrfs_super_compat_ro_flags(disk_super) & ~BTRFS_FEATURE_COMPAT_RO_SUPP; -- cgit v1.2.2 From 20b450773d17e325190c158e10bfdb25dc21d2d6 Mon Sep 17 00:00:00 2001 From: Dave Young Date: Sat, 8 Jan 2011 10:09:13 +0000 Subject: btrfs: mount failure return value fix I happened to pass swap partition as root partition in cmdline, then kernel panic and tell me about "Cannot open root device". It is not correct, in fact it is a fs type mismatch instead of 'no device'. Eventually I found btrfs mounting failed with -EIO, it should be -EINVAL. The logic in init/do_mounts.c: for (p = fs_names; *p; p += strlen(p)+1) { int err = do_mount_root(name, p, flags, root_mount_data); switch (err) { case 0: goto out; case -EACCES: flags |= MS_RDONLY; goto retry; case -EINVAL: continue; } print "Cannot open root device" panic } SO fs type after btrfs will have no chance to mount Here fix the return value as -EINVAL Signed-off-by: Dave Young Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index f88eb2ce7919..f9efb68fc2e3 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1713,8 +1713,10 @@ struct btrfs_root *open_ctree(struct super_block *sb, fs_info, BTRFS_ROOT_TREE_OBJECTID); bh = btrfs_read_dev_super(fs_devices->latest_bdev); - if (!bh) + if (!bh) { + err = -EINVAL; goto fail_iput; + } memcpy(&fs_info->super_copy, bh->b_data, sizeof(fs_info->super_copy)); memcpy(&fs_info->super_for_commit, &fs_info->super_copy, -- cgit v1.2.2 From 91ca338d776e0cefb255bf2979b6448febd880f5 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Wed, 5 Jan 2011 02:32:22 +0000 Subject: btrfs: check NULL or not Should check if functions returns NULL or not. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- 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 f9efb68fc2e3..a0c37b2ee9ed 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -353,6 +353,10 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) WARN_ON(len == 0); eb = alloc_extent_buffer(tree, start, len, page, GFP_NOFS); + 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); @@ -427,6 +431,10 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, WARN_ON(len == 0); eb = alloc_extent_buffer(tree, start, len, page, GFP_NOFS); + if (eb == NULL) { + ret = -EIO; + goto out; + } found_start = btrfs_header_bytenr(eb); if (found_start != start) { -- cgit v1.2.2 From 5e540f7715b8cd83b8e60beaaa525b125cc122de Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Mon, 27 Dec 2010 06:53:10 +0000 Subject: btrfs: Fix memory leak in btrfs_read_fs_root_no_radix() In btrfs_read_fs_root_no_radix(), 'root' is not freed if btrfs_search_slot() returns error. Signed-off-by: Tsutomu Itoh 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 a0c37b2ee9ed..9b1dd4138072 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1153,6 +1153,7 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root, } btrfs_free_path(path); if (ret) { + kfree(root); if (ret > 0) ret = -ENOENT; return ERR_PTR(ret); -- cgit v1.2.2 From acce952b0263825da32cf10489413dec78053347 Mon Sep 17 00:00:00 2001 From: liubo Date: Thu, 6 Jan 2011 19:30:25 +0800 Subject: Btrfs: forced readonly mounts on errors This patch comes from "Forced readonly mounts on errors" ideas. As we know, this is the first step in being more fault tolerant of disk corruptions instead of just using BUG() statements. The major content: - add a framework for generating errors that should result in filesystems going readonly. - keep FS state in disk super block. - make sure that all of resource will be freed and released at umount time. - make sure that fter FS is forced readonly on error, there will be no more disk change before FS is corrected. For this, we should stop write operation. After this patch is applied, the conversion from BUG() to such a framework can happen incrementally. Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 391 ++++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 389 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 9b1dd4138072..1a3af9e8e0c4 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -44,6 +44,20 @@ 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, + int read_only); +static int btrfs_destroy_ordered_operations(struct btrfs_root *root); +static int 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 int btrfs_destroy_marked_extents(struct btrfs_root *root, + struct extent_io_tree *dirty_pages, + 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 @@ -1738,6 +1752,11 @@ struct btrfs_root *open_ctree(struct super_block *sb, if (!btrfs_super_root(disk_super)) goto fail_iput; + /* 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_parse_options(tree_root, options); if (ret) { err = ret; @@ -1968,7 +1987,9 @@ struct btrfs_root *open_ctree(struct super_block *sb, btrfs_set_opt(fs_info->mount_opt, SSD); } - if (btrfs_super_log_root(disk_super) != 0) { + /* do not make disk changes in broken FS */ + if (btrfs_super_log_root(disk_super) != 0 && + !(fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)) { u64 bytenr = btrfs_super_log_root(disk_super); if (fs_devices->rw_devices == 0) { @@ -2464,8 +2485,28 @@ int close_ctree(struct btrfs_root *root) smp_mb(); btrfs_put_block_group_cache(fs_info); + + /* + * Here come 2 situations when btrfs is broken to flip readonly: + * + * 1. when btrfs flips readonly somewhere else before + * btrfs_commit_super, sb->s_flags has MS_RDONLY flag, + * and btrfs will skip to write sb directly to keep + * ERROR state on disk. + * + * 2. when btrfs flips readonly just in btrfs_commit_super, + * and in such case, btrfs cannnot write sb via btrfs_commit_super, + * and since fs_state has been set BTRFS_SUPER_FLAG_ERROR flag, + * btrfs will cleanup all FS resources first and write sb then. + */ if (!(fs_info->sb->s_flags & MS_RDONLY)) { - ret = btrfs_commit_super(root); + ret = btrfs_commit_super(root); + if (ret) + printk(KERN_ERR "btrfs: commit super ret %d\n", ret); + } + + if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + ret = btrfs_error_commit_super(root); if (ret) printk(KERN_ERR "btrfs: commit super ret %d\n", ret); } @@ -2641,6 +2682,352 @@ out: return 0; } +static void btrfs_check_super_valid(struct btrfs_fs_info *fs_info, + int read_only) +{ + if (read_only) + return; + + if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) + printk(KERN_WARNING "warning: mount fs with errors, " + "running btrfsck is recommended\n"); +} + +int btrfs_error_commit_super(struct btrfs_root *root) +{ + int ret; + + mutex_lock(&root->fs_info->cleaner_mutex); + btrfs_run_delayed_iputs(root); + mutex_unlock(&root->fs_info->cleaner_mutex); + + down_write(&root->fs_info->cleanup_work_sem); + up_write(&root->fs_info->cleanup_work_sem); + + /* cleanup FS via transaction */ + btrfs_cleanup_transaction(root); + + ret = write_ctree_super(NULL, root, 0); + + return ret; +} + +static int btrfs_destroy_ordered_operations(struct btrfs_root *root) +{ + struct btrfs_inode *btrfs_inode; + struct list_head splice; + + INIT_LIST_HEAD(&splice); + + mutex_lock(&root->fs_info->ordered_operations_mutex); + spin_lock(&root->fs_info->ordered_extent_lock); + + list_splice_init(&root->fs_info->ordered_operations, &splice); + while (!list_empty(&splice)) { + btrfs_inode = list_entry(splice.next, struct btrfs_inode, + ordered_operations); + + list_del_init(&btrfs_inode->ordered_operations); + + btrfs_invalidate_inodes(btrfs_inode->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) +{ + struct list_head splice; + struct btrfs_ordered_extent *ordered; + struct inode *inode; + + INIT_LIST_HEAD(&splice); + + spin_lock(&root->fs_info->ordered_extent_lock); + + list_splice_init(&root->fs_info->ordered_extents, &splice); + while (!list_empty(&splice)) { + ordered = list_entry(splice.next, struct btrfs_ordered_extent, + root_extent_list); + + list_del_init(&ordered->root_extent_list); + atomic_inc(&ordered->refs); + + /* the inode may be getting freed (in sys_unlink path). */ + inode = igrab(ordered->inode); + + spin_unlock(&root->fs_info->ordered_extent_lock); + if (inode) + iput(inode); + + atomic_set(&ordered->refs, 1); + btrfs_put_ordered_extent(ordered); + + spin_lock(&root->fs_info->ordered_extent_lock); + } + + spin_unlock(&root->fs_info->ordered_extent_lock); + + return 0; +} + +static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, + struct btrfs_root *root) +{ + struct rb_node *node; + struct btrfs_delayed_ref_root *delayed_refs; + struct btrfs_delayed_ref_node *ref; + int ret = 0; + + delayed_refs = &trans->delayed_refs; + + spin_lock(&delayed_refs->lock); + if (delayed_refs->num_entries == 0) { + printk(KERN_INFO "delayed_refs has NO entry\n"); + return ret; + } + + node = rb_first(&delayed_refs->root); + while (node) { + ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node); + node = rb_next(node); + + ref->in_tree = 0; + rb_erase(&ref->rb_node, &delayed_refs->root); + delayed_refs->num_entries--; + + atomic_set(&ref->refs, 1); + if (btrfs_delayed_ref_is_head(ref)) { + struct btrfs_delayed_ref_head *head; + + head = btrfs_delayed_node_to_head(ref); + mutex_lock(&head->mutex); + kfree(head->extent_op); + delayed_refs->num_heads--; + if (list_empty(&head->cluster)) + delayed_refs->num_heads_ready--; + list_del_init(&head->cluster); + mutex_unlock(&head->mutex); + } + + spin_unlock(&delayed_refs->lock); + btrfs_put_delayed_ref(ref); + + cond_resched(); + spin_lock(&delayed_refs->lock); + } + + spin_unlock(&delayed_refs->lock); + + return ret; +} + +static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t) +{ + struct btrfs_pending_snapshot *snapshot; + struct list_head splice; + + INIT_LIST_HEAD(&splice); + + list_splice_init(&t->pending_snapshots, &splice); + + while (!list_empty(&splice)) { + snapshot = list_entry(splice.next, + struct btrfs_pending_snapshot, + list); + + list_del_init(&snapshot->list); + + kfree(snapshot); + } + + return 0; +} + +static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root) +{ + struct btrfs_inode *btrfs_inode; + struct list_head splice; + + INIT_LIST_HEAD(&splice); + + list_splice_init(&root->fs_info->delalloc_inodes, &splice); + + spin_lock(&root->fs_info->delalloc_lock); + + while (!list_empty(&splice)) { + btrfs_inode = list_entry(splice.next, struct btrfs_inode, + delalloc_inodes); + + list_del_init(&btrfs_inode->delalloc_inodes); + + btrfs_invalidate_inodes(btrfs_inode->root); + } + + spin_unlock(&root->fs_info->delalloc_lock); + + return 0; +} + +static int btrfs_destroy_marked_extents(struct btrfs_root *root, + struct extent_io_tree *dirty_pages, + int mark) +{ + int ret; + struct page *page; + struct inode *btree_inode = root->fs_info->btree_inode; + struct extent_buffer *eb; + u64 start = 0; + u64 end; + u64 offset; + unsigned long index; + + while (1) { + ret = find_first_extent_bit(dirty_pages, start, &start, &end, + mark); + if (ret) + break; + + clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS); + while (start <= end) { + index = start >> PAGE_CACHE_SHIFT; + start = (u64)(index + 1) << PAGE_CACHE_SHIFT; + page = find_get_page(btree_inode->i_mapping, index); + if (!page) + continue; + offset = page_offset(page); + + spin_lock(&dirty_pages->buffer_lock); + eb = radix_tree_lookup( + &(&BTRFS_I(page->mapping->host)->io_tree)->buffer, + offset >> PAGE_CACHE_SHIFT); + spin_unlock(&dirty_pages->buffer_lock); + if (eb) { + ret = test_and_clear_bit(EXTENT_BUFFER_DIRTY, + &eb->bflags); + atomic_set(&eb->refs, 1); + } + if (PageWriteback(page)) + end_page_writeback(page); + + lock_page(page); + if (PageDirty(page)) { + clear_page_dirty_for_io(page); + spin_lock_irq(&page->mapping->tree_lock); + radix_tree_tag_clear(&page->mapping->page_tree, + page_index(page), + PAGECACHE_TAG_DIRTY); + spin_unlock_irq(&page->mapping->tree_lock); + } + + page->mapping->a_ops->invalidatepage(page, 0); + unlock_page(page); + } + } + + return ret; +} + +static int btrfs_destroy_pinned_extent(struct btrfs_root *root, + struct extent_io_tree *pinned_extents) +{ + struct extent_io_tree *unpin; + u64 start; + u64 end; + int ret; + + unpin = pinned_extents; + while (1) { + ret = find_first_extent_bit(unpin, 0, &start, &end, + EXTENT_DIRTY); + if (ret) + break; + + /* opt_discard */ + ret = btrfs_error_discard_extent(root, start, end + 1 - start); + + clear_extent_dirty(unpin, start, end, GFP_NOFS); + btrfs_error_unpin_extent_range(root, start, end); + cond_resched(); + } + + return 0; +} + +static int btrfs_cleanup_transaction(struct btrfs_root *root) +{ + struct btrfs_transaction *t; + LIST_HEAD(list); + + WARN_ON(1); + + mutex_lock(&root->fs_info->trans_mutex); + mutex_lock(&root->fs_info->transaction_kthread_mutex); + + list_splice_init(&root->fs_info->trans_list, &list); + while (!list_empty(&list)) { + t = list_entry(list.next, struct btrfs_transaction, list); + if (!t) + break; + + btrfs_destroy_ordered_operations(root); + + btrfs_destroy_ordered_extents(root); + + btrfs_destroy_delayed_refs(t, root); + + btrfs_block_rsv_release(root, + &root->fs_info->trans_block_rsv, + t->dirty_pages.dirty_bytes); + + /* FIXME: cleanup wait for commit */ + t->in_commit = 1; + t->blocked = 1; + if (waitqueue_active(&root->fs_info->transaction_blocked_wait)) + wake_up(&root->fs_info->transaction_blocked_wait); + + t->blocked = 0; + if (waitqueue_active(&root->fs_info->transaction_wait)) + wake_up(&root->fs_info->transaction_wait); + mutex_unlock(&root->fs_info->trans_mutex); + + mutex_lock(&root->fs_info->trans_mutex); + t->commit_done = 1; + if (waitqueue_active(&t->commit_wait)) + wake_up(&t->commit_wait); + mutex_unlock(&root->fs_info->trans_mutex); + + mutex_lock(&root->fs_info->trans_mutex); + + btrfs_destroy_pending_snapshots(t); + + btrfs_destroy_delalloc_inodes(root); + + spin_lock(&root->fs_info->new_trans_lock); + root->fs_info->running_transaction = NULL; + spin_unlock(&root->fs_info->new_trans_lock); + + btrfs_destroy_marked_extents(root, &t->dirty_pages, + EXTENT_DIRTY); + + btrfs_destroy_pinned_extent(root, + root->fs_info->pinned_extents); + + t->use_count = 0; + list_del_init(&t->list); + memset(t, 0, sizeof(*t)); + kmem_cache_free(btrfs_transaction_cachep, t); + } + + mutex_unlock(&root->fs_info->transaction_kthread_mutex); + mutex_unlock(&root->fs_info->trans_mutex); + + return 0; +} + 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, -- cgit v1.2.2 From 83a4d54840c88a4a45c49670f044b8c7ddeaa8c7 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Mon, 27 Dec 2010 16:19:53 +0800 Subject: Btrfs: Fix memory leak at umount fs_info, which is allocated in open_ctree(), should be freed in close_ctree(). Signed-off-by: Li Zefan --- fs/btrfs/disk-io.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index a5d2249e6da5..089871e5cd5a 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2513,6 +2513,8 @@ int close_ctree(struct btrfs_root *root) kfree(fs_info->chunk_root); kfree(fs_info->dev_root); kfree(fs_info->csum_root); + kfree(fs_info); + return 0; } -- cgit v1.2.2 From 3612b49598c303cfb22a4b609427f829828e2427 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Tue, 25 Jan 2011 02:51:38 +0000 Subject: btrfs: fix return value check of btrfs_join_transaction() The error check of btrfs_join_transaction()/btrfs_join_transaction_nolock() is added, and the mistake of the error check in several places is corrected. For more stable Btrfs, I think that we should reduce BUG_ON(). But, I think that long time is necessary for this. So, I propose this patch as a short-term solution. With this patch: - To more stable Btrfs, the part that should be corrected is clarified. - The panic isn't done by the NULL pointer reference etc. (even if BUG_ON() is increased temporarily) - The error code is returned in the place where the error can be easily returned. As a long-term plan: - BUG_ON() is reduced by using the forced-readonly framework, etc. Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 2887b8be6fdd..b36eeef19194 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1550,6 +1550,7 @@ static int transaction_kthread(void *arg) spin_unlock(&root->fs_info->new_trans_lock); trans = btrfs_join_transaction(root, 1); + BUG_ON(IS_ERR(trans)); if (transid == trans->transid) { ret = btrfs_commit_transaction(trans, root); BUG_ON(ret); @@ -2464,10 +2465,14 @@ int btrfs_commit_super(struct btrfs_root *root) up_write(&root->fs_info->cleanup_work_sem); trans = btrfs_join_transaction(root, 1); + if (IS_ERR(trans)) + return PTR_ERR(trans); ret = btrfs_commit_transaction(trans, root); BUG_ON(ret); /* run commit again to drop the original snapshot */ trans = btrfs_join_transaction(root, 1); + if (IS_ERR(trans)) + return PTR_ERR(trans); btrfs_commit_transaction(trans, root); ret = btrfs_write_and_wait_transaction(NULL, root); BUG_ON(ret); -- cgit v1.2.2 From eb14ab8ed24a0405fd056068b28c33a1cd846024 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 10 Feb 2011 12:35:00 -0500 Subject: Btrfs: fix page->private races There is a race where btrfs_releasepage can drop the page->private contents just as alloc_extent_buffer is setting up pages for metadata. Because of how the Btrfs page flags work, this results in us skipping the crc on the page during IO. This patch sovles the race by waiting until after the extent buffer is inserted into the radix tree before it sets page private. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 8 ++++++-- 1 file changed, 6 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 b36eeef19194..3e1ea3e0477e 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -359,10 +359,14 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page) tree = &BTRFS_I(page->mapping->host)->io_tree; - if (page->private == EXTENT_PAGE_PRIVATE) + if (page->private == EXTENT_PAGE_PRIVATE) { + WARN_ON(1); goto out; - if (!page->private) + } + if (!page->private) { + WARN_ON(1); goto out; + } len = page->private >> 2; WARN_ON(len == 0); -- cgit v1.2.2 From ae0e47f02aaedbfdd5e4bec73f79b714d758223d Mon Sep 17 00:00:00 2001 From: "Justin P. Mattock" Date: Tue, 1 Mar 2011 15:06:02 +0100 Subject: Remove one to many n's in a word Signed-off-by: Justin P. Mattock Signed-off-by: Jiri Kosina --- 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 fdce8799b98d..7b658d2107b4 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2489,7 +2489,7 @@ int close_ctree(struct btrfs_root *root) * ERROR state on disk. * * 2. when btrfs flips readonly just in btrfs_commit_super, - * and in such case, btrfs cannnot write sb via btrfs_commit_super, + * and in such case, btrfs cannot write sb via btrfs_commit_super, * and since fs_state has been set BTRFS_SUPER_FLAG_ERROR flag, * btrfs will cleanup all FS resources first and write sb then. */ -- cgit v1.2.2 From 7eaceaccab5f40bbfda044629a6298616aeaed50 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Thu, 10 Mar 2011 08:52:07 +0100 Subject: block: remove per-queue plugging Code has been converted over to the new explicit on-stack plugging, and delay users have been converted to use the new API for that. So lets kill off the old plugging along with aops->sync_page(). Signed-off-by: Jens Axboe --- fs/btrfs/disk-io.c | 79 ------------------------------------------------------ 1 file changed, 79 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index e1aa8d607bc7..ada1f6bd0a57 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -847,7 +847,6 @@ static const struct address_space_operations btree_aops = { .writepages = btree_writepages, .releasepage = btree_releasepage, .invalidatepage = btree_invalidatepage, - .sync_page = block_sync_page, #ifdef CONFIG_MIGRATION .migratepage = btree_migratepage, #endif @@ -1330,82 +1329,6 @@ static int btrfs_congested_fn(void *congested_data, int bdi_bits) return ret; } -/* - * this unplugs every device on the box, and it is only used when page - * is null - */ -static void __unplug_io_fn(struct backing_dev_info *bdi, struct page *page) -{ - struct btrfs_device *device; - struct btrfs_fs_info *info; - - info = (struct btrfs_fs_info *)bdi->unplug_io_data; - list_for_each_entry(device, &info->fs_devices->devices, dev_list) { - if (!device->bdev) - continue; - - bdi = blk_get_backing_dev_info(device->bdev); - if (bdi->unplug_io_fn) - bdi->unplug_io_fn(bdi, page); - } -} - -static void btrfs_unplug_io_fn(struct backing_dev_info *bdi, struct page *page) -{ - struct inode *inode; - struct extent_map_tree *em_tree; - struct extent_map *em; - struct address_space *mapping; - u64 offset; - - /* the generic O_DIRECT read code does this */ - if (1 || !page) { - __unplug_io_fn(bdi, page); - return; - } - - /* - * page->mapping may change at any time. Get a consistent copy - * and use that for everything below - */ - smp_mb(); - mapping = page->mapping; - if (!mapping) - return; - - inode = mapping->host; - - /* - * don't do the expensive searching for a small number of - * devices - */ - if (BTRFS_I(inode)->root->fs_info->fs_devices->open_devices <= 2) { - __unplug_io_fn(bdi, page); - return; - } - - offset = page_offset(page); - - em_tree = &BTRFS_I(inode)->extent_tree; - read_lock(&em_tree->lock); - em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE); - read_unlock(&em_tree->lock); - if (!em) { - __unplug_io_fn(bdi, page); - return; - } - - if (em->block_start >= EXTENT_MAP_LAST_BYTE) { - free_extent_map(em); - __unplug_io_fn(bdi, page); - return; - } - offset = offset - em->start; - btrfs_unplug_page(&BTRFS_I(inode)->root->fs_info->mapping_tree, - em->block_start + offset, page); - free_extent_map(em); -} - /* * If this fails, caller must call bdi_destroy() to get rid of the * bdi again. @@ -1420,8 +1343,6 @@ static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi) return err; bdi->ra_pages = default_backing_dev_info.ra_pages; - bdi->unplug_io_fn = btrfs_unplug_io_fn; - bdi->unplug_io_data = info; bdi->congested_fn = btrfs_congested_fn; bdi->congested_data = info; return 0; -- cgit v1.2.2 From 66b4ffd110f9b48b8d8c1319ee446b53b8d073bf Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 31 Jan 2011 16:22:42 -0500 Subject: Btrfs: handle errors in btrfs_orphan_cleanup If we cannot truncate an inode for some reason we will never delete the orphan item associated with that inode, which means that we will loop forever in btrfs_orphan_cleanup. Instead of doing this just return error so we fail to mount. It sucks, but hey it's better than hanging. