diff options
Diffstat (limited to 'fs/ocfs2')
-rw-r--r-- | fs/ocfs2/alloc.c | 80 | ||||
-rw-r--r-- | fs/ocfs2/blockcheck.c | 184 | ||||
-rw-r--r-- | fs/ocfs2/blockcheck.h | 29 | ||||
-rw-r--r-- | fs/ocfs2/cluster/masklog.h | 35 | ||||
-rw-r--r-- | fs/ocfs2/cluster/tcp.c | 7 | ||||
-rw-r--r-- | fs/ocfs2/dir.c | 21 | ||||
-rw-r--r-- | fs/ocfs2/dlmglue.c | 144 | ||||
-rw-r--r-- | fs/ocfs2/dlmglue.h | 31 | ||||
-rw-r--r-- | fs/ocfs2/file.c | 62 | ||||
-rw-r--r-- | fs/ocfs2/inode.c | 11 | ||||
-rw-r--r-- | fs/ocfs2/journal.c | 126 | ||||
-rw-r--r-- | fs/ocfs2/journal.h | 4 | ||||
-rw-r--r-- | fs/ocfs2/namei.c | 15 | ||||
-rw-r--r-- | fs/ocfs2/ocfs2.h | 26 | ||||
-rw-r--r-- | fs/ocfs2/ocfs2_lockid.h | 5 | ||||
-rw-r--r-- | fs/ocfs2/quota_global.c | 4 | ||||
-rw-r--r-- | fs/ocfs2/quota_local.c | 21 | ||||
-rw-r--r-- | fs/ocfs2/stack_o2cb.c | 11 | ||||
-rw-r--r-- | fs/ocfs2/stack_user.c | 8 | ||||
-rw-r--r-- | fs/ocfs2/stackglue.c | 13 | ||||
-rw-r--r-- | fs/ocfs2/stackglue.h | 6 | ||||
-rw-r--r-- | fs/ocfs2/suballoc.c | 28 | ||||
-rw-r--r-- | fs/ocfs2/super.c | 103 | ||||
-rw-r--r-- | fs/ocfs2/sysfile.c | 19 | ||||
-rw-r--r-- | fs/ocfs2/xattr.c | 5 |
25 files changed, 845 insertions, 153 deletions
diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c index 678a067d925..9edcde4974a 100644 --- a/fs/ocfs2/alloc.c +++ b/fs/ocfs2/alloc.c @@ -475,6 +475,12 @@ struct ocfs2_path { #define path_leaf_el(_path) ((_path)->p_node[(_path)->p_tree_depth].el) #define path_num_items(_path) ((_path)->p_tree_depth + 1) +static int ocfs2_find_path(struct inode *inode, struct ocfs2_path *path, + u32 cpos); +static void ocfs2_adjust_rightmost_records(struct inode *inode, + handle_t *handle, + struct ocfs2_path *path, + struct ocfs2_extent_rec *insert_rec); /* * Reset the actual path elements so that we can re-use the structure * to build another path. Generally, this involves freeing the buffer @@ -1013,6 +1019,54 @@ static inline u32 ocfs2_sum_rightmost_rec(struct ocfs2_extent_list *el) } /* + * Change range of the branches in the right most path according to the leaf + * extent block's rightmost record. + */ +static int ocfs2_adjust_rightmost_branch(handle_t *handle, + struct inode *inode, + struct ocfs2_extent_tree *et) +{ + int status; + struct ocfs2_path *path = NULL; + struct ocfs2_extent_list *el; + struct ocfs2_extent_rec *rec; + + path = ocfs2_new_path_from_et(et); + if (!path) { + status = -ENOMEM; + return status; + } + + status = ocfs2_find_path(inode, path, UINT_MAX); + if (status < 0) { + mlog_errno(status); + goto out; + } + + status = ocfs2_extend_trans(handle, path_num_items(path) + + handle->h_buffer_credits); + if (status < 0) { + mlog_errno(status); + goto out; + } + + status = ocfs2_journal_access_path(inode, handle, path); + if (status < 0) { + mlog_errno(status); + goto out; + } + + el = path_leaf_el(path); + rec = &el->l_recs[le32_to_cpu(el->l_next_free_rec) - 1]; + + ocfs2_adjust_rightmost_records(inode, handle, path, rec); + +out: + ocfs2_free_path(path); + return status; +} + +/* * Add an entire tree branch to our inode. eb_bh is the extent block * to start at, if we don't want to start the branch at the dinode * structure. @@ -1038,7 +1092,7 @@ static int ocfs2_add_branch(struct ocfs2_super *osb, struct ocfs2_extent_block *eb; struct ocfs2_extent_list *eb_el; struct ocfs2_extent_list *el; - u32 new_cpos; + u32 new_cpos, root_end; mlog_entry_void(); @@ -1055,6 +1109,27 @@ static int ocfs2_add_branch(struct ocfs2_super *osb, new_blocks = le16_to_cpu(el->l_tree_depth); + eb = (struct ocfs2_extent_block *)(*last_eb_bh)->b_data; + new_cpos = ocfs2_sum_rightmost_rec(&eb->h_list); + root_end = ocfs2_sum_rightmost_rec(et->et_root_el); + + /* + * If there is a gap before the root end and the real end + * of the righmost leaf block, we need to remove the gap + * between new_cpos and root_end first so that the tree + * is consistent after we add a new branch(it will start + * from new_cpos). + */ + if (root_end > new_cpos) { + mlog(0, "adjust the cluster end from %u to %u\n", + root_end, new_cpos); + status = ocfs2_adjust_rightmost_branch(handle, inode, et); + if (status) { + mlog_errno(status); + goto bail; + } + } + /* allocate the number of new eb blocks we need */ new_eb_bhs = kcalloc(new_blocks, sizeof(struct buffer_head *), GFP_KERNEL); @@ -1071,9 +1146,6 @@ static int ocfs2_add_branch(struct ocfs2_super *osb, goto bail; } - eb = (struct ocfs2_extent_block *)(*last_eb_bh)->b_data; - new_cpos = ocfs2_sum_rightmost_rec(&eb->h_list); - /* Note: new_eb_bhs[new_blocks - 1] is the guy which will be * linked with the rest of the tree. * conversly, new_eb_bhs[0] is the new bottommost leaf. diff --git a/fs/ocfs2/blockcheck.c b/fs/ocfs2/blockcheck.c index 2a947c44e59..a1163b8b417 100644 --- a/fs/ocfs2/blockcheck.c +++ b/fs/ocfs2/blockcheck.c @@ -22,6 +22,9 @@ #include <linux/crc32.h> #include <linux/buffer_head.h> #include <linux/bitops.h> +#include <linux/debugfs.h> +#include <linux/module.h> +#include <linux/fs.h> #include <asm/byteorder.h> #include <cluster/masklog.h> @@ -222,6 +225,155 @@ void ocfs2_hamming_fix_block(void *data, unsigned int blocksize, ocfs2_hamming_fix(data, blocksize * 8, 0, fix); } + +/* + * Debugfs handling. + */ + +#ifdef CONFIG_DEBUG_FS + +static int blockcheck_u64_get(void *data, u64 *val) +{ + *val = *(u64 *)data; + return 0; +} +DEFINE_SIMPLE_ATTRIBUTE(blockcheck_fops, blockcheck_u64_get, NULL, "%llu\n"); + +static struct dentry *blockcheck_debugfs_create(const char *name, + struct dentry *parent, + u64 *value) +{ + return debugfs_create_file(name, S_IFREG | S_IRUSR, parent, value, + &blockcheck_fops); +} + +static void ocfs2_blockcheck_debug_remove(struct ocfs2_blockcheck_stats *stats) +{ + if (stats) { + debugfs_remove(stats->b_debug_check); + stats->b_debug_check = NULL; + debugfs_remove(stats->b_debug_failure); + stats->b_debug_failure = NULL; + debugfs_remove(stats->b_debug_recover); + stats->b_debug_recover = NULL; + debugfs_remove(stats->b_debug_dir); + stats->b_debug_dir = NULL; + } +} + +static int ocfs2_blockcheck_debug_install(struct ocfs2_blockcheck_stats *stats, + struct dentry *parent) +{ + int rc = -EINVAL; + + if (!stats) + goto out; + + stats->b_debug_dir = debugfs_create_dir("blockcheck", parent); + if (!stats->b_debug_dir) + goto out; + + stats->b_debug_check = + blockcheck_debugfs_create("blocks_checked", + stats->b_debug_dir, + &stats->b_check_count); + + stats->b_debug_failure = + blockcheck_debugfs_create("checksums_failed", + stats->b_debug_dir, + &stats->b_failure_count); + + stats->b_debug_recover = + blockcheck_debugfs_create("ecc_recoveries", + stats->b_debug_dir, + &stats->b_recover_count); + if (stats->b_debug_check && stats->b_debug_failure && + stats->b_debug_recover) + rc = 0; + +out: + if (rc) + ocfs2_blockcheck_debug_remove(stats); + return rc; +} +#else +static inline int ocfs2_blockcheck_debug_install(struct ocfs2_blockcheck_stats *stats, + struct dentry *parent) +{ + return 0; +} + +static inline void ocfs2_blockcheck_debug_remove(struct ocfs2_blockcheck_stats *stats) +{ +} +#endif /* CONFIG_DEBUG_FS */ + +/* Always-called wrappers for starting and stopping the debugfs files */ +int ocfs2_blockcheck_stats_debugfs_install(struct ocfs2_blockcheck_stats *stats, + struct dentry *parent) +{ + return ocfs2_blockcheck_debug_install(stats, parent); +} + +void ocfs2_blockcheck_stats_debugfs_remove(struct ocfs2_blockcheck_stats *stats) +{ + ocfs2_blockcheck_debug_remove(stats); +} + +static void ocfs2_blockcheck_inc_check(struct ocfs2_blockcheck_stats *stats) +{ + u64 new_count; + + if (!stats) + return; + + spin_lock(&stats->b_lock); + stats->b_check_count++; + new_count = stats->b_check_count; + spin_unlock(&stats->b_lock); + + if (!new_count) + mlog(ML_NOTICE, "Block check count has wrapped\n"); +} + +static void ocfs2_blockcheck_inc_failure(struct ocfs2_blockcheck_stats *stats) +{ + u64 new_count; + + if (!stats) + return; + + spin_lock(&stats->b_lock); + stats->b_failure_count++; + new_count = stats->b_failure_count; + spin_unlock(&stats->b_lock); + + if (!new_count) + mlog(ML_NOTICE, "Checksum failure count has wrapped\n"); +} + +static void ocfs2_blockcheck_inc_recover(struct ocfs2_blockcheck_stats *stats) +{ + u64 new_count; + + if (!stats) + return; + + spin_lock(&stats->b_lock); + stats->b_recover_count++; + new_count = stats->b_recover_count; + spin_unlock(&stats->b_lock); + + if (!new_count) + mlog(ML_NOTICE, "ECC recovery count has wrapped\n"); +} + + + +/* + * These are the low-level APIs for using the ocfs2_block_check structure. + */ + /* * This function generates check information for a block. * data is the block to be checked. bc is a pointer to the @@ -266,12 +418,15 @@ void ocfs2_block_check_compute(void *data, size_t blocksize, * Again, the data passed in should be the on-disk endian. */ int ocfs2_block_check_validate(void *data, size_t blocksize, - struct ocfs2_block_check *bc) + struct ocfs2_block_check *bc, + struct ocfs2_blockcheck_stats *stats) { int rc = 0; struct ocfs2_block_check check; u32 crc, ecc; + ocfs2_blockcheck_inc_check(stats); + check.bc_crc32e = le32_to_cpu(bc->bc_crc32e); check.bc_ecc = le16_to_cpu(bc->bc_ecc); @@ -282,6 +437,7 @@ int ocfs2_block_check_validate(void *data, size_t blocksize, if (crc == check.bc_crc32e) goto out; + ocfs2_blockcheck_inc_failure(stats); mlog(ML_ERROR, "CRC32 failed: stored: %u, computed %u. Applying ECC.\n", (unsigned int)check.bc_crc32e, (unsigned int)crc); @@ -292,8 +448,10 @@ int ocfs2_block_check_validate(void *data, size_t blocksize, /* And check the crc32 again */ crc = crc32_le(~0, data, blocksize); - if (crc == check.bc_crc32e) + if (crc == check.bc_crc32e) { + ocfs2_blockcheck_inc_recover(stats); goto out; + } mlog(ML_ERROR, "Fixed CRC32 failed: stored: %u, computed %u\n", (unsigned int)check.bc_crc32e, (unsigned int)crc); @@ -366,7 +524,8 @@ void ocfs2_block_check_compute_bhs(struct buffer_head **bhs, int nr, * Again, the data passed in should be the on-disk endian. */ int ocfs2_block_check_validate_bhs(struct buffer_head **bhs, int nr, - struct ocfs2_block_check *bc) + struct ocfs2_block_check *bc, + struct ocfs2_blockcheck_stats *stats) { int i, rc = 0; struct ocfs2_block_check check; @@ -377,6 +536,8 @@ int ocfs2_block_check_validate_bhs(struct buffer_head **bhs, int nr, if (!nr) return 0; + ocfs2_blockcheck_inc_check(stats); + check.bc_crc32e = le32_to_cpu(bc->bc_crc32e); check.bc_ecc = le16_to_cpu(bc->bc_ecc); @@ -388,6 +549,7 @@ int ocfs2_block_check_validate_bhs(struct buffer_head **bhs, int nr, if (crc == check.bc_crc32e) goto out; + ocfs2_blockcheck_inc_failure(stats); mlog(ML_ERROR, "CRC32 failed: stored: %u, computed %u. Applying ECC.\n", (unsigned int)check.bc_crc32e, (unsigned int)crc); @@ -416,8 +578,10 @@ int ocfs2_block_check_validate_bhs(struct buffer_head **bhs, int nr, /* And check the crc32 again */ for (i = 0, crc = ~0; i < nr; i++) crc = crc32_le(crc, bhs[i]->b_data, bhs[i]->b_size); - if (crc == check.bc_crc32e) + if (crc == check.bc_crc32e) { + ocfs2_blockcheck_inc_recover(stats); goto out; + } mlog(ML_ERROR, "Fixed CRC32 failed: stored: %u, computed %u\n", (unsigned int)check.bc_crc32e, (unsigned int)crc); @@ -448,9 +612,11 @@ int ocfs2_validate_meta_ecc(struct super_block *sb, void *data, struct ocfs2_block_check *bc) { int rc = 0; + struct ocfs2_super *osb = OCFS2_SB(sb); - if (ocfs2_meta_ecc(OCFS2_SB(sb))) - rc = ocfs2_block_check_validate(data, sb->s_blocksize, bc); + if (ocfs2_meta_ecc(osb)) + rc = ocfs2_block_check_validate(data, sb->s_blocksize, bc, + &osb->osb_ecc_stats); return rc; } @@ -468,9 +634,11 @@ int ocfs2_validate_meta_ecc_bhs(struct super_block *sb, struct ocfs2_block_check *bc) { int rc = 0; + struct ocfs2_super *osb = OCFS2_SB(sb); - if (ocfs2_meta_ecc(OCFS2_SB(sb))) - rc = ocfs2_block_check_validate_bhs(bhs, nr, bc); + if (ocfs2_meta_ecc(osb)) + rc = ocfs2_block_check_validate_bhs(bhs, nr, bc, + &osb->osb_ecc_stats); return rc; } diff --git a/fs/ocfs2/blockcheck.h b/fs/ocfs2/blockcheck.h index 70ec3feda32..d4b69febf70 100644 --- a/fs/ocfs2/blockcheck.h +++ b/fs/ocfs2/blockcheck.h @@ -21,6 +21,24 @@ #define OCFS2_BLOCKCHECK_H +/* Count errors and error correction from blockcheck.c */ +struct ocfs2_blockcheck_stats { + spinlock_t b_lock; + u64 b_check_count; /* Number of blocks we've checked */ + u64 b_failure_count; /* Number of failed checksums */ + u64 b_recover_count; /* Number of blocks fixed by ecc */ + + /* + * debugfs entries, used if this is passed to + * ocfs2_blockcheck_stats_debugfs_install() + */ + struct dentry *b_debug_dir; /* Parent of the debugfs files */ + struct dentry *b_debug_check; /* Exposes b_check_count */ + struct dentry *b_debug_failure; /* Exposes b_failure_count */ + struct dentry *b_debug_recover; /* Exposes b_recover_count */ +}; + + /* High level block API */ void ocfs2_compute_meta_ecc(struct super_block *sb, void *data, struct ocfs2_block_check *bc); @@ -37,11 +55,18 @@ int ocfs2_validate_meta_ecc_bhs(struct super_block *sb, void ocfs2_block_check_compute(void *data, size_t blocksize, struct ocfs2_block_check *bc); int ocfs2_block_check_validate(void *data, size_t blocksize, - struct ocfs2_block_check *bc); + struct ocfs2_block_check *bc, + struct ocfs2_blockcheck_stats *stats); void ocfs2_block_check_compute_bhs(struct buffer_head **bhs, int nr, struct ocfs2_block_check *bc); int ocfs2_block_check_validate_bhs(struct buffer_head **bhs, int nr, - struct ocfs2_block_check *bc); + struct ocfs2_block_check *bc, + struct ocfs2_blockcheck_stats *stats); + +/* Debug Initialization */ +int ocfs2_blockcheck_stats_debugfs_install(struct ocfs2_blockcheck_stats *stats, + struct dentry *parent); +void ocfs2_blockcheck_stats_debugfs_remove(struct ocfs2_blockcheck_stats *stats); /* * Hamming code functions diff --git a/fs/ocfs2/cluster/masklog.h b/fs/ocfs2/cluster/masklog.h index 7e72a81bc2d..696c32e5071 100644 --- a/fs/ocfs2/cluster/masklog.h +++ b/fs/ocfs2/cluster/masklog.h @@ -48,34 +48,33 @@ * only emit the appropriage printk() when the caller passes in a constant * mask, as is almost always the case. * - * All this bitmask nonsense is hidden from the /proc interface so that Joel - * doesn't have an aneurism. Reading the file gives a straight forward - * indication of which bits are on or off: - * ENTRY off - * EXIT off + * All this bitmask nonsense is managed from the files under + * /sys/fs/o2cb/logmask/. Reading the files gives a straightforward + * indication of which bits are allowed (allow) or denied (off/deny). + * ENTRY deny + * EXIT deny * TCP off * MSG off * SOCKET off - * ERROR off - * NOTICE on + * ERROR allow + * NOTICE allow * * Writing changes the state of a given bit and requires a strictly formatted * single write() call: * - * write(fd, "ENTRY on", 8); + * write(fd, "allow", 5); * - * would turn the entry bit on. "1" is also accepted in the place of "on", and - * "off" and "0" behave as expected. + * Echoing allow/deny/off string into the logmask files can flip the bits + * on or off as expected; here is the bash script for example: * - * Some trivial shell can flip all the bits on or off: + * log_mask="/sys/fs/o2cb/log_mask" + * for node in ENTRY EXIT TCP MSG SOCKET ERROR NOTICE; do + * echo allow >"$log_mask"/"$node" + * done * - * log_mask="/proc/fs/ocfs2_nodemanager/log_mask" - * cat $log_mask | ( - * while read bit status; do - * # $1 is "on" or "off", say - * echo "$bit $1" > $log_mask - * done - * ) + * The debugfs.ocfs2 tool can also flip the bits with the -l option: + * + * debugfs.ocfs2 -l TCP allow */ /* for task_struct */ diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c index 9fbe849f634..334f231a422 100644 --- a/fs/ocfs2/cluster/tcp.c +++ b/fs/ocfs2/cluster/tcp.c @@ -974,7 +974,7 @@ static int o2net_tx_can_proceed(struct o2net_node *nn, int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *caller_vec, size_t caller_veclen, u8 target_node, int *status) { - int ret, error = 0; + int ret; struct o2net_msg *msg = NULL; size_t veclen, caller_bytes = 0; struct kvec *vec = NULL; @@ -1015,10 +1015,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *caller_vec, o2net_set_nst_sock_time(&nst); - ret = wait_event_interruptible(nn->nn_sc_wq, - o2net_tx_can_proceed(nn, &sc, &error)); - if (!ret && error) - ret = error; + wait_event(nn->nn_sc_wq, o2net_tx_can_proceed(nn, &sc, &ret)); if (ret) goto out; diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c index c5752305627..b358f3bf896 100644 --- a/fs/ocfs2/dir.c +++ b/fs/ocfs2/dir.c @@ -2900,6 +2900,8 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh, alloc = ocfs2_clusters_for_bytes(sb, bytes); dx_alloc = 0; + down_write(&oi->ip_alloc_sem); + if (ocfs2_supports_indexed_dirs(osb)) { credits += ocfs2_add_dir_index_credits(sb); @@ -2940,8 +2942,6 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh, goto out; } - down_write(&oi->ip_alloc_sem); - /* * Prepare for worst case allocation scenario of two separate * extents in the unindexed tree. @@ -2953,7 +2953,7 @@ static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh, if (IS_ERR(handle)) { ret = PTR_ERR(handle); mlog_errno(ret); - goto out_sem; + goto out; } if (vfs_dq_alloc_space_nodirty(dir, @@ -3172,10 +3172,8 @@ out_commit: ocfs2_commit_trans(osb, handle); -out_sem: - up_write(&oi->ip_alloc_sem); - out: + up_write(&oi->ip_alloc_sem); if (data_ac) ocfs2_free_alloc_context(data_ac); if (meta_ac) @@ -3322,11 +3320,15 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb, brelse(new_bh); new_bh = NULL; + down_write(&OCFS2_I(dir)->ip_alloc_sem); + drop_alloc_sem = 1; dir_i_size = i_size_read(dir); credits = OCFS2_SIMPLE_DIR_EXTEND_CREDITS; goto do_extend; } + down_write(&OCFS2_I(dir)->ip_alloc_sem); + drop_alloc_sem = 1; dir_i_size = i_size_read(dir); mlog(0, "extending dir %llu (i_size = %lld)\n", (unsigned long long)OCFS2_I(dir)->ip_blkno, dir_i_size); @@ -3370,9 +3372,6 @@ do_extend: credits++; /* For attaching the new dirent block to the * dx_root */ - down_write(&OCFS2_I(dir)->ip_alloc_sem); - drop_alloc_sem = 1; - handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) { status = PTR_ERR(handle); @@ -3435,10 +3434,10 @@ bail_bh: *new_de_bh = new_bh; get_bh(*new_de_bh); bail: - if (drop_alloc_sem) - up_write(&OCFS2_I(dir)->ip_alloc_sem); if (handle) ocfs2_commit_trans(osb, handle); + if (drop_alloc_sem) + up_write(&OCFS2_I(dir)->ip_alloc_sem); if (data_ac) ocfs2_free_alloc_context(data_ac); diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index e15fc7d5082..110bb57c46a 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -92,6 +92,9 @@ struct ocfs2_unblock_ctl { enum ocfs2_unblock_action unblock_action; }; +/* Lockdep class keys */ +struct lock_class_key lockdep_keys[OCFS2_NUM_LOCK_TYPES]; + static int ocfs2_check_meta_downconvert(struct ocfs2_lock_res *lockres, int new_level); static void ocfs2_set_meta_lvb(struct ocfs2_lock_res *lockres); @@ -248,6 +251,10 @@ static struct ocfs2_lock_res_ops ocfs2_nfs_sync_lops = { .flags = 0, }; +static struct ocfs2_lock_res_ops ocfs2_orphan_scan_lops = { + .flags = LOCK_TYPE_REQUIRES_REFRESH|LOCK_TYPE_USES_LVB, +}; + static struct ocfs2_lock_res_ops ocfs2_dentry_lops = { .get_osb = ocfs2_get_dentry_osb, .post_unlock = ocfs2_dentry_post_unlock, @@ -313,9 +320,16 @@ static int ocfs2_lock_create(struct ocfs2_super *osb, u32 dlm_flags); static inline int ocfs2_may_continue_on_blocked_lock(struct ocfs2_lock_res *lockres, int wanted); -static void ocfs2_cluster_unlock(struct ocfs2_super *osb, - struct ocfs2_lock_res *lockres, - int level); +static void __ocfs2_cluster_unlock(struct ocfs2_super *osb, + struct ocfs2_lock_res *lockres, + int level, unsigned long caller_ip); +static inline void ocfs2_cluster_unlock(struct ocfs2_super *osb, + struct ocfs2_lock_res *lockres, + int level) +{ + __ocfs2_cluster_unlock(osb, lockres, level, _RET_IP_); +} + static inline void ocfs2_generic_handle_downconvert_action(struct ocfs2_lock_res *lockres); static inline void ocfs2_generic_handle_convert_action(struct ocfs2_lock_res *lockres); static inline void ocfs2_generic_handle_attach_action(struct ocfs2_lock_res *lockres); @@ -485,6 +499,13 @@ static void ocfs2_lock_res_init_common(struct ocfs2_super *osb, ocfs2_add_lockres_tracking(res, osb->osb_dlm_debug); ocfs2_init_lock_stats(res); +#ifdef CONFIG_DEBUG_LOCK_ALLOC + if (type != OCFS2_LOCK_TYPE_OPEN) + lockdep_init_map(&res->l_lockdep_map, ocfs2_lock_type_strings[type], + &lockdep_keys[type], 0); + else + res->l_lockdep_map.key = NULL; +#endif } void ocfs2_lock_res_init_once(struct ocfs2_lock_res *res) @@ -637,6 +658,15 @@ static void ocfs2_nfs_sync_lock_res_init(struct ocfs2_lock_res *res, &ocfs2_nfs_sync_lops, osb); } +static void ocfs2_orphan_scan_lock_res_init(struct ocfs2_lock_res *res, + struct ocfs2_super *osb) +{ + ocfs2_lock_res_init_once(res); + ocfs2_build_lock_name(OCFS2_LOCK_TYPE_ORPHAN_SCAN, 0, 0, res->l_name); + ocfs2_lock_res_init_common(osb, res, OCFS2_LOCK_TYPE_ORPHAN_SCAN, + &ocfs2_orphan_scan_lops, osb); +} + void ocfs2_file_lock_res_init(struct ocfs2_lock_res *lockres, struct ocfs2_file_private *fp) { @@ -1239,11 +1269,13 @@ static int ocfs2_wait_for_mask_interruptible(struct ocfs2_mask_waiter *mw, return ret; } -static int ocfs2_cluster_lock(struct ocfs2_super *osb, - struct ocfs2_lock_res *lockres, - int level, - u32 lkm_flags, - int arg_flags) +static int __ocfs2_cluster_lock(struct ocfs2_super *osb, + struct ocfs2_lock_res *lockres, + int level, + u32 lkm_flags, + int arg_flags, + int l_subclass, + unsigned long caller_ip) { struct ocfs2_mask_waiter mw; int wait, catch_signals = !(osb->s_mount_opt & OCFS2_MOUNT_NOINTR); @@ -1386,13 +1418,37 @@ out: } ocfs2_update_lock_stats(lockres, level, &mw, ret); +#ifdef CONFIG_DEBUG_LOCK_ALLOC + if (!ret && lockres->l_lockdep_map.key != NULL) { + if (level == DLM_LOCK_PR) + rwsem_acquire_read(&lockres->l_lockdep_map, l_subclass, + !!(arg_flags & OCFS2_META_LOCK_NOQUEUE), + caller_ip); + else + rwsem_acquire(&lockres->l_lockdep_map, l_subclass, + !!(arg_flags & OCFS2_META_LOCK_NOQUEUE), + caller_ip); + } +#endif mlog_exit(ret); return ret; } -static void ocfs2_cluster_unlock(struct ocfs2_super *osb, - struct ocfs2_lock_res *lockres, - int level) +static inline int ocfs2_cluster_lock(struct ocfs2_super *osb, + struct ocfs2_lock_res *lockres, + int level, + u32 lkm_flags, + int arg_flags) +{ + return __ocfs2_cluster_lock(osb, lockres, level, lkm_flags, arg_flags, + 0, _RET_IP_); +} + + +static void __ocfs2_cluster_unlock(struct ocfs2_super *osb, + struct ocfs2_lock_res *lockres, + int level, + unsigned long caller_ip) { unsigned long flags; @@ -1401,6 +1457,10 @@ static void ocfs2_cluster_unlock(struct ocfs2_super *osb, ocfs2_dec_holders(lockres, level); ocfs2_downconvert_on_unlock(osb, lockres); spin_unlock_irqrestore(&lockres->l_lock, flags); +#ifdef CONFIG_DEBUG_LOCK_ALLOC + if (lockres->l_lockdep_map.key != NULL) + rwsem_release(&lockres->l_lockdep_map, 1, caller_ip); +#endif mlog_exit_void(); } @@ -1972,7 +2032,8 @@ static inline int ocfs2_meta_lvb_is_trustable(struct inode *inode, { struct ocfs2_meta_lvb *lvb = ocfs2_dlm_lvb(&lockres->l_lksb); - if (lvb->lvb_version == OCFS2_LVB_VERSION + if (ocfs2_dlm_lvb_valid(&lockres->l_lksb) + && lvb->lvb_version == OCFS2_LVB_VERSION && be32_to_cpu(lvb->lvb_igeneration) == inode->i_generation) return 1; return 0; @@ -2145,10 +2206,11 @@ static int ocfs2_assign_bh(struct inode *inode, * returns < 0 error if the callback will never be called, otherwise * the result of the lock will be communicated via the callback. */ -int ocfs2_inode_lock_full(struct inode *inode, - struct buffer_head **ret_bh, - int ex, - int arg_flags) +int ocfs2_inode_lock_full_nested(struct inode *inode, + struct buffer_head **ret_bh, + int ex, + int arg_flags, + int subclass) { int status, level, acquired; u32 dlm_flags; @@ -2186,7 +2248,8 @@ int ocfs2_inode_lock_full(struct inode *inode, if (arg_flags & OCFS2_META_LOCK_NOQUEUE) dlm_flags |= DLM_LKF_NOQUEUE; - status = ocfs2_cluster_lock(osb, lockres, level, dlm_flags, arg_flags); + status = __ocfs2_cluster_lock(osb, lockres, level, dlm_flags, + arg_flags, subclass, _RET_IP_); if (status < 0) { if (status != -EAGAIN && status != -EIOCBRETRY) mlog_errno(status); @@ -2352,6 +2415,47 @@ void ocfs2_inode_unlock(struct inode *inode, mlog_exit_void(); } +int ocfs2_orphan_scan_lock(struct ocfs2_super *osb, u32 *seqno) +{ + struct ocfs2_lock_res *lockres; + struct ocfs2_orphan_scan_lvb *lvb; + int status = 0; + + if (ocfs2_is_hard_readonly(osb)) + return -EROFS; + + if (ocfs2_mount_local(osb)) + return 0; + + lockres = &osb->osb_orphan_scan.os_lockres; + status = ocfs2_cluster_lock(osb, lockres, DLM_LOCK_EX, 0, 0); + if (status < 0) + return status; + + lvb = ocfs2_dlm_lvb(&lockres->l_lksb); + if (ocfs2_dlm_lvb_valid(&lockres->l_lksb) && + lvb->lvb_version == OCFS2_ORPHAN_LVB_VERSION) + *seqno = be32_to_cpu(lvb->lvb_os_seqno); + else + *seqno = osb->osb_orphan_scan.os_seqno + 1; + + return status; +} + +void ocfs2_orphan_scan_unlock(struct ocfs2_super *osb, u32 seqno) +{ + struct ocfs2_lock_res *lockres; + struct ocfs2_orphan_scan_lvb *lvb; + + if (!ocfs2_is_hard_readonly(osb) && !ocfs2_mount_local(osb)) { + lockres = &osb->osb_orphan_scan.os_lockres; + lvb = ocfs2_dlm_lvb(&lockres->l_lksb); + lvb->lvb_version = OCFS2_ORPHAN_LVB_VERSION; + lvb->lvb_os_seqno = cpu_to_be32(seqno); + ocfs2_cluster_unlock(osb, lockres, DLM_LOCK_EX); + } +} + int ocfs2_super_lock(struct ocfs2_super *osb, int ex) { @@ -2842,6 +2946,7 @@ local: ocfs2_super_lock_res_init(&osb->osb_super_lockres, osb); ocfs2_rename_lock_res_init(&osb->osb_rename_lockres, osb); ocfs2_nfs_sync_lock_res_init(&osb->osb_nfs_sync_lockres, osb); + ocfs2_orphan_scan_lock_res_init(&osb->osb_orphan_scan.os_lockres, osb); osb->cconn = conn; @@ -2878,6 +2983,7 @@ void ocfs2_dlm_shutdown(struct ocfs2_super *osb, ocfs2_lock_res_free(&osb->osb_super_lockres); ocfs2_lock_res_free(&osb->osb_rename_lockres); ocfs2_lock_res_free(&osb->osb_nfs_sync_lockres); + ocfs2_lock_res_free(&osb->osb_orphan_scan.os_lockres); ocfs2_cluster_disconnect(osb->cconn, hangup_pending); osb->cconn = NULL; @@ -3061,6 +3167,7 @@ static void ocfs2_drop_osb_locks(struct ocfs2_super *osb) ocfs2_simple_drop_lockres(osb, &osb->osb_super_lockres); ocfs2_simple_drop_lockres(osb, &osb->osb_rename_lockres); ocfs2_simple_drop_lockres(osb, &osb->osb_nfs_sync_lockres); + ocfs2_simple_drop_lockres(osb, &osb->osb_orphan_scan.os_lockres); } int ocfs2_drop_inode_locks(struct inode *inode) @@ -3576,7 +3683,8 @@ static int ocfs2_refresh_qinfo(struct ocfs2_mem_dqinfo *oinfo) struct ocfs2_global_disk_dqinfo *gdinfo; int status = 0; - if (lvb->lvb_version == OCFS2_QINFO_LVB_VERSION) { + if (ocfs2_dlm_lvb_valid(&lockres->l_lksb) && + lvb->lvb_version == OCFS2_QINFO_LVB_VERSION) { info->dqi_bgrace = be32_to_cpu(lvb->lvb_bgrace); info->dqi_igrace = be32_to_cpu(lvb->lvb_igrace); oinfo->dqi_syncms = be32_to_cpu(lvb->lvb_syncms); diff --git a/fs/ocfs2/dlmglue.h b/fs/ocfs2/dlmglue.h index e1fd5721cd7..7553836931d 100644 --- a/fs/ocfs2/dlmglue.h +++ b/fs/ocfs2/dlmglue.h @@ -62,6 +62,14 @@ struct ocfs2_qinfo_lvb { __be32 lvb_free_entry; }; +#define OCFS2_ORPHAN_LVB_VERSION 1 + +struct ocfs2_orphan_scan_lvb { + __u8 lvb_version; + __u8 lvb_reserved[3]; + __be32 lvb_os_seqno; +}; + /* ocfs2_inode_lock_full() 'arg_flags' flags */ /* don't wait on recovery. */ #define OCFS2_META_LOCK_RECOVERY (0x01) @@ -70,6 +78,14 @@ struct ocfs2_qinfo_lvb { /* don't block waiting for the downconvert thread, instead return -EAGAIN */ #define OCFS2_LOCK_NONBLOCK (0x04) +/* Locking subclasses of inode cluster lock */ +enum { + OI_LS_NORMAL = 0, + OI_LS_PARENT, + OI_LS_RENAME1, + OI_LS_RENAME2, +}; + int ocfs2_dlm_init(struct ocfs2_super *osb); void ocfs2_dlm_shutdown(struct ocfs2_super *osb, int hangup_pending); void ocfs2_lock_res_init_once(struct ocfs2_lock_res *res); @@ -96,23 +112,32 @@ void ocfs2_open_unlock(struct inode *inode); int ocfs2_inode_lock_atime(struct inode *inode, struct vfsmount *vfsmnt, int *level); -int ocfs2_inode_lock_full(struct inode *inode, +int ocfs2_inode_lock_full_nested(struct inode *inode, struct buffer_head **ret_bh, int ex, - int arg_flags); + int arg_flags, + int subclass); int ocfs2_inode_lock_with_page(struct inode *inode, struct buffer_head **ret_bh, int ex, struct page *page); +/* Variants without special locking class or flags */ +#define ocfs2_inode_lock_full(i, r, e, f)\ + ocfs2_inode_lock_full_nested(i, r, e, f, OI_LS_NORMAL) +#define ocfs2_inode_lock_nested(i, b, e, s)\ + ocfs2_inode_lock_full_nested(i, b, e, 0, s) /* 99% of the time we don't want to supply any additional flags -- * those are for very specific cases only. */ -#define ocfs2_inode_lock(i, b, e) ocfs2_inode_lock_full(i, b, e, 0) +#define ocfs2_inode_lock(i, b, e) ocfs2_inode_lock_full_nested(i, b, e, 0, OI_LS_NORMAL) void ocfs2_inode_unlock(struct inode *inode, int ex); int ocfs2_super_lock(struct ocfs2_super *osb, int ex); void ocfs2_super_unlock(struct ocfs2_super *osb, int ex); +int ocfs2_orphan_scan_lock(struct ocfs2_super *osb, u32 *seqno); +void ocfs2_orphan_scan_unlock(struct ocfs2_super *osb, u32 seqno); + int ocfs2_rename_lock(struct ocfs2_super *osb); void ocfs2_rename_unlock(struct ocfs2_super *osb); int ocfs2_nfs_sync_lock(struct ocfs2_super *osb, int ex); diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index c2a87c885b7..62442e413a0 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -187,6 +187,9 @@ static int ocfs2_sync_file(struct file *file, if (err) goto bail; + if (datasync && !(inode->i_state & I_DIRTY_DATASYNC)) + goto bail; + journal = osb->journal->j_journal; err = jbd2_journal_force_commit(journal); @@ -894,9 +897,9 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr) struct ocfs2_super *osb = OCFS2_SB(sb); struct buffer_head *bh = NULL; handle_t *handle = NULL; - int locked[MAXQUOTAS] = {0, 0}; - int credits, qtype; - struct ocfs2_mem_dqinfo *oinfo; + int qtype; + struct dquot *transfer_from[MAXQUOTAS] = { }; + struct dquot *transfer_to[MAXQUOTAS] = { }; mlog_entry("(0x%p, '%.