mirror of
https://github.com/CloverHackyColor/CloverBootloader.git
synced 2024-11-22 11:25:42 +01:00
95be95c531
Clean warnings reactivated. Clean build flags in the 4 configurations to make RELEASE_XCODE8 works in QEMU.
875 lines
34 KiB
C
875 lines
34 KiB
C
/**
|
|
* \file fsw_reiserfs.c
|
|
* ReiserFS file system driver code.
|
|
*/
|
|
|
|
/*-
|
|
* Copyright (c) 2006 Christoph Pfisterer
|
|
*
|
|
* This program is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU General Public License
|
|
* as published by the Free Software Foundation; either version 2
|
|
* of the License, or (at your option) any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program; if not, write to the Free Software
|
|
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
|
*/
|
|
|
|
#include "fsw_reiserfs.h"
|
|
|
|
|
|
// functions
|
|
|
|
static fsw_status_t fsw_reiserfs_volume_mount(struct fsw_reiserfs_volume *vol);
|
|
static void fsw_reiserfs_volume_free(struct fsw_reiserfs_volume *vol);
|
|
static fsw_status_t fsw_reiserfs_volume_stat(struct fsw_reiserfs_volume *vol, struct fsw_volume_stat *sb);
|
|
|
|
static fsw_status_t fsw_reiserfs_dnode_fill(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno);
|
|
static void fsw_reiserfs_dnode_free(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno);
|
|
static fsw_status_t fsw_reiserfs_dnode_stat(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_dnode_stat_str *sb);
|
|
static fsw_status_t fsw_reiserfs_get_extent(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_extent *extent);
|
|
|
|
static fsw_status_t fsw_reiserfs_dir_lookup(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_string *lookup_name, struct fsw_reiserfs_dnode **child_dno);
|
|
static fsw_status_t fsw_reiserfs_dir_read(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_shandle *shand, struct fsw_reiserfs_dnode **child_dno);
|
|
|
|
static fsw_status_t fsw_reiserfs_readlink(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_string *link);
|
|
|
|
static fsw_status_t fsw_reiserfs_item_search(struct fsw_reiserfs_volume *vol,
|
|
fsw_u32 dir_id, fsw_u32 objectid, fsw_u64 offset,
|
|
struct fsw_reiserfs_item *item);
|
|
static fsw_status_t fsw_reiserfs_item_next(struct fsw_reiserfs_volume *vol,
|
|
struct fsw_reiserfs_item *item);
|
|
static void fsw_reiserfs_item_release(struct fsw_reiserfs_volume *vol,
|
|
struct fsw_reiserfs_item *item);
|
|
|
|
//
|
|
// Dispatch Table
|
|
//
|
|
|
|
struct fsw_fstype_table FSW_FSTYPE_TABLE_NAME(reiserfs) = {
|
|
{ FSW_STRING_TYPE_ISO88591, 8, 8, "reiserfs" },
|
|
sizeof(struct fsw_reiserfs_volume),
|
|
sizeof(struct fsw_reiserfs_dnode),
|
|
|
|
fsw_reiserfs_volume_mount,
|
|
fsw_reiserfs_volume_free,
|
|
fsw_reiserfs_volume_stat,
|
|
fsw_reiserfs_dnode_fill,
|
|
fsw_reiserfs_dnode_free,
|
|
fsw_reiserfs_dnode_stat,
|
|
fsw_reiserfs_get_extent,
|
|
fsw_reiserfs_dir_lookup,
|
|
fsw_reiserfs_dir_read,
|
|
fsw_reiserfs_readlink,
|
|
};
|
|
|
|
// misc data
|
|
|
|
static fsw_u32 superblock_offsets[3] = {
|
|
REISERFS_DISK_OFFSET_IN_BYTES >> REISERFS_SUPERBLOCK_BLOCKSIZEBITS,
|
|
REISERFS_OLD_DISK_OFFSET_IN_BYTES >> REISERFS_SUPERBLOCK_BLOCKSIZEBITS,
|
|
0
|
|
};
|
|
|
|
/**
|
|
