2018-09-12 09:16:07 +08:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
2013-11-10 23:13:19 +08:00
|
|
|
/*
|
|
|
|
|
* fs/f2fs/inline.c
|
|
|
|
|
* Copyright (c) 2013, Intel Corporation
|
|
|
|
|
* Authors: Huajun Li <huajun.li@intel.com>
|
|
|
|
|
* Haicheng Li <haicheng.li@intel.com>
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#include <linux/fs.h>
|
|
|
|
|
#include <linux/f2fs_fs.h>
|
2020-05-23 15:30:11 +08:00
|
|
|
#include <linux/fiemap.h>
|
2013-11-10 23:13:19 +08:00
|
|
|
|
|
|
|
|
#include "f2fs.h"
|
2015-10-16 02:34:49 +08:00
|
|
|
#include "node.h"
|
2020-06-29 20:13:13 +08:00
|
|
|
#include <trace/events/f2fs.h>
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2015-04-24 01:27:21 +08:00
|
|
|
bool f2fs_may_inline_data(struct inode *inode)
|
2013-11-10 23:13:19 +08:00
|
|
|
{
|
2014-10-07 08:39:50 +08:00
|
|
|
if (f2fs_is_atomic_file(inode))
|
|
|
|
|
return false;
|
|
|
|
|
|
2015-03-19 13:23:48 +08:00
|
|
|
if (!S_ISREG(inode->i_mode) && !S_ISLNK(inode->i_mode))
|
2013-11-10 23:13:19 +08:00
|
|
|
return false;
|
|
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
if (i_size_read(inode) > MAX_INLINE_DATA(inode))
|
2014-11-12 06:10:01 +08:00
|
|
|
return false;
|
|
|
|
|
|
f2fs: refactor read path to allow multiple postprocessing steps
Currently f2fs's ->readpage() and ->readpages() assume that either the
data undergoes no postprocessing, or decryption only. But with
fs-verity, there will be an additional authenticity verification step,
and it may be needed either by itself, or combined with decryption.
To support this, store a 'struct bio_post_read_ctx' in ->bi_private
which contains a work struct, a bitmask of postprocessing steps that are
enabled, and an indicator of the current step. The bio completion
routine, if there was no I/O error, enqueues the first postprocessing
step. When that completes, it continues to the next step. Pages that
fail any postprocessing step have PageError set. Once all steps have
completed, pages without PageError set are set Uptodate, and all pages
are unlocked.
Also replace f2fs_encrypted_file() with a new function
f2fs_post_read_required() in places like direct I/O and garbage
collection that really should be testing whether the file needs special
I/O processing, not whether it is encrypted specifically.
This may also be useful for other future f2fs features such as
compression.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-19 02:09:48 +08:00
|
|
|
if (f2fs_post_read_required(inode))
|
2015-04-22 11:39:58 +08:00
|
|
|
return false;
|
|
|
|
|
|
2013-11-10 23:13:19 +08:00
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2015-04-24 01:27:21 +08:00
|
|
|
bool f2fs_may_inline_dentry(struct inode *inode)
|
|
|
|
|
{
|
|
|
|
|
if (!test_opt(F2FS_I_SB(inode), INLINE_DENTRY))
|
|
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
if (!S_ISDIR(inode->i_mode))
|
|
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
void f2fs_do_read_inline_data(struct page *page, struct page *ipage)
|
2013-11-10 23:13:19 +08:00
|
|
|
{
|
2017-07-19 00:19:05 +08:00
|
|
|
struct inode *inode = page->mapping->host;
|
2013-11-10 23:13:19 +08:00
|
|
|
void *src_addr, *dst_addr;
|
|
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
if (PageUptodate(page))
|
|
|
|
|
return;
|
2013-12-30 18:36:23 +08:00
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
f2fs_bug_on(F2FS_P_SB(page), page->index);
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
zero_user_segment(page, MAX_INLINE_DATA(inode), PAGE_SIZE);
|
2013-11-10 23:13:19 +08:00
|
|
|
|
|
|
|
|
/* Copy the whole inline data block */
|
2017-07-19 00:19:05 +08:00
|
|
|
src_addr = inline_data_addr(inode, ipage);
|
2014-10-19 14:41:38 +08:00
|
|
|
dst_addr = kmap_atomic(page);
|
2017-07-19 00:19:05 +08:00
|
|
|
memcpy(dst_addr, src_addr, MAX_INLINE_DATA(inode));
|
2014-10-27 13:59:27 +08:00
|
|
|
flush_dcache_page(page);
|
2014-10-19 14:41:38 +08:00
|
|
|
kunmap_atomic(dst_addr);
|
2016-07-01 09:49:15 +08:00
|
|
|
if (!PageUptodate(page))
|
|
|
|
|
SetPageUptodate(page);
|
2014-10-24 10:48:09 +08:00
|
|
|
}
|
|
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
void f2fs_truncate_inline_inode(struct inode *inode,
|
|
|
|
|
struct page *ipage, u64 from)
|
2015-01-26 20:22:55 +08:00
|
|
|
{
|
2015-03-10 13:16:25 +08:00
|
|
|
void *addr;
|
|
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
if (from >= MAX_INLINE_DATA(inode))
|
2017-03-10 20:43:20 +08:00
|
|
|
return;
|
2015-03-10 13:16:25 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
addr = inline_data_addr(inode, ipage);
|
2015-03-10 13:16:25 +08:00
|
|
|
|
2018-12-25 17:43:42 +08:00
|
|
|
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
|
2017-07-19 00:19:05 +08:00
|
|
|
memset(addr + from, 0, MAX_INLINE_DATA(inode) - from);
|
2016-05-21 07:32:49 +08:00
|
|
|
set_page_dirty(ipage);
|
2017-03-10 20:43:20 +08:00
|
|
|
|
|
|
|
|
if (from == 0)
|
|
|
|
|
clear_inode_flag(inode, FI_DATA_EXIST);
|
2015-01-26 20:22:55 +08:00
|
|
|
}
|
|
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
int f2fs_read_inline_data(struct inode *inode, struct page *page)
|
|
|
|
|
{
|
|
|
|
|
struct page *ipage;
|
|
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(F2FS_I_SB(inode), inode->i_ino);
|
2014-10-24 10:48:09 +08:00
|
|
|
if (IS_ERR(ipage)) {
|
|
|
|
|
unlock_page(page);
|
|
|
|
|
return PTR_ERR(ipage);
|
|
|
|
|
}
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
if (!f2fs_has_inline_data(inode)) {
|
|
|
|
|
f2fs_put_page(ipage, 1);
|
|
|
|
|
return -EAGAIN;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (page->index)
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 20:29:47 +08:00
|
|
|
zero_user_segment(page, 0, PAGE_SIZE);
|
2014-10-24 10:48:09 +08:00
|
|
|
else
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_do_read_inline_data(page, ipage);
|
2014-10-24 10:48:09 +08:00
|
|
|
|
2016-07-01 09:49:15 +08:00
|
|
|
if (!PageUptodate(page))
|
|
|
|
|
SetPageUptodate(page);
|
2014-10-24 10:48:09 +08:00
|
|
|
f2fs_put_page(ipage, 1);
|
|
|
|
|
unlock_page(page);
|
2013-11-10 23:13:19 +08:00
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page)
|
2013-11-10 23:13:19 +08:00
|
|
|
{
|
|
|
|
|
struct f2fs_io_info fio = {
|
2015-04-24 05:38:15 +08:00
|
|
|
.sbi = F2FS_I_SB(dn->inode),
|
2017-09-29 13:59:38 +08:00
|
|
|
.ino = dn->inode->i_ino,
|
2013-11-10 23:13:19 +08:00
|
|
|
.type = DATA,
|
2016-06-06 03:31:55 +08:00
|
|
|
.op = REQ_OP_WRITE,
|
2016-11-01 21:40:10 +08:00
|
|
|
.op_flags = REQ_SYNC | REQ_PRIO,
|
2015-04-24 05:38:15 +08:00
|
|
|
.page = page,
|
2015-04-24 03:04:33 +08:00
|
|
|
.encrypted_page = NULL,
|
2017-08-02 23:21:48 +08:00
|
|
|
.io_type = FS_DATA_IO,
|
2013-11-10 23:13:19 +08:00
|
|
|
};
|
2018-07-17 00:02:17 +08:00
|
|
|
struct node_info ni;
|
2014-11-26 03:34:02 +08:00
|
|
|
int dirty, err;
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
if (!f2fs_exist_data(dn->inode))
|
|
|
|
|
goto clear_out;
|
2014-08-19 05:41:11 +08:00
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
err = f2fs_reserve_block(dn, 0);
|
2014-04-16 13:22:50 +08:00
|
|
|
if (err)
|
2014-10-24 10:48:09 +08:00
|
|
|
return err;
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2018-07-17 00:02:17 +08:00
|
|
|
err = f2fs_get_node_info(fio.sbi, dn->nid, &ni);
|
|
|
|
|
if (err) {
|
2019-09-03 10:06:25 +08:00
|
|
|
f2fs_truncate_data_blocks_range(dn, 1);
|
2018-07-17 00:02:17 +08:00
|
|
|
f2fs_put_dnode(dn);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fio.version = ni.version;
|
|
|
|
|
|
f2fs: fix to do sanity check with reserved blkaddr of inline inode
As Wen Xu reported in bugzilla, after image was injected with random data
by fuzzing, inline inode would contain invalid reserved blkaddr, then
during inline conversion, we will encounter illegal memory accessing
reported by KASAN, the root cause of this is when writing out converted
inline page, we will use invalid reserved blkaddr to update sit bitmap,
result in accessing memory beyond sit bitmap boundary.
