WSL2-Linux-Kernel/fs/f2fs/inline.c

802 строки
19 KiB
C
Исходник Обычный вид История

// SPDX-License-Identifier: GPL-2.0
/*
* 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>
#include <linux/fiemap.h>
#include "f2fs.h"
#include "node.h"
#include <trace/events/f2fs.h>
bool f2fs_may_inline_data(struct inode *inode)
{
if (f2fs_is_atomic_file(inode))
return false;
if (!S_ISREG(inode->i_mode) && !S_ISLNK(inode->i_mode))
return false;
if (i_size_read(inode) > MAX_INLINE_DATA(inode))
return false;
if (f2fs_post_read_required(inode))
return false;
return true;
}
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>
2018-05-29 19:20:41 +03:00
void f2fs_do_read_inline_data(struct page *page, struct page *ipage)
{
struct inode *inode = page->mapping->host;
void *src_addr, *dst_addr;
if (PageUptodate(page))
return;
f2fs_bug_on(F2FS_P_SB(page), page->index);
zero_user_segment(page, MAX_INLINE_DATA(inode), PAGE_SIZE);
/* Copy the whole inline data block */
src_addr = inline_data_addr(inode, ipage);
dst_addr = kmap_atomic(page);
memcpy(dst_addr, src_addr, MAX_INLINE_DATA(inode));
flush_dcache_page(page);
kunmap_atomic(dst_addr);
if (!PageUptodate(page))
SetPageUptodate(page);
}
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>
2018-05-29 19:20:41 +03:00
void f2fs_truncate_inline_inode(struct inode *inode,
struct page *ipage, u64 from)
{
void *addr;
if (from >= MAX_INLINE_DATA(inode))
return;
addr = inline_data_addr(inode, ipage);
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
memset(addr + from, 0, MAX_INLINE_DATA(inode) - from);
set_page_dirty(ipage);
if (from == 0)
clear_inode_flag(inode, FI_DATA_EXIST);
}
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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(F2FS_I_SB(inode), inode->i_ino);
if (IS_ERR(ipage)) {
unlock_page(page);
return PTR_ERR(ipage);
}
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 15:29:47 +03:00
zero_user_segment(page, 0, PAGE_SIZE);
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>
2018-05-29 19:20:41 +03:00
f2fs_do_read_inline_data(page, ipage);
if (!PageUptodate(page))
SetPageUptodate(page);
f2fs_put_page(ipage, 1);
unlock_page(page);
return 0;
}
int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page)
{
struct f2fs_io_info fio = {
.sbi = F2FS_I_SB(dn->inode),
.ino = dn->inode->i_ino,
.type = DATA,
.op = REQ_OP_WRITE,
.op_flags = REQ_SYNC | REQ_PRIO,
.page = page,
.encrypted_page = NULL,
.io_type = FS_DATA_IO,
};
struct node_info ni;
int dirty, err;
if (!f2fs_exist_data(dn->inode))
goto clear_out;
err = f2fs_reserve_block(dn, 0);
if (err)
return err;
err = f2fs_get_node_info(fio.sbi, dn->nid, &ni, false);
if (err) {
f2fs_truncate_data_blocks_range(dn, 1);
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 13:13:40 +03:00
if (unlikely(dn->data_blkaddr != NEW_ADDR)) {
f2fs_put_dnode(dn);
set_sbi_flag(fio.sbi, SBI_NEED_FSCK);
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);
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 13:13:40 +03:00
}
f2fs_bug_on(F2FS_P_SB(page), PageWriteback(page));
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>
2018-05-29 19:20:41 +03:00
f2fs_do_read_inline_data(page, dn->inode_page);
f2fs: call set_page_dirty to attach i_wb for cgroup The cgroup attaches inode->i_wb via mark_inode_dirty and when set_page_writeback is called, __inc_wb_stat() updates i_wb's stat. So, we need to explicitly call set_page_dirty->__mark_inode_dirty in prior to any writebacking pages. This patch should resolve the following kernel panic reported by Andreas Reis. https://bugzilla.kernel.org/show_bug.cgi?id=101801 --- Comment #2 from Andreas Reis <andreas.reis@gmail.com> --- BUG: unable to handle kernel NULL pointer dereference at 00000000000000a8 IP: [<ffffffff8149deea>] __percpu_counter_add+0x1a/0x90 PGD 2951ff067 PUD 2df43f067 PMD 0 Oops: 0000 [#1] PREEMPT SMP Modules linked in: CPU: 7 PID: 10356 Comm: gcc Tainted: G W 4.2.0-1-cu #1 Hardware name: Gigabyte Technology Co., Ltd. G1.Sniper M5/G1.Sniper M5, BIOS T01 02/03/2015 task: ffff880295044f80 ti: ffff880295140000 task.ti: ffff880295140000 RIP: 0010:[<ffffffff8149deea>] [<ffffffff8149deea>] __percpu_counter_add+0x1a/0x90 