* Mount an reiserfs volume. Reads the superblock and constructs the
|
|
* root directory dnode.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_volume_mount(struct fsw_reiserfs_volume *vol)
|
|
{
|
|
fsw_status_t status;
|
|
void *buffer;
|
|
fsw_u32 blocksize;
|
|
int i;
|
|
struct fsw_string s;
|
|
|
|
// allocate memory to keep the superblock around
|
|
status = fsw_alloc(sizeof(struct reiserfs_super_block), &vol->sb);
|
|
if (status)
|
|
return status;
|
|
|
|
// read the superblock into its buffer
|
|
fsw_set_blocksize(vol, REISERFS_SUPERBLOCK_BLOCKSIZE, REISERFS_SUPERBLOCK_BLOCKSIZE);
|
|
for (i = 0; superblock_offsets[i]; i++) {
|
|
status = fsw_block_get(vol, superblock_offsets[i], 0, &buffer);
|
|
if (status)
|
|
return status;
|
|
fsw_memcpy(vol->sb, buffer, sizeof(struct reiserfs_super_block));
|
|
fsw_block_release(vol, superblock_offsets[i], buffer);
|
|
|
|
// check for one of the magic strings
|
|
if (fsw_memeq(vol->sb->s_v1.s_magic,
|
|
REISERFS_SUPER_MAGIC_STRING, 8)) {
|
|
vol->version = REISERFS_VERSION_1;
|
|
break;
|
|
} else if (fsw_memeq(vol->sb->s_v1.s_magic,
|
|
REISER2FS_SUPER_MAGIC_STRING, 9)) {
|
|
vol->version = REISERFS_VERSION_2;
|
|
break;
|
|
} else if (fsw_memeq(vol->sb->s_v1.s_magic,
|
|
REISER2FS_JR_SUPER_MAGIC_STRING, 9)) {
|
|
vol->version = vol->sb->s_v1.s_version;
|
|
if (vol->version == REISERFS_VERSION_1 || vol->version == REISERFS_VERSION_2)
|
|
break;
|
|
}
|
|
}
|
|
if (superblock_offsets[i] == 0)
|
|
return FSW_UNSUPPORTED;
|
|
|
|
// check the superblock
|
|
if (vol->sb->s_v1.s_root_block == (__le32)-1) // unfinished 'reiserfsck --rebuild-tree'
|
|
return FSW_VOLUME_CORRUPTED;
|
|
|
|
/*
|
|
if (vol->sb->s_rev_level != EXT2_GOOD_OLD_REV &&
|
|
vol->sb->s_rev_level != EXT2_DYNAMIC_REV)
|
|
return FSW_UNSUPPORTED;
|
|
if (vol->sb->s_rev_level == EXT2_DYNAMIC_REV &&
|
|
(vol->sb->s_feature_incompat & ~(EXT2_FEATURE_INCOMPAT_FILETYPE | EXT3_FEATURE_INCOMPAT_RECOVER)))
|
|
return FSW_UNSUPPORTED;
|
|
*/
|
|
|
|
// set real blocksize
|
|
blocksize = vol->sb->s_v1.s_blocksize;
|
|
fsw_set_blocksize(vol, blocksize, blocksize);
|
|
|
|
// get other info from superblock
|
|
/*
|
|
vol->ind_bcnt = EXT2_ADDR_PER_BLOCK(vol->sb);
|
|
vol->dind_bcnt = vol->ind_bcnt * vol->ind_bcnt;
|
|
vol->inode_size = EXT2_INODE_SIZE(vol->sb);
|
|
*/
|
|
|
|
for (i = 0; i < 16; i++)
|
|
if (vol->sb->s_label[i] == 0)
|
|
break;
|
|
s.type = FSW_STRING_TYPE_ISO88591;
|
|
s.size = s.len = i;
|
|
s.data = vol->sb->s_label;
|
|
status = fsw_strdup_coerce(&vol->g.label, vol->g.host_string_type, &s);
|
|
if (status)
|
|
return status;
|
|
|
|
// setup the root dnode
|
|
status = fsw_dnode_create_root(vol, REISERFS_ROOT_OBJECTID, &vol->g.root);
|
|
if (status)
|
|
return status;
|
|
vol->g.root->dir_id = REISERFS_ROOT_PARENT_OBJECTID;
|
|
|
|
FSW_MSG_DEBUG((FSW_MSGSTR("fsw_reiserfs_volume_mount: success, blocksize %d tree height %d\n"),
|
|
blocksize, vol->sb->s_v1.s_tree_height));
|
|
|
|
return FSW_SUCCESS;
|
|
}
|
|
|
|
/**
|
|
* Free the volume data structure. Called by the core after an unmount or after
|
|
* an unsuccessful mount to release the memory used by the file system type specific
|
|
* part of the volume structure.
