linux/fs/f2fs/segment.c

4614 lines
116 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* fs/f2fs/segment.c
*
* Copyright (c) 2012 Samsung Electronics Co., Ltd.
* http://www.samsung.com/
*/
#include <linux/fs.h>
#include <linux/f2fs_fs.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/prefetch.h>
#include <linux/kthread.h>
#include <linux/swap.h>
#include <linux/timer.h>
#include <linux/freezer.h>
#include <linux/sched/signal.h>
#include "f2fs.h"
#include "segment.h"
#include "node.h"
#include "gc.h"
#include "trace.h"
#include <trace/events/f2fs.h>
#define __reverse_ffz(x) __reverse_ffs(~(x))
static struct kmem_cache *discard_entry_slab;
static struct kmem_cache *discard_cmd_slab;
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
static struct kmem_cache *sit_entry_set_slab;
static struct kmem_cache *inmem_entry_slab;
static unsigned long __reverse_ulong(unsigned char *str)
{
unsigned long tmp = 0;
int shift = 24, idx = 0;
#if BITS_PER_LONG == 64
shift = 56;
#endif
while (shift >= 0) {
tmp |= (unsigned long)str[idx++] << shift;
shift -= BITS_PER_BYTE;
}
return tmp;
}
/*
* __reverse_ffs is copied from include/asm-generic/bitops/__ffs.h since
* MSB and LSB are reversed in a byte by f2fs_set_bit.
*/
static inline unsigned long __reverse_ffs(unsigned long word)
{
int num = 0;
#if BITS_PER_LONG == 64
if ((word & 0xffffffff00000000UL) == 0)
num += 32;
else
word >>= 32;
#endif
if ((word & 0xffff0000) == 0)
num += 16;
else
word >>= 16;
if ((word & 0xff00) == 0)
num += 8;
else
word >>= 8;
if ((word & 0xf0) == 0)
num += 4;
else
word >>= 4;
if ((word & 0xc) == 0)
num += 2;
else
word >>= 2;
if ((word & 0x2) == 0)
num += 1;
return num;
}
/*
* __find_rev_next(_zero)_bit is copied from lib/find_next_bit.c because
* f2fs_set_bit makes MSB and LSB reversed in a byte.
* @size must be integral times of unsigned long.
* Example:
* MSB <--> LSB
* f2fs_set_bit(0, bitmap) => 1000 0000
* f2fs_set_bit(7, bitmap) => 0000 0001
*/
static unsigned long __find_rev_next_bit(const unsigned long *addr,
unsigned long size, unsigned long offset)
{
const unsigned long *p = addr + BIT_WORD(offset);
unsigned long result = size;
unsigned long tmp;
if (offset >= size)
return size;
size -= (offset & ~(BITS_PER_LONG - 1));
offset %= BITS_PER_LONG;
while (1) {
if (*p == 0)
goto pass;
tmp = __reverse_ulong((unsigned char *)p);
tmp &= ~0UL >> offset;
if (size < BITS_PER_LONG)
tmp &= (~0UL << (BITS_PER_LONG - size));
if (tmp)
goto found;
pass:
if (size <= BITS_PER_LONG)
break;
size -= BITS_PER_LONG;
offset = 0;
p++;
}
return result;
found:
return result - size + __reverse_ffs(tmp);
}
static unsigned long __find_rev_next_zero_bit(const unsigned long *addr,
unsigned long size, unsigned long offset)
{
const unsigned long *p = addr + BIT_WORD(offset);
unsigned long result = size;
unsigned long tmp;
if (offset >= size)
return size;
size -= (offset & ~(BITS_PER_LONG - 1));
offset %= BITS_PER_LONG;
while (1) {
if (*p == ~0UL)
goto pass;
tmp = __reverse_ulong((unsigned char *)p);
if (offset)
tmp |= ~0UL << (BITS_PER_LONG - offset);
if (size < BITS_PER_LONG)
tmp |= ~0UL >> size;
if (tmp != ~0UL)
goto found;
pass:
if (size <= BITS_PER_LONG)
break;
size -= BITS_PER_LONG;
offset = 0;
p++;
}
return result;
found:
return result - size + __reverse_ffz(tmp);
}
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 16:20:41 +00:00
bool f2fs_need_SSR(struct f2fs_sb_info *sbi)
{
int node_secs = get_blocktype_secs(sbi, F2FS_DIRTY_NODES);
int dent_secs = get_blocktype_secs(sbi, F2FS_DIRTY_DENTS);
int imeta_secs = get_blocktype_secs(sbi, F2FS_DIRTY_IMETA);
if (test_opt(sbi, LFS))
return false;
if (sbi->gc_mode == GC_URGENT)
return true;
