kernel_optimize_test/fs/sysv/super.c

564 lines
14 KiB
C
Raw Normal View History

/*
* linux/fs/sysv/inode.c
*
* minix/inode.c
* Copyright (C) 1991, 1992 Linus Torvalds
*
* xenix/inode.c
* Copyright (C) 1992 Doug Evans
*
* coh/inode.c
* Copyright (C) 1993 Pascal Haible, Bruno Haible
*
* sysv/inode.c
* Copyright (C) 1993 Paul B. Monday
*
* sysv/inode.c
* Copyright (C) 1993 Bruno Haible
* Copyright (C) 1997, 1998 Krzysztof G. Baranowski
*
* This file contains code for read/parsing the superblock.
*/
#include <linux/module.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/buffer_head.h>
#include "sysv.h"
/*
* The following functions try to recognize specific filesystems.
*
* We recognize:
* - Xenix FS by its magic number.
* - SystemV FS by its magic number.
* - Coherent FS by its funny fname/fpack field.
* - SCO AFS by s_nfree == 0xffff
* - V7 FS has no distinguishing features.
*
* We discriminate among SystemV4 and SystemV2 FS by the assumption that
* the time stamp is not < 01-01-1980.
*/
enum {
JAN_1_1980 = (10*365 + 2) * 24 * 60 * 60
};
static void detected_xenix(struct sysv_sb_info *sbi)
{
struct buffer_head *bh1 = sbi->s_bh1;
struct buffer_head *bh2 = sbi->s_bh2;
struct xenix_super_block * sbd1;
struct xenix_super_block * sbd2;
if (bh1 != bh2)
sbd1 = sbd2 = (struct xenix_super_block *) bh1->b_data;
else {
/* block size = 512, so bh1 != bh2 */
sbd1 = (struct xenix_super_block *) bh1->b_data;
sbd2 = (struct xenix_super_block *) (bh2->b_data - 512);
}
sbi->s_link_max = XENIX_LINK_MAX;
sbi->s_fic_size = XENIX_NICINOD;
sbi->s_flc_size = XENIX_NICFREE;
sbi->s_sbd1 = (char *)sbd1;
sbi->s_sbd2 = (char *)sbd2;
sbi->s_sb_fic_count = &sbd1->s_ninode;
sbi->s_sb_fic_inodes = &sbd1->s_inode[0];
sbi->s_sb_total_free_inodes = &sbd2->s_tinode;
sbi->s_bcache_count = &sbd1->s_nfree;
sbi->s_bcache = &sbd1->s_free[0];
sbi->s_free_blocks = &sbd2->s_tfree;
sbi->s_sb_time = &sbd2->s_time;
sbi->s_firstdatazone = fs16_to_cpu(sbi, sbd1->s_isize);
sbi->s_nzones = fs32_to_cpu(sbi, sbd1->s_fsize);
}
static void detected_sysv4(struct sysv_sb_info *sbi)
{
struct sysv4_super_block * sbd;
struct buffer_head *bh1 = sbi->s_bh1;
struct buffer_head *bh2 = sbi->s_bh2;
if (bh1 == bh2)
sbd = (struct sysv4_super_block *) (bh1->b_data + BLOCK_SIZE/2);
else
sbd = (struct sysv4_super_block *) bh2->b_data;
sbi->s_link_max = SYSV_LINK_MAX;
sbi->s_fic_size = SYSV_NICINOD;
sbi->s_flc_size = SYSV_NICFREE;
sbi->s_sbd1 = (char *)sbd;
sbi->s_sbd2 = (char *)sbd;
sbi->s_sb_fic_count = &sbd->s_ninode;
sbi->s_sb_fic_inodes = &sbd->s_inode[0];
sbi->s_sb_total_free_inodes = &sbd->s_tinode;
sbi->s_bcache_count = &sbd->s_nfree;
sbi->s_bcache = &sbd->s_free[0];
sbi->s_free_blocks = &sbd->s_tfree;
sbi->s_sb_time = &sbd->s_time;
sbi->s_sb_state = &sbd->s_state;
sbi->s_firstdatazone = fs16_to_cpu(sbi, sbd->s_isize);
sbi->s_nzones = fs32_to_cpu(sbi, sbd->s_fsize);
}
static void detected_sysv2(struct sysv_sb_info *sbi)
{
struct sysv2_super_block *sbd;
struct buffer_head *bh1 = sbi->s_bh1;
struct buffer_head *bh2 = sbi->s_bh2;
if (bh1 == bh2)
sbd = (struct sysv2_super_block *) (bh1->b_data + BLOCK_SIZE/2);
else
sbd = (struct sysv2_super_block *) bh2->b_data;
sbi->s_link_max = SYSV_LINK_MAX;
sbi->s_fic_size = SYSV_NICINOD;
sbi->s_flc_size = SYSV_NICFREE;
sbi->s_sbd1 = (char *)sbd;
sbi->s_sbd2 = (char *)sbd;
sbi->s_sb_fic_count = &sbd->s_ninode;
sbi->s_sb_fic_inodes = &sbd->s_inode[0];
sbi->s_sb_total_free_inodes = &sbd->s_tinode;