RSP: 0018:ffff880295143ac8 EFLAGS: 00010082 RAX: 0000000000000003 RBX: ffffea000a526d40 RCX: 0000000000000001 RDX: 0000000000000020 RSI: 0000000000000001 RDI: 0000000000000088 RBP: ffff880295143ae8 R08: 0000000000000000 R09: ffff88008f69bb30 R10: 00000000fffffffa R11: 0000000000000000 R12: 0000000000000088 R13: 0000000000000001 R14: ffff88041d099000 R15: ffff880084a205d0 FS: 00007f8549374700(0000) GS:ffff88042f3c0000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000000a8 CR3: 000000033e1d5000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Stack: 0000000000000000 ffffea000a526d40 ffff880084a20738 ffff880084a20750 ffff880295143b48 ffffffff811cc91e ffff880000000000 0000000000000296 0000000000000000 ffff880417090198 0000000000000000 ffffea000a526d40 Call Trace: [<ffffffff811cc91e>] __test_set_page_writeback+0xde/0x1d0 [<ffffffff813fee87>] do_write_data_page+0xe7/0x3a0 [<ffffffff813faeea>] gc_data_segment+0x5aa/0x640 [<ffffffff813fb0b8>] do_garbage_collect+0x138/0x150 [<ffffffff813fb3fe>] f2fs_gc+0x1be/0x3e0 [<ffffffff81405541>] f2fs_balance_fs+0x81/0x90 [<ffffffff813ee357>] f2fs_unlink+0x47/0x1d0 [<ffffffff81239329>] vfs_unlink+0x109/0x1b0 [<ffffffff8123e3d7>] do_unlinkat+0x287/0x2c0 [<ffffffff8123ebc6>] SyS_unlink+0x16/0x20 [<ffffffff81942e2e>] entry_SYSCALL_64_fastpath+0x12/0x71 Code: 41 5e 5d c3 0f 1f 00 66 2e 0f 1f 84 00 00 00 00 00 55 48 89 e5 41 55 49 89 f5 41 54 49 89 fc 53 48 83 ec 08 65 ff 05 e6 d9 b6 7e <48> 8b 47 20 48 63 ca 65 8b 18 48 63 db 48 01 f3 48 39 cb 7d 0a RIP [<ffffffff8149deea>] __percpu_counter_add+0x1a/0x90 RSP <ffff880295143ac8> CR2: 00000000000000a8 ---[ end trace 5132449a58ed93a3 ]--- note: gcc[10356] exited with preempt_count 2 Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-07-25 10:29:17 +03:00
set_page_dirty(page);
/* clear dirty state */
dirty = clear_page_dirty_for_io(page);
/* write data page to try to make data consistent */
set_page_writeback(page);
ClearPageError(page);
fio.old_blkaddr = dn->data_blkaddr;
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>
2018-05-29 19:20:41 +03:00
f2fs_outplace_write_data(dn, &fio);
f2fs_wait_on_page_writeback(page, DATA, true, true);
if (dirty) {
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>
2018-05-29 19:20:41 +03:00
f2fs_remove_dirty_inode(dn->inode);
}
/* this converted inline_data should be recovered. */
set_inode_flag(dn->inode, FI_APPEND_WRITE);
/* 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>
2018-05-29 19:20:41 +03:00
f2fs_truncate_inline_inode(dn->inode, dn->inode_page, 0);
f2fs: restructure f2fs page.private layout Restruct f2fs page private layout for below reasons: There are some cases that f2fs wants to set a flag in a page to indicate a specified status of page: a) page is in transaction list for atomic write b) page contains dummy data for aligned write c) page is migrating for GC d) page contains inline data for inline inode flush e) page belongs to merkle tree, and is verified for fsverity f) page is dirty and has filesystem/inode reference count for writeback g) page is temporary and has decompress io context reference for compression There are existed places in page structure we can use to store f2fs private status/data: - page.flags: PG_checked, PG_private - page.private However it was a mess when we using them, which may cause potential confliction: page.private PG_private PG_checked page._refcount (+1 at most) a) -1 set +1 b) -2 set c), d), e) set f) 0 set +1 g) pointer set The other problem is page.flags has no free slot, if we can avoid set zero to page.private and set PG_private flag, then we use non-zero value to indicate PG_private status, so that we may have chance to reclaim PG_private slot for other usage. [1] The other concern is f2fs has bad scalability in aspect of indicating more page status. So in this patch, let's restructure f2fs' page.private as below to solve above issues: Layout A: lowest bit should be 1 | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | bit 0 PAGE_PRIVATE_NOT_POINTER bit 1 PAGE_PRIVATE_ATOMIC_WRITE bit 2 PAGE_PRIVATE_DUMMY_WRITE bit 3 PAGE_PRIVATE_ONGOING_MIGRATION bit 4 PAGE_PRIVATE_INLINE_INODE bit 5 PAGE_PRIVATE_REF_RESOURCE bit 6- f2fs private data Layout B: lowest bit should be 0 page.private is a wrapped pointer. After the change: page.private PG_private PG_checked page._refcount (+1 at most) a) 11 set +1 b) 101 set +1 c) 1001 set +1 d) 10001 set +1 e) set f) 100001 set +1 g) pointer set +1 [1] https://lore.kernel.org/linux-f2fs-devel/20210422154705.GO3596236@casper.infradead.org/T/#u Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-04-28 12:20:31 +03:00