|
|
*/
|
|
|
|
static void fsw_reiserfs_volume_free(struct fsw_reiserfs_volume *vol)
|
|
{
|
|
if (vol->sb)
|
|
fsw_free(vol->sb);
|
|
}
|
|
|
|
/**
|
|
* Get in-depth information on a volume.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_volume_stat(struct fsw_reiserfs_volume *vol, struct fsw_volume_stat *sb)
|
|
{
|
|
sb->total_bytes = (fsw_u64)vol->sb->s_v1.s_block_count * vol->g.log_blocksize;
|
|
sb->free_bytes = (fsw_u64)vol->sb->s_v1.s_free_blocks * vol->g.log_blocksize;
|
|
return FSW_SUCCESS;
|
|
}
|
|
|
|
/**
|
|
* Get full information on a dnode from disk. This function is called by the core
|
|
* whenever it needs to access fields in the dnode structure that may not
|
|
* be filled immediately upon creation of the dnode. In the case of reiserfs, we
|
|
* delay fetching of the stat data until dnode_fill is called. The size and
|
|
* type fields are invalid until this function has been called.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_dnode_fill(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno)
|
|
{
|
|
fsw_status_t status;
|
|
fsw_u32 item_len, mode;
|
|
struct fsw_reiserfs_item item;
|
|
|
|
if (dno->sd_v1 || dno->sd_v2)
|
|
return FSW_SUCCESS;
|
|
|
|
FSW_MSG_DEBUG((FSW_MSGSTR("fsw_reiserfs_dnode_fill: object %d/%d\n"), dno->dir_id, dno->g.dnode_id));
|
|
|
|
// find stat data item in reiserfs tree
|
|
status = fsw_reiserfs_item_search(vol, dno->dir_id, dno->g.dnode_id, 0, &item);
|
|
if (status == FSW_NOT_FOUND) {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_dnode_fill: cannot find stat_data for object %d/%d\n"),
|
|
dno->dir_id, dno->g.dnode_id));
|
|
return FSW_VOLUME_CORRUPTED;
|
|
}
|
|
if (status)
|
|
return status;
|
|
if (item.item_offset != 0) {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_dnode_fill: got item that's not stat_data\n")));
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
return FSW_VOLUME_CORRUPTED;
|
|
}
|
|
item_len = item.ih.ih_item_len;
|
|
|
|
// get data in appropriate version
|
|
if (item.ih.ih_version == KEY_FORMAT_3_5 && item_len == SD_V1_SIZE) {
|
|
// have stat_data_v1 structure
|
|
status = fsw_memdup((void **)&dno->sd_v1, item.item_data, item_len);
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
if (status)
|
|
return status;
|
|
|
|
// get info from the inode
|
|
dno->g.size = dno->sd_v1->sd_size;
|
|
mode = dno->sd_v1->sd_mode;
|
|
|
|
} else if (item.ih.ih_version == KEY_FORMAT_3_6 && item_len == SD_V2_SIZE) {
|
|
// have stat_data_v2 structure
|
|
status = fsw_memdup((void **)&dno->sd_v2, item.item_data, item_len);
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
if (status)
|
|
return status;
|
|
|
|
// get info from the inode
|
|
dno->g.size = dno->sd_v2->sd_size;
|
|
mode = dno->sd_v2->sd_mode;
|
|
|
|
} else {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_dnode_fill: version %d(%d) and size %d(%d) not recognized for stat_data\n"),
|
|
item.ih.ih_version, KEY_FORMAT_3_6, item_len, SD_V2_SIZE));
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
return FSW_VOLUME_CORRUPTED;
|
|
}
|
|
|
|
// get node type from mode field
|
|
if (S_ISREG(mode))
|
|
dno->g.type = FSW_DNODE_TYPE_FILE;
|
|
else if (S_ISDIR(mode))
|
|
dno->g.type = FSW_DNODE_TYPE_DIR;
|
|
else if (S_ISLNK(mode))
|
|
dno->g.type = FSW_DNODE_TYPE_SYMLINK;
|
|
else
|
|
dno->g.type = FSW_DNODE_TYPE_SPECIAL;
|
|
|
|
return FSW_SUCCESS;
|
|
}
|
|
|
|
/**
|
|
* Free the dnode data structure. Called by the core when deallocating a dnode
|
|
* structure to release the memory used by the file system type specific part
|
|
* of the dnode structure.
|
|
*/
|
|
|
|
static void fsw_reiserfs_dnode_free(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno)
|
|
{
|
|
if (dno->sd_v1)
|
|
fsw_free(dno->sd_v1);
|
|
if (dno->sd_v2)
|
|
fsw_free(dno->sd_v2);
|
|
}
|
|
|
|
/**
|
|
* Get in-depth information on a dnode. The core makes sure that fsw_reiserfs_dnode_fill
|
|
* has been called on the dnode before this function is called. Note that some
|
|
* data is not directly stored into the structure, but passed to a host-specific
|
|
* callback that converts it to the host-specific format.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_dnode_stat(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_dnode_stat_str *sb)
|
|
{
|
|
if (dno->sd_v1) {
|
|
if (dno->g.type == FSW_DNODE_TYPE_SPECIAL)
|
|
sb->used_bytes = 0;
|
|
else
|
|
sb->used_bytes = dno->sd_v1->u.sd_blocks * vol->g.log_blocksize;
|
|
sb->store_time_posix(sb, FSW_DNODE_STAT_CTIME, dno->sd_v1->sd_ctime);
|
|
sb->store_time_posix(sb, FSW_DNODE_STAT_ATIME, dno->sd_v1->sd_atime);
|
|
sb->store_time_posix(sb, FSW_DNODE_STAT_MTIME, dno->sd_v1->sd_mtime);
|
|
sb->store_attr_posix(sb, dno->sd_v1->sd_mode);
|
|
} else if (dno->sd_v2) {
|
|
sb->used_bytes = dno->sd_v2->sd_blocks * vol->g.log_blocksize;
|
|
sb->store_time_posix(sb, FSW_DNODE_STAT_CTIME, dno->sd_v2->sd_ctime);
|
|
sb->store_time_posix(sb, FSW_DNODE_STAT_ATIME, dno->sd_v2->sd_atime);
|
|
sb->store_time_posix(sb, FSW_DNODE_STAT_MTIME, dno->sd_v2->sd_mtime);
|
|
sb->store_attr_posix(sb, dno->sd_v2->sd_mode);
|
|
}
|
|
|
|
return FSW_SUCCESS;
|
|
}
|
|
|
|
/**
|
|
* Retrieve file data mapping information. This function is called by the core when
|
|
* fsw_shandle_read needs to know where on the disk the required piece of the file's
|
|
* data can be found. The core makes sure that fsw_reiserfs_dnode_fill has been called
|
|
* on the dnode before. Our task here is to get the physical disk block number for
|
|
* the requested logical block number.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_get_extent(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_extent *extent)
|
|
{
|
|
fsw_status_t status;
|
|
fsw_u64 search_offset, intra_offset;
|
|
struct fsw_reiserfs_item item;
|
|
fsw_u32 intra_bno, nr_item;
|
|
|
|
// Preconditions: The caller has checked that the requested logical block
|
|
// is within the file's size. The dnode has complete information, i.e.
