Minki/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2024-11-26 22:21:42 +00:00
Code Issues Packages Projects Releases Wiki Activity
7b4e4f7f81
linux/fs/ufs/inode.c

1352 lines
36 KiB
C
Raw Normal View History

Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* linux/fs/ufs/inode.c
*
* Copyright (C) 1998
* Daniel Pirkl <daniel.pirkl@email.cz>
* Charles University, Faculty of Mathematics and Physics
*
* from
*
* linux/fs/ext2/inode.c
*
* Copyright (C) 1992, 1993, 1994, 1995
* Remy Card (card@masi.ibp.fr)
* Laboratoire MASI - Institut Blaise Pascal
* Universite Pierre et Marie Curie (Paris VI)
*
* from
*
* linux/fs/minix/inode.c
*
* Copyright (C) 1991, 1992 Linus Torvalds
*
* Goal-directed block allocation by Stephen Tweedie (sct@dcs.ed.ac.uk), 1993
* Big-endian to little-endian byte-swapping/bitmaps by
* David S. Miller (davem@caip.rutgers.edu), 1995
*/
#include <asm/uaccess.h>
#include <linux/errno.h>
#include <linux/fs.h>
#include <linux/time.h>
#include <linux/stat.h>
#include <linux/string.h>
#include <linux/mm.h>
#include <linux/buffer_head.h>
pass writeback_control to ->write_inode This gives the filesystem more information about the writeback that is happening. Trond requested this for the NFS unstable write handling, and other filesystems might benefit from this too by beeing able to distinguish between the different callers in more detail. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-03-05 08:21:37 +00:00
#include <linux/writeback.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
drop linux/ufs_fs.h from userspace export and relocate it to fs/ufs/ufs_fs.h Per previous discussions about cleaning up ufs_fs.h, people just want this straight up dropped from userspace export. The only remaining consumer (silo) has been fixed a while ago to not rely on this header. This allows use to move it completely from include/linux/ to fs/ufs/ seeing as how the only in-kernel consumer is fs/ufs/. Signed-off-by: Mike Frysinger <vapier@gentoo.org> Cc: Jan Kara <jack@ucw.cz> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-08 12:21:31 +00:00
#include "ufs_fs.h"
ufs: move non-layout parts of ufs_fs.h to fs/ufs/ Move prototypes and in-core structures to fs/ufs/ similar to what most other filesystems already do. I made little modifications: move also ufs debug macros and mount options constants into fs/ufs/ufs.h, this stuff also private for ufs. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-17 06:26:51 +00:00
#include "ufs.h"
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#include "swab.h"
#include "util.h"
ufs: the offsets ufs_block_to_path() puts into array are not sector_t type makes no sense - those are indices in block number arrays, not block numbers. And no, UFS is not likely to grow indirect blocks with 4Gpointers in them... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-04 18:13:14 +00:00
static int ufs_block_to_path(struct inode *inode, sector_t i_block, unsigned offsets[4])
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct ufs_sb_private_info *uspi = UFS_SB(inode->i_sb)->s_uspi;
int ptrs = uspi->s_apb;
int ptrs_bits = uspi->s_apbshift;
const long direct_blocks = UFS_NDADDR,
indirect_blocks = ptrs,
double_blocks = (1 << (ptrs_bits * 2));
int n = 0;
[PATCH] ufs: easy debug Currently to turn on debug mode "user" has to edit ~10 files, to turn off he has to do it again. This patch introduce such changes: 1)turn on(off) debug messages via ".config" 2)remove unnecessary duplication of code 3)make "UFSD" macros more similar to function 4)fix some compiler warnings Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:24 +00:00
UFSD("ptrs=uspi->s_apb = %d,double_blocks=%ld \n",ptrs,double_blocks);
ufs: sector_t cannot be negative unsigned i_block,fragment cannot be negative. Signed-off-by: Roel Kluin <roel.kluin@gmail.com> Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-17 23:26:28 +00:00
if (i_block < direct_blocks) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
offsets[n++] = i_block;
} else if ((i_block -= direct_blocks) < indirect_blocks) {
offsets[n++] = UFS_IND_BLOCK;
offsets[n++] = i_block;
} else if ((i_block -= indirect_blocks) < double_blocks) {
offsets[n++] = UFS_DIND_BLOCK;
offsets[n++] = i_block >> ptrs_bits;
offsets[n++] = i_block & (ptrs - 1);
} else if (((i_block -= double_blocks) >> (ptrs_bits * 2)) < ptrs) {
offsets[n++] = UFS_TIND_BLOCK;
offsets[n++] = i_block >> (ptrs_bits * 2);
offsets[n++] = (i_block >> ptrs_bits) & (ptrs - 1);
offsets[n++] = i_block & (ptrs - 1);
} else {
ufs_warning(inode->i_sb, "ufs_block_to_path", "block > big");
}
return n;
}
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
typedef struct {
void *p;
union {
__fs32 key32;
__fs64 key64;
};
struct buffer_head *bh;
} Indirect;
static inline int grow_chain32(struct ufs_inode_info *ufsi,
struct buffer_head *bh, __fs32 *v,
Indirect *from, Indirect *to)
{
Indirect *p;
unsigned seq;
to->bh = bh;
do {
seq = read_seqbegin(&ufsi->meta_lock);
to->key32 = *(__fs32 *)(to->p = v);
for (p = from; p <= to && p->key32 == *(__fs32 *)p->p; p++)
;
} while (read_seqretry(&ufsi->meta_lock, seq));
return (p > to);
}
static inline int grow_chain64(struct ufs_inode_info *ufsi,
struct buffer_head *bh, __fs64 *v,
Indirect *from, Indirect *to)
{
Indirect *p;
unsigned seq;
to->bh = bh;
do {
seq = read_seqbegin(&ufsi->meta_lock);
to->key64 = *(__fs64 *)(to->p = v);
for (p = from; p <= to && p->key64 == *(__fs64 *)p->p; p++)
;
} while (read_seqretry(&ufsi->meta_lock, seq));
return (p > to);
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* Returns the location of the fragment from
Fix common misspellings Fixes generated by 'codespell' and manually reviewed. Signed-off-by: Lucas De Marchi <lucas.demarchi@profusion.mobi>
2011-03-31 01:57:33 +00:00
* the beginning of the filesystem.
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
*/
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
static u64 ufs_frag_map(struct inode *inode, sector_t frag)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct ufs_inode_info *ufsi = UFS_I(inode);
struct super_block *sb = inode->i_sb;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
u64 mask = (u64) uspi->s_apbmask>>uspi->s_fpbshift;
int shift = uspi->s_apbshift-uspi->s_fpbshift;
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
unsigned offsets[4], *p;
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
Indirect chain[4], *q = chain;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
int depth = ufs_block_to_path(inode, frag >> uspi->s_fpbshift, offsets);
unsigned flags = UFS_SB(sb)->s_flags;
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
u64 res = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[PATCH] ufs: easy debug Currently to turn on debug mode "user" has to edit ~10 files, to turn off he has to do it again. This patch introduce such changes: 1)turn on(off) debug messages via ".config" 2)remove unnecessary duplication of code 3)make "UFSD" macros more similar to function 4)fix some compiler warnings Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:24 +00:00
UFSD(": frag = %llu depth = %d\n", (unsigned long long)frag, depth);
[PATCH] ufs: printk() fix fs/ufs/inode.c: In function `ufs_frag_map': fs/ufs/inode.c:101: warning: long long unsigned int format, u64 arg (arg 4) fs/ufs/inode.c: In function `ufs_getfrag_block': fs/ufs/inode.c:432: warning: long long unsigned int format, u64 arg (arg 2) Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-29 09:24:29 +00:00
UFSD(": uspi->s_fpbshift = %d ,uspi->s_apbmask = %x, mask=%llx\n",
uspi->s_fpbshift, uspi->s_apbmask,
(unsigned long long)mask);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (depth == 0)
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
goto no_block;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
again:
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
p = offsets;
if ((flags & UFS_TYPE_MASK) == UFS_TYPE_UFS2)
goto ufs2;
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
if (!grow_chain32(ufsi, NULL, &ufsi->i_u1.i_data[*p++], chain, q))
goto changed;
if (!q->key32)
goto no_block;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
while (--depth) {
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
__fs32 *ptr;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct buffer_head *bh;
ufs: the offsets ufs_block_to_path() puts into array are not sector_t type makes no sense - those are indices in block number arrays, not block numbers. And no, UFS is not likely to grow indirect blocks with 4Gpointers in them... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-04 18:13:14 +00:00
unsigned n = *p++;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
bh = sb_bread(sb, uspi->s_sbbase +
fs32_to_cpu(sb, q->key32) + (n>>shift));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (!bh)
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
goto no_block;
ptr = (__fs32 *)bh->b_data + (n & mask);
if (!grow_chain32(ufsi, bh, ptr, chain, ++q))
goto changed;
if (!q->key32)
goto no_block;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
res = fs32_to_cpu(sb, q->key32);
goto found;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
ufs2:
if (!grow_chain64(ufsi, NULL, &ufsi->i_u1.u2_i_data[*p++], chain, q))
goto changed;
if (!q->key64)
goto no_block;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
while (--depth) {
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
__fs64 *ptr;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct buffer_head *bh;
ufs: the offsets ufs_block_to_path() puts into array are not sector_t type makes no sense - those are indices in block number arrays, not block numbers. And no, UFS is not likely to grow indirect blocks with 4Gpointers in them... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-04 18:13:14 +00:00
unsigned n = *p++;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
bh = sb_bread(sb, uspi->s_sbbase +
fs64_to_cpu(sb, q->key64) + (n>>shift));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (!bh)
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
goto no_block;
ptr = (__fs64 *)bh->b_data + (n & mask);
if (!grow_chain64(ufsi, bh, ptr, chain, ++q))
goto changed;
if (!q->key64)
goto no_block;
}
res = fs64_to_cpu(sb, q->key64);
found:
res += uspi->s_sbbase + (frag & uspi->s_fpbmask);
no_block:
while (q > chain) {
brelse(q->bh);
q--;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
return res;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
changed:
while (q > chain) {
brelse(q->bh);
q--;
}
goto again;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
/**
* ufs_inode_getfrag() - allocate new fragment(s)
fs/ufs/inode.c: kernel-doc warning fixes Signed-off-by: Fabian Frederick <fabf@skynet.be> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-08 21:21:08 +00:00
* @inode: pointer to inode
* @fragment: number of `fragment' which hold pointer
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
* to new allocated fragment(s)
fs/ufs/inode.c: kernel-doc warning fixes Signed-off-by: Fabian Frederick <fabf@skynet.be> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-08 21:21:08 +00:00
* @new_fragment: number of new allocated fragment(s)
* @required: how many fragment(s) we require
* @err: we set it if something wrong
* @phys: pointer to where we save physical number of new allocated fragments,
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
* NULL if we allocate not data(indirect blocks for example).