if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
return true;
return free_sections(sbi) <= (node_secs + 2 * dent_secs + imeta_secs +
SM_I(sbi)->min_ssr_sections + reserved_sections(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 16:20:41 +00:00
void f2fs_register_inmem_page(struct inode *inode, struct page *page)
{
struct inmem_pages *new;
f2fs_trace_pid(page);
f2fs_set_page_private(page, (unsigned long)ATOMIC_WRITTEN_PAGE);
new = f2fs_kmem_cache_alloc(inmem_entry_slab, GFP_NOFS);
/* add atomic page indices to the list */
new->page = page;
INIT_LIST_HEAD(&new->list);
/* increase reference count with clean state */
get_page(page);
f2fs: avoid infinite GC loop due to stale atomic files If committing atomic pages is failed when doing f2fs_do_sync_file(), we can get commited pages but atomic_file being still set like: - inmem: 0, atomic IO: 4 (Max. 10), volatile IO: 0 (Max. 0) If GC selects this block, we can get an infinite loop like this: f2fs_submit_page_bio: dev = (253,7), ino = 2, page_index = 0x2359a8, oldaddr = 0x2359a8, newaddr = 0x2359a8, rw = READ(), type = COLD_DATA f2fs_submit_read_bio: dev = (253,7)/(253,7), rw = READ(), DATA, sector = 18533696, size = 4096 f2fs_get_victim: dev = (253,7), type = No TYPE, policy = (Foreground GC, LFS-mode, Greedy), victim = 4355, cost = 1, ofs_unit = 1, pre_victim_secno = 4355, prefree = 0, free = 234 f2fs_iget: dev = (253,7), ino = 6247, pino = 5845, i_mode = 0x81b0, i_size = 319488, i_nlink = 1, i_blocks = 624, i_advise = 0x2c f2fs_submit_page_bio: dev = (253,7), ino = 2, page_index = 0x2359a8, oldaddr = 0x2359a8, newaddr = 0x2359a8, rw = READ(), type = COLD_DATA f2fs_submit_read_bio: dev = (253,7)/(253,7), rw = READ(), DATA, sector = 18533696, size = 4096 f2fs_get_victim: dev = (253,7), type = No TYPE, policy = (Foreground GC, LFS-mode, Greedy), victim = 4355, cost = 1, ofs_unit = 1, pre_victim_secno = 4355, prefree = 0, free = 234 f2fs_iget: dev = (253,7), ino = 6247, pino = 5845, i_mode = 0x81b0, i_size = 319488, i_nlink = 1, i_blocks = 624, i_advise = 0x2c In that moment, we can observe: [Before] Try to move 5084219 blocks (BG: 384508) - data blocks : 4962373 (274483) - node blocks : 121846 (110025) Skipped : atomic write 4534686 (10) [After] Try to move 5088973 blocks (BG: 384508) - data blocks : 4967127 (274483) - node blocks : 121846 (110025) Skipped : atomic write 4539440 (10) So, refactor atomic_write flow like this: 1. start_atomic_write - add inmem_list and set atomic_file 2. write() - register it in inmem_pages 3. commit_atomic_write - if no error, f2fs_drop_inmem_pages() - f2fs_commit_inmme_pages() failed : __revoked_inmem_pages() was done - f2fs_do_sync_file failed : abort_atomic_write later 4. abort_atomic_write - f2fs_drop_inmem_pages 5. f2fs_drop_inmem_pages - clear atomic_file - remove inmem_list Based on this change, when GC fails to move block in atomic_file, f2fs_drop_inmem_pages_all() can call f2fs_drop_inmem_pages(). Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-09-09 12:10:59 +00:00
mutex_lock(&F2FS_I(inode)->inmem_lock);
list_add_tail(&new->list, &F2FS_I(inode)->inmem_pages);
inc_page_count(F2FS_I_SB(inode), F2FS_INMEM_PAGES);