|
|
// fsw_reiserfs_dnode_read_info was called successfully on it.
|
|
|
|
FSW_MSG_DEBUG((FSW_MSGSTR("fsw_reiserfs_get_extent: mapping block %d of object %d/%d\n"),
|
|
extent->log_start, dno->dir_id, dno->g.dnode_id));
|
|
|
|
extent->type = FSW_EXTENT_TYPE_SPARSE;
|
|
extent->log_count = 1;
|
|
|
|
// get the item for the requested block
|
|
search_offset = (fsw_u64)extent->log_start * vol->g.log_blocksize + 1;
|
|
status = fsw_reiserfs_item_search(vol, dno->dir_id, dno->g.dnode_id, search_offset, &item);
|
|
if (status)
|
|
return status;
|
|
if (item.item_offset == 0) {
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
return FSW_SUCCESS; // no data items found, assume all-sparse file
|
|
}
|
|
intra_offset = search_offset - item.item_offset;
|
|
|
|
// check the kind of block
|
|
if (item.item_type == TYPE_INDIRECT || item.item_type == V1_INDIRECT_UNIQUENESS) {
|
|
// indirect item, contains block numbers
|
|
|
|
if (intra_offset & (vol->g.log_blocksize - 1)) {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_get_extent: intra_offset not block-aligned for indirect block\n")));
|
|
goto bail;
|
|
}
|
|
intra_bno = (fsw_u32)FSW_U64_DIV(intra_offset, vol->g.log_blocksize);
|
|
nr_item = item.ih.ih_item_len / sizeof(fsw_u32);
|
|
if (intra_bno >= nr_item) {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_get_extent: indirect block too small\n")));
|
|
goto bail;
|
|
}
|
|
extent->type = FSW_EXTENT_TYPE_PHYSBLOCK;
|
|
extent->phys_start = ((fsw_u32 *)item.item_data)[intra_bno];
|
|
|
|
// TODO: check if the following blocks can be aggregated into one extent
|
|
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
return FSW_SUCCESS;
|
|
|
|
} else if (item.item_type == TYPE_DIRECT || item.item_type == V1_DIRECT_UNIQUENESS) {
|
|
// direct item, contains file data
|
|
|
|
// TODO: Check if direct items always start on block boundaries. If not, we may have
|
|
// to do extra work here.
|
|
|
|
if (intra_offset != 0) {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_get_extent: intra_offset not aligned for direct block\n")));
|
|
goto bail;
|
|
}
|
|
|
|
extent->type = FSW_EXTENT_TYPE_BUFFER;
|
|
status = fsw_memdup(&extent->buffer, item.item_data, item.ih.ih_item_len);
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
if (status)
|
|
return status;
|
|
|
|
return FSW_SUCCESS;
|
|
|
|
}
|
|
|
|
bail:
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
return FSW_VOLUME_CORRUPTED;
|
|
|
|
/*
|
|
// check if the following blocks can be aggregated into one extent
|
|
file_bcnt = (fsw_u32)((dno->g.size + vol->g.log_blocksize - 1) & (vol->g.log_blocksize - 1));
|
|
while (path[i] + extent->log_count < buf_bcnt && // indirect block has more block pointers
|
|
extent->log_start + extent->log_count < file_bcnt) { // file has more blocks
|
|
if (buffer[path[i] + extent->log_count] == buffer[path[i] + extent->log_count - 1] + 1)
|
|
extent->log_count++;
|
|
else
|
|
break;
|
|
}
|
|
*/
|
|
}
|
|
|
|
/**
|
|
* Lookup a directory's child dnode by name. This function is called on a directory
|
|
* to retrieve the directory entry with the given name. A dnode is constructed for
|
|
* this entry and returned. The core makes sure that fsw_reiserfs_dnode_fill has been called
|
|
* and the dnode is actually a directory.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_dir_lookup(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_string *lookup_name, struct fsw_reiserfs_dnode **child_dno_out)
|
|
{
|
|
fsw_status_t status;
|
|
struct fsw_reiserfs_item item;
|
|
fsw_u32 nr_item, i, name_offset, next_name_offset, name_len;
|
|
fsw_u32 child_dir_id;
|
|
struct reiserfs_de_head *dhead;
|
|
struct fsw_string entry_name;
|
|
|
|
// Preconditions: The caller has checked that dno is a directory node.