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/disk-io.c | 15 ++++++++++++--- 1 file changed, 12 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 e1aa8d607bc7..495b1ac45f8c 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2058,9 +2058,14 @@ struct btrfs_root *open_ctree(struct super_block *sb, if (!(sb->s_flags & MS_RDONLY)) { down_read(&fs_info->cleanup_work_sem); - btrfs_orphan_cleanup(fs_info->fs_root); - btrfs_orphan_cleanup(fs_info->tree_root); + err = btrfs_orphan_cleanup(fs_info->fs_root); + if (!err) + err = btrfs_orphan_cleanup(fs_info->tree_root); up_read(&fs_info->cleanup_work_sem); + if (err) { + close_ctree(tree_root); + return ERR_PTR(err); + } } return tree_root; @@ -2435,8 +2440,12 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info) root_objectid = gang[ret - 1]->root_key.objectid + 1; for (i = 0; i < ret; i++) { + int err; + root_objectid = gang[i]->root_key.objectid; - btrfs_orphan_cleanup(gang[i]); + err = btrfs_orphan_cleanup(gang[i]); + if (err) + return err; } root_objectid++; } -- cgit v1.2.2 From a826d6dcb32d811b4c81df57a5ef1367516586b0 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 16 Mar 2011 13:42:43 -0400 Subject: Btrfs: check items for correctness as we search Currently if we have corrupted items things will blow up in spectacular ways. So as we read in blocks and they are leaves, check the entire leaf to make sure all of the items are correct and point to valid parts in the leaf for the item data the are responsible for. If the item is corrupt we will kick back EIO and not read any of the copies since they are likely to not be correct either. This will catch generic corruptions, it will be up to the individual callers of btrfs_search_slot to make sure their items are right. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/disk-io.c | 90 +++++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 89 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 495b1ac45f8c..9f31e110b481 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -323,6 +323,7 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root, int num_copies = 0; int mirror_num = 0; + clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags); io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree; while (1) { ret = read_extent_buffer_pages(io_tree, eb, start, 1, @@ -331,6 +332,14 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root, !verify_parent_transid(io_tree, eb, parent_transid)) return ret; + /* + * This buffer's crc is fine, but its contents are corrupted, so + * there is no reason to read the other copies, they won't be + * any less wrong. + */ + if (test_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags)) + return ret; + num_copies = btrfs_num_copies(&root->fs_info->mapping_tree, eb->start, eb->len); if (num_copies == 1) @@ -419,6 +428,73 @@ static int check_tree_block_fsid(struct btrfs_root *root, return ret; } +#define CORRUPT(reason, eb, root, slot) \ + printk(KERN_CRIT "btrfs: corrupt leaf, %s: block=%llu," \ + "root=%llu, slot=%d\n", reason, \ + (unsigned long long)btrfs_header_bytenr(eb), \ + (unsigned long long)root->objectid, slot) + +static noinline int check_leaf(struct btrfs_root *root, + struct extent_buffer *leaf) +{ + struct btrfs_key key; + struct btrfs_key leaf_key; + u32 nritems = btrfs_header_nritems(leaf); + int slot; + + if (nritems == 0) + return 0; + + /* Check the 0 item */ + if (btrfs_item_offset_nr(leaf, 0) + btrfs_item_size_nr(leaf, 0) != + BTRFS_LEAF_DATA_SIZE(root)) { + CORRUPT("invalid item offset size pair", leaf, root, 0); + return -EIO; + } + + /* + * Check to make sure each items keys are in the correct order and their + * offsets make sense. We only have to loop through nritems-1 because + * we check the current slot against the next slot, which verifies the + * next slot's offset+size makes sense and that the current's slot + * offset is correct. + */ + for (slot = 0; slot < nritems - 1; slot++) { + btrfs_item_key_to_cpu(leaf, &leaf_key, slot); + btrfs_item_key_to_cpu(leaf, &key, slot + 1); + + /* Make sure the keys are in the right order */ + if (btrfs_comp_cpu_keys(&leaf_key, &key) >= 0) { + CORRUPT("bad key order", leaf, root, slot); + return -EIO; + } + + /* + * Make sure the offset and ends are right, remember that the + * item data starts at the end of the leaf and grows towards the + * front. + */ + if (btrfs_item_offset_nr(leaf, slot) != + btrfs_item_end_nr(leaf, slot + 1)) { + CORRUPT("slot offset bad", leaf, root, slot); + return -EIO; + } + + /* + * Check to make sure that we don't point outside of the leaf, + * just incase all the items are consistent to eachother, but + * all point outside of the leaf. + */ + if (btrfs_item_end_nr(leaf, slot) > + BTRFS_LEAF_DATA_SIZE(root)) { + CORRUPT("slot end outside of leaf", leaf, root, slot); + return -EIO; + } + } + + return 0; +} + #ifdef CONFIG_DEBUG_LOCK_ALLOC void btrfs_set_buffer_lockdep_class(struct extent_buffer *eb, int level) { @@ -485,8 +561,20 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, btrfs_set_buffer_lockdep_class(eb, found_level); ret = csum_tree_block(root, eb, 1); - if (ret) + if (ret) { ret = -EIO; + goto err; + } + + /* + * If this is a leaf block and it is corrupt, set the corrupt bit so + * that we don't try and read the other copies of this block, just + * return -EIO. + */ + if (found_level == 0 && check_leaf(root, eb)) { + set_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags); + ret = -EIO; + } end = min_t(u64, eb->len, PAGE_CACHE_SIZE); end = eb->start + end - 1; -- cgit v1.2.2 From db5b493ac78e46c7b6bad22cd25d8041564cd8ea Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Wed, 23 Mar 2011 08:14:16 +0000 Subject: Btrfs: cleanup some BUG_ON() This patch changes some BUG_ON() to the error return. (but, most callers still use BUG_ON()) Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 9f31e110b481..00cbb41af660 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1248,7 +1248,10 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root, root, fs_info, location->objectid); path = btrfs_alloc_path(); - BUG_ON(!path); + if (!path) { + kfree(root); + return ERR_PTR(-ENOMEM); + } ret = btrfs_search_slot(NULL, tree_root, location, path, 0, 0); if (ret == 0) { l = path->nodes[0]; -- cgit v1.2.2 From 7e75bf3ff3a716d7b21d8fb43bf823115801c1e9 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Fri, 18 Mar 2011 22:56:43 +0000 Subject: btrfs: properly access unaligned checksum buffer On Fri, Mar 18, 2011 at 11:56:53AM -0400, Chris Mason wrote: > Thanks for fielding this one. Does put_unaligned_le32 optimize away on > platforms with efficient access? It would be great if we didn't need > the #ifdef. (quicktest: assembly output is same for put_unaligned_le32 and direct assignment on my x86_64) I was originally following examples in Documentation/unaligned-memory-access.txt. From other code it seems to me that the define CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS is intended for larger portions of code. Macros/wrappers for {put,get}_unaligned* are chosen via arch//include/asm/unaligned.h accordingly, therefore it's safe to use put_unaligned_le32 without the ifdef. dave Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 00cbb41af660..2bdb124333ab 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -29,6 +29,7 @@ #include #include #include +#include #include "compat.h" #include "ctree.h" #include "disk-io.h" @@ -198,7 +199,7 @@ u32 btrfs_csum_data(struct btrfs_root *root, char *data, u32 seed, size_t len) void btrfs_csum_final(u32 crc, char *result) { - *(__le32 *)result = ~cpu_to_le32(crc); + put_unaligned_le32(~crc, result); } /* -- cgit v1.2.2 From 75e7cb7fe0c391561bd3af36515be3f3c64a04c6 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Tue, 22 Mar 2011 10:12:20 +0000 Subject: Btrfs: Per file/directory controls for COW and compression Data compression and data cow are controlled across the entire FS by mount options right now. ioctls are needed to set this on a per file or per directory basis. This has been proposed previously, but VFS developers wanted us to use generic ioctls rather than btrfs-specific ones. According to Chris's comment, there should be just one true compression method(probably LZO) stored in the super. However, before this, we would wait for that one method is stable enough to be adopted into the super. So I list it as a long term goal, and just store it in ram today. After applying this patch, we can use the generic "FS_IOC_SETFLAGS" ioctl to control file and directory's datacow and compression attribute. NOTE: - The compression type is selected by such rules: If we mount btrfs with compress options, ie, zlib/lzo, the type is it. Otherwise, we'll use the default compress type (zlib today). v1->v2: - rebase to the latest btrfs. v2->v3: - fix a problem, i.e. when a file is set NOCOW via mount option, then this NOCOW will be screwed by inheritance from parent directory. Signed-off-by: Liu Bo 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 2bdb124333ab..125639ddaffe 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1854,6 +1854,12 @@ struct btrfs_root *open_ctree(struct super_block *sb, btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY); + /* + * In the long term, we'll store the compression type in the super + * block, and it'll be used for per file compression control. + */ + fs_info->compress_type = BTRFS_COMPRESS_ZLIB; + ret = btrfs_parse_options(tree_root, options); if (ret) { err = ret; -- cgit v1.2.2 From 5378e60734f5b7bfe1b43dc191aaf6131c1befe7 Mon Sep 17 00:00:00 2001 From: Li Dongyang Date: Thu, 24 Mar 2011 10:24:27 +0000 Subject: Btrfs: adjust btrfs_discard_extent() return errors and trimmed bytes Callers of btrfs_discard_extent() should check if we are mounted with -o discard, as we want to make fitrim to work even the fs is not mounted with -o discard. Also we