diff options
Diffstat (limited to 'fs/ocfs2/namei.c')
| -rw-r--r-- | fs/ocfs2/namei.c | 2624 |
1 files changed, 1512 insertions, 1112 deletions
diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c index f6b77ff1d2b..8add6f1030d 100644 --- a/fs/ocfs2/namei.c +++ b/fs/ocfs2/namei.c @@ -40,8 +40,8 @@ #include <linux/types.h> #include <linux/slab.h> #include <linux/highmem.h> +#include <linux/quotaops.h> -#define MLOG_MASK_PREFIX ML_NAMEI #include <cluster/masklog.h> #include "ocfs2.h" @@ -56,115 +56,66 @@ #include "journal.h" #include "namei.h" #include "suballoc.h" +#include "super.h" #include "symlink.h" #include "sysfile.h" #include "uptodate.h" -#include "vote.h" +#include "xattr.h" +#include "acl.h" +#include "ocfs2_trace.h" #include "buffer_head_io.h" -#define NAMEI_RA_CHUNKS 2 -#define NAMEI_RA_BLOCKS 4 -#define NAMEI_RA_SIZE (NAMEI_RA_CHUNKS * NAMEI_RA_BLOCKS) -#define NAMEI_RA_INDEX(c,b) (((c) * NAMEI_RA_BLOCKS) + (b)) - -static int inline ocfs2_search_dirblock(struct buffer_head *bh, - struct inode *dir, - const char *name, int namelen, - unsigned long offset, - struct ocfs2_dir_entry **res_dir); - -static int ocfs2_delete_entry(struct ocfs2_journal_handle *handle, - struct inode *dir, - struct ocfs2_dir_entry *de_del, - struct buffer_head *bh); - -static int __ocfs2_add_entry(struct ocfs2_journal_handle *handle, - struct inode *dir, - const char *name, int namelen, - struct inode *inode, u64 blkno, - struct buffer_head *parent_fe_bh, - struct buffer_head *insert_bh); - static int ocfs2_mknod_locked(struct ocfs2_super *osb, struct inode *dir, - struct dentry *dentry, int mode, + struct inode *inode, dev_t dev, struct buffer_head **new_fe_bh, struct buffer_head *parent_fe_bh, - struct ocfs2_journal_handle *handle, - struct inode **ret_inode, + handle_t *handle, struct ocfs2_alloc_context *inode_ac); -static int ocfs2_fill_new_dir(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - struct inode *parent, - struct inode *inode, - struct buffer_head *fe_bh, - struct ocfs2_alloc_context *data_ac); - -static int ocfs2_double_lock(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - struct buffer_head **bh1, - struct inode *inode1, - struct buffer_head **bh2, - struct inode *inode2); - static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - struct inode *inode, + struct inode **ret_orphan_dir, + u64 blkno, char *name, - struct buffer_head **de_bh); + struct ocfs2_dir_lookup_result *lookup); static int ocfs2_orphan_add(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, - struct ocfs2_dinode *fe, + struct buffer_head *fe_bh, char *name, - struct buffer_head *de_bh); + struct ocfs2_dir_lookup_result *lookup, + struct inode *orphan_dir_inode); static int ocfs2_create_symlink_data(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, const char *symname); -static inline int ocfs2_add_entry(struct ocfs2_journal_handle *handle, - struct dentry *dentry, - struct inode *inode, u64 blkno, - struct buffer_head *parent_fe_bh, - struct buffer_head *insert_bh) -{ - return __ocfs2_add_entry(handle, dentry->d_parent->d_inode, - dentry->d_name.name, dentry->d_name.len, - inode, blkno, parent_fe_bh, insert_bh); -} - /* An orphan dir name is an 8 byte value, printed as a hex string */ #define OCFS2_ORPHAN_NAMELEN ((int)(2 * sizeof(u64))) static struct dentry *ocfs2_lookup(struct inode *dir, struct dentry *dentry, - struct nameidata *nd) + unsigned int flags) { int status; u64 blkno; - struct buffer_head *dirent_bh = NULL; struct inode *inode = NULL; struct dentry *ret; - struct ocfs2_dir_entry *dirent; struct ocfs2_inode_info *oi; - mlog_entry("(0x%p, 0x%p, '%.*s')\n", dir, dentry, - dentry->d_name.len, dentry->d_name.name); + trace_ocfs2_lookup(dir, dentry, dentry->d_name.len, + dentry->d_name.name, + (unsigned long long)OCFS2_I(dir)->ip_blkno, 0); if (dentry->d_name.len > OCFS2_MAX_FILENAME_LEN) { ret = ERR_PTR(-ENAMETOOLONG); goto bail; } - mlog(0, "find name %.*s in directory %"MLFu64"\n", dentry->d_name.len, - dentry->d_name.name, OCFS2_I(dir)->ip_blkno); - - status = ocfs2_meta_lock(dir, NULL, NULL, 0); + status = ocfs2_inode_lock_nested(dir, NULL, 0, OI_LS_PARENT); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -172,15 +123,13 @@ static struct dentry *ocfs2_lookup(struct inode *dir, struct dentry *dentry, goto bail; } - status = ocfs2_find_files_on_disk(dentry->d_name.name, - dentry->d_name.len, &blkno, - dir, &dirent_bh, &dirent); + status = ocfs2_lookup_ino_from_name(dir, dentry->d_name.name, + dentry->d_name.len, &blkno); if (status < 0) goto bail_add; - inode = ocfs2_iget(OCFS2_SB(dir->i_sb), blkno); + inode = ocfs2_iget(OCFS2_SB(dir->i_sb), blkno, 0, 0); if (IS_ERR(inode)) { - mlog(ML_ERROR, "Unable to create inode %"MLFu64"\n", blkno); ret = ERR_PTR(-EACCES); goto bail_unlock; } @@ -193,145 +142,135 @@ static struct dentry *ocfs2_lookup(struct inode *dir, struct dentry *dentry, * unlink. */ spin_lock(&oi->ip_lock); oi->ip_flags &= ~OCFS2_INODE_MAYBE_ORPHANED; - oi->ip_orphaned_slot = OCFS2_INVALID_SLOT; spin_unlock(&oi->ip_lock); bail_add: - - dentry->d_op = &ocfs2_dentry_ops; ret = d_splice_alias(inode, dentry); + if (inode) { + /* + * If d_splice_alias() finds a DCACHE_DISCONNECTED + * dentry, it will d_move() it on top of ourse. The + * return value will indicate this however, so in + * those cases, we switch them around for the locking + * code. + * + * NOTE: This dentry already has ->d_op set from + * ocfs2_get_parent() and ocfs2_get_dentry() + */ + if (ret) + dentry = ret; + + status = ocfs2_dentry_attach_lock(dentry, inode, + OCFS2_I(dir)->ip_blkno); + if (status) { + mlog_errno(status); + ret = ERR_PTR(status); + goto bail_unlock; + } + } else + ocfs2_dentry_attach_gen(dentry); + bail_unlock: /* Don't drop the cluster lock until *after* the d_add -- * unlink on another node will message us to remove that * dentry under this lock so otherwise we can race this with - * the vote thread and have a stale dentry. */ - ocfs2_meta_unlock(dir, 0); + * the downconvert thread and have a stale dentry. */ + ocfs2_inode_unlock(dir, 0); bail: - if (dirent_bh) - brelse(dirent_bh); - mlog_exit_ptr(ret); + trace_ocfs2_lookup_ret(ret); return ret; } -static int ocfs2_fill_new_dir(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - struct inode *parent, - struct inode *inode, - struct buffer_head *fe_bh, - struct ocfs2_alloc_context *data_ac) +static struct inode *ocfs2_get_init_inode(struct inode *dir, umode_t mode) { - int status; - struct buffer_head *new_bh = NULL; - struct ocfs2_dir_entry *de = NULL; - - mlog_entry_void(); - - status = ocfs2_do_extend_dir(osb->sb, handle, inode, fe_bh, - data_ac, NULL, &new_bh); - if (status < 0) { - mlog_errno(status); - goto bail; - } - - ocfs2_set_new_buffer_uptodate(inode, new_bh); - - status = ocfs2_journal_access(handle, inode, new_bh, - OCFS2_JOURNAL_ACCESS_CREATE); - if (status < 0) { - mlog_errno(status); - goto bail; - } - memset(new_bh->b_data, 0, osb->sb->s_blocksize); - - de = (struct ocfs2_dir_entry *) new_bh->b_data; - de->inode = cpu_to_le64(OCFS2_I(inode)->ip_blkno); - de->name_len = 1; - de->rec_len = - cpu_to_le16(OCFS2_DIR_REC_LEN(de->name_len)); - strcpy(de->name, "."); - ocfs2_set_de_type(de, S_IFDIR); - de = (struct ocfs2_dir_entry *) ((char *)de + le16_to_cpu(de->rec_len)); - de->inode = cpu_to_le64(OCFS2_I(parent)->ip_blkno); - de->rec_len = cpu_to_le16(inode->i_sb->s_blocksize - - OCFS2_DIR_REC_LEN(1)); - de->name_len = 2; - strcpy(de->name, ".."); - ocfs2_set_de_type(de, S_IFDIR); - - status = ocfs2_journal_dirty(handle, new_bh); - if (status < 0) { - mlog_errno(status); - goto bail; - } + struct inode *inode; - i_size_write(inode, inode->i_sb->s_blocksize); - inode->i_nlink = 2; - inode->i_blocks = ocfs2_align_bytes_to_sectors(inode->i_sb->s_blocksize); - status = ocfs2_mark_inode_dirty(handle, inode, fe_bh); - if (status < 0) { - mlog_errno(status); - goto bail; + inode = new_inode(dir->i_sb); + if (!inode) { + mlog(ML_ERROR, "new_inode failed!\n"); + return NULL; } - status = 0; -bail: - if (new_bh) - brelse(new_bh); + /* populate as many fields early on as possible - many of + * these are used by the support functions here and in + * callers. */ + if (S_ISDIR(mode)) + set_nlink(inode, 2); + inode_init_owner(inode, dir, mode); + dquot_initialize(inode); + return inode; +} - mlog_exit(status); - return status; +static void ocfs2_cleanup_add_entry_failure(struct ocfs2_super *osb, + struct dentry *dentry, struct inode *inode) +{ + struct ocfs2_dentry_lock *dl = dentry->d_fsdata; + + ocfs2_simple_drop_lockres(osb, &dl->dl_lockres); + ocfs2_lock_res_free(&dl->dl_lockres); + BUG_ON(dl->dl_count != 1); + spin_lock(&dentry_attach_lock); + dentry->d_fsdata = NULL; + spin_unlock(&dentry_attach_lock); + kfree(dl); + iput(inode); } static int ocfs2_mknod(struct inode *dir, struct dentry *dentry, - int mode, + umode_t mode, dev_t dev) { int status = 0; struct buffer_head *parent_fe_bh = NULL; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_super *osb; struct ocfs2_dinode *dirfe; struct buffer_head *new_fe_bh = NULL; - struct buffer_head *de_bh = NULL; struct inode *inode = NULL; struct ocfs2_alloc_context *inode_ac = NULL; struct ocfs2_alloc_context *data_ac = NULL; - - mlog_entry("(0x%p, 0x%p, %d, %lu, '%.*s')\n", dir, dentry, mode, - (unsigned long)dev, dentry->d_name.len, - dentry->d_name.name); + struct ocfs2_alloc_context *meta_ac = NULL; + int want_clusters = 0; + int want_meta = 0; + int xattr_credits = 0; + struct ocfs2_security_xattr_info si = { + .enable = 1, + }; + int did_quota_inode = 0; + struct ocfs2_dir_lookup_result lookup = { NULL, }; + sigset_t oldset; + int did_block_signals = 0; + struct posix_acl *default_acl = NULL, *acl = NULL; + struct ocfs2_dentry_lock *dl = NULL; + + trace_ocfs2_mknod(dir, dentry, dentry->d_name.len, dentry->d_name.name, + (unsigned long long)OCFS2_I(dir)->ip_blkno, + (unsigned long)dev, mode); + + dquot_initialize(dir); /* get our super block */ osb = OCFS2_SB(dir->i_sb); - if (S_ISDIR(mode) && (dir->i_nlink >= OCFS2_LINK_MAX)) { - mlog(ML_ERROR, "inode %"MLFu64" has i_nlink of %u\n", - OCFS2_I(dir)->ip_blkno, dir->i_nlink); - status = -EMLINK; - goto leave; - } - - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto leave; - } - - status = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1); + status = ocfs2_inode_lock(dir, &parent_fe_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); + return status; + } + + if (S_ISDIR(mode) && (dir->i_nlink >= ocfs2_link_max(osb))) { + status = -EMLINK; goto leave; } dirfe = (struct ocfs2_dinode *) parent_fe_bh->b_data; - if (!dirfe->i_links_count) { + if (!ocfs2_read_links_count(dirfe)) { /* can't make a file in a deleted directory. */ status = -ENOENT; goto leave; @@ -345,32 +284,79 @@ static int ocfs2_mknod(struct inode *dir, /* get a spot inside the dir. */ status = ocfs2_prepare_dir_for_insert(osb, dir, parent_fe_bh, dentry->d_name.name, - dentry->d_name.len, &de_bh); + dentry->d_name.len, &lookup); if (status < 0) { mlog_errno(status); goto leave; } /* reserve an inode spot */ - status = ocfs2_reserve_new_inode(osb, handle, &inode_ac); + status = ocfs2_reserve_new_inode(osb, &inode_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); goto leave; } - /* are we making a directory? If so, reserve a cluster for his - * 1st extent. */ - if (S_ISDIR(mode)) { - status = ocfs2_reserve_clusters(osb, handle, 1, &data_ac); - if (status < 0) { - if (status != -ENOSPC) - mlog_errno(status); + inode = ocfs2_get_init_inode(dir, mode); + if (!inode) { + status = -ENOMEM; + mlog_errno(status); + goto leave; + } + + /* get security xattr */ + status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si); + if (status) { + if (status == -EOPNOTSUPP) + si.enable = 0; + else { + mlog_errno(status); goto