f2fs: avoid infinite GC loop due to stale atomic files If committing atomic pages is failed when doing f2fs_do_sync_file(), we can get commited pages but atomic_file being still set like: - inmem: 0, atomic IO: 4 (Max. 10), volatile IO: 0 (Max. 0) If GC selects this block, we can get an infinite loop like this: f2fs_submit_page_bio: dev = (253,7), ino = 2, page_index = 0x2359a8, oldaddr = 0x2359a8, newaddr = 0x2359a8, rw = READ(), type = COLD_DATA f2fs_submit_read_bio: dev = (253,7)/(253,7), rw = READ(), DATA, sector = 18533696, size = 4096 f2fs_get_victim: dev = (253,7), type = No TYPE, policy = (Foreground GC, LFS-mode, Greedy), victim = 4355, cost = 1, ofs_unit = 1, pre_victim_secno = 4355, prefree = 0, free = 234 f2fs_iget: dev = (253,7), ino = 6247, pino = 5845, i_mode = 0x81b0, i_size = 319488, i_nlink = 1, i_blocks = 624, i_advise = 0x2c f2fs_submit_page_bio: dev = (253,7), ino = 2, page_index = 0x2359a8, oldaddr = 0x2359a8, newaddr = 0x2359a8, rw = READ(), type = COLD_DATA f2fs_submit_read_bio: dev = (253,7)/(253,7), rw = READ(), DATA, sector = 18533696, size = 4096 f2fs_get_victim: dev = (253,7), type = No TYPE, policy = (Foreground GC, LFS-mode, Greedy), victim = 4355, cost = 1, ofs_unit = 1, pre_victim_secno = 4355, prefree = 0, free = 234 f2fs_iget: dev = (253,7), ino = 6247, pino = 5845, i_mode = 0x81b0, i_size = 319488, i_nlink = 1, i_blocks = 624, i_advise = 0x2c In that moment, we can observe: [Before] Try to move 5084219 blocks (BG: 384508) - data blocks : 4962373 (274483) - node blocks : 121846 (110025) Skipped : atomic write 4534686 (10) [After] Try to move 5088973 blocks (BG: 384508) - data blocks : 4967127 (274483) - node blocks : 121846 (110025) Skipped : atomic write 4539440 (10) So, refactor atomic_write flow like this: 1. start_atomic_write - add inmem_list and set atomic_file 2. write() - register it in inmem_pages 3. commit_atomic_write - if no error, f2fs_drop_inmem_pages() - f2fs_commit_inmme_pages() failed : __revoked_inmem_pages() was done - f2fs_do_sync_file failed : abort_atomic_write later 4. abort_atomic_write - f2fs_drop_inmem_pages 5. f2fs_drop_inmem_pages - clear atomic_file - remove inmem_list Based on this change, when GC fails to move block in atomic_file, f2fs_drop_inmem_pages_all() can call f2fs_drop_inmem_pages(). Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-09-09 12:10:59 +00:00
mutex_unlock(&F2FS_I(inode)->inmem_lock);
trace_f2fs_register_inmem_page(page, INMEM);
}
static int __revoke_inmem_pages(struct inode *inode,
f2fs: fix to avoid deadlock of atomic file operations Thread A Thread B - __fput - f2fs_release_file - drop_inmem_pages - mutex_lock(&fi->inmem_lock) - __revoke_inmem_pages - lock_page(page) - open - f2fs_setattr - truncate_setsize - truncate_inode_pages_range - lock_page(page) - truncate_cleanup_page - f2fs_invalidate_page - drop_inmem_page - mutex_lock(&fi->inmem_lock); We may encounter above ABBA deadlock as reported by Kyungtae Kim: I'm reporting a bug in linux-4.17.19: "INFO: task hung in drop_inmem_page" (no reproducer) I think this might be somehow related to the following: https://groups.google.com/forum/#!searchin/syzkaller-bugs/INFO$3A$20task$20hung$20in$20%7Csort:date/syzkaller-bugs/c6soBTrdaIo/AjAzPeIzCgAJ ========================================= INFO: task syz-executor7:10822 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D27024 10822 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 schedule_preempt_disabled+0x18/0x30 kernel/sched/core.c:3617 __mutex_lock_common kernel/locking/mutex.c:833 [inline] __mutex_lock+0x5bd/0x1410 kernel/locking/mutex.c:893 mutex_lock_nested+0x1b/0x20 kernel/locking/mutex.c:908 drop_inmem_page+0xcb/0x810 fs/f2fs/segment.c:327 f2fs_invalidate_page+0x337/0x5e0 fs/f2fs/data.c:2401 do_invalidatepage mm/truncate.c:165 [inline] truncate_cleanup_page+0x261/0x330 mm/truncate.c:187 truncate_inode_pages_range+0x552/0x1610 mm/truncate.c:367 truncate_inode_pages mm/truncate.c:478 [inline] truncate_pagecache+0x6d/0x90 mm/truncate.c:801 truncate_setsize+0x81/0xa0 mm/truncate.c:826 f2fs_setattr+0x44f/0x1270 fs/f2fs/file.c:781 