should use REQ_DISCARD to map the free extent to get a full mapping, last we only return errors if 1. the error is not a EOPNOTSUPP 2. no device supports discard Signed-off-by: Li Dongyang Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 125639ddaffe..8ecc5419d8b6 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -3054,7 +3054,10 @@ static int btrfs_destroy_pinned_extent(struct btrfs_root *root, break; /* opt_discard */ - ret = btrfs_error_discard_extent(root, start, end + 1 - start); + if (btrfs_test_opt(root, DISCARD)) + ret = btrfs_error_discard_extent(root, start, + end + 1 - start, + NULL); clear_extent_dirty(unpin, start, end, GFP_NOFS); btrfs_error_unpin_extent_range(root, start, end); -- cgit v1.2.2 From c59021f846881a957ac5afe456d0f59d6a517b61 Mon Sep 17 00:00:00 2001 From: liubo Date: Mon, 7 Mar 2011 02:13:14 +0000 Subject: Btrfs: fix OOPS of empty filesystem after balance btrfs will remove unused block groups after balance. When a empty filesystem is balanced, the block group with tag "DATA" may be dropped, and after umount and mount again, it will not find "DATA" space_info and lead to OOPS. So we initial the necessary space_infos(DATA, SYSTEM, METADATA) to avoid OOPS. Reported-by: Daniel J Blueman Signed-off-by: Liu Bo 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 8ecc5419d8b6..b3fc8475870f 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2065,6 +2065,12 @@ struct btrfs_root *open_ctree(struct super_block *sb, fs_info->metadata_alloc_profile = (u64)-1; fs_info->system_alloc_profile = fs_info->metadata_alloc_profile; + ret = btrfs_init_space_info(fs_info); + if (ret) { + printk(KERN_ERR "Failed to initial space info: %d\n", ret); + goto fail_block_groups; + } + ret = btrfs_read_block_groups(extent_root); if (ret) { printk(KERN_ERR "Failed to read block groups: %d\n", ret); -- cgit v1.2.2 From 1561deda687eef0e95065f1268d680ddc5976ee7 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Sun, 27 Mar 2011 16:07:36 +0800 Subject: btrfs: fix possible deadlock by clearing __GFP_FS flag Using the GFP_HIGHUSER_MOVABLE flag to allocate the metadata's page may cause deadlock. Task1 open() ... btrfs_search_slot() ... btrfs_cow_block() ... alloc_page() wait for reclaiming shrink_slab() ... shrink_icache_memory() ... btrfs_evict_inode() ... btrfs_search_slot() If the path is locked by task1, the deadlock happens. So the btree's page cache is different with the file's page cache, it can not allocate pages by GFP_HIGHUSER_MOVABLE flag, we must clear __GFP_FS flag in GFP_HIGHUSER_MOVABLE flag. Reported-by: Itaru Kitayama Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index b3fc8475870f..5cf3aa7b125c 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1724,6 +1724,8 @@ struct btrfs_root *open_ctree(struct super_block *sb, goto fail_bdi; } + fs_info->btree_inode->i_mapping->flags &= ~__GFP_FS; + INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC); INIT_LIST_HEAD(&fs_info->trans_list); INIT_LIST_HEAD(&fs_info->dead_roots); -- cgit v1.2.2 From 08fe4db170b4193603d9d31f40ebaf652d07ac9c Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Mon, 28 Mar 2011 02:01:25 +0000 Subject: Btrfs: Fix uninitialized root flags for subvolumes root_item->flags and root_item->byte_limit are not initialized when a subvolume is created. This bug is not revealed until we added readonly snapshot support - now you mount a btrfs filesystem and you may find the subvolumes in it are readonly. To work around this problem, we steal a bit from root_item->inode_item->flags, and use it to indicate if those fields have been properly initialized. When we read a tree root from disk, we check if the bit is set, and if not we'll set the flag and initialize the two fields of the root item. Reported-by: Andreas Philipp Signed-off-by: Li Zefan Tested-by: Andreas Philipp cc: stable@kernel.org Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 5cf3aa7b125c..a272bfd74ea0 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1276,8 +1276,10 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root, root->commit_root = btrfs_root_node(root); BUG_ON(!root->node); out: - if (location->objectid != BTRFS_TREE_LOG_OBJECTID) + if (location->objectid != BTRFS_TREE_LOG_OBJECTID) { root->ref_cows = 1; + btrfs_check_and_init_root_item(&root->root_item); + } return root; } -- cgit v1.2.2 From 13c5a93e7005d7dae0b6d070d25203593e692d13 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 11 Apr 2011 15:45:29 -0400 Subject: Btrfs: avoid taking the trans_mutex in btrfs_end_transaction I've been working on making our O_DIRECT latency not suck and I noticed we were taking the trans_mutex in btrfs_end_transaction. So to do this we convert num_writers and use_count to atomic_t's and just decrement them in btrfs_end_transaction. Instead of deleting the transaction from the trans list in put_transaction we do that in btrfs_commit_transaction() since that's the only time it actually needs to be removed from the list. Thanks, Signed-off-by: Josef Bacik --- 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 a272bfd74ea0..ef6865c17cd6 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -3136,7 +3136,7 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root) btrfs_destroy_pinned_extent(root, root->fs_info->pinned_extents); - t->use_count = 0; + atomic_set(&t->use_count, 0); list_del_init(&t->list); memset(t, 0, sizeof(*t)); kmem_cache_free(btrfs_transaction_cachep, t); -- cgit v1.2.2 From 581bb050941b4f220f84d3e5ed6dace3d42dd382 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Wed, 20 Apr 2011 10:06:11 +0800 Subject: Btrfs: Cache free inode numbers in memory Currently btrfs stores the highest objectid of the fs tree, and it always returns (highest+1) inode number when we create a file, so inode numbers won't be reclaimed when we delete files, so we'll run out of inode numbers as we keep create/delete files in 32bits machines. This fixes it, and it works similarly to how we cache free space in block cgroups. We start a kernel thread to read the file tree. By scanning inode items, we know which chunks of inode numbers are free, and we cache them in an rb-tree. Because we are searching the commit root, we have to carefully handle the cross-transaction case. The rb-tree is a hybrid extent+bitmap tree, so if we have too many small chunks of inode numbers, we'll use bitmaps. Initially we allow 16K ram of extents, and a bitmap will be used if we exceed this threshold. The extents threshold is adjusted in runtime. Signed-off-by: Li Zefan --- fs/btrfs/disk-io.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index ef6865c17cd6..d02683b1ee16 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -41,6 +41,7 @@ #include "locking.h" #include "tree-log.h" #include "free-space-cache.h" +#include "inode-map.h" static struct extent_io_ops btree_extent_io_ops; static void end_workqueue_fn(struct btrfs_work *work); @@ -1327,6 +1328,19 @@ again: if (IS_ERR(root)) return root; + root->free_ino_ctl = kzalloc(sizeof(*root->free_ino_ctl), GFP_NOFS); + if (!root->free_ino_ctl) + goto fail; + root->free_ino_pinned = kzalloc(sizeof(*root->free_ino_pinned), + GFP_NOFS); + if (!root->free_ino_pinned) + goto fail; + + btrfs_init_free_ino_ctl(root); + mutex_init(&root->fs_commit_mutex); + spin_lock_init(&root->cache_lock); + init_waitqueue_head(&root->cache_wait); + set_anon_super(&root->anon_super, NULL); if (btrfs_root_refs(&root->root_item) == 0) { @@ -2483,6 +2497,8 @@ int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root) if (btrfs_root_refs(&root->root_item) == 0) synchronize_srcu(&fs_info->subvol_srcu); + __btrfs_remove_free_space_cache(root->free_ino_pinned); + __btrfs_remove_free_space_cache(root->free_ino_ctl); free_fs_root(root); return 0; } @@ -2496,6 +2512,8 @@ static void free_fs_root(struct btrfs_root *root) } free_extent_buffer(root->node); free_extent_buffer(root->commit_root); + kfree(root->free_ino_ctl); + kfree(root->free_ino_pinned); kfree(root->name); kfree(root); } -- cgit v1.2.2 From 82d5902d9c681be37ffa9d70482907f9f0b7ec1f Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Wed, 20 Apr 2011 10:33:24 +0800 Subject: Btrfs: Support reading/writing on disk free ino cache This is similar to block group caching. We dedicate a special inode in fs tree to save free ino cache. At the very first time we create/delete a file after mount, the free ino cache will be loaded from disk into memory. When the fs tree is commited, the cache will be written back to disk. To keep compatibility, we check the root generation against the generation of the special inode when loading the cache, so the loading will fail if the btrfs filesystem was mounted in an older kernel before. Signed-off-by: Li Zefan --- 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 d02683b1ee16..4f12c30a5470 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2505,6 +2505,7 @@ int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root) static void free_fs_root(struct btrfs_root *root) { + iput(root->cache_inode); WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree)); if (root->anon_super.s_dev) { down_write(&root->anon_super.s_umount); -- cgit v1.2.2 From cfece4db110dacfd6b4b87b912c59e77e6846fc0 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Mon, 25 Apr 2011 19:43:52 -0400 Subject: btrfs: add missing spin_unlock to a rare exit path Signed-off-by: David Sterba 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 ef6865c17cd6..fe5aec9b3924 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2903,6 +2903,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, spin_lock(&delayed_refs->lock); if (delayed_refs->num_entries == 0) { + spin_unlock(&delayed_refs->lock); printk(KERN_INFO "delayed_refs has NO entry\n"); return ret; } -- cgit v1.2.2 From 306e16ce13c0f3d4fc071b45803b5b83c2606011 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Tue, 19 Apr 2011 14:29:38 +0200 Subject: btrfs: rename variables clashing with global function names reported by gcc -Wshadow: page_index, page_offset, new_inode, dev_name Signed-off-by: David Sterba --- 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 228cf36ece83..990afa8656a2 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -137,7 +137,7 @@ static const char *btrfs_eb_name[BTRFS_MAX_LEVEL + 1] = { * that covers the entire device */ static struct extent_map *btree_get_extent(struct inode *inode, - struct page *page, size_t page_offset, u64 start, u64 len, + struct page *page, size_t pg_offset, u64 start, u64 len, int create) { struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; -- cgit v1.2.2 From 4891aca2dac612a2f21a3278d9906ade13b55788 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Tue, 19 Apr 2011 16:45:00 +0200 Subject: btrfs: fix dereference before check The superblock's ->s_fs_info is properly set in btrfs_fill_super, after a call to open_ctree, which derefereces it before check. Although tree_root is set via btrfs_set_super, let's be defensive and leave the check in place. Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 5 +++-- 1 file changed, 3 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 990afa8656a2..25e4b8f1d0ef 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1611,7 +1611,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, struct btrfs_root *csum_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS); struct btrfs_root *tree_root = btrfs_sb(sb); - struct btrfs_fs_info *fs_info = tree_root->fs_info; + struct btrfs_fs_info *fs_info = NULL; struct btrfs_root *chunk_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS); struct btrfs_root *dev_root = kzalloc(sizeof(struct btrfs_root), @@ -1623,11 +1623,12 @@ struct btrfs_root *open_ctree(struct super_block *sb, struct btrfs_super_block *disk_super; - if (!extent_root || !tree_root || !fs_info || + if (!extent_root || !tree_root || !tree_root->fs_info || !chunk_root || !dev_root || !csum_root) { err = -ENOMEM; goto fail; } + fs_info = tree_root->fs_info; ret = init_srcu_struct(&fs_info->subvol_srcu); if (ret) { -- cgit v1.2.2 From f993c883ad8e111fb9e9ae603540acbe94f7246c Mon Sep 17 00:00:00 2001 From: David Sterba Date: Wed, 20 Apr 2011 23:35:57 +0200 Subject: btrfs: drop unused argument from extent_io_tree_init all callers pass GFP_NOFS, but the GFP mask argument is not used in the function; GFP_ATOMIC is passed to radix tree initialization and it's the only correct one, since we're using the preload/insert mechanism of radix tree. Let's drop the gfp mask from btrfs function, this will not change behaviour. Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 9 ++++----- 1 file changed, 4 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 25e4b8f1d0ef..3ce80f71e98a 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1080,7 +1080,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->log_transid = 0; root->last_log_commit = 0; extent_io_tree_init(&root->dirty_log_pages, - fs_info->btree_inode->i_mapping, GFP_NOFS); + fs_info->btree_inode->i_mapping); memset(&root->root_key, 0, sizeof(root->root_key)); memset(&root->root_item, 0, sizeof(root->root_item)); @@ -1712,8 +1712,7 @@ struct btrfs_root *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, - GFP_NOFS); + fs_info->btree_inode->i_mapping); extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree, GFP_NOFS); @@ -1729,9 +1728,9 @@ struct btrfs_root *open_ctree(struct super_block *sb, fs_info->block_group_cache_tree = RB_ROOT; extent_io_tree_init(&fs_info->freed_extents[0], - fs_info->btree_inode->i_mapping, GFP_NOFS); + fs_info->btree_inode->i_mapping); extent_io_tree_init(&fs_info->freed_extents[1], - fs_info->btree_inode->i_mapping, GFP_NOFS); + fs_info->btree_inode->i_mapping); fs_info->pinned_extents = &fs_info->freed_extents[0]; fs_info->do_barriers = 1; -- cgit v1.2.2 From a8067e022ab54fde8953880a64572c3acca644dc Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 21 Apr 2011 00:34:43 +0200 Subject: btrfs: drop unused parameter from extent_map_tree_init the GFP flags are not stored anywhere and all allocations are done via alloc_extent_map(GFP_NOFS). Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 3ce80f71e98a..f2ee584b8efd 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1713,8 +1713,7 @@ struct btrfs_root *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); - extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree, - GFP_NOFS); + 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; -- cgit v1.2.2 From 172ddd60a662c4d8bf2809462866ddddd6431ea5 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 21 Apr 2011 00:48:27 +0200 Subject: btrfs: drop gfp parameter from alloc_extent_map pass GFP_NOFS directly to kmem_cache_alloc Signed-off-by: David Sterba --- 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 f2ee584b8efd..e1e55679d061 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -154,7 +154,7 @@ static struct extent_map *btree_get_extent(struct inode *inode, } read_unlock(&em_tree->lock); - em = alloc_extent_map(GFP_NOFS); + em = alloc_extent_map(); if (!em) { em = ERR_PTR(-ENOMEM); goto out; -- cgit v1.2.2 From f09d1f60e6aa82fb4cfaa525e21f6287fc1516f4 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 21 Apr 2011 01:08:01 +0200 Subject: btrfs: drop gfp parameter from find_extent_buffer pass GFP_NOFS directly to kmem_cache_alloc Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 4 ++-- 1 file changed, 2 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 e1e55679d061..1c0752e99066 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -963,7 +963,7 @@ struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root, struct inode *btree_inode = root->fs_info->btree_inode; struct extent_buffer *eb; eb = find_extent_buffer(&BTRFS_I(btree_inode)->io_tree, - bytenr, blocksize, GFP_NOFS); + bytenr, blocksize); return eb; } @@ -2696,7 +2696,7 @@ int btree_lock_page_hook(struct page *page) goto out; len = page->private >> 2; - eb = find_extent_buffer(io_tree, bytenr, len, GFP_NOFS); + eb = find_extent_buffer(io_tree, bytenr, len); if (!eb) goto out; -- cgit v1.2.2 From ba14419264684b290f0d0b7f48d26eafb11fc0c6 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 21 Apr 2011 01:12:06 +0200 Subject: btrfs: drop gfp parameter from alloc_extent_buffer pass GFP_NOFS directly to kmem_cache_alloc Signed-off-by: David Sterba --- 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 1c0752e99066..4084959b36fd 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -380,7 +380,7 @@ 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, GFP_NOFS); + eb = alloc_extent_buffer(tree, start, len, page); if (eb == NULL) { WARN_ON(1); goto out; @@ -525,7 +525,7 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, len = page->private >> 2; WARN_ON(len == 0); - eb = alloc_extent_buffer(tree, start, len, page, GFP_NOFS); + eb = alloc_extent_buffer(tree, start, len, page); if (eb == NULL) { ret = -EIO; goto out; @@ -974,7 +974,7 @@ 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, GFP_NOFS); + bytenr, blocksize, NULL); return eb; } -- cgit v1.2.2 From f2a97a9dbd86eb1ef956bdf20e05c507b32beb96 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 5 May 2011 12:44:41 +0200 Subject: btrfs: remove all unused functions Remove static and global declarations and/or definitions. Reduces size of btrfs.ko by ~3.4kB. text data bss dec hex filename 402081 7464 200 409745 64091 btrfs.ko.base 398620 7144 200 405964 631cc btrfs.ko.remove-all Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 27 --------------------------- 1 file changed, 27 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 4084959b36fd..fa287c551ffc 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -650,12 +650,6 @@ unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info) return 256 * limit; } -int btrfs_congested_async(struct btrfs_fs_info *info, int iodone) -{ - return atomic_read(&info->nr_async_bios) > - btrfs_async_submit_limit(info); -} - static void run_one_async_start(struct btrfs_work *work) { struct async_submit_bio *async; @@ -1283,21 +1277,6 @@ out: return root; } -struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info, - u64 root_objectid) -{ - struct btrfs_root *root; - - if (root_objectid == BTRFS_ROOT_TREE_OBJECTID) - return fs_info->tree_root; - if (root_objectid == BTRFS_EXTENT_TREE_OBJECTID) - return fs_info->extent_root; - - root = radix_tree_lookup(&fs_info->fs_roots_radix, - (unsigned long)root_objectid); - return root; -} - struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info, struct btrfs_key *location) { @@ -1369,11 +1348,6 @@ fail: return ERR_PTR(ret); } -struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info, - struct btrfs_key *location, - const char *name, int namelen) -{ - return btrfs_read_fs_root_no_name(fs_info, location); #if 0 struct btrfs_root *root; int ret; @@ -1402,7 +1376,6 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info, root->in_sysfs = 1; return root; #endif -} static int btrfs_congested_fn(void *congested_data, int bdi_bits) { -- cgit v1.2.2 From 182608c8294b5fe90d7bbd4b026c82bf0a24b736 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 5 May 2011 13:13:16 +0200 Subject: btrfs: remove old unused commented out code Remove code which has been #if0-ed out for a very long time and does not seem to be related to current codebase anymore. Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 29 ----------------------------- 1 file changed, 29 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index fa287c551ffc..de7b4770ab17 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1348,35 +1348,6 @@ fail: return ERR_PTR(ret); } -#if 0 - struct btrfs_root *root; - int ret; - - root = btrfs_read_fs_root_no_name(fs_info, location); - if (!root) - return NULL; - - if (root->in_sysfs) - return root; - - ret = btrfs_set_root_name(root, name, namelen); - if (ret) { - free_extent_buffer(root->node); - kfree(root); - return ERR_PTR(ret); - } - - ret = btrfs_sysfs_add_root(root); - if (ret) { - free_extent_buffer(root->node); - kfree(root->name); - kfree(root); - return ERR_PTR(ret); - } - root->in_sysfs = 1; - return root; -#endif - static int btrfs_congested_fn(void *congested_data, int bdi_bits) { struct btrfs_fs_info *info = (struct btrfs_fs_info *)congested_data; -- cgit v1.2.2 From a2de733c78fa7af51ba9670482fa7d392aa67c57 Mon Sep 17 00:00:00 2001 From: Arne Jansen Date: Tue, 8 Mar 2011 14:14:00 +0100 Subject: btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen --- fs/btrfs/disk-io.