/* * These are distinct checks because future versions of the file system will * want to have a trailing dirent structure independent of indexing.
*/ staticint ocfs2_supports_dir_trailer(struct inode *dir)
{ struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) return 0;
/* * "new' here refers to the point at which we're creating a new * directory via "mkdir()", but also when we're expanding an inline * directory. In either case, we don't yet have the indexing bit set * on the directory, so the standard checks will fail in when metaecc * is turned off. Only directory-initialization type functions should * use this then. Everything else wants ocfs2_supports_dir_trailer()
*/ staticint ocfs2_new_dir_wants_trailer(struct inode *dir)
{ struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
/* XXX ocfs2_block_dqtrailer() is similar but not quite - can we make
* them more consistent? */ struct ocfs2_dir_block_trailer *ocfs2_dir_trailer_from_size(int blocksize, void *data)
{ char *p = data;
/* * XXX: This is executed once on every dirent. We should consider optimizing * it.
*/ staticint ocfs2_skip_dir_trailer(struct inode *dir, struct ocfs2_dir_entry *de, unsignedlong offset, unsignedlong blklen)
{ unsignedlong toff = blklen - sizeof(struct ocfs2_dir_block_trailer);
trailer = ocfs2_trailer_from_bh(bh, inode->i_sb);
strcpy(trailer->db_signature, OCFS2_DIR_TRAILER_SIGNATURE);
trailer->db_compat_rec_len =
cpu_to_le16(sizeof(struct ocfs2_dir_block_trailer));
trailer->db_parent_dinode = cpu_to_le64(OCFS2_I(inode)->ip_blkno);
trailer->db_blkno = cpu_to_le64(bh->b_blocknr);
trailer->db_free_rec_len = cpu_to_le16(rec_len);
} /* * Link an unindexed block with a dir trailer structure into the index free * list. This function will modify dirdata_bh, but assumes you've already * passed it to the journal.
*/ staticint ocfs2_dx_dir_link_trailer(struct inode *dir, handle_t *handle, struct buffer_head *dx_root_bh, struct buffer_head *dirdata_bh)
{ int ret; struct ocfs2_dx_root_block *dx_root; struct ocfs2_dir_block_trailer *trailer;
staticvoid TEA_transform(__u32 buf[4], __u32 const in[])
{
__u32 sum = 0;
__u32 b0 = buf[0], b1 = buf[1];
__u32 a = in[0], b = in[1], c = in[2], d = in[3]; int n = 16;
do {
sum += DELTA;
b0 += ((b1 << 4)+a) ^ (b1+sum) ^ ((b1 >> 5)+b);
b1 += ((b0 << 4)+c) ^ (b0+sum) ^ ((b0 >> 5)+d);
} while (--n);
buf[0] += b0;
buf[1] += b1;
}
staticvoid str2hashbuf(constchar *msg, int len, __u32 *buf, int num)
{
__u32 pad, val; int i;
pad = (__u32)len | ((__u32)len << 8);
pad |= pad << 16;
val = pad; if (len > num*4)
len = num * 4; for (i = 0; i < len; i++) { if ((i % 4) == 0)
val = pad;
val = msg[i] + (val << 8); if ((i % 4) == 3) {
*buf++ = val;
val = pad;
num--;
}
} if (--num >= 0)
*buf++ = val; while (--num >= 0)
*buf++ = pad;
}
/* * XXX: Is this really necessary, if the index is never looked * at by readdir? Is a hash value of '0' a bad idea?
*/ if ((len == 1 && !strncmp(".", name, 1)) ||
(len == 2 && !strncmp("..", name, 2))) {
buf[0] = buf[1] = 0; goto out;
}
#ifdef OCFS2_DEBUG_DX_DIRS /* * This makes it very easy to debug indexing problems. We * should never allow this to be selected without hand editing * this file though.
*/
buf[0] = buf[1] = len; goto out; #endif
memcpy(buf, osb->osb_dx_seed, sizeof(buf));
p = name; while (len > 0) {
str2hashbuf(p, len, in, 4);
TEA_transform(buf, in);
len -= 16;
p += 16;
}
/* * bh passed here can be an inode block or a dir data block, depending * on the inode inline data flag.