sbi->s_bcache_count = &sbd->s_nfree;
sbi->s_bcache = &sbd->s_free[0];
sbi->s_free_blocks = &sbd->s_tfree;
sbi->s_sb_time = &sbd->s_time;
sbi->s_sb_state = &sbd->s_state;
sbi->s_firstdatazone = fs16_to_cpu(sbi, sbd->s_isize);
sbi->s_nzones = fs32_to_cpu(sbi, sbd->s_fsize);
}
static void detected_coherent(struct sysv_sb_info *sbi)
{
struct coh_super_block * sbd;
struct buffer_head *bh1 = sbi->s_bh1;
sbd = (struct coh_super_block *) bh1->b_data;
sbi->s_link_max = COH_LINK_MAX;
sbi->s_fic_size = COH_NICINOD;
sbi->s_flc_size = COH_NICFREE;
sbi->s_sbd1 = (char *)sbd;
sbi->s_sbd2 = (char *)sbd;
sbi->s_sb_fic_count = &sbd->s_ninode;
sbi->s_sb_fic_inodes = &sbd->s_inode[0];
sbi->s_sb_total_free_inodes = &sbd->s_tinode;
sbi->s_bcache_count = &sbd->s_nfree;
sbi->s_bcache = &sbd->s_free[0];
sbi->s_free_blocks = &sbd->s_tfree;
sbi->s_sb_time = &sbd->s_time;
sbi->s_firstdatazone = fs16_to_cpu(sbi, sbd->s_isize);
sbi->s_nzones = fs32_to_cpu(sbi, sbd->s_fsize);
}
static void detected_v7(struct sysv_sb_info *sbi)
{
struct buffer_head *bh2 = sbi->s_bh2;
struct v7_super_block *sbd = (struct v7_super_block *)bh2->b_data;
sbi->s_link_max = V7_LINK_MAX;
sbi->s_fic_size = V7_NICINOD;
sbi->s_flc_size = V7_NICFREE;
sbi->s_sbd1 = (char *)sbd;
sbi->s_sbd2 = (char *)sbd;
sbi->s_sb_fic_count = &sbd->s_ninode;
sbi->s_sb_fic_inodes = &sbd->s_inode[0];
sbi->s_sb_total_free_inodes = &sbd->s_tinode;
sbi->s_bcache_count = &sbd->s_nfree;
sbi->s_bcache = &sbd->s_free[0];
sbi->s_free_blocks = &sbd->s_tfree;
sbi->s_sb_time = &sbd->s_time;
sbi->s_firstdatazone = fs16_to_cpu(sbi, sbd->s_isize);
sbi->s_nzones = fs32_to_cpu(sbi, sbd->s_fsize);
}
static int detect_xenix(struct sysv_sb_info *sbi, struct buffer_head *bh)
{
struct xenix_super_block *sbd = (struct xenix_super_block *)bh->b_data;
if (*(__le32 *)&sbd->s_magic == cpu_to_le32(0x2b5544))
sbi->s_bytesex = BYTESEX_LE;
else if (*(__be32 *)&sbd->s_magic == cpu_to_be32(0x2b5544))
sbi->s_bytesex = BYTESEX_BE;
else
return 0;
switch (fs32_to_cpu(sbi, sbd->s_type)) {
case 1:
sbi->s_type = FSTYPE_XENIX;
return 1;
case 2:
sbi->s_type = FSTYPE_XENIX;
return 2;
default:
return 0;
}
}
static int detect_sysv(struct sysv_sb_info *sbi, struct buffer_head *bh)
{
struct super_block *sb = sbi->s_sb;
/* All relevant fields are at the same offsets in R2 and R4 */
struct sysv4_super_block * sbd;
u32 type;
sbd = (struct sysv4_super_block *) (bh->b_data + BLOCK_SIZE/2);
if (*(__le32 *)&sbd->s_magic == cpu_to_le32(0xfd187e20))
sbi->s_bytesex = BYTESEX_LE;
else if (*(__be32 *)&sbd->s_magic == cpu_to_be32(0xfd187e20))
sbi->s_bytesex = BYTESEX_BE;
else
return 0;
type = fs32_to_cpu(sbi, sbd->s_type);
if (fs16_to_cpu(sbi, sbd->s_nfree) == 0xffff) {
sbi->s_type = FSTYPE_AFS;
sbi->s_forced_ro = 1;
if (!(sb->s_flags & MS_RDONLY)) {
printk("SysV FS: SCO EAFS on %s detected, "
"forcing read-only mode.\n",
sb->s_id);
}
return type;
}
if (fs32_to_cpu(sbi, sbd->s_time) < JAN_1_1980) {
/* this is likely to happen on SystemV2 FS */
if (type > 3 || type < 1)
return 0;
sbi->s_type = FSTYPE_SYSV2;
return type;
}
if ((type > 3 || type < 1) && (type > 0x30 || type < 0x10))
return 0;
/* On Interactive Unix (ISC) Version 4.0/3.x s_type field = 0x10,
0x20 or 0x30 indicates that symbolic links and the 14-character
filename limit is gone. Due to lack of information about this
feature read-only mode seems to be a reasonable approach... -KGB */
if (type >= 0x10) {
printk("SysV FS: can't handle long file names on %s, "