fs/ufs/inode.c: kernel-doc warning fixes Signed-off-by: Fabian Frederick <fabf@skynet.be> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-08 21:21:08 +00:00
* @new: we set it if we allocate new block
* @locked_page: for ufs_new_fragments()
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
*/
static struct buffer_head *
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
ufs_inode_getfrag(struct inode *inode, u64 fragment,
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
sector_t new_fragment, unsigned int required, int *err,
long *phys, int *new, struct page *locked_page)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct ufs_inode_info *ufsi = UFS_I(inode);
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
struct super_block *sb = inode->i_sb;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct buffer_head * result;
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
unsigned blockoff, lastblockoff;
u64 tmp, goal, lastfrag, block, lastblock;
void *p, *p2;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
UFSD("ENTER, ino %lu, fragment %llu, new_fragment %llu, required %u, "
"metadata %d\n", inode->i_ino, (unsigned long long)fragment,
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
(unsigned long long)new_fragment, required, !phys);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/* TODO : to be done for write support
if ( (flags & UFS_TYPE_MASK) == UFS_TYPE_UFS2)
goto ufs2;
*/
block = ufs_fragstoblks (fragment);
blockoff = ufs_fragnum (fragment);
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
p = ufs_get_direct_data_ptr(uspi, ufsi, block);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
goal = 0;
repeat:
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
tmp = ufs_data_ptr_to_cpu(sb, p);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
lastfrag = ufsi->i_lastfrag;
if (tmp && fragment < lastfrag) {
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
if (!phys) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
result = sb_getblk(sb, uspi->s_sbbase + tmp + blockoff);
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
if (tmp == ufs_data_ptr_to_cpu(sb, p)) {
UFSD("EXIT, result %llu\n",
(unsigned long long)tmp + blockoff);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return result;
}
brelse (result);
goto repeat;
} else {
[PATCH] ufs: zeroize the rest of block in truncate This patch fix behaviour in such test scenario: lseek(fd, BIG_OFFSET) write(fd, buf, sizeof(buf)) truncate(BIG_OFFSET) truncate(BIG_OFFSET + sizeof(buf)) read(fd, buf...) Because of if file big enough(BIG_OFFSET) we start allocate space by block, ordinary block size > page size, so we should zeroize the rest of block in truncate(except last framgnet, about which VFS should care), to not get garbage, when we extend file. Also patch corrects conversion from pointer to block to physical block number, this helps in case of not common used UFS types. And add to debug output inode number. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-03-16 21:38:09 +00:00
*phys = uspi->s_sbbase + tmp + blockoff;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return NULL;
}
}
lastblock = ufs_fragstoblks (lastfrag);
lastblockoff = ufs_fragnum (lastfrag);
/*
* We will extend file into new block beyond last allocated block
*/
if (lastblock < block) {
/*
* We must reallocate last allocated block
*/
if (lastblockoff) {
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
p2 = ufs_get_direct_data_ptr(uspi, ufsi, lastblock);
tmp = ufs_new_fragments(inode, p2, lastfrag,
ufs_data_ptr_to_cpu(sb, p2),
uspi->s_fpb - lastblockoff,
err, locked_page);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (!tmp) {
if (lastfrag != ufsi->i_lastfrag)
goto repeat;
else
return NULL;
}
lastfrag = ufsi->i_lastfrag;
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
tmp = ufs_data_ptr_to_cpu(sb,
ufs_get_direct_data_ptr(uspi, ufsi,
lastblock));
[PATCH] ufs: write to hole in big file On UFS, this scenario: open(O_TRUNC) lseek(1024 * 1024 * 80) write("A") lseek(1024 * 2) write("A") may cause access to invalid address. This happened because of "goal" is calculated in wrong way in block allocation path, as I see this problem exists also in 2.4. We use construction like this i_data[lastfrag], i_data array of pointers to direct blocks, indirect and so on, it has ceratain size ~20 elements, and lastfrag may have value for example 40000. Also this patch fixes related to handling such scenario issues, wrong zeroing metadata, in case of block(not fragment) allocation, and wrong goal calculation, when we allocate block Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-08-27 08:23:45 +00:00
if (tmp)
goal = tmp + uspi->s_fpb;
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
tmp = ufs_new_fragments (inode, p, fragment - blockoff,
[PATCH] ufs: change block number on the fly First of all some necessary notes about UFS by it self: To avoid waste of disk space the tail of file consists not from blocks (which is ordinary big enough, 16K usually), it consists from fragments(which is ordinary 2K). When file is growing its tail occupy 1 fragment, 2 fragments... At some stage decision to allocate whole block is made and all fragments are moved to one block. How this situation was handled before: ufs_prepare_write ->block_prepare_write ->ufs_getfrag_block ->... ->ufs_new_fragments: bh = sb_bread bh->b_blocknr = result + i; mark_buffer_dirty (bh); This is wrong solution, because: - it didn't take into consideration that there is another cache: "inode page cache" - because of sb_getblk uses not b_blocknr, (it uses page->index) to find certain block, this breaks sb_getblk. How this situation is handled now: we go though all "page inode cache", if there are no such page in cache we load it into cache, and change b_blocknr. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:20 +00:00
goal, required + blockoff,
[PATCH] ufs: alloc metadata null page fix These series of patches result of UFS1 write support stress testing, like running fsx-linux, untar and build linux kernel etc We pass from ufs::get_block_t to levels below: pointer to the current page, to make possible things like reallocation of blocks on the fly, and we also uses this pointer for indication, what actually we allocate data block or meta data block, but currently we make decision about what we allocate on the wrong level, this may and cause oops if we allocate blocks in some special order. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-01-29 21:19:54 +00:00
err,
phys != NULL ? locked_page : NULL);
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
} else if (lastblock == block) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* We will extend last allocated block
*/
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
tmp = ufs_new_fragments(inode, p, fragment -
(blockoff - lastblockoff),
ufs_data_ptr_to_cpu(sb, p),
required + (blockoff - lastblockoff),
[PATCH] ufs: alloc metadata null page fix These series of patches result of UFS1 write support stress testing, like running fsx-linux, untar and build linux kernel etc We pass from ufs::get_block_t to levels below: pointer to the current page, to make possible things like reallocation of blocks on the fly, and we also uses this pointer for indication, what actually we allocate data block or meta data block, but currently we make decision about what we allocate on the wrong level, this may and cause oops if we allocate blocks in some special order. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-01-29 21:19:54 +00:00
err, phys != NULL ? locked_page : NULL);
[PATCH] ufs: write to hole in big file On UFS, this scenario: open(O_TRUNC) lseek(1024 * 1024 * 80) write("A") lseek(1024 * 2) write("A") may cause access to invalid address. This happened because of "goal" is calculated in wrong way in block allocation path, as I see this problem exists also in 2.4. We use construction like this i_data[lastfrag], i_data array of pointers to direct blocks, indirect and so on, it has ceratain size ~20 elements, and lastfrag may have value for example 40000. Also this patch fixes related to handling such scenario issues, wrong zeroing metadata, in case of block(not fragment) allocation, and wrong goal calculation, when we allocate block Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-08-27 08:23:45 +00:00
} else /* (lastblock > block) */ {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* We will allocate new block before last allocated block
*/
[PATCH] ufs: write to hole in big file On UFS, this scenario: open(O_TRUNC) lseek(1024 * 1024 * 80) write("A") lseek(1024 * 2) write("A") may cause access to invalid address. This happened because of "goal" is calculated in wrong way in block allocation path, as I see this problem exists also in 2.4. We use construction like this i_data[lastfrag], i_data array of pointers to direct blocks, indirect and so on, it has ceratain size ~20 elements, and lastfrag may have value for example 40000. Also this patch fixes related to handling such scenario issues, wrong zeroing metadata, in case of block(not fragment) allocation, and wrong goal calculation, when we allocate block Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-08-27 08:23:45 +00:00
if (block) {
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
tmp = ufs_data_ptr_to_cpu(sb,
ufs_get_direct_data_ptr(uspi, ufsi, block - 1));
[PATCH] ufs: write to hole in big file On UFS, this scenario: open(O_TRUNC) lseek(1024 * 1024 * 80) write("A") lseek(1024 * 2) write("A") may cause access to invalid address. This happened because of "goal" is calculated in wrong way in block allocation path, as I see this problem exists also in 2.4. We use construction like this i_data[lastfrag], i_data array of pointers to direct blocks, indirect and so on, it has ceratain size ~20 elements, and lastfrag may have value for example 40000. Also this patch fixes related to handling such scenario issues, wrong zeroing metadata, in case of block(not fragment) allocation, and wrong goal calculation, when we allocate block Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-08-27 08:23:45 +00:00
if (tmp)
goal = tmp + uspi->s_fpb;
}
[PATCH] ufs: change block number on the fly First of all some necessary notes about UFS by it self: To avoid waste of disk space the tail of file consists not from blocks (which is ordinary big enough, 16K usually), it consists from fragments(which is ordinary 2K). When file is growing its tail occupy 1 fragment, 2 fragments... At some stage decision to allocate whole block is made and all fragments are moved to one block. How this situation was handled before: ufs_prepare_write ->block_prepare_write ->ufs_getfrag_block ->... ->ufs_new_fragments: bh = sb_bread bh->b_blocknr = result + i; mark_buffer_dirty (bh); This is wrong solution, because: - it didn't take into consideration that there is another cache: "inode page cache" - because of sb_getblk uses not b_blocknr, (it uses page->index) to find certain block, this breaks sb_getblk. How this situation is handled now: we go though all "page inode cache", if there are no such page in cache we load it into cache, and change b_blocknr. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:20 +00:00
tmp = ufs_new_fragments(inode, p, fragment - blockoff,
[PATCH] ufs: alloc metadata null page fix These series of patches result of UFS1 write support stress testing, like running fsx-linux, untar and build linux kernel etc We pass from ufs::get_block_t to levels below: pointer to the current page, to make possible things like reallocation of blocks on the fly, and we also uses this pointer for indication, what actually we allocate data block or meta data block, but currently we make decision about what we allocate on the wrong level, this may and cause oops if we allocate blocks in some special order. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-01-29 21:19:54 +00:00
goal, uspi->s_fpb, err,
phys != NULL ? locked_page : NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
if (!tmp) {
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
if ((!blockoff && ufs_data_ptr_to_cpu(sb, p)) ||
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
(blockoff && lastfrag != ufsi->i_lastfrag))
goto repeat;
*err = -ENOSPC;
return NULL;
}
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
if (!phys) {
[PATCH] ufs: zeroize the rest of block in truncate This patch fix behaviour in such test scenario: lseek(fd, BIG_OFFSET) write(fd, buf, sizeof(buf)) truncate(BIG_OFFSET) truncate(BIG_OFFSET + sizeof(buf)) read(fd, buf...) Because of if file big enough(BIG_OFFSET) we start allocate space by block, ordinary block size > page size, so we should zeroize the rest of block in truncate(except last framgnet, about which VFS should care), to not get garbage, when we extend file. Also patch corrects conversion from pointer to block to physical block number, this helps in case of not common used UFS types. And add to debug output inode number. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-03-16 21:38:09 +00:00
result = sb_getblk(sb, uspi->s_sbbase + tmp + blockoff);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
} else {
[PATCH] ufs: zeroize the rest of block in truncate This patch fix behaviour in such test scenario: lseek(fd, BIG_OFFSET) write(fd, buf, sizeof(buf)) truncate(BIG_OFFSET) truncate(BIG_OFFSET + sizeof(buf)) read(fd, buf...) Because of if file big enough(BIG_OFFSET) we start allocate space by block, ordinary block size > page size, so we should zeroize the rest of block in truncate(except last framgnet, about which VFS should care), to not get garbage, when we extend file. Also patch corrects conversion from pointer to block to physical block number, this helps in case of not common used UFS types. And add to debug output inode number. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-03-16 21:38:09 +00:00
*phys = uspi->s_sbbase + tmp + blockoff;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
result = NULL;
*err = 0;
*new = 1;
}
inode->i_ctime = CURRENT_TIME_SEC;
if (IS_SYNC(inode))
ufs_sync_inode (inode);
mark_inode_dirty(inode);
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
UFSD("EXIT, result %llu\n", (unsigned long long)tmp + blockoff);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return result;
/* This part : To be implemented ....