leave; } } - handle = ocfs2_start_trans(osb, handle, OCFS2_MKNOD_CREDITS); + /* calculate meta data/clusters for setting security and acl xattr */ + status = ocfs2_calc_xattr_init(dir, parent_fe_bh, mode, + &si, &want_clusters, + &xattr_credits, &want_meta); + if (status < 0) { + mlog_errno(status); + goto leave; + } + + /* Reserve a cluster if creating an extent based directory. */ + if (S_ISDIR(mode) && !ocfs2_supports_inline_data(osb)) { + want_clusters += 1; + + /* Dir indexing requires extra space as well */ + if (ocfs2_supports_indexed_dirs(osb)) + want_meta++; + } + + status = ocfs2_reserve_new_metadata_blocks(osb, want_meta, &meta_ac); + if (status < 0) { + if (status != -ENOSPC) + mlog_errno(status); + goto leave; + } + + status = ocfs2_reserve_clusters(osb, want_clusters, &data_ac); + if (status < 0) { + if (status != -ENOSPC) + mlog_errno(status); + goto leave; + } + + status = posix_acl_create(dir, &mode, &default_acl, &acl); + if (status) { + mlog_errno(status); + goto leave; + } + + handle = ocfs2_start_trans(osb, ocfs2_mknod_credits(osb->sb, + S_ISDIR(mode), + xattr_credits)); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -378,10 +364,19 @@ static int ocfs2_mknod(struct inode *dir, goto leave; } + /* Starting to change things, restart is no longer possible. */ + ocfs2_block_signals(&oldset); + did_block_signals = 1; + + status = dquot_alloc_inode(inode); + if (status) + goto leave; + did_quota_inode = 1; + /* do the real work now. */ - status = ocfs2_mknod_locked(osb, dir, dentry, mode, dev, + status = ocfs2_mknod_locked(osb, dir, inode, dev, &new_fe_bh, parent_fe_bh, handle, - &inode, inode_ac); + inode_ac); if (status < 0) { mlog_errno(status); goto leave; @@ -389,57 +384,93 @@ static int ocfs2_mknod(struct inode *dir, if (S_ISDIR(mode)) { status = ocfs2_fill_new_dir(osb, handle, dir, inode, - new_fe_bh, data_ac); + new_fe_bh, data_ac, meta_ac); if (status < 0) { mlog_errno(status); goto leave; } - status = ocfs2_journal_access(handle, dir, parent_fe_bh, - OCFS2_JOURNAL_ACCESS_WRITE); + status = ocfs2_journal_access_di(handle, INODE_CACHE(dir), + parent_fe_bh, + OCFS2_JOURNAL_ACCESS_WRITE); if (status < 0) { mlog_errno(status); goto leave; } - le16_add_cpu(&dirfe->i_links_count, 1); - status = ocfs2_journal_dirty(handle, parent_fe_bh); + ocfs2_add_links_count(dirfe, 1); + ocfs2_journal_dirty(handle, parent_fe_bh); + inc_nlink(dir); + } + + if (default_acl) { + status = ocfs2_set_acl(handle, inode, new_fe_bh, + ACL_TYPE_DEFAULT, default_acl, + meta_ac, data_ac); + } + if (!status && acl) { + status = ocfs2_set_acl(handle, inode, new_fe_bh, + ACL_TYPE_ACCESS, acl, + meta_ac, data_ac); + } + + if (status < 0) { + mlog_errno(status); + goto leave; + } + + if (si.enable) { + status = ocfs2_init_security_set(handle, inode, new_fe_bh, &si, + meta_ac, data_ac); if (status < 0) { mlog_errno(status); goto leave; } - dir->i_nlink++; } + /* + * Do this before adding the entry to the directory. We add + * also set d_op after success so that ->d_iput() will cleanup + * the dentry lock even if ocfs2_add_entry() fails below. + */ + status = ocfs2_dentry_attach_lock(dentry, inode, + OCFS2_I(dir)->ip_blkno); + if (status) { + mlog_errno(status); + goto leave; + } + + dl = dentry->d_fsdata; + status = ocfs2_add_entry(handle, dentry, inode, OCFS2_I(inode)->ip_blkno, parent_fe_bh, - de_bh); + &lookup); if (status < 0) { mlog_errno(status); goto leave; } insert_inode_hash(inode); - dentry->d_op = &ocfs2_dentry_ops; d_instantiate(dentry, inode); status = 0; leave: + if (default_acl) + posix_acl_release(default_acl); + if (acl) + posix_acl_release(acl); + if (status < 0 && did_quota_inode) + dquot_free_inode(inode); if (handle) - ocfs2_commit_trans(handle); - - if (status == -ENOSPC) - mlog(0, "Disk is full\n"); - - if (new_fe_bh) - brelse(new_fe_bh); + ocfs2_commit_trans(osb, handle); - if (de_bh) - brelse(de_bh); + ocfs2_inode_unlock(dir, 1); + if (did_block_signals) + ocfs2_unblock_signals(&oldset); - if (parent_fe_bh) - brelse(parent_fe_bh); + brelse(new_fe_bh); + brelse(parent_fe_bh); + kfree(si.value); - if ((status < 0) && inode) - iput(inode); + ocfs2_free_dir_lookup_result(&lookup); if (inode_ac) ocfs2_free_alloc_context(inode_ac); @@ -447,73 +478,67 @@ leave: if (data_ac) ocfs2_free_alloc_context(data_ac); - mlog_exit(status); + if (meta_ac) + ocfs2_free_alloc_context(meta_ac); + + /* + * We should call iput after the i_mutex of the bitmap been + * unlocked in ocfs2_free_alloc_context, or the + * ocfs2_delete_inode will mutex_lock again. + */ + if ((status < 0) && inode) { + if (dl) + ocfs2_cleanup_add_entry_failure(osb, dentry, inode); + + OCFS2_I(inode)->ip_flags |= OCFS2_INODE_SKIP_ORPHAN_DIR; + clear_nlink(inode); + iput(inode); + } + + if (status) + mlog_errno(status); return status; } -static int ocfs2_mknod_locked(struct ocfs2_super *osb, - struct inode *dir, - struct dentry *dentry, int mode, - dev_t dev, - struct buffer_head **new_fe_bh, - struct buffer_head *parent_fe_bh, - struct ocfs2_journal_handle *handle, - struct inode **ret_inode, - struct ocfs2_alloc_context *inode_ac) +static int __ocfs2_mknod_locked(struct inode *dir, + struct inode *inode, + dev_t dev, + struct buffer_head **new_fe_bh, + struct buffer_head *parent_fe_bh, + handle_t *handle, + struct ocfs2_alloc_context *inode_ac, + u64 fe_blkno, u64 suballoc_loc, u16 suballoc_bit) { int status = 0; + struct ocfs2_super *osb = OCFS2_SB(dir->i_sb); struct ocfs2_dinode *fe = NULL; struct ocfs2_extent_list *fel; - u64 fe_blkno = 0; - u16 suballoc_bit; - struct inode *inode = NULL; - - mlog_entry("(0x%p, 0x%p, %d, %lu, '%.*s')\n", dir, dentry, mode, - (unsigned long)dev, dentry->d_name.len, - dentry->d_name.name); + u16 feat; + struct ocfs2_inode_info *oi = OCFS2_I(inode); *new_fe_bh = NULL; - *ret_inode = NULL; - - status = ocfs2_claim_new_inode(osb, handle, inode_ac, &suballoc_bit, - &fe_blkno); - if (status < 0) { - mlog_errno(status); - goto leave; - } - - inode = new_inode(dir->i_sb); - if (IS_ERR(inode)) { - status = PTR_ERR(inode); - mlog(ML_ERROR, "new_inode failed!\n"); - goto leave; - } /* populate as many fields early on as possible - many of * these are used by the support functions here and in * callers. */ inode->i_ino = ino_from_blkno(osb->sb, fe_blkno); OCFS2_I(inode)->ip_blkno = fe_blkno; - if (S_ISDIR(mode)) - inode->i_nlink = 2; - else - inode->i_nlink = 1; - inode->i_mode = mode; spin_lock(&osb->osb_lock); inode->i_generation = osb->s_next_generation++; spin_unlock(&osb->osb_lock); *new_fe_bh = sb_getblk(osb->sb, fe_blkno); if (!*new_fe_bh) { - status = -EIO; + status = -ENOMEM; mlog_errno(status); goto leave; } - ocfs2_set_new_buffer_uptodate(inode, *new_fe_bh); + ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), *new_fe_bh); - status = ocfs2_journal_access(handle, inode, *new_fe_bh, - OCFS2_JOURNAL_ACCESS_CREATE); + status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), + *new_fe_bh, + OCFS2_JOURNAL_ACCESS_CREATE); if (status < 0) { mlog_errno(status); goto leave; @@ -525,97 +550,124 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb, fe->i_generation = cpu_to_le32(inode->i_generation); fe->i_fs_generation = cpu_to_le32(osb->fs_generation); fe->i_blkno = cpu_to_le64(fe_blkno); + fe->i_suballoc_loc = cpu_to_le64(suballoc_loc); fe->i_suballoc_bit = cpu_to_le16(suballoc_bit); - fe->i_suballoc_slot = cpu_to_le16(osb->slot_num); - fe->i_uid = cpu_to_le32(current->fsuid); - if (dir->i_mode & S_ISGID) { - fe->i_gid = cpu_to_le32(dir->i_gid); - if (S_ISDIR(mode)) - mode |= S_ISGID; - } else - fe->i_gid = cpu_to_le32(current->fsgid); - fe->i_mode = cpu_to_le16(mode); - if (S_ISCHR(mode) || S_ISBLK(mode)) + fe->i_suballoc_slot = cpu_to_le16(inode_ac->ac_alloc_slot); + fe->i_uid = cpu_to_le32(i_uid_read(inode)); + fe->i_gid = cpu_to_le32(i_gid_read(inode)); + fe->i_mode = cpu_to_le16(inode->i_mode); + if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) fe->id1.dev1.i_rdev = cpu_to_le64(huge_encode_dev(dev)); - fe->i_links_count = cpu_to_le16(inode->i_nlink); + ocfs2_set_links_count(fe, inode->i_nlink); fe->i_last_eb_blk = 0; strcpy(fe->i_signature, OCFS2_INODE_SIGNATURE); - le32_add_cpu(&fe->i_flags, OCFS2_VALID_FL); + fe->i_flags |= cpu_to_le32(OCFS2_VALID_FL); fe->i_atime = fe->i_ctime = fe->i_mtime = cpu_to_le64(CURRENT_TIME.tv_sec); fe->i_mtime_nsec = fe->i_ctime_nsec = fe->i_atime_nsec = cpu_to_le32(CURRENT_TIME.tv_nsec); fe->i_dtime = 0; - fel = &fe->id2.i_list; - fel->l_tree_depth = 0; - fel->l_next_free_rec = 0; - fel->l_count = cpu_to_le16(ocfs2_extent_recs_per_inode(osb->sb)); + /* + * If supported, directories start with inline data. If inline + * isn't supported, but indexing is, we start them as indexed. + */ + feat = le16_to_cpu(fe->i_dyn_features); + if (S_ISDIR(inode->i_mode) && ocfs2_supports_inline_data(osb)) { + fe->i_dyn_features = cpu_to_le16(feat | OCFS2_INLINE_DATA_FL); - status = ocfs2_journal_dirty(handle, *new_fe_bh); - if (status < 0) { - mlog_errno(status); - goto leave; + fe->id2.i_data.id_count = cpu_to_le16( + ocfs2_max_inline_data_with_xattr(osb->sb, fe)); + } else { + fel = &fe->id2.i_list; + fel->l_tree_depth = 0; + fel->l_next_free_rec = 0; + fel->l_count = cpu_to_le16(ocfs2_extent_recs_per_inode(osb->sb)); } - if (ocfs2_populate_inode(inode, fe, 1) < 0) { - mlog(ML_ERROR, "populate inode failed! bh->b_blocknr=%llu, " - "i_blkno=%"MLFu64", i_ino=%lu\n", - (unsigned long long) (*new_fe_bh)->b_blocknr, - fe->i_blkno, inode->i_ino); - BUG(); - } + ocfs2_journal_dirty(handle, *new_fe_bh); - ocfs2_inode_set_new(osb, inode); - status = ocfs2_create_new_inode_locks(inode); - if (status < 0) - mlog_errno(status); + ocfs2_populate_inode(inode, fe, 1); + ocfs2_ci_set_new(osb, INODE_CACHE(inode)); + if (!ocfs2_mount_local(osb)) { + status = ocfs2_create_new_inode_locks(inode); + if (status < 0) + mlog_errno(status); + } - status = 0; /* error in ocfs2_create_new_inode_locks is not - * critical */ + oi->i_sync_tid = handle->h_transaction->t_tid; + oi->i_datasync_tid = handle->h_transaction->t_tid; - *ret_inode = inode; leave: if (status < 0) { if (*new_fe_bh) { brelse(*new_fe_bh); *new_fe_bh = NULL; } - if (inode) - iput(inode); } - mlog_exit(status); + if (status) + mlog_errno(status); return status; } +static int ocfs2_mknod_locked(struct ocfs2_super *osb, + struct inode *dir, + struct inode *inode, + dev_t dev, + struct buffer_head **new_fe_bh, + struct buffer_head *parent_fe_bh, + handle_t *handle, + struct ocfs2_alloc_context *inode_ac) +{ + int status = 0; + u64 suballoc_loc, fe_blkno = 0; + u16 suballoc_bit; + + *new_fe_bh = NULL; + + status = ocfs2_claim_new_inode(handle, dir, parent_fe_bh, + inode_ac, &suballoc_loc, + &suballoc_bit, &fe_blkno); + if (status < 0) { + mlog_errno(status); + return status; + } + + return __ocfs2_mknod_locked(dir, inode, dev, new_fe_bh, + parent_fe_bh, handle, inode_ac, + fe_blkno, suballoc_loc, suballoc_bit); +} + static int ocfs2_mkdir(struct inode *dir, struct dentry *dentry, - int mode) + umode_t mode) { int ret; - mlog_entry("(0x%p, 0x%p, %d, '%.*s')\n", dir, dentry, mode, - dentry->d_name.len, dentry->d_name.name); + trace_ocfs2_mkdir(dir, dentry, dentry->d_name.len, dentry->d_name.name, + OCFS2_I(dir)->ip_blkno, mode); ret = ocfs2_mknod(dir, dentry, mode | S_IFDIR, 0); - mlog_exit(ret); + if (ret) + mlog_errno(ret); return ret; } static int ocfs2_create(struct inode *dir, struct dentry *dentry, - int mode, - struct nameidata *nd) + umode_t mode, + bool excl) { int ret; - mlog_entry("(0x%p, 0x%p, %d, '%.