"forcing read-only mode.\n", sb->s_id);
sbi->s_forced_ro = 1;
}
sbi->s_type = FSTYPE_SYSV4;
return type >= 0x10 ? type >> 4 : type;
}
static int detect_coherent(struct sysv_sb_info *sbi, struct buffer_head *bh)
{
struct coh_super_block * sbd;
sbd = (struct coh_super_block *) (bh->b_data + BLOCK_SIZE/2);
if ((memcmp(sbd->s_fname,"noname",6) && memcmp(sbd->s_fname,"xxxxx ",6))
|| (memcmp(sbd->s_fpack,"nopack",6) && memcmp(sbd->s_fpack,"xxxxx\n",6)))
return 0;
sbi->s_bytesex = BYTESEX_PDP;
sbi->s_type = FSTYPE_COH;
return 1;
}
static int detect_sysv_odd(struct sysv_sb_info *sbi, struct buffer_head *bh)
{
int size = detect_sysv(sbi, bh);
return size>2 ? 0 : size;
}
static struct {
int block;
int (*test)(struct sysv_sb_info *, struct buffer_head *);
} flavours[] = {
{1, detect_xenix},
{0, detect_sysv},
{0, detect_coherent},
{9, detect_sysv_odd},
{15,detect_sysv_odd},
{18,detect_sysv},
};
static char *flavour_names[] = {
[FSTYPE_XENIX] = "Xenix",
[FSTYPE_SYSV4] = "SystemV",
[FSTYPE_SYSV2] = "SystemV Release 2",
[FSTYPE_COH] = "Coherent",
[FSTYPE_V7] = "V7",
[FSTYPE_AFS] = "AFS",
};
static void (*flavour_setup[])(struct sysv_sb_info *) = {
[FSTYPE_XENIX] = detected_xenix,
[FSTYPE_SYSV4] = detected_sysv4,
[FSTYPE_SYSV2] = detected_sysv2,
[FSTYPE_COH] = detected_coherent,
[FSTYPE_V7] = detected_v7,
[FSTYPE_AFS] = detected_sysv4,
};
static int complete_read_super(struct super_block *sb, int silent, int size)
{
struct sysv_sb_info *sbi = SYSV_SB(sb);
struct inode *root_inode;
char *found = flavour_names[sbi->s_type];
u_char n_bits = size+8;
int bsize = 1 << n_bits;
int bsize_4 = bsize >> 2;
sbi->s_firstinodezone = 2;
flavour_setup[sbi->s_type](sbi);
sbi->s_truncate = 1;
sbi->s_ndatazones = sbi->s_nzones - sbi->s_firstdatazone;
sbi->s_inodes_per_block = bsize >> 6;
sbi->s_inodes_per_block_1 = (bsize >> 6)-1;
sbi->s_inodes_per_block_bits = n_bits-6;
sbi->s_ind_per_block = bsize_4;
sbi->s_ind_per_block_2 = bsize_4*bsize_4;
sbi->s_toobig_block = 10 + bsize_4 * (1 + bsize_4 * (1 + bsize_4));
sbi->s_ind_per_block_bits = n_bits-2;
sbi->s_ninodes = (sbi->s_firstdatazone - sbi->s_firstinodezone)
<< sbi->s_inodes_per_block_bits;
if (!silent)
printk("VFS: Found a %s FS (block size = %ld) on device %s\n",
found, sb->s_blocksize, sb->s_id);
sb->s_magic = SYSV_MAGIC_BASE + sbi->s_type;
/* set up enough so that it can read an inode */
sb->s_op = &sysv_sops;
root_inode = iget(sb,SYSV_ROOT_INO);
if (!root_inode || is_bad_inode(root_inode)) {
printk("SysV FS: get root inode failed\n");
return 0;
}
sb->s_root = d_alloc_root(root_inode);
if (!sb->s_root) {
iput(root_inode);
printk("SysV FS: get root dentry failed\n");
return 0;
}
if (sbi->s_forced_ro)
sb->s_flags |= MS_RDONLY;
if (sbi->s_truncate)
sb->s_root->d_op = &sysv_dentry_operations;
sb->s_dirt = 1;
return 1;
}
static int sysv_fill_super(struct super_block *sb, void *data, int silent)
{
struct buffer_head *bh1, *bh = NULL;
struct sysv_sb_info *sbi;
unsigned long blocknr;
int size = 0, i;
BUILD_BUG_ON(1024 != sizeof (struct xenix_super_block));
BUILD_BUG_ON(512 != sizeof (struct sysv4_super_block));
BUILD_BUG_ON(512 != sizeof (struct sysv2_super_block));
BUILD_BUG_ON(500 != sizeof (struct coh_super_block));
BUILD_BUG_ON(64 != sizeof (struct sysv_inode));
sbi = kzalloc(sizeof(struct sysv_sb_info), GFP_KERNEL);
if (!sbi)
return -ENOMEM;
sbi->s_sb = sb;
sbi->s_block_base = 0;
sb->s_fs_info = sbi;
sb_set_blocksize(sb, BLOCK_SIZE);