In order to fix this issue, let's do sanity check with reserved block
address of inline inode to avoid above condition.
https://bugzilla.kernel.org/show_bug.cgi?id=200179
[ 1428.846352] BUG: KASAN: use-after-free in update_sit_entry+0x80/0x7f0
[ 1428.846618] Read of size 4 at addr ffff880194483540 by task a.out/2741
[ 1428.846855] CPU: 0 PID: 2741 Comm: a.out Tainted: G W 4.17.0+ #1
[ 1428.846858] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
[ 1428.846860] Call Trace:
[ 1428.846868] dump_stack+0x71/0xab
[ 1428.846875] print_address_description+0x6b/0x290
[ 1428.846881] kasan_report+0x28e/0x390
[ 1428.846888] ? update_sit_entry+0x80/0x7f0
[ 1428.846898] update_sit_entry+0x80/0x7f0
[ 1428.846906] f2fs_allocate_data_block+0x6db/0xc70
[ 1428.846914] ? f2fs_get_node_info+0x14f/0x590
[ 1428.846920] do_write_page+0xc8/0x150
[ 1428.846928] f2fs_outplace_write_data+0xfe/0x210
[ 1428.846935] ? f2fs_do_write_node_page+0x170/0x170
[ 1428.846941] ? radix_tree_tag_clear+0xff/0x130
[ 1428.846946] ? __mod_node_page_state+0x22/0xa0
[ 1428.846951] ? inc_zone_page_state+0x54/0x100
[ 1428.846956] ? __test_set_page_writeback+0x336/0x5d0
[ 1428.846964] f2fs_convert_inline_page+0x407/0x6d0
[ 1428.846971] ? f2fs_read_inline_data+0x3b0/0x3b0
[ 1428.846978] ? __get_node_page+0x335/0x6b0
[ 1428.846987] f2fs_convert_inline_inode+0x41b/0x500
[ 1428.846994] ? f2fs_convert_inline_page+0x6d0/0x6d0
[ 1428.847000] ? kasan_unpoison_shadow+0x31/0x40
[ 1428.847005] ? kasan_kmalloc+0xa6/0xd0
[ 1428.847024] f2fs_file_mmap+0x79/0xc0
[ 1428.847029] mmap_region+0x58b/0x880
[ 1428.847037] ? arch_get_unmapped_area+0x370/0x370
[ 1428.847042] do_mmap+0x55b/0x7a0
[ 1428.847048] vm_mmap_pgoff+0x16f/0x1c0
[ 1428.847055] ? vma_is_stack_for_current+0x50/0x50
[ 1428.847062] ? __fsnotify_update_child_dentry_flags.part.1+0x160/0x160
[ 1428.847068] ? do_sys_open+0x206/0x2a0
[ 1428.847073] ? __fget+0xb4/0x100
[ 1428.847079] ksys_mmap_pgoff+0x278/0x360
[ 1428.847085] ? find_mergeable_anon_vma+0x50/0x50
[ 1428.847091] do_syscall_64+0x73/0x160
[ 1428.847098] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847102] RIP: 0033:0x7fb1430766ba
[ 1428.847103] Code: 89 f5 41 54 49 89 fc 55 53 74 35 49 63 e8 48 63 da 4d 89 f9 49 89 e8 4d 63 d6 48 89 da 4c 89 ee 4c 89 e7 b8 09 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 56 5b 5d 41 5c 41 5d 41 5e 41 5f c3 0f 1f 00
[ 1428.847162] RSP: 002b:00007ffc651d9388 EFLAGS: 00000246 ORIG_RAX: 0000000000000009
[ 1428.847167] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007fb1430766ba
[ 1428.847170] RDX: 0000000000000001 RSI: 0000000000001000 RDI: 0000000000000000
[ 1428.847173] RBP: 0000000000000003 R08: 0000000000000003 R09: 0000000000000000
[ 1428.847176] R10: 0000000000008002 R11: 0000000000000246 R12: 0000000000000000
[ 1428.847179] R13: 0000000000001000 R14: 0000000000008002 R15: 0000000000000000
[ 1428.847252] Allocated by task 2683:
[ 1428.847372] kasan_kmalloc+0xa6/0xd0
[ 1428.847380] kmem_cache_alloc+0xc8/0x1e0
[ 1428.847385] getname_flags+0x73/0x2b0
[ 1428.847390] user_path_at_empty+0x1d/0x40
[ 1428.847395] vfs_statx+0xc1/0x150
[ 1428.847401] __do_sys_newlstat+0x7e/0xd0
[ 1428.847405] do_syscall_64+0x73/0x160
[ 1428.847411] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847466] Freed by task 2683:
[ 1428.847566] __kasan_slab_free+0x137/0x190
[ 1428.847571] kmem_cache_free+0x85/0x1e0
[ 1428.847575] filename_lookup+0x191/0x280
[ 1428.847580] vfs_statx+0xc1/0x150
[ 1428.847585] __do_sys_newlstat+0x7e/0xd0
[ 1428.847590] do_syscall_64+0x73/0x160
[ 1428.847596] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847648] The buggy address belongs to the object at ffff880194483300
which belongs to the cache names_cache of size 4096
[ 1428.847946] The buggy address is located 576 bytes inside of
4096-byte region [ffff880194483300, ffff880194484300)
[ 1428.848234] The buggy address belongs to the page:
[ 1428.848366] page:ffffea0006512000 count:1 mapcount:0 mapping:ffff8801f3586380 index:0x0 compound_mapcount: 0
[ 1428.848606] flags: 0x17fff8000008100(slab|head)
[ 1428.848737] raw: 017fff8000008100 dead000000000100 dead000000000200 ffff8801f3586380
[ 1428.848931] raw: 0000000000000000 0000000000070007 00000001ffffffff 0000000000000000
[ 1428.849122] page dumped because: kasan: bad access detected
[ 1428.849305] Memory state around the buggy address:
[ 1428.849436] ffff880194483400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849620] ffff880194483480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849804] >ffff880194483500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849985] ^
[ 1428.850120] ffff880194483580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.850303] ffff880194483600: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.850498] ==================================================================
Reported-by: Wen Xu <wen.xu@gatech.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-30 18:13:40 +08:00
|
|
|
if (unlikely(dn->data_blkaddr != NEW_ADDR)) {
|
|
|
|
|
f2fs_put_dnode(dn);
|
|
|
|
|
set_sbi_flag(fio.sbi, SBI_NEED_FSCK);
|
2019-06-18 17:48:42 +08:00
|
|
|
f2fs_warn(fio.sbi, "%s: corrupted inline inode ino=%lx, i_addr[0]:0x%x, run fsck to fix.",
|
|
|
|
|
__func__, dn->inode->i_ino, dn->data_blkaddr);
|
2019-06-20 11:36:14 +08:00
|
|
|
return -EFSCORRUPTED;
|
f2fs: fix to do sanity check with reserved blkaddr of inline inode
As Wen Xu reported in bugzilla, after image was injected with random data
by fuzzing, inline inode would contain invalid reserved blkaddr, then
during inline conversion, we will encounter illegal memory accessing
reported by KASAN, the root cause of this is when writing out converted
inline page, we will use invalid reserved blkaddr to update sit bitmap,
result in accessing memory beyond sit bitmap boundary.
In order to fix this issue, let's do sanity check with reserved block
address of inline inode to avoid above condition.
https://bugzilla.kernel.org/show_bug.cgi?id=200179
[ 1428.846352] BUG: KASAN: use-after-free in update_sit_entry+0x80/0x7f0
[ 1428.846618] Read of size 4 at addr ffff880194483540 by task a.out/2741
[ 1428.846855] CPU: 0 PID: 2741 Comm: a.out Tainted: G W 4.17.0+ #1
[ 1428.846858] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
[ 1428.846860] Call Trace:
[ 1428.846868] dump_stack+0x71/0xab
[ 1428.846875] print_address_description+0x6b/0x290
[ 1428.846881] kasan_report+0x28e/0x390
[ 1428.846888] ? update_sit_entry+0x80/0x7f0
[ 1428.846898] update_sit_entry+0x80/0x7f0
[ 1428.846906] f2fs_allocate_data_block+0x6db/0xc70
[ 1428.846914] ? f2fs_get_node_info+0x14f/0x590
[ 1428.846920] do_write_page+0xc8/0x150
[ 1428.846928] f2fs_outplace_write_data+0xfe/0x210
[ 1428.846935] ? f2fs_do_write_node_page+0x170/0x170
[ 1428.846941] ? radix_tree_tag_clear+0xff/0x130
[ 1428.846946] ? __mod_node_page_state+0x22/0xa0
[ 1428.846951] ? inc_zone_page_state+0x54/0x100
[ 1428.846956] ? __test_set_page_writeback+0x336/0x5d0
[ 1428.846964] f2fs_convert_inline_page+0x407/0x6d0
[ 1428.846971] ? f2fs_read_inline_data+0x3b0/0x3b0
[ 1428.846978] ? __get_node_page+0x335/0x6b0
[ 1428.846987] f2fs_convert_inline_inode+0x41b/0x500
[ 1428.846994] ? f2fs_convert_inline_page+0x6d0/0x6d0
[ 1428.847000] ? kasan_unpoison_shadow+0x31/0x40
[ 1428.847005] ? kasan_kmalloc+0xa6/0xd0
[ 1428.847024] f2fs_file_mmap+0x79/0xc0
[ 1428.847029] mmap_region+0x58b/0x880
[ 1428.847037] ? arch_get_unmapped_area+0x370/0x370
[ 1428.847042] do_mmap+0x55b/0x7a0
[ 1428.847048] vm_mmap_pgoff+0x16f/0x1c0
[ 1428.847055] ? vma_is_stack_for_current+0x50/0x50
[ 1428.847062] ? __fsnotify_update_child_dentry_flags.part.1+0x160/0x160
[ 1428.847068] ? do_sys_open+0x206/0x2a0
[ 1428.847073] ? __fget+0xb4/0x100
[ 1428.847079] ksys_mmap_pgoff+0x278/0x360
[ 1428.847085] ? find_mergeable_anon_vma+0x50/0x50
[ 1428.847091] do_syscall_64+0x73/0x160
[ 1428.847098] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847102] RIP: 0033:0x7fb1430766ba
[ 1428.847103] Code: 89 f5 41 54 49 89 fc 55 53 74 35 49 63 e8 48 63 da 4d 89 f9 49 89 e8 4d 63 d6 48 89 da 4c 89 ee 4c 89 e7 b8 09 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 56 5b 5d 41 5c 41 5d 41 5e 41 5f c3 0f 1f 00
[ 1428.847162] RSP: 002b:00007ffc651d9388 EFLAGS: 00000246 ORIG_RAX: 0000000000000009
[ 1428.847167] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007fb1430766ba
[ 1428.847170] RDX: 0000000000000001 RSI: 0000000000001000 RDI: 0000000000000000
[ 1428.847173] RBP: 0000000000000003 R08: 0000000000000003 R09: 0000000000000000
[ 1428.847176] R10: 0000000000008002 R11: 0000000000000246 R12: 0000000000000000
[ 1428.847179] R13: 0000000000001000 R14: 0000000000008002 R15: 0000000000000000
[ 1428.847252] Allocated by task 2683:
[ 1428.847372] kasan_kmalloc+0xa6/0xd0
[ 1428.847380] kmem_cache_alloc+0xc8/0x1e0
[ 1428.847385] getname_flags+0x73/0x2b0
[ 1428.847390] user_path_at_empty+0x1d/0x40
[ 1428.847395] vfs_statx+0xc1/0x150
[ 1428.847401] __do_sys_newlstat+0x7e/0xd0
[ 1428.847405] do_syscall_64+0x73/0x160
[ 1428.847411] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847466] Freed by task 2683:
[ 1428.847566] __kasan_slab_free+0x137/0x190
[ 1428.847571] kmem_cache_free+0x85/0x1e0
[ 1428.847575] filename_lookup+0x191/0x280
[ 1428.847580] vfs_statx+0xc1/0x150
[ 1428.847585] __do_sys_newlstat+0x7e/0xd0
[ 1428.847590] do_syscall_64+0x73/0x160
[ 1428.847596] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847648] The buggy address belongs to the object at ffff880194483300
which belongs to the cache names_cache of size 4096
[ 1428.847946] The buggy address is located 576 bytes inside of
4096-byte region [ffff880194483300, ffff880194484300)
[ 1428.848234] The buggy address belongs to the page:
[ 1428.848366] page:ffffea0006512000 count:1 mapcount:0 mapping:ffff8801f3586380 index:0x0 compound_mapcount: 0
[ 1428.848606] flags: 0x17fff8000008100(slab|head)
[ 1428.848737] raw: 017fff8000008100 dead000000000100 dead000000000200 ffff8801f3586380
[ 1428.848931] raw: 0000000000000000 0000000000070007 00000001ffffffff 0000000000000000
[ 1428.849122] page dumped because: kasan: bad access detected
[ 1428.849305] Memory state around the buggy address:
[ 1428.849436] ffff880194483400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849620] ffff880194483480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849804] >ffff880194483500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849985] ^
[ 1428.850120] ffff880194483580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.850303] ffff880194483600: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.850498] ==================================================================
Reported-by: Wen Xu <wen.xu@gatech.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-30 18:13:40 +08:00
|
|
|
}
|
|
|
|
|
|
2016-02-03 10:26:13 +08:00
|
|
|
f2fs_bug_on(F2FS_P_SB(page), PageWriteback(page));
|
2014-10-24 10:48:09 +08:00
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_do_read_inline_data(page, dn->inode_page);
|
2015-07-25 15:29:17 +08:00
|
|
|
set_page_dirty(page);
|
|
|
|
|
|
2014-11-26 03:34:02 +08:00
|
|
|
/* clear dirty state */
|
|
|
|
|
dirty = clear_page_dirty_for_io(page);
|
|
|
|
|
|
2013-11-10 23:13:19 +08:00
|
|
|
/* write data page to try to make data consistent */
|
|
|
|
|
set_page_writeback(page);
|
2018-04-12 14:09:04 +08:00
|
|
|
ClearPageError(page);
|
f2fs: trace old block address for CoWed page
This patch enables to trace old block address of CoWed page for better
debugging.