clear_page_private_inline(dn->inode_page);
clear_out:
stat_dec_inline_inode(dn->inode);
clear_inode_flag(dn->inode, FI_INLINE_DATA);
f2fs_put_dnode(dn);
return 0;
}
int f2fs_convert_inline_inode(struct inode *inode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct dnode_of_data dn;
struct page *ipage, *page;
int err = 0;
if (!f2fs_has_inline_data(inode) ||
f2fs_hw_is_readonly(sbi) || f2fs_readonly(sbi->sb))
return 0;
err = f2fs_dquot_initialize(inode);
f2fs: fix to avoid inconsistent quota data 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>
2021-01-28 12:02:56 +03:00
if (err)
return err;
page = f2fs_grab_cache_page(inode->i_mapping, 0, false);
if (!page)
return -ENOMEM;
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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(sbi, inode->i_ino);
if (IS_ERR(ipage)) {
err = PTR_ERR(ipage);
goto out;
}
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);
out:
f2fs_unlock_op(sbi);
f2fs_put_page(page, 1);
if (!err)
f2fs_balance_fs(sbi, dn.node_changed);
return err;
}
int f2fs_write_inline_data(struct inode *inode, struct page *page)
{
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>
2018-05-29 19:20:41 +03:00
err = f2fs_get_dnode_of_data(&dn, 0, LOOKUP_NODE);
if (err)
return err;
if (!f2fs_has_inline_data(inode)) {
f2fs_put_dnode(&dn);
return -EAGAIN;
}
f2fs_bug_on(F2FS_I_SB(inode), page->index);
f2fs_wait_on_page_writeback(dn.inode_page, NODE, true, true);
src_addr = kmap_atomic(page);
dst_addr = inline_data_addr(inode, dn.inode_page);
memcpy(dst_addr, src_addr, MAX_INLINE_DATA(inode));
kunmap_atomic(src_addr);
set_page_dirty(dn.inode_page);
f2fs_clear_page_cache_dirty_tag(page);
set_inode_flag(inode, FI_APPEND_WRITE);
set_inode_flag(inode, FI_DATA_EXIST);
f2fs: restructure f2fs page.private layout Restruct f2fs page private layout for below reasons: There are some cases that f2fs wants to set a flag in a page to indicate a specified status of page: a) page is in transaction list for atomic write b) page contains dummy data for aligned write c) page is migrating for GC d) page contains inline data for inline inode flush e) page belongs to merkle tree, and is verified for fsverity f) page is dirty and has filesystem/inode reference count for writeback g) page is temporary and has decompress io context reference for compression There are existed places in page structure we can use to store f2fs private status/data: - page.flags: PG_checked, PG_private - page.private However it was a mess when we using them, which may cause potential confliction: page.private PG_private PG_checked page._refcount (+1 at most) a) -1 set +1 b) -2 set c), d), e) set f) 0 set +1 g) pointer set The other problem is page.flags has no free slot, if we can avoid set zero to page.private and set PG_private flag, then we use non-zero value to indicate PG_private status, so that we may have chance to reclaim PG_private slot for other usage. [1] The other concern is f2fs has bad scalability in aspect of indicating more page status. So in this patch, let's restructure f2fs' page.private as below to solve above issues: Layout A: lowest bit should be 1 | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | bit 0 PAGE_PRIVATE_NOT_POINTER bit 1 PAGE_PRIVATE_ATOMIC_WRITE bit 2 PAGE_PRIVATE_DUMMY_WRITE bit 3 PAGE_PRIVATE_ONGOING_MIGRATION bit 4 PAGE_PRIVATE_INLINE_INODE bit 5 PAGE_PRIVATE_REF_RESOURCE bit 6- f2fs private data Layout B: lowest bit should be 0 page.private is a wrapped pointer. After the change: page.private PG_private PG_checked page._refcount (+1 at most) a) 11 set +1 b) 101 set +1 c) 1001 set +1 d) 10001 set +1 e) set f) 100001 set +1 g) pointer set +1 [1] https://lore.kernel.org/linux-f2fs-devel/20210422154705.GO3596236@casper.infradead.org/T/#u Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2021-04-28 12:20:31 +03:00