notify_change+0xa62/0xe80 fs/attr.c:313 do_truncate+0x12e/0x1e0 fs/open.c:63 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e459c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e45a6cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071bea0 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e45a700 INFO: task syz-executor7:10858 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D28880 10858 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 __rwsem_down_write_failed_common kernel/locking/rwsem-xadd.c:565 [inline] rwsem_down_write_failed+0x5e6/0xc90 kernel/locking/rwsem-xadd.c:594 call_rwsem_down_write_failed+0x17/0x30 arch/x86/lib/rwsem.S:117 __down_write arch/x86/include/asm/rwsem.h:142 [inline] down_write+0x58/0xa0 kernel/locking/rwsem.c:72 inode_lock include/linux/fs.h:713 [inline] do_truncate+0x120/0x1e0 fs/open.c:61 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e3b4c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e3b56cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071c238 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e3b5700 INFO: task syz-executor5:10829 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor5 D28760 10829 6308 0x80000002 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 io_schedule+0x21/0x80 kernel/sched/core.c:5179 wait_on_page_bit_common mm/filemap.c:1100 [inline] __lock_page+0x2b5/0x390 mm/filemap.c:1273 lock_page include/linux/pagemap.h:483 [inline] __revoke_inmem_pages+0xb35/0x11c0 fs/f2fs/segment.c:231 drop_inmem_pages+0xa3/0x3e0 fs/f2fs/segment.c:306 f2fs_release_file+0x2c7/0x330 fs/f2fs/file.c:1556 __fput+0x2c7/0x780 fs/file_table.c:209 ____fput+0x1a/0x20 fs/file_table.c:243 task_work_run+0x151/0x1d0 kernel/task_work.c:113 exit_task_work include/linux/task_work.h:22 [inline] do_exit+0x8ba/0x30a0 kernel/exit.c:865 do_group_exit+0x13b/0x3a0 kernel/exit.c:968 get_signal+0x6bb/0x1650 kernel/signal.c:2482 do_signal+0x84/0x1b70 arch/x86/kernel/signal.c:810 exit_to_usermode_loop+0x155/0x190 arch/x86/entry/common.c:162 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x445/0x4e0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f1c68e74ce8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: fffffffffffffe00 RBX: 000000000071bf80 RCX: 00000000004497b9 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000071bf80 RBP: 000000000071bf80 R08: 0000000000000000 R09: 000000000071bf58 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 0000000000000000 R14: 00007f1c68e759c0 R15: 00007f1c68e75700 This patch tries to use trylock_page to mitigate such deadlock condition for fix. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-25 09:11:03 +00:00
struct list_head *head, bool drop, bool recover,
bool trylock)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct inmem_pages *cur, *tmp;
int err = 0;
list_for_each_entry_safe(cur, tmp, head, list) {
struct page *page = cur->page;
if (drop)
trace_f2fs_commit_inmem_page(page, INMEM_DROP);
f2fs: fix to avoid deadlock of atomic file operations Thread A Thread B - __fput - f2fs_release_file - drop_inmem_pages - mutex_lock(&fi->inmem_lock) - __revoke_inmem_pages - lock_page(page) - open - f2fs_setattr - truncate_setsize - truncate_inode_pages_range - lock_page(page) - truncate_cleanup_page - f2fs_invalidate_page - drop_inmem_page - mutex_lock(&fi->inmem_lock); We may encounter above ABBA deadlock as reported by Kyungtae Kim: I'm reporting a bug in linux-4.17.19: "INFO: task hung in drop_inmem_page" (no reproducer) I think this might be somehow related to the following: https://groups.google.com/forum/#!searchin/syzkaller-bugs/INFO$3A$20task$20hung$20in$20%7Csort:date/syzkaller-bugs/c6soBTrdaIo/AjAzPeIzCgAJ ========================================= INFO: task syz-executor7:10822 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D27024 10822 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 schedule_preempt_disabled+0x18/0x30 kernel/sched/core.c:3617 __mutex_lock_common kernel/locking/mutex.c:833 [inline] __mutex_lock+0x5bd/0x1410 kernel/locking/mutex.c:893 mutex_lock_nested+0x1b/0x20 kernel/locking/mutex.c:908 drop_inmem_page+0xcb/0x810 fs/f2fs/segment.c:327 f2fs_invalidate_page+0x337/0x5e0 fs/f2fs/data.c:2401 do_invalidatepage mm/truncate.c:165 [inline] truncate_cleanup_page+0x261/0x330 mm/truncate.c:187 truncate_inode_pages_range+0x552/0x1610 mm/truncate.c:367 truncate_inode_pages mm/truncate.c:478 [inline] truncate_pagecache+0x6d/0x90 mm/truncate.c:801 truncate_setsize+0x81/0xa0 mm/truncate.c:826 f2fs_setattr+0x44f/0x1270 fs/f2fs/file.c:781 notify_change+0xa62/0xe80 fs/attr.c:313 do_truncate+0x12e/0x1e0 fs/open.c:63 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e459c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e45a6cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071bea0 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e45a700 INFO: task syz-executor7:10858 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D28880 10858 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 __rwsem_down_write_failed_common kernel/locking/rwsem-xadd.c:565 [inline] rwsem_down_write_failed+0x5e6/0xc90 kernel/locking/rwsem-xadd.c:594 call_rwsem_down_write_failed+0x17/0x30 arch/x86/lib/rwsem.S:117 __down_write arch/x86/include/asm/rwsem.h:142 [inline] down_write+0x58/0xa0 kernel/locking/rwsem.c:72 inode_lock include/linux/fs.h:713 [inline] do_truncate+0x120/0x1e0 fs/open.c:61 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e3b4c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e3b56cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071c238 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e3b5700 INFO: task syz-executor5:10829 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor5 D28760 10829 6308 0x80000002 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 io_schedule+0x21/0x80 kernel/sched/core.c:5179 wait_on_page_bit_common mm/filemap.c:1100 [inline] __lock_page+0x2b5/0x390 mm/filemap.c:1273 lock_page include/linux/pagemap.h:483 [inline] __revoke_inmem_pages+0xb35/0x11c0 fs/f2fs/segment.c:231 drop_inmem_pages+0xa3/0x3e0 fs/f2fs/segment.c:306 f2fs_release_file+0x2c7/0x330 fs/f2fs/file.c:1556 __fput+0x2c7/0x780 fs/file_table.c:209 ____fput+0x1a/0x20 fs/file_table.c:243 task_work_run+0x151/0x1d0 kernel/task_work.c:113 exit_task_work include/linux/task_work.h:22 [inline] do_exit+0x8ba/0x30a0 kernel/exit.c:865 do_group_exit+0x13b/0x3a0 kernel/exit.c:968 get_signal+0x6bb/0x1650 kernel/signal.c:2482 do_signal+0x84/0x1b70 arch/x86/kernel/signal.c:810 exit_to_usermode_loop+0x155/0x190 arch/x86/entry/common.c:162 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x445/0x4e0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f1c68e74ce8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: fffffffffffffe00 RBX: 000000000071bf80 RCX: 00000000004497b9 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000071bf80 RBP: 000000000071bf80 R08: 0000000000000000 R09: 000000000071bf58 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 0000000000000000 R14: 00007f1c68e759c0 R15: 00007f1c68e75700 This patch tries to use trylock_page to mitigate such deadlock condition for fix. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-25 09:11:03 +00:00
if (trylock) {
/*
* to avoid deadlock in between page lock and
* inmem_lock.