|
|
|
|
// BIG TODOS: Use the hash function to start with the item containing the entry.
|
|
// Use binary search within the item.
|
|
|
|
entry_name.type = FSW_STRING_TYPE_ISO88591;
|
|
|
|
// get the item for that position
|
|
status = fsw_reiserfs_item_search(vol, dno->dir_id, dno->g.dnode_id, FIRST_ITEM_OFFSET, &item);
|
|
if (status)
|
|
return status;
|
|
if (item.item_offset == 0) {
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
return FSW_NOT_FOUND; // empty directory or something
|
|
}
|
|
|
|
for(;;) {
|
|
|
|
// search the directory item
|
|
dhead = (struct reiserfs_de_head *)item.item_data;
|
|
nr_item = item.ih.u.ih_entry_count;
|
|
next_name_offset = item.ih.ih_item_len;
|
|
for (i = 0; i < nr_item; i++, dhead++, next_name_offset = name_offset) {
|
|
// get the name
|
|
name_offset = dhead->deh_location;
|
|
name_len = next_name_offset - name_offset;
|
|
while (name_len > 0 && item.item_data[name_offset + name_len - 1] == 0)
|
|
name_len--;
|
|
|
|
entry_name.len = entry_name.size = name_len;
|
|
entry_name.data = item.item_data + name_offset;
|
|
|
|
// compare name
|
|
if (fsw_streq(lookup_name, &entry_name)) {
|
|
// found the entry we're looking for!
|
|
|
|
// setup a dnode for the child item
|
|
status = fsw_dnode_create(dno, dhead->deh_objectid, FSW_DNODE_TYPE_UNKNOWN, &entry_name, child_dno_out);
|
|
child_dir_id = dhead->deh_dir_id;
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
if (status)
|
|
return status;
|
|
(*child_dno_out)->dir_id = child_dir_id;
|
|
|
|
return FSW_SUCCESS;
|
|
}
|
|
}
|
|
|
|
// We didn't find the next directory entry in this item. Look for the next
|
|
// item of the directory.
|
|
|
|
status = fsw_reiserfs_item_next(vol, &item);
|
|
if (status)
|
|
return status;
|
|
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Get the next directory entry when reading a directory. This function is called during
|
|
* directory iteration to retrieve the next directory entry. A dnode is constructed for
|
|
* the entry and returned. The core makes sure that fsw_reiserfs_dnode_fill has been called
|
|
* and the dnode is actually a directory. The shandle provided by the caller is used to
|
|
* record the position in the directory between calls.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_dir_read(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_shandle *shand, struct fsw_reiserfs_dnode **child_dno_out)
|
|
{
|
|
fsw_status_t status;
|
|
struct fsw_reiserfs_item item;
|
|
fsw_u32 nr_item, i, name_offset, next_name_offset, name_len;
|
|
fsw_u32 child_dir_id;
|
|
struct reiserfs_de_head *dhead;
|
|
struct fsw_string entry_name;
|
|
|
|
// Preconditions: The caller has checked that dno is a directory node. The caller
|
|
// has opened a storage handle to the directory's storage and keeps it around between
|
|
// calls.
|
|
|
|
// BIG TODOS: Use binary search within the item.
|
|
|
|
// adjust pointer to first entry if necessary
|
|
if (shand->pos == 0)
|
|
shand->pos = FIRST_ITEM_OFFSET;
|
|
|
|
// get the item for that position
|
|
status = fsw_reiserfs_item_search(vol, dno->dir_id, dno->g.dnode_id, shand->pos, &item);
|
|
if (status)
|
|
return status;
|
|
if (item.item_offset == 0) {
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
return FSW_NOT_FOUND; // empty directory or something
|
|
}
|
|
|
|
for(;;) {
|
|
|
|
// search the directory item
|
|
dhead = (struct reiserfs_de_head *)item.item_data;
|
|
nr_item = item.ih.u.ih_entry_count;
|
|
for (i = 0; i < nr_item; i++, dhead++) {
|
|
if (dhead->deh_offset < shand->pos)
|
|
continue; // not yet past the last entry returned
|
|
if (dhead->deh_offset == DOT_OFFSET || dhead->deh_offset == DOT_DOT_OFFSET)
|
|
continue; // never report . or ..
|
|
|
|
// get the name
|
|
name_offset = dhead->deh_location;
|
|
if (i == 0)
|
|
next_name_offset = item.ih.ih_item_len;
|
|
else
|
|
next_name_offset = dhead[-1].deh_location;
|
|
name_len = next_name_offset - name_offset;
|
|
while (name_len > 0 && item.item_data[name_offset + name_len - 1] == 0)
|
|
name_len--;
|
|
|
|
entry_name.type = FSW_STRING_TYPE_ISO88591;
|
|
entry_name.len = entry_name.size = name_len;
|
|
entry_name.data = item.item_data + name_offset;
|
|
|
|
if (fsw_streq_cstr(&entry_name, ".reiserfs_priv"))
|
|
continue; // never report this special file
|
|
|
|
// found the next entry!