clear_page_private_inline(dn.inode_page);
f2fs_put_dnode(&dn);
return 0;
}
int f2fs_recover_inline_data(struct inode *inode, struct page *npage)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
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 data blocks, and then recover inline_data
* x x -> recover data blocks
*/
if (IS_INODE(npage))
ri = F2FS_INODE(npage);
if (f2fs_has_inline_data(inode) &&
ri && (ri->i_inline & F2FS_INLINE_DATA)) {
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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(sbi, inode->i_ino);
if (IS_ERR(ipage))
return PTR_ERR(ipage);
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
src_addr = inline_data_addr(inode, npage);
dst_addr = inline_data_addr(inode, ipage);
memcpy(dst_addr, src_addr, MAX_INLINE_DATA(inode));
set_inode_flag(inode, FI_INLINE_DATA);
set_inode_flag(inode, FI_DATA_EXIST);
set_page_dirty(ipage);
f2fs_put_page(ipage, 1);
return 1;
}
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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(sbi, inode->i_ino);
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>
2018-05-29 19:20:41 +03:00
f2fs_truncate_inline_inode(inode, ipage, 0);
stat_dec_inline_inode(inode);
clear_inode_flag(inode, FI_INLINE_DATA);
f2fs_put_page(ipage, 1);
} else if (ri && (ri->i_inline & F2FS_INLINE_DATA)) {
int ret;
ret = f2fs_truncate_blocks(inode, 0, false);
if (ret)
return ret;
stat_inc_inline_inode(inode);
goto process_inline;
}
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>
2018-05-29 19:20:41 +03:00
struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
2020-05-07 10:59:04 +03:00
const struct f2fs_filename *fname,
struct page **res_page)
{
struct f2fs_sb_info *sbi = F2FS_SB(dir->i_sb);
struct f2fs_dir_entry *de;
struct f2fs_dentry_ptr d;
struct page *ipage;
void *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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(sbi, dir->i_ino);
if (IS_ERR(ipage)) {
*res_page = ipage;
return NULL;
}
inline_dentry = inline_data_addr(dir, ipage);
make_dentry_ptr_inline(dir, &d, inline_dentry);
2020-05-07 10:59:04 +03:00
de = f2fs_find_target_dentry(&d, fname, NULL);
unlock_page(ipage);
if (IS_ERR(de)) {
*res_page = ERR_CAST(de);
de = NULL;
}
if (de)
*res_page = ipage;
else
f2fs_put_page(ipage, 0);
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>
2018-05-29 19:20:41 +03:00
int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
struct page *ipage)
{
struct f2fs_dentry_ptr d;
void *inline_dentry;
inline_dentry = inline_data_addr(inode, ipage);
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>
2018-05-29 19:20:41 +03:00
f2fs_do_make_empty_dir(inode, parent, &d);
set_page_dirty(ipage);
/* update i_size to MAX_INLINE_DATA */
if (i_size_read(inode) < MAX_INLINE_DATA(inode))
f2fs_i_size_write(inode, MAX_INLINE_DATA(inode));
return 0;
}
/*
* 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 13:29:18 +03:00
static int f2fs_move_inline_dirents(struct inode *dir, struct page *ipage,
void *inline_dentry)
{
struct page *page;
struct dnode_of_data dn;
struct f2fs_dentry_block *dentry_blk;
struct f2fs_dentry_ptr src, dst;
int err;
page = f2fs_grab_cache_page(dir->i_mapping, 0, true);
if (!page) {
f2fs_put_page(ipage, 1);
return -ENOMEM;
}
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 13:13:40 +03:00
if (unlikely(dn.data_blkaddr != NEW_ADDR)) {
f2fs_put_dnode(&dn);
set_sbi_flag(F2FS_P_SB(page), SBI_NEED_FSCK);
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);
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 13:13:40 +03:00
goto out;
}
f2fs_wait_on_page_writeback(page, DATA, true, true);
dentry_blk = page_address(page);
make_dentry_ptr_inline(dir, &src, inline_dentry);
make_dentry_ptr_block(dir, &dst, dentry_blk);
/* copy data from inline dentry block to new dentry block */
memcpy(dst.bitmap, src.bitmap, src.nr_bitmap);
memset(dst.bitmap + src.nr_bitmap, 0, dst.nr_bitmap - src.nr_bitmap);
/*
* 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.