*s')\n", dir, dentry, mode, - dentry->d_name.len, dentry->d_name.name); + trace_ocfs2_create(dir, dentry, dentry->d_name.len, dentry->d_name.name, + (unsigned long long)OCFS2_I(dir)->ip_blkno, mode); ret = ocfs2_mknod(dir, dentry, mode | S_IFREG, 0); - mlog_exit(ret); + if (ret) + mlog_errno(ret); return ret; } @@ -624,172 +676,215 @@ static int ocfs2_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry) { - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle; struct inode *inode = old_dentry->d_inode; int err; struct buffer_head *fe_bh = NULL; struct buffer_head *parent_fe_bh = NULL; - struct buffer_head *de_bh = NULL; struct ocfs2_dinode *fe = NULL; struct ocfs2_super *osb = OCFS2_SB(dir->i_sb); + struct ocfs2_dir_lookup_result lookup = { NULL, }; + sigset_t oldset; + u64 old_de_ino; - mlog_entry("(inode=%lu, old='%.*s' new='%.*s')\n", inode->i_ino, - old_dentry->d_name.len, old_dentry->d_name.name, - dentry->d_name.len, dentry->d_name.name); + trace_ocfs2_link((unsigned long long)OCFS2_I(inode)->ip_blkno, + old_dentry->d_name.len, old_dentry->d_name.name, + dentry->d_name.len, dentry->d_name.name); - if (S_ISDIR(inode->i_mode)) { - err = -EPERM; - goto bail; + if (S_ISDIR(inode->i_mode)) + return -EPERM; + + dquot_initialize(dir); + + err = ocfs2_inode_lock_nested(dir, &parent_fe_bh, 1, OI_LS_PARENT); + if (err < 0) { + if (err != -ENOENT) + mlog_errno(err); + return err; } - if (inode->i_nlink >= OCFS2_LINK_MAX) { - err = -EMLINK; - goto bail; + if (!dir->i_nlink) { + err = -ENOENT; + goto out; } - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - err = -ENOMEM; - goto bail; + err = ocfs2_lookup_ino_from_name(dir, old_dentry->d_name.name, + old_dentry->d_name.len, &old_de_ino); + if (err) { + err = -ENOENT; + goto out; } - err = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1); - if (err < 0) { - if (err != -ENOENT) - mlog_errno(err); - goto bail; + /* + * Check whether another node removed the source inode while we + * were in the vfs. + */ + if (old_de_ino != OCFS2_I(inode)->ip_blkno) { + err = -ENOENT; + goto out; } err = ocfs2_check_dir_for_entry(dir, dentry->d_name.name, dentry->d_name.len); if (err) - goto bail; + goto out; err = ocfs2_prepare_dir_for_insert(osb, dir, parent_fe_bh, dentry->d_name.name, - dentry->d_name.len, &de_bh); + dentry->d_name.len, &lookup); if (err < 0) { mlog_errno(err); - goto bail; + goto out; } - err = ocfs2_meta_lock(inode, handle, &fe_bh, 1); + err = ocfs2_inode_lock(inode, &fe_bh, 1); if (err < 0) { if (err != -ENOENT) mlog_errno(err); - goto bail; + goto out; } fe = (struct ocfs2_dinode *) fe_bh->b_data; - if (le16_to_cpu(fe->i_links_count) >= OCFS2_LINK_MAX) { + if (ocfs2_read_links_count(fe) >= ocfs2_link_max(osb)) { err = -EMLINK; - goto bail; + goto out_unlock_inode; } - handle = ocfs2_start_trans(osb, handle, OCFS2_LINK_CREDITS); + handle = ocfs2_start_trans(osb, ocfs2_link_credits(osb->sb)); if (IS_ERR(handle)) { err = PTR_ERR(handle); handle = NULL; mlog_errno(err); - goto bail; + goto out_unlock_inode; } - err = ocfs2_journal_access(handle, inode, fe_bh, - OCFS2_JOURNAL_ACCESS_WRITE); + /* Starting to change things, restart is no longer possible. */ + ocfs2_block_signals(&oldset); + + err = ocfs2_journal_access_di(handle, INODE_CACHE(inode), fe_bh, + OCFS2_JOURNAL_ACCESS_WRITE); if (err < 0) { mlog_errno(err); - goto bail; + goto out_commit; } - inode->i_nlink++; + inc_nlink(inode); inode->i_ctime = CURRENT_TIME; - fe->i_links_count = cpu_to_le16(inode->i_nlink); + ocfs2_set_links_count(fe, inode->i_nlink); fe->i_ctime = cpu_to_le64(inode->i_ctime.tv_sec); fe->i_ctime_nsec = cpu_to_le32(inode->i_ctime.tv_nsec); + ocfs2_journal_dirty(handle, fe_bh); - err = ocfs2_journal_dirty(handle, fe_bh); - if (err < 0) { - le16_add_cpu(&fe->i_links_count, -1); - inode->i_nlink--; + err = ocfs2_add_entry(handle, dentry, inode, + OCFS2_I(inode)->ip_blkno, + parent_fe_bh, &lookup); + if (err) { + ocfs2_add_links_count(fe, -1); + drop_nlink(inode); mlog_errno(err); - goto bail; + goto out_commit; } - err = ocfs2_add_entry(handle, dentry, inode, - OCFS2_I(inode)->ip_blkno, - parent_fe_bh, de_bh); + err = ocfs2_dentry_attach_lock(dentry, inode, OCFS2_I(dir)->ip_blkno); if (err) { - le16_add_cpu(&fe->i_links_count, -1); - inode->i_nlink--; mlog_errno(err); - goto bail; + goto out_commit; } - atomic_inc(&inode->i_count); - dentry->d_op = &ocfs2_dentry_ops; + ihold(inode); d_instantiate(dentry, inode); -bail: - if (handle) - ocfs2_commit_trans(handle); - if (de_bh) - brelse(de_bh); - if (fe_bh) - brelse(fe_bh); - if (parent_fe_bh) - brelse(parent_fe_bh); - mlog_exit(err); +out_commit: + ocfs2_commit_trans(osb, handle); + ocfs2_unblock_signals(&oldset); +out_unlock_inode: + ocfs2_inode_unlock(inode, 1); + +out: + ocfs2_inode_unlock(dir, 1); + + brelse(fe_bh); + brelse(parent_fe_bh); + + ocfs2_free_dir_lookup_result(&lookup); + + if (err) + mlog_errno(err); return err; } +/* + * Takes and drops an exclusive lock on the given dentry. This will + * force other nodes to drop it. + */ +static int ocfs2_remote_dentry_delete(struct dentry *dentry) +{ + int ret; + + ret = ocfs2_dentry_lock(dentry, 1); + if (ret) + mlog_errno(ret); + else + ocfs2_dentry_unlock(dentry, 1); + + return ret; +} + +static inline int ocfs2_inode_is_unlinkable(struct inode *inode) +{ + if (S_ISDIR(inode->i_mode)) { + if (inode->i_nlink == 2) + return 1; + return 0; + } + + if (inode->i_nlink == 1) + return 1; + return 0; +} + static int ocfs2_unlink(struct inode *dir, struct dentry *dentry) { int status; - unsigned int saved_nlink = 0; + int child_locked = 0; + bool is_unlinkable = false; struct inode *inode = dentry->d_inode; + struct inode *orphan_dir = NULL; struct ocfs2_super *osb = OCFS2_SB(dir->i_sb); u64 blkno; struct ocfs2_dinode *fe = NULL; struct buffer_head *fe_bh = NULL; struct buffer_head *parent_node_bh = NULL; - struct ocfs2_journal_handle *handle = NULL; - struct ocfs2_dir_entry *dirent = NULL; - struct buffer_head *dirent_bh = NULL; + handle_t *handle = NULL; char orphan_name[OCFS2_ORPHAN_NAMELEN + 1]; - struct buffer_head *orphan_entry_bh = NULL; + struct ocfs2_dir_lookup_result lookup = { NULL, }; + struct ocfs2_dir_lookup_result orphan_insert = { NULL, }; - mlog_entry("(0x%p, 0x%p, '%.*s')\n", dir, dentry, - dentry->d_name.len, dentry->d_name.name); + trace_ocfs2_unlink(dir, dentry, dentry->d_name.len, + dentry->d_name.name, + (unsigned long long)OCFS2_I(dir)->ip_blkno, + (unsigned long long)OCFS2_I(inode)->ip_blkno); - BUG_ON(dentry->d_parent->d_inode != dir); - - mlog(0, "ino = %"MLFu64"\n", OCFS2_I(inode)->ip_blkno); + dquot_initialize(dir); - if (inode == osb->root_inode) { - mlog(0, "Cannot delete the root directory\n"); - status = -EPERM; - goto leave; - } + BUG_ON(dentry->d_parent->d_inode != dir); - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto leave; - } + if (inode == osb->root_inode) + return -EPERM; - status = ocfs2_meta_lock(dir, handle, &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); - goto leave; + return status; } status = ocfs2_find_files_on_disk(dentry->d_name.name, - dentry->d_name.len, &blkno, - dir, &dirent_bh, &dirent); + dentry->d_name.len, &blkno, dir, + &lookup); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -799,59 +894,48 @@ static int ocfs2_unlink(struct inode *dir, if (OCFS2_I(inode)->ip_blkno != blkno) { status = -ENOENT; - mlog(0, "ip_blkno (%"MLFu64") != dirent blkno (%"MLFu64") " - "ip_flags = %x\n", OCFS2_I(inode)->ip_blkno, blkno, - OCFS2_I(inode)->ip_flags); + trace_ocfs2_unlink_noent( + (unsigned long long)OCFS2_I(inode)->ip_blkno, + (unsigned long long)blkno, + OCFS2_I(inode)->ip_flags); goto leave; } - status = ocfs2_meta_lock(inode, handle, &fe_bh, 1); + status = ocfs2_inode_lock(inode, &fe_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); goto leave; } + child_locked = 1; if (S_ISDIR(inode->i_mode)) { - if (!ocfs2_empty_dir(inode)) { - status = -ENOTEMPTY; - goto leave; - } else if (inode->i_nlink != 2) { + if (inode->i_nlink != 2 || !ocfs2_empty_dir(inode)) { status = -ENOTEMPTY; goto leave; } } - /* There are still a few steps left until we can consider the - * unlink to have succeeded. Save off nlink here before - * modification so we can set it back in case we hit an issue - * before commit. */ - saved_nlink = inode->i_nlink; - if (S_ISDIR(inode->i_mode)) - inode->i_nlink = 0; - else - inode->i_nlink--; - - status = ocfs2_request_unlink_vote(inode, dentry, - (unsigned int) inode->i_nlink); + status = ocfs2_remote_dentry_delete(dentry); if (status < 0) { - /* This vote should succeed under all normal + /* This remote delete should succeed under all normal * circumstances. */ mlog_errno(status); goto leave; } - if (!inode->i_nlink) { - status = ocfs2_prepare_orphan_dir(osb, handle, inode, - orphan_name, - &orphan_entry_bh); + if (ocfs2_inode_is_unlinkable(inode)) { + status = ocfs2_prepare_orphan_dir(osb, &orphan_dir, + OCFS2_I(inode)->ip_blkno, + orphan_name, &orphan_insert); if (status < 0) { mlog_errno(status); goto leave; } + is_unlinkable = true; } - handle = ocfs2_start_trans(osb, handle, OCFS2_UNLINK_CREDITS); + handle = ocfs2_start_trans(osb, ocfs2_unlink_credits(osb->sb)); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -859,8 +943,8 @@ static int ocfs2_unlink(struct inode *dir, goto leave; } - status = ocfs2_journal_access(handle, inode, fe_bh, - OCFS2_JOURNAL_ACCESS_WRITE); + status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), fe_bh, + OCFS2_JOURNAL_ACCESS_WRITE); if (status < 0) { mlog_errno(status); goto leave; @@ -868,99 +952,171 @@ static int ocfs2_unlink(struct inode *dir, fe = (struct ocfs2_dinode *) fe_bh->b_data; - if (!inode->i_nlink) { - status = ocfs2_orphan_add(osb, handle, inode, fe, orphan_name, - orphan_entry_bh); - if (status < 0) { - mlog_errno(status); - goto leave; - } - } - /* delete the name from the parent dir */ - status = ocfs2_delete_entry(handle, dir, dirent, dirent_bh); + status = ocfs2_delete_entry(handle, dir, &lookup); if (status < 0) { mlog_errno(status); goto leave; } - /* We can set nlink on the dinode now. clear the saved version - * so that it doesn't get set later. */ - fe->i_links_count = cpu_to_le16(inode->i_nlink); - saved_nlink = 0; + if (S_ISDIR(inode->i_mode)) + drop_nlink(inode); + drop_nlink(inode); + ocfs2_set_links_count(fe, inode->i_nlink); + ocfs2_journal_dirty(handle, fe_bh); - status = ocfs2_journal_dirty(handle, fe_bh); + dir->i_ctime = dir->i_mtime = CURRENT_TIME; + if (S_ISDIR(inode->i_mode)) + drop_nlink(dir); + + status = ocfs2_mark_inode_dirty(handle, dir, parent_node_bh); if (status < 0) { mlog_errno(status); + if (S_ISDIR(inode->i_mode)) + inc_nlink(dir); goto leave; } - if (S_ISDIR(inode->i_mode)) { - dir->i_nlink--; - status = ocfs2_mark_inode_dirty(handle, dir, - parent_node_bh); - if (status < 0) { + if (is_unlinkable) { + status = ocfs2_orphan_add(osb, handle, inode, fe_bh, + orphan_name, &orphan_insert, orphan_dir); + if (status < 0) mlog_errno(status); - dir->i_nlink++; - } } leave: - if (status < 0 && saved_nlink) - inode->i_nlink = saved_nlink; - if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); - if (fe_bh) - brelse(fe_bh); + if (child_locked) + ocfs2_inode_unlock(inode, 1); - if (dirent_bh) - brelse(dirent_bh); + ocfs2_inode_unlock(dir, 1); - if (parent_node_bh) - brelse(parent_node_bh); + if (orphan_dir) { + /* This was locked for us in ocfs2_prepare_orphan_dir() */ + ocfs2_inode_unlock(orphan_dir, 1); + mutex_unlock(&orphan_dir->i_mutex); + iput(orphan_dir); + } + + brelse(fe_bh); + brelse(parent_node_bh); - if (orphan_entry_bh) - brelse(orphan_entry_bh); + ocfs2_free_dir_lookup_result(&orphan_insert); + ocfs2_free_dir_lookup_result(&lookup); - mlog_exit(status); + if (status && (status != -ENOTEMPTY) && (status != -ENOENT)) + mlog_errno(status); return status; } +static int ocfs2_check_if_ancestor(struct ocfs2_super *osb, + u64 src_inode_no, u64 dest_inode_no) +{ + int ret = 0, i = 0; + u64 parent_inode_no = 0; + u64 child_inode_no = src_inode_no; + struct inode *child_inode; + +#define MAX_LOOKUP_TIMES 32 + while (1) { + child_inode = ocfs2_iget(osb, child_inode_no, 0, 0); + if (IS_ERR(child_inode)) { + ret = PTR_ERR(child_inode); + break; + } + + ret = ocfs2_inode_lock(child_inode, NULL, 0); + if (ret < 0) { + iput(child_inode); + if (ret != -ENOENT) + mlog_errno(ret); + break; + } + + ret = ocfs2_lookup_ino_from_name(child_inode, "..", 2, + &parent_inode_no); + ocfs2_inode_unlock(child_inode, 0); + iput(child_inode); + if (ret < 0) { + ret = -ENOENT; + break; + } + + if (parent_inode_no == dest_inode_no) { + ret = 1; + break; + } + + if (parent_inode_no == osb->root_inode->i_ino) { + ret = 0; + break; + } + + child_inode_no = parent_inode_no; + + if (++i >= MAX_LOOKUP_TIMES) { + mlog(ML_NOTICE, "max lookup times reached, filesystem " + "may have nested directories, " + "src inode: %llu, dest inode: %llu.