for (i = 0; i < ARRAY_SIZE(flavours) && !size; i++) {
brelse(bh);
bh = sb_bread(sb, flavours[i].block);
if (!bh)
continue;
size = flavours[i].test(SYSV_SB(sb), bh);
}
if (!size)
goto Eunknown;
switch (size) {
case 1:
blocknr = bh->b_blocknr << 1;
brelse(bh);
sb_set_blocksize(sb, 512);
bh1 = sb_bread(sb, blocknr);
bh = sb_bread(sb, blocknr + 1);
break;
case 2:
bh1 = bh;
break;
case 3:
blocknr = bh->b_blocknr >> 1;
brelse(bh);
sb_set_blocksize(sb, 2048);
bh1 = bh = sb_bread(sb, blocknr);
break;
default:
goto Ebadsize;
}
if (bh && bh1) {
sbi->s_bh1 = bh1;
sbi->s_bh2 = bh;
if (complete_read_super(sb, silent, size))
return 0;
}
brelse(bh1);
brelse(bh);
sb_set_blocksize(sb, BLOCK_SIZE);
printk("oldfs: cannot read superblock\n");
failed:
kfree(sbi);
return -EINVAL;
Eunknown:
brelse(bh);
if (!silent)
printk("VFS: unable to find oldfs superblock on device %s\n",
sb->s_id);
goto failed;
Ebadsize:
brelse(bh);
if (!silent)
printk("VFS: oldfs: unsupported block size (%dKb)\n",
1<<(size-2));
goto failed;
}
static int v7_fill_super(struct super_block *sb, void *data, int silent)
{
struct sysv_sb_info *sbi;
struct buffer_head *bh, *bh2 = NULL;
struct v7_super_block *v7sb;
struct sysv_inode *v7i;
if (440 != sizeof (struct v7_super_block))
panic("V7 FS: bad super-block size");
if (64 != sizeof (struct sysv_inode))
panic("sysv fs: bad i-node size");
sbi = kzalloc(sizeof(struct sysv_sb_info), GFP_KERNEL);
if (!sbi)
return -ENOMEM;
sbi->s_sb = sb;
sbi->s_block_base = 0;
sbi->s_type = FSTYPE_V7;
sbi->s_bytesex = BYTESEX_PDP;
sb->s_fs_info = sbi;
sb_set_blocksize(sb, 512);
if ((bh = sb_bread(sb, 1)) == NULL) {
if (!silent)
printk("VFS: unable to read V7 FS superblock on "
"device %s.\n", sb->s_id);
goto failed;
}
/* plausibility check on superblock */
v7sb = (struct v7_super_block *) bh->b_data;
if (fs16_to_cpu(sbi, v7sb->s_nfree) > V7_NICFREE ||
fs16_to_cpu(sbi, v7sb->s_ninode) > V7_NICINOD ||
fs32_to_cpu(sbi, v7sb->s_time) == 0)
goto failed;
/* plausibility check on root inode: it is a directory,
with a nonzero size that is a multiple of 16 */
if ((bh2 = sb_bread(sb, 2)) == NULL)
goto failed;
v7i = (struct sysv_inode *)(bh2->b_data + 64);
if ((fs16_to_cpu(sbi, v7i->i_mode) & ~0777) != S_IFDIR ||
(fs32_to_cpu(sbi, v7i->i_size) == 0) ||
(fs32_to_cpu(sbi, v7i->i_size) & 017) != 0)
goto failed;
brelse(bh2);
bh2 = NULL;
sbi->s_bh1 = bh;
sbi->s_bh2 = bh;
if (complete_read_super(sb, silent, 1))
return 0;
failed:
brelse(bh2);
brelse(bh);
kfree(sbi);
return -EINVAL;
}
/* Every kernel module contains stuff like this. */
[PATCH] VFS: Permit filesystem to override root dentry on mount Extend the get_sb() filesystem operation to take an extra argument that permits the VFS to pass in the target vfsmount that defines the mountpoint. The filesystem is then required to manually set the superblock and root dentry pointers. For most filesystems, this should be done with simple_set_mnt() which will set the superblock pointer and then set the root dentry to the superblock's s_root (as per the old default behaviour). The get_sb() op now returns an integer as there's now no need to return the superblock pointer. This patch permits a superblock to be implicitly shared amongst several mount points, such as can be done with NFS to avoid potential inode aliasing. In such a case, simple_set_mnt() would not be called, and instead the mnt_root and mnt_sb would be set directly. The patch also makes the