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index fe5aec9b3924..e48e8095c61f 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1773,6 +1773,17 @@ struct btrfs_root *open_ctree(struct super_block *sb, INIT_LIST_HEAD(&fs_info->ordered_extents); spin_lock_init(&fs_info->ordered_extent_lock); + mutex_init(&fs_info->scrub_lock); + atomic_set(&fs_info->scrubs_running, 0); + atomic_set(&fs_info->scrub_pause_req, 0); + atomic_set(&fs_info->scrubs_paused, 0); + atomic_set(&fs_info->scrub_cancel_req, 0); + init_waitqueue_head(&fs_info->scrub_pause_wait); + init_rwsem(&fs_info->scrub_super_lock); + fs_info->scrub_workers_refcnt = 0; + btrfs_init_workers(&fs_info->scrub_workers, "scrub", + fs_info->thread_pool_size, &fs_info->generic_worker); + sb->s_blocksize = 4096; sb->s_blocksize_bits = blksize_bits(4096); sb->s_bdi = &fs_info->bdi; @@ -2599,6 +2610,7 @@ int close_ctree(struct btrfs_root *root) fs_info->closing = 1; smp_mb(); + btrfs_scrub_cancel(root); btrfs_put_block_group_cache(fs_info); /* -- cgit v1.2.2 From 7a36ddec1003a4e84e79f28ee714a142ed6bc529 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Fri, 6 May 2011 15:33:15 +0200 Subject: btrfs: use printk_ratelimited instead of printk_ratelimit As per printk_ratelimit comment, it should not be used. Signed-off-by: David Sterba --- fs/btrfs/disk-io.c | 21 ++++++--------------- 1 file changed, 6 insertions(+), 15 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index de7b4770ab17..cb9d1b8bfe74 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -29,6 +29,7 @@ #include #include #include +#include #include #include "compat.h" #include "ctree.h" @@ -254,14 +255,12 @@ static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf, memcpy(&found, result, csum_size); read_extent_buffer(buf, &val, 0, csum_size); - if (printk_ratelimit()) { - printk(KERN_INFO "btrfs: %s checksum verify " + printk_ratelimited(KERN_INFO "btrfs: %s checksum verify " "failed on %llu wanted %X found %X " "level %d\n", root->fs_info->sb->s_id, (unsigned long long)buf->start, val, found, btrfs_header_level(buf)); - } if (result != (char *)&inline_result) kfree(result); return 1; @@ -296,13 +295,11 @@ static int verify_parent_transid(struct extent_io_tree *io_tree, ret = 0; goto out; } - if (printk_ratelimit()) { - printk("parent transid verify failed on %llu wanted %llu " + printk_ratelimited("parent transid verify failed on %llu wanted %llu " "found %llu\n", (unsigned long long)eb->start, (unsigned long long)parent_transid, (unsigned long long)btrfs_header_generation(eb)); - } ret = 1; clear_extent_buffer_uptodate(io_tree, eb, &cached_state); out: @@ -533,12 +530,10 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, found_start = btrfs_header_bytenr(eb); if (found_start != start) { - if (printk_ratelimit()) { - printk(KERN_INFO "btrfs bad tree block start " + printk_ratelimited(KERN_INFO "btrfs bad tree block start " "%llu %llu\n", (unsigned long long)found_start, (unsigned long long)eb->start); - } ret = -EIO; goto err; } @@ -550,10 +545,8 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, goto err; } if (check_tree_block_fsid(root, eb)) { - if (printk_ratelimit()) { - printk(KERN_INFO "btrfs bad fsid on block %llu\n", + printk_ratelimited(KERN_INFO "btrfs bad fsid on block %llu\n", (unsigned long long)eb->start); - } ret = -EIO; goto err; } @@ -2108,11 +2101,9 @@ static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate) if (uptodate) { set_buffer_uptodate(bh); } else { - if (printk_ratelimit()) { - printk(KERN_WARNING "lost page write due to " + printk_ratelimited(KERN_WARNING "lost page write due to " "I/O error on %s\n", bdevname(bh->b_bdev, b)); - } /* note, we dont' set_buffer_write_io_error because we have * our own ways of dealing with the IO errors */ -- cgit v1.2.2 From 16cdcec736cd214350cdb591bf1091f8beedefa0 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Fri, 22 Apr 2011 18:12:22 +0800 Subject: btrfs: implement delayed inode items operation Changelog V5 -> V6: - Fix oom when the memory load is high, by storing the delayed nodes into the root's radix tree, and letting btrfs inodes go. Changelog V4 -> V5: - Fix the race on adding the delayed node to the inode, which is spotted by Chris Mason. - Merge Chris Mason's incremental patch into this patch. - Fix deadlock between readdir() and memory fault, which is reported by Itaru Kitayama. Changelog V3 -> V4: - Fix nested lock, which is reported by Itaru Kitayama, by updating space cache inode in time. Changelog V2 -> V3: - Fix the race between the delayed worker and the task which does delayed items balance, which is reported by Tsutomu Itoh. - Modify the patch address David Sterba's comment. - Fix the bug of the cpu recursion spinlock, reported by Chris Mason Changelog V1 -> V2: - break up the global rb-tree, use a list to manage the delayed nodes, which is created for every directory and file, and used to manage the delayed directory name index items and the delayed inode item. - introduce a worker to deal with the delayed nodes. Compare with Ext3/4, the performance of file creation and deletion on btrfs is very poor. the reason is that btrfs must do a lot of b+ tree insertions, such as inode item, directory name item, directory name index and so on. If we can do some delayed b+ tree insertion or deletion, we can improve the performance, so we made this patch which implemented delayed directory name index insertion/deletion and delayed inode update. Implementation: - introduce a delayed root object into the filesystem, that use two lists to manage the delayed nodes which are created for every file/directory. One is used to manage all the delayed nodes that have delayed items. And the other is used to manage the delayed nodes which is waiting to be dealt with by the work thread. - Every delayed node has two rb-tree, one is used to manage the directory name index which is going to be inserted into b+ tree, and the other is used to manage the directory name index which is going to be deleted from b+ tree. - introduce a worker to deal with the delayed operation. This worker is used to deal with the works of the delayed directory name index items insertion and deletion and the delayed inode update. When the delayed items is beyond the lower limit, we create works for some delayed nodes and insert them into the work queue of the worker, and then go back. When the delayed items is beyond the upper bound, we create works for all the delayed nodes that haven't been dealt with, and insert them into the work queue of the worker, and then wait for that the untreated items is below some threshold value. - When we want to insert a directory name index into b+ tree, we just add the information into the delayed inserting rb-tree. And then we check the number of the delayed items and do delayed items balance. (The balance policy is above.) - When we want to delete a directory name index from the b+ tree, we search it in the inserting rb-tree at first. If we look it up, just drop it. If not, add the key of it into the delayed deleting rb-tree. Similar to the delayed inserting rb-tree, we also check the number of the delayed items and do delayed items balance. (The same to inserting manipulation) - When we want to update the metadata of some inode, we cached the data of the inode into the delayed node. the worker will flush it into the b+ tree after dealing with the delayed insertion and deletion. - We will move the delayed node to the tail of the list after we access the delayed node, By this way, we can cache more delayed items and merge more inode updates. - If we want to commit transaction, we will deal with all the delayed node. - the delayed node will be freed when we free the btrfs inode. - Before we log the inode items, we commit all the directory name index items and the delayed inode update. I did a quick test by the benchmark tool[1] and found we can improve the performance of file creation by ~15%, and file deletion by ~20%. Before applying this patch: Create files: Total files: 50000 Total time: 1.096108 Average time: 0.000022 Delete files: Total files: 50000 Total time: 1.510403 Average time: 0.000030 After applying this patch: Create files: Total files: 50000 Total time: 0.932899 Average time: 0.000019 Delete files: Total files: 50000 Total time: 1.215732 Average time: 0.000024 [1] http://marc.info/?l=linux-btrfs&m=128212635122920&q=p3 Many thanks for Kitayama-san's help! Signed-off-by: Miao Xie Reviewed-by: David Sterba Tested-by: Tsutomu Itoh Tested-by: Itaru Kitayama Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 50 +++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 45 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 228cf36ece83..22c3c9586049 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1058,6 +1058,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->name = NULL; root->in_sysfs = 0; root->inode_tree = RB_ROOT; + INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC); root->block_rsv = NULL; root->orphan_block_rsv = NULL; @@ -1693,6 +1694,13 @@ struct btrfs_root *open_ctree(struct super_block *sb, INIT_LIST_HEAD(&fs_info->ordered_extents); spin_lock_init(&fs_info->ordered_extent_lock); + fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root), + GFP_NOFS); + if (!fs_info->delayed_root) { + err = -ENOMEM; + goto fail_iput; + } + btrfs_init_delayed_root(fs_info->delayed_root); sb->s_blocksize = 4096; sb->s_blocksize_bits = blksize_bits(4096); @@ -1760,7 +1768,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, bh = btrfs_read_dev_super(fs_devices->latest_bdev); if (!bh) { err = -EINVAL; - goto fail_iput; + goto fail_alloc; } memcpy(&fs_info->super_copy, bh->b_data, sizeof(fs_info->super_copy)); @@ -1772,7 +1780,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, disk_super = &fs_info->super_copy; if (!btrfs_super_root(disk_super)) - goto fail_iput; + goto fail_alloc; /* check FS state, whether FS is broken. */ fs_info->fs_state |= btrfs_super_flags(disk_super); @@ -1788,7 +1796,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, ret = btrfs_parse_options(tree_root, options); if (ret) { err = ret; - goto fail_iput; + goto fail_alloc; } features = btrfs_super_incompat_flags(disk_super) & @@ -1798,7 +1806,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, "unsupported optional features (%Lx).\n", (unsigned long long)features); err = -EINVAL; - goto fail_iput; + goto fail_alloc; } features = btrfs_super_incompat_flags(disk_super); @@ -1814,7 +1822,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, "unsupported option features (%Lx).