*/
if (!trylock_page(page))
continue;
} else {
lock_page(page);
}
f2fs_wait_on_page_writeback(page, DATA, true, true);
if (recover) {
struct dnode_of_data dn;
struct node_info ni;
trace_f2fs_commit_inmem_page(page, INMEM_REVOKE);
retry:
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 16:20:41 +00:00
err = f2fs_get_dnode_of_data(&dn, page->index,
LOOKUP_NODE);
if (err) {
if (err == -ENOMEM) {
congestion_wait(BLK_RW_ASYNC, HZ/50);
cond_resched();
goto retry;
}
err = -EAGAIN;
goto next;
}
err = f2fs_get_node_info(sbi, dn.nid, &ni);
if (err) {
f2fs_put_dnode(&dn);
return err;
}
if (cur->old_addr == NEW_ADDR) {
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 16:20:41 +00:00
f2fs_invalidate_blocks(sbi, dn.data_blkaddr);
f2fs_update_data_blkaddr(&dn, NEW_ADDR);
} else
f2fs_replace_block(sbi, &dn, dn.data_blkaddr,
cur->old_addr, ni.version, true, true);
f2fs_put_dnode(&dn);
}
next:
/* we don't need to invalidate this in the sccessful status */
if (drop || recover) {
ClearPageUptodate(page);
clear_cold_data(page);
}
f2fs_clear_page_private(page);
f2fs_put_page(page, 1);
list_del(&cur->list);
kmem_cache_free(inmem_entry_slab, cur);
dec_page_count(F2FS_I_SB(inode), F2FS_INMEM_PAGES);
}
return err;
}
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 16:20:41 +00:00
void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure)
{
struct list_head *head = &sbi->inode_list[ATOMIC_FILE];
struct inode *inode;
struct f2fs_inode_info *fi;
unsigned int count = sbi->atomic_files;
unsigned int looped = 0;
next:
spin_lock(&sbi->inode_lock[ATOMIC_FILE]);
if (list_empty(head)) {
spin_unlock(&sbi->inode_lock[ATOMIC_FILE]);
return;
}
fi = list_first_entry(head, struct f2fs_inode_info, inmem_ilist);
inode = igrab(&fi->vfs_inode);
if (inode)
list_move_tail(&fi->inmem_ilist, head);
spin_unlock(&sbi->inode_lock[ATOMIC_FILE]);
if (inode) {
if (gc_failure) {
if (!fi->i_gc_failures[GC_FAILURE_ATOMIC])
goto skip;
}
set_inode_flag(inode, FI_ATOMIC_REVOKE_REQUEST);
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 16:20:41 +00:00
f2fs_drop_inmem_pages(inode);
skip:
iput(inode);
}
congestion_wait(BLK_RW_ASYNC, HZ/50);
cond_resched();
if (gc_failure) {
if (++looped >= count)
return;
}
goto next;
}
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 16:20:41 +00:00
void f2fs_drop_inmem_pages(struct inode *inode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct f2fs_inode_info *fi = F2FS_I(inode);
f2fs: fix to avoid deadlock of atomic file operations Thread A Thread B - __fput - f2fs_release_file - drop_inmem_pages - mutex_lock(&fi->inmem_lock) - __revoke_inmem_pages - lock_page(page) - open - f2fs_setattr - truncate_setsize - truncate_inode_pages_range - lock_page(page) - truncate_cleanup_page - f2fs_invalidate_page - drop_inmem_page - mutex_lock(&fi->inmem_lock); We may encounter above ABBA deadlock as reported by Kyungtae Kim: I'm reporting a bug in linux-4.17.19: "INFO: task hung in drop_inmem_page" (no