*s')\n", dentry, dentry->d_name.len, dentry->d_name.name); @@ -969,30 +972,37 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr) if ((attr->ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) || (attr->ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) { - credits = OCFS2_INODE_UPDATE_CREDITS; + /* + * Gather pointers to quota structures so that allocation / + * freeing of quota structures happens here and not inside + * vfs_dq_transfer() where we have problems with lock ordering + */ if (attr->ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid && OCFS2_HAS_RO_COMPAT_FEATURE(sb, OCFS2_FEATURE_RO_COMPAT_USRQUOTA)) { - oinfo = sb_dqinfo(sb, USRQUOTA)->dqi_priv; - status = ocfs2_lock_global_qf(oinfo, 1); - if (status < 0) + transfer_to[USRQUOTA] = dqget(sb, attr->ia_uid, + USRQUOTA); + transfer_from[USRQUOTA] = dqget(sb, inode->i_uid, + USRQUOTA); + if (!transfer_to[USRQUOTA] || !transfer_from[USRQUOTA]) { + status = -ESRCH; goto bail_unlock; - credits += ocfs2_calc_qinit_credits(sb, USRQUOTA) + - ocfs2_calc_qdel_credits(sb, USRQUOTA); - locked[USRQUOTA] = 1; + } } if (attr->ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid && OCFS2_HAS_RO_COMPAT_FEATURE(sb, OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)) { - oinfo = sb_dqinfo(sb, GRPQUOTA)->dqi_priv; - status = ocfs2_lock_global_qf(oinfo, 1); - if (status < 0) + transfer_to[GRPQUOTA] = dqget(sb, attr->ia_gid, + GRPQUOTA); + transfer_from[GRPQUOTA] = dqget(sb, inode->i_gid, + GRPQUOTA); + if (!transfer_to[GRPQUOTA] || !transfer_from[GRPQUOTA]) { + status = -ESRCH; goto bail_unlock; - credits += ocfs2_calc_qinit_credits(sb, GRPQUOTA) + - ocfs2_calc_qdel_credits(sb, GRPQUOTA); - locked[GRPQUOTA] = 1; + } } - handle = ocfs2_start_trans(osb, credits); + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS + + 2 * ocfs2_quota_trans_credits(sb)); if (IS_ERR(handle)) { status = PTR_ERR(handle); mlog_errno(status); @@ -1030,12 +1040,6 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr) bail_commit: ocfs2_commit_trans(osb, handle); bail_unlock: - for (qtype = 0; qtype < MAXQUOTAS; qtype++) { - if (!locked[qtype]) - continue; - oinfo = sb_dqinfo(sb, qtype)->dqi_priv; - ocfs2_unlock_global_qf(oinfo, 1); - } ocfs2_inode_unlock(inode, 1); bail_unlock_rw: if (size_change) @@ -1043,6 +1047,12 @@ bail_unlock_rw: bail: brelse(bh); + /* Release quota pointers in case we acquired them */ + for (qtype = 0; qtype < MAXQUOTAS; qtype++) { + dqput(transfer_to[qtype]); + dqput(transfer_from[qtype]); + } + if (!status && attr->ia_valid & ATTR_MODE) { status = ocfs2_acl_chmod(inode); if (status < 0) @@ -2016,7 +2026,7 @@ static ssize_t ocfs2_file_splice_read(struct file *in, size_t len, unsigned int flags) { - int ret = 0; + int ret = 0, lock_level = 0; struct inode *inode = in->f_path.dentry->d_inode; mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", in, pipe, @@ -2027,12 +2037,12 @@ static ssize_t ocfs2_file_splice_read(struct file *in, /* * See the comment in ocfs2_file_aio_read() */ - ret = ocfs2_inode_lock(inode, NULL, 0); + ret = ocfs2_inode_lock_atime(inode, in->f_vfsmnt, &lock_level); if (ret < 0) { mlog_errno(ret); goto bail; } - ocfs2_inode_unlock(inode, 0); + ocfs2_inode_unlock(inode, lock_level); ret = generic_file_splice_read(in, ppos, pipe, len, flags); diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c index 10e1fa87396..4dc8890ba31 100644 --- a/fs/ocfs2/inode.c +++ b/fs/ocfs2/inode.c @@ -215,6 +215,8 @@ bail: static int ocfs2_init_locked_inode(struct inode *inode, void *opaque) { struct ocfs2_find_inode_args *args = opaque; + static struct lock_class_key ocfs2_quota_ip_alloc_sem_key, + ocfs2_file_ip_alloc_sem_key; mlog_entry("inode = %p, opaque = %p\n", inode, opaque); @@ -223,6 +225,15 @@ static int ocfs2_init_locked_inode(struct inode *inode, void *opaque) if (args->fi_sysfile_type != 0) lockdep_set_class(&inode->i_mutex, &ocfs2_sysfile_lock_key[args->fi_sysfile_type]); + if (args->fi_sysfile_type == USER_QUOTA_SYSTEM_INODE || + args->fi_sysfile_type == GROUP_QUOTA_SYSTEM_INODE || + args->fi_sysfile_type == LOCAL_USER_QUOTA_SYSTEM_INODE || + args->fi_sysfile_type == LOCAL_GROUP_QUOTA_SYSTEM_INODE) + lockdep_set_class(&OCFS2_I(inode)->ip_alloc_sem, + &ocfs2_quota_ip_alloc_sem_key); + else + lockdep_set_class(&OCFS2_I(inode)->ip_alloc_sem, + &ocfs2_file_ip_alloc_sem_key); mlog_exit(0); return 0; diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c index a20a0f1e37f..f033760ecbe 100644 --- a/fs/ocfs2/journal.c +++ b/fs/ocfs2/journal.c @@ -28,6 +28,8 @@ #include <linux/slab.h> #include <linux/highmem.h> #include <linux/kthread.h> +#include <linux/time.h> +#include <linux/random.h> #define MLOG_MASK_PREFIX ML_JOURNAL #include <cluster/masklog.h> @@ -52,6 +54,8 @@ DEFINE_SPINLOCK(trans_inc_lock); +#define ORPHAN_SCAN_SCHEDULE_TIMEOUT 300000 + static int ocfs2_force_read_journal(struct inode *inode); static int ocfs2_recover_node(struct ocfs2_super *osb, int node_num, int slot_num); @@ -1841,6 +1845,128 @@ bail: return status; } +/* + * Scan timer should get fired every ORPHAN_SCAN_SCHEDULE_TIMEOUT. Add some + * randomness to the timeout to minimize multple nodes firing the timer at the + * same time. + */ +static inline unsigned long ocfs2_orphan_scan_timeout(void) +{ + unsigned long time; + + get_random_bytes(&time, sizeof(time)); + time = ORPHAN_SCAN_SCHEDULE_TIMEOUT + (time % 5000); + return msecs_to_jiffies(time); +} + +/* + * ocfs2_queue_orphan_scan calls ocfs2_queue_recovery_completion for + * every slot, queuing a recovery of the slot on the ocfs2_wq thread. This + * is done to catch any orphans that are left over in orphan directories. + * + * ocfs2_queue_orphan_scan gets called every ORPHAN_SCAN_SCHEDULE_TIMEOUT + * seconds. It gets an EX lock on os_lockres and checks sequence number + * stored in LVB. If the sequence number has changed, it means some other + * node has done the scan. This node skips the scan and tracks the + * sequence number. If the sequence number didn't change, it means a scan + * hasn't happened. The node queues a scan and increments the + * sequence number in the LVB. + */ +void ocfs2_queue_orphan_scan(struct ocfs2_super *osb) +{ + struct ocfs2_orphan_scan *os; + int status, i; + u32 seqno = 0; + + os = &osb->osb_orphan_scan; + + if (atomic_read(&os->os_state) == ORPHAN_SCAN_INACTIVE) + goto out; + + status = ocfs2_orphan_scan_lock(osb, &seqno); + if (status < 0) { + if (status != -EAGAIN) + mlog_errno(status); + goto out; + } + + /* Do no queue the tasks if the volume is being umounted */ + if (atomic_read(&os->os_state) == ORPHAN_SCAN_INACTIVE) + goto unlock; + + if (os->os_seqno != seqno) { + os->os_seqno = seqno; + goto unlock; + } + + for (i = 0; i < osb->max_slots; i++) + ocfs2_queue_recovery_completion(osb->journal, i, NULL, NULL, + NULL); + /* + * We queued a recovery on orphan slots, increment the sequence + * number and update LVB so other node will skip the scan for a while + */ + seqno++; + os->os_count++; + os->os_scantime = CURRENT_TIME; +unlock: + ocfs2_orphan_scan_unlock(osb, seqno); +out: + return; +} + +/* Worker task that gets fired every ORPHAN_SCAN_SCHEDULE_TIMEOUT millsec */ +void ocfs2_orphan_scan_work(struct work_struct *work) +{ + struct ocfs2_orphan_scan *os; + struct ocfs2_super *osb; + + os = container_of(work, struct ocfs2_orphan_scan, + os_orphan_scan_work.work); + osb = os->os_osb; + + mutex_lock(&os->os_lock); + ocfs2_queue_orphan_scan(osb); + if (atomic_read(&os->os_state) == ORPHAN_SCAN_ACTIVE) + schedule_delayed_work(&os->os_orphan_scan_work, + ocfs2_orphan_scan_timeout()); + mutex_unlock(&os->os_lock); +} + +void ocfs2_orphan_scan_stop(struct ocfs2_super *osb) +{ + struct ocfs2_orphan_scan *os; + + os = &osb->osb_orphan_scan; + if (atomic_read(&os->os_state) == ORPHAN_SCAN_ACTIVE) { + atomic_set(&os->os_state, ORPHAN_SCAN_INACTIVE); + mutex_lock(&os->os_lock); + cancel_delayed_work(&os->os_orphan_scan_work); + mutex_unlock(&os->os_lock); + } +} + +void ocfs2_orphan_scan_init(struct ocfs2_super *osb) +{ + struct ocfs2_orphan_scan *os; + + os = &osb->osb_orphan_scan; + os->os_osb = osb; + os->os_count = 0; + os->os_seqno = 0; + os->os_scantime = CURRENT_TIME; + mutex_init(&os->os_lock); + INIT_DELAYED_WORK(&os->os_orphan_scan_work, ocfs2_orphan_scan_work); + + if (ocfs2_is_hard_readonly(osb) || ocfs2_mount_local(osb)) + atomic_set(&os->os_state, ORPHAN_SCAN_INACTIVE); + else { + atomic_set(&os->os_state, ORPHAN_SCAN_ACTIVE); + schedule_delayed_work(&os->os_orphan_scan_work, + ocfs2_orphan_scan_timeout()); + } +} + struct ocfs2_orphan_filldir_priv { struct inode *head; struct ocfs2_super *osb; diff --git a/fs/ocfs2/journal.h b/fs/ocfs2/journal.h index eb7b76331eb..5432c7f79cc 100644 --- a/fs/ocfs2/journal.h +++ b/fs/ocfs2/journal.h @@ -144,6 +144,10 @@ static inline void ocfs2_inode_set_new(struct ocfs2_super *osb, } /* Exported only for the journal struct init code in super.c. Do not call. */ +void ocfs2_orphan_scan_init(struct ocfs2_super *osb); +void ocfs2_orphan_scan_stop(struct ocfs2_super *osb); +void ocfs2_orphan_scan_exit(struct ocfs2_super *osb); + void ocfs2_complete_recovery(struct work_struct *work); void ocfs2_wait_for_recovery(struct ocfs2_super *osb); diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c index 33464c6b60a..8601f934010 100644 --- a/fs/ocfs2/namei.c +++ b/fs/ocfs2/namei.c @@ -118,7 +118,7 @@ static struct dentry *ocfs2_lookup(struct inode *dir, struct dentry *dentry, mlog(0, "find name %.