\n", (unsigned long long)features); err = -EINVAL; - goto fail_iput; + goto fail_alloc; } btrfs_init_workers(&fs_info->generic_worker, @@ -1861,6 +1869,9 @@ struct btrfs_root *open_ctree(struct super_block *sb, &fs_info->generic_worker); btrfs_init_workers(&fs_info->endio_freespace_worker, "freespace-write", 1, &fs_info->generic_worker); + btrfs_init_workers(&fs_info->delayed_workers, "delayed-meta", + fs_info->thread_pool_size, + &fs_info->generic_worker); /* * endios are largely parallel and should have a very @@ -1882,6 +1893,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, btrfs_start_workers(&fs_info->endio_meta_write_workers, 1); btrfs_start_workers(&fs_info->endio_write_workers, 1); btrfs_start_workers(&fs_info->endio_freespace_worker, 1); + btrfs_start_workers(&fs_info->delayed_workers, 1); fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super); fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages, @@ -2138,6 +2150,9 @@ fail_sb_buffer: btrfs_stop_workers(&fs_info->endio_write_workers); btrfs_stop_workers(&fs_info->endio_freespace_worker); btrfs_stop_workers(&fs_info->submit_workers); + btrfs_stop_workers(&fs_info->delayed_workers); +fail_alloc: + kfree(fs_info->delayed_root); fail_iput: invalidate_inode_pages2(fs_info->btree_inode->i_mapping); iput(fs_info->btree_inode); @@ -2578,6 +2593,7 @@ int close_ctree(struct btrfs_root *root) del_fs_roots(fs_info); iput(fs_info->btree_inode); + kfree(fs_info->delayed_root); btrfs_stop_workers(&fs_info->generic_worker); btrfs_stop_workers(&fs_info->fixup_workers); @@ -2589,6 +2605,7 @@ int close_ctree(struct btrfs_root *root) btrfs_stop_workers(&fs_info->endio_write_workers); btrfs_stop_workers(&fs_info->endio_freespace_worker); btrfs_stop_workers(&fs_info->submit_workers); + btrfs_stop_workers(&fs_info->delayed_workers); btrfs_close_devices(fs_info->fs_devices); btrfs_mapping_tree_free(&fs_info->mapping_tree); @@ -2662,6 +2679,29 @@ void btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr) u64 num_dirty; unsigned long thresh = 32 * 1024 * 1024; + if (current->flags & PF_MEMALLOC) + return; + + btrfs_balance_delayed_items(root); + + num_dirty = root->fs_info->dirty_metadata_bytes; + + if (num_dirty > thresh) { + balance_dirty_pages_ratelimited_nr( + root->fs_info->btree_inode->i_mapping, 1); + } + return; +} + +void __btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr) +{ + /* + * looks as though older kernels can get into trouble with + * this code, they end up stuck in balance_dirty_pages forever + */ + u64 num_dirty; + unsigned long thresh = 32 * 1024 * 1024; + if (current->flags & PF_MEMALLOC) return; -- cgit v1.2.2 From 7a7eaa40a39bde4eefc91aadeb1ce3dc4e6a1252 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 13 Apr 2011 12:54:33 -0400 Subject: Btrfs: take away the num_items argument from btrfs_join_transaction I keep forgetting that btrfs_join_transaction() just ignores the num_items argument, which leads me to sending pointless patches and looking stupid :). So just kill the num_items argument from btrfs_join_transaction and btrfs_start_ioctl_transaction, since neither of them use it. Thanks, Signed-off-by: Josef Bacik --- 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 228cf36ece83..9d6c9e332ca3 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1568,7 +1568,7 @@ static int transaction_kthread(void *arg) transid = cur->transid; spin_unlock(&root->fs_info->new_trans_lock); - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); BUG_ON(IS_ERR(trans)); if (transid == trans->transid) { ret = btrfs_commit_transaction(trans, root); @@ -2495,13 +2495,13 @@ int btrfs_commit_super(struct btrfs_root *root) down_write(&root->fs_info->cleanup_work_sem); up_write(&root->fs_info->cleanup_work_sem); - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); if (IS_ERR(trans)) return PTR_ERR(trans); ret = btrfs_commit_transaction(trans, root); BUG_ON(ret); /* run commit again to drop the original snapshot */ - trans = btrfs_join_transaction(root, 1); + trans = btrfs_join_transaction(root); if (IS_ERR(trans)) return PTR_ERR(trans); btrfs_commit_transaction(trans, root); -- cgit v1.2.2 From a4abeea41adfa3c143c289045f4625dfaeba2212 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 11 Apr 2011 17:25:13 -0400 Subject: Btrfs: kill trans_mutex We use trans_mutex for lots of things, here's a basic list 1) To serialize trans_handles joining the currently running transaction 2) To make sure that no new trans handles are started while we are committing 3) To protect the dead_roots list and the transaction lists Really the serializing trans_handles joining is not too hard, and can really get bogged down in acquiring a reference to the transaction. So replace the trans_mutex with a trans_lock spinlock and use it to do the following 1) Protect fs_info->running_transaction. All trans handles have to do is check this, and then take a reference of the transaction and keep on going. 2) Protect the fs_info->trans_list. This doesn't get used too much, basically it just holds the current transactions, which will usually just be the currently committing transaction and the currently running transaction at most. 3) Protect the dead roots list. This is only ever processed by splicing the list so this is relatively simple. 4) Protect the fs_info->reloc_ctl stuff. This is very lightweight and was using the trans_mutex before, so this is a pretty straightforward change. 5) Protect fs_info->no_trans_join. Because we don't hold the trans_lock over the entirety of the commit we need to have a way to block new people from creating a new transaction while we're doing our work. So we set no_trans_join and in join_transaction we test to see if that is set, and if it is we do a wait_on_commit. 6) Make the transaction use count atomic so we don't need to take locks to modify it when we're dropping references. 7) Add a commit_lock to the transaction to make sure multiple people trying to commit the same transaction don't race and commit at the same time. 8) Make open_ioctl_trans an atomic so we don't have to take any locks for ioctl trans. I have tested this with xfstests, but obviously it is a pretty hairy change so lots of testing is greatly appreciated. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/disk-io.c | 30 +++++++++++++++--------------- 1 file changed, 15 insertions(+), 15 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 9d6c9e332ca3..93ef254ec432 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1551,22 +1551,22 @@ static int transaction_kthread(void *arg) vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE); mutex_lock(&root->fs_info->transaction_kthread_mutex); - spin_lock(&root->fs_info->new_trans_lock); + spin_lock(&root->fs_info->trans_lock); cur = root->fs_info->running_transaction; if (!cur) { - spin_unlock(&root->fs_info->new_trans_lock); + spin_unlock(&root->fs_info->trans_lock); goto sleep; } now = get_seconds(); if (!cur->blocked && (now < cur->start_time || now - cur->start_time < 30)) { - spin_unlock(&root->fs_info->new_trans_lock); + spin_unlock(&root->fs_info->trans_lock); delay = HZ * 5; goto sleep; } transid = cur->transid; - spin_unlock(&root->fs_info->new_trans_lock); + spin_unlock(&root->fs_info->trans_lock); trans = btrfs_join_transaction(root); BUG_ON(IS_ERR(trans)); @@ -1658,7 +1658,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, INIT_LIST_HEAD(&fs_info->ordered_operations); INIT_LIST_HEAD(&fs_info->caching_block_groups); spin_lock_init(&fs_info->delalloc_lock); - spin_lock_init(&fs_info->new_trans_lock); + spin_lock_init(&fs_info->trans_lock); spin_lock_init(&fs_info->ref_cache_lock); spin_lock_init(&fs_info->fs_roots_radix_lock); spin_lock_init(&fs_info->delayed_iput_lock); @@ -1687,6 +1687,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, fs_info->sb = sb; fs_info->max_inline = 8192 * 1024; fs_info->metadata_ratio = 0; + fs_info->trans_no_join = 0; fs_info->thread_pool_size = min_t(unsigned long, num_online_cpus() + 2, 8); @@ -1735,7 +1736,6 @@ struct btrfs_root *open_ctree(struct super_block *sb, fs_info->do_barriers = 1; - mutex_init(&fs_info->trans_mutex); mutex_init(&fs_info->ordered_operations_mutex); mutex_init(&fs_info->tree_log_mutex); mutex_init(&fs_info->chunk_mutex); @@ -3006,10 +3006,13 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root) WARN_ON(1); - mutex_lock(&root->fs_info->trans_mutex); mutex_lock(&root->fs_info->transaction_kthread_mutex); + spin_lock(&root->fs_info->trans_lock); list_splice_init(&root->fs_info->trans_list, &list); + root->fs_info->trans_no_join = 1; + spin_unlock(&root->fs_info->trans_lock); + while (!list_empty(&list)) { t = list_entry(list.next, struct btrfs_transaction, list); if (!t) @@ -3034,23 +3037,18 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root) t->blocked = 0; if (waitqueue_active(&root->fs_info->transaction_wait)) wake_up(&root->fs_info->transaction_wait); - mutex_unlock(&root->fs_info->trans_mutex); - mutex_lock(&root->fs_info->trans_mutex); t->commit_done = 1; if (waitqueue_active(&t->commit_wait)) wake_up(&t->commit_wait); - mutex_unlock(&root->fs_info->trans_mutex); - - mutex_lock(&root->fs_info->trans_mutex); btrfs_destroy_pending_snapshots(t); btrfs_destroy_delalloc_inodes(root); - spin_lock(&root->fs_info->new_trans_lock); + spin_lock(&root->fs_info->trans_lock); root->fs_info->running_transaction = NULL; - spin_unlock(&root->fs_info->new_trans_lock); + spin_unlock(&root->fs_info->trans_lock); btrfs_destroy_marked_extents(root, &t->dirty_pages, EXTENT_DIRTY); @@ -3064,8 +3062,10 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root) kmem_cache_free(btrfs_transaction_cachep, t); } + spin_lock(&root->fs_info->trans_lock); + root->fs_info->trans_no_join = 0; + spin_unlock(&root->fs_info->trans_lock); mutex_unlock(&root->fs_info->transaction_kthread_mutex); - mutex_unlock(&root->fs_info->trans_mutex); return 0; } -- cgit v1.2.2 From 0956c798ef8dbe0fc215870eb68bd2d8e789f86a Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Wed, 18 May 2011 00:11:22 +0000 Subject: BTRFS: Remove unused node_lock 240f62c8756 replaced the node_lock with rcu_read_lock, but forgot to remove the actual lock in the data structure. Remove it here. Signed-off-by: Andi Kleen Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 1 - 1 file changed, 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 228cf36ece83..64b289690f9d 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1064,7 +1064,6 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, INIT_LIST_HEAD(&root->dirty_list); INIT_LIST_HEAD(&root->orphan_list); INIT_LIST_HEAD(&root->root_list); - spin_lock_init(&root->node_lock); spin_lock_init(&root->orphan_lock); spin_lock_init(&root->inode_lock); spin_lock_init(&root->accounting_lock); -- cgit v1.2.2 From c9513edb0079f97749c2ac00c887a22c4ba44792 Mon Sep 17 00:00:00 2001 From: Xiao Guangrong Date: Wed, 20 Apr 2011 10:07:30 +0000 Subject: Btrfs: fix the race between reading and updating devices On btrfs_congested_fn and __unplug_io_fn paths, we should hold device_list_mutex to avoid remove/add device path to update fs_devices->devices On __btrfs_close_devices and btrfs_prepare_sprout paths, the devices in fs_devices->devices or fs_devices->devices is updated, so we should hold the mutex to avoid the reader side to reach them Signed-off-by: Xiao Guangrong Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 64b289690f9d..4e53a4fc467f 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1410,6 +1410,7 @@ static int btrfs_congested_fn(void *congested_data, int bdi_bits) struct btrfs_device *device; struct backing_dev_info *bdi; + mutex_lock(&info->fs_devices->device_list_mutex); list_for_each_entry(device, &info->fs_devices->devices, dev_list) { if (!device->bdev) continue; @@ -1419,6 +1420,7 @@ static int btrfs_congested_fn(void *congested_data, int bdi_bits) break; } } + mutex_unlock(&info->fs_devices->device_list_mutex); return ret; } -- cgit v1.2.2 From 1f78160ce1b1b8e657e2248118c4d91f881763f0 Mon Sep 17 00:00:00 2001 From: Xiao Guangrong Date: Wed, 20 Apr 2011 10:09:16 +0000 Subject: Btrfs: using rcu lock in the reader side of devices list fs_devices->devices is only updated on remove and add device paths, so we can use rcu to protect it in the reader side Signed-off-by: Xiao Guangrong Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 4e53a4fc467f..deba3d9c8853 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1410,8 +1410,8 @@ static int btrfs_congested_fn(void *congested_data, int bdi_bits) struct btrfs_device *device; struct backing_dev_info *bdi; - mutex_lock(&info->fs_devices->device_list_mutex); - list_for_each_entry(device, &info->fs_devices->devices, dev_list) { + rcu_read_lock(); + list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) { if (!device->bdev) continue; bdi = blk_get_backing_dev_info(device->bdev); @@ -1420,7 +1420,7 @@ static int btrfs_congested_fn(void *congested_data, int bdi_bits) break; } } - mutex_unlock(&info->fs_devices->device_list_mutex); + rcu_read_unlock(); return ret; } @@ -2332,9 +2332,9 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors) sb = &root->fs_info->super_for_commit; dev_item = &sb->dev_item; - mutex_lock(&root->fs_info->fs_devices->device_list_mutex); + rcu_read_lock(); head = &root->fs_info->fs_devices->devices; - list_for_each_entry(dev, head, dev_list) { + list_for_each_entry_rcu(dev, head, dev_list) { if (!dev->bdev) { total_errors++; continue; @@ -2367,7 +2367,7 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors) } total_errors = 0; - list_for_each_entry(dev, head, dev_list) { + list_for_each_entry_rcu(dev, head, dev_list) { if (!dev->bdev) continue; if (!dev->in_fs_metadata || !dev->writeable) @@ -2377,7 +2377,7 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors) if (ret) total_errors++; } - mutex_unlock(&root->fs_info->fs_devices->device_list_mutex); + rcu_read_unlock(); if (total_errors > max_errors) { printk(KERN_ERR "btrfs: %d errors while writing supers\n", total_errors); -- cgit v1.2.2 From 4cb5300bc839b8a943eb19c9f27f25470e22d0ca Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Tue, 24 May 2011 15:35:30 -0400 Subject: Btrfs: add mount -o auto_defrag This will detect small random writes into files and queue the up for an auto defrag process. It isn't well suited to database workloads yet, but works for smaller files such as rpm, sqlite or bdb databases. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 16d335b342a2..b2588a552658 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1475,6 +1475,7 @@ static int cleaner_kthread(void *arg) btrfs_run_delayed_iputs(root); btrfs_clean_old_snapshots(root); mutex_unlock(&root->fs_info->cleaner_mutex); + btrfs_run_defrag_inodes(root->fs_info); } if (freezing(current)) { @@ -1616,6 +1617,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, spin_lock_init(&fs_info->ref_cache_lock); spin_lock_init(&fs_info->fs_roots_radix_lock); spin_lock_init(&fs_info->delayed_iput_lock); + spin_lock_init(&fs_info->defrag_inodes_lock); init_completion(&fs_info->kobj_unregister); fs_info->tree_root = tree_root; @@ -1638,9 +1640,11 @@ struct btrfs_root *open_ctree(struct super_block *sb, atomic_set(&fs_info->async_delalloc_pages, 0); atomic_set(&fs_info->async_submit_draining, 0); atomic_set(&fs_info->nr_async_bios, 0); + atomic_set(&fs_info->defrag_running, 0); fs_info->sb = sb; fs_info->max_inline = 8192 * 1024; fs_info->metadata_ratio = 0; + fs_info->defrag_inodes = RB_ROOT; fs_info->thread_pool_size = min_t(unsigned long, num_online_cpus() + 2, 8); @@ -2501,6 +2505,14 @@ int close_ctree(struct btrfs_root *root) smp_mb(); btrfs_scrub_cancel(root); + + /* wait for any defraggers to finish */ + wait_event(fs_info->transaction_wait, + (atomic_read(&fs_info->defrag_running) == 0)); + + /* clear out the rbtree of defraggable inodes */ + btrfs_run_defrag_inodes(root->fs_info); + btrfs_put_block_group_cache(fs_info); /* -- cgit v1.2.2 From 174ba50915b08dcfd07c8b5fb795b46a165fa09a Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 27 May 2011 10:03:58 -0400 Subject: Btrfs: use the device_list_mutex during write_dev_supers write_dev_supers was changed to use RCU to protect the list of devices, but it was then sleeping while it actually wrote the supers. This fixes it to just use the mutex, since we really don't any concurrency in write_dev_supers anyway. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 4 ++-- 1 file changed, 2 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 b2588a552658..98b6a71decba 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2310,7 +2310,7 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors) sb = &root->fs_info->super_for_commit; dev_item = &sb->dev_item; - rcu_read_lock(); + 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) { if (!dev->bdev) { @@ -2355,7 +2355,7 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors) if (ret) total_errors++; } - rcu_read_unlock(); + 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); -- cgit v1.2.2 From 632dd772fcbde2ba37c0e8983bd38ef4a1eac906 Mon Sep 17 00:00:00 2001 From: Arne Jansen Date: Fri, 10 Jun 2011 12:07:07 +0200 Subject: btrfs: reinitialize scrub workers Scrub starts the workers each time a scrub starts and stops them after it finished. This patch adds an initialization for the workers before each start, otherwise the workers behave strangely. Signed-off-by: Arne Jansen --- fs/btrfs/disk-io.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index a203d363184d..7bbbfebe47e4 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1668,8 +1668,6 @@ struct btrfs_root *open_ctree(struct super_block *sb, init_waitqueue_head(&fs_info->scrub_pause_wait); init_rwsem(&fs_info->scrub_super_lock); fs_info->scrub_workers_refcnt = 0; - btrfs_init_workers(&fs_info->scrub_workers, "scrub", - fs_info->thread_pool_size, &fs_info->generic_worker); sb->s_blocksize = 4096; sb->s_blocksize_bits = blksize_bits(4096); -- cgit v1.2.2 From 5be76758f35ec6578e5b9b150aa513ac26bd9c54 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 9 Jun 2011 10:02:51 +0000 Subject: btrfs: fix unlocked access of delalloc_inodes list_splice_init will make delalloc_inodes empty, but without a spinlock around, this may produce corrupted list head, accessed in many placess, The race window is very tight and nobody seems to have hit it so far. Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index a203d363184d..33b744a5ac03 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2911,9 +2911,8 @@ static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root) INIT_LIST_HEAD(&splice); - list_splice_init(&root->fs_info->delalloc_inodes, &splice); - spin_lock(&root->fs_info->delalloc_lock); + list_splice_init(&root->fs_info->delalloc_inodes, &splice); while (!list_empty(&splice)) { btrfs_inode = list_entry(splice.next, struct btrfs_inode, -- cgit v1.2.2 From ac08aedfa5d3de0dcb3825b598d16c2e57991f54 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 13 Jun 2011 11:28:50 -0400 Subject: Btrfs: check the return value from set_anon_super Al Viro noticed we weren't checking for set_anon_super failures. This adds the required checks. Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 9f68c6898653..20c111b3fa0d 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1312,7 +1312,9 @@ again: spin_lock_init(&root->cache_lock); init_waitqueue_head(&root->cache_wait); - set_anon_super(&root->anon_super, NULL); + ret = set_anon_super(&root->anon_super, NULL); + if (ret) + goto fail; if (btrfs_root_refs(&root->root_item) == 0) { ret = -ENOENT; -- cgit v1.2.2 From 7585717f304f5ed005cc4ad933a69aab3efbd136 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 13 Jun 2011 20:00:16 -0400 Subject: Btrfs: fix relocation races The recent commit to get rid of our trans_mutex introduced some races with block group relocation. The problem is that relocation needs to do some record keeping about each root, and it was relying on the transaction mutex to coordinate things in subtle ways. This fix adds a mutex just for the relocation code and makes sure it doesn't have a big impact on normal operations. The race is really fixed in btrfs_record_root_in_trans, which is where we step back and wait for the relocation code to finish accounting setup. 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 20c111b3fa0d..0b2b4b759136 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1620,6 +1620,7 @@ struct btrfs_root *open_ctree(struct super_block *sb, spin_lock_init(&fs_info->fs_roots_radix_lock); spin_lock_init(&fs_info->delayed_iput_lock); spin_lock_init(&fs_info->defrag_inodes_lock); + mutex_init(&fs_info->reloc_mutex); init_completion(&fs_info->kobj_unregister); fs_info->tree_root = tree_root; -- cgit v1.2.2 From 9fe6a50fb764f508dd2de47a66e62e51388791fb Mon Sep 17 00:00:00 2001 From: Maarten Lankhorst Date: Thu, 16 Jun 2011 09:04:57 +0000 Subject: btrfs: Remove unused sysfs code Removes code no longer used. The sysfs file itself is kept, because the btrfs developers expressed interest in putting new entries to sysfs. Signed-off-by: Maarten Lankhorst Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 1 - 1 file changed, 1 deletion(-) (limited to 'fs/btrfs/disk-io.c') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 0b2b4b759136..c25ef5a0ccd6 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1044,7 +1044,6 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->last_trans = 0; root->highest_objectid = 0; root->name = NULL; - root->in_sysfs = 0; root->inode_tree = RB_ROOT; INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC); root->block_rsv = NULL; -- cgit v1.2.2 From 35a30d7ce54e087d8025a725d4e5a2fdee723a9f Mon Sep 17 00:00:00 2001 From: David Sterba Date: Mon, 13 Jun 2011 15:18:23 +0000 Subject: btrfs: fix uninitialized return value When allocation fails in btrfs_read_fs_root_no_name, ret is not set although it is returned, holding a garbage value. Signed-off-by: David Sterba Reviewed-by: Li Zefan Signed-off-by: Chris Mason --- 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 c25ef5a0ccd6..1ac8db5dc0a3 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1299,12 +1299,12 @@ again: return root; root->free_ino_ctl = kzalloc(sizeof(*root->free_ino_ctl), GFP_NOFS); - if (!root->free_ino_ctl) - goto fail; root->free_ino_pinned = kzalloc(sizeof(*root->free_ino_pinned), GFP_NOFS); - if (!root->free_ino_pinned) + if (!root->free_ino_pinned || !root->free_ino_ctl) { + ret = -ENOMEM; goto fail; + } btrfs_init_free_ino_ctl(root); mutex_init(&root->fs_commit_mutex); -- cgit v1.2.2