Required only for writing, not required for READ-ONLY.
ufs2:
u2_block = ufs_fragstoblks(fragment);
u2_blockoff = ufs_fragnum(fragment);
p = ufsi->i_u1.u2_i_data + block;
goal = 0;
repeat2:
tmp = fs32_to_cpu(sb, *p);
lastfrag = ufsi->i_lastfrag;
*/
}
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
/**
* ufs_inode_getblock() - allocate new block
fs/ufs/inode.c: kernel-doc warning fixes Signed-off-by: Fabian Frederick <fabf@skynet.be> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-08 21:21:08 +00:00
* @inode: pointer to inode
* @bh: pointer to block which hold "pointer" to new allocated block
* @fragment: number of `fragment' which hold pointer
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
* to new allocated block
fs/ufs/inode.c: kernel-doc warning fixes Signed-off-by: Fabian Frederick <fabf@skynet.be> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-08 21:21:08 +00:00
* @new_fragment: number of new allocated fragment
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
* (block will hold this fragment and also uspi->s_fpb-1)
fs/ufs/inode.c: kernel-doc warning fixes Signed-off-by: Fabian Frederick <fabf@skynet.be> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-08 21:21:08 +00:00
* @err: see ufs_inode_getfrag()
* @phys: see ufs_inode_getfrag()
* @new: see ufs_inode_getfrag()
* @locked_page: see ufs_inode_getfrag()
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
*/
static struct buffer_head *
ufs_inode_getblock(struct inode *inode, struct buffer_head *bh,
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
u64 fragment, sector_t new_fragment, int *err,
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
long *phys, int *new, struct page *locked_page)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
struct super_block *sb = inode->i_sb;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct buffer_head * result;
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
unsigned blockoff;
u64 tmp, goal, block;
void *p;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
block = ufs_fragstoblks (fragment);
blockoff = ufs_fragnum (fragment);
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
UFSD("ENTER, ino %lu, fragment %llu, new_fragment %llu, metadata %d\n",
inode->i_ino, (unsigned long long)fragment,
(unsigned long long)new_fragment, !phys);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
result = NULL;
if (!bh)
goto out;
if (!buffer_uptodate(bh)) {
ll_rw_block (READ, 1, &bh);
wait_on_buffer (bh);
if (!buffer_uptodate(bh))
goto out;
}
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
if (uspi->fs_magic == UFS2_MAGIC)
p = (__fs64 *)bh->b_data + block;
else
p = (__fs32 *)bh->b_data + block;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
repeat:
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
tmp = ufs_data_ptr_to_cpu(sb, p);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (tmp) {
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
if (!phys) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
result = sb_getblk(sb, uspi->s_sbbase + tmp + blockoff);
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
if (tmp == ufs_data_ptr_to_cpu(sb, p))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
goto out;
brelse (result);
goto repeat;
} else {
[PATCH] ufs: zeroize the rest of block in truncate This patch fix behaviour in such test scenario: lseek(fd, BIG_OFFSET) write(fd, buf, sizeof(buf)) truncate(BIG_OFFSET) truncate(BIG_OFFSET + sizeof(buf)) read(fd, buf...) Because of if file big enough(BIG_OFFSET) we start allocate space by block, ordinary block size > page size, so we should zeroize the rest of block in truncate(except last framgnet, about which VFS should care), to not get garbage, when we extend file. Also patch corrects conversion from pointer to block to physical block number, this helps in case of not common used UFS types. And add to debug output inode number. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-03-16 21:38:09 +00:00
*phys = uspi->s_sbbase + tmp + blockoff;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
goto out;
}
}
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
if (block && (uspi->fs_magic == UFS2_MAGIC ?
(tmp = fs64_to_cpu(sb, ((__fs64 *)bh->b_data)[block-1])) :
(tmp = fs32_to_cpu(sb, ((__fs32 *)bh->b_data)[block-1]))))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
goal = tmp + uspi->s_fpb;
else
goal = bh->b_blocknr + uspi->s_fpb;
[PATCH] ufs: change block number on the fly First of all some necessary notes about UFS by it self: To avoid waste of disk space the tail of file consists not from blocks (which is ordinary big enough, 16K usually), it consists from fragments(which is ordinary 2K). When file is growing its tail occupy 1 fragment, 2 fragments... At some stage decision to allocate whole block is made and all fragments are moved to one block. How this situation was handled before: ufs_prepare_write ->block_prepare_write ->ufs_getfrag_block ->... ->ufs_new_fragments: bh = sb_bread bh->b_blocknr = result + i; mark_buffer_dirty (bh); This is wrong solution, because: - it didn't take into consideration that there is another cache: "inode page cache" - because of sb_getblk uses not b_blocknr, (it uses page->index) to find certain block, this breaks sb_getblk. How this situation is handled now: we go though all "page inode cache", if there are no such page in cache we load it into cache, and change b_blocknr. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:20 +00:00
tmp = ufs_new_fragments(inode, p, ufs_blknum(new_fragment), goal,
uspi->s_fpb, err, locked_page);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (!tmp) {
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
if (ufs_data_ptr_to_cpu(sb, p))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
goto repeat;
goto out;
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[PATCH] ufs: right block allocation * After block allocation, we map it on the same "address" as 8 others blocks * We nullify block several times: once in ufs/block.c and once in block_*write_full_page, and use different "caches" for this. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:19 +00:00
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
if (!phys) {
[PATCH] ufs: zeroize the rest of block in truncate This patch fix behaviour in such test scenario: lseek(fd, BIG_OFFSET) write(fd, buf, sizeof(buf)) truncate(BIG_OFFSET) truncate(BIG_OFFSET + sizeof(buf)) read(fd, buf...) Because of if file big enough(BIG_OFFSET) we start allocate space by block, ordinary block size > page size, so we should zeroize the rest of block in truncate(except last framgnet, about which VFS should care), to not get garbage, when we extend file. Also patch corrects conversion from pointer to block to physical block number, this helps in case of not common used UFS types. And add to debug output inode number. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-03-16 21:38:09 +00:00
result = sb_getblk(sb, uspi->s_sbbase + tmp + blockoff);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
} else {
[PATCH] ufs: zeroize the rest of block in truncate This patch fix behaviour in such test scenario: lseek(fd, BIG_OFFSET) write(fd, buf, sizeof(buf)) truncate(BIG_OFFSET) truncate(BIG_OFFSET + sizeof(buf)) read(fd, buf...) Because of if file big enough(BIG_OFFSET) we start allocate space by block, ordinary block size > page size, so we should zeroize the rest of block in truncate(except last framgnet, about which VFS should care), to not get garbage, when we extend file. Also patch corrects conversion from pointer to block to physical block number, this helps in case of not common used UFS types. And add to debug output inode number. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-03-16 21:38:09 +00:00
*phys = uspi->s_sbbase + tmp + blockoff;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
*new = 1;
}
mark_buffer_dirty(bh);
if (IS_SYNC(inode))
sync_dirty_buffer(bh);
inode->i_ctime = CURRENT_TIME_SEC;
mark_inode_dirty(inode);
[PATCH] ufs2 write: block allocation update Patch adds ability to work with 64bit metadata, this made by replacing work with 32bit pointers by inline functions. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:32 +00:00
UFSD("result %llu\n", (unsigned long long)tmp + blockoff);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
out:
brelse (bh);
[PATCH] ufs: easy debug Currently to turn on debug mode "user" has to edit ~10 files, to turn off he has to do it again. This patch introduce such changes: 1)turn on(off) debug messages via ".config" 2)remove unnecessary duplication of code 3)make "UFSD" macros more similar to function 4)fix some compiler warnings Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:24 +00:00
UFSD("EXIT\n");
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return result;
}
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
/**
ufs: Fix a typo Signed-off-by: Alessio Igor Bogani <abogani@kernel.org> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
2011-04-08 17:33:07 +00:00
* ufs_getfrag_block() - `get_block_t' function, interface between UFS and
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
* readpage, writepage and so on
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
*/
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
static int ufs_getfrag_block(struct inode *inode, sector_t fragment, struct buffer_head *bh_result, int create)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct super_block * sb = inode->i_sb;
ufs: remove the BKL This introduces a new per-superblock mutex in UFS to replace the big kernel lock. I have been careful to avoid nested calls to lock_ufs and to get the lock order right with respect to other mutexes, in particular lock_super. I did not make any attempt to prove that the big kernel lock is not needed in a particular place in the code, which is very possible. The mutex has a significant performance impact, so it is only used on SMP or PREEMPT configurations. As Nick Piggin noticed, any allocation inside of the lock may end up deadlocking when we get to ufs_getfrag_block in the reclaim task, so we now use GFP_NOFS. Signed-off-by: Arnd Bergmann <arnd@arndb.de> Tested-by: Nick Bowler <nbowler@elliptictech.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Cc: Nick Piggin <npiggin@gmail.com>
2011-01-24 09:14:12 +00:00
struct ufs_sb_info * sbi = UFS_SB(sb);
struct ufs_sb_private_info * uspi = sbi->s_uspi;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
struct buffer_head * bh;
int ret, err, new;
unsigned long ptr,phys;
u64 phys64 = 0;
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (!create) {
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
phys64 = ufs_frag_map(inode, fragment);
[PATCH] ufs: printk() fix fs/ufs/inode.c: In function `ufs_frag_map': fs/ufs/inode.c:101: warning: long long unsigned int format, u64 arg (arg 4) fs/ufs/inode.c: In function `ufs_getfrag_block': fs/ufs/inode.c:432: warning: long long unsigned int format, u64 arg (arg 2) Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-29 09:24:29 +00:00
UFSD("phys64 = %llu\n", (unsigned long long)phys64);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (phys64)
map_bh(bh_result, sb, phys64);
return 0;
}
/* This code entered only while writing ....? */
err = -EIO;
new = 0;
ret = 0;
bh = NULL;
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
mutex_lock(&UFS_I(inode)->truncate_mutex);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[PATCH] ufs: easy debug Currently to turn on debug mode "user" has to edit ~10 files, to turn off he has to do it again. This patch introduce such changes: 1)turn on(off) debug messages via ".config" 2)remove unnecessary duplication of code 3)make "UFSD" macros more similar to function 4)fix some compiler warnings Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:24 +00:00
UFSD("ENTER, ino %lu, fragment %llu\n", inode->i_ino, (unsigned long long)fragment);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (fragment >
((UFS_NDADDR + uspi->s_apb + uspi->s_2apb + uspi->s_3apb)
<< uspi->s_fpbshift))
goto abort_too_big;
err = 0;
ptr = fragment;
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* ok, these macros clean the logic up a bit and make
* it much more readable:
*/
#define GET_INODE_DATABLOCK(x) \
[PATCH] ufs: alloc metadata null page fix These series of patches result of UFS1 write support stress testing, like running fsx-linux, untar and build linux kernel etc We pass from ufs::get_block_t to levels below: pointer to the current page, to make possible things like reallocation of blocks on the fly, and we also uses this pointer for indication, what actually we allocate data block or meta data block, but currently we make decision about what we allocate on the wrong level, this may and cause oops if we allocate blocks in some special order. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-01-29 21:19:54 +00:00
ufs_inode_getfrag(inode, x, fragment, 1, &err, &phys, &new,\
bh_result->b_page)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#define GET_INODE_PTR(x) \
[PATCH] ufs: alloc metadata null page fix These series of patches result of UFS1 write support stress testing, like running fsx-linux, untar and build linux kernel etc We pass from ufs::get_block_t to levels below: pointer to the current page, to make possible things like reallocation of blocks on the fly, and we also uses this pointer for indication, what actually we allocate data block or meta data block, but currently we make decision about what we allocate on the wrong level, this may and cause oops if we allocate blocks in some special order. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-01-29 21:19:54 +00:00
ufs_inode_getfrag(inode, x, fragment, uspi->s_fpb, &err, NULL, NULL,\
bh_result->b_page)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#define GET_INDIRECT_DATABLOCK(x) \
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
ufs_inode_getblock(inode, bh, x, fragment, \
[PATCH] fix garbage instead of zeroes in UFS Looks like this is the problem, which point Al Viro some time ago: ufs's get_block callback allocates 16k of disk at a time, and links that entire 16k into the file's metadata. But because get_block is called for only a single buffer_head (a 2k buffer_head in this case?) we are only able to tell the VFS that this 2k is buffer_new(). So when ufs_getfrag_block() is later called to map some more data in the file, and when that data resides within the remaining 14k of this fragment, ufs_getfrag_block() will incorrectly return a !buffer_new() buffer_head. I don't see _right_ way to do nullification of whole block, if use inode page cache, some pages may be outside of inode limits (inode size), and will be lost; if use blockdev page cache it is possible to zero real data, if later inode page cache will be used. The simpliest way, as can I see usage of block device page cache, but not only mark dirty, but also sync it during "nullification". I use my simple tests collection, which I used for check that create,open,write,read,close works on ufs, and I see that this patch makes ufs code 18% slower then before. Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:37:04 +00:00
&err, &phys, &new, bh_result->b_page)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
#define GET_INDIRECT_PTR(x) \
[PATCH] ufs: zero metadata Presently if we allocate several "metadata" blocks (pointers to indirect blocks for example), we fill with zeroes only the first block. This cause some problems in "truncate" function. Also this patch remove some unused arguments from several functions and add comments. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:27 +00:00
ufs_inode_getblock(inode, bh, x, fragment, \
[PATCH] fix garbage instead of zeroes in UFS Looks like this is the problem, which point Al Viro some time ago: ufs's get_block callback allocates 16k of disk at a time, and links that entire 16k into the file's metadata. But because get_block is called for only a single buffer_head (a 2k buffer_head in this case?) we are only able to tell the VFS that this 2k is buffer_new(). So when ufs_getfrag_block() is later called to map some more data in the file, and when that data resides within the remaining 14k of this fragment, ufs_getfrag_block() will incorrectly return a !buffer_new() buffer_head. I don't see _right_ way to do nullification of whole block, if use inode page cache, some pages may be outside of inode limits (inode size), and will be lost; if use blockdev page cache it is possible to zero real data, if later inode page cache will be used. The simpliest way, as can I see usage of block device page cache, but not only mark dirty, but also sync it during "nullification". I use my simple tests collection, which I used for check that create,open,write,read,close works on ufs, and I see that this patch makes ufs code 18% slower then before. Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:37:04 +00:00