f2fs_submit_page_mbio: dev = (1,0), ino = 1, page_index = 0x1d4f0, oldaddr = 0xfe8ab, newaddr = 0xfee90 rw = WRITE_SYNC, type = NODE
f2fs_submit_page_mbio: dev = (1,0), ino = 1, page_index = 0x1d4f8, oldaddr = 0xfe8b0, newaddr = 0xfee91 rw = WRITE_SYNC, type = NODE
f2fs_submit_page_mbio: dev = (1,0), ino = 1, page_index = 0x1d4fa, oldaddr = 0xfe8ae, newaddr = 0xfee92 rw = WRITE_SYNC, type = NODE
f2fs_submit_page_mbio: dev = (1,0), ino = 134824, page_index = 0x96, oldaddr = 0xf049b, newaddr = 0x2bbe rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 134824, page_index = 0x97, oldaddr = 0xf049c, newaddr = 0x2bbf rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 134824, page_index = 0x98, oldaddr = 0xf049d, newaddr = 0x2bc0 rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 135260, page_index = 0x47, oldaddr = 0xffffffff, newaddr = 0xf2631 rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 135260, page_index = 0x48, oldaddr = 0xffffffff, newaddr = 0xf2632 rw = WRITE, type = DATA
f2fs_submit_page_mbio: dev = (1,0), ino = 135260, page_index = 0x49, oldaddr = 0xffffffff, newaddr = 0xf2633 rw = WRITE, type = DATA
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:36:38 +08:00
|
|
|
fio.old_blkaddr = dn->data_blkaddr;
|
2017-03-25 08:05:13 +08:00
|
|
|
set_inode_flag(dn->inode, FI_HOT_DATA);
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_outplace_write_data(dn, &fio);
|
2018-12-25 17:43:42 +08:00
|
|
|
f2fs_wait_on_page_writeback(page, DATA, true, true);
|
2016-10-11 22:57:01 +08:00
|
|
|
if (dirty) {
|
2014-11-26 03:34:02 +08:00
|
|
|
inode_dec_dirty_pages(dn->inode);
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_remove_dirty_inode(dn->inode);
|
2016-10-11 22:57:01 +08:00
|
|
|
}
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2014-11-26 09:27:38 +08:00
|
|
|
/* this converted inline_data should be recovered. */
|
2016-05-21 01:13:22 +08:00
|
|
|
set_inode_flag(dn->inode, FI_APPEND_WRITE);
|
2014-11-26 09:27:38 +08:00
|
|
|
|
2013-11-10 23:13:19 +08:00
|
|
|
/* clear inline data and flag after data writeback */
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_truncate_inline_inode(dn->inode, dn->inode_page, 0);
|
2016-01-25 21:57:05 +08:00
|
|
|
clear_inline_node(dn->inode_page);
|
2014-10-24 10:48:09 +08:00
|
|
|
clear_out:
|
|
|
|
|
stat_dec_inline_inode(dn->inode);
|
2017-03-10 20:43:20 +08:00
|
|
|
clear_inode_flag(dn->inode, FI_INLINE_DATA);
|
2014-10-24 10:48:09 +08:00
|
|
|
f2fs_put_dnode(dn);
|
|
|
|
|
return 0;
|
2013-11-10 23:13:19 +08:00
|
|
|
}
|
|
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
int f2fs_convert_inline_inode(struct inode *inode)
|
2013-11-10 23:13:19 +08:00
|
|
|
{
|
2014-10-24 10:48:09 +08:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
|
|
|
|
|
struct dnode_of_data dn;
|
|
|
|
|
struct page *ipage, *page;
|
|
|
|
|
int err = 0;
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2015-12-23 03:09:35 +08:00
|
|
|
if (!f2fs_has_inline_data(inode))
|
|
|
|
|
return 0;
|
|
|
|
|
|
f2fs: fix to avoid inconsistent quota data
[ Upstream commit 25fb04dbce6a0e165d28fd1fa8a1d7018c637fe8 ]
Occasionally, quota data may be corrupted detected by fsck:
Info: checkpoint state = 45 : crc compacted_summary unmount
[QUOTA WARNING] Usage inconsistent for ID 0:actual (1543036928, 762) != expected (1543032832, 762)
[ASSERT] (fsck_chk_quota_files:1986) --> Quota file is missing or invalid quota file content found.
[QUOTA WARNING] Usage inconsistent for ID 0:actual (1352478720, 344) != expected (1352474624, 344)
[ASSERT] (fsck_chk_quota_files:1986) --> Quota file is missing or invalid quota file content found.
[FSCK] Unreachable nat entries [Ok..] [0x0]
[FSCK] SIT valid block bitmap checking [Ok..]
[FSCK] Hard link checking for regular file [Ok..] [0x0]
[FSCK] valid_block_count matching with CP [Ok..] [0xdf299]
[FSCK] valid_node_count matcing with CP (de lookup) [Ok..] [0x2b01]
[FSCK] valid_node_count matcing with CP (nat lookup) [Ok..] [0x2b01]
[FSCK] valid_inode_count matched with CP [Ok..] [0x2665]
[FSCK] free segment_count matched with CP [Ok..] [0xcb04]
[FSCK] next block offset is free [Ok..]
[FSCK] fixing SIT types
[FSCK] other corrupted bugs [Fail]
The root cause is:
If we open file w/ readonly flag, disk quota info won't be initialized
for this file, however, following mmap() will force to convert inline
inode via f2fs_convert_inline_inode(), which may increase block usage
for this inode w/o updating quota data, it causes inconsistent disk quota
info.
The issue will happen in following stack:
open(file, O_RDONLY)
mmap(file)
- f2fs_convert_inline_inode
- f2fs_convert_inline_page
- f2fs_reserve_block
- f2fs_reserve_new_block
- f2fs_reserve_new_blocks
- f2fs_i_blocks_write
- dquot_claim_block
inode->i_blocks increase, but the dqb_curspace keep the size for the dquots
is NULL.
To fix this issue, let's call dquot_initialize() anyway in both
f2fs_truncate() and f2fs_convert_inline_inode() functions to avoid potential
inconsistent quota data issue.