*s in directory %llu\n", dentry->d_name.len, dentry->d_name.name, (unsigned long long)OCFS2_I(dir)->ip_blkno); - status = ocfs2_inode_lock(dir, NULL, 0); + status = ocfs2_inode_lock_nested(dir, NULL, 0, OI_LS_PARENT); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -636,7 +636,7 @@ static int ocfs2_link(struct dentry *old_dentry, if (S_ISDIR(inode->i_mode)) return -EPERM; - err = ocfs2_inode_lock(dir, &parent_fe_bh, 1); + err = ocfs2_inode_lock_nested(dir, &parent_fe_bh, 1, OI_LS_PARENT); if (err < 0) { if (err != -ENOENT) mlog_errno(err); @@ -800,7 +800,8 @@ static int ocfs2_unlink(struct inode *dir, return -EPERM; } - status = ocfs2_inode_lock(dir, &parent_node_bh, 1); + status = ocfs2_inode_lock_nested(dir, &parent_node_bh, 1, + OI_LS_PARENT); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -978,7 +979,8 @@ static int ocfs2_double_lock(struct ocfs2_super *osb, inode1 = tmpinode; } /* lock id2 */ - status = ocfs2_inode_lock(inode2, bh2, 1); + status = ocfs2_inode_lock_nested(inode2, bh2, 1, + OI_LS_RENAME1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -987,7 +989,7 @@ static int ocfs2_double_lock(struct ocfs2_super *osb, } /* lock id1 */ - status = ocfs2_inode_lock(inode1, bh1, 1); + status = ocfs2_inode_lock_nested(inode1, bh1, 1, OI_LS_RENAME2); if (status < 0) { /* * An error return must mean that no cluster locks @@ -1103,7 +1105,8 @@ static int ocfs2_rename(struct inode *old_dir, * won't have to concurrently downconvert the inode and the * dentry locks. */ - status = ocfs2_inode_lock(old_inode, &old_inode_bh, 1); + status = ocfs2_inode_lock_nested(old_inode, &old_inode_bh, 1, + OI_LS_PARENT); if (status < 0) { if (status != -ENOENT) mlog_errno(status); diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h index 1386281950d..c9345ebb849 100644 --- a/fs/ocfs2/ocfs2.h +++ b/fs/ocfs2/ocfs2.h @@ -34,6 +34,7 @@ #include <linux/workqueue.h> #include <linux/kref.h> #include <linux/mutex.h> +#include <linux/lockdep.h> #ifndef CONFIG_OCFS2_COMPAT_JBD # include <linux/jbd2.h> #else @@ -47,6 +48,9 @@ #include "ocfs2_fs.h" #include "ocfs2_lockid.h" +/* For struct ocfs2_blockcheck_stats */ +#include "blockcheck.h" + /* Most user visible OCFS2 inodes will have very few pieces of * metadata, but larger files (including bitmaps, etc) must be taken * into account when designing an access scheme. We allow a small @@ -149,6 +153,25 @@ struct ocfs2_lock_res { unsigned int l_lock_max_exmode; /* Max wait for EX */ unsigned int l_lock_refresh; /* Disk refreshes */ #endif +#ifdef CONFIG_DEBUG_LOCK_ALLOC + struct lockdep_map l_lockdep_map; +#endif +}; + +enum ocfs2_orphan_scan_state { + ORPHAN_SCAN_ACTIVE, + ORPHAN_SCAN_INACTIVE +}; + +struct ocfs2_orphan_scan { + struct mutex os_lock; + struct ocfs2_super *os_osb; + struct ocfs2_lock_res os_lockres; /* lock to synchronize scans */ + struct delayed_work os_orphan_scan_work; + struct timespec os_scantime; /* time this node ran the scan */ + u32 os_count; /* tracks node specific scans */ + u32 os_seqno; /* tracks cluster wide scans */ + atomic_t os_state; /* ACTIVE or INACTIVE */ }; struct ocfs2_dlm_debug { @@ -295,6 +318,7 @@ struct ocfs2_super struct ocfs2_dinode *local_alloc_copy; struct ocfs2_quota_recovery *quota_rec; + struct ocfs2_blockcheck_stats osb_ecc_stats; struct ocfs2_alloc_stats alloc_stats; char dev_str[20]; /* "major,minor" of the device */ @@ -341,6 +365,8 @@ struct ocfs2_super unsigned int *osb_orphan_wipes; wait_queue_head_t osb_wipe_event; + struct ocfs2_orphan_scan osb_orphan_scan; + /* used to protect metaecc calculation check of xattr. */ spinlock_t osb_xattr_lock; diff --git a/fs/ocfs2/ocfs2_lockid.h b/fs/ocfs2/ocfs2_lockid.h index a53ce87481b..fcdba091af3 100644 --- a/fs/ocfs2/ocfs2_lockid.h +++ b/fs/ocfs2/ocfs2_lockid.h @@ -48,6 +48,7 @@ enum ocfs2_lock_type { OCFS2_LOCK_TYPE_FLOCK, OCFS2_LOCK_TYPE_QINFO, OCFS2_LOCK_TYPE_NFS_SYNC, + OCFS2_LOCK_TYPE_ORPHAN_SCAN, OCFS2_NUM_LOCK_TYPES }; @@ -85,6 +86,9 @@ static inline char ocfs2_lock_type_char(enum ocfs2_lock_type type) case OCFS2_LOCK_TYPE_NFS_SYNC: c = 'Y'; break; + case OCFS2_LOCK_TYPE_ORPHAN_SCAN: + c = 'P'; + break; default: c = '\0'; } @@ -104,6 +108,7 @@ static char *ocfs2_lock_type_strings[] = { [OCFS2_LOCK_TYPE_OPEN] = "Open", [OCFS2_LOCK_TYPE_FLOCK] = "Flock", [OCFS2_LOCK_TYPE_QINFO] = "Quota", + [OCFS2_LOCK_TYPE_ORPHAN_SCAN] = "OrphanScan", }; static inline const char *ocfs2_lock_type_string(enum ocfs2_lock_type type) diff --git a/fs/ocfs2/quota_global.c b/fs/ocfs2/quota_global.c index 1ed0f7c8686..edfa60cd155 100644 --- a/fs/ocfs2/quota_global.c +++ b/fs/ocfs2/quota_global.c @@ -421,6 +421,7 @@ int ocfs2_global_read_dquot(struct dquot *dquot) OCFS2_DQUOT(dquot)->dq_originodes = dquot->dq_dqb.dqb_curinodes; if (!dquot->dq_off) { /* No real quota entry? */ /* Upgrade to exclusive lock for allocation */ + ocfs2_qinfo_unlock(info, 0); err = ocfs2_qinfo_lock(info, 1); if (err < 0) goto out_qlock; @@ -435,7 +436,8 @@ int ocfs2_global_read_dquot(struct dquot *dquot) out_qlock: if (ex) ocfs2_qinfo_unlock(info, 1); - ocfs2_qinfo_unlock(info, 0); + else + ocfs2_qinfo_unlock(info, 0); out: if (err < 0) mlog_errno(err); diff --git a/fs/ocfs2/quota_local.c b/fs/ocfs2/quota_local.c index 07deec5e972..5a460fa8255 100644 --- a/fs/ocfs2/quota_local.c +++ b/fs/ocfs2/quota_local.c @@ -444,10 +444,6 @@ static int ocfs2_recover_local_quota_file(struct inode *lqinode, mlog_entry("ino=%lu type=%u", (unsigned long)lqinode->i_ino, type); - status = ocfs2_lock_global_qf(oinfo, 1); - if (status < 0) - goto out; - list_for_each_entry_safe(rchunk, next, &(rec->r_list[type]), rc_list) { chunk = rchunk->rc_chunk; hbh = NULL; @@ -480,12 +476,18 @@ static int ocfs2_recover_local_quota_file(struct inode *lqinode, type); goto out_put_bh; } + status = ocfs2_lock_global_qf(oinfo, 1); + if (status < 0) { + mlog_errno(status); + goto out_put_dquot; + } + handle = ocfs2_start_trans(OCFS2_SB(sb), OCFS2_QSYNC_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); mlog_errno(status); - goto out_put_dquot; + goto out_drop_lock; } mutex_lock(&sb_dqopt(sb)->dqio_mutex); spin_lock(&dq_data_lock); @@ -523,6 +525,8 @@ static int ocfs2_recover_local_quota_file(struct inode *lqinode, out_commit: mutex_unlock(&sb_dqopt(sb)->dqio_mutex); ocfs2_commit_trans(OCFS2_SB(sb), handle); +out_drop_lock: + ocfs2_unlock_global_qf(oinfo, 1); out_put_dquot: dqput(dquot); out_put_bh: @@ -537,8 +541,6 @@ out_put_bh: if (status < 0) break; } - ocfs2_unlock_global_qf(oinfo, 1); -out: if (status < 0) free_recovery_list(&(rec->r_list[type])); mlog_exit(status); @@ -655,6 +657,9 @@ static int ocfs2_local_read_info(struct super_block *sb, int type) struct ocfs2_quota_recovery *rec; int locked = 0; + /* We don't need the lock and we have to acquire quota file locks + * which will later depend on this lock */ + mutex_unlock(&sb_dqopt(sb)->dqio_mutex); info->dqi_maxblimit = 0x7fffffffffffffffLL; info->dqi_maxilimit = 0x7fffffffffffffffLL; oinfo = kmalloc(sizeof(struct