&err, NULL, NULL, NULL)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (ptr < UFS_NDIR_FRAGMENT) {
bh = GET_INODE_DATABLOCK(ptr);
goto out;
}
ptr -= UFS_NDIR_FRAGMENT;
if (ptr < (1 << (uspi->s_apbshift + uspi->s_fpbshift))) {
bh = GET_INODE_PTR(UFS_IND_FRAGMENT + (ptr >> uspi->s_apbshift));
goto get_indirect;
}
ptr -= 1 << (uspi->s_apbshift + uspi->s_fpbshift);
if (ptr < (1 << (uspi->s_2apbshift + uspi->s_fpbshift))) {
bh = GET_INODE_PTR(UFS_DIND_FRAGMENT + (ptr >> uspi->s_2apbshift));
goto get_double;
}
ptr -= 1 << (uspi->s_2apbshift + uspi->s_fpbshift);
bh = GET_INODE_PTR(UFS_TIND_FRAGMENT + (ptr >> uspi->s_3apbshift));
bh = GET_INDIRECT_PTR((ptr >> uspi->s_2apbshift) & uspi->s_apbmask);
get_double:
bh = GET_INDIRECT_PTR((ptr >> uspi->s_apbshift) & uspi->s_apbmask);
get_indirect:
bh = GET_INDIRECT_DATABLOCK(ptr & uspi->s_apbmask);
#undef GET_INODE_DATABLOCK
#undef GET_INODE_PTR
#undef GET_INDIRECT_DATABLOCK
#undef GET_INDIRECT_PTR
out:
if (err)
goto abort;
if (new)
set_buffer_new(bh_result);
map_bh(bh_result, sb, phys);
abort:
ufs: don't use lock_ufs() for block pointers tree protection * stores to block pointers are under per-inode seqlock (meta_lock) and mutex (truncate_mutex) * fetches of block pointers are either under truncate_mutex, or wrapped into seqretry loop on meta_lock * all changes of ->i_size are under truncate_mutex and i_mutex * all changes of ->i_lastfrag are under truncate_mutex It's similar to what ext2 is doing; the main difference is that unlike ext2 we can't rely upon the atomicity of stores into block pointers - on UFS2 they are 64bit. So we can't cut the corner when switching a pointer from NULL to non-NULL as we could in ext2_splice_branch() and need to use meta_lock on all modifications. We use seqlock where ext2 uses rwlock; ext2 could probably also benefit from such change... Another non-trivial difference is that with UFS we *cannot* have reader grab truncate_mutex in case of race - it has to keep retrying. That might be possible to change, but not until we lift tail unpacking several levels up in call chain. After that commit we do *NOT* hold fs-wide serialization on accesses to block pointers anymore. Moreover, lock_ufs() can become a normal mutex now - it's only used on statfs, remount and sync_fs and none of those uses are recursive. As the matter of fact, *now* it can be collapsed with ->s_lock, and be eventually replaced with saner per-cylinder-group spinlocks, but that's a separate story. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:02:56 +00:00
mutex_unlock(&UFS_I(inode)->truncate_mutex);
ufs: remove the BKL This introduces a new per-superblock mutex in UFS to replace the big kernel lock. I have been careful to avoid nested calls to lock_ufs and to get the lock order right with respect to other mutexes, in particular lock_super. I did not make any attempt to prove that the big kernel lock is not needed in a particular place in the code, which is very possible. The mutex has a significant performance impact, so it is only used on SMP or PREEMPT configurations. As Nick Piggin noticed, any allocation inside of the lock may end up deadlocking when we get to ufs_getfrag_block in the reclaim task, so we now use GFP_NOFS. Signed-off-by: Arnd Bergmann <arnd@arndb.de> Tested-by: Nick Bowler <nbowler@elliptictech.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Cc: Nick Piggin <npiggin@gmail.com>
2011-01-24 09:14:12 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return err;
abort_too_big:
ufs_warning(sb, "ufs_get_block", "block > big");
goto abort;
}
static int ufs_writepage(struct page *page, struct writeback_control *wbc)
{
return block_write_full_page(page,ufs_getfrag_block,wbc);
}
ufs: convert to new aops Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:19 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
static int ufs_readpage(struct file *file, struct page *page)
{
return block_read_full_page(page,ufs_getfrag_block);
}
ufs: convert to new aops Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:19 +00:00
clean up write_begin usage for directories in pagecache For filesystem that implement directories in pagecache we call block_write_begin with an already allocated page for this code, while the normal regular file write path uses the default block_write_begin behaviour. Get rid of the __foofs_write_begin helper and opencode the normal write_begin call in foofs_write_begin, while adding a new foofs_prepare_chunk helper for the directory code. The added benefit is that foofs_prepare_chunk has a much saner calling convention. Note that the interruptible flag passed into block_write_begin is always ignored if we already pass in a page (see next patch for details), and we never were doing truncations of exessive blocks for this case either so we can switch directly to block_write_begin_newtrunc. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-04 09:29:56 +00:00
int ufs_prepare_chunk(struct page *page, loff_t pos, unsigned len)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
introduce __block_write_begin Split up the block_write_begin implementation - __block_write_begin is a new trivial wrapper for block_prepare_write that always takes an already allocated page and can be either called from block_write_begin or filesystem code that already has a page allocated. Remove the handling of already allocated pages from block_write_begin after switching all callers that do it to __block_write_begin. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-04 09:29:57 +00:00
return __block_write_begin(page, pos, len, ufs_getfrag_block);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
ufs: convert to new aops Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:19 +00:00
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
static void ufs_truncate_blocks(struct inode *);
ufs: drop vmtruncate Removed vmtruncate Signed-off-by: Marco Stornelli <marco.stornelli@gmail.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2012-12-15 10:45:14 +00:00
static void ufs_write_failed(struct address_space *mapping, loff_t to)
{
struct inode *inode = mapping->host;
ufs: free excessive blocks upon ->write_begin() failure/short copy Broken in "[PATCH] ufs: truncate should allocate block for last byte"; all way back in 2006. ufs_setattr() hadn't been the only user of vmtruncate() and eliminating ->truncate() method required corrections in a bunch of places. Eventually those places had migrated into ->write_begin() failure exit and ->write_end() after short copy... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-16 22:06:40 +00:00
if (to > inode->i_size) {
truncate: drop 'oldsize' truncate_pagecache() parameter truncate_pagecache() doesn't care about old size since commit cedabed49b39 ("vfs: Fix vmtruncate() regression"). Let's drop it. Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 22:13:56 +00:00
truncate_pagecache(inode, inode->i_size);
ufs: free excessive blocks upon ->write_begin() failure/short copy Broken in "[PATCH] ufs: truncate should allocate block for last byte"; all way back in 2006. ufs_setattr() hadn't been the only user of vmtruncate() and eliminating ->truncate() method required corrections in a bunch of places. Eventually those places had migrated into ->write_begin() failure exit and ->write_end() after short copy... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-16 22:06:40 +00:00
ufs_truncate_blocks(inode);
}
ufs: drop vmtruncate Removed vmtruncate Signed-off-by: Marco Stornelli <marco.stornelli@gmail.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2012-12-15 10:45:14 +00:00
}
ufs: convert to new aops Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:19 +00:00
static int ufs_write_begin(struct file *file, struct address_space *mapping,
loff_t pos, unsigned len, unsigned flags,
struct page **pagep, void **fsdata)
{
get rid of block_write_begin_newtrunc Move the call to vmtruncate to get rid of accessive blocks to the callers in preparation of the new truncate sequence and rename the non-truncating version to block_write_begin. While we're at it also remove several unused arguments to block_write_begin. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-04 09:29:58 +00:00
int ret;
ret = block_write_begin(mapping, pos, len, flags, pagep,
clean up write_begin usage for directories in pagecache For filesystem that implement directories in pagecache we call block_write_begin with an already allocated page for this code, while the normal regular file write path uses the default block_write_begin behaviour. Get rid of the __foofs_write_begin helper and opencode the normal write_begin call in foofs_write_begin, while adding a new foofs_prepare_chunk helper for the directory code. The added benefit is that foofs_prepare_chunk has a much saner calling convention. Note that the interruptible flag passed into block_write_begin is always ignored if we already pass in a page (see next patch for details), and we never were doing truncations of exessive blocks for this case either so we can switch directly to block_write_begin_newtrunc. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-04 09:29:56 +00:00
ufs_getfrag_block);
ufs: drop vmtruncate Removed vmtruncate Signed-off-by: Marco Stornelli <marco.stornelli@gmail.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2012-12-15 10:45:14 +00:00
if (unlikely(ret))
ufs_write_failed(mapping, pos + len);
get rid of block_write_begin_newtrunc Move the call to vmtruncate to get rid of accessive blocks to the callers in preparation of the new truncate sequence and rename the non-truncating version to block_write_begin. While we're at it also remove several unused arguments to block_write_begin. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-04 09:29:58 +00:00
return ret;
ufs: convert to new aops Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:19 +00:00
}
ufs: free excessive blocks upon ->write_begin() failure/short copy Broken in "[PATCH] ufs: truncate should allocate block for last byte"; all way back in 2006. ufs_setattr() hadn't been the only user of vmtruncate() and eliminating ->truncate() method required corrections in a bunch of places. Eventually those places had migrated into ->write_begin() failure exit and ->write_end() after short copy... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-16 22:06:40 +00:00
static int ufs_write_end(struct file *file, struct address_space *mapping,
loff_t pos, unsigned len, unsigned copied,
struct page *page, void *fsdata)
{
int ret;
ret = generic_write_end(file, mapping, pos, len, copied, page, fsdata);
if (ret < len)
ufs_write_failed(mapping, pos + len);
return ret;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
static sector_t ufs_bmap(struct address_space *mapping, sector_t block)
{
return generic_block_bmap(mapping,block,ufs_getfrag_block);
}
ufs: convert to new aops Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:19 +00:00
[PATCH] mark address_space_operations const Same as with already do with the file operations: keep them in .rodata and prevents people from doing runtime patching. Signed-off-by: Christoph Hellwig <hch@lst.de> Cc: Steven French <sfrench@us.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-28 11:26:44 +00:00
const struct address_space_operations ufs_aops = {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
.readpage = ufs_readpage,
.writepage = ufs_writepage,
ufs: convert to new aops Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 08:25:19 +00:00
.write_begin = ufs_write_begin,
ufs: free excessive blocks upon ->write_begin() failure/short copy Broken in "[PATCH] ufs: truncate should allocate block for last byte"; all way back in 2006. ufs_setattr() hadn't been the only user of vmtruncate() and eliminating ->truncate() method required corrections in a bunch of places. Eventually those places had migrated into ->write_begin() failure exit and ->write_end() after short copy... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-16 22:06:40 +00:00
.write_end = ufs_write_end,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
.bmap = ufs_bmap
};
[PATCH] ufs: directory and page cache: install aops This series of patches finished "bugs fixing" mentioned here http://lkml.org/lkml/2006/1/31/275 . The main bugs: * for i in `seq 1 1000`; do touch $i; done - crash system * mkdir create directory without "." and ".." entries The suggested solution is work with page cache instead of straight work with blocks. Such solution has following advantages * reduce code size and its complexity * some global locks go away * fix bugs The most part of code is stolen from ext2, because of it has similar directory structure. Patches testes with UFS1 and UFS2 file systems. This patch installs i_mapping->a_ops for directory inodes and removes some duplicated code. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:21 +00:00
static void ufs_set_inode_ops(struct inode *inode)
{
if (S_ISREG(inode->i_mode)) {
inode->i_op = &ufs_file_inode_operations;
inode->i_fop = &ufs_file_operations;
inode->i_mapping->a_ops = &ufs_aops;
} else if (S_ISDIR(inode->i_mode)) {
inode->i_op = &ufs_dir_inode_operations;
inode->i_fop = &ufs_dir_operations;
inode->i_mapping->a_ops = &ufs_aops;
} else if (S_ISLNK(inode->i_mode)) {
ufs: switch to simple_follow_link() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-05-02 14:28:56 +00:00
if (!inode->i_blocks) {
[PATCH] ufs: directory and page cache: install aops This series of patches finished "bugs fixing" mentioned here http://lkml.org/lkml/2006/1/31/275 . The main bugs: * for i in `seq 1 1000`; do touch $i; done - crash system * mkdir create directory without "." and ".." entries The suggested solution is work with page cache instead of straight work with blocks. Such solution has following advantages * reduce code size and its complexity * some global locks go away * fix bugs The most part of code is stolen from ext2, because of it has similar directory structure. Patches testes with UFS1 and UFS2 file systems. This patch installs i_mapping->a_ops for directory inodes and removes some duplicated code. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:21 +00:00
inode->i_op = &ufs_fast_symlink_inode_operations;
ufs: switch to simple_follow_link() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-05-02 14:28:56 +00:00
inode->i_link = (char *)UFS_I(inode)->i_u1.i_symlink;
} else {
ufs: add ufs speciffic ->setattr call generic setattr not longer responsible for quota transfer. use ufs_setattr for all ufs's inodes. Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org> Signed-off-by: Jan Kara <jack@suse.cz>
2010-04-14 22:56:58 +00:00
inode->i_op = &ufs_symlink_inode_operations;
[PATCH] ufs: directory and page cache: install aops This series of patches finished "bugs fixing" mentioned here http://lkml.org/lkml/2006/1/31/275 . The main bugs: * for i in `seq 1 1000`; do touch $i; done - crash system * mkdir create directory without "." and ".." entries The suggested solution is work with page cache instead of straight work with blocks. Such solution has following advantages * reduce code size and its complexity * some global locks go away * fix bugs The most part of code is stolen from ext2, because of it has similar directory structure. Patches testes with UFS1 and UFS2 file systems. This patch installs i_mapping->a_ops for directory inodes and removes some duplicated code. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:21 +00:00
inode->i_mapping->a_ops = &ufs_aops;
}
} else
init_special_inode(inode, inode->i_mode,
ufs_get_inode_dev(inode->i_sb, UFS_I(inode)));
}
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
static int ufs1_read_inode(struct inode *inode, struct ufs_inode *ufs_inode)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
struct ufs_inode_info *ufsi = UFS_I(inode);
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
struct super_block *sb = inode->i_sb;
ufs: propagate umode_t Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-26 06:49:13 +00:00
umode_t mode;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* Copy data to the in-core inode.