Fixes: 0abd675e97e6 ("f2fs: support plain user/group quota")
Signed-off-by: Daiyue Zhang <zhangdaiyue1@huawei.com>
Signed-off-by: Dehe Gu <gudehe@huawei.com>
Signed-off-by: Junchao Jiang <jiangjunchao1@huawei.com>
Signed-off-by: Ge Qiu <qiuge@huawei.com>
Signed-off-by: Yi Chen <chenyi77@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2021-01-28 17:02:56 +08:00
|
|
|
err = dquot_initialize(inode);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
|
2016-04-30 07:11:53 +08:00
|
|
|
page = f2fs_grab_cache_page(inode->i_mapping, 0, false);
|
2014-10-24 10:48:09 +08:00
|
|
|
if (!page)
|
|
|
|
|
return -ENOMEM;
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
f2fs_lock_op(sbi);
|
|
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(sbi, inode->i_ino);
|
2014-10-24 10:48:09 +08:00
|
|
|
if (IS_ERR(ipage)) {
|
2014-11-18 08:06:55 +08:00
|
|
|
err = PTR_ERR(ipage);
|
|
|
|
|
goto out;
|
2014-08-08 07:32:25 +08:00
|
|
|
}
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
set_new_dnode(&dn, inode, ipage, ipage, 0);
|
|
|
|
|
|
|
|
|
|
if (f2fs_has_inline_data(inode))
|
|
|
|
|
err = f2fs_convert_inline_page(&dn, page);
|
|
|
|
|
|
|
|
|
|
f2fs_put_dnode(&dn);
|
2014-11-18 08:06:55 +08:00
|
|
|
out:
|
2014-10-24 10:48:09 +08:00
|
|
|
f2fs_unlock_op(sbi);
|
|
|
|
|
|
|
|
|
|
f2fs_put_page(page, 1);
|
2015-12-23 05:23:35 +08:00
|
|
|
|
2021-03-04 17:21:18 +08:00
|
|
|
if (!err)
|
|
|
|
|
f2fs_balance_fs(sbi, dn.node_changed);
|
2015-12-23 05:23:35 +08:00
|
|
|
|
2013-11-10 23:13:19 +08:00
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
int f2fs_write_inline_data(struct inode *inode, struct page *page)
|
2013-11-10 23:13:19 +08:00
|
|
|
{
|
|
|
|
|
void *src_addr, *dst_addr;
|
|
|
|
|
struct dnode_of_data dn;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
set_new_dnode(&dn, inode, NULL, NULL, 0);
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
err = f2fs_get_dnode_of_data(&dn, 0, LOOKUP_NODE);
|
2013-11-10 23:13:19 +08:00
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
|
2014-10-16 01:16:54 +08:00
|
|
|
if (!f2fs_has_inline_data(inode)) {
|
2014-10-24 10:48:09 +08:00
|
|
|
f2fs_put_dnode(&dn);
|
|
|
|
|
return -EAGAIN;
|
2014-10-16 01:16:54 +08:00
|
|
|
}
|
|
|
|
|
|
2014-10-24 10:48:09 +08:00
|
|
|
f2fs_bug_on(F2FS_I_SB(inode), page->index);
|
|
|
|
|
|
2018-12-25 17:43:42 +08:00
|
|
|
f2fs_wait_on_page_writeback(dn.inode_page, NODE, true, true);
|
2014-10-19 14:41:38 +08:00
|
|
|
src_addr = kmap_atomic(page);
|
2017-07-19 00:19:05 +08:00
|
|
|
dst_addr = inline_data_addr(inode, dn.inode_page);
|
|
|
|
|
memcpy(dst_addr, src_addr, MAX_INLINE_DATA(inode));
|
2014-10-19 14:41:38 +08:00
|
|
|
kunmap_atomic(src_addr);
|
2016-05-21 07:32:49 +08:00
|
|
|
set_page_dirty(dn.inode_page);
|
2013-11-10 23:13:19 +08:00
|
|
|
|
2017-12-05 09:25:25 +08:00
|
|
|
f2fs_clear_page_cache_dirty_tag(page);
|
2017-09-11 15:30:28 +08:00
|
|
|
|
2016-05-21 01:13:22 +08:00
|
|
|
set_inode_flag(inode, FI_APPEND_WRITE);
|
|
|
|
|
set_inode_flag(inode, FI_DATA_EXIST);
|
2014-10-24 10:48:09 +08:00
|
|
|
|
2016-01-25 21:57:05 +08:00
|
|
|
clear_inline_node(dn.inode_page);
|
2013-11-10 23:13:19 +08:00
|
|
|
f2fs_put_dnode(&dn);
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
2013-12-26 11:49:48 +08:00
|
|
|
|
2020-07-06 18:23:36 +08:00
|
|
|
int f2fs_recover_inline_data(struct inode *inode, struct page *npage)
|
2013-12-26 11:49:48 +08:00
|
|
|
{
|
2014-09-03 06:31:18 +08:00
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
|
2013-12-26 11:49:48 +08:00
|
|
|
struct f2fs_inode *ri = NULL;
|
|
|
|
|
void *src_addr, *dst_addr;
|
|
|
|
|
struct page *ipage;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* The inline_data recovery policy is as follows.
|
|
|
|
|
* [prev.] [next] of inline_data flag
|
|
|
|
|
* o o -> recover inline_data
|
|
|
|
|
* o x -> remove inline_data, and then recover data blocks
|
|
|
|
|
* x o -> remove inline_data, and then recover inline_data
|
|
|
|
|
* x x -> recover data blocks
|
|
|
|
|
*/
|
|
|
|
|
if (IS_INODE(npage))
|
|
|
|
|
ri = F2FS_INODE(npage);
|
|
|
|
|
|
|
|
|
|
if (f2fs_has_inline_data(inode) &&
|
2014-08-08 07:57:17 +08:00
|
|
|
ri && (ri->i_inline & F2FS_INLINE_DATA)) {
|
2013-12-26 11:49:48 +08:00
|
|
|
process_inline:
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(sbi, inode->i_ino);
|
2020-07-06 18:23:36 +08:00
|
|
|
if (IS_ERR(ipage))
|
|
|
|
|
return PTR_ERR(ipage);
|
2013-12-26 11:49:48 +08:00
|
|
|
|
2018-12-25 17:43:42 +08:00
|
|
|
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
|
2014-04-29 16:28:32 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
src_addr = inline_data_addr(inode, npage);
|
|
|
|
|
dst_addr = inline_data_addr(inode, ipage);
|
|
|
|
|
memcpy(dst_addr, src_addr, MAX_INLINE_DATA(inode));
|
2014-10-24 10:48:09 +08:00
|
|
|
|
2016-05-21 01:13:22 +08:00
|
|
|
set_inode_flag(inode, FI_INLINE_DATA);
|
|
|
|
|
set_inode_flag(inode, FI_DATA_EXIST);
|
2014-10-24 10:48:09 +08:00
|
|
|
|
2016-05-21 07:32:49 +08:00
|
|
|
set_page_dirty(ipage);
|
2013-12-26 11:49:48 +08:00
|
|
|
f2fs_put_page(ipage, 1);
|
2020-07-06 18:23:36 +08:00
|
|
|
return 1;
|
2013-12-26 11:49:48 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (f2fs_has_inline_data(inode)) {
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(sbi, inode->i_ino);
|
2020-07-06 18:23:36 +08:00
|
|
|
if (IS_ERR(ipage))
|
|
|
|
|
return PTR_ERR(ipage);
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_truncate_inline_inode(inode, ipage, 0);
|
2017-03-10 20:43:20 +08:00
|
|
|
clear_inode_flag(inode, FI_INLINE_DATA);
|
2013-12-26 11:49:48 +08:00
|
|
|
f2fs_put_page(ipage, 1);
|
2014-08-08 07:57:17 +08:00
|
|
|
} else if (ri && (ri->i_inline & F2FS_INLINE_DATA)) {
|
2020-07-06 18:23:36 +08:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
ret = f2fs_truncate_blocks(inode, 0, false);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2013-12-26 11:49:48 +08:00
|
|
|
goto process_inline;
|
|
|
|
|
}
|
2020-07-06 18:23:36 +08:00
|
|
|
return 0;
|
2013-12-26 11:49:48 +08:00
|
|
|
}
|
2014-09-24 18:17:53 +08:00
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
const struct f2fs_filename *fname,
|
|
|
|
|
struct page **res_page)
|
2014-09-24 18:17:53 +08:00
|
|
|
{
|
|
|
|
|
struct f2fs_sb_info *sbi = F2FS_SB(dir->i_sb);
|
|
|
|
|
struct f2fs_dir_entry *de;
|
2014-10-19 13:52:52 +08:00
|
|
|
struct f2fs_dentry_ptr d;
|
2014-10-14 08:26:14 +08:00
|
|
|
struct page *ipage;
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry;
|
2014-09-24 18:17:53 +08:00
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(sbi, dir->i_ino);
|
2016-05-26 05:29:11 +08:00
|
|
|
if (IS_ERR(ipage)) {
|
|
|
|
|
*res_page = ipage;
|
2014-09-24 18:17:53 +08:00
|
|
|
return NULL;
|
2016-05-26 05:29:11 +08:00
|
|
|
}
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
inline_dentry = inline_data_addr(dir, ipage);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
make_dentry_ptr_inline(dir, &d, inline_dentry);
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
de = f2fs_find_target_dentry(&d, fname, NULL);
|
2014-09-24 18:17:53 +08:00
|
|
|
unlock_page(ipage);
|
2014-10-14 08:26:14 +08:00
|
|
|
if (de)
|
|
|
|
|
*res_page = ipage;
|
|
|
|
|
else
|
|
|
|
|
f2fs_put_page(ipage, 0);
|
|
|
|
|
|
2014-09-24 18:17:53 +08:00
|
|
|
return de;
|
|
|
|
|
}
|
|
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
|
2014-09-24 18:17:53 +08:00
|
|
|
struct page *ipage)
|
|
|
|
|
{
|
2014-10-19 14:06:41 +08:00
|
|
|
struct f2fs_dentry_ptr d;
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry;
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
inline_dentry = inline_data_addr(inode, ipage);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
make_dentry_ptr_inline(inode, &d, inline_dentry);
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_do_make_empty_dir(inode, parent, &d);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
|
|
|
|
set_page_dirty(ipage);
|
|
|
|
|
|
|
|
|
|
/* update i_size to MAX_INLINE_DATA */
|
2017-07-19 00:19:05 +08:00
|
|
|
if (i_size_read(inode) < MAX_INLINE_DATA(inode))
|
|
|
|
|
f2fs_i_size_write(inode, MAX_INLINE_DATA(inode));
|
2014-09-24 18:17:53 +08:00
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2015-07-14 18:14:06 +08:00
|
|
|
/*
|
|
|
|
|
* NOTE: ipage is grabbed by caller, but if any error occurs, we should
|
|
|
|
|
* release ipage in this function.
|
|
|
|
|
*/
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
static int f2fs_move_inline_dirents(struct inode *dir, struct page *ipage,
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry)
|
2014-09-24 18:17:53 +08:00
|
|
|
{
|
|
|
|
|
struct page *page;
|
|
|
|
|
struct dnode_of_data dn;
|
|
|
|
|
struct f2fs_dentry_block *dentry_blk;
|
2017-07-16 15:08:54 +08:00
|
|
|
struct f2fs_dentry_ptr src, dst;
|
2014-09-24 18:17:53 +08:00
|
|
|
int err;
|
|
|
|
|
|
2019-12-04 11:07:53 +08:00
|
|
|
page = f2fs_grab_cache_page(dir->i_mapping, 0, true);
|
2015-07-14 18:14:06 +08:00
|
|
|
if (!page) {
|
|
|
|
|
f2fs_put_page(ipage, 1);
|
2014-09-24 18:17:53 +08:00
|
|
|
return -ENOMEM;
|
2015-07-14 18:14:06 +08:00
|
|
|
}
|
2014-09-24 18:17:53 +08:00
|
|
|
|
|
|
|
|
set_new_dnode(&dn, dir, ipage, NULL, 0);
|
|
|
|
|
err = f2fs_reserve_block(&dn, 0);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out;
|
|
|
|
|
|
f2fs: fix to do sanity check with reserved blkaddr of inline inode
As Wen Xu reported in bugzilla, after image was injected with random data
by fuzzing, inline inode would contain invalid reserved blkaddr, then
during inline conversion, we will encounter illegal memory accessing
reported by KASAN, the root cause of this is when writing out converted
inline page, we will use invalid reserved blkaddr to update sit bitmap,
result in accessing memory beyond sit bitmap boundary.