*/ staticint ocfs2_check_dir_entry(struct inode *dir, struct ocfs2_dir_entry *de, struct buffer_head *bh, char *buf, unsignedint size, unsignedlong offset)
{ constchar *error_msg = NULL; constint rlen = le16_to_cpu(de->rec_len); constunsignedlong next_offset = ((char *) de - buf) + rlen;
if (unlikely(rlen < OCFS2_DIR_REC_LEN(1)))
error_msg = "rec_len is smaller than minimal"; elseif (unlikely(rlen % 4 != 0))
error_msg = "rec_len % 4 != 0"; elseif (unlikely(rlen < OCFS2_DIR_REC_LEN(de->name_len)))
error_msg = "rec_len is too small for name_len"; elseif (unlikely(next_offset > size))
error_msg = "directory entry overrun"; elseif (unlikely(next_offset > size - OCFS2_DIR_REC_LEN(1)) &&
next_offset != size)
error_msg = "directory entry too close to end";
/* * Returns 0 if not found, -1 on failure, and 1 on success
*/ staticinlineint ocfs2_search_dirblock(struct buffer_head *bh, struct inode *dir, constchar *name, int namelen, unsignedlong offset, char *first_de, unsignedint bytes, struct ocfs2_dir_entry **res_dir)
{ struct ocfs2_dir_entry *de; char *dlimit, *de_buf; int de_len; int ret = 0;
de_buf = first_de;
dlimit = de_buf + bytes;
while (de_buf < dlimit - OCFS2_DIR_MEMBER_LEN) { /* this code is executed quadratically often */ /* do minimal checking `by hand' */
de = (struct ocfs2_dir_entry *) de_buf;
if (de->name + 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, first_de,
bytes, 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;
}
/* * We don't validate dirents here, that's handled * in-place when the code walks them.
*/
trace_ocfs2_validate_dir_block((unsignedlonglong)bh->b_blocknr);
BUG_ON(!buffer_uptodate(bh));
/* * If the ecc fails, we return the error but otherwise * leave the filesystem running. We know any error is * local to this block. * * Note that we are safe to call this even if the directory * doesn't have a trailer. Filesystems without metaecc will do * nothing, and filesystems with it will have one.
*/
rc = ocfs2_validate_meta_ecc(sb, bh->b_data, &trailer->db_check); if (rc)
mlog(ML_ERROR, "Checksum failed for dinode %llu\n",
(unsignedlonglong)bh->b_blocknr);
return rc;
}
/* * Validate a directory trailer. * * We check the trailer here rather than in ocfs2_validate_dir_block() * because that function doesn't have the inode to test.
*/ staticint ocfs2_check_dir_trailer(struct inode *dir, struct buffer_head *bh)
{ int rc = 0; struct ocfs2_dir_block_trailer *trailer;
trailer = ocfs2_trailer_from_bh(bh, dir->i_sb); if (!OCFS2_IS_VALID_DIR_TRAILER(trailer)) {
rc = ocfs2_error(dir->i_sb, "Invalid dirblock #%llu: signature = %.*s\n",
(unsignedlonglong)bh->b_blocknr, 7,
trailer->db_signature); goto out;
} if (le64_to_cpu(trailer->db_blkno) != bh->b_blocknr) {
rc = ocfs2_error(dir->i_sb, "Directory block #%llu has an invalid db_blkno of %llu\n",
(unsignedlonglong)bh->b_blocknr,
(unsignedlonglong)le64_to_cpu(trailer->db_blkno)); goto out;
} if (le64_to_cpu(trailer->db_parent_dinode) !=
OCFS2_I(dir)->ip_blkno) {
rc = ocfs2_error(dir->i_sb, "Directory block #%llu on dinode #%llu has an invalid parent_dinode of %llu\n",
(unsignedlonglong)bh->b_blocknr,
(unsignedlonglong)OCFS2_I(dir)->ip_blkno,
(unsignedlonglong)le64_to_cpu(trailer->db_blkno)); goto out;
}
out: return rc;
}
/* * This function forces all errors to -EIO for consistency with its * predecessor, ocfs2_bread(). We haven't audited what returning the * real error codes would do to callers. We log the real codes with * mlog_errno() before we squash them.
*/ staticint ocfs2_read_dir_block(struct inode *inode, u64 v_block, struct buffer_head **bh, int flags)
{ int rc = 0; struct buffer_head *tmp = *bh;
if (!(flags & OCFS2_BH_READAHEAD) &&
ocfs2_supports_dir_trailer(inode)) {
rc = ocfs2_check_dir_trailer(inode, tmp); if (rc) { if (!*bh)
brelse(tmp);
mlog_errno(rc); goto out;
}
}
/* If ocfs2_read_virt_blocks() got us a new bh, pass it up. */ if (!*bh)
*bh = tmp;
out: return rc ? -EIO : 0;
}
/* * Read the block at 'phys' which belongs to this directory * inode. This function does no virtual->physical block translation - * what's passed in is assumed to be a valid directory block.