\n", + (unsigned long long)src_inode_no, + (unsigned long long)dest_inode_no); + ret = 0; + break; + } + } + + return ret; +} + /* * The only place this should be used is rename! * if they have the same id, then the 1st one is the only one locked. */ static int ocfs2_double_lock(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, struct buffer_head **bh1, struct inode *inode1, struct buffer_head **bh2, struct inode *inode2) { int status; + int inode1_is_ancestor, inode2_is_ancestor; struct ocfs2_inode_info *oi1 = OCFS2_I(inode1); struct ocfs2_inode_info *oi2 = OCFS2_I(inode2); struct buffer_head **tmpbh; struct inode *tmpinode; - mlog_entry("(inode1 = %"MLFu64", inode2 = %"MLFu64")\n", - oi1->ip_blkno, oi2->ip_blkno); - - BUG_ON(!handle); + trace_ocfs2_double_lock((unsigned long long)oi1->ip_blkno, + (unsigned long long)oi2->ip_blkno); if (*bh1) *bh1 = NULL; if (*bh2) *bh2 = NULL; - /* we always want to lock the one with the lower lockid first. */ + /* we always want to lock the one with the lower lockid first. + * and if they are nested, we lock ancestor first */ if (oi1->ip_blkno != oi2->ip_blkno) { - if (oi1->ip_blkno < oi2->ip_blkno) { + inode1_is_ancestor = ocfs2_check_if_ancestor(osb, oi2->ip_blkno, + oi1->ip_blkno); + if (inode1_is_ancestor < 0) { + status = inode1_is_ancestor; + goto bail; + } + + inode2_is_ancestor = ocfs2_check_if_ancestor(osb, oi1->ip_blkno, + oi2->ip_blkno); + if (inode2_is_ancestor < 0) { + status = inode2_is_ancestor; + goto bail; + } + + if ((inode1_is_ancestor == 1) || + (oi1->ip_blkno < oi2->ip_blkno && + inode2_is_ancestor == 0)) { /* switch id1 and id2 around */ - mlog(0, "switching them around...\n"); tmpbh = bh2; bh2 = bh1; bh1 = tmpbh; @@ -970,63 +1126,87 @@ static int ocfs2_double_lock(struct ocfs2_super *osb, inode1 = tmpinode; } /* lock id2 */ - status = ocfs2_meta_lock(inode2, handle, bh2, 1); + status = ocfs2_inode_lock_nested(inode2, bh2, 1, + OI_LS_RENAME1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); goto bail; } } + /* lock id1 */ - status = ocfs2_meta_lock(inode1, handle, 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 + * were held on function exit. + */ + if (oi1->ip_blkno != oi2->ip_blkno) { + ocfs2_inode_unlock(inode2, 1); + brelse(*bh2); + *bh2 = NULL; + } + if (status != -ENOENT) mlog_errno(status); - goto bail; } + + trace_ocfs2_double_lock_end( + (unsigned long long)OCFS2_I(inode1)->ip_blkno, + (unsigned long long)OCFS2_I(inode2)->ip_blkno); + bail: - mlog_exit(status); + if (status) + mlog_errno(status); return status; } -#define PARENT_INO(buffer) \ - ((struct ocfs2_dir_entry *) \ - ((char *)buffer + \ - le16_to_cpu(((struct ocfs2_dir_entry *)buffer)->rec_len)))->inode +static void ocfs2_double_unlock(struct inode *inode1, struct inode *inode2) +{ + ocfs2_inode_unlock(inode1, 1); + + if (inode1 != inode2) + ocfs2_inode_unlock(inode2, 1); +} static int ocfs2_rename(struct inode *old_dir, struct dentry *old_dentry, struct inode *new_dir, struct dentry *new_dentry) { - int status = 0, rename_lock = 0; + int status = 0, rename_lock = 0, parents_locked = 0, target_exists = 0; + int old_child_locked = 0, new_child_locked = 0, update_dot_dot = 0; struct inode *old_inode = old_dentry->d_inode; struct inode *new_inode = new_dentry->d_inode; + struct inode *orphan_dir = NULL; struct ocfs2_dinode *newfe = NULL; char orphan_name[OCFS2_ORPHAN_NAMELEN + 1]; - struct buffer_head *orphan_entry_bh = NULL; struct buffer_head *newfe_bh = NULL; - struct buffer_head *insert_entry_bh = NULL; + struct buffer_head *old_inode_bh = NULL; struct ocfs2_super *osb = NULL; - u64 newfe_blkno; - struct ocfs2_journal_handle *handle = NULL; + u64 newfe_blkno, old_de_ino; + handle_t *handle = NULL; struct buffer_head *old_dir_bh = NULL; struct buffer_head *new_dir_bh = NULL; - struct ocfs2_dir_entry *old_de = NULL, *new_de = NULL; // dirent for old_dentry - // and new_dentry - struct buffer_head *new_de_bh = NULL, *old_de_bh = NULL; // bhs for above - struct buffer_head *old_inode_de_bh = NULL; // if old_dentry is a dir, - // this is the 1st dirent bh - nlink_t old_dir_nlink = old_dir->i_nlink, new_dir_nlink = new_dir->i_nlink; - unsigned int links_count; + u32 old_dir_nlink = old_dir->i_nlink; + struct ocfs2_dinode *old_di; + struct ocfs2_dir_lookup_result old_inode_dot_dot_res = { NULL, }; + struct ocfs2_dir_lookup_result target_lookup_res = { NULL, }; + struct ocfs2_dir_lookup_result old_entry_lookup = { NULL, }; + struct ocfs2_dir_lookup_result orphan_insert = { NULL, }; + struct ocfs2_dir_lookup_result target_insert = { NULL, }; + bool should_add_orphan = false; /* At some point it might be nice to break this function up a * bit. */ - mlog_entry("(0x%p, 0x%p, 0x%p, 0x%p, from='%.*s' to='%.*s')\n", - old_dir, old_dentry, new_dir, new_dentry, - old_dentry->d_name.len, old_dentry->d_name.name, - new_dentry->d_name.len, new_dentry->d_name.name); + trace_ocfs2_rename(old_dir, old_dentry, new_dir, new_dentry, + old_dentry->d_name.len, old_dentry->d_name.name, + new_dentry->d_name.len, new_dentry->d_name.name); + + dquot_initialize(old_dir); + dquot_initialize(new_dir); osb = OCFS2_SB(old_dir->i_sb); @@ -1035,15 +1215,7 @@ static int ocfs2_rename(struct inode *old_dir, BUG(); } - if (atomic_read(&old_dentry->d_count) > 2) { - shrink_dcache_parent(old_dentry); - if (atomic_read(&old_dentry->d_count) > 2) { - status = -EBUSY; - goto bail; - } - } - - /* Assume a directory heirarchy thusly: + /* Assume a directory hierarchy thusly: * a/b/c * a/d * a,b,c, and d are all directories. @@ -1054,30 +1226,38 @@ static int ocfs2_rename(struct inode *old_dir, * * And that's why, just like the VFS, we need a file system * rename lock. */ - if (old_dentry != new_dentry) { + if (old_dir != new_dir && S_ISDIR(old_inode->i_mode)) { status = ocfs2_rename_lock(osb); if (status < 0) { mlog_errno(status); goto bail; } rename_lock = 1; - } - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto bail; + /* here we cannot guarantee the inodes haven't just been + * changed, so check if they are nested again */ + status = ocfs2_check_if_ancestor(osb, new_dir->i_ino, + old_inode->i_ino); + if (status < 0) { + mlog_errno(status); + goto bail; + } else if (status == 1) { + status = -EPERM; + trace_ocfs2_rename_not_permitted( + (unsigned long long)old_inode->i_ino, + (unsigned long long)new_dir->i_ino); + goto bail; + } } /* if old and new are the same, this'll just do one lock. */ - status = ocfs2_double_lock(osb, handle, - &old_dir_bh, old_dir, - &new_dir_bh, new_dir); + status = ocfs2_double_lock(osb, &old_dir_bh, old_dir, + &new_dir_bh, new_dir); if (status < 0) { mlog_errno(status); goto bail; } + parents_locked = 1; /* make sure both dirs have bhs * get an extra ref on old_dir_bh if old==new */ @@ -1092,52 +1272,58 @@ static int ocfs2_rename(struct inode *old_dir, } } + /* + * Aside from allowing a meta data update, the locking here + * also ensures that the downconvert thread on other nodes + * won't have to concurrently downconvert the inode and the + * dentry locks. + */ + status = ocfs2_inode_lock_nested(old_inode, &old_inode_bh, 1, + OI_LS_PARENT); + if (status < 0) { + if (status != -ENOENT) + mlog_errno(status); + goto bail; + } + old_child_locked = 1; + + status = ocfs2_remote_dentry_delete(old_dentry); + if (status < 0) { + mlog_errno(status); + goto bail; + } + if (S_ISDIR(old_inode->i_mode)) { - /* Directories actually require metadata updates to - * the directory info so we can't get away with not - * doing node locking on it. */ - status = ocfs2_meta_lock(old_inode, handle, NULL, 1); - if (status < 0) { - if (status != -ENOENT) - mlog_errno(status); - goto bail; - } + u64 old_inode_parent; - status = ocfs2_request_rename_vote(old_inode, old_dentry); - if (status < 0) { - mlog_errno(status); + update_dot_dot = 1; + status = ocfs2_find_files_on_disk("..", 2, &old_inode_parent, + old_inode, + &old_inode_dot_dot_res); + if (status) { + status = -EIO; goto bail; } - status = -EIO; - old_inode_de_bh = ocfs2_bread(old_inode, 0, &status, 0); - if (!old_inode_de_bh) + if (old_inode_parent != OCFS2_I(old_dir)->ip_blkno) { + status = -EIO; goto bail; + } - status = -EIO; - if (le64_to_cpu(PARENT_INO(old_inode_de_bh->b_data)) != - OCFS2_I(old_dir)->ip_blkno) - goto bail; - status = -EMLINK; - if (!new_inode && new_dir!=old_dir && - new_dir->i_nlink >= OCFS2_LINK_MAX) - goto bail; - } else { - /* Ah, the simple case - we're a file so just send a - * message. */ - status = ocfs2_request_rename_vote(old_inode, old_dentry); - if (status < 0) { - mlog_errno(status); + if (!new_inode && new_dir != old_dir && + new_dir->i_nlink >= ocfs2_link_max(osb)) { + status = -EMLINK; goto bail; } } - status = -ENOENT; - old_de_bh = ocfs2_find_entry(old_dentry->d_name.name, - old_dentry->d_name.len, - old_dir, &old_de); - if (!old_de_bh) + status = ocfs2_lookup_ino_from_name(old_dir, old_dentry->d_name.name, + old_dentry->d_name.len, + &old_de_ino); + if (status) { + status = -ENOENT; goto bail; + } /* * Check for inode number is _not_ due to possible IO errors. @@ -1145,15 +1331,17 @@ static int ocfs2_rename(struct inode *old_dir, * and merrily kill the link to whatever was created under the * same name. Goodbye sticky bit ;-< */ - if (le64_to_cpu(old_de->inode) != OCFS2_I(old_inode)->ip_blkno) + if (old_de_ino != OCFS2_I(old_inode)->ip_blkno) { + status = -ENOENT; goto bail; + } /* check if the target already exists (in which case we need * to delete it */ status = ocfs2_find_files_on_disk(new_dentry->d_name.name, new_dentry->d_name.len, - &newfe_blkno, new_dir, &new_de_bh, - &new_de); + &newfe_blkno, new_dir, + &target_lookup_res); /* The only error we allow here is -ENOENT because the new * file not existing is perfectly valid. */ if ((status < 0) && (status != -ENOENT)) { @@ -1162,14 +1350,23 @@ static int ocfs2_rename(struct inode *old_dir, mlog_errno(status); goto bail; } + if (status == 0) + target_exists = 1; - if (!new_de && new_inode) - mlog(ML_ERROR, "inode %lu does not exist in it's parent " - "directory!", new_inode->i_ino); + if (!target_exists && new_inode) { + /* + * Target was unlinked by another node while we were + * waiting to get to ocfs2_rename(). There isn't + * anything we can do here to help the situation, so + * bubble up the appropriate error. + */ + status = -ENOENT; + goto bail; + } /* In case we need to overwrite an existing file, we blow it * away first */ - if (new_de) { + if (target_exists) { /* VFS didn't think there existed an inode here, but * someone else in the cluster must have raced our * rename to create one. Today we error cleanly, in @@ -1178,36 +1375,30 @@ static int ocfs2_rename(struct inode *old_dir, if (!new_inode) { status = -EACCES; - mlog(0, "We found an inode for name %.*s but VFS " - "didn't give us one.