In order to fix this issue, let's do sanity check with reserved block
address of inline inode to avoid above condition.
https://bugzilla.kernel.org/show_bug.cgi?id=200179
[ 1428.846352] BUG: KASAN: use-after-free in update_sit_entry+0x80/0x7f0
[ 1428.846618] Read of size 4 at addr ffff880194483540 by task a.out/2741
[ 1428.846855] CPU: 0 PID: 2741 Comm: a.out Tainted: G W 4.17.0+ #1
[ 1428.846858] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
[ 1428.846860] Call Trace:
[ 1428.846868] dump_stack+0x71/0xab
[ 1428.846875] print_address_description+0x6b/0x290
[ 1428.846881] kasan_report+0x28e/0x390
[ 1428.846888] ? update_sit_entry+0x80/0x7f0
[ 1428.846898] update_sit_entry+0x80/0x7f0
[ 1428.846906] f2fs_allocate_data_block+0x6db/0xc70
[ 1428.846914] ? f2fs_get_node_info+0x14f/0x590
[ 1428.846920] do_write_page+0xc8/0x150
[ 1428.846928] f2fs_outplace_write_data+0xfe/0x210
[ 1428.846935] ? f2fs_do_write_node_page+0x170/0x170
[ 1428.846941] ? radix_tree_tag_clear+0xff/0x130
[ 1428.846946] ? __mod_node_page_state+0x22/0xa0
[ 1428.846951] ? inc_zone_page_state+0x54/0x100
[ 1428.846956] ? __test_set_page_writeback+0x336/0x5d0
[ 1428.846964] f2fs_convert_inline_page+0x407/0x6d0
[ 1428.846971] ? f2fs_read_inline_data+0x3b0/0x3b0
[ 1428.846978] ? __get_node_page+0x335/0x6b0
[ 1428.846987] f2fs_convert_inline_inode+0x41b/0x500
[ 1428.846994] ? f2fs_convert_inline_page+0x6d0/0x6d0
[ 1428.847000] ? kasan_unpoison_shadow+0x31/0x40
[ 1428.847005] ? kasan_kmalloc+0xa6/0xd0
[ 1428.847024] f2fs_file_mmap+0x79/0xc0
[ 1428.847029] mmap_region+0x58b/0x880
[ 1428.847037] ? arch_get_unmapped_area+0x370/0x370
[ 1428.847042] do_mmap+0x55b/0x7a0
[ 1428.847048] vm_mmap_pgoff+0x16f/0x1c0
[ 1428.847055] ? vma_is_stack_for_current+0x50/0x50
[ 1428.847062] ? __fsnotify_update_child_dentry_flags.part.1+0x160/0x160
[ 1428.847068] ? do_sys_open+0x206/0x2a0
[ 1428.847073] ? __fget+0xb4/0x100
[ 1428.847079] ksys_mmap_pgoff+0x278/0x360
[ 1428.847085] ? find_mergeable_anon_vma+0x50/0x50
[ 1428.847091] do_syscall_64+0x73/0x160
[ 1428.847098] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847102] RIP: 0033:0x7fb1430766ba
[ 1428.847103] Code: 89 f5 41 54 49 89 fc 55 53 74 35 49 63 e8 48 63 da 4d 89 f9 49 89 e8 4d 63 d6 48 89 da 4c 89 ee 4c 89 e7 b8 09 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 56 5b 5d 41 5c 41 5d 41 5e 41 5f c3 0f 1f 00
[ 1428.847162] RSP: 002b:00007ffc651d9388 EFLAGS: 00000246 ORIG_RAX: 0000000000000009
[ 1428.847167] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007fb1430766ba
[ 1428.847170] RDX: 0000000000000001 RSI: 0000000000001000 RDI: 0000000000000000
[ 1428.847173] RBP: 0000000000000003 R08: 0000000000000003 R09: 0000000000000000
[ 1428.847176] R10: 0000000000008002 R11: 0000000000000246 R12: 0000000000000000
[ 1428.847179] R13: 0000000000001000 R14: 0000000000008002 R15: 0000000000000000
[ 1428.847252] Allocated by task 2683:
[ 1428.847372] kasan_kmalloc+0xa6/0xd0
[ 1428.847380] kmem_cache_alloc+0xc8/0x1e0
[ 1428.847385] getname_flags+0x73/0x2b0
[ 1428.847390] user_path_at_empty+0x1d/0x40
[ 1428.847395] vfs_statx+0xc1/0x150
[ 1428.847401] __do_sys_newlstat+0x7e/0xd0
[ 1428.847405] do_syscall_64+0x73/0x160
[ 1428.847411] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847466] Freed by task 2683:
[ 1428.847566] __kasan_slab_free+0x137/0x190
[ 1428.847571] kmem_cache_free+0x85/0x1e0
[ 1428.847575] filename_lookup+0x191/0x280
[ 1428.847580] vfs_statx+0xc1/0x150
[ 1428.847585] __do_sys_newlstat+0x7e/0xd0
[ 1428.847590] do_syscall_64+0x73/0x160
[ 1428.847596] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847648] The buggy address belongs to the object at ffff880194483300
which belongs to the cache names_cache of size 4096
[ 1428.847946] The buggy address is located 576 bytes inside of
4096-byte region [ffff880194483300, ffff880194484300)
[ 1428.848234] The buggy address belongs to the page:
[ 1428.848366] page:ffffea0006512000 count:1 mapcount:0 mapping:ffff8801f3586380 index:0x0 compound_mapcount: 0
[ 1428.848606] flags: 0x17fff8000008100(slab|head)
[ 1428.848737] raw: 017fff8000008100 dead000000000100 dead000000000200 ffff8801f3586380
[ 1428.848931] raw: 0000000000000000 0000000000070007 00000001ffffffff 0000000000000000
[ 1428.849122] page dumped because: kasan: bad access detected
[ 1428.849305] Memory state around the buggy address:
[ 1428.849436] ffff880194483400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849620] ffff880194483480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849804] >ffff880194483500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849985] ^
[ 1428.850120] ffff880194483580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.850303] ffff880194483600: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.850498] ==================================================================
Reported-by: Wen Xu <wen.xu@gatech.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-30 18:13:40 +08:00
|
|
|
if (unlikely(dn.data_blkaddr != NEW_ADDR)) {
|
|
|
|
|
f2fs_put_dnode(&dn);
|
|
|
|
|
set_sbi_flag(F2FS_P_SB(page), SBI_NEED_FSCK);
|
2019-06-18 17:48:42 +08:00
|
|
|
f2fs_warn(F2FS_P_SB(page), "%s: corrupted inline inode ino=%lx, i_addr[0]:0x%x, run fsck to fix.",
|
|
|
|
|
__func__, dir->i_ino, dn.data_blkaddr);
|
2019-06-20 11:36:14 +08:00
|
|
|
err = -EFSCORRUPTED;
|
f2fs: fix to do sanity check with reserved blkaddr of inline inode
As Wen Xu reported in bugzilla, after image was injected with random data
by fuzzing, inline inode would contain invalid reserved blkaddr, then
during inline conversion, we will encounter illegal memory accessing
reported by KASAN, the root cause of this is when writing out converted
inline page, we will use invalid reserved blkaddr to update sit bitmap,
result in accessing memory beyond sit bitmap boundary.
In order to fix this issue, let's do sanity check with reserved block
address of inline inode to avoid above condition.