*/
memcpy(dst.dentry, src.dentry, SIZE_OF_DIR_ENTRY * src.max);
memcpy(dst.filename, src.filename, src.max * F2FS_SLOT_LEN);
if (!PageUptodate(page))
SetPageUptodate(page);
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>
2018-05-29 19:20:41 +03:00
f2fs_truncate_inline_inode(dir, ipage, 0);
stat_dec_inline_dir(dir);
clear_inode_flag(dir, FI_INLINE_DENTRY);
/*
* 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;
f2fs_i_depth_write(dir, 1);
if (i_size_read(dir) < PAGE_SIZE)
f2fs_i_size_write(dir, PAGE_SIZE);
out:
f2fs_put_page(page, 1);
return err;
}
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 13:29:18 +03:00
{
struct f2fs_dentry_ptr d;
unsigned long bit_pos = 0;
int err = 0;
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 13:29:18 +03:00
while (bit_pos < d.max) {
struct f2fs_dir_entry *de;
2020-05-07 10:59:04 +03:00
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 13:29:18 +03:00
nid_t ino;
umode_t fake_mode;
if (!test_bit_le(bit_pos, d.bitmap)) {
bit_pos++;
continue;
}
de = &d.dentry[bit_pos];
if (unlikely(!de->name_len)) {
bit_pos++;
continue;
}
2020-05-07 10:59:04 +03:00
/*
* 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 13:29:18 +03:00
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>
2018-05-29 19:20:41 +03:00
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 13:29:18 +03:00
2020-05-07 10:59:04 +03:00
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 13:29:18 +03:00
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);
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>
2018-05-29 19:20:41 +03:00
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 13:29:18 +03:00
return err;
}
static int f2fs_move_rehashed_dirents(struct inode *dir, struct page *ipage,
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 13:29:18 +03:00
{
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 13:29:18 +03:00
int err;
backup_dentry = f2fs_kmalloc(F2FS_I_SB(dir),
MAX_INLINE_DATA(dir), GFP_F2FS_ZERO);
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 13:29:18 +03:00
return -ENOMEM;
}
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 13:29:18 +03:00
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>
2018-05-29 19:20:41 +03:00
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 13:29:18 +03:00
unlock_page(ipage);
err = f2fs_add_inline_entries(dir, backup_dentry);
if (err)
goto recover;
lock_page(ipage);
stat_dec_inline_dir(dir);
clear_inode_flag(dir, FI_INLINE_DENTRY);
/*
* 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;
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 13:29:18 +03:00
return 0;
recover:
lock_page(ipage);
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
memcpy(inline_dentry, backup_dentry, MAX_INLINE_DATA(dir));
f2fs_i_depth_write(dir, 0);
f2fs_i_size_write(dir, MAX_INLINE_DATA(dir));
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 13:29:18 +03:00
f2fs_put_page(ipage, 1);
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 13:29:18 +03:00
return err;
}
static int do_convert_inline_dir(struct inode *dir, struct page *ipage,
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 13:29:18 +03:00
{
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);
}
int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(dir);
struct page *ipage;
2020-05-07 10:59:04 +03:00
struct f2fs_filename fname;
void *inline_dentry = NULL;
int err = 0;
if (!f2fs_has_inline_dentry(dir))
return 0;
f2fs_lock_op(sbi);
2020-05-07 10:59:04 +03:00
err = f2fs_setup_filename(dir, &dentry->d_name, 0, &fname);
if (err)
goto out;
ipage = f2fs_get_node_page(sbi, dir->i_ino);
if (IS_ERR(ipage)) {
err = PTR_ERR(ipage);
goto out_fname;
}
if (f2fs_has_enough_room(dir, ipage, &fname)) {
f2fs_put_page(ipage, 1);
goto out_fname;
}
inline_dentry = inline_data_addr(dir, ipage);
err = do_convert_inline_dir(dir, ipage, inline_dentry);
if (!err)
f2fs_put_page(ipage, 1);
out_fname:
2020-05-07 10:59:04 +03:00
f2fs_free_filename(&fname);
out:
f2fs_unlock_op(sbi);
return err;
}
2020-05-07 10:59:04 +03:00
int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