ocfs2_mem_dqinfo), GFP_NOFS); @@ -733,6 +738,7 @@ static int ocfs2_local_read_info(struct super_block *sb, int type) goto out_err; } + mutex_lock(&sb_dqopt(sb)->dqio_mutex); return 0; out_err: if (oinfo) { @@ -746,6 +752,7 @@ out_err: kfree(oinfo); } brelse(bh); + mutex_lock(&sb_dqopt(sb)->dqio_mutex); return -1; } diff --git a/fs/ocfs2/stack_o2cb.c b/fs/ocfs2/stack_o2cb.c index fcd120f1493..3f661376a2d 100644 --- a/fs/ocfs2/stack_o2cb.c +++ b/fs/ocfs2/stack_o2cb.c @@ -236,6 +236,16 @@ static int o2cb_dlm_lock_status(union ocfs2_dlm_lksb *lksb) return dlm_status_to_errno(lksb->lksb_o2dlm.status); } +/* + * o2dlm aways has a "valid" LVB. If the dlm loses track of the LVB + * contents, it will zero out the LVB. Thus the caller can always trust + * the contents. + */ +static int o2cb_dlm_lvb_valid(union ocfs2_dlm_lksb *lksb) +{ + return 1; +} + static void *o2cb_dlm_lvb(union ocfs2_dlm_lksb *lksb) { return (void *)(lksb->lksb_o2dlm.lvb); @@ -354,6 +364,7 @@ static struct ocfs2_stack_operations o2cb_stack_ops = { .dlm_lock = o2cb_dlm_lock, .dlm_unlock = o2cb_dlm_unlock, .lock_status = o2cb_dlm_lock_status, + .lvb_valid = o2cb_dlm_lvb_valid, .lock_lvb = o2cb_dlm_lvb, .dump_lksb = o2cb_dump_lksb, }; diff --git a/fs/ocfs2/stack_user.c b/fs/ocfs2/stack_user.c index 9b76d41a8ac..ff4c798a563 100644 --- a/fs/ocfs2/stack_user.c +++ b/fs/ocfs2/stack_user.c @@ -738,6 +738,13 @@ static int user_dlm_lock_status(union ocfs2_dlm_lksb *lksb) return lksb->lksb_fsdlm.sb_status; } +static int user_dlm_lvb_valid(union ocfs2_dlm_lksb *lksb) +{ + int invalid = lksb->lksb_fsdlm.sb_flags & DLM_SBF_VALNOTVALID; + + return !invalid; +} + static void *user_dlm_lvb(union ocfs2_dlm_lksb *lksb) { if (!lksb->lksb_fsdlm.sb_lvbptr) @@ -873,6 +880,7 @@ static struct ocfs2_stack_operations ocfs2_user_plugin_ops = { .dlm_lock = user_dlm_lock, .dlm_unlock = user_dlm_unlock, .lock_status = user_dlm_lock_status, + .lvb_valid = user_dlm_lvb_valid, .lock_lvb = user_dlm_lvb, .plock = user_plock, .dump_lksb = user_dlm_dump_lksb, diff --git a/fs/ocfs2/stackglue.c b/fs/ocfs2/stackglue.c index 68b668b0e60..3f2f1c45b7b 100644 --- a/fs/ocfs2/stackglue.c +++ b/fs/ocfs2/stackglue.c @@ -6,7 +6,7 @@ * Code which implements an OCFS2 specific interface to underlying * cluster stacks. * - * Copyright (C) 2007 Oracle. All rights reserved. + * Copyright (C) 2007, 2009 Oracle. All rights reserved. * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public @@ -271,11 +271,12 @@ int ocfs2_dlm_lock_status(union ocfs2_dlm_lksb *lksb) } EXPORT_SYMBOL_GPL(ocfs2_dlm_lock_status); -/* - * Why don't we cast to ocfs2_meta_lvb? The "clean" answer is that we - * don't cast at the glue level. The real answer is that the header - * ordering is nigh impossible. - */ +int ocfs2_dlm_lvb_valid(union ocfs2_dlm_lksb *lksb) +{ + return active_stack->sp_ops->lvb_valid(lksb); +} +EXPORT_SYMBOL_GPL(ocfs2_dlm_lvb_valid); + void *ocfs2_dlm_lvb(union ocfs2_dlm_lksb *lksb) { return active_stack->sp_ops->lock_lvb(lksb); diff --git a/fs/ocfs2/stackglue.h b/fs/ocfs2/stackglue.h index c571af375ef..03a44d60eac 100644 --- a/fs/ocfs2/stackglue.h +++ b/fs/ocfs2/stackglue.h @@ -186,6 +186,11 @@ struct ocfs2_stack_operations { int (*lock_status)(union ocfs2_dlm_lksb *lksb); /* + * Return non-zero if the LVB is valid. + */ + int (*lvb_valid)(union ocfs2_dlm_lksb *lksb); + + /* * Pull the lvb pointer off of the stack-specific lksb. */ void *(*lock_lvb)(union ocfs2_dlm_lksb *lksb); @@ -252,6 +257,7 @@ int ocfs2_dlm_unlock(struct ocfs2_cluster_connection *conn, struct ocfs2_lock_res *astarg); int ocfs2_dlm_lock_status(union ocfs2_dlm_lksb *lksb); +int ocfs2_dlm_lvb_valid(union ocfs2_dlm_lksb *lksb); void *ocfs2_dlm_lvb(union ocfs2_dlm_lksb *lksb); void ocfs2_dlm_dump_lksb(union ocfs2_dlm_lksb *lksb); diff --git a/fs/ocfs2/suballoc.c b/fs/ocfs2/suballoc.c index 8439f6b324b..73a16d4666d 100644 --- a/fs/ocfs2/suballoc.c +++ b/fs/ocfs2/suballoc.c @@ -923,14 +923,23 @@ static int ocfs2_test_bg_bit_allocatable(struct buffer_head *bg_bh, int nr) { struct ocfs2_group_desc *bg = (struct ocfs2_group_desc *) bg_bh->b_data; + int ret; if (ocfs2_test_bit(nr, (unsigned long *)bg->bg_bitmap)) return 0; - if (!buffer_jbd(bg_bh) || !bh2jh(bg_bh)->b_committed_data) + + if (!buffer_jbd(bg_bh)) return 1; + jbd_lock_bh_state(bg_bh); bg = (struct ocfs2_group_desc *) bh2jh(bg_bh)->b_committed_data; - return !ocfs2_test_bit(nr, (unsigned long *)bg->bg_bitmap); + if (bg) + ret = !ocfs2_test_bit(nr, (unsigned long *)bg->bg_bitmap); + else + ret = 1; + jbd_unlock_bh_state(bg_bh); + + return ret; } static int ocfs2_block_group_find_clear_bits(struct ocfs2_super *osb, @@ -1885,6 +1894,7 @@ static inline int ocfs2_block_group_clear_bits(handle_t *handle, unsigned int tmp; int journal_type = OCFS2_JOURNAL_ACCESS_WRITE; struct ocfs2_group_desc *undo_bg = NULL; + int cluster_bitmap = 0; mlog_entry_void(); @@ -1905,18 +1915,28 @@ static inline int ocfs2_block_group_clear_bits(handle_t *handle, } if (ocfs2_is_cluster_bitmap(alloc_inode)) - undo_bg = (struct ocfs2_group_desc *) bh2jh(group_bh)->b_committed_data; + cluster_bitmap = 1; + + if (cluster_bitmap) { + jbd_lock_bh_state(group_bh); + undo_bg = (struct ocfs2_group_desc *) + bh2jh(group_bh)->b_committed_data; + BUG_ON(!undo_bg); + } tmp = num_bits; while(tmp--) { ocfs2_clear_bit((bit_off + tmp), (unsigned long *) bg->bg_bitmap); - if (ocfs2_is_cluster_bitmap(alloc_inode)) + if (cluster_bitmap) ocfs2_set_bit(bit_off + tmp, (unsigned long *) undo_bg->bg_bitmap); } le16_add_cpu(&bg->bg_free_bits_count, num_bits); + if (cluster_bitmap) + jbd_unlock_bh_state(group_bh); + status = ocfs2_journal_dirty(handle, group_bh); if (status < 0) mlog_errno(status); diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c index 201b40a441f..7efb349fb9b 100644 --- a/fs/ocfs2/super.c +++ b/fs/ocfs2/super.c @@ -119,10 +119,12 @@ static void ocfs2_release_system_inodes(struct ocfs2_super *osb); static int ocfs2_check_volume(struct ocfs2_super *osb); static int ocfs2_verify_volume(struct ocfs2_dinode *di, struct buffer_head *bh, - u32 sectsize); + u32 sectsize, + struct ocfs2_blockcheck_stats *stats); static int ocfs2_initialize_super(struct super_block *sb, struct buffer_head *bh, - int sector_size); + int sector_size, + struct ocfs2_blockcheck_stats *stats); static int ocfs2_get_sector(struct super_block *sb, struct buffer_head **bh, int block, @@ -203,10 +205,10 @@ static const match_table_t tokens = { #ifdef CONFIG_DEBUG_FS static int ocfs2_osb_dump(struct ocfs2_super *osb, char *buf, int len) { - int out = 0; - int i; struct ocfs2_cluster_connection *cconn = osb->cconn; struct ocfs2_recovery_map *rm = osb->recovery_map; + struct ocfs2_orphan_scan *os = &osb->osb_orphan_scan; + int i, out = 0; out += snprintf(buf + out, len - out, "%10s => Id: %-s Uuid: %-s Gen: 0x%X Label: %-s\n", @@ -231,20 +233,24 @@ static int ocfs2_osb_dump(struct ocfs2_super *osb, char *buf, int len) "%10s => Opts: 0x%lX AtimeQuanta: %u\n", "Mount", osb->s_mount_opt, osb->s_atime_quantum); - out += snprintf(buf + out, len - out, - "%10s => Stack: %s Name: %*s Version: %d.%d\n", - "Cluster", - (*osb->osb_cluster_stack == '\0' ? - "o2cb" : osb->osb_cluster_stack), - cconn->cc_namelen, cconn->cc_name, - cconn->cc_version.pv_major, cconn->cc_version.pv_minor); + if (cconn) { + out += snprintf(buf + out, len - out, + "%10s => Stack: %s Name: %*s " + "Version: %d.