https://bugzilla.kernel.org/show_bug.cgi?id=200179
[ 1428.846352] BUG: KASAN: use-after-free in update_sit_entry+0x80/0x7f0
[ 1428.846618] Read of size 4 at addr ffff880194483540 by task a.out/2741
[ 1428.846855] CPU: 0 PID: 2741 Comm: a.out Tainted: G W 4.17.0+ #1
[ 1428.846858] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
[ 1428.846860] Call Trace:
[ 1428.846868] dump_stack+0x71/0xab
[ 1428.846875] print_address_description+0x6b/0x290
[ 1428.846881] kasan_report+0x28e/0x390
[ 1428.846888] ? update_sit_entry+0x80/0x7f0
[ 1428.846898] update_sit_entry+0x80/0x7f0
[ 1428.846906] f2fs_allocate_data_block+0x6db/0xc70
[ 1428.846914] ? f2fs_get_node_info+0x14f/0x590
[ 1428.846920] do_write_page+0xc8/0x150
[ 1428.846928] f2fs_outplace_write_data+0xfe/0x210
[ 1428.846935] ? f2fs_do_write_node_page+0x170/0x170
[ 1428.846941] ? radix_tree_tag_clear+0xff/0x130
[ 1428.846946] ? __mod_node_page_state+0x22/0xa0
[ 1428.846951] ? inc_zone_page_state+0x54/0x100
[ 1428.846956] ? __test_set_page_writeback+0x336/0x5d0
[ 1428.846964] f2fs_convert_inline_page+0x407/0x6d0
[ 1428.846971] ? f2fs_read_inline_data+0x3b0/0x3b0
[ 1428.846978] ? __get_node_page+0x335/0x6b0
[ 1428.846987] f2fs_convert_inline_inode+0x41b/0x500
[ 1428.846994] ? f2fs_convert_inline_page+0x6d0/0x6d0
[ 1428.847000] ? kasan_unpoison_shadow+0x31/0x40
[ 1428.847005] ? kasan_kmalloc+0xa6/0xd0
[ 1428.847024] f2fs_file_mmap+0x79/0xc0
[ 1428.847029] mmap_region+0x58b/0x880
[ 1428.847037] ? arch_get_unmapped_area+0x370/0x370
[ 1428.847042] do_mmap+0x55b/0x7a0
[ 1428.847048] vm_mmap_pgoff+0x16f/0x1c0
[ 1428.847055] ? vma_is_stack_for_current+0x50/0x50
[ 1428.847062] ? __fsnotify_update_child_dentry_flags.part.1+0x160/0x160
[ 1428.847068] ? do_sys_open+0x206/0x2a0
[ 1428.847073] ? __fget+0xb4/0x100
[ 1428.847079] ksys_mmap_pgoff+0x278/0x360
[ 1428.847085] ? find_mergeable_anon_vma+0x50/0x50
[ 1428.847091] do_syscall_64+0x73/0x160
[ 1428.847098] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847102] RIP: 0033:0x7fb1430766ba
[ 1428.847103] Code: 89 f5 41 54 49 89 fc 55 53 74 35 49 63 e8 48 63 da 4d 89 f9 49 89 e8 4d 63 d6 48 89 da 4c 89 ee 4c 89 e7 b8 09 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 56 5b 5d 41 5c 41 5d 41 5e 41 5f c3 0f 1f 00
[ 1428.847162] RSP: 002b:00007ffc651d9388 EFLAGS: 00000246 ORIG_RAX: 0000000000000009
[ 1428.847167] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007fb1430766ba
[ 1428.847170] RDX: 0000000000000001 RSI: 0000000000001000 RDI: 0000000000000000
[ 1428.847173] RBP: 0000000000000003 R08: 0000000000000003 R09: 0000000000000000
[ 1428.847176] R10: 0000000000008002 R11: 0000000000000246 R12: 0000000000000000
[ 1428.847179] R13: 0000000000001000 R14: 0000000000008002 R15: 0000000000000000
[ 1428.847252] Allocated by task 2683:
[ 1428.847372] kasan_kmalloc+0xa6/0xd0
[ 1428.847380] kmem_cache_alloc+0xc8/0x1e0
[ 1428.847385] getname_flags+0x73/0x2b0
[ 1428.847390] user_path_at_empty+0x1d/0x40
[ 1428.847395] vfs_statx+0xc1/0x150
[ 1428.847401] __do_sys_newlstat+0x7e/0xd0
[ 1428.847405] do_syscall_64+0x73/0x160
[ 1428.847411] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847466] Freed by task 2683:
[ 1428.847566] __kasan_slab_free+0x137/0x190
[ 1428.847571] kmem_cache_free+0x85/0x1e0
[ 1428.847575] filename_lookup+0x191/0x280
[ 1428.847580] vfs_statx+0xc1/0x150
[ 1428.847585] __do_sys_newlstat+0x7e/0xd0
[ 1428.847590] do_syscall_64+0x73/0x160
[ 1428.847596] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 1428.847648] The buggy address belongs to the object at ffff880194483300
which belongs to the cache names_cache of size 4096
[ 1428.847946] The buggy address is located 576 bytes inside of
4096-byte region [ffff880194483300, ffff880194484300)
[ 1428.848234] The buggy address belongs to the page:
[ 1428.848366] page:ffffea0006512000 count:1 mapcount:0 mapping:ffff8801f3586380 index:0x0 compound_mapcount: 0
[ 1428.848606] flags: 0x17fff8000008100(slab|head)
[ 1428.848737] raw: 017fff8000008100 dead000000000100 dead000000000200 ffff8801f3586380
[ 1428.848931] raw: 0000000000000000 0000000000070007 00000001ffffffff 0000000000000000
[ 1428.849122] page dumped because: kasan: bad access detected
[ 1428.849305] Memory state around the buggy address:
[ 1428.849436] ffff880194483400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849620] ffff880194483480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849804] >ffff880194483500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.849985] ^
[ 1428.850120] ffff880194483580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.850303] ffff880194483600: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 1428.850498] ==================================================================
Reported-by: Wen Xu <wen.xu@gatech.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-06-30 18:13:40 +08:00
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2018-12-25 17:43:42 +08:00
|
|
|
f2fs_wait_on_page_writeback(page, DATA, true, true);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2018-02-28 20:31:52 +08:00
|
|
|
dentry_blk = page_address(page);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
make_dentry_ptr_inline(dir, &src, inline_dentry);
|
|
|
|
|
make_dentry_ptr_block(dir, &dst, dentry_blk);
|
2017-07-16 15:08:54 +08:00
|
|
|
|
2014-09-24 18:17:53 +08:00
|
|
|
/* copy data from inline dentry block to new dentry block */
|
2017-07-16 15:08:54 +08:00
|
|
|
memcpy(dst.bitmap, src.bitmap, src.nr_bitmap);
|
|
|
|
|
memset(dst.bitmap + src.nr_bitmap, 0, dst.nr_bitmap - src.nr_bitmap);
|
2015-08-24 17:36:25 +08:00
|
|
|
/*
|
|
|
|
|
* we do not need to zero out remainder part of dentry and filename
|
|
|
|
|
* field, since we have used bitmap for marking the usage status of
|
|
|
|
|
* them, besides, we can also ignore copying/zeroing reserved space
|
|
|
|
|
* of dentry block, because them haven't been used so far.
|
|
|
|
|
*/
|
2017-07-16 15:08:54 +08:00
|
|
|
memcpy(dst.dentry, src.dentry, SIZE_OF_DIR_ENTRY * src.max);
|
|
|
|
|
memcpy(dst.filename, src.filename, src.max * F2FS_SLOT_LEN);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2016-07-01 09:49:15 +08:00
|
|
|
if (!PageUptodate(page))
|
|
|
|
|
SetPageUptodate(page);
|
2014-09-24 18:17:53 +08:00
|
|
|
set_page_dirty(page);
|
|
|
|
|
|
|
|
|
|
/* clear inline dir and flag after data writeback */
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_truncate_inline_inode(dir, ipage, 0);
|
2014-10-24 10:48:09 +08:00
|
|
|
|
2014-10-14 11:00:16 +08:00
|
|
|
stat_dec_inline_dir(dir);
|
2016-05-21 01:13:22 +08:00
|
|
|
clear_inode_flag(dir, FI_INLINE_DENTRY);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
f2fs: fix to retrieve inline xattr space
With below mkfs and mount option, generic/339 of fstest will report that
scratch image becomes corrupted.
MKFS_OPTIONS -- -O extra_attr -O project_quota -O inode_checksum -O flexible_inline_xattr -O inode_crtime -f /dev/zram1
MOUNT_OPTIONS -- -o acl,user_xattr -o discard,noinline_xattr /dev/zram1 /mnt/scratch_f2fs
[ASSERT] (f2fs_check_dirent_position:1315) --> Wrong position of dirent pino:1970, name: (...)
level:8, dir_level:0, pgofs:951, correct range:[900, 901]
In old kernel, inline data and directory always reserved 200 bytes in
inode layout, even if inline_xattr is disabled, then new kernel tries
to retrieve that space for non-inline xattr inode, but for inline dentry,
its layout size should be fixed, so we just keep that reserved space.
But the problem here is that, after inline dentry conversion, inline
dentry layout no longer exists, if we still reserve inline xattr space,
after dents updates, there will be a hole in inline xattr space, which
can break hierarchy hash directory structure.
This patch fixes this issue by retrieving inline xattr space after
inline dentry conversion.
Fixes: 6afc662e68b5 ("f2fs: support flexible inline xattr size")
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-04-11 11:48:09 +08:00
|
|
|
/*
|
|
|
|
|
* should retrieve reserved space which was used to keep
|
|
|
|
|
* inline_dentry's structure for backward compatibility.
|
|
|
|
|
*/
|
|
|
|
|
if (!f2fs_sb_has_flexible_inline_xattr(F2FS_I_SB(dir)) &&
|
|
|
|
|
!f2fs_has_inline_xattr(dir))
|
|
|
|
|
F2FS_I(dir)->i_inline_xattr_size = 0;
|
|
|
|
|
|
2016-05-21 00:52:20 +08:00
|
|
|
f2fs_i_depth_write(dir, 1);
|
2016-05-21 07:32:49 +08:00
|
|
|
if (i_size_read(dir) < PAGE_SIZE)
|
2016-05-21 00:22:03 +08:00
|
|
|
f2fs_i_size_write(dir, PAGE_SIZE);
|
2014-09-24 18:17:53 +08:00
|
|
|
out:
|
|
|
|
|
f2fs_put_page(page, 1);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
static int f2fs_add_inline_entries(struct inode *dir, void *inline_dentry)
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
{
|
|
|
|
|
struct f2fs_dentry_ptr d;
|
|
|
|
|
unsigned long bit_pos = 0;
|
|
|
|
|
int err = 0;
|
|
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
make_dentry_ptr_inline(dir, &d, inline_dentry);
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
|
|
|
|
|
while (bit_pos < d.max) {
|
|
|
|
|
struct f2fs_dir_entry *de;
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
struct f2fs_filename fname;
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
nid_t ino;
|
|
|
|
|
umode_t fake_mode;
|
|
|
|
|
|
|
|
|
|
if (!test_bit_le(bit_pos, d.bitmap)) {
|
|
|
|
|
bit_pos++;
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
de = &d.dentry[bit_pos];
|
2016-04-27 22:22:20 +08:00
|
|
|
|
|
|
|
|
if (unlikely(!de->name_len)) {
|
|
|
|
|
bit_pos++;
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
/*
|
|
|
|
|
* We only need the disk_name and hash to move the dentry.
|
|
|
|
|
* We don't need the original or casefolded filenames.
|
|
|
|
|
*/
|
|
|
|
|
memset(&fname, 0, sizeof(fname));
|
|
|
|
|
fname.disk_name.name = d.filename[bit_pos];
|
|
|
|
|
fname.disk_name.len = le16_to_cpu(de->name_len);
|
|
|
|
|
fname.hash = de->hash_code;
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
|
|
|
|
|
ino = le32_to_cpu(de->ino);
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
fake_mode = f2fs_get_de_type(de) << S_SHIFT;
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
err = f2fs_add_regular_entry(dir, &fname, NULL, ino, fake_mode);
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
if (err)
|
|
|
|
|
goto punch_dentry_pages;
|
|
|
|
|
|
|
|
|
|
bit_pos += GET_DENTRY_SLOTS(le16_to_cpu(de->name_len));
|
|
|
|
|
}
|
|
|
|
|
return 0;
|
|
|
|
|
punch_dentry_pages:
|
|
|
|
|
truncate_inode_pages(&dir->i_data, 0);
|
2019-02-02 17:33:01 +08:00
|
|
|
f2fs_truncate_blocks(dir, 0, false);
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_remove_dirty_inode(dir);
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int f2fs_move_rehashed_dirents(struct inode *dir, struct page *ipage,
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry)
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
{
|
2017-07-19 00:19:05 +08:00
|
|
|
void *backup_dentry;
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
int err;
|
|
|
|
|
|
2016-09-23 21:30:09 +08:00
|
|
|
backup_dentry = f2fs_kmalloc(F2FS_I_SB(dir),
|
2017-07-19 00:19:05 +08:00
|
|
|
MAX_INLINE_DATA(dir), GFP_F2FS_ZERO);
|
2016-05-14 19:03:53 +08:00
|
|
|
if (!backup_dentry) {
|
|
|
|
|
f2fs_put_page(ipage, 1);
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
return -ENOMEM;
|
2016-05-14 19:03:53 +08:00
|
|
|
}
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
memcpy(backup_dentry, inline_dentry, MAX_INLINE_DATA(dir));
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_truncate_inline_inode(dir, ipage, 0);
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
|
|
|
|
|
unlock_page(ipage);
|
|
|
|
|
|
|
|
|
|
err = f2fs_add_inline_entries(dir, backup_dentry);
|
|
|
|
|
if (err)
|
|
|
|
|
goto recover;
|
|
|
|
|
|
|
|
|
|
lock_page(ipage);
|
|
|
|
|
|
|
|
|
|
stat_dec_inline_dir(dir);
|
2016-05-21 01:13:22 +08:00
|
|
|
clear_inode_flag(dir, FI_INLINE_DENTRY);
|
f2fs: fix to retrieve inline xattr space
With below mkfs and mount option, generic/339 of fstest will report that
scratch image becomes corrupted.