following changes: (*) the get_sb_*() convenience functions in the core kernel now take a vfsmount pointer argument and return an integer, so most filesystems have to change very little. (*) If one of the convenience function is not used, then get_sb() should normally call simple_set_mnt() to instantiate the vfsmount. This will always return 0, and so can be tail-called from get_sb(). (*) generic_shutdown_super() now calls shrink_dcache_sb() to clean up the dcache upon superblock destruction rather than shrink_dcache_anon(). This is required because the superblock may now have multiple trees that aren't actually bound to s_root, but that still need to be cleaned up. The currently called functions assume that the whole tree is rooted at s_root, and that anonymous dentries are not the roots of trees which results in dentries being left unculled. However, with the way NFS superblock sharing are currently set to be implemented, these assumptions are violated: the root of the filesystem is simply a dummy dentry and inode (the real inode for '/' may well be inaccessible), and all the vfsmounts are rooted on anonymous[*] dentries with child trees. [*] Anonymous until discovered from another tree. (*) The documentation has been adjusted, including the additional bit of changing ext2_* into foo_* in the documentation. [akpm@osdl.org: convert ipath_fs, do other stuff] Signed-off-by: David Howells <dhowells@redhat.com> Acked-by: Al Viro <viro@zeniv.linux.org.uk> Cc: Nathan Scott <nathans@sgi.com> Cc: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 17:02:57 +08:00
static int sysv_get_sb(struct file_system_type *fs_type,
int flags, const char *dev_name, void *data, struct vfsmount *mnt)
{
[PATCH] VFS: Permit filesystem to override root dentry on mount Extend the get_sb() filesystem operation to take an extra argument that permits the VFS to pass in the target vfsmount that defines the mountpoint. The filesystem is then required to manually set the superblock and root dentry pointers. For most filesystems, this should be done with simple_set_mnt() which will set the superblock pointer and then set the root dentry to the superblock's s_root (as per the old default behaviour). The get_sb() op now returns an integer as there's now no need to return the superblock pointer. This patch permits a superblock to be implicitly shared amongst several mount points, such as can be done with NFS to avoid potential inode aliasing. In such a case, simple_set_mnt() would not be called, and instead the mnt_root and mnt_sb would be set directly. The patch also makes the following changes: (*) the get_sb_*() convenience functions in the core kernel now take a vfsmount pointer argument and return an integer, so most filesystems have to change very little. (*) If one of the convenience function is not used, then get_sb() should normally call simple_set_mnt() to instantiate the vfsmount. This will always return 0, and so can be tail-called from get_sb(). (*) generic_shutdown_super() now calls shrink_dcache_sb() to clean up the dcache upon superblock destruction rather than shrink_dcache_anon(). This is required because the superblock may now have multiple trees that aren't actually bound to s_root, but that still need to be cleaned up. The currently called functions assume that the whole tree is rooted at s_root, and that anonymous dentries are not the roots of trees which results in dentries being