From 53b381b3abeb86f12787a6c40fee9b2f71edc23b Mon Sep 17 00:00:00 2001 From: David Woodhouse Date: Tue, 29 Jan 2013 18:40:14 -0500 Subject: Btrfs: RAID5 and RAID6 This builds on David Woodhouse's original Btrfs raid5/6 implementation. The code has changed quite a bit, blame Chris Mason for any bugs. Read/modify/write is done after the higher levels of the filesystem have prepared a given bio. This means the higher layers are not responsible for building full stripes, and they don't need to query for the topology of the extents that may get allocated during delayed allocation runs. It also means different files can easily share the same stripe. But, it does expose us to incorrect parity if we crash or lose power while doing a read/modify/write cycle. This will be addressed in a later commit. Scrub is unable to repair crc errors on raid5/6 chunks. Discard does not work on raid5/6 (yet) The stripe size is fixed at 64KiB per disk. This will be tunable in a later commit. Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index bdbb94f245c9..bc35ed4238b8 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -28,6 +28,7 @@ #include "dev-replace.h" #include "check-integrity.h" #include "rcu-string.h" +#include "raid56.h" /* * This is only the first step towards a full-features scrub. It reads all @@ -2246,6 +2247,13 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx, struct btrfs_device *extent_dev; int extent_mirror_num; + if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6)) { + if (num >= nr_data_stripes(map)) { + return 0; + } + } + nstripes = length; offset = 0; do_div(nstripes, map->stripe_len); -- cgit v1.2.3 From 6f1c36055f96e80031c7fdda3fd5be826b8d7782 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Tue, 29 Jan 2013 03:22:10 +0000 Subject: Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder Signed-off-by: Liu Bo Signed-off-by: Josef Bacik --- fs/btrfs/file.c | 22 ++++++++++++++++++---- fs/btrfs/scrub.c | 25 ++++++++++++++++++++----- 2 files changed, 38 insertions(+), 9 deletions(-) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index a902faab7161..b06d289f998f 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -293,15 +293,24 @@ static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info, struct btrfs_key key; struct btrfs_ioctl_defrag_range_args range; int num_defrag; + int index; + int ret; /* get the inode */ key.objectid = defrag->root; btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); key.offset = (u64)-1; + + index = srcu_read_lock(&fs_info->subvol_srcu); + inode_root = btrfs_read_fs_root_no_name(fs_info, &key); if (IS_ERR(inode_root)) { - kmem_cache_free(btrfs_inode_defrag_cachep, defrag); - return PTR_ERR(inode_root); + ret = PTR_ERR(inode_root); + goto cleanup; + } + if (btrfs_root_refs(&inode_root->root_item) == 0) { + ret = -ENOENT; + goto cleanup; } key.objectid = defrag->ino; @@ -309,9 +318,10 @@ static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info, key.offset = 0; inode = btrfs_iget(fs_info->sb, &key, inode_root, NULL); if (IS_ERR(inode)) { - kmem_cache_free(btrfs_inode_defrag_cachep, defrag); - return PTR_ERR(inode); + ret = PTR_ERR(inode); + goto cleanup; } + srcu_read_unlock(&fs_info->subvol_srcu, index); /* do a chunk of defrag */ clear_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags); @@ -346,6 +356,10 @@ static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info, iput(inode); return 0; +cleanup: + srcu_read_unlock(&fs_info->subvol_srcu, index); + kmem_cache_free(btrfs_inode_defrag_cachep, defrag); + return ret; } /* diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index bdbb94f245c9..67783e03d121 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -580,20 +580,29 @@ static int scrub_fixup_readpage(u64 inum, u64 offset, u64 root, void *fixup_ctx) int corrected = 0; struct btrfs_key key; struct inode *inode = NULL; + struct btrfs_fs_info *fs_info; u64 end = offset + PAGE_SIZE - 1; struct btrfs_root *local_root; + int srcu_index; key.objectid = root; key.type = BTRFS_ROOT_ITEM_KEY; key.offset = (u64)-1; - local_root = btrfs_read_fs_root_no_name(fixup->root->fs_info, &key); - if (IS_ERR(local_root)) + + fs_info = fixup->root->fs_info; + srcu_index = srcu_read_lock(&fs_info->subvol_srcu); + + local_root = btrfs_read_fs_root_no_name(fs_info, &key); + if (IS_ERR(local_root)) { + srcu_read_unlock(&fs_info->subvol_srcu, srcu_index); return PTR_ERR(local_root); + } key.type = BTRFS_INODE_ITEM_KEY; key.objectid = inum; key.offset = 0; - inode = btrfs_iget(fixup->root->fs_info->sb, &key, local_root, NULL); + inode = btrfs_iget(fs_info->sb, &key, local_root, NULL); + srcu_read_unlock(&fs_info->subvol_srcu, srcu_index); if (IS_ERR(inode)) return PTR_ERR(inode); @@ -606,7 +615,6 @@ static int scrub_fixup_readpage(u64 inum, u64 offset, u64 root, void *fixup_ctx) } if (PageUptodate(page)) { - struct btrfs_fs_info *fs_info; if (PageDirty(page)) { /* * we need to write the data to the defect sector. the @@ -3180,18 +3188,25 @@ static int copy_nocow_pages_for_inode(u64 inum, u64 offset, u64 root, void *ctx) u64 physical_for_dev_replace; u64 len; struct btrfs_fs_info *fs_info = nocow_ctx->sctx->dev_root->fs_info; + int srcu_index; key.objectid = root; key.type = BTRFS_ROOT_ITEM_KEY; key.offset = (u64)-1; + + srcu_index = srcu_read_lock(&fs_info->subvol_srcu); + local_root = btrfs_read_fs_root_no_name(fs_info, &key); - if (IS_ERR(local_root)) + if (IS_ERR(local_root)) { + srcu_read_unlock(&fs_info->subvol_srcu, srcu_index); return PTR_ERR(local_root); + } key.type = BTRFS_INODE_ITEM_KEY; key.objectid = inum; key.offset = 0; inode = btrfs_iget(fs_info->sb, &key, local_root, NULL); + srcu_read_unlock(&fs_info->subvol_srcu, srcu_index); if (IS_ERR(inode)) return PTR_ERR(inode); -- cgit v1.2.3 From 87533c475187c1420794a2e164bc67a7974f1327 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:14:48 +0000 Subject: Btrfs: use bit operation for ->fs_state There is no lock to protect fs_info->fs_state, it will introduce some problems, such as the value may be covered by the other task when several tasks modify it. For example: Task0 - CPU0 Task1 - CPU1 mov %fs_state rax or $0x1 rax mov %fs_state rax or $0x2 rax mov rax %fs_state mov rax %fs_state The expected value is 3, but in fact, it is 2. Though this problem doesn't happen now (because there is only one flag currently), the code is error prone, if we add other flags, the above problem will happen to a certainty. Now we use bit operation for it to fix the above problem. In this way, we can make the code more robust and be easy to add new flags. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 4 +++- fs/btrfs/disk-io.c | 5 +++-- fs/btrfs/file.c | 2 +- fs/btrfs/scrub.c | 2 +- fs/btrfs/super.c | 4 ++-- fs/btrfs/transaction.c | 9 ++++----- 6 files changed, 14 insertions(+), 12 deletions(-) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 5f5c30aaef36..78fba44ad3d9 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -338,7 +338,9 @@ static inline unsigned long btrfs_chunk_item_size(int num_stripes) /* * File system states */ +#define BTRFS_FS_STATE_ERROR 0 +/* Super block flags */ /* Errors detected */ #define BTRFS_SUPER_FLAG_ERROR (1ULL << 2) @@ -1549,7 +1551,7 @@ struct btrfs_fs_info { u64 qgroup_seq; /* filesystem state */ - u64 fs_state; + unsigned long fs_state; struct btrfs_delayed_root *delayed_root; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 00b6742fdde7..8e58a1f90546 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2201,7 +2201,8 @@ int open_ctree(struct super_block *sb, goto fail_alloc; /* check FS state, whether FS is broken. */ - fs_info->fs_state |= btrfs_super_flags(disk_super); + if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_ERROR) + set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state); ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY); if (ret) { @@ -3359,7 +3360,7 @@ int close_ctree(struct btrfs_root *root) printk(KERN_ERR "btrfs: commit super ret %d\n", ret); } - if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) + if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) btrfs_error_commit_super(root); btrfs_put_block_group_cache(fs_info); diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 13c78ea3ebce..75d0fe134be3 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1545,7 +1545,7 @@ static ssize_t btrfs_file_aio_write(struct kiocb *iocb, * although we have opened a file as writable, we have * to stop this write operation to ensure FS consistency. */ - if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) { mutex_unlock(&inode->i_mutex); err = -EROFS; goto out; diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index 67783e03d121..c78b2a3fc335 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -2708,7 +2708,7 @@ static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx, int ret; struct btrfs_root *root = sctx->dev_root; - if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) + if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) return -EIO; gen = root->fs_info->last_trans_committed; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index eda330df45a4..4152f9ea34f5 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -98,7 +98,7 @@ static void __save_error_info(struct btrfs_fs_info *fs_info) * today we only save the error info into ram. Long term we'll * also send it down to the disk */ - fs_info->fs_state = BTRFS_SUPER_FLAG_ERROR; + set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state); } static void save_error_info(struct btrfs_fs_info *fs_info) @@ -114,7 +114,7 @@ static void btrfs_handle_error(struct btrfs_fs_info *fs_info) if (sb->s_flags & MS_RDONLY) return; - if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { sb->s_flags |= MS_RDONLY; printk(KERN_INFO "btrfs is forced readonly\n"); /* diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 34610dc6d140..baf6d74fd0f2 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -61,7 +61,7 @@ static noinline int join_transaction(struct btrfs_root *root, int type) spin_lock(&fs_info->trans_lock); loop: /* The file system has been taken offline. No new transactions. */ - if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { spin_unlock(&fs_info->trans_lock); return -EROFS; } @@ -113,7 +113,7 @@ loop: kmem_cache_free(btrfs_transaction_cachep, cur_trans); cur_trans = fs_info->running_transaction; goto loop; - } else if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + } else if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { spin_unlock(&fs_info->trans_lock); kmem_cache_free(btrfs_transaction_cachep, cur_trans); return -EROFS; @@ -301,7 +301,7 @@ start_transaction(struct btrfs_root *root, u64 num_items, int type, int ret; u64 qgroup_reserved = 0; - if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) + if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) return ERR_PTR(-EROFS); if (current->journal_info) { @@ -645,9 +645,8 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans, btrfs_run_delayed_iputs(root); if (trans->aborted || - root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) err = -EIO; - } assert_qgroups_uptodate(trans); kmem_cache_free(btrfs_trans_handle_cachep, trans); -- cgit v1.2.3