*/
inode->i_mode = mode = fs16_to_cpu(sb, ufs_inode->ui_mode);
filesystems: add set_nlink() Replace remaining direct i_nlink updates with a new set_nlink() updater function. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Tested-by: Toshiyuki Okajima <toshi.okajima@jp.fujitsu.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2011-10-28 12:13:29 +00:00
set_nlink(inode, fs16_to_cpu(sb, ufs_inode->ui_nlink));
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
if (inode->i_nlink == 0) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs_error (sb, "ufs_read_inode", "inode %lu has zero nlink\n", inode->i_ino);
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
return -1;
}
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* Linux now has 32-bit uid and gid, so we can support EFT.
*/
userns: Convert the ufs filesystem to use kuid/kgid where appropriate Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Serge Hallyn <serge.hallyn@canonical.com> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
2012-02-10 20:21:22 +00:00
i_uid_write(inode, ufs_get_inode_uid(sb, ufs_inode));
i_gid_write(inode, ufs_get_inode_gid(sb, ufs_inode));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
inode->i_size = fs64_to_cpu(sb, ufs_inode->ui_size);
inode->i_atime.tv_sec = fs32_to_cpu(sb, ufs_inode->ui_atime.tv_sec);
inode->i_ctime.tv_sec = fs32_to_cpu(sb, ufs_inode->ui_ctime.tv_sec);
inode->i_mtime.tv_sec = fs32_to_cpu(sb, ufs_inode->ui_mtime.tv_sec);
inode->i_mtime.tv_nsec = 0;
inode->i_atime.tv_nsec = 0;
inode->i_ctime.tv_nsec = 0;
inode->i_blocks = fs32_to_cpu(sb, ufs_inode->ui_blocks);
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
inode->i_generation = fs32_to_cpu(sb, ufs_inode->ui_gen);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufsi->i_flags = fs32_to_cpu(sb, ufs_inode->ui_flags);
ufsi->i_shadow = fs32_to_cpu(sb, ufs_inode->ui_u3.ui_sun.ui_shadow);
ufsi->i_oeftflag = fs32_to_cpu(sb, ufs_inode->ui_u3.ui_sun.ui_oeftflag);
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
if (S_ISCHR(mode) || S_ISBLK(mode) || inode->i_blocks) {
ufs: don't truncate longer ufs2 fast symlinks ufs2 fast symlinks can be twice as long as ufs ones, however the code was using the ufs size in various places. Fix that so ufs2 symlinks over 60 characters aren't truncated. Note that we copy the entire area instead of using the maxsymlinklen field from the superblock. This way we will be more robust against corruption (of the superblock). While we are at it, use memcpy instead of open-coding it with for loops. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:49 +00:00
memcpy(ufsi->i_u1.i_data, &ufs_inode->ui_u2.ui_addr,
sizeof(ufs_inode->ui_u2.ui_addr));
[PATCH] ufs: little directory lookup optimization This patch make little optimization of ufs_find_entry like "ext2" does. Save number of page and reuse it again in the next call. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:25 +00:00
} else {
ufs: don't truncate longer ufs2 fast symlinks ufs2 fast symlinks can be twice as long as ufs ones, however the code was using the ufs size in various places. Fix that so ufs2 symlinks over 60 characters aren't truncated. Note that we copy the entire area instead of using the maxsymlinklen field from the superblock. This way we will be more robust against corruption (of the superblock). While we are at it, use memcpy instead of open-coding it with for loops. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:49 +00:00
memcpy(ufsi->i_u1.i_symlink, ufs_inode->ui_u2.ui_symlink,
ufs: ensure fast symlinks are NUL-terminated Ensure fast symlink targets are NUL-terminated, even if corrupted on-disk. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:50 +00:00
sizeof(ufs_inode->ui_u2.ui_symlink) - 1);
ufsi->i_u1.i_symlink[sizeof(ufs_inode->ui_u2.ui_symlink) - 1] = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
return 0;
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
static int ufs2_read_inode(struct inode *inode, struct ufs2_inode *ufs2_inode)
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
{
struct ufs_inode_info *ufsi = UFS_I(inode);
struct super_block *sb = inode->i_sb;
ufs: propagate umode_t Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2011-07-26 06:49:13 +00:00
umode_t mode;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[PATCH] ufs: easy debug Currently to turn on debug mode "user" has to edit ~10 files, to turn off he has to do it again. This patch introduce such changes: 1)turn on(off) debug messages via ".config" 2)remove unnecessary duplication of code 3)make "UFSD" macros more similar to function 4)fix some compiler warnings Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:24 +00:00
UFSD("Reading ufs2 inode, ino %lu\n", inode->i_ino);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* Copy data to the in-core inode.
*/
inode->i_mode = mode = fs16_to_cpu(sb, ufs2_inode->ui_mode);
filesystems: add set_nlink() Replace remaining direct i_nlink updates with a new set_nlink() updater function. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Tested-by: Toshiyuki Okajima <toshi.okajima@jp.fujitsu.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2011-10-28 12:13:29 +00:00
set_nlink(inode, fs16_to_cpu(sb, ufs2_inode->ui_nlink));
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
if (inode->i_nlink == 0) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs_error (sb, "ufs_read_inode", "inode %lu has zero nlink\n", inode->i_ino);
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
return -1;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
/*
* Linux now has 32-bit uid and gid, so we can support EFT.