|
|
shand->pos = dhead->deh_offset + 1;
|
|
|
|
// setup a dnode for the child item
|
|
status = fsw_dnode_create(dno, dhead->deh_objectid, FSW_DNODE_TYPE_UNKNOWN, &entry_name, child_dno_out);
|
|
child_dir_id = dhead->deh_dir_id;
|
|
fsw_reiserfs_item_release(vol, &item);
|
|
if (status)
|
|
return status;
|
|
(*child_dno_out)->dir_id = child_dir_id;
|
|
|
|
return FSW_SUCCESS;
|
|
}
|
|
|
|
// We didn't find the next directory entry in this item. Look for the next
|
|
// item of the directory.
|
|
|
|
status = fsw_reiserfs_item_next(vol, &item);
|
|
if (status)
|
|
return status;
|
|
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Get the target path of a symbolic link. This function is called when a symbolic
|
|
* link needs to be resolved. The core makes sure that the fsw_reiserfs_dnode_fill has been
|
|
* called on the dnode and that it really is a symlink.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_readlink(struct fsw_reiserfs_volume *vol, struct fsw_reiserfs_dnode *dno,
|
|
struct fsw_string *link_target)
|
|
{
|
|
return fsw_dnode_readlink_data(dno, link_target);
|
|
}
|
|
|
|
/**
|
|
* Compare an on-disk tree key against the search key.
|
|
*/
|
|
|
|
static int fsw_reiserfs_compare_key(struct reiserfs_key *key,
|
|
fsw_u32 dir_id, fsw_u32 objectid, fsw_u64 offset)
|
|
{
|
|
fsw_u32 key_type;
|
|
fsw_u64 key_offset;
|
|
|
|
if (key->k_dir_id > dir_id)
|
|
return FIRST_GREATER;
|
|
if (key->k_dir_id < dir_id)
|
|
return SECOND_GREATER;
|
|
|
|
if (key->k_objectid > objectid)
|
|
return FIRST_GREATER;
|
|
if (key->k_objectid < objectid)
|
|
return SECOND_GREATER;
|
|
|
|
// determine format of the on-disk key
|
|
key_type = (fsw_u32)FSW_U64_SHR(key->u.k_offset_v2.v, 60);
|
|
if (key_type != TYPE_DIRECT && key_type != TYPE_INDIRECT && key_type != TYPE_DIRENTRY) {
|
|
// detected 3.5 format (_v1)
|
|
key_offset = key->u.k_offset_v1.k_offset;
|
|
} else {
|
|
// detected 3.6 format (_v2)
|
|
key_offset = key->u.k_offset_v2.v & (~0ULL >> 4);
|
|
}
|
|
if (key_offset > offset)
|
|
return FIRST_GREATER;
|
|
if (key_offset < offset)
|
|
return SECOND_GREATER;
|
|
return KEYS_IDENTICAL;
|
|
}
|
|
|
|
/**
|
|
* Find an item by key in the reiserfs tree.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_item_search(struct fsw_reiserfs_volume *vol,
|
|
fsw_u32 dir_id, fsw_u32 objectid, fsw_u64 offset,
|
|
struct fsw_reiserfs_item *item)
|
|
{
|
|
fsw_status_t status;
|
|
int comp_result;
|
|
fsw_u32 tree_bno, next_tree_bno, tree_level, nr_item, i;
|
|
fsw_u8 *buffer;
|
|
struct block_head *bhead;
|
|
struct reiserfs_key *key;
|
|
struct item_head *ihead;
|
|
|
|
FSW_MSG_DEBUG((FSW_MSGSTR("fsw_reiserfs_item_search: searching %d/%d/%lld\n"), dir_id, objectid, offset));
|
|
|
|
// BIG TODOS: Use binary search within the item.
|
|
// Remember tree path for "get next item" function.