left unculled. However, with the way NFS superblock sharing are currently set to be implemented, these assumptions are violated: the root of the filesystem is simply a dummy dentry and inode (the real inode for '/' may well be inaccessible), and all the vfsmounts are rooted on anonymous[*] dentries with child trees. [*] Anonymous until discovered from another tree. (*) The documentation has been adjusted, including the additional bit of changing ext2_* into foo_* in the documentation. [akpm@osdl.org: convert ipath_fs, do other stuff] Signed-off-by: David Howells <dhowells@redhat.com> Acked-by: Al Viro <viro@zeniv.linux.org.uk> Cc: Nathan Scott <nathans@sgi.com> Cc: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 17:02:57 +08:00
return get_sb_bdev(fs_type, flags, dev_name, data, sysv_fill_super,
mnt);
}
[PATCH] VFS: Permit filesystem to override root dentry on mount Extend the get_sb() filesystem operation to take an extra argument that permits the VFS to pass in the target vfsmount that defines the mountpoint. The filesystem is then required to manually set the superblock and root dentry pointers. For most filesystems, this should be done with simple_set_mnt() which will set the superblock pointer and then set the root dentry to the superblock's s_root (as per the old default behaviour). The get_sb() op now returns an integer as there's now no need to return the superblock pointer. This patch permits a superblock to be implicitly shared amongst several mount points, such as can be done with NFS to avoid potential inode aliasing. In such a case, simple_set_mnt() would not be called, and instead the mnt_root and mnt_sb would be set directly. The patch also makes the following changes: (*) the get_sb_*() convenience functions in the core kernel now take a vfsmount pointer argument and return an integer, so most filesystems have to change very little. (*) If one of the convenience function is not used, then get_sb() should normally call simple_set_mnt() to instantiate the vfsmount. This will always return 0, and so can be tail-called from get_sb(). (*) generic_shutdown_super() now calls shrink_dcache_sb() to clean up the dcache upon superblock destruction rather than shrink_dcache_anon(). This is required because the superblock may now have multiple trees that aren't actually bound to s_root, but that still need to be cleaned up. The currently called functions assume that the whole tree is rooted at s_root, and that anonymous dentries are not the roots of trees which results in dentries being left unculled. However, with the way NFS superblock sharing are currently set to be implemented, these assumptions are violated: the root of the filesystem is simply a dummy dentry and inode (the real inode for '/' may well be inaccessible), and all the vfsmounts are rooted on anonymous[*] dentries with child trees. [*] Anonymous until discovered from another tree. (*) The documentation has been adjusted, including the additional bit of changing ext2_* into foo_* in the documentation. [akpm@osdl.org: convert ipath_fs, do other stuff] Signed-off-by: David Howells <dhowells@redhat.com> Acked-by: Al Viro <viro@zeniv.linux.org.uk> Cc: Nathan Scott <nathans@sgi.com> Cc: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 17:02:57 +08:00
static int v7_get_sb(struct file_system_type *fs_type,
int flags, const char *dev_name, void *data, struct vfsmount *mnt)
{