*/ staticint ocfs2_read_dir_block_direct(struct inode *dir, u64 phys, struct buffer_head **bh)
{ int ret; struct buffer_head *tmp = *bh;
ret = ocfs2_read_block(INODE_CACHE(dir), phys, &tmp,
ocfs2_validate_dir_block); if (ret) {
mlog_errno(ret); goto out;
}
if (ocfs2_supports_dir_trailer(dir)) {
ret = ocfs2_check_dir_trailer(dir, tmp); if (ret) { if (!*bh)
brelse(tmp);
mlog_errno(ret); goto out;
}
}
ret = ocfs2_validate_meta_ecc(sb, bh->b_data, &dx_root->dr_check); if (ret) {
mlog(ML_ERROR, "Checksum failed for dir index root block %llu\n",
(unsignedlonglong)bh->b_blocknr); return ret;
}
if (!OCFS2_IS_VALID_DX_ROOT(dx_root)) {
ret = ocfs2_error(sb, "Dir Index Root # %llu has bad signature %.*s\n",
(unsignedlonglong)le64_to_cpu(dx_root->dr_blkno),
7, dx_root->dr_signature);
}
ret = ocfs2_validate_meta_ecc(sb, bh->b_data, &dx_leaf->dl_check); if (ret) {
mlog(ML_ERROR, "Checksum failed for dir index leaf block %llu\n",
(unsignedlonglong)bh->b_blocknr); return ret;
}
if (!OCFS2_IS_VALID_DX_LEAF(dx_leaf)) {
ret = ocfs2_error(sb, "Dir Index Leaf has bad signature %.*s\n",
7, dx_leaf->dl_signature);
}
ret = ocfs2_read_block(INODE_CACHE(dir), blkno, &tmp,
ocfs2_validate_dx_leaf);
/* If ocfs2_read_block() got us a new bh, pass it up. */ if (!ret && !*dx_leaf_bh)
*dx_leaf_bh = tmp;
return ret;
}
/* * Read a series of dx_leaf blocks. This expects all buffer_head * pointers to be NULL on function entry.
*/ staticint ocfs2_read_dx_leaves(struct inode *dir, u64 start, int num, struct buffer_head **dx_leaf_bhs)
{ int ret;
ret = ocfs2_read_blocks(INODE_CACHE(dir), start, num, dx_leaf_bhs, 0,
ocfs2_validate_dx_leaf); if (ret)
mlog_errno(ret);
return ret;
}
staticstruct buffer_head *ocfs2_find_entry_el(constchar *name, int namelen, struct inode *dir, struct ocfs2_dir_entry **res_dir)
{ struct super_block *sb; struct buffer_head *bh_use[NAMEI_RA_SIZE]; struct buffer_head *bh, *ret = NULL; unsignedlong 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;
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++;
bh = NULL;
ocfs2_read_dir_block(dir, b++, &bh,
OCFS2_BH_READAHEAD);
bh_use[ra_max] = bh;
}
} if ((bh = bh_use[ra_ptr++]) == NULL) goto next; if (ocfs2_read_dir_block(dir, block, &bh, 0)) { /* read error, skip block & hope for the best.
* ocfs2_read_dir_block() has released the bh. */
mlog(ML_ERROR, "reading directory %llu, " "offset %lu\n",
(unsignedlonglong)OCFS2_I(dir)->ip_blkno,
block); goto next;
}
i = ocfs2_search_dirblock(bh, dir, name, namelen,
block << sb->s_blocksize_bits,
bh->b_data, sb->s_blocksize,
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);
/* * If the directory has grown while we were searching, then * search the last part of the directory before giving up.
*/
block = nblocks;
nblocks = i_size_read(dir) >> sb->s_blocksize_bits; if (block < nblocks) {
start = 0; goto restart;
}
cleanup_and_exit: /* Clean up the read-ahead blocks */ for (; ra_ptr < ra_max; ra_ptr++)
brelse(bh_use[ra_ptr]);
if (el->l_tree_depth) {
ret = ocfs2_find_leaf(INODE_CACHE(inode), el, major_hash,
&eb_bh); if (ret) {
mlog_errno(ret); goto out;
}
eb = (struct ocfs2_extent_block *) eb_bh->b_data;
el = &eb->h_list;
if (el->l_tree_depth) {
ret = ocfs2_error(inode->i_sb, "Inode %lu has non zero tree depth in btree tree block %llu\n",
inode->i_ino,
(unsignedlonglong)eb_bh->b_blocknr); goto out;
}
}
if (le16_to_cpu(el->l_next_free_rec) == 0) {
ret = ocfs2_error(inode->i_sb, "Inode %lu has empty extent list at depth %u\n",
inode->i_ino,
le16_to_cpu(el->l_tree_depth)); goto out;
}
found = 0; for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
rec = &el->l_recs[i];
if (le32_to_cpu(rec->e_cpos) <= major_hash) {
found = 1; break;
}
}
if (!found) {
ret = ocfs2_error(inode->i_sb, "Inode %lu has bad extent record (%u, %u, 0) in btree\n",
inode->i_ino,
le32_to_cpu(rec->e_cpos),
ocfs2_rec_clusters(el, rec)); goto out;
}
if (ret_phys_blkno)
*ret_phys_blkno = le64_to_cpu(rec->e_blkno); if (ret_cpos)
*ret_cpos = le32_to_cpu(rec->e_cpos); if (ret_clen)
*ret_clen = le16_to_cpu(rec->e_leaf_clusters);
out:
brelse(eb_bh); return ret;
}
/* * Returns the block index, from the start of the cluster which this * hash belongs too.