|
|
|
|
item->valid = 0;
|
|
item->block_bno = 0;
|
|
|
|
// walk the tree
|
|
tree_bno = vol->sb->s_v1.s_root_block;
|
|
for (tree_level = vol->sb->s_v1.s_tree_height - 1; ; tree_level--) {
|
|
|
|
// get the current tree block into memory
|
|
status = fsw_block_get(vol, tree_bno, tree_level, (void **)&buffer);
|
|
if (status)
|
|
return status;
|
|
bhead = (struct block_head *)buffer;
|
|
if (bhead->blk_level != tree_level) {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_item_search: tree block %d has not expected level %d\n"), tree_bno, tree_level));
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
return FSW_VOLUME_CORRUPTED;
|
|
}
|
|
nr_item = bhead->blk_nr_item;
|
|
FSW_MSG_DEBUGV((FSW_MSGSTR("fsw_reiserfs_item_search: visiting block %d level %d items %d\n"), tree_bno, tree_level, nr_item));
|
|
item->path_bno[tree_level] = tree_bno;
|
|
|
|
// check if we have reached a leaf block
|
|
if (tree_level == DISK_LEAF_NODE_LEVEL)
|
|
break;
|
|
|
|
// search internal node block, look for the path to follow
|
|
key = (struct reiserfs_key *)(buffer + BLKH_SIZE);
|
|
for (i = 0; i < nr_item; i++, key++) {
|
|
if (fsw_reiserfs_compare_key(key, dir_id, objectid, offset) == FIRST_GREATER)
|
|
break;
|
|
}
|
|
item->path_index[tree_level] = i;
|
|
next_tree_bno = ((struct disk_child *)(buffer + BLKH_SIZE + nr_item * KEY_SIZE))[i].dc_block_number;
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
tree_bno = next_tree_bno;
|
|
}
|
|
|
|
// search leaf node block, look for our data
|
|
ihead = (struct item_head *)(buffer + BLKH_SIZE);
|
|
for (i = 0; i < nr_item; i++, ihead++) {
|
|
comp_result = fsw_reiserfs_compare_key(&ihead->ih_key, dir_id, objectid, offset);
|
|
if (comp_result == KEYS_IDENTICAL)
|
|
break;
|
|
if (comp_result == FIRST_GREATER) {
|
|
// Current key is greater than the search key. Use the last key before this
|
|
// one as the preliminary result.
|
|
if (i == 0) {
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
return FSW_NOT_FOUND;
|
|
}
|
|
i--, ihead--;
|
|
break;
|
|
}
|
|
}
|
|
if (i >= nr_item) {
|
|
// Go back to the last key, it was smaller than the search key.
|
|
// NOTE: The first key of the next leaf block is guaranteed to be greater than
|
|
// our search key.
|
|
i--, ihead--;
|
|
}
|
|
item->path_index[tree_level] = i;
|
|
// Since we may have a key that is smaller than the search key, verify that
|
|
// it is for the same object.
|
|
if (ihead->ih_key.k_dir_id != dir_id || ihead->ih_key.k_objectid != objectid) {
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
return FSW_NOT_FOUND; // Found no key for this object at all
|
|
}
|
|
|
|
// return results
|
|
fsw_memcpy(&item->ih, ihead, sizeof(struct item_head));
|
|
item->item_type = (fsw_u32)FSW_U64_SHR(ihead->ih_key.u.k_offset_v2.v, 60);
|
|
if (item->item_type != TYPE_DIRECT &&
|
|
item->item_type != TYPE_INDIRECT &&
|
|
item->item_type != TYPE_DIRENTRY) {
|
|
// 3.5 format (_v1)
|
|
item->item_type = ihead->ih_key.u.k_offset_v1.k_uniqueness;
|
|
item->item_offset = ihead->ih_key.u.k_offset_v1.k_offset;
|
|
} else {
|
|
// 3.6 format (_v2)
|
|
item->item_offset = ihead->ih_key.u.k_offset_v2.v & (~0ULL >> 4);
|
|
}
|
|
item->item_data = buffer + ihead->ih_item_location;
|
|
item->valid = 1;
|
|
|
|
// add information for block release
|
|
item->block_bno = tree_bno;
|
|
item->block_buffer = buffer;
|
|
|
|
FSW_MSG_DEBUG((FSW_MSGSTR("fsw_reiserfs_item_search: found %d/%d/%lld (%d)\n"),
|
|
ihead->ih_key.k_dir_id, ihead->ih_key.k_objectid, item->item_offset, item->item_type));
|
|
return FSW_SUCCESS;
|
|
}
|
|
|
|
/**
|
|
* Find the next item in the reiserfs tree for an already-found item.
|
|
*/
|
|
|
|
static fsw_status_t fsw_reiserfs_item_next(struct fsw_reiserfs_volume *vol,
|
|
struct fsw_reiserfs_item *item)
|
|
{
|
|
fsw_status_t status;
|
|
fsw_u32 dir_id, objectid;
|
|
// fsw_u64 offset;
|
|
fsw_u32 tree_bno, next_tree_bno, tree_level, nr_item, nr_ptr_item;
|
|
fsw_u8 *buffer;
|
|
struct block_head *bhead;
|
|
struct item_head *ihead;
|
|
|
|
if (!item->valid)
|
|
return FSW_NOT_FOUND;
|
|
fsw_reiserfs_item_release(vol, item); // TODO: maybe delay this and/or use the cached block!