[PATCH] VFS: Permit filesystem to override root dentry on mount Extend the get_sb() filesystem operation to take an extra argument that permits the VFS to pass in the target vfsmount that defines the mountpoint. The filesystem is then required to manually set the superblock and root dentry pointers. For most filesystems, this should be done with simple_set_mnt() which will set the superblock pointer and then set the root dentry to the superblock's s_root (as per the old default behaviour). The get_sb() op now returns an integer as there's now no need to return the superblock pointer. This patch permits a superblock to be implicitly shared amongst several mount points, such as can be done with NFS to avoid potential inode aliasing. In such a case, simple_set_mnt() would not be called, and instead the mnt_root and mnt_sb would be set directly. The patch also makes the following changes: (*) the get_sb_*() convenience functions in the core kernel now take a vfsmount pointer argument and return an integer, so most filesystems have to change very little. (*) If one of the convenience function is not used, then get_sb() should normally call simple_set_mnt() to instantiate the vfsmount. This will always return 0, and so can be tail-called from get_sb(). (*) generic_shutdown_super() now calls shrink_dcache_sb() to clean up the dcache upon superblock destruction rather than shrink_dcache_anon(). This is required because the superblock may now have multiple trees that aren't actually bound to s_root, but that still need to be cleaned up. The currently called functions assume that the whole tree is rooted at s_root, and that anonymous dentries are not the roots of trees which results in dentries being left unculled. However, with the way NFS superblock sharing are currently set to be implemented, these assumptions are violated: the root of the filesystem is simply a dummy dentry and inode (the real inode for '/' may well be inaccessible), and all the vfsmounts are rooted on anonymous[*] dentries with child trees. [*] Anonymous until discovered from another tree. (*) The documentation has been adjusted, including the additional bit of changing ext2_* into foo_* in the documentation. [akpm@osdl.org: convert ipath_fs, do other stuff] Signed-off-by: David Howells <dhowells@redhat.com> Acked-by: Al Viro <viro@zeniv.linux.org.uk> Cc: Nathan Scott <nathans@sgi.com> Cc: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 17:02:57 +08:00
return get_sb_bdev(fs_type, flags, dev_name, data, v7_fill_super, mnt);
}
static struct file_system_type sysv_fs_type = {
.owner = THIS_MODULE,
.name = "sysv",
.get_sb = sysv_get_sb,
.kill_sb = kill_block_super,
.fs_flags = FS_REQUIRES_DEV,
};
static struct file_system_type v7_fs_type = {
.owner = THIS_MODULE,
.name = "v7",
.get_sb = v7_get_sb,
.kill_sb = kill_block_super,
.fs_flags = FS_REQUIRES_DEV,
};
static int __init init_sysv_fs(void)
{
int error;
error = sysv_init_icache();
if (error)
goto out;
error = register_filesystem(&sysv_fs_type);
if (error)
goto destroy_icache;
error = register_filesystem(&v7_fs_type);
if (error)
goto unregister;
return 0;
unregister:
unregister_filesystem(&sysv_fs_type);
destroy_icache:
sysv_destroy_icache();
out:
return error;
}
static void __exit exit_sysv_fs(void)
{
unregister_filesystem(&sysv_fs_type);
unregister_filesystem(&v7_fs_type);
sysv_destroy_icache();
}
module_init(init_sysv_fs)
module_exit(exit_sysv_fs)
MODULE_LICENSE("GPL");