*/ staticinlineunsignedint __ocfs2_dx_dir_hash_idx(struct ocfs2_super *osb,
u32 minor_hash)
{ return minor_hash & osb->osb_dx_mask;
}
/* * We now have the cluster which should hold our entry. To * find the exact block from the start of the cluster to * search, we take the lower bits of the hash.
*/
blkno += ocfs2_dx_dir_hash_idx(OCFS2_SB(inode->i_sb), hinfo);
if (ret_phys_blkno)
*ret_phys_blkno = blkno; if (ret_cpos)
*ret_cpos = cpos;
search: /* * Empty leaf is legal, so no need to check for that.
*/
found = 0; for (i = 0; i < le16_to_cpu(entry_list->de_num_used); i++) {
dx_entry = &entry_list->de_entries[i];
if (hinfo->major_hash != le32_to_cpu(dx_entry->dx_major_hash)
|| hinfo->minor_hash != le32_to_cpu(dx_entry->dx_minor_hash)) continue;
/* * Search unindexed leaf block now. We're not * guaranteed to find anything.
*/
ret = ocfs2_read_dir_block_direct(dir,
le64_to_cpu(dx_entry->dx_dirent_blk),
&dir_ent_bh); if (ret) {
mlog_errno(ret); goto out;
}
/* * XXX: We should check the unindexed block here, * before using it.
*/
found = ocfs2_search_dirblock(dir_ent_bh, dir, name, namelen,
0, dir_ent_bh->b_data,
dir->i_sb->s_blocksize, &dir_ent); if (found == 1) break;
if (found == -1) { /* This means we found a bad directory entry. */
ret = -EIO;
mlog_errno(ret); goto out;
}
/* * Try to find an entry of the provided name within 'dir'. * * If nothing was found, -ENOENT is returned. Otherwise, zero is * returned and the struct 'res' will contain information useful to * other directory manipulation functions. * * Caller can NOT assume anything about the contents of the * buffer_heads - they are passed back only so that it can be passed * into any one of the manipulation functions (add entry, delete * entry, etc). As an example, bh in the extent directory case is a * data block, in the inline-data case it actually points to an inode, * in the indexed directory case, multiple buffers are involved.
*/ int ocfs2_find_entry(constchar *name, int namelen, struct inode *dir, struct ocfs2_dir_lookup_result *lookup)
{ struct buffer_head *bh; struct ocfs2_dir_entry *res_dir = NULL; int ret = 0;
if (ocfs2_dir_indexed(dir)) return ocfs2_find_entry_dx(name, namelen, dir, lookup);
if (unlikely(i_size_read(dir) <= 0)) {
ret = -EFSCORRUPTED;
mlog_errno(ret); goto out;
} /* * The unindexed dir code only uses part of the lookup * structure, so there's no reason to push it down further * than this.
*/ if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) { if (unlikely(i_size_read(dir) > dir->i_sb->s_blocksize)) {
ret = -EFSCORRUPTED;
mlog_errno(ret); goto out;
}
bh = ocfs2_find_entry_id(name, namelen, dir, &res_dir);
} else {
bh = ocfs2_find_entry_el(name, namelen, dir, &res_dir);
}
/* * Update inode number and type of a previously found directory entry.
*/ int ocfs2_update_entry(struct inode *dir, handle_t *handle, struct ocfs2_dir_lookup_result *res, struct inode *new_entry_inode)
{ int ret;
ocfs2_journal_access_func access = ocfs2_journal_access_db; struct ocfs2_dir_entry *de = res->dl_entry; struct buffer_head *de_bh = res->dl_leaf_bh;
/* * The same code works fine for both inline-data and extent * based directories, so no need to split this up. The only * difference is the journal_access function.
*/
if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
access = ocfs2_journal_access_di;
ret = access(handle, INODE_CACHE(dir), de_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret) {
mlog_errno(ret); goto out;
}
/* * This function gets a bit messy because we might have to * modify the root block, regardless of whether the indexed * entries are stored inline.
*/
/* * *Only* set 'entry_list' here, based on where we're looking * for the indexed entries. Later, we might still want to * journal both blocks, based on free list state.
*/
dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data; if (ocfs2_dx_root_inline(dx_root)) {
entry_list = &dx_root->dr_entries;
} else {
dx_leaf = (struct ocfs2_dx_leaf *) lookup->dl_dx_leaf_bh->b_data;
entry_list = &dx_leaf->dl_list;
}
/* Neither of these are a disk corruption - that should have
* been caught by lookup, before we got here. */
BUG_ON(le16_to_cpu(entry_list->de_count) <= 0);
BUG_ON(le16_to_cpu(entry_list->de_num_used) <= 0);
index = (char *)dx_entry - (char *)entry_list->de_entries;
index /= sizeof(*dx_entry);
/* * We know that removal of this dirent will leave enough room * for a new one, so add this block to the free list if it * isn't already there.
*/
trailer = ocfs2_trailer_from_bh(leaf_bh, dir->i_sb); if (trailer->db_free_rec_len == 0)
add_to_free_list = 1;
/* * Add the block holding our index into the journal before * removing the unindexed entry. If we get an error return * from __ocfs2_delete_entry(), then it hasn't removed the * entry yet. Likewise, successful return means we *must* * remove the indexed entry. * * We're also careful to journal the root tree block here as * the entry count needs to be updated. Also, we might be * adding to the start of the free list.
*/
ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret) {
mlog_errno(ret); goto out;
}
if (!ocfs2_dx_root_inline(dx_root)) {
ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir),
lookup->dl_dx_leaf_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret) {
mlog_errno(ret); goto out;
}
}
/* * Delete a directory entry. Hide the details of directory * implementation from the caller.
*/ int ocfs2_delete_entry(handle_t *handle, struct inode *dir, struct ocfs2_dir_lookup_result *res)
{ if (ocfs2_dir_indexed(dir)) return ocfs2_delete_entry_dx(handle, dir, res);
if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) return ocfs2_delete_entry_id(handle, dir, res->dl_entry,
res->dl_leaf_bh);
/* * Check whether 'de' has enough room to hold an entry of * 'new_rec_len' bytes.
*/ staticinlineint ocfs2_dirent_would_fit(struct ocfs2_dir_entry *de, unsignedint new_rec_len)
{ unsignedint de_really_used;
/* Check whether this is an empty record with enough space */ if (le64_to_cpu(de->inode) == 0 &&
le16_to_cpu(de->rec_len) >= new_rec_len) return 1;
/* * Record might have free space at the end which we can * use.
*/
de_really_used = OCFS2_DIR_REC_LEN(de->name_len); if (le16_to_cpu(de->rec_len) >= (de_really_used + new_rec_len)) return 1;
return 0;
}
staticvoid ocfs2_dx_dir_leaf_insert_tail(struct ocfs2_dx_leaf *dx_leaf, struct ocfs2_dx_entry *dx_new_entry)
{ int i;
i = le16_to_cpu(dx_leaf->dl_list.de_num_used);
dx_leaf->dl_list.de_entries[i] = *dx_new_entry;
/* * This expects that a journal write has been reserved on * lookup->dl_prev_leaf_bh or lookup->dl_dx_root_bh
*/ staticvoid ocfs2_recalc_free_list(struct inode *dir, handle_t *handle, struct ocfs2_dir_lookup_result *lookup)
{ int max_rec_len; struct ocfs2_dir_block_trailer *trailer;
/* Walk dl_leaf_bh to figure out what the new free rec_len is. */
max_rec_len = ocfs2_find_max_rec_len(dir->i_sb, lookup->dl_leaf_bh); if (max_rec_len) { /* * There's still room in this block, so no need to remove it * from the free list. In this case, we just want to update * the rec len accounting.
*/
trailer = ocfs2_trailer_from_bh(lookup->dl_leaf_bh, dir->i_sb);
trailer->db_free_rec_len = cpu_to_le16(max_rec_len);
ocfs2_journal_dirty(handle, lookup->dl_leaf_bh);
} else {
ocfs2_remove_block_from_free_list(dir, handle, lookup);
}
}
/* we don't always have a dentry for what we want to add, so people * like orphan dir can call this instead. * * The lookup context must have been filled from * ocfs2_prepare_dir_for_insert.