\n", new_dentry->d_name.len, - new_dentry->d_name.name); + trace_ocfs2_rename_target_exists(new_dentry->d_name.len, + new_dentry->d_name.name); goto bail; } if (OCFS2_I(new_inode)->ip_blkno != newfe_blkno) { status = -EACCES; - mlog(0, "Inode blkno (%"MLFu64") and dir (%"MLFu64") " - "disagree. ip_flags = %x\n", - OCFS2_I(new_inode)->ip_blkno, newfe_blkno, + trace_ocfs2_rename_disagree( + (unsigned long long)OCFS2_I(new_inode)->ip_blkno, + (unsigned long long)newfe_blkno, OCFS2_I(new_inode)->ip_flags); goto bail; } - status = ocfs2_meta_lock(new_inode, handle, &newfe_bh, 1); + status = ocfs2_inode_lock(new_inode, &newfe_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); goto bail; } + new_child_locked = 1; - if (S_ISDIR(new_inode->i_mode)) - links_count = 0; - else - links_count = (unsigned int) (new_inode->i_nlink - 1); - - status = ocfs2_request_unlink_vote(new_inode, new_dentry, - links_count); + status = ocfs2_remote_dentry_delete(new_dentry); if (status < 0) { mlog_errno(status); goto bail; @@ -1215,20 +1406,19 @@ static int ocfs2_rename(struct inode *old_dir, newfe = (struct ocfs2_dinode *) newfe_bh->b_data; - mlog(0, "aha rename over existing... new_de=%p " - "new_blkno=%"MLFu64" newfebh=%p bhblocknr=%llu\n", - new_de, newfe_blkno, newfe_bh, newfe_bh ? + trace_ocfs2_rename_over_existing( + (unsigned long long)newfe_blkno, newfe_bh, newfe_bh ? (unsigned long long)newfe_bh->b_blocknr : 0ULL); if (S_ISDIR(new_inode->i_mode) || (new_inode->i_nlink == 1)) { - status = ocfs2_prepare_orphan_dir(osb, handle, - new_inode, - orphan_name, - &orphan_entry_bh); + status = ocfs2_prepare_orphan_dir(osb, &orphan_dir, + OCFS2_I(new_inode)->ip_blkno, + orphan_name, &orphan_insert); if (status < 0) { mlog_errno(status); goto bail; } + should_add_orphan = true; } } else { BUG_ON(new_dentry->d_parent->d_inode != new_dir); @@ -1242,14 +1432,14 @@ static int ocfs2_rename(struct inode *old_dir, status = ocfs2_prepare_dir_for_insert(osb, new_dir, new_dir_bh, new_dentry->d_name.name, new_dentry->d_name.len, - &insert_entry_bh); + &target_insert); if (status < 0) { mlog_errno(status); goto bail; } } - handle = ocfs2_start_trans(osb, handle, OCFS2_RENAME_CREDITS); + handle = ocfs2_start_trans(osb, ocfs2_rename_credits(osb->sb)); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -1257,167 +1447,182 @@ static int ocfs2_rename(struct inode *old_dir, goto bail; } - if (new_de) { + if (target_exists) { if (S_ISDIR(new_inode->i_mode)) { - if (!ocfs2_empty_dir(new_inode) || - new_inode->i_nlink != 2) { + if (new_inode->i_nlink != 2 || + !ocfs2_empty_dir(new_inode)) { status = -ENOTEMPTY; goto bail; } } - status = ocfs2_journal_access(handle, new_inode, newfe_bh, - OCFS2_JOURNAL_ACCESS_WRITE); + status = ocfs2_journal_access_di(handle, INODE_CACHE(new_inode), + newfe_bh, + OCFS2_JOURNAL_ACCESS_WRITE); if (status < 0) { mlog_errno(status); goto bail; } - if (S_ISDIR(new_inode->i_mode) || - (newfe->i_links_count == cpu_to_le16(1))){ - status = ocfs2_orphan_add(osb, handle, new_inode, - newfe, orphan_name, - orphan_entry_bh); - if (status < 0) { - mlog_errno(status); - goto bail; - } - } - /* change the dirent to point to the correct inode */ - status = ocfs2_journal_access(handle, new_dir, new_de_bh, - OCFS2_JOURNAL_ACCESS_WRITE); + status = ocfs2_update_entry(new_dir, handle, &target_lookup_res, + old_inode); if (status < 0) { mlog_errno(status); goto bail; } - new_de->inode = cpu_to_le64(OCFS2_I(old_inode)->ip_blkno); - new_de->file_type = old_de->file_type; new_dir->i_version++; - status = ocfs2_journal_dirty(handle, new_de_bh); - if (status < 0) { - mlog_errno(status); - goto bail; - } if (S_ISDIR(new_inode->i_mode)) - newfe->i_links_count = 0; + ocfs2_set_links_count(newfe, 0); else - le16_add_cpu(&newfe->i_links_count, -1); - - status = ocfs2_journal_dirty(handle, newfe_bh); - if (status < 0) { - mlog_errno(status); - goto bail; + ocfs2_add_links_count(newfe, -1); + ocfs2_journal_dirty(handle, newfe_bh); + if (should_add_orphan) { + status = ocfs2_orphan_add(osb, handle, new_inode, + newfe_bh, orphan_name, + &orphan_insert, orphan_dir); + if (status < 0) { + mlog_errno(status); + goto bail; + } } } else { /* if the name was not found in new_dir, add it now */ status = ocfs2_add_entry(handle, new_dentry, old_inode, OCFS2_I(old_inode)->ip_blkno, - new_dir_bh, insert_entry_bh); + new_dir_bh, &target_insert); } old_inode->i_ctime = CURRENT_TIME; mark_inode_dirty(old_inode); - /* now that the name has been added to new_dir, remove the old name */ - status = ocfs2_delete_entry(handle, old_dir, old_de, old_de_bh); + status = ocfs2_journal_access_di(handle, INODE_CACHE(old_inode), + old_inode_bh, + OCFS2_JOURNAL_ACCESS_WRITE); + if (status >= 0) { + old_di = (struct ocfs2_dinode *) old_inode_bh->b_data; + + old_di->i_ctime = cpu_to_le64(old_inode->i_ctime.tv_sec); + old_di->i_ctime_nsec = cpu_to_le32(old_inode->i_ctime.tv_nsec); + ocfs2_journal_dirty(handle, old_inode_bh); + } else + mlog_errno(status); + + /* + * Now that the name has been added to new_dir, remove the old name. + * + * We don't keep any directory entry context around until now + * because the insert might have changed the type of directory + * we're dealing with. + */ + status = ocfs2_find_entry(old_dentry->d_name.name, + old_dentry->d_name.len, old_dir, + &old_entry_lookup); + if (status) + goto bail; + + status = ocfs2_delete_entry(handle, old_dir, &old_entry_lookup); if (status < 0) { mlog_errno(status); goto bail; } if (new_inode) { - new_inode->i_nlink--; + drop_nlink(new_inode); new_inode->i_ctime = CURRENT_TIME; } old_dir->i_ctime = old_dir->i_mtime = CURRENT_TIME; - if (old_inode_de_bh) { - status = ocfs2_journal_access(handle, old_inode, - old_inode_de_bh, - OCFS2_JOURNAL_ACCESS_WRITE); - PARENT_INO(old_inode_de_bh->b_data) = - cpu_to_le64(OCFS2_I(new_dir)->ip_blkno); - status = ocfs2_journal_dirty(handle, old_inode_de_bh); - old_dir->i_nlink--; + + if (update_dot_dot) { + status = ocfs2_update_entry(old_inode, handle, + &old_inode_dot_dot_res, new_dir); + drop_nlink(old_dir); if (new_inode) { - new_inode->i_nlink--; + drop_nlink(new_inode); } else { - new_dir->i_nlink++; + inc_nlink(new_dir); mark_inode_dirty(new_dir); } } mark_inode_dirty(old_dir); - if (new_inode) + ocfs2_mark_inode_dirty(handle, old_dir, old_dir_bh); + if (new_inode) { mark_inode_dirty(new_inode); + ocfs2_mark_inode_dirty(handle, new_inode, newfe_bh); + } - if (old_dir != new_dir) - if (new_dir_nlink != new_dir->i_nlink) { - if (!new_dir_bh) { - mlog(ML_ERROR, "need to change nlink for new " - "dir %"MLFu64" from %d to %d but bh is " - "NULL\n", OCFS2_I(new_dir)->ip_blkno, - (int)new_dir_nlink, new_dir->i_nlink); - } else { - struct ocfs2_dinode *fe; - status = ocfs2_journal_access(handle, - new_dir, - new_dir_bh, - OCFS2_JOURNAL_ACCESS_WRITE); - fe = (struct ocfs2_dinode *) new_dir_bh->b_data; - fe->i_links_count = cpu_to_le16(new_dir->i_nlink); - status = ocfs2_journal_dirty(handle, new_dir_bh); - } - } + if (old_dir != new_dir) { + /* Keep the same times on both directories.*/ + new_dir->i_ctime = new_dir->i_mtime = old_dir->i_ctime; + + /* + * This will also pick up the i_nlink change from the + * block above. + */ + ocfs2_mark_inode_dirty(handle, new_dir, new_dir_bh); + } if (old_dir_nlink != old_dir->i_nlink) { if (!old_dir_bh) { mlog(ML_ERROR, "need to change nlink for old dir " - "%"MLFu64" from %d to %d but bh is NULL!\n", - OCFS2_I(old_dir)->ip_blkno, - (int)old_dir_nlink, - old_dir->i_nlink); + "%llu from %d to %d but bh is NULL!\n", + (unsigned long long)OCFS2_I(old_dir)->ip_blkno, + (int)old_dir_nlink, old_dir->i_nlink); } else { struct ocfs2_dinode *fe; - status = ocfs2_journal_access(handle, old_dir, - old_dir_bh, - OCFS2_JOURNAL_ACCESS_WRITE); + status = ocfs2_journal_access_di(handle, + INODE_CACHE(old_dir), + old_dir_bh, + OCFS2_JOURNAL_ACCESS_WRITE); fe = (struct ocfs2_dinode *) old_dir_bh->b_data; - fe->i_links_count = cpu_to_le16(old_dir->i_nlink); - status = ocfs2_journal_dirty(handle, old_dir_bh); + ocfs2_set_links_count(fe, old_dir->i_nlink); + ocfs2_journal_dirty(handle, old_dir_bh); } } - + ocfs2_dentry_move(old_dentry, new_dentry, old_dir, new_dir); status = 0; bail: if (rename_lock) ocfs2_rename_unlock(osb); if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); + + if (parents_locked) + ocfs2_double_unlock(old_dir, new_dir); + + if (old_child_locked) + ocfs2_inode_unlock(old_inode, 1); + + if (new_child_locked) + ocfs2_inode_unlock(new_inode, 1); + + if (orphan_dir) { + /* This was locked for us in ocfs2_prepare_orphan_dir() */ + ocfs2_inode_unlock(orphan_dir, 1); + mutex_unlock(&orphan_dir->i_mutex); + iput(orphan_dir); + } if (new_inode) sync_mapping_buffers(old_inode->i_mapping); if (new_inode) iput(new_inode); - if (newfe_bh) - brelse(newfe_bh); - if (old_dir_bh) - brelse(old_dir_bh); - if (new_dir_bh) - brelse(new_dir_bh); - if (new_de_bh) - brelse(new_de_bh); - if (old_de_bh) - brelse(old_de_bh); - if (old_inode_de_bh) - brelse(old_inode_de_bh); - if (orphan_entry_bh) - brelse(orphan_entry_bh); - if (insert_entry_bh) - brelse(insert_entry_bh); - - mlog_exit(status); + + ocfs2_free_dir_lookup_result(&target_lookup_res); + ocfs2_free_dir_lookup_result(&old_entry_lookup); + ocfs2_free_dir_lookup_result(&old_inode_dot_dot_res); + ocfs2_free_dir_lookup_result(&orphan_insert); + ocfs2_free_dir_lookup_result(&target_insert); + + brelse(newfe_bh); + brelse(old_inode_bh); + brelse(old_dir_bh); + brelse(new_dir_bh); + + if (status) + mlog_errno(status); return status; } @@ -1427,15 +1632,14 @@ bail: * data, including the null terminator. */ static int ocfs2_create_symlink_data(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, const char *symname) { struct buffer_head **bhs = NULL; const char *c; struct super_block *sb = osb->sb; - u64 p_blkno; - int p_blocks; + u64 p_blkno, p_blocks; int virtual, blocks, status, i, bytes_left; bytes_left = i_size_read(inode) + 1; @@ -1443,8 +1647,8 @@ static int ocfs2_create_symlink_data(struct ocfs2_super *osb, * write i_size + 1 bytes. */ blocks = (bytes_left + sb->s_blocksize - 1) >> sb->s_blocksize_bits; - mlog_entry("i_blocks = %lu, i_size = %llu, blocks = %d\n", - inode->i_blocks, i_size_read(inode), blocks); + trace_ocfs2_create_symlink_data((unsigned long long)inode->i_blocks, + i_size_read(inode), blocks); /* Sanity check -- make sure we're going to fit. */ if (bytes_left > @@ -1461,8 +1665,8 @@ static int ocfs2_create_symlink_data(struct ocfs2_super *osb, goto bail; } - status = ocfs2_extent_map_get_blocks(inode, 0, 1, &p_blkno, - &p_blocks); + status = ocfs2_extent_map_get_blocks(inode, 0, &p_blkno, &p_blocks, + NULL); if (status < 0) { mlog_errno(status); goto bail; @@ -1487,9 +1691,11 @@ static int ocfs2_create_symlink_data(struct ocfs2_super *osb, mlog_errno(status); goto bail; } - ocfs2_set_new_buffer_uptodate(inode, bhs[virtual]); + ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), + bhs[virtual]); - status = ocfs2_journal_access(handle, inode, bhs[virtual], + status = ocfs2_journal_access(handle, INODE_CACHE(inode), + bhs[virtual], OCFS2_JOURNAL_ACCESS_CREATE); if (status < 0) { mlog_errno(status); @@ -1502,11 +1708,7 @@ static int ocfs2_create_symlink_data(struct ocfs2_super *osb, (bytes_left > sb->s_blocksize) ? sb->s_blocksize : bytes_left); - status = ocfs2_journal_dirty(handle, bhs[virtual]); - if (status < 0) { - mlog_errno(status); - goto bail; - } + ocfs2_journal_dirty(handle, bhs[virtual]); virtual++; p_blkno++; @@ -1518,12 +1720,12 @@ bail: if (bhs) { for(i = 0; i < blocks; i++) - if (bhs[i]) - brelse(bhs[i]); + brelse(bhs[i]); kfree(bhs); } - mlog_exit(status); + if (status) + mlog_errno(status); return status; } @@ -1537,16 +1739,28 @@ static int ocfs2_symlink(struct inode *dir, struct inode *inode = NULL; struct super_block *sb; struct buffer_head *new_fe_bh = NULL; - struct buffer_head *de_bh = NULL; struct buffer_head *parent_fe_bh = NULL; struct ocfs2_dinode *fe = NULL; struct ocfs2_dinode *dirfe; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_alloc_context *inode_ac = NULL; struct ocfs2_alloc_context *data_ac = NULL; - - mlog_entry("(0x%p, 0x%p, symname='%s' actual='%.