*/
userns: Convert the ufs filesystem to use kuid/kgid where appropriate Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Serge Hallyn <serge.hallyn@canonical.com> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
2012-02-10 20:21:22 +00:00
i_uid_write(inode, fs32_to_cpu(sb, ufs2_inode->ui_uid));
i_gid_write(inode, fs32_to_cpu(sb, ufs2_inode->ui_gid));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
inode->i_size = fs64_to_cpu(sb, ufs2_inode->ui_size);
[PATCH] ufs2: more correct work with time This patch corrects work with time in UFS2 case. 1) According to UFS2 disk layout modification/access and so on "time" should be hold in two variables one 64bit for seconds and another 32bit for nanoseconds, at now for some unknown reason we suppose that "inode time" holds in three variables 32bit for seconds, 32bit for milliseconds and 32bit for nanoseconds. 2) We set amount of nanoseconds in "VFS inode" to 0 during read, instead of getting values from "on disk inode"(this should close http://bugzilla.kernel.org/show_bug.cgi?id=7991). Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Bjoern Jacke <bjoern@j3e.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-03-16 21:38:07 +00:00
inode->i_atime.tv_sec = fs64_to_cpu(sb, ufs2_inode->ui_atime);
inode->i_ctime.tv_sec = fs64_to_cpu(sb, ufs2_inode->ui_ctime);
inode->i_mtime.tv_sec = fs64_to_cpu(sb, ufs2_inode->ui_mtime);
inode->i_atime.tv_nsec = fs32_to_cpu(sb, ufs2_inode->ui_atimensec);
inode->i_ctime.tv_nsec = fs32_to_cpu(sb, ufs2_inode->ui_ctimensec);
inode->i_mtime.tv_nsec = fs32_to_cpu(sb, ufs2_inode->ui_mtimensec);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
inode->i_blocks = fs64_to_cpu(sb, ufs2_inode->ui_blocks);
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
inode->i_generation = fs32_to_cpu(sb, ufs2_inode->ui_gen);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufsi->i_flags = fs32_to_cpu(sb, ufs2_inode->ui_flags);
/*
ufsi->i_shadow = fs32_to_cpu(sb, ufs_inode->ui_u3.ui_sun.ui_shadow);
ufsi->i_oeftflag = fs32_to_cpu(sb, ufs_inode->ui_u3.ui_sun.ui_oeftflag);
*/
if (S_ISCHR(mode) || S_ISBLK(mode) || inode->i_blocks) {
ufs: don't truncate longer ufs2 fast symlinks ufs2 fast symlinks can be twice as long as ufs ones, however the code was using the ufs size in various places. Fix that so ufs2 symlinks over 60 characters aren't truncated. Note that we copy the entire area instead of using the maxsymlinklen field from the superblock. This way we will be more robust against corruption (of the superblock). While we are at it, use memcpy instead of open-coding it with for loops. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:49 +00:00
memcpy(ufsi->i_u1.u2_i_data, &ufs2_inode->ui_u2.ui_addr,
sizeof(ufs2_inode->ui_u2.ui_addr));
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
} else {
ufs: don't truncate longer ufs2 fast symlinks ufs2 fast symlinks can be twice as long as ufs ones, however the code was using the ufs size in various places. Fix that so ufs2 symlinks over 60 characters aren't truncated. Note that we copy the entire area instead of using the maxsymlinklen field from the superblock. This way we will be more robust against corruption (of the superblock). While we are at it, use memcpy instead of open-coding it with for loops. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:49 +00:00
memcpy(ufsi->i_u1.i_symlink, ufs2_inode->ui_u2.ui_symlink,
ufs: ensure fast symlinks are NUL-terminated Ensure fast symlink targets are NUL-terminated, even if corrupted on-disk. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:50 +00:00
sizeof(ufs2_inode->ui_u2.ui_symlink) - 1);
ufsi->i_u1.i_symlink[sizeof(ufs2_inode->ui_u2.ui_symlink) - 1] = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
return 0;
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
}
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
struct inode *ufs_iget(struct super_block *sb, unsigned long ino)
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
{
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
struct ufs_inode_info *ufsi;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
struct buffer_head * bh;
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
struct inode *inode;
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
int err;
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
UFSD("ENTER, ino %lu\n", ino);
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
if (ino < UFS_ROOTINO || ino > (uspi->s_ncg * uspi->s_ipg)) {
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
ufs_warning(sb, "ufs_read_inode", "bad inode number (%lu)\n",
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
ino);
return ERR_PTR(-EIO);
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
}
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
inode = iget_locked(sb, ino);
if (!inode)
return ERR_PTR(-ENOMEM);
if (!(inode->i_state & I_NEW))
return inode;
ufsi = UFS_I(inode);
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
bh = sb_bread(sb, uspi->s_sbbase + ufs_inotofsba(inode->i_ino));
if (!bh) {
ufs_warning(sb, "ufs_read_inode", "unable to read inode %lu\n",
inode->i_ino);
goto bad_inode;
}
if ((UFS_SB(sb)->s_flags & UFS_TYPE_MASK) == UFS_TYPE_UFS2) {
struct ufs2_inode *ufs2_inode = (struct ufs2_inode *)bh->b_data;
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
err = ufs2_read_inode(inode,
ufs2_inode + ufs_inotofsbo(inode->i_ino));
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
} else {
struct ufs_inode *ufs_inode = (struct ufs_inode *)bh->b_data;
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
err = ufs1_read_inode(inode,
ufs_inode + ufs_inotofsbo(inode->i_ino));
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
}
ufs proper handling of zero link case This patch should fix or partly fix this bug: http://bugzilla.kernel.org/show_bug.cgi?id=8276 The problem is: - if we see "zero link case" during reading inode operation, we call ufs_error(which remount fs readonly), but not "mark" inode as bad (1) - in readonly case we do not fill some data structures, which are used in read and write case (2) - VFS call ufs_delete_inode if link count is zero (3) so (1)->(3)->(2) cause oops, this patch should fix such scenario Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Jim Paris <jim@jtan.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-04-17 05:53:24 +00:00
if (err)
goto bad_inode;
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
inode->i_version++;
ufsi->i_lastfrag =
(inode->i_size + uspi->s_fsize - 1) >> uspi->s_fshift;
ufsi->i_dir_start_lookup = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufsi->i_osync = 0;
[PATCH] ufs: directory and page cache: install aops This series of patches finished "bugs fixing" mentioned here http://lkml.org/lkml/2006/1/31/275 . The main bugs: * for i in `seq 1 1000`; do touch $i; done - crash system * mkdir create directory without "." and ".." entries The suggested solution is work with page cache instead of straight work with blocks. Such solution has following advantages * reduce code size and its complexity * some global locks go away * fix bugs The most part of code is stolen from ext2, because of it has similar directory structure. Patches testes with UFS1 and UFS2 file systems. This patch installs i_mapping->a_ops for directory inodes and removes some duplicated code. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:21 +00:00
ufs_set_inode_ops(inode);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
brelse(bh);
[PATCH] ufs: easy debug Currently to turn on debug mode "user" has to edit ~10 files, to turn off he has to do it again. This patch introduce such changes: 1)turn on(off) debug messages via ".config" 2)remove unnecessary duplication of code 3)make "UFSD" macros more similar to function 4)fix some compiler warnings Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:24 +00:00
UFSD("EXIT\n");
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
unlock_new_inode(inode);
return inode;
[PATCH] ufs: ufs_read_inode cleanup Add missed ufsi->i_dir_start_lookup initialization in ufs_read_inode in UFS2 case. Also it cleans ufs_read_inode function to prevent such kind of situation in the future: it move depend on UFS type parts of code into separate functions and leaves in ufs_read_inode only generic code. It cleans code and avoids duplication. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-27 09:53:59 +00:00
bad_inode:
iget: stop UFS from using iget() and read_inode() Stop the UFS filesystem from using iget() and read_inode(). Replace ufs_read_inode() with ufs_iget(), and call that instead of iget(). ufs_iget() then uses iget_locked() directly and returns a proper error code instead of an inode in the event of an error. ufs_fill_super() returns any error incurred when getting the root inode instead of EINVAL. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-07 08:15:48 +00:00
iget_failed(inode);
return ERR_PTR(-EIO);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
static void ufs1_update_inode(struct inode *inode, struct ufs_inode *ufs_inode)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
struct super_block *sb = inode->i_sb;
struct ufs_inode_info *ufsi = UFS_I(inode);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs_inode->ui_mode = cpu_to_fs16(sb, inode->i_mode);
ufs_inode->ui_nlink = cpu_to_fs16(sb, inode->i_nlink);
userns: Convert the ufs filesystem to use kuid/kgid where appropriate Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Serge Hallyn <serge.hallyn@canonical.com> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
2012-02-10 20:21:22 +00:00
ufs_set_inode_uid(sb, ufs_inode, i_uid_read(inode));
ufs_set_inode_gid(sb, ufs_inode, i_gid_read(inode));
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs_inode->ui_size = cpu_to_fs64(sb, inode->i_size);
ufs_inode->ui_atime.tv_sec = cpu_to_fs32(sb, inode->i_atime.tv_sec);
ufs_inode->ui_atime.tv_usec = 0;
ufs_inode->ui_ctime.tv_sec = cpu_to_fs32(sb, inode->i_ctime.tv_sec);
ufs_inode->ui_ctime.tv_usec = 0;
ufs_inode->ui_mtime.tv_sec = cpu_to_fs32(sb, inode->i_mtime.tv_sec);
ufs_inode->ui_mtime.tv_usec = 0;
ufs_inode->ui_blocks = cpu_to_fs32(sb, inode->i_blocks);
ufs_inode->ui_flags = cpu_to_fs32(sb, ufsi->i_flags);
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
ufs_inode->ui_gen = cpu_to_fs32(sb, inode->i_generation);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
if ((UFS_SB(sb)->s_flags & UFS_UID_MASK) == UFS_UID_EFT) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
ufs_inode->ui_u3.ui_sun.ui_shadow = cpu_to_fs32(sb, ufsi->i_shadow);
ufs_inode->ui_u3.ui_sun.ui_oeftflag = cpu_to_fs32(sb, ufsi->i_oeftflag);
}
if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) {
/* ufs_inode->ui_u2.ui_addr.ui_db[0] = cpu_to_fs32(sb, inode->i_rdev); */
ufs_inode->ui_u2.ui_addr.ui_db[0] = ufsi->i_u1.i_data[0];
} else if (inode->i_blocks) {
ufs: don't truncate longer ufs2 fast symlinks ufs2 fast symlinks can be twice as long as ufs ones, however the code was using the ufs size in various places. Fix that so ufs2 symlinks over 60 characters aren't truncated. Note that we copy the entire area instead of using the maxsymlinklen field from the superblock. This way we will be more robust against corruption (of the superblock). While we are at it, use memcpy instead of open-coding it with for loops. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:49 +00:00
memcpy(&ufs_inode->ui_u2.ui_addr, ufsi->i_u1.i_data,
sizeof(ufs_inode->ui_u2.ui_addr));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
else {
ufs: don't truncate longer ufs2 fast symlinks ufs2 fast symlinks can be twice as long as ufs ones, however the code was using the ufs size in various places. Fix that so ufs2 symlinks over 60 characters aren't truncated. Note that we copy the entire area instead of using the maxsymlinklen field from the superblock. This way we will be more robust against corruption (of the superblock). While we are at it, use memcpy instead of open-coding it with for loops. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:49 +00:00
memcpy(&ufs_inode->ui_u2.ui_symlink, ufsi->i_u1.i_symlink,
sizeof(ufs_inode->ui_u2.ui_symlink));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
if (!inode->i_nlink)
memset (ufs_inode, 0, sizeof(struct ufs_inode));
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
}
static void ufs2_update_inode(struct inode *inode, struct ufs2_inode *ufs_inode)
{
struct super_block *sb = inode->i_sb;
struct ufs_inode_info *ufsi = UFS_I(inode);
UFSD("ENTER\n");
ufs_inode->ui_mode = cpu_to_fs16(sb, inode->i_mode);
ufs_inode->ui_nlink = cpu_to_fs16(sb, inode->i_nlink);
userns: Convert the ufs filesystem to use kuid/kgid where appropriate Cc: Evgeniy Dushistov <dushistov@mail.ru> Acked-by: Serge Hallyn <serge.hallyn@canonical.com> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
2012-02-10 20:21:22 +00:00
ufs_inode->ui_uid = cpu_to_fs32(sb, i_uid_read(inode));
ufs_inode->ui_gid = cpu_to_fs32(sb, i_gid_read(inode));
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
ufs_inode->ui_size = cpu_to_fs64(sb, inode->i_size);