*/ int __ocfs2_add_entry(handle_t *handle, struct inode *dir, constchar *name, int namelen, struct inode *inode, u64 blkno, struct buffer_head *parent_fe_bh, struct ocfs2_dir_lookup_result *lookup)
{ unsignedlong offset; unsignedshort rec_len; struct ocfs2_dir_entry *de, *de1; struct ocfs2_dinode *di = (struct ocfs2_dinode *)parent_fe_bh->b_data; struct super_block *sb = dir->i_sb; int retval; unsignedint size = sb->s_blocksize; struct buffer_head *insert_bh = lookup->dl_leaf_bh; char *data_start = insert_bh->b_data;
if (ocfs2_dir_indexed(dir)) { struct buffer_head *bh;
/* * An indexed dir may require that we update the free space * list. Reserve a write to the previous node in the list so * that we don't fail later. * * XXX: This can be either a dx_root_block, or an unindexed * directory tree leaf block.
*/ if (ocfs2_free_list_at_root(lookup)) {
bh = lookup->dl_dx_root_bh;
retval = ocfs2_journal_access_dr(handle,
INODE_CACHE(dir), bh,
OCFS2_JOURNAL_ACCESS_WRITE);
} else {
bh = lookup->dl_prev_leaf_bh;
retval = ocfs2_journal_access_db(handle,
INODE_CACHE(dir), bh,
OCFS2_JOURNAL_ACCESS_WRITE);
} if (retval) {
mlog_errno(retval); return retval;
}
} elseif (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
data_start = di->id2.i_data.id_data;
size = i_size_read(dir);
/* 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, data_start,
size, offset)) {
retval = -ENOENT; goto bail;
} if (ocfs2_match(namelen, name, de)) {
retval = -EEXIST; goto bail;
}
/* We're guaranteed that we should have space, so we
* can't possibly have hit the trailer...right? */
mlog_bug_on_msg(ocfs2_skip_dir_trailer(dir, de, offset, size), "Hit dir trailer trying to insert %.*s " "(namelen %d) into directory %llu. " "offset is %lu, trailer offset is %d\n",
namelen, name, namelen,
(unsignedlonglong)parent_fe_bh->b_blocknr,
offset, ocfs2_dir_trailer_blk_off(dir->i_sb));
if (ocfs2_dirent_would_fit(de, rec_len)) {
inode_set_mtime_to_ts(dir,
inode_set_ctime_current(dir));
retval = ocfs2_mark_inode_dirty(handle, dir, parent_fe_bh); if (retval < 0) {
mlog_errno(retval); goto bail;
}
ret = ocfs2_read_inode_block(inode, &di_bh); if (ret) {
mlog(ML_ERROR, "Unable to read inode block for dir %llu\n",
(unsignedlonglong)OCFS2_I(inode)->ip_blkno); goto out;
}
di = (struct ocfs2_dinode *)di_bh->b_data;
data = &di->id2.i_data;
while (ctx->pos < i_size_read(inode)) { /* If the dir block has changed since the last call to * readdir(2), then we might be pointing to an invalid * dirent right now. Scan from the start of the block
* to make sure. */ if (!inode_eq_iversion(inode, *f_version)) { for (i = 0; i < i_size_read(inode) && i < offset; ) {
de = (struct ocfs2_dir_entry *)
(data->id_data + i); /* It's too expensive to do a full * dirent test each time round this * loop, but we do have to test at * least that it is non-zero. A * failure will be detected in the
* dirent test below. */ if (le16_to_cpu(de->rec_len) <
OCFS2_DIR_REC_LEN(1)) break;
i += le16_to_cpu(de->rec_len);
}
ctx->pos = offset = i;
*f_version = inode_query_iversion(inode);
}
de = (struct ocfs2_dir_entry *) (data->id_data + ctx->pos); if (!ocfs2_check_dir_entry(inode, de, di_bh, (char *)data->id_data,
i_size_read(inode), ctx->pos)) { /* On error, skip the f_pos to the end. */
ctx->pos = i_size_read(inode); break;
}
offset += le16_to_cpu(de->rec_len); if (le64_to_cpu(de->inode)) { if (!dir_emit(ctx, de->name, de->name_len,
le64_to_cpu(de->inode),
fs_ftype_to_dtype(de->file_type))) goto out;
}
ctx->pos += le16_to_cpu(de->rec_len);
}
out:
brelse(di_bh); return 0;
}
/* * NOTE: This function can be called against unindexed directories, * and indexed ones.