struct inode *inode, nid_t ino, umode_t mode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(dir);
struct page *ipage;
unsigned int bit_pos;
void *inline_dentry = NULL;
struct f2fs_dentry_ptr d;
2020-05-07 10:59:04 +03:00
int slots = GET_DENTRY_SLOTS(fname->disk_name.len);
struct page *page = NULL;
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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(sbi, dir->i_ino);
if (IS_ERR(ipage))
return PTR_ERR(ipage);
inline_dentry = inline_data_addr(dir, ipage);
make_dentry_ptr_inline(dir, &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>
2018-05-29 19:20:41 +03:00
bit_pos = f2fs_room_for_filename(d.bitmap, slots, d.max);
if (bit_pos >= d.max) {
err = do_convert_inline_dir(dir, ipage, inline_dentry);
if (err)
return err;
err = -EAGAIN;
goto out;
}
if (inode) {
f2fs_down_write(&F2FS_I(inode)->i_sem);
2020-05-07 10:59:04 +03:00
page = f2fs_init_inode_metadata(inode, dir, fname, ipage);
if (IS_ERR(page)) {
err = PTR_ERR(page);
goto fail;
}
}
f2fs: avoid deadlock on init_inode_metadata Previously, init_inode_metadata does not hold any parent directory's inode page. So, f2fs_init_acl can grab its parent inode page without any problem. But, when we use inline_dentry, that page is grabbed during f2fs_add_link, so that we can fall into deadlock condition like below. INFO: task mknod:11006 blocked for more than 120 seconds. Tainted: G OE 3.17.0-rc1+ #13 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. mknod D ffff88003fc94580 0 11006 11004 0x00000000 ffff880007717b10 0000000000000002 ffff88003c323220 ffff880007717fd8 0000000000014580 0000000000014580 ffff88003daecb30 ffff88003c323220 ffff88003fc94e80 ffff88003ffbb4e8 ffff880007717ba0 0000000000000002 Call Trace: [<ffffffff8173dc40>] ? bit_wait+0x50/0x50 [<ffffffff8173d4cd>] io_schedule+0x9d/0x130 [<ffffffff8173dc6c>] bit_wait_io+0x2c/0x50 [<ffffffff8173da3b>] __wait_on_bit_lock+0x4b/0xb0 [<ffffffff811640a7>] __lock_page+0x67/0x70 [<ffffffff810acf50>] ? autoremove_wake_function+0x40/0x40 [<ffffffff811652cc>] pagecache_get_page+0x14c/0x1e0 [<ffffffffa029afa9>] get_node_page+0x59/0x130 [f2fs] [<ffffffffa02a63ad>] read_all_xattrs+0x24d/0x430 [f2fs] [<ffffffffa02a6ca2>] f2fs_getxattr+0x52/0xe0 [f2fs] [<ffffffffa02a7481>] f2fs_get_acl+0x41/0x2d0 [f2fs] [<ffffffff8122d847>] get_acl+0x47/0x70 [<ffffffff8122db5a>] posix_acl_create+0x5a/0x150 [<ffffffffa02a7759>] f2fs_init_acl+0x29/0xcb [f2fs] [<ffffffffa0286a8d>] init_inode_metadata+0x5d/0x340 [f2fs] [<ffffffffa029253a>] f2fs_add_inline_entry+0x12a/0x2e0 [f2fs] [<ffffffffa0286ea5>] __f2fs_add_link+0x45/0x4a0 [f2fs] [<ffffffffa028b5b6>] ? f2fs_new_inode+0x146/0x220 [f2fs] [<ffffffffa028b816>] f2fs_mknod+0x86/0xf0 [f2fs] [<ffffffff811e3ec1>] vfs_mknod+0xe1/0x160 [<ffffffff811e4b26>] SyS_mknod+0x1f6/0x200 [<ffffffff81741d7f>] tracesys+0xe1/0xe6 Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-10-14 06:42:53 +04:00
f2fs_wait_on_page_writeback(ipage, NODE, true, true);
2020-05-07 10:59:04 +03:00
f2fs_update_dentry(ino, mode, &d, &fname->disk_name, fname->hash,
bit_pos);
set_page_dirty(ipage);
/* we don't need to mark_inode_dirty now */
if (inode) {
f2fs_i_pino_write(inode, dir->i_ino);
/* synchronize inode page's data from inode cache */
if (is_inode_flag_set(inode, FI_NEW_INODE))
f2fs_update_inode(inode, page);
f2fs_put_page(page, 1);
}
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>
2018-05-29 19:20:41 +03:00
f2fs_update_parent_metadata(dir, inode, 0);
fail:
if (inode)
f2fs_up_write(&F2FS_I(inode)->i_sem);
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)
{
struct f2fs_dentry_ptr d;
void *inline_dentry;
int slots = GET_DENTRY_SLOTS(le16_to_cpu(dentry->name_len));
unsigned int bit_pos;
int i;
lock_page(page);
f2fs_wait_on_page_writeback(page, NODE, true, true);
inline_dentry = inline_data_addr(dir, page);
make_dentry_ptr_inline(dir, &d, inline_dentry);
bit_pos = dentry - d.dentry;
for (i = 0; i < slots; i++)
__clear_bit_le(bit_pos + i, d.bitmap);
set_page_dirty(page);
f2fs_put_page(page, 1);
dir->i_ctime = dir->i_mtime = current_time(dir);
f2fs_mark_inode_dirty_sync(dir, false);
if (inode)