*s')\n", dir, - dentry, symname, dentry->d_name.len, dentry->d_name.name); + struct ocfs2_alloc_context *xattr_ac = NULL; + int want_clusters = 0; + int xattr_credits = 0; + struct ocfs2_security_xattr_info si = { + .enable = 1, + }; + int did_quota = 0, did_quota_inode = 0; + struct ocfs2_dir_lookup_result lookup = { NULL, }; + sigset_t oldset; + int did_block_signals = 0; + struct ocfs2_dentry_lock *dl = NULL; + + trace_ocfs2_symlink_begin(dir, dentry, symname, + dentry->d_name.len, dentry->d_name.name); + + dquot_initialize(dir); sb = dir->i_sb; osb = OCFS2_SB(sb); @@ -1555,23 +1769,16 @@ static int ocfs2_symlink(struct inode *dir, credits = ocfs2_calc_symlink_credits(sb); - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - /* lock the parent directory */ - status = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1); + status = ocfs2_inode_lock(dir, &parent_fe_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); - goto bail; + return status; } dirfe = (struct ocfs2_dinode *) parent_fe_bh->b_data; - if (!dirfe->i_links_count) { + if (!ocfs2_read_links_count(dirfe)) { /* can't make a file in a deleted directory. */ status = -ENOENT; goto bail; @@ -1584,30 +1791,59 @@ static int ocfs2_symlink(struct inode *dir, status = ocfs2_prepare_dir_for_insert(osb, dir, parent_fe_bh, dentry->d_name.name, - dentry->d_name.len, &de_bh); + dentry->d_name.len, &lookup); if (status < 0) { mlog_errno(status); goto bail; } - status = ocfs2_reserve_new_inode(osb, handle, &inode_ac); + status = ocfs2_reserve_new_inode(osb, &inode_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); goto bail; } - /* don't reserve bitmap space for fast symlinks. */ - if (l > ocfs2_fast_symlink_chars(sb)) { - status = ocfs2_reserve_clusters(osb, handle, 1, &data_ac); + inode = ocfs2_get_init_inode(dir, S_IFLNK | S_IRWXUGO); + if (!inode) { + status = -ENOMEM; + mlog_errno(status); + goto bail; + } + + /* get security xattr */ + status = ocfs2_init_security_get(inode, dir, &dentry->d_name, &si); + if (status) { + if (status == -EOPNOTSUPP) + si.enable = 0; + else { + mlog_errno(status); + goto bail; + } + } + + /* calculate meta data/clusters for setting security xattr */ + if (si.enable) { + status = ocfs2_calc_security_init(dir, &si, &want_clusters, + &xattr_credits, &xattr_ac); if (status < 0) { - if (status != -ENOSPC) - mlog_errno(status); + mlog_errno(status); goto bail; } } - handle = ocfs2_start_trans(osb, handle, credits); + /* don't reserve bitmap space for fast symlinks. */ + if (l > ocfs2_fast_symlink_chars(sb)) + want_clusters += 1; + + status = ocfs2_reserve_clusters(osb, want_clusters, &data_ac); + if (status < 0) { + if (status != -ENOSPC) + mlog_errno(status); + goto bail; + } + + handle = ocfs2_start_trans(osb, credits + xattr_credits); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -1615,10 +1851,23 @@ static int ocfs2_symlink(struct inode *dir, goto bail; } - status = ocfs2_mknod_locked(osb, dir, dentry, - S_IFLNK | S_IRWXUGO, 0, - &new_fe_bh, parent_fe_bh, handle, - &inode, inode_ac); + /* Starting to change things, restart is no longer possible. */ + ocfs2_block_signals(&oldset); + did_block_signals = 1; + + status = dquot_alloc_inode(inode); + if (status) + goto bail; + did_quota_inode = 1; + + trace_ocfs2_symlink_create(dir, dentry, dentry->d_name.len, + dentry->d_name.name, + (unsigned long long)OCFS2_I(dir)->ip_blkno, + inode->i_mode); + + status = ocfs2_mknod_locked(osb, dir, inode, + 0, &new_fe_bh, parent_fe_bh, handle, + inode_ac); if (status < 0) { mlog_errno(status); goto bail; @@ -1627,25 +1876,34 @@ static int ocfs2_symlink(struct inode *dir, fe = (struct ocfs2_dinode *) new_fe_bh->b_data; inode->i_rdev = 0; newsize = l - 1; + inode->i_op = &ocfs2_symlink_inode_operations; if (l > ocfs2_fast_symlink_chars(sb)) { - inode->i_op = &ocfs2_symlink_inode_operations; - status = ocfs2_do_extend_allocation(osb, inode, 1, new_fe_bh, - handle, data_ac, NULL, - NULL); + u32 offset = 0; + + status = dquot_alloc_space_nodirty(inode, + ocfs2_clusters_to_bytes(osb->sb, 1)); + if (status) + goto bail; + did_quota = 1; + inode->i_mapping->a_ops = &ocfs2_aops; + status = ocfs2_add_inode_data(osb, inode, &offset, 1, 0, + new_fe_bh, + handle, data_ac, NULL, + NULL); if (status < 0) { if (status != -ENOSPC && status != -EINTR) { - mlog(ML_ERROR, "Failed to extend file to " - "%"MLFu64"\n", - newsize); + mlog(ML_ERROR, + "Failed to extend file to %llu\n", + (unsigned long long)newsize); mlog_errno(status); status = -ENOSPC; } goto bail; } i_size_write(inode, newsize); - inode->i_blocks = ocfs2_align_bytes_to_sectors(newsize); + inode->i_blocks = ocfs2_inode_sector_count(inode); } else { - inode->i_op = &ocfs2_fast_symlink_inode_operations; + inode->i_mapping->a_ops = &ocfs2_fast_symlink_aops; memcpy((char *) fe->id2.i_symlink, symname, l); i_size_write(inode, newsize); inode->i_blocks = 0; @@ -1666,395 +1924,319 @@ static int ocfs2_symlink(struct inode *dir, } } + if (si.enable) { + status = ocfs2_init_security_set(handle, inode, new_fe_bh, &si, + xattr_ac, data_ac); + if (status < 0) { + mlog_errno(status); + goto bail; + } + } + + /* + * Do this before adding the entry to the directory. We add + * also set d_op after success so that ->d_iput() will cleanup + * the dentry lock even if ocfs2_add_entry() fails below. + */ + status = ocfs2_dentry_attach_lock(dentry, inode, OCFS2_I(dir)->ip_blkno); + if (status) { + mlog_errno(status); + goto bail; + } + + dl = dentry->d_fsdata; + status = ocfs2_add_entry(handle, dentry, inode, le64_to_cpu(fe->i_blkno), parent_fe_bh, - de_bh); + &lookup); if (status < 0) { mlog_errno(status); goto bail; } insert_inode_hash(inode); - dentry->d_op = &ocfs2_dentry_ops; d_instantiate(dentry, inode); bail: + if (status < 0 && did_quota) + dquot_free_space_nodirty(inode, + ocfs2_clusters_to_bytes(osb->sb, 1)); + if (status < 0 && did_quota_inode) + dquot_free_inode(inode); if (handle) - ocfs2_commit_trans(handle); - if (new_fe_bh) - brelse(new_fe_bh); - if (parent_fe_bh) - brelse(parent_fe_bh); - if (de_bh) - brelse(de_bh); + ocfs2_commit_trans(osb, handle); + + ocfs2_inode_unlock(dir, 1); + if (did_block_signals) + ocfs2_unblock_signals(&oldset); + + brelse(new_fe_bh); + brelse(parent_fe_bh); + kfree(si.value); + ocfs2_free_dir_lookup_result(&lookup); if (inode_ac) ocfs2_free_alloc_context(inode_ac); if (data_ac) ocfs2_free_alloc_context(data_ac); - if ((status < 0) && inode) + if (xattr_ac) + ocfs2_free_alloc_context(xattr_ac); + if ((status < 0) && inode) { + if (dl) + ocfs2_cleanup_add_entry_failure(osb, dentry, inode); + + OCFS2_I(inode)->ip_flags |= OCFS2_INODE_SKIP_ORPHAN_DIR; + clear_nlink(inode); iput(inode); + } - mlog_exit(status); + if (status) + mlog_errno(status); return status; } -int ocfs2_check_dir_entry(struct inode * dir, - struct ocfs2_dir_entry * de, - struct buffer_head * bh, - unsigned long offset) +static int ocfs2_blkno_stringify(u64 blkno, char *name) { - const char *error_msg = NULL; - const int rlen = le16_to_cpu(de->rec_len); - - if (rlen < OCFS2_DIR_REC_LEN(1)) - error_msg = "rec_len is smaller than minimal"; - else if (rlen % 4 != 0) - error_msg = "rec_len % 4 != 0"; - else if (rlen < OCFS2_DIR_REC_LEN(de->name_len)) - error_msg = "rec_len is too small for name_len"; - else if (((char *) de - bh->b_data) + rlen > dir->i_sb->s_blocksize) - error_msg = "directory entry across blocks"; - - if (error_msg != NULL) - mlog(ML_ERROR, "bad entry in directory #%"MLFu64": %s - " - "offset=%lu, inode=%"MLFu64", rec_len=%d, name_len=%d\n", - OCFS2_I(dir)->ip_blkno, error_msg, offset, - le64_to_cpu(de->inode), rlen, de->name_len); - return error_msg == NULL ? 1 : 0; + int status, namelen; + + namelen = snprintf(name, OCFS2_ORPHAN_NAMELEN + 1, "%016llx", + (long long)blkno); + if (namelen <= 0) { + if (namelen) + status = namelen; + else + status = -EINVAL; + mlog_errno(status); + goto bail; + } + if (namelen != OCFS2_ORPHAN_NAMELEN) { + status = -EINVAL; + mlog_errno(status); + goto bail; + } + + trace_ocfs2_blkno_stringify(blkno, name, namelen); + + status = 0; +bail: + if (status < 0) + mlog_errno(status); + return status; } -/* we don't always have a dentry for what we want to add, so people - * like orphan dir can call this instead. - * - * If you pass me insert_bh, I'll skip the search of the other dir - * blocks and put the record in there. - */ -static int __ocfs2_add_entry(struct ocfs2_journal_handle *handle, - struct inode *dir, - const char *name, int namelen, - struct inode *inode, u64 blkno, - struct buffer_head *parent_fe_bh, - struct buffer_head *insert_bh) +static int ocfs2_lookup_lock_orphan_dir(struct ocfs2_super *osb, + struct inode **ret_orphan_dir, + struct buffer_head **ret_orphan_dir_bh) { - unsigned long offset; - unsigned short rec_len; - struct ocfs2_dir_entry *de, *de1; - struct super_block *sb; - int retval, status; + struct inode *orphan_dir_inode; + struct buffer_head *orphan_dir_bh = NULL; + int ret = 0; - mlog_entry_void(); + orphan_dir_inode = ocfs2_get_system_file_inode(osb, + ORPHAN_DIR_SYSTEM_INODE, + osb->slot_num); + if (!orphan_dir_inode) { + ret = -ENOENT; + mlog_errno(ret); + return ret; + } - sb = dir->i_sb; + mutex_lock(&orphan_dir_inode->i_mutex); - if (!namelen) - return -EINVAL; + ret = ocfs2_inode_lock(orphan_dir_inode, &orphan_dir_bh, 1); + if (ret < 0) { + mutex_unlock(&orphan_dir_inode->i_mutex); + iput(orphan_dir_inode); - rec_len = OCFS2_DIR_REC_LEN(namelen); - offset = 0; - de = (struct ocfs2_dir_entry *) insert_bh->b_data; - while (1) { - BUG_ON((char *)de >= sb->s_blocksize + insert_bh->b_data); - /* These checks should've already been passed by the - * prepare function, but I guess we can leave them - * here anyway. */ - if (!ocfs2_check_dir_entry(dir, de, insert_bh, offset)) { - retval = -ENOENT; - goto bail; - } - if (ocfs2_match(namelen, name, de)) { - retval = -EEXIST; - goto bail; - } - if (((le64_to_cpu(de->inode) == 0) && - (le16_to_cpu(de->rec_len) >= rec_len)) || - (le16_to_cpu(de->rec_len) >= - (OCFS2_DIR_REC_LEN(de->name_len) + rec_len))) { - status = ocfs2_journal_access(handle, dir, insert_bh, - OCFS2_JOURNAL_ACCESS_WRITE); - /* By now the buffer is marked for journaling */ - offset += le16_to_cpu(de->rec_len); - if (le64_to_cpu(de->inode)) { - de1 = (struct ocfs2_dir_entry *)((char *) de + - OCFS2_DIR_REC_LEN(de->name_len)); - de1->rec_len = - cpu_to_le16(le16_to_cpu(de->rec_len) - - OCFS2_DIR_REC_LEN(de->name_len)); - de->rec_len = cpu_to_le16(OCFS2_DIR_REC_LEN(de->name_len)); - de = de1; - } - de->file_type = OCFS2_FT_UNKNOWN; - if (blkno) { - de->inode = cpu_to_le64(blkno); - ocfs2_set_de_type(de, inode->i_mode); - } else - de->inode = 0; - de->name_len = namelen; - memcpy(de->name, name, namelen); - - dir->i_mtime = dir->i_ctime = CURRENT_TIME; - dir->i_version++; - status = ocfs2_journal_dirty(handle, insert_bh); - retval = 0; - goto bail; - } - offset += le16_to_cpu(de->rec_len); - de = (struct ocfs2_dir_entry *) ((char *) de + le16_to_cpu(de->rec_len)); + mlog_errno(ret); + return ret; } - /* when you think about it, the assert above should prevent us - * from ever getting here. */ - retval = -ENOSPC; -bail: + *ret_orphan_dir = orphan_dir_inode; + *ret_orphan_dir_bh = orphan_dir_bh; - mlog_exit(retval); - return retval; + return 0; } - -/* - * ocfs2_delete_entry deletes a directory entry by merging it with the - * previous entry - */ -static int ocfs2_delete_entry(struct ocfs2_journal_handle *handle, - struct inode *dir, - struct ocfs2_dir_entry *de_del, - struct buffer_head *bh) +static int __ocfs2_prepare_orphan_dir(struct inode *orphan_dir_inode, + struct buffer_head *orphan_dir_bh, + u64 blkno, + char *name, + struct ocfs2_dir_lookup_result *lookup) { - struct ocfs2_dir_entry *de, *pde; - int i, status = -ENOENT; + int ret; + struct ocfs2_super *osb = OCFS2_SB(orphan_dir_inode->i_sb); - mlog_entry("(0x%p, 0x%p, 0x%p, 0x%p)\n", handle, dir, de_del, bh); + ret = ocfs2_blkno_stringify(blkno, name); + if (ret < 0) { + mlog_errno(ret); + return ret; + } - i = 0; - pde = NULL; - de = (struct ocfs2_dir_entry *) bh->b_data; - while (i < bh->b_size) { - if (!ocfs2_check_dir_entry(dir, de, bh, i)) { - status = -EIO; - mlog_errno(status); - goto bail; - } - if (de == de_del) { - status = ocfs2_journal_access(handle, dir, bh, - OCFS2_JOURNAL_ACCESS_WRITE); - if (status < 0) { - status = -EIO; - mlog_errno(status); - goto bail; - } - if (pde) - pde->rec_len = - cpu_to_le16(le16_to_cpu(pde->rec_len) + - le16_to_cpu(de->rec_len)); - else - de->inode = 0; - dir->i_version++; - status = ocfs2_journal_dirty(handle, bh); - goto bail; - } - i += le16_to_cpu(de->rec_len); - pde = de; - de = (struct ocfs2_dir_entry *)((char *)de + le16_to_cpu(de->rec_len)); + ret = ocfs2_prepare_dir_for_insert(osb, orphan_dir_inode, + orphan_dir_bh, name, + OCFS2_ORPHAN_NAMELEN, lookup); + if (ret < 0) { + mlog_errno(ret); + return ret; } -bail: - mlog_exit(status); - return status; + + return 0; } -/* - * Returns 0 if not found, -1 on failure, and 1 on success +/** + * ocfs2_prepare_orphan_dir() - Prepare an orphan directory for + * insertion of an orphan. + * @osb: ocfs2 file system + * @ret_orphan_dir: Orphan dir inode - returned locked! + * @blkno: Actual block number of the inode to be inserted into orphan dir. + * @lookup: dir lookup result, to be passed back into functions like + * ocfs2_orphan_add + * + * Returns zero on success and the ret_orphan_dir, name and lookup + * fields will be populated. + * + * Returns non-zero on failure. */ -static int inline ocfs2_search_dirblock(struct buffer_head *bh, - struct inode *dir, - const char *name, int namelen, - unsigned long offset, - struct ocfs2_dir_entry **res_dir) +static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb, + struct inode **ret_orphan_dir, + u64 blkno, + char *name, + struct ocfs2_dir_lookup_result *lookup) { - struct ocfs2_dir_entry *de; - char *dlimit, *de_buf; - int de_len; + struct inode *orphan_dir_inode = NULL; + struct buffer_head *orphan_dir_bh = NULL; int ret = 0; - mlog_entry_void(); + ret = ocfs2_lookup_lock_orphan_dir(osb, &orphan_dir_inode, + &orphan_dir_bh); + if (ret < 0) { + mlog_errno(ret); + return ret; + } - de_buf = bh->b_data; - dlimit = de_buf + dir->i_sb->s_blocksize; + ret = __ocfs2_prepare_orphan_dir(orphan_dir_inode, orphan_dir_bh, + blkno, name, lookup); + if (ret < 0) { + mlog_errno(ret); + goto out; + } - while (de_buf < dlimit) { - /* this code is executed quadratically often */ - /* do minimal checking `by hand' */ + *ret_orphan_dir = orphan_dir_inode; - de = (struct ocfs2_dir_entry *) de_buf; +out: + brelse(orphan_dir_bh); - if (de_buf + namelen <= dlimit && - ocfs2_match(namelen, name, de)) { - /* found a match - just to be sure, do a full check */ - if (!ocfs2_check_dir_entry(dir, de, bh, offset)) { - ret = -1; - goto bail; - } - *res_dir = de; - ret = 1; - goto bail; - } - - /* prevent looping on a bad block */ - de_len = le16_to_cpu(de->rec_len); - if (de_len <= 0) { - ret = -1; - goto bail; - } - - de_buf += de_len; - offset += de_len; + if (ret) { + ocfs2_inode_unlock(orphan_dir_inode, 1); + mutex_unlock(&orphan_dir_inode->i_mutex); + iput(orphan_dir_inode); } -bail: - mlog_exit(ret); + if (ret) + mlog_errno(ret); return ret; } -struct buffer_head *ocfs2_find_entry(const char *name, int namelen, - struct inode *dir, - struct ocfs2_dir_entry **res_dir) +static int ocfs2_orphan_add(struct ocfs2_super *osb, + handle_t *handle, + struct inode *inode, + struct buffer_head *fe_bh, + char *name, + struct ocfs2_dir_lookup_result *lookup, + struct inode *orphan_dir_inode) { - struct super_block *sb; - struct buffer_head *bh_use[NAMEI_RA_SIZE]; - struct buffer_head *bh, *ret = NULL; - unsigned long start, block, b; - int ra_max = 0; /* Number of bh's in the readahead - buffer, bh_use[] */ - int ra_ptr = 0; /* Current index into readahead - buffer */ - int num = 0; - int nblocks, i, err; - - mlog_entry_void(); - - *res_dir = NULL; - sb = dir->i_sb; + struct buffer_head *orphan_dir_bh = NULL; + int status = 0; + struct ocfs2_dinode *orphan_fe; + struct ocfs2_dinode *fe = (struct ocfs2_dinode *) fe_bh->b_data; - nblocks = i_size_read(dir) >> sb->s_blocksize_bits; - start = OCFS2_I(dir)->ip_dir_start_lookup; - if (start >= nblocks) - start = 0; - block = start; + trace_ocfs2_orphan_add_begin( + (unsigned long long)OCFS2_I(inode)->ip_blkno); -restart: - do { - /* - * We deal with the read-ahead logic here. - */ - if (ra_ptr >= ra_max) { - /* Refill the readahead buffer */ - ra_ptr = 0; - b = block; - for (ra_max = 0; ra_max < NAMEI_RA_SIZE; ra_max++) { - /* - * Terminate if we reach the end of the - * directory and must wrap, or if our - * search has finished at this block. - */ - if (b >= nblocks || (num && block == start)) { - bh_use[ra_max] = NULL; - break; - } - num++; - - /* XXX: questionable readahead stuff here */ - bh = ocfs2_bread(dir, b++, &err, 1); - bh_use[ra_max] = bh; -#if 0 // ??? - if (bh) - ll_rw_block(READ, 1, &bh); -#endif - } - } - if ((bh = bh_use[ra_ptr++]) == NULL) - goto next; - wait_on_buffer(bh); - if (!buffer_uptodate(bh)) { - /* read error, skip block & hope for the best */ - brelse(bh); - goto next; - } - i = ocfs2_search_dirblock(bh, dir, name, namelen, - block << sb->s_blocksize_bits, - res_dir); - if (i == 1) { - OCFS2_I(dir)->ip_dir_start_lookup = block; - ret = bh; - goto cleanup_and_exit; - } else { - brelse(bh); - if (i < 0) - goto cleanup_and_exit; - } - next: - if (++block >= nblocks) - block = 0; - } while (block != start); + status = ocfs2_read_inode_block(orphan_dir_inode, &orphan_dir_bh); + if (status < 0) { + mlog_errno(status); + goto leave; + } + + status = ocfs2_journal_access_di(handle, + INODE_CACHE(orphan_dir_inode), + orphan_dir_bh, + OCFS2_JOURNAL_ACCESS_WRITE); + if (status < 0) { + mlog_errno(status); + goto leave; + } /* - * If the directory has grown while we were searching, then - * search the last part of the directory before giving up. + * We're going to journal the change of i_flags and i_orphaned_slot. + * It's safe anyway, though some callers may duplicate the journaling. + * Journaling within the func just make the logic look more + * straightforward. */ - block = nblocks; - nblocks = i_size_read(dir) >> sb->s_blocksize_bits; - if (block < nblocks) { - start = 0; - goto restart; + status = ocfs2_journal_access_di(handle, + INODE_CACHE(inode), + fe_bh, + OCFS2_JOURNAL_ACCESS_WRITE); + if (status < 0) { + mlog_errno(status); + goto leave; } -cleanup_and_exit: - /* Clean up the read-ahead blocks */ - for (; ra_ptr < ra_max; ra_ptr++) - brelse(bh_use[ra_ptr]); + /* we're a cluster, and nlink can change on disk from + * underneath us... */ + orphan_fe = (struct ocfs2_dinode *) orphan_dir_bh->b_data; + if (S_ISDIR(inode->i_mode)) + ocfs2_add_links_count(orphan_fe, 1); + set_nlink(orphan_dir_inode, ocfs2_read_links_count(orphan_fe)); + ocfs2_journal_dirty(handle, orphan_dir_bh); - mlog_exit_ptr(ret); - return ret; -} + status = __ocfs2_add_entry(handle, orphan_dir_inode, name, + OCFS2_ORPHAN_NAMELEN, inode, + OCFS2_I(inode)->ip_blkno, + orphan_dir_bh, lookup); + if (status < 0) { + mlog_errno(status); + goto rollback; + } -static int ocfs2_blkno_stringify(u64 blkno, char *name) -{ - int status, namelen; + fe->i_flags |= cpu_to_le32(OCFS2_ORPHANED_FL); + OCFS2_I(inode)->ip_flags &= ~OCFS2_INODE_SKIP_ORPHAN_DIR; - mlog_entry_void(); + /* Record which orphan dir our inode now resides + * in. delete_inode will use this to determine which orphan + * dir to lock. */ + fe->i_orphaned_slot = cpu_to_le16(osb->slot_num); - namelen = snprintf(name, OCFS2_ORPHAN_NAMELEN + 1, "%016"MLFx64, - blkno); - if (namelen <= 0) { - if (namelen) - status = namelen; - else - status = -EINVAL; - mlog_errno(status); - goto bail; - } - if (namelen != OCFS2_ORPHAN_NAMELEN) { - status = -EINVAL; - mlog_errno(status); - goto bail; + ocfs2_journal_dirty(handle, fe_bh); + + trace_ocfs2_orphan_add_end((unsigned long long)OCFS2_I(inode)->ip_blkno, + osb->slot_num); + +rollback: + if (status < 0) { + if (S_ISDIR(inode->i_mode)) + ocfs2_add_links_count(orphan_fe, -1); + set_nlink(orphan_dir_inode, ocfs2_read_links_count(orphan_fe)); } - mlog(0, "built filename '%s' for orphan dir (len=%d)\n", name, - namelen); +leave: + brelse(orphan_dir_bh); - status = 0; -bail: - mlog_exit(status); return status; } -static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - struct inode *inode, - char *name, - struct buffer_head **de_bh) +/* unlike orphan_add, we expect the orphan dir to already be locked here. */ +int ocfs2_orphan_del(struct ocfs2_super *osb, + handle_t *handle, + struct inode *orphan_dir_inode, + struct inode *inode, + struct buffer_head *orphan_dir_bh) { - struct inode *orphan_dir_inode = NULL; - struct buffer_head *orphan_dir_bh = NULL; + char name[OCFS2_ORPHAN_NAMELEN + 1]; + struct ocfs2_dinode *orphan_fe; int status = 0; + struct ocfs2_dir_lookup_result lookup = { NULL, }; status = ocfs2_blkno_stringify(OCFS2_I(inode)->ip_blkno, name); if (status < 0) { @@ -2062,194 +2244,404 @@ static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb, goto leave; } - orphan_dir_inode = ocfs2_get_system_file_inode(osb, - ORPHAN_DIR_SYSTEM_INODE, - osb->slot_num); - if (!orphan_dir_inode) { - status = -ENOENT; + trace_ocfs2_orphan_del( + (unsigned long long)OCFS2_I(orphan_dir_inode)->ip_blkno, + name, OCFS2_ORPHAN_NAMELEN); + + /* find it's spot in the orphan directory */ + status = ocfs2_find_entry(name, OCFS2_ORPHAN_NAMELEN, orphan_dir_inode, + &lookup); + if (status) { mlog_errno(status); goto leave; } - ocfs2_handle_add_inode(handle, orphan_dir_inode); - status = ocfs2_meta_lock(orphan_dir_inode, handle, &orphan_dir_bh, 1); + /* remove it from the orphan directory */ + status = ocfs2_delete_entry(handle, orphan_dir_inode, &lookup); if (status < 0) { mlog_errno(status); goto leave; } - status = ocfs2_prepare_dir_for_insert(osb, orphan_dir_inode, - orphan_dir_bh, name, - OCFS2_ORPHAN_NAMELEN, de_bh); + status = ocfs2_journal_access_di(handle, + INODE_CACHE(orphan_dir_inode), + orphan_dir_bh, + OCFS2_JOURNAL_ACCESS_WRITE); if (status < 0) { mlog_errno(status); goto leave; } -leave: - if (orphan_dir_inode) - iput(orphan_dir_inode); + /* do the i_nlink dance! :) */ + orphan_fe = (struct ocfs2_dinode *) orphan_dir_bh->b_data; + if (S_ISDIR(inode->i_mode)) + ocfs2_add_links_count(orphan_fe, -1); + set_nlink(orphan_dir_inode, ocfs2_read_links_count(orphan_fe)); + ocfs2_journal_dirty(handle, orphan_dir_bh); - if (orphan_dir_bh) - brelse(orphan_dir_bh); +leave: + ocfs2_free_dir_lookup_result(&lookup); - mlog_exit(status); + if (status) + mlog_errno(status); return status; } -static int ocfs2_orphan_add(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - struct inode *inode, - struct ocfs2_dinode *fe, - char *name, - struct buffer_head *de_bh) +/** + * ocfs2_prep_new_orphaned_file() - Prepare the orphan dir to receive a newly + * allocated file. This is different from the typical 'add to orphan dir' + * operation in that the inode does not yet exist. This is a problem because + * the orphan dir stringifies the inode block number to come up with it's + * dirent. Obviously if the inode does not yet exist we have a chicken and egg + * problem. This function works around it by calling deeper into the orphan + * and suballoc code than other callers. Use this only by necessity. + * @dir: The directory which this inode will ultimately wind up under - not the + * orphan dir! + * @dir_bh: buffer_head the @dir inode block + * @orphan_name: string of length (CFS2_ORPHAN_NAMELEN + 1). Will be filled + * with the string to be used for orphan dirent. Pass back to the orphan dir + * code. + * @ret_orphan_dir: orphan dir inode returned to be passed back into orphan + * dir code. + * @ret_di_blkno: block number where the new inode will be allocated. + * @orphan_insert: Dir insert context to be passed back into orphan dir code. + * @ret_inode_ac: Inode alloc context to be passed back to the allocator. + * + * Returns zero on success and the ret_orphan_dir, name and lookup + * fields will be populated. + * + * Returns non-zero on failure. + */ +static int ocfs2_prep_new_orphaned_file(struct inode *dir, + struct buffer_head *dir_bh, + char *orphan_name, + struct inode **ret_orphan_dir, + u64 *ret_di_blkno, + struct ocfs2_dir_lookup_result *orphan_insert, + struct ocfs2_alloc_context **ret_inode_ac) { - struct inode *orphan_dir_inode = NULL; + int ret; + u64 di_blkno; + struct ocfs2_super *osb = OCFS2_SB(dir->i_sb); + struct inode *orphan_dir = NULL; struct buffer_head *orphan_dir_bh = NULL; - int status = 0; - struct ocfs2_dinode *orphan_fe; + struct ocfs2_alloc_context *inode_ac = NULL; - mlog_entry("(inode->i_ino = %lu)\n", inode->i_ino); + ret = ocfs2_lookup_lock_orphan_dir(osb, &orphan_dir, &orphan_dir_bh); + if (ret < 0) { + mlog_errno(ret); + return ret; + } - orphan_dir_inode = ocfs2_get_system_file_inode(osb, - ORPHAN_DIR_SYSTEM_INODE, - osb->slot_num); - if (!orphan_dir_inode) { - status = -ENOENT; + /* reserve an inode spot */ + ret = ocfs2_reserve_new_inode(osb, &inode_ac); + if (ret < 0) { + if (ret != -ENOSPC) + mlog_errno(ret); + goto out; + } + + ret = ocfs2_find_new_inode_loc(dir, dir_bh, inode_ac, + &di_blkno); + if (ret) { + mlog_errno(ret); + goto out; + } + + ret = __ocfs2_prepare_orphan_dir(orphan_dir, orphan_dir_bh, + di_blkno, orphan_name, orphan_insert); + if (ret < 0) { + mlog_errno(ret); + goto out; + } + +out: + if (ret == 0) { + *ret_orphan_dir = orphan_dir; + *ret_di_blkno = di_blkno; + *ret_inode_ac = inode_ac; + /* + * orphan_name and orphan_insert are already up to + * date via prepare_orphan_dir + */ + } else { + /* Unroll reserve_new_inode* */ + if (inode_ac) + ocfs2_free_alloc_context(inode_ac); + + /* Unroll orphan dir locking */ + mutex_unlock(&orphan_dir->i_mutex); + ocfs2_inode_unlock(orphan_dir, 1); + iput(orphan_dir); + } + + brelse(orphan_dir_bh); + + return ret; +} + +int ocfs2_create_inode_in_orphan(struct inode *dir, + int mode, + struct inode **new_inode) +{ + int status, did_quota_inode = 0; + struct inode *inode = NULL; + struct inode *orphan_dir = NULL; + struct ocfs2_super *osb = OCFS2_SB(dir->i_sb); + struct ocfs2_dinode *di = NULL; + handle_t *handle = NULL; + char orphan_name[OCFS2_ORPHAN_NAMELEN + 1]; + struct buffer_head *parent_di_bh = NULL; + struct buffer_head *new_di_bh = NULL; + struct ocfs2_alloc_context *inode_ac = NULL; + struct ocfs2_dir_lookup_result orphan_insert = { NULL, }; + u64 uninitialized_var(di_blkno), suballoc_loc; + u16 suballoc_bit; + + status = ocfs2_inode_lock(dir, &parent_di_bh, 1); + if (status < 0) { + if (status != -ENOENT) + mlog_errno(status); + return status; + } + + status = ocfs2_prep_new_orphaned_file(dir, parent_di_bh, + orphan_name, &orphan_dir, + &di_blkno, &orphan_insert, &inode_ac); + if (status < 0) { + if (status != -ENOSPC) + mlog_errno(status); + goto leave; + } + + inode = ocfs2_get_init_inode(dir, mode); + if (!inode) { + status = -ENOMEM; mlog_errno(status); goto leave; } - status = ocfs2_read_block(osb, - OCFS2_I(orphan_dir_inode)->ip_blkno, - &orphan_dir_bh, OCFS2_BH_CACHED, - orphan_dir_inode); - if (status < 0) { + handle = ocfs2_start_trans(osb, ocfs2_mknod_credits(osb->sb, 0, 0)); + if (IS_ERR(handle)) { + status = PTR_ERR(handle); + handle = NULL; mlog_errno(status); goto leave; } - status = ocfs2_journal_access(handle, orphan_dir_inode, orphan_dir_bh, - OCFS2_JOURNAL_ACCESS_WRITE); + status = dquot_alloc_inode(inode); + if (status) + goto leave; + did_quota_inode = 1; + + status = ocfs2_claim_new_inode_at_loc(handle, dir, inode_ac, + &suballoc_loc, + &suballoc_bit, di_blkno); if (status < 0) { mlog_errno(status); goto leave; } - /* we're a cluster, and nlink can change on disk from - * underneath us... */ - orphan_fe = (struct ocfs2_dinode *) orphan_dir_bh->b_data; - if (S_ISDIR(inode->i_mode)) - le16_add_cpu(&orphan_fe->i_links_count, 1); - orphan_dir_inode->i_nlink = le16_to_cpu(orphan_fe->i_links_count); - - status = ocfs2_journal_dirty(handle, orphan_dir_bh); + clear_nlink(inode); + /* do the real work now. */ + status = __ocfs2_mknod_locked(dir, inode, + 0, &new_di_bh, parent_di_bh, handle, + inode_ac, di_blkno, suballoc_loc, + suballoc_bit); if (status < 0) { mlog_errno(status); goto leave; } - status = __ocfs2_add_entry(handle, orphan_dir_inode, name, - OCFS2_ORPHAN_NAMELEN, inode, - OCFS2_I(inode)->ip_blkno, - orphan_dir_bh, de_bh); + di = (struct ocfs2_dinode *)new_di_bh->b_data; + status = ocfs2_orphan_add(osb, handle, inode, new_di_bh, orphan_name, + &orphan_insert, orphan_dir); if (status < 0) { mlog_errno(status); goto leave; } - le32_add_cpu(&fe->i_flags, OCFS2_ORPHANED_FL); + /* get open lock so that only nodes can't remove it from orphan dir. */ + status = ocfs2_open_lock(inode); + if (status < 0) + mlog_errno(status); - /* Record which orphan dir our inode now resides - * in. delete_inode will use this to determine which orphan - * dir to lock. */ - spin_lock(&OCFS2_I(inode)->ip_lock); - OCFS2_I(inode)->ip_orphaned_slot = osb->slot_num; - spin_unlock(&OCFS2_I(inode)->ip_lock); + insert_inode_hash(inode); +leave: + if (status < 0 && did_quota_inode) + dquot_free_inode(inode); + if (handle) + ocfs2_commit_trans(osb, handle); - mlog(0, "Inode %"MLFu64" orphaned in slot %d\n", - OCFS2_I(inode)->ip_blkno, osb->slot_num); + if (orphan_dir) { + /* This was locked for us in ocfs2_prepare_orphan_dir() */ + ocfs2_inode_unlock(orphan_dir, 1); + mutex_unlock(&orphan_dir->i_mutex); + iput(orphan_dir); + } -leave: - if (orphan_dir_inode) - iput(orphan_dir_inode); + if ((status < 0) && inode) { + clear_nlink(inode); + iput(inode); + } + + if (inode_ac) + ocfs2_free_alloc_context(inode_ac); + + brelse(new_di_bh); + + if (!status) + *new_inode = inode; - if (orphan_dir_bh) - brelse(orphan_dir_bh); + ocfs2_free_dir_lookup_result(&orphan_insert); - mlog_exit(status); + ocfs2_inode_unlock(dir, 1); + brelse(parent_di_bh); return status; } -/* unlike orphan_add, we expect the orphan dir to already be locked here. */ -int ocfs2_orphan_del(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - struct inode *orphan_dir_inode, - struct inode *inode, - struct buffer_head *orphan_dir_bh) +int ocfs2_mv_orphaned_inode_to_new(struct inode *dir, + struct inode *inode, + struct dentry *dentry) { - char name[OCFS2_ORPHAN_NAMELEN + 1]; - struct ocfs2_dinode *orphan_fe; int status = 0; - struct buffer_head *target_de_bh = NULL; - struct ocfs2_dir_entry *target_de = NULL; + struct buffer_head *parent_di_bh = NULL; + handle_t *handle = NULL; + struct ocfs2_super *osb = OCFS2_SB(dir->i_sb); + struct ocfs2_dinode *dir_di, *di; + struct inode *orphan_dir_inode = NULL; + struct buffer_head *orphan_dir_bh = NULL; + struct buffer_head *di_bh = NULL; + struct ocfs2_dir_lookup_result lookup = { NULL, }; - mlog_entry_void(); + trace_ocfs2_mv_orphaned_inode_to_new(dir, dentry, + dentry->d_name.len, dentry->d_name.name, + (unsigned long long)OCFS2_I(dir)->ip_blkno, + (unsigned long long)OCFS2_I(inode)->ip_blkno); - status = ocfs2_blkno_stringify(OCFS2_I(inode)->ip_blkno, name); + status = ocfs2_inode_lock(dir, &parent_di_bh, 1); if (status < 0) { - mlog_errno(status); + if (status != -ENOENT) + mlog_errno(status); + return status; + } + + dir_di = (struct ocfs2_dinode *) parent_di_bh->b_data; + if (!dir_di->i_links_count) { + /* can't make a file in a deleted directory. */ + status = -ENOENT; goto leave; } - mlog(0, "removing '%s' from orphan dir %"MLFu64" (namelen=%d)\n", - name, OCFS2_I(orphan_dir_inode)->ip_blkno, OCFS2_ORPHAN_NAMELEN); + status = ocfs2_check_dir_for_entry(dir, dentry->d_name.name, + dentry->d_name.len); + if (status) + goto leave; - /* find it's spot in the orphan directory */ - target_de_bh = ocfs2_find_entry(name, OCFS2_ORPHAN_NAMELEN, - orphan_dir_inode, &target_de); - if (!target_de_bh) { - status = -ENOENT; + /* get a spot inside the dir. */ + status = ocfs2_prepare_dir_for_insert(osb, dir, parent_di_bh, + dentry->d_name.name, + dentry->d_name.len, &lookup); + if (status < 0) { mlog_errno(status); goto leave; } - /* remove it from the orphan directory */ - status = ocfs2_delete_entry(handle, orphan_dir_inode, target_de, - target_de_bh); - if (status < 0) { + orphan_dir_inode = ocfs2_get_system_file_inode(osb, + ORPHAN_DIR_SYSTEM_INODE, + osb->slot_num); + if (!orphan_dir_inode) { + status = -EEXIST; mlog_errno(status); goto leave; } - status = ocfs2_journal_access(handle,orphan_dir_inode, orphan_dir_bh, - OCFS2_JOURNAL_ACCESS_WRITE); + mutex_lock(&orphan_dir_inode->i_mutex); + + status = ocfs2_inode_lock(orphan_dir_inode, &orphan_dir_bh, 1); if (status < 0) { mlog_errno(status); + mutex_unlock(&orphan_dir_inode->i_mutex); + iput(orphan_dir_inode); goto leave; } - /* do the i_nlink dance! :) */ - orphan_fe = (struct ocfs2_dinode *) orphan_dir_bh->b_data; - if (S_ISDIR(inode->i_mode)) - le16_add_cpu(&orphan_fe->i_links_count, -1); - orphan_dir_inode->i_nlink = le16_to_cpu(orphan_fe->i_links_count); + status = ocfs2_read_inode_block(inode, &di_bh); + if (status < 0) { + mlog_errno(status); + goto orphan_unlock; + } - status = ocfs2_journal_dirty(handle, orphan_dir_bh); + handle = ocfs2_start_trans(osb, ocfs2_rename_credits(osb->sb)); + if (IS_ERR(handle)) { + status = PTR_ERR(handle); + handle = NULL; + mlog_errno(status); + goto orphan_unlock; + } + + status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), + di_bh, OCFS2_JOURNAL_ACCESS_WRITE); if (status < 0) { mlog_errno(status); - goto leave; + goto out_commit; } + status = ocfs2_orphan_del(osb, handle, orphan_dir_inode, inode, + orphan_dir_bh); + if (status < 0) { + mlog_errno(status); + goto out_commit; + } + + di = (struct ocfs2_dinode *)di_bh->b_data; + di->i_flags &= ~cpu_to_le32(OCFS2_ORPHANED_FL); + di->i_orphaned_slot = 0; + set_nlink(inode, 1); + ocfs2_set_links_count(di, inode->i_nlink); + ocfs2_update_inode_fsync_trans(handle, inode, 1); + ocfs2_journal_dirty(handle, di_bh); + + status = ocfs2_add_entry(handle, dentry, inode, + OCFS2_I(inode)->ip_blkno, parent_di_bh, + &lookup); + if (status < 0) { + mlog_errno(status); + goto out_commit; + } + + status = ocfs2_dentry_attach_lock(dentry, inode, + OCFS2_I(dir)->ip_blkno); + if (status) { + mlog_errno(status); + goto out_commit; + } + + d_instantiate(dentry, inode); + status = 0; +out_commit: + ocfs2_commit_trans(osb, handle); +orphan_unlock: + ocfs2_inode_unlock(orphan_dir_inode, 1); + mutex_unlock(&orphan_dir_inode->i_mutex); + iput(orphan_dir_inode); leave: - if (target_de_bh) - brelse(target_de_bh); - mlog_exit(status); + ocfs2_inode_unlock(dir, 1); + + brelse(di_bh); + brelse(parent_di_bh); + brelse(orphan_dir_bh); + + ocfs2_free_dir_lookup_result(&lookup); + + if (status) + mlog_errno(status); + return status; } -struct inode_operations ocfs2_dir_iops = { +const struct inode_operations ocfs2_dir_iops = { .create = ocfs2_create, .lookup = ocfs2_lookup, .link = ocfs2_link, @@ -2261,4 +2653,12 @@ struct inode_operations ocfs2_dir_iops = { .rename = ocfs2_rename, .setattr = ocfs2_setattr, .getattr = ocfs2_getattr, + .permission = ocfs2_permission, + .setxattr = generic_setxattr, + .getxattr = generic_getxattr, + .listxattr = ocfs2_listxattr, + .removexattr = generic_removexattr, + .fiemap = ocfs2_fiemap, + .get_acl = ocfs2_iop_get_acl, + .set_acl = ocfs2_iop_set_acl, }; |