*/ staticint ocfs2_dir_foreach_blk_el(struct inode *inode,
u64 *f_version, struct dir_context *ctx, bool persist)
{ unsignedlong offset, blk, last_ra_blk = 0; int i; struct buffer_head * bh, * tmp; struct ocfs2_dir_entry * de; struct super_block * sb = inode->i_sb; unsignedint ra_sectors = 16; int stored = 0;
bh = NULL;
offset = ctx->pos & (sb->s_blocksize - 1);
while (ctx->pos < i_size_read(inode)) {
blk = ctx->pos >> sb->s_blocksize_bits; if (ocfs2_read_dir_block(inode, blk, &bh, 0)) { /* Skip the corrupt dirblock and keep trying */
ctx->pos += sb->s_blocksize - offset; continue;
}
/* The idea here is to begin with 8k read-ahead and to stay * 4k ahead of our current position. * * TODO: Use the pagecache for this. We just need to
* make sure it's cluster-safe... */ if (!last_ra_blk
|| (((last_ra_blk - blk) << 9) <= (ra_sectors / 2))) { for (i = ra_sectors >> (sb->s_blocksize_bits - 9);
i > 0; i--) {
tmp = NULL; if (!ocfs2_read_dir_block(inode, ++blk, &tmp,
OCFS2_BH_READAHEAD))
brelse(tmp);
}
last_ra_blk = blk;
ra_sectors = 8;
}
/* If the dir block has changed since the last call to * readdir(2), then we might be pointing to an invalid * dirent right now. Scan from the start of the block
* to make sure. */ if (!inode_eq_iversion(inode, *f_version)) { for (i = 0; i < sb->s_blocksize && i < offset; ) {
de = (struct ocfs2_dir_entry *) (bh->b_data + i); /* It's too expensive to do a full * dirent test each time round this * loop, but we do have to test at * least that it is non-zero. A * failure will be detected in the
* dirent test below. */ if (le16_to_cpu(de->rec_len) <
OCFS2_DIR_REC_LEN(1)) break;
i += le16_to_cpu(de->rec_len);
}
offset = i;
ctx->pos = (ctx->pos & ~(sb->s_blocksize - 1))
| offset;
*f_version = inode_query_iversion(inode);
}
while (ctx->pos < i_size_read(inode)
&& offset < sb->s_blocksize) {
de = (struct ocfs2_dir_entry *) (bh->b_data + offset); if (!ocfs2_check_dir_entry(inode, de, bh, bh->b_data,
sb->s_blocksize, offset)) { /* On error, skip the f_pos to the
next block. */
ctx->pos = (ctx->pos | (sb->s_blocksize - 1)) + 1; break;
} if (le64_to_cpu(de->inode)) { if (!dir_emit(ctx, de->name,
de->name_len,
le64_to_cpu(de->inode),
fs_ftype_to_dtype(de->file_type))) {
brelse(bh); return 0;
}
stored++;
}
offset += le16_to_cpu(de->rec_len);
ctx->pos += le16_to_cpu(de->rec_len);
}
offset = 0;
brelse(bh);
bh = NULL; if (!persist && stored) break;
} return 0;
}
/* * This is intended to be called from inside other kernel functions, * so we fake some arguments.
*/ int ocfs2_dir_foreach(struct inode *inode, struct dir_context *ctx)
{
u64 version = inode_query_iversion(inode);
ocfs2_dir_foreach_blk(inode, &version, ctx, true); return 0;
}
error = ocfs2_inode_lock_atime(inode, file->f_path.mnt, &lock_level, 1); if (lock_level && error >= 0) { /* We release EX lock which used to update atime * and get PR lock again to reduce contention
* on commonly accessed directories. */
ocfs2_inode_unlock(inode, 1);
lock_level = 0;
error = ocfs2_inode_lock(inode, NULL, 0);
} if (error < 0) { if (error != -ENOENT)
mlog_errno(error); /* we haven't got any yet, so propagate the error. */ goto bail_nolock;
}
ocfs2_inode_unlock(inode, lock_level); if (error)
mlog_errno(error);
bail_nolock:
return error;
}
/* * NOTE: this should always be called with parent dir i_rwsem taken.
*/ int ocfs2_find_files_on_disk(constchar *name, int namelen,
u64 *blkno, struct inode *inode, struct ocfs2_dir_lookup_result *lookup)
{ int status = -ENOENT;
status = ocfs2_find_entry(name, namelen, inode, lookup); if (status) goto leave;
*blkno = le64_to_cpu(lookup->dl_entry->inode);
status = 0;
leave:
return status;
}
/* * Convenience function for callers which just want the block number * mapped to a name and don't require the full dirent info, etc.