MKFS_OPTIONS -- -O extra_attr -O project_quota -O inode_checksum -O flexible_inline_xattr -O inode_crtime -f /dev/zram1
MOUNT_OPTIONS -- -o acl,user_xattr -o discard,noinline_xattr /dev/zram1 /mnt/scratch_f2fs
[ASSERT] (f2fs_check_dirent_position:1315) --> Wrong position of dirent pino:1970, name: (...)
level:8, dir_level:0, pgofs:951, correct range:[900, 901]
In old kernel, inline data and directory always reserved 200 bytes in
inode layout, even if inline_xattr is disabled, then new kernel tries
to retrieve that space for non-inline xattr inode, but for inline dentry,
its layout size should be fixed, so we just keep that reserved space.
But the problem here is that, after inline dentry conversion, inline
dentry layout no longer exists, if we still reserve inline xattr space,
after dents updates, there will be a hole in inline xattr space, which
can break hierarchy hash directory structure.
This patch fixes this issue by retrieving inline xattr space after
inline dentry conversion.
Fixes: 6afc662e68b5 ("f2fs: support flexible inline xattr size")
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-04-11 11:48:09 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* should retrieve reserved space which was used to keep
|
|
|
|
|
* inline_dentry's structure for backward compatibility.
|
|
|
|
|
*/
|
|
|
|
|
if (!f2fs_sb_has_flexible_inline_xattr(F2FS_I_SB(dir)) &&
|
|
|
|
|
!f2fs_has_inline_xattr(dir))
|
|
|
|
|
F2FS_I(dir)->i_inline_xattr_size = 0;
|
|
|
|
|
|
2020-09-14 16:47:00 +08:00
|
|
|
kfree(backup_dentry);
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
return 0;
|
|
|
|
|
recover:
|
|
|
|
|
lock_page(ipage);
|
2018-12-25 17:43:42 +08:00
|
|
|
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
|
2017-07-19 00:19:05 +08:00
|
|
|
memcpy(inline_dentry, backup_dentry, MAX_INLINE_DATA(dir));
|
2016-05-21 00:52:20 +08:00
|
|
|
f2fs_i_depth_write(dir, 0);
|
2017-07-19 00:19:05 +08:00
|
|
|
f2fs_i_size_write(dir, MAX_INLINE_DATA(dir));
|
2016-05-21 07:32:49 +08:00
|
|
|
set_page_dirty(ipage);
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
f2fs_put_page(ipage, 1);
|
|
|
|
|
|
2020-09-14 16:47:00 +08:00
|
|
|
kfree(backup_dentry);
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2019-12-10 11:03:05 +08:00
|
|
|
static int do_convert_inline_dir(struct inode *dir, struct page *ipage,
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry)
|
f2fs: fix to convert inline directory correctly
With below serials, we will lose parts of dirents:
1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir
ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root 0 Feb 19 15:12 1
-rw-r--r-- 1 root root 0 Feb 19 15:12 10
-rw-r--r-- 1 root root 0 Feb 19 15:12 100
-????????? ? ? ? ? ? 101
-????????? ? ? ? ? ? 102
-????????? ? ? ? ? ? 103
...
The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.
By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.
However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.
This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2016-02-22 18:29:18 +08:00
|
|
|
{
|
|
|
|
|
if (!F2FS_I(dir)->i_dir_level)
|
|
|
|
|
return f2fs_move_inline_dirents(dir, ipage, inline_dentry);
|
|
|
|
|
else
|
|
|
|
|
return f2fs_move_rehashed_dirents(dir, ipage, inline_dentry);
|
|
|
|
|
}
|
|
|
|
|
|
2019-12-10 11:03:05 +08:00
|
|
|
int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry)
|
|
|
|
|
{
|
|
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(dir);
|
|
|
|
|
struct page *ipage;
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
struct f2fs_filename fname;
|
2019-12-10 11:03:05 +08:00
|
|
|
void *inline_dentry = NULL;
|
|
|
|
|
int err = 0;
|
|
|
|
|
|
|
|
|
|
if (!f2fs_has_inline_dentry(dir))
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
f2fs_lock_op(sbi);
|
|
|
|
|
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
err = f2fs_setup_filename(dir, &dentry->d_name, 0, &fname);
|
2019-12-10 11:03:05 +08:00
|
|
|
if (err)
|
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
|
|
ipage = f2fs_get_node_page(sbi, dir->i_ino);
|
|
|
|
|
if (IS_ERR(ipage)) {
|
|
|
|
|
err = PTR_ERR(ipage);
|
2020-05-07 15:59:02 +08:00
|
|
|
goto out_fname;
|
2019-12-10 11:03:05 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (f2fs_has_enough_room(dir, ipage, &fname)) {
|
|
|
|
|
f2fs_put_page(ipage, 1);
|
2020-05-07 15:59:02 +08:00
|
|
|
goto out_fname;
|
2019-12-10 11:03:05 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
inline_dentry = inline_data_addr(dir, ipage);
|
|
|
|
|
|
|
|
|
|
err = do_convert_inline_dir(dir, ipage, inline_dentry);
|
|
|
|
|
if (!err)
|
|
|
|
|
f2fs_put_page(ipage, 1);
|
2020-05-07 15:59:02 +08:00
|
|
|
out_fname:
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
f2fs_free_filename(&fname);
|
2019-12-10 11:03:05 +08:00
|
|
|
out:
|
|
|
|
|
f2fs_unlock_op(sbi);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
|
|
|
|
|
struct inode *inode, nid_t ino, umode_t mode)
|
2014-09-24 18:17:53 +08:00
|
|
|
{
|
|
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(dir);
|
|
|
|
|
struct page *ipage;
|
|
|
|
|
unsigned int bit_pos;
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry = NULL;
|
2015-02-16 16:17:20 +08:00
|
|
|
struct f2fs_dentry_ptr d;
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
int slots = GET_DENTRY_SLOTS(fname->disk_name.len);
|
2015-03-31 06:07:16 +08:00
|
|
|
struct page *page = NULL;
|
2014-09-24 18:17:53 +08:00
|
|
|
int err = 0;
|
|
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(sbi, dir->i_ino);
|
2014-09-24 18:17:53 +08:00
|
|
|
if (IS_ERR(ipage))
|
|
|
|
|
return PTR_ERR(ipage);
|
|
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
inline_dentry = inline_data_addr(dir, ipage);
|
|
|
|
|
make_dentry_ptr_inline(dir, &d, inline_dentry);
|
2017-07-16 15:08:54 +08:00
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
bit_pos = f2fs_room_for_filename(d.bitmap, slots, d.max);
|
2017-07-16 15:08:54 +08:00
|
|
|
if (bit_pos >= d.max) {
|
2019-12-10 11:03:05 +08:00
|
|
|
err = do_convert_inline_dir(dir, ipage, inline_dentry);
|
2015-07-14 18:14:06 +08:00
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
err = -EAGAIN;
|
2014-09-24 18:17:53 +08:00
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2015-03-31 06:07:16 +08:00
|
|
|
if (inode) {
|
|
|
|
|
down_write(&F2FS_I(inode)->i_sem);
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
page = f2fs_init_inode_metadata(inode, dir, fname, ipage);
|
2015-03-31 06:07:16 +08:00
|
|
|
if (IS_ERR(page)) {
|
|
|
|
|
err = PTR_ERR(page);
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
2014-09-24 18:17:53 +08:00
|
|
|
}
|
2014-10-14 10:42:53 +08:00
|
|
|
|
2018-12-25 17:43:42 +08:00
|
|
|
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
|
2015-02-16 16:17:20 +08:00
|
|
|
|
f2fs: rework filename handling
Rework f2fs's handling of filenames to use a new 'struct f2fs_filename'.
Similar to 'struct ext4_filename', this stores the usr_fname, disk_name,
dirhash, crypto_buf, and casefolded name. Some of these names can be
NULL in some cases. 'struct f2fs_filename' differs from
'struct fscrypt_name' mainly in that the casefolded name is included.
For user-initiated directory operations like lookup() and create(),
initialize the f2fs_filename by translating the corresponding
fscrypt_name, then computing the dirhash and casefolded name if needed.
This makes the dirhash and casefolded name be cached for each syscall,
so we don't have to recompute them repeatedly. (Previously, f2fs
computed the dirhash once per directory level, and the casefolded name
once per directory block.) This improves performance.
This rework also makes it much easier to correctly handle all
combinations of normal, encrypted, casefolded, and encrypted+casefolded
directories. (The fourth isn't supported yet but is being worked on.)