f2fs_drop_nlink(dir, inode);
}
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;
void *inline_dentry;
struct f2fs_dentry_ptr d;
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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(sbi, dir->i_ino);
if (IS_ERR(ipage))
return false;
inline_dentry = inline_data_addr(dir, ipage);
make_dentry_ptr_inline(dir, &d, inline_dentry);
bit_pos = find_next_bit_le(d.bitmap, d.max, bit_pos);
f2fs_put_page(ipage, 1);
if (bit_pos < d.max)
return false;
return true;
}
int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
struct fscrypt_str *fstr)
{
struct inode *inode = file_inode(file);
struct page *ipage = NULL;
struct f2fs_dentry_ptr d;
void *inline_dentry = NULL;
int err;
make_dentry_ptr_inline(inode, &d, inline_dentry);
if (ctx->pos == d.max)
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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(F2FS_I_SB(inode), inode->i_ino);
if (IS_ERR(ipage))
return PTR_ERR(ipage);
f2fs: fix to avoid deadlock in f2fs_read_inline_dir() As Jiqun Li reported in bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=202883 sometimes, dead lock when make system call SYS_getdents64 with fsync() is called by another process. monkey running on android9.0 1. task 9785 held sbi->cp_rwsem and waiting lock_page() 2. task 10349 held mm_sem and waiting sbi->cp_rwsem 3. task 9709 held lock_page() and waiting mm_sem so this is a dead lock scenario. task stack is show by crash tools as following crash_arm64> bt ffffffc03c354080 PID: 9785 TASK: ffffffc03c354080 CPU: 1 COMMAND: "RxIoScheduler-3" >> #7 [ffffffc01b50fac0] __lock_page at ffffff80081b11e8 crash-arm64> bt 10349 PID: 10349 TASK: ffffffc018b83080 CPU: 1 COMMAND: "BUGLY_ASYNC_UPL" >> #3 [ffffffc01f8cfa40] rwsem_down_read_failed at ffffff8008a93afc PC: 00000033 LR: 00000000 SP: 00000000 PSTATE: ffffffffffffffff crash-arm64> bt 9709 PID: 9709 TASK: ffffffc03e7f3080 CPU: 1 COMMAND: "IntentService[A" >> #3 [ffffffc001e67850] rwsem_down_read_failed at ffffff8008a93afc >> #8 [ffffffc001e67b80] el1_ia at ffffff8008084fc4 PC: ffffff8008274114 [compat_filldir64+120] LR: ffffff80083584d4 [f2fs_fill_dentries+448] SP: ffffffc001e67b80 PSTATE: 80400145 X29: ffffffc001e67b80 X28: 0000000000000000 X27: 000000000000001a X26: 00000000000093d7 X25: ffffffc070d52480 X24: 0000000000000008 X23: 0000000000000028 X22: 00000000d43dfd60 X21: ffffffc001e67e90 X20: 0000000000000011 X19: ffffff80093a4000 X18: 0000000000000000 X17: 0000000000000000 X16: 0000000000000000 X15: 0000000000000000 X14: ffffffffffffffff X13: 0000000000000008 X12: 0101010101010101 X11: 7f7f7f7f7f7f7f7f X10: 6a6a6a6a6a6a6a6a X9: 7f7f7f7f7f7f7f7f X8: 0000000080808000 X7: ffffff800827409c X6: 0000000080808000 X5: 0000000000000008 X4: 00000000000093d7 X3: 000000000000001a X2: 0000000000000011 X1: ffffffc070d52480 X0: 0000000000800238 >> #9 [ffffffc001e67be0] f2fs_fill_dentries at ffffff80083584d0 PC: 0000003c LR: 00000000 SP: 00000000 PSTATE: 000000d9 X12: f48a02ff X11: d4678960 X10: d43dfc00 X9: d4678ae4 X8: 00000058 X7: d4678994 X6: d43de800 X5: 000000d9 X4: d43dfc0c X3: d43dfc10 X2: d46799c8 X1: 00000000 X0: 00001068 Below potential deadlock will happen between three threads: Thread A Thread B Thread C - f2fs_do_sync_file - f2fs_write_checkpoint - down_write(&sbi->node_change) -- 1) - do_page_fault - down_write(&mm->mmap_sem) -- 2) - do_wp_page - f2fs_vm_page_mkwrite - getdents64 - f2fs_read_inline_dir - lock_page -- 3) - f2fs_sync_node_pages - lock_page -- 3) - __do_map_lock - down_read(&sbi->node_change) -- 1) - f2fs_fill_dentries - dir_emit - compat_filldir64 - do_page_fault - down_read(&mm->mmap_sem) -- 2) Since f2fs_readdir is protected by inode.i_rwsem, there should not be any updates in inode page, we're safe to lookup dents in inode page without its lock held, so taking off the lock to improve concurrency of readdir and avoid potential deadlock. Reported-by: Jiqun Li <jiqun.li@unisoc.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-03-12 10:44:27 +03:00
/*
* f2fs_readdir was protected by inode.i_rwsem, it is safe to access
* ipage without page's lock held.