|
|
|
|
dir_id = item->ih.ih_key.k_dir_id;
|
|
objectid = item->ih.ih_key.k_objectid;
|
|
// offset = item->item_offset;
|
|
|
|
FSW_MSG_DEBUG((FSW_MSGSTR("fsw_reiserfs_item_next: next for %d/%d/%lld\n"), dir_id, objectid, item->item_offset));
|
|
|
|
// find a node that has more items, moving up until we find one
|
|
|
|
for (tree_level = DISK_LEAF_NODE_LEVEL; tree_level < vol->sb->s_v1.s_tree_height; tree_level++) {
|
|
|
|
// get the current tree block into memory
|
|
tree_bno = item->path_bno[tree_level];
|
|
status = fsw_block_get(vol, tree_bno, tree_level, (void **)&buffer);
|
|
if (status)
|
|
return status;
|
|
bhead = (struct block_head *)buffer;
|
|
if (bhead->blk_level != tree_level) {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_item_next: tree block %d has not expected level %d\n"), tree_bno, tree_level));
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
return FSW_VOLUME_CORRUPTED;
|
|
}
|
|
nr_item = bhead->blk_nr_item;
|
|
FSW_MSG_DEBUGV((FSW_MSGSTR("fsw_reiserfs_item_next: visiting block %d level %d items %d\n"), tree_bno, tree_level, nr_item));
|
|
|
|
nr_ptr_item = nr_item + ((tree_level > DISK_LEAF_NODE_LEVEL) ? 1 : 0); // internal nodes have (nr_item) keys and (nr_item+1) pointers
|
|
item->path_index[tree_level]++;
|
|
if (item->path_index[tree_level] >= nr_ptr_item) {
|
|
item->path_index[tree_level] = 0;
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
continue; // this node doesn't have any more items, move up one level
|
|
}
|
|
|
|
// we have a new path to follow, move down to the leaf node again
|
|
while (tree_level > DISK_LEAF_NODE_LEVEL) {
|
|
// get next pointer from current block
|
|
next_tree_bno = ((struct disk_child *)(buffer + BLKH_SIZE + nr_item * KEY_SIZE))[item->path_index[tree_level]].dc_block_number;
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
tree_bno = next_tree_bno;
|
|
tree_level--;
|
|
|
|
// get the current tree block into memory
|
|
status = fsw_block_get(vol, tree_bno, tree_level, (void **)&buffer);
|
|
if (status)
|
|
return status;
|
|
bhead = (struct block_head *)buffer;
|
|
if (bhead->blk_level != tree_level) {
|
|
FSW_MSG_ASSERT((FSW_MSGSTR("fsw_reiserfs_item_next: tree block %d has not expected level %d\n"), tree_bno, tree_level));
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
return FSW_VOLUME_CORRUPTED;
|
|
}
|
|
nr_item = bhead->blk_nr_item;
|
|
FSW_MSG_DEBUGV((FSW_MSGSTR("fsw_reiserfs_item_next: visiting block %d level %d items %d\n"), tree_bno, tree_level, nr_item));
|
|
item->path_bno[tree_level] = tree_bno;
|
|
}
|
|
|
|
// get the item from the leaf node
|
|
ihead = ((struct item_head *)(buffer + BLKH_SIZE)) + item->path_index[tree_level];
|
|
|
|
// We now have the item that follows the previous one in the tree. Check that it
|
|
// belongs to the same object.
|
|
if (ihead->ih_key.k_dir_id != dir_id || ihead->ih_key.k_objectid != objectid) {
|
|
fsw_block_release(vol, tree_bno, buffer);
|
|
return FSW_NOT_FOUND; // Found no next key for this object
|
|
}
|
|
|
|
// return results
|
|
fsw_memcpy(&item->ih, ihead, sizeof(struct item_head));
|
|
item->item_type = (fsw_u32)FSW_U64_SHR(ihead->ih_key.u.k_offset_v2.v, 60);
|
|
if (item->item_type != TYPE_DIRECT &&
|
|
item->item_type != TYPE_INDIRECT &&
|
|
item->item_type != TYPE_DIRENTRY) {
|
|
// 3.5 format (_v1)
|
|
item->item_type = ihead->ih_key.u.k_offset_v1.k_uniqueness;
|
|
item->item_offset = ihead->ih_key.u.k_offset_v1.k_offset;
|
|
} else {
|
|
// 3.6 format (_v2)
|
|
item->item_offset = ihead->ih_key.u.k_offset_v2.v & (~0ULL >> 4);
|
|
}
|
|
item->item_data = buffer + ihead->ih_item_location;
|
|
item->valid = 1;
|
|
|
|
// add information for block release
|
|
item->block_bno = tree_bno;
|
|
item->block_buffer = buffer;
|
|
|
|
FSW_MSG_DEBUG((FSW_MSGSTR("fsw_reiserfs_item_next: found %d/%d/%lld (%d)\n"),
|
|
ihead->ih_key.k_dir_id, ihead->ih_key.k_objectid, item->item_offset, item->item_type));
|
|
return FSW_SUCCESS;
|
|
}
|
|
|
|
// we went to the highest level node and there still were no more items...
|
|
return FSW_NOT_FOUND;
|
|
}
|
|
|
|
/**
|
|
* Release the disk block still referenced by an item search result.
|
|
*/
|
|
|
|
static void fsw_reiserfs_item_release(struct fsw_reiserfs_volume *vol,
|
|
struct fsw_reiserfs_item *item)
|
|
{
|
|
if (!item->valid)
|
|
return;
|
|
|
|
if (item->block_bno > 0) {
|
|
fsw_block_release(vol, item->block_bno, item->block_buffer);
|
|
item->block_bno = 0;
|
|
}
|
|
}
|
|
|
|
// EOF
|