The only other cases where an f2fs_filename gets initialized are for two
filesystem-internal operations: (1) when converting an inline directory
to a regular one, we grab the needed disk_name and hash from an existing
f2fs_dir_entry; and (2) when roll-forward recovering a new dentry, we
grab the needed disk_name from f2fs_inode::i_name and compute the hash.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2020-05-07 15:59:04 +08:00
|
|
|
f2fs_update_dentry(ino, mode, &d, &fname->disk_name, fname->hash,
|
|
|
|
|
bit_pos);
|
2015-02-16 16:17:20 +08:00
|
|
|
|
2014-09-24 18:17:53 +08:00
|
|
|
set_page_dirty(ipage);
|
|
|
|
|
|
|
|
|
|
/* we don't need to mark_inode_dirty now */
|
2015-03-31 06:07:16 +08:00
|
|
|
if (inode) {
|
2016-05-21 00:52:20 +08:00
|
|
|
f2fs_i_pino_write(inode, dir->i_ino);
|
2019-09-10 09:14:16 +08:00
|
|
|
|
|
|
|
|
/* synchronize inode page's data from inode cache */
|
|
|
|
|
if (is_inode_flag_set(inode, FI_NEW_INODE))
|
|
|
|
|
f2fs_update_inode(inode, page);
|
|
|
|
|
|
2015-03-31 06:07:16 +08:00
|
|
|
f2fs_put_page(page, 1);
|
|
|
|
|
}
|
2014-09-24 18:17:53 +08:00
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
f2fs_update_parent_metadata(dir, inode, 0);
|
2014-09-24 18:17:53 +08:00
|
|
|
fail:
|
2015-03-31 06:07:16 +08:00
|
|
|
if (inode)
|
|
|
|
|
up_write(&F2FS_I(inode)->i_sem);
|
2014-09-24 18:17:53 +08:00
|
|
|
out:
|
|
|
|
|
f2fs_put_page(ipage, 1);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, struct page *page,
|
|
|
|
|
struct inode *dir, struct inode *inode)
|
|
|
|
|
{
|
2017-07-16 15:08:54 +08:00
|
|
|
struct f2fs_dentry_ptr d;
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry;
|
2014-09-24 18:17:53 +08:00
|
|
|
int slots = GET_DENTRY_SLOTS(le16_to_cpu(dentry->name_len));
|
|
|
|
|
unsigned int bit_pos;
|
|
|
|
|
int i;
|
|
|
|
|
|
|
|
|
|
lock_page(page);
|
2018-12-25 17:43:42 +08:00
|
|
|
f2fs_wait_on_page_writeback(page, NODE, true, true);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
inline_dentry = inline_data_addr(dir, page);
|
|
|
|
|
make_dentry_ptr_inline(dir, &d, inline_dentry);
|
2017-07-16 15:08:54 +08:00
|
|
|
|
|
|
|
|
bit_pos = dentry - d.dentry;
|
2014-09-24 18:17:53 +08:00
|
|
|
for (i = 0; i < slots; i++)
|
2017-07-16 15:08:54 +08:00
|
|
|
__clear_bit_le(bit_pos + i, d.bitmap);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
|
|
|
|
set_page_dirty(page);
|
2016-06-02 12:18:25 +08:00
|
|
|
f2fs_put_page(page, 1);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2016-09-14 22:48:04 +08:00
|
|
|
dir->i_ctime = dir->i_mtime = current_time(dir);
|
2016-10-15 02:51:23 +08:00
|
|
|
f2fs_mark_inode_dirty_sync(dir, false);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
|
|
|
|
if (inode)
|
2016-06-02 12:18:25 +08:00
|
|
|
f2fs_drop_nlink(dir, inode);
|
2014-09-24 18:17:53 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bool f2fs_empty_inline_dir(struct inode *dir)
|
|
|
|
|
{
|
|
|
|
|
struct f2fs_sb_info *sbi = F2FS_I_SB(dir);
|
|
|
|
|
struct page *ipage;
|
|
|
|
|
unsigned int bit_pos = 2;
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry;
|
2017-07-16 15:08:54 +08:00
|
|
|
struct f2fs_dentry_ptr d;
|
2014-09-24 18:17:53 +08:00
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(sbi, dir->i_ino);
|
2014-09-24 18:17:53 +08:00
|
|
|
if (IS_ERR(ipage))
|
|
|
|
|
return false;
|
|
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
inline_dentry = inline_data_addr(dir, ipage);
|
|
|
|
|
make_dentry_ptr_inline(dir, &d, inline_dentry);
|
2017-07-16 15:08:54 +08:00
|
|
|
|
|
|
|
|
bit_pos = find_next_bit_le(d.bitmap, d.max, bit_pos);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
|
|
|
|
f2fs_put_page(ipage, 1);
|
|
|
|
|
|
2017-07-16 15:08:54 +08:00
|
|
|
if (bit_pos < d.max)
|
2014-09-24 18:17:53 +08:00
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2015-04-28 07:26:24 +08:00
|
|
|
int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
|
2015-05-16 07:26:10 +08:00
|
|
|
struct fscrypt_str *fstr)
|
2014-09-24 18:17:53 +08:00
|
|
|
{
|
|
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct page *ipage = NULL;
|
2014-10-19 13:52:52 +08:00
|
|
|
struct f2fs_dentry_ptr d;
|
2017-07-19 00:19:05 +08:00
|
|
|
void *inline_dentry = NULL;
|
2016-10-29 18:46:34 +08:00
|
|
|
int err;
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2017-07-16 15:08:54 +08:00
|
|
|
make_dentry_ptr_inline(inode, &d, inline_dentry);
|
|
|
|
|
|
|
|
|
|
if (ctx->pos == d.max)
|
2014-09-24 18:17:53 +08:00
|
|
|
return 0;
|
|
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(F2FS_I_SB(inode), inode->i_ino);
|
2014-09-24 18:17:53 +08:00
|
|
|
if (IS_ERR(ipage))
|
|
|
|
|
return PTR_ERR(ipage);
|
|
|
|
|
|
2019-03-12 15:44:27 +08:00
|
|
|
/*
|
|
|
|
|
* f2fs_readdir was protected by inode.i_rwsem, it is safe to access
|
|
|
|
|
* ipage without page's lock held.
|
|
|
|
|
*/
|
|
|
|
|
unlock_page(ipage);
|
|
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
inline_dentry = inline_data_addr(inode, ipage);
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2017-04-04 18:01:22 +08:00
|
|
|
make_dentry_ptr_inline(inode, &d, inline_dentry);
|
2014-10-19 13:52:52 +08:00
|
|
|
|
2016-10-29 18:46:34 +08:00
|
|
|
err = f2fs_fill_dentries(ctx, &d, 0, fstr);
|
|
|
|
|
if (!err)
|
2017-07-16 15:08:54 +08:00
|
|
|
ctx->pos = d.max;
|
2014-09-24 18:17:53 +08:00
|
|
|
|
2019-03-12 15:44:27 +08:00
|
|
|
f2fs_put_page(ipage, 0);
|
2016-10-29 18:46:34 +08:00
|
|
|
return err < 0 ? err : 0;
|
2014-09-24 18:17:53 +08:00
|
|
|
}
|
2015-10-16 02:34:49 +08:00
|
|
|
|
|
|
|
|
int f2fs_inline_data_fiemap(struct inode *inode,
|
|
|
|
|
struct fiemap_extent_info *fieinfo, __u64 start, __u64 len)
|
|
|
|
|
{
|
|
|
|
|
__u64 byteaddr, ilen;
|
|
|
|
|
__u32 flags = FIEMAP_EXTENT_DATA_INLINE | FIEMAP_EXTENT_NOT_ALIGNED |
|
|
|
|
|
FIEMAP_EXTENT_LAST;
|
|
|
|
|
struct node_info ni;
|
|
|
|
|
struct page *ipage;
|
|
|
|
|
int err = 0;
|
|
|
|
|
|
f2fs: clean up symbol namespace
As Ted reported:
"Hi, I was looking at f2fs's sources recently, and I noticed that there
is a very large number of non-static symbols which don't have a f2fs
prefix. There's well over a hundred (see attached below).
As one example, in fs/f2fs/dir.c there is:
unsigned char get_de_type(struct f2fs_dir_entry *de)
This function is clearly only useful for f2fs, but it has a generic
name. This means that if any other file system tries to have the same
symbol name, there will be a symbol conflict and the kernel would not
successfully build. It also means that when someone is looking f2fs
sources, it's not at all obvious whether a function such as
read_data_page(), invalidate_blocks(), is a generic kernel function
found in the fs, mm, or block layers, or a f2fs specific function.
You might want to fix this at some point. Hopefully Kent's bcachefs
isn't similarly using genericly named functions, since that might
cause conflicts with f2fs's functions --- but just as this would be a
problem that we would rightly insist that Kent fix, this is something
that we should have rightly insisted that f2fs should have fixed
before it was integrated into the mainline kernel.
acquire_orphan_inode
add_ino_entry
add_orphan_inode
allocate_data_block
allocate_new_segments
alloc_nid
alloc_nid_done
alloc_nid_failed
available_free_memory
...."
This patch adds "f2fs_" prefix for all non-static symbols in order to:
a) avoid conflict with other kernel generic symbols;
b) to indicate the function is f2fs specific one instead of generic
one;
Reported-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-30 00:20:41 +08:00
|
|
|
ipage = f2fs_get_node_page(F2FS_I_SB(inode), inode->i_ino);
|
2015-10-16 02:34:49 +08:00
|
|
|
if (IS_ERR(ipage))
|
|
|
|
|
return PTR_ERR(ipage);
|
|
|
|
|
|
2019-07-22 18:03:50 +08:00
|
|
|
if ((S_ISREG(inode->i_mode) || S_ISLNK(inode->i_mode)) &&
|
|
|
|
|
!f2fs_has_inline_data(inode)) {
|
|
|
|
|
err = -EAGAIN;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (S_ISDIR(inode->i_mode) && !f2fs_has_inline_dentry(inode)) {
|
2015-10-16 02:34:49 +08:00
|
|
|
err = -EAGAIN;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2017-07-19 00:19:05 +08:00
|
|
|
ilen = min_t(size_t, MAX_INLINE_DATA(inode), i_size_read(inode));
|
2015-10-16 02:34:49 +08:00
|
|
|
if (start >= ilen)
|
|
|
|
|
goto out;
|
|
|
|
|
if (start + len < ilen)
|
|
|
|
|
ilen = start + len;
|
|
|
|
|
ilen -= start;
|
|
|
|
|
|
2018-07-17 00:02:17 +08:00
|
|
|
err = f2fs_get_node_info(F2FS_I_SB(inode), inode->i_ino, &ni);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out;
|
|
|
|
|
|
2015-10-16 02:34:49 +08:00
|
|
|
byteaddr = (__u64)ni.blk_addr << inode->i_sb->s_blocksize_bits;
|
2017-07-19 00:19:05 +08:00
|
|
|
byteaddr += (char *)inline_data_addr(inode, ipage) -
|
|
|
|
|
(char *)F2FS_INODE(ipage);
|
2015-10-16 02:34:49 +08:00
|
|
|
err = fiemap_fill_next_extent(fieinfo, start, byteaddr, ilen, flags);
|
2020-06-29 20:13:13 +08:00
|
|
|
trace_f2fs_fiemap(inode, start, byteaddr, ilen, flags, err);
|
2015-10-16 02:34:49 +08:00
|
|
|
out:
|
|
|
|
|
f2fs_put_page(ipage, 1);
|
|
|
|
|
return err;
|
|
|
|
|
}
|