*/ int ocfs2_lookup_ino_from_name(struct inode *dir, constchar *name, int namelen, u64 *blkno)
{ int ret; struct ocfs2_dir_lookup_result lookup = { NULL, };
ret = ocfs2_find_files_on_disk(name, namelen, blkno, dir, &lookup);
ocfs2_free_dir_lookup_result(&lookup);
return ret;
}
/* Check for a name within a directory. * * Return 0 if the name does not exist * Return -EEXIST if the directory contains the name * Return -EFSCORRUPTED if found corruption * * Callers should have i_rwsem + a cluster lock on dir
*/ int ocfs2_check_dir_for_entry(struct inode *dir, constchar *name, int namelen)
{ int ret = 0; struct ocfs2_dir_lookup_result lookup = { NULL, };
/* * Check the positions of "." and ".." records to be sure * they're in the correct place. * * Indexed directories don't need to proceed past the first * two entries, so we end the scan after seeing '..'. Despite * that, we allow the scan to proceed In the event that we * have a corrupted indexed directory (no dot or dot dot * entries). This allows us to double check for existing * entries which might not have been found in the index.
*/ if (name_len == 1 && !strncmp(".", name, 1) && pos == 0) {
p->seen_dot = 1; returntrue;
}
/* * routine to check that the specified directory is empty (for rmdir) * * Returns 1 if dir is empty, zero otherwise. * * XXX: This is a performance problem for unindexed directories.
*/ int ocfs2_empty_dir(struct inode *inode)
{ int ret; struct ocfs2_empty_dir_priv priv = {
.ctx.actor = ocfs2_empty_dir_filldir,
};
if (ocfs2_dir_indexed(inode)) {
ret = ocfs2_empty_dir_dx(inode, &priv); if (ret)
mlog_errno(ret); /* * We still run ocfs2_dir_foreach to get the checks * for "." and "..".
*/
}
ret = ocfs2_dir_foreach(inode, &priv.ctx); if (ret)
mlog_errno(ret);
if (!priv.seen_dot || !priv.seen_dot_dot) {
mlog(ML_ERROR, "bad directory (dir #%llu) - no `.' or `..'\n",
(unsignedlonglong)OCFS2_I(inode)->ip_blkno); /* * XXX: Is it really safe to allow an unlink to continue?
*/ return 1;
}
return !priv.seen_other;
}
/* * Fills "." and ".." dirents in a new directory block. Returns dirent for * "..", which might be used during creation of a directory with a trailing * header. It is otherwise safe to ignore the return code.
*/ staticstruct ocfs2_dir_entry *ocfs2_fill_initial_dirents(struct inode *inode, struct inode *parent, char *start, unsignedint size)
{ struct ocfs2_dir_entry *de = (struct ocfs2_dir_entry *)start;
/* * This works together with code in ocfs2_mknod_locked() which sets * the inline-data flag and initializes the inline-data section.
*/ staticint ocfs2_fill_new_dir_id(struct ocfs2_super *osb,
handle_t *handle, struct inode *parent, struct inode *inode, struct buffer_head *di_bh)
{ int ret; struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data; struct ocfs2_inline_data *data = &di->id2.i_data; unsignedint size = le16_to_cpu(data->id_count);
ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret) {
mlog_errno(ret); goto out;
}
status = ocfs2_journal_access_db(handle, INODE_CACHE(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 = ocfs2_fill_initial_dirents(inode, parent, new_bh->b_data, size); if (ocfs2_new_dir_wants_trailer(inode)) { int size = le16_to_cpu(de->rec_len);
/* * Figure out the size of the hole left over after * insertion of '.' and '..'. The trailer wants this * information.
*/
size -= OCFS2_DIR_REC_LEN(2);
size -= sizeof(struct ocfs2_dir_block_trailer);
/* * Allocates and formats a new cluster for use in an indexed dir * leaf. This version will not do the extent insert, so that it can be * used by operations which need careful ordering.
*/ staticint __ocfs2_dx_dir_new_cluster(struct inode *dir,
u32 cpos, handle_t *handle, struct ocfs2_alloc_context *data_ac, struct buffer_head **dx_leaves, int num_dx_leaves, u64 *ret_phys_blkno)
{ int ret;
u32 phys, num;
u64 phys_blkno; struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
/* * XXX: For create, this should claim cluster for the index * *before* the unindexed insert so that we have a better * chance of contiguousness as the directory grows in number * of entries.
*/
ret = __ocfs2_claim_clusters(handle, data_ac, 1, 1, &phys, &num); if (ret) {
mlog_errno(ret); goto out;
}
/* * Format the new cluster first. That way, we're inserting * valid data.
*/
phys_blkno = ocfs2_clusters_to_blocks(osb->sb, phys);
ret = ocfs2_dx_dir_format_cluster(osb, handle, dir, dx_leaves,
num_dx_leaves, phys_blkno); if (ret) {
mlog_errno(ret); goto out;
}
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.