reproducer) I think this might be somehow related to the following: https://groups.google.com/forum/#!searchin/syzkaller-bugs/INFO$3A$20task$20hung$20in$20%7Csort:date/syzkaller-bugs/c6soBTrdaIo/AjAzPeIzCgAJ ========================================= INFO: task syz-executor7:10822 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D27024 10822 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 schedule_preempt_disabled+0x18/0x30 kernel/sched/core.c:3617 __mutex_lock_common kernel/locking/mutex.c:833 [inline] __mutex_lock+0x5bd/0x1410 kernel/locking/mutex.c:893 mutex_lock_nested+0x1b/0x20 kernel/locking/mutex.c:908 drop_inmem_page+0xcb/0x810 fs/f2fs/segment.c:327 f2fs_invalidate_page+0x337/0x5e0 fs/f2fs/data.c:2401 do_invalidatepage mm/truncate.c:165 [inline] truncate_cleanup_page+0x261/0x330 mm/truncate.c:187 truncate_inode_pages_range+0x552/0x1610 mm/truncate.c:367 truncate_inode_pages mm/truncate.c:478 [inline] truncate_pagecache+0x6d/0x90 mm/truncate.c:801 truncate_setsize+0x81/0xa0 mm/truncate.c:826 f2fs_setattr+0x44f/0x1270 fs/f2fs/file.c:781 notify_change+0xa62/0xe80 fs/attr.c:313 do_truncate+0x12e/0x1e0 fs/open.c:63 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e459c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e45a6cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071bea0 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e45a700 INFO: task syz-executor7:10858 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D28880 10858 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 __rwsem_down_write_failed_common kernel/locking/rwsem-xadd.c:565 [inline] rwsem_down_write_failed+0x5e6/0xc90 kernel/locking/rwsem-xadd.c:594 call_rwsem_down_write_failed+0x17/0x30 arch/x86/lib/rwsem.S:117 __down_write arch/x86/include/asm/rwsem.h:142 [inline] down_write+0x58/0xa0 kernel/locking/rwsem.c:72 inode_lock include/linux/fs.h:713 [inline] do_truncate+0x120/0x1e0 fs/open.c:61 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e3b4c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e3b56cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071c238 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e3b5700 INFO: task syz-executor5:10829 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor5 D28760 10829 6308 0x80000002 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 io_schedule+0x21/0x80 kernel/sched/core.c:5179 wait_on_page_bit_common mm/filemap.c:1100 [inline] __lock_page+0x2b5/0x390 mm/filemap.c:1273 lock_page include/linux/pagemap.h:483 [inline] __revoke_inmem_pages+0xb35/0x11c0 fs/f2fs/segment.c:231 drop_inmem_pages+0xa3/0x3e0 fs/f2fs/segment.c:306 f2fs_release_file+0x2c7/0x330 fs/f2fs/file.c:1556 __fput+0x2c7/0x780 fs/file_table.c:209 ____fput+0x1a/0x20 fs/file_table.c:243 task_work_run+0x151/0x1d0 kernel/task_work.c:113 exit_task_work include/linux/task_work.h:22 [inline] do_exit+0x8ba/0x30a0 kernel/exit.c:865 do_group_exit+0x13b/0x3a0 kernel/exit.c:968