[PATCH] ufs2: more correct work with time This patch corrects work with time in UFS2 case. 1) According to UFS2 disk layout modification/access and so on "time" should be hold in two variables one 64bit for seconds and another 32bit for nanoseconds, at now for some unknown reason we suppose that "inode time" holds in three variables 32bit for seconds, 32bit for milliseconds and 32bit for nanoseconds. 2) We set amount of nanoseconds in "VFS inode" to 0 during read, instead of getting values from "on disk inode"(this should close http://bugzilla.kernel.org/show_bug.cgi?id=7991). Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Cc: Bjoern Jacke <bjoern@j3e.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-03-16 21:38:07 +00:00
ufs_inode->ui_atime = cpu_to_fs64(sb, inode->i_atime.tv_sec);
ufs_inode->ui_atimensec = cpu_to_fs32(sb, inode->i_atime.tv_nsec);
ufs_inode->ui_ctime = cpu_to_fs64(sb, inode->i_ctime.tv_sec);
ufs_inode->ui_ctimensec = cpu_to_fs32(sb, inode->i_ctime.tv_nsec);
ufs_inode->ui_mtime = cpu_to_fs64(sb, inode->i_mtime.tv_sec);
ufs_inode->ui_mtimensec = cpu_to_fs32(sb, inode->i_mtime.tv_nsec);
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
ufs_inode->ui_blocks = cpu_to_fs64(sb, inode->i_blocks);
ufs_inode->ui_flags = cpu_to_fs32(sb, ufsi->i_flags);
ufs_inode->ui_gen = cpu_to_fs32(sb, inode->i_generation);
if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) {
/* ufs_inode->ui_u2.ui_addr.ui_db[0] = cpu_to_fs32(sb, inode->i_rdev); */
ufs_inode->ui_u2.ui_addr.ui_db[0] = ufsi->i_u1.u2_i_data[0];
} else if (inode->i_blocks) {
ufs: don't truncate longer ufs2 fast symlinks ufs2 fast symlinks can be twice as long as ufs ones, however the code was using the ufs size in various places. Fix that so ufs2 symlinks over 60 characters aren't truncated. Note that we copy the entire area instead of using the maxsymlinklen field from the superblock. This way we will be more robust against corruption (of the superblock). While we are at it, use memcpy instead of open-coding it with for loops. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:49 +00:00
memcpy(&ufs_inode->ui_u2.ui_addr, ufsi->i_u1.u2_i_data,
sizeof(ufs_inode->ui_u2.ui_addr));
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
} else {
ufs: don't truncate longer ufs2 fast symlinks ufs2 fast symlinks can be twice as long as ufs ones, however the code was using the ufs size in various places. Fix that so ufs2 symlinks over 60 characters aren't truncated. Note that we copy the entire area instead of using the maxsymlinklen field from the superblock. This way we will be more robust against corruption (of the superblock). While we are at it, use memcpy instead of open-coding it with for loops. Signed-off-by: Duane Griffin <duaneg@dghda.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2009-01-08 22:43:49 +00:00
memcpy(&ufs_inode->ui_u2.ui_symlink, ufsi->i_u1.i_symlink,
sizeof(ufs_inode->ui_u2.ui_symlink));
[PATCH] ufs2 write: inodes write This patch adds into write inode path function to write UFS2 inode, and modifys allocate inode path to allocate and init additional inode chunks. Also some cleanups: - remove not used parameters in some functions - remove i_gen field from ufs_inode_info structure, there is i_generation in inode structure with same purposes. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-02-12 08:54:31 +00:00
}
if (!inode->i_nlink)
memset (ufs_inode, 0, sizeof(struct ufs2_inode));
UFSD("EXIT\n");
}
static int ufs_update_inode(struct inode * inode, int do_sync)
{
struct super_block *sb = inode->i_sb;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
struct buffer_head * bh;
UFSD("ENTER, ino %lu\n", inode->i_ino);
if (inode->i_ino < UFS_ROOTINO ||
inode->i_ino > (uspi->s_ncg * uspi->s_ipg)) {
ufs_warning (sb, "ufs_read_inode", "bad inode number (%lu)\n", inode->i_ino);
return -1;
}
bh = sb_bread(sb, ufs_inotofsba(inode->i_ino));
if (!bh) {
ufs_warning (sb, "ufs_read_inode", "unable to read inode %lu\n", inode->i_ino);
return -1;
}
if (uspi->fs_magic == UFS2_MAGIC) {
struct ufs2_inode *ufs2_inode = (struct ufs2_inode *)bh->b_data;
ufs2_update_inode(inode,
ufs2_inode + ufs_inotofsbo(inode->i_ino));
} else {
struct ufs_inode *ufs_inode = (struct ufs_inode *) bh->b_data;
ufs1_update_inode(inode, ufs_inode + ufs_inotofsbo(inode->i_ino));
}
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
mark_buffer_dirty(bh);
if (do_sync)
sync_dirty_buffer(bh);
brelse (bh);
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
[PATCH] ufs: easy debug Currently to turn on debug mode "user" has to edit ~10 files, to turn off he has to do it again. This patch introduce such changes: 1)turn on(off) debug messages via ".config" 2)remove unnecessary duplication of code 3)make "UFSD" macros more similar to function 4)fix some compiler warnings Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 12:47:24 +00:00
UFSD("EXIT\n");
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
return 0;
}
pass writeback_control to ->write_inode This gives the filesystem more information about the writeback that is happening. Trond requested this for the NFS unstable write handling, and other filesystems might benefit from this too by beeing able to distinguish between the different callers in more detail. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-03-05 08:21:37 +00:00
int ufs_write_inode(struct inode *inode, struct writeback_control *wbc)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
ufs: kill more lock_ufs() calls a) move it inside ufs_truncate() b) ufs_free_inode() doesn't need it - it's serialized on ->s_lock c) ufs_write_inode() doesn't need it either (and can be called without it anyway). Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-16 06:35:14 +00:00
return ufs_update_inode(inode, wbc->sync_mode == WB_SYNC_ALL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
int ufs_sync_inode (struct inode *inode)
{
return ufs_update_inode (inode, 1);
}
switch ufs to ->evict_inode() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-05 23:40:56 +00:00
void ufs_evict_inode(struct inode * inode)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
{
switch ufs to ->evict_inode() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-05 23:40:56 +00:00
int want_delete = 0;
if (!inode->i_nlink && !is_bad_inode(inode))
want_delete = 1;
[PATCH] ufs: truncate should allocate block for last byte This patch fixes buggy behaviour of UFS in such kind of scenario: open(, O_TRUNC...) ftruncate(, 1024) ftruncate(, 0) Such a scenario causes ufs_panic and remount read-only. This happen because of according to specification UFS should always allocate block for last byte, and many parts of our implementation rely on this, but `ufs_truncate' doesn't care about this. To make possible return error code and to know about old size, this patch removes `truncate' from ufs inode_operations and uses `setattr' method to call ufs_truncate. Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-01 11:36:24 +00:00
mm + fs: store shadow entries in page cache Reclaim will be leaving shadow entries in the page cache radix tree upon evicting the real page. As those pages are found from the LRU, an iput() can lead to the inode being freed concurrently. At this point, reclaim must no longer install shadow pages because the inode freeing code needs to ensure the page tree is really empty. Add an address_space flag, AS_EXITING, that the inode freeing code sets under the tree lock before doing the final truncate. Reclaim will check for this flag before installing shadow pages. Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Reviewed-by: Rik van Riel <riel@redhat.com> Reviewed-by: Minchan Kim <minchan@kernel.org> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Bob Liu <bob.liu@oracle.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Dave Chinner <david@fromorbit.com> Cc: Greg Thelen <gthelen@google.com> Cc: Hugh Dickins <hughd@google.com> Cc: Jan Kara <jack@suse.cz> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Luigi Semenzato <semenzato@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Metin Doslu <metin@citusdata.com> Cc: Michel Lespinasse <walken@google.com> Cc: Ozgun Erdogan <ozgun@citusdata.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Roman Gushchin <klamm@yandex-team.ru> Cc: Ryan Mallon <rmallon@gmail.com> Cc: Tejun Heo <tj@kernel.org> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-04-03 21:47:49 +00:00
truncate_inode_pages_final(&inode->i_data);
switch ufs to ->evict_inode() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-05 23:40:56 +00:00
if (want_delete) {
inode->i_size = 0;
ufs: switch ufs_evict_inode() to trimmed-down variant of ufs_truncate() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-16 22:04:16 +00:00
if (inode->i_blocks)
ufs_truncate_blocks(inode);
switch ufs to ->evict_inode() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-05 23:40:56 +00:00
}
invalidate_inode_buffers(inode);
vfs: Rename end_writeback() to clear_inode() After we moved inode_sync_wait() from end_writeback() it doesn't make sense to call the function end_writeback() anymore. Rename it to clear_inode() which well says what the function really does - set I_CLEAR flag. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
2012-05-03 12:48:02 +00:00
clear_inode(inode);
switch ufs to ->evict_inode() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-06-05 23:40:56 +00:00
ufs: kill more lock_ufs() calls a) move it inside ufs_truncate() b) ufs_free_inode() doesn't need it - it's serialized on ->s_lock c) ufs_write_inode() doesn't need it either (and can be called without it anyway). Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-16 06:35:14 +00:00
if (want_delete)
ufs: fix deadlocks introduced by sb mutex merge Commit 0244756edc4b ("ufs: sb mutex merge + mutex_destroy") introduces deadlocks in ufs_new_inode() and ufs_free_inode(). Most callers of that functions acqure the mutex by themselves and ufs_{new,free}_inode() do that via lock_ufs(), i.e we have an unavoidable double lock. The patch proposes to resolve the issue by making sure that ufs_{new,free}_inode() are not called with the mutex held. Found by Linux Driver Verification project (linuxtesting.org). Cc: stable@vger.kernel.org # 3.16 Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2014-09-02 07:40:17 +00:00
ufs_free_inode(inode);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 22:20:36 +00:00
}
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
ufs: unify the logics for collecting adjacent data blocks to free open-coded in several places... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 06:18:54 +00:00
struct to_free {
struct inode *inode;
u64 to;
unsigned count;
};
static inline void free_data(struct to_free *ctx, u64 from, unsigned count)
{
if (ctx->count && ctx->to != from) {
ufs_free_blocks(ctx->inode, ctx->to - ctx->count, ctx->count);
ctx->count = 0;
}
ctx->count += count;
ctx->to = from + count;
}
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
#define DIRECT_BLOCK ((inode->i_size + uspi->s_bsize - 1) >> uspi->s_bshift)
#define DIRECT_FRAGMENT ((inode->i_size + uspi->s_fsize - 1) >> uspi->s_fshift)
static void ufs_trunc_direct(struct inode *inode)
{
struct ufs_inode_info *ufsi = UFS_I(inode);
struct super_block * sb;
struct ufs_sb_private_info * uspi;
void *p;
u64 frag1, frag2, frag3, frag4, block1, block2;
ufs: unify the logics for collecting adjacent data blocks to free open-coded in several places... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 06:18:54 +00:00
struct to_free ctx = {.inode = inode};
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
unsigned i, tmp;
UFSD("ENTER: ino %lu\n", inode->i_ino);
sb = inode->i_sb;
uspi = UFS_SB(sb)->s_uspi;
frag1 = DIRECT_FRAGMENT;
frag4 = min_t(u64, UFS_NDIR_FRAGMENT, ufsi->i_lastfrag);
frag2 = ((frag1 & uspi->s_fpbmask) ? ((frag1 | uspi->s_fpbmask) + 1) : frag1);
frag3 = frag4 & ~uspi->s_fpbmask;
block1 = block2 = 0;
if (frag2 > frag3) {
frag2 = frag4;
frag3 = frag4 = 0;
} else if (frag2 < frag3) {
block1 = ufs_fragstoblks (frag2);
block2 = ufs_fragstoblks (frag3);
}
UFSD("ino %lu, frag1 %llu, frag2 %llu, block1 %llu, block2 %llu,"
" frag3 %llu, frag4 %llu\n", inode->i_ino,
(unsigned long long)frag1, (unsigned long long)frag2,
(unsigned long long)block1, (unsigned long long)block2,
(unsigned long long)frag3, (unsigned long long)frag4);
if (frag1 >= frag2)
goto next1;
/*
* Free first free fragments
*/
p = ufs_get_direct_data_ptr(uspi, ufsi, ufs_fragstoblks(frag1));
tmp = ufs_data_ptr_to_cpu(sb, p);
if (!tmp )
ufs_panic (sb, "ufs_trunc_direct", "internal error");
frag2 -= frag1;
frag1 = ufs_fragnum (frag1);
ufs_free_fragments(inode, tmp + frag1, frag2);
mark_inode_dirty(inode);
next1:
/*
* Free whole blocks
*/
for (i = block1 ; i < block2; i++) {
p = ufs_get_direct_data_ptr(uspi, ufsi, i);
tmp = ufs_data_ptr_to_cpu(sb, p);
if (!tmp)
continue;
write_seqlock(&ufsi->meta_lock);
ufs_data_ptr_clear(uspi, p);
write_sequnlock(&ufsi->meta_lock);
ufs: unify the logics for collecting adjacent data blocks to free open-coded in several places... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 06:18:54 +00:00
free_data(&ctx, tmp, uspi->s_fpb);
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
mark_inode_dirty(inode);
}
ufs: unify the logics for collecting adjacent data blocks to free open-coded in several places... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 06:18:54 +00:00
free_data(&ctx, 0, 0);
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