*/
unlock_page(ipage);
inline_dentry = inline_data_addr(inode, ipage);
make_dentry_ptr_inline(inode, &d, inline_dentry);
err = f2fs_fill_dentries(ctx, &d, 0, fstr);
if (!err)
ctx->pos = d.max;
f2fs: fix to avoid deadlock in f2fs_read_inline_dir() As Jiqun Li reported in bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=202883 sometimes, dead lock when make system call SYS_getdents64 with fsync() is called by another process. monkey running on android9.0 1. task 9785 held sbi->cp_rwsem and waiting lock_page() 2. task 10349 held mm_sem and waiting sbi->cp_rwsem 3. task 9709 held lock_page() and waiting mm_sem so this is a dead lock scenario. task stack is show by crash tools as following crash_arm64> bt ffffffc03c354080 PID: 9785 TASK: ffffffc03c354080 CPU: 1 COMMAND: "RxIoScheduler-3" >> #7 [ffffffc01b50fac0] __lock_page at ffffff80081b11e8 crash-arm64> bt 10349 PID: 10349 TASK: ffffffc018b83080 CPU: 1 COMMAND: "BUGLY_ASYNC_UPL" >> #3 [ffffffc01f8cfa40] rwsem_down_read_failed at ffffff8008a93afc PC: 00000033 LR: 00000000 SP: 00000000 PSTATE: ffffffffffffffff crash-arm64> bt 9709 PID: 9709 TASK: ffffffc03e7f3080 CPU: 1 COMMAND: "IntentService[A" >> #3 [ffffffc001e67850] rwsem_down_read_failed at ffffff8008a93afc >> #8 [ffffffc001e67b80] el1_ia at ffffff8008084fc4 PC: ffffff8008274114 [compat_filldir64+120] LR: ffffff80083584d4 [f2fs_fill_dentries+448] SP: ffffffc001e67b80 PSTATE: 80400145 X29: ffffffc001e67b80 X28: 0000000000000000 X27: 000000000000001a X26: 00000000000093d7 X25: ffffffc070d52480 X24: 0000000000000008 X23: 0000000000000028 X22: 00000000d43dfd60 X21: ffffffc001e67e90 X20: 0000000000000011 X19: ffffff80093a4000 X18: 0000000000000000 X17: 0000000000000000 X16: 0000000000000000 X15: 0000000000000000 X14: ffffffffffffffff X13: 0000000000000008 X12: 0101010101010101 X11: 7f7f7f7f7f7f7f7f X10: 6a6a6a6a6a6a6a6a X9: 7f7f7f7f7f7f7f7f X8: 0000000080808000 X7: ffffff800827409c X6: 0000000080808000 X5: 0000000000000008 X4: 00000000000093d7 X3: 000000000000001a X2: 0000000000000011 X1: ffffffc070d52480 X0: 0000000000800238 >> #9 [ffffffc001e67be0] f2fs_fill_dentries at ffffff80083584d0 PC: 0000003c LR: 00000000 SP: 00000000 PSTATE: 000000d9 X12: f48a02ff X11: d4678960 X10: d43dfc00 X9: d4678ae4 X8: 00000058 X7: d4678994 X6: d43de800 X5: 000000d9 X4: d43dfc0c X3: d43dfc10 X2: d46799c8 X1: 00000000 X0: 00001068 Below potential deadlock will happen between three threads: Thread A Thread B Thread C - f2fs_do_sync_file - f2fs_write_checkpoint - down_write(&sbi->node_change) -- 1) - do_page_fault - down_write(&mm->mmap_sem) -- 2) - do_wp_page - f2fs_vm_page_mkwrite - getdents64 - f2fs_read_inline_dir - lock_page -- 3) - f2fs_sync_node_pages - lock_page -- 3) - __do_map_lock - down_read(&sbi->node_change) -- 1) - f2fs_fill_dentries - dir_emit - compat_filldir64 - do_page_fault - down_read(&mm->mmap_sem) -- 2) Since f2fs_readdir is protected by inode.i_rwsem, there should not be any updates in inode page, we're safe to lookup dents in inode page without its lock held, so taking off the lock to improve concurrency of readdir and avoid potential deadlock. Reported-by: Jiqun Li <jiqun.li@unisoc.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-03-12 10:44:27 +03:00
f2fs_put_page(ipage, 0);
return err < 0 ? err : 0;
}
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>
2018-05-29 19:20:41 +03:00
ipage = f2fs_get_node_page(F2FS_I_SB(inode), inode->i_ino);
if (IS_ERR(ipage))
return PTR_ERR(ipage);
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)) {
err = -EAGAIN;
goto out;
}
ilen = min_t(size_t, MAX_INLINE_DATA(inode), i_size_read(inode));
if (start >= ilen)
goto out;
if (start + len < ilen)
ilen = start + len;
ilen -= start;
err = f2fs_get_node_info(F2FS_I_SB(inode), inode->i_ino, &ni, false);
if (err)
goto out;
byteaddr = (__u64)ni.blk_addr << inode->i_sb->s_blocksize_bits;
byteaddr += (char *)inline_data_addr(inode, ipage) -
(char *)F2FS_INODE(ipage);
err = fiemap_fill_next_extent(fieinfo, start, byteaddr, ilen, flags);
trace_f2fs_fiemap(inode, start, byteaddr, ilen, flags, err);
out:
f2fs_put_page(ipage, 1);
return err;
}