%d\n", "Cluster", + (*osb->osb_cluster_stack == '\0' ? + "o2cb" : osb->osb_cluster_stack), + cconn->cc_namelen, cconn->cc_name, + cconn->cc_version.pv_major, + cconn->cc_version.pv_minor); + } spin_lock(&osb->dc_task_lock); out += snprintf(buf + out, len - out, "%10s => Pid: %d Count: %lu WakeSeq: %lu " "WorkSeq: %lu\n", "DownCnvt", - task_pid_nr(osb->dc_task), osb->blocked_lock_count, - osb->dc_wake_sequence, osb->dc_work_sequence); + (osb->dc_task ? task_pid_nr(osb->dc_task) : -1), + osb->blocked_lock_count, osb->dc_wake_sequence, + osb->dc_work_sequence); spin_unlock(&osb->dc_task_lock); spin_lock(&osb->osb_lock); @@ -264,14 +270,15 @@ static int ocfs2_osb_dump(struct ocfs2_super *osb, char *buf, int len) out += snprintf(buf + out, len - out, "%10s => Pid: %d Interval: %lu Needs: %d\n", "Commit", - task_pid_nr(osb->commit_task), osb->osb_commit_interval, + (osb->commit_task ? task_pid_nr(osb->commit_task) : -1), + osb->osb_commit_interval, atomic_read(&osb->needs_checkpoint)); out += snprintf(buf + out, len - out, - "%10s => State: %d NumTxns: %d TxnId: %lu\n", + "%10s => State: %d TxnId: %lu NumTxns: %d\n", "Journal", osb->journal->j_state, - atomic_read(&osb->journal->j_num_trans), - osb->journal->j_trans_id); + osb->journal->j_trans_id, + atomic_read(&osb->journal->j_num_trans)); out += snprintf(buf + out, len - out, "%10s => GlobalAllocs: %d LocalAllocs: %d " @@ -297,9 +304,18 @@ static int ocfs2_osb_dump(struct ocfs2_super *osb, char *buf, int len) atomic_read(&osb->s_num_inodes_stolen)); spin_unlock(&osb->osb_lock); + out += snprintf(buf + out, len - out, "OrphanScan => "); + out += snprintf(buf + out, len - out, "Local: %u Global: %u ", + os->os_count, os->os_seqno); + out += snprintf(buf + out, len - out, " Last Scan: "); + if (atomic_read(&os->os_state) == ORPHAN_SCAN_INACTIVE) + out += snprintf(buf + out, len - out, "Disabled\n"); + else + out += snprintf(buf + out, len - out, "%lu seconds ago\n", + (get_seconds() - os->os_scantime.tv_sec)); + out += snprintf(buf + out, len - out, "%10s => %3s %10s\n", "Slots", "Num", "RecoGen"); - for (i = 0; i < osb->max_slots; ++i) { out += snprintf(buf + out, len - out, "%10s %c %3d %10d\n", @@ -542,7 +558,7 @@ static unsigned long long ocfs2_max_file_offset(unsigned int bbits, */ #if BITS_PER_LONG == 32 -# if defined(CONFIG_LBD) +# if defined(CONFIG_LBDAF) BUILD_BUG_ON(sizeof(sector_t) != 8); /* * We might be limited by page cache size. @@ -693,7 +709,8 @@ out: static int ocfs2_sb_probe(struct super_block *sb, struct buffer_head **bh, - int *sector_size) + int *sector_size, + struct ocfs2_blockcheck_stats *stats) { int status, tmpstat; struct ocfs1_vol_disk_hdr *hdr; @@ -759,7 +776,8 @@ static int ocfs2_sb_probe(struct super_block *sb, goto bail; } di = (struct ocfs2_dinode *) (*bh)->b_data; - status = ocfs2_verify_volume(di, *bh, blksize); + memset(stats, 0, sizeof(struct ocfs2_blockcheck_stats)); + status = ocfs2_verify_volume(di, *bh, blksize, stats); if (status >= 0) goto bail; brelse(*bh); @@ -965,6 +983,7 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent) struct ocfs2_super *osb = NULL; struct buffer_head *bh = NULL; char nodestr[8]; + struct ocfs2_blockcheck_stats stats; mlog_entry("%p, %p, %i", sb, data, silent); @@ -974,13 +993,13 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent) } /* probe for superblock */ - status = ocfs2_sb_probe(sb, &bh, §or_size); + status = ocfs2_sb_probe(sb, &bh, §or_size, &stats); if (status < 0) { mlog(ML_ERROR, "superblock probe failed!\n"); goto read_super_error; } - status = ocfs2_initialize_super(sb, bh, sector_size); + status = ocfs2_initialize_super(sb, bh, sector_size, &stats); osb = OCFS2_SB(sb); if (status < 0) { mlog_errno(status); @@ -1090,6 +1109,18 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent) goto read_super_error; } + if (ocfs2_meta_ecc(osb)) { + status = ocfs2_blockcheck_stats_debugfs_install( + &osb->osb_ecc_stats, + osb->osb_debug_root); + if (status) { + mlog(ML_ERROR, + "Unable to create blockcheck statistics " + "files\n"); + goto read_super_error; + } + } + status = ocfs2_mount_volume(sb); if (osb->root_inode) inode = igrab(osb->root_inode); @@ -1150,6 +1181,9 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent) atomic_set(&osb->vol_state, VOLUME_MOUNTED_QUOTAS); wake_up(&osb->osb_mount_event); + /* Start this when the mount is almost sure of being successful */ + ocfs2_orphan_scan_init(osb); + mlog_exit(status); return status; @@ -1760,13 +1794,8 @@ static int ocfs2_mount_volume(struct super_block *sb) } status = ocfs2_truncate_log_init(osb); - if (status < 0) { + if (status < 0) mlog_errno(status); - goto leave; - } - - if (ocfs2_mount_local(osb)) - goto leave; leave: if (unlock_super) @@ -1790,6 +1819,9 @@ static void ocfs2_dismount_volume(struct super_block *sb, int mnt_err) debugfs_remove(osb->osb_ctxt); + /* Orphan scan should be stopped as early as possible */ + ocfs2_orphan_scan_stop(osb); + ocfs2_disable_quotas(osb); ocfs2_shutdown_local_alloc(osb); @@ -1833,6 +1865,7 @@ static void ocfs2_dismount_volume(struct super_block *sb, int mnt_err) if (osb->cconn) ocfs2_dlm_shutdown(osb, hangup_needed); + ocfs2_blockcheck_stats_debugfs_remove(&osb->osb_ecc_stats); debugfs_remove(osb->osb_debug_root); if (hangup_needed) @@ -1880,7 +1913,8 @@ static int ocfs2_setup_osb_uuid(struct ocfs2_super *osb, const unsigned char *uu static int ocfs2_initialize_super(struct super_block *sb, struct buffer_head *bh, - int sector_size) + int sector_size, + struct ocfs2_blockcheck_stats *stats) { int status; int i, cbits, bbits; @@ -1939,6 +1973,9 @@ static int ocfs2_initialize_super(struct super_block *sb, atomic_set(&osb->alloc_stats.bg_allocs, 0); atomic_set(&osb->alloc_stats.bg_extends, 0); + /* Copy the blockcheck stats from the superblock probe */ + osb->osb_ecc_stats = *stats; + ocfs2_init_node_maps(osb); snprintf(osb->dev_str, sizeof(osb->dev_str), "%u,%u", @@ -2169,7 +2206,8 @@ bail: */ static int ocfs2_verify_volume(struct ocfs2_dinode *di, struct buffer_head *bh, - u32 blksz) + u32 blksz, + struct ocfs2_blockcheck_stats *stats) { int status = -EAGAIN; @@ -2182,7 +2220,8 @@ static int ocfs2_verify_volume(struct ocfs2_dinode *di, OCFS2_FEATURE_INCOMPAT_META_ECC) { status = ocfs2_block_check_validate(bh->b_data, bh->b_size, - &di->i_check); + &di->i_check, + stats); if (status) goto out; } diff --git a/fs/ocfs2/sysfile.c b/fs/ocfs2/sysfile.c index ab713ebdd54..40e53702948 100644 --- a/fs/ocfs2/sysfile.c +++ b/fs/ocfs2/sysfile.c @@ -50,6 +50,10 @@ static inline int is_in_system_inode_array(struct ocfs2_super *osb, int type, u32 slot); +#ifdef CONFIG_DEBUG_LOCK_ALLOC +static struct lock_class_key ocfs2_sysfile_cluster_lock_key[NUM_SYSTEM_INODES]; +#endif + static inline int is_global_system_inode(int type) { return type >= OCFS2_FIRST_ONLINE_SYSTEM_INODE && @@ -118,6 +122,21 @@ static struct inode * _ocfs2_get_system_file_inode(struct ocfs2_super *osb, inode = NULL; goto bail; } +#ifdef CONFIG_DEBUG_LOCK_ALLOC + if (type == LOCAL_USER_QUOTA_SYSTEM_INODE || + type == LOCAL_GROUP_QUOTA_SYSTEM_INODE || + type == JOURNAL_SYSTEM_INODE) { + /* Ignore inode lock on these inodes as the lock does not + * really belong to any process and lockdep cannot handle + * that */ + OCFS2_I(inode)->ip_inode_lockres.l_lockdep_map.key = NULL; + } else { + lockdep_init_map(&OCFS2_I(inode)->ip_inode_lockres. + l_lockdep_map, + ocfs2_system_inodes[type].si_name, + &ocfs2_sysfile_cluster_lock_key[type], 0); + } +#endif bail: return inode; diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c index 15631019dc6..ba320e25074 100644 --- a/fs/ocfs2/xattr.c +++ b/fs/ocfs2/xattr.c @@ -3154,7 +3154,7 @@ static int ocfs2_iterate_xattr_buckets(struct inode *inode, le32_to_cpu(bucket_xh(bucket)->xh_entries[0].xe_name_hash)); if (func) { ret = func(inode, bucket, para); - if (ret) + if (ret && ret != -ERANGE) mlog_errno(ret); /* Fall through to bucket_relse() */ } @@ -3261,7 +3261,8 @@ static int ocfs2_xattr_tree_list_index_block(struct inode *inode, ocfs2_list_xattr_bucket, &xl); if (ret) { - mlog_errno(ret); + if (ret != -ERANGE) + mlog_errno(ret); goto out; } |