if (frag3 >= frag4)
goto next3;
/*
* Free last free fragments
*/
p = ufs_get_direct_data_ptr(uspi, ufsi, ufs_fragstoblks(frag3));
tmp = ufs_data_ptr_to_cpu(sb, p);
if (!tmp )
ufs_panic(sb, "ufs_truncate_direct", "internal error");
frag4 = ufs_fragnum (frag4);
write_seqlock(&ufsi->meta_lock);
ufs_data_ptr_clear(uspi, p);
write_sequnlock(&ufsi->meta_lock);
ufs_free_fragments (inode, tmp, frag4);
mark_inode_dirty(inode);
next3:
UFSD("EXIT: ino %lu\n", inode->i_ino);
}
split ufs_truncate_branch() into full- and partial-branch variants Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 21:11:49 +00:00
static void free_full_branch(struct inode *inode, int depth, void *p)
{
struct super_block *sb = inode->i_sb;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
struct ufs_inode_info *ufsi = UFS_I(inode);
struct ufs_buffer_head *ubh;
u64 tmp;
unsigned i;
tmp = ufs_data_ptr_to_cpu(sb, p);
if (!tmp)
return;
ubh = ubh_bread (sb, tmp, uspi->s_bsize);
if (!ubh) {
write_seqlock(&ufsi->meta_lock);
ufs_data_ptr_clear(uspi, p);
write_sequnlock(&ufsi->meta_lock);
return;
}
if (--depth) {
for (i = 0 ; i < uspi->s_apb ; i++) {
void *ind = ubh_get_data_ptr(uspi, ubh, i);
free_full_branch(inode, depth, ind);
ubh_mark_buffer_dirty(ubh);
}
} else {
struct to_free ctx = {.inode = inode};
for (i = 0; i < uspi->s_apb; i++) {
void *ind = ubh_get_data_ptr(uspi, ubh, i);
tmp = ufs_data_ptr_to_cpu(sb, ind);
if (!tmp)
continue;
write_seqlock(&UFS_I(inode)->meta_lock);
ufs_data_ptr_clear(uspi, ind);
write_sequnlock(&UFS_I(inode)->meta_lock);
ubh_mark_buffer_dirty(ubh);
free_data(&ctx, tmp, uspi->s_fpb);
mark_inode_dirty(inode);
}
free_data(&ctx, 0, 0);
}
tmp = ufs_data_ptr_to_cpu(sb, p);
write_seqlock(&ufsi->meta_lock);
ufs_data_ptr_clear(uspi, p);
write_sequnlock(&ufsi->meta_lock);
ubh_bforget(ubh);
ufs_free_blocks(inode, tmp, uspi->s_fpb);
mark_inode_dirty(inode);
}
ufs_trunc_branch(): kill recursion turn recursion into a pair of loops Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 23:13:02 +00:00
static void free_branch_tail(struct inode *inode, unsigned from, struct ufs_buffer_head *ubh, int depth)
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
{
ufs_trunc_dindirect(): pass the number of blocks to keep same as the previous two. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 18:21:09 +00:00
struct super_block *sb = inode->i_sb;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
unsigned i;
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
unify ufs_trunc_..indirect() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:33:47 +00:00
if (--depth) {
ufs_trunc_branch(): kill recursion turn recursion into a pair of loops Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 23:13:02 +00:00
for (i = from; i < uspi->s_apb ; i++) {
ufs_trunc_branch(): separate the calls with non-NULL offsets Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 20:13:56 +00:00
void *ind = ubh_get_data_ptr(uspi, ubh, i);
split ufs_truncate_branch() into full- and partial-branch variants Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 21:11:49 +00:00
free_full_branch(inode, depth, ind);
ufs_trunc_branch(): separate the calls with non-NULL offsets Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 20:13:56 +00:00
ubh_mark_buffer_dirty(ubh);
}
unify ufs_trunc_..indirect() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:33:47 +00:00
} else {
ufs: unify the logics for collecting adjacent data blocks to free open-coded in several places... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 06:18:54 +00:00
struct to_free ctx = {.inode = inode};
unify ufs_trunc_..indirect() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:33:47 +00:00
for (i = from; i < uspi->s_apb; i++) {
void *ind = ubh_get_data_ptr(uspi, ubh, i);
ufs_trunc_branch(): kill recursion turn recursion into a pair of loops Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 23:13:02 +00:00
u64 tmp = ufs_data_ptr_to_cpu(sb, ind);
unify ufs_trunc_..indirect() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:33:47 +00:00
if (!tmp)
continue;
write_seqlock(&UFS_I(inode)->meta_lock);
ufs_data_ptr_clear(uspi, ind);
write_sequnlock(&UFS_I(inode)->meta_lock);
ubh_mark_buffer_dirty(ubh);
ufs: unify the logics for collecting adjacent data blocks to free open-coded in several places... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 06:18:54 +00:00
free_data(&ctx, tmp, uspi->s_fpb);
unify ufs_trunc_..indirect() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:33:47 +00:00
mark_inode_dirty(inode);
}
ufs: unify the logics for collecting adjacent data blocks to free open-coded in several places... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 06:18:54 +00:00
free_data(&ctx, 0, 0);
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
}
unify ufs_trunc_..indirect() Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:33:47 +00:00
if (IS_SYNC(inode) && ubh_buffer_dirty(ubh))
ubh_sync_block(ubh);
ubh_brelse(ubh);
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
}
static int ufs_alloc_lastblock(struct inode *inode, loff_t size)
{
int err = 0;
struct super_block *sb = inode->i_sb;
struct address_space *mapping = inode->i_mapping;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
unsigned i, end;
sector_t lastfrag;
struct page *lastpage;
struct buffer_head *bh;
u64 phys64;
lastfrag = (size + uspi->s_fsize - 1) >> uspi->s_fshift;
if (!lastfrag)
goto out;
lastfrag--;
lastpage = ufs_get_locked_page(mapping, lastfrag >>
(PAGE_CACHE_SHIFT - inode->i_blkbits));
if (IS_ERR(lastpage)) {
err = -EIO;
goto out;
}
end = lastfrag & ((1 << (PAGE_CACHE_SHIFT - inode->i_blkbits)) - 1);
bh = page_buffers(lastpage);
for (i = 0; i < end; ++i)
bh = bh->b_this_page;
err = ufs_getfrag_block(inode, lastfrag, bh, 1);
if (unlikely(err))
goto out_unlock;
if (buffer_new(bh)) {
clear_buffer_new(bh);
unmap_underlying_metadata(bh->b_bdev,
bh->b_blocknr);
/*
* we do not zeroize fragment, because of
* if it maped to hole, it already contains zeroes
*/
set_buffer_uptodate(bh);
mark_buffer_dirty(bh);
set_page_dirty(lastpage);
}
if (lastfrag >= UFS_IND_FRAGMENT) {
end = uspi->s_fpb - ufs_fragnum(lastfrag) - 1;
phys64 = bh->b_blocknr + 1;
for (i = 0; i < end; ++i) {
bh = sb_getblk(sb, i + phys64);
lock_buffer(bh);
memset(bh->b_data, 0, sb->s_blocksize);
set_buffer_uptodate(bh);
mark_buffer_dirty(bh);
unlock_buffer(bh);
sync_dirty_buffer(bh);
brelse(bh);
}
}
out_unlock:
ufs_put_locked_page(lastpage);
out:
return err;
}
static void __ufs_truncate_blocks(struct inode *inode)
{
struct ufs_inode_info *ufsi = UFS_I(inode);
struct super_block *sb = inode->i_sb;
struct ufs_sb_private_info *uspi = UFS_SB(sb)->s_uspi;
ufs_trunc_dindirect(): pass the number of blocks to keep same as the previous two. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 18:21:09 +00:00
unsigned offsets[4];
ufs: beginning of __ufs_truncate_block() massage Use ufs_block_to_path() to find the cutoff path in the block pointers' tree. For now just use the information about the depth (to bypass the fully preserved subtrees); subsequent commits will use the information about actual path. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 05:10:03 +00:00
int depth = ufs_block_to_path(inode, DIRECT_BLOCK, offsets);
ufs_trunc_..indirect(): more massage towards unifying Instead of manually checking that the array contains only zeroes, find the position of the last non-zero (in __ufs_truncate(), where we can conveniently do that) and use that to tell if there's any non-zero in the array tail passed to ufs_trunc_...indirect(). The goal of all that clumsiness is to get fold these functions together. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 18:55:50 +00:00
int depth2;
__ufs_trunc_blocks(): turn the part after switch into a loop ... and turn the switch into if (), since all cases with depth != 1 have just become identical. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:47:17 +00:00
unsigned i;
ufs_trunc_branch(): kill recursion turn recursion into a pair of loops Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 23:13:02 +00:00
struct ufs_buffer_head *ubh[3];
void *p;
u64 block;
ufs_trunc_..indirect(): more massage towards unifying Instead of manually checking that the array contains only zeroes, find the position of the last non-zero (in __ufs_truncate(), where we can conveniently do that) and use that to tell if there's any non-zero in the array tail passed to ufs_trunc_...indirect(). The goal of all that clumsiness is to get fold these functions together. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 18:55:50 +00:00
if (!depth)
return;
/* find the last non-zero in offsets[] */
for (depth2 = depth - 1; depth2; depth2--)
if (offsets[depth2])
break;
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
mutex_lock(&ufsi->truncate_mutex);
__ufs_trunc_blocks(): turn the part after switch into a loop ... and turn the switch into if (), since all cases with depth != 1 have just become identical. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:47:17 +00:00
if (depth == 1) {
ufs: beginning of __ufs_truncate_block() massage Use ufs_block_to_path() to find the cutoff path in the block pointers' tree. For now just use the information about the depth (to bypass the fully preserved subtrees); subsequent commits will use the information about actual path. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 05:10:03 +00:00
ufs_trunc_direct(inode);
__ufs_trunc_blocks(): turn the part after switch into a loop ... and turn the switch into if (), since all cases with depth != 1 have just become identical. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:47:17 +00:00
offsets[0] = UFS_IND_BLOCK;
} else {
ufs_trunc_branch(): kill recursion turn recursion into a pair of loops Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 23:13:02 +00:00
/* get the blocks that should be partially emptied */
p = ufs_get_direct_data_ptr(uspi, ufsi, offsets[0]);
for (i = 0; i < depth2; i++) {
offsets[i]++; /* next branch is fully freed */
block = ufs_data_ptr_to_cpu(sb, p);
if (!block)
break;
ubh[i] = ubh_bread(sb, block, uspi->s_bsize);
if (!ubh[i]) {
write_seqlock(&ufsi->meta_lock);
ufs_data_ptr_clear(uspi, p);
write_sequnlock(&ufsi->meta_lock);
break;
}
p = ubh_get_data_ptr(uspi, ubh[i], offsets[i + 1]);
}
while (i--) {
ubh_mark_buffer_dirty(ubh[i]);
free_branch_tail(inode, offsets[i + 1], ubh[i], depth - i - 1);
}
__ufs_trunc_blocks(): turn the part after switch into a loop ... and turn the switch into if (), since all cases with depth != 1 have just become identical. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:47:17 +00:00
}
for (i = offsets[0]; i <= UFS_TIND_BLOCK; i++) {
split ufs_truncate_branch() into full- and partial-branch variants Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 21:11:49 +00:00
free_full_branch(inode, i - UFS_IND_BLOCK + 1,
__ufs_trunc_blocks(): turn the part after switch into a loop ... and turn the switch into if (), since all cases with depth != 1 have just become identical. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-18 19:47:17 +00:00
ufs_get_direct_data_ptr(uspi, ufsi, i));
ufs: beginning of __ufs_truncate_block() massage Use ufs_block_to_path() to find the cutoff path in the block pointers' tree. For now just use the information about the depth (to bypass the fully preserved subtrees); subsequent commits will use the information about actual path. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 05:10:03 +00:00
}
ufs: move truncate code into inode.c It is closely tied to block pointers handling there, can benefit from existing helpers, etc. - no point keeping them apart. Trimmed the trailing whitespaces in inode.c at the same time. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-06-17 16:44:14 +00:00
ufsi->i_lastfrag = DIRECT_FRAGMENT;
mutex_unlock(&ufsi->truncate_mutex);
}
static int ufs_truncate(struct inode *inode, loff_t size)
{
int err = 0;
UFSD("ENTER: ino %lu, i_size: %llu, old_i_size: %llu\n",
inode->i_ino, (unsigned long long)size,
(unsigned long long)i_size_read(inode));
if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
S_ISLNK(inode->i_mode)))
return -EINVAL;
if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
return -EPERM;
err = ufs_alloc_lastblock(inode, size);
if (err)
goto out;
block_truncate_page(inode->i_mapping, size, ufs_getfrag_block);
truncate_setsize(inode, size);
__ufs_truncate_blocks(inode);
inode->i_mtime = inode->i_ctime = CURRENT_TIME_SEC;
mark_inode_dirty(inode);
out:
UFSD("EXIT: err %d\n", err);
return err;
}
void ufs_truncate_blocks(struct inode *inode)
{
if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
S_ISLNK(inode->i_mode)))
return;
if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
return;
__ufs_truncate_blocks(inode);
}
int ufs_setattr(struct dentry *dentry, struct iattr *attr)
{
struct inode *inode = d_inode(dentry);
unsigned int ia_valid = attr->ia_valid;
int error;
error = inode_change_ok(inode, attr);
if (error)
return error;
if (ia_valid & ATTR_SIZE && attr->ia_size != inode->i_size) {
error = ufs_truncate(inode, attr->ia_size);
if (error)
return error;
}
setattr_copy(inode, attr);
mark_inode_dirty(inode);
return 0;
}
const struct inode_operations ufs_file_inode_operations = {
.setattr = ufs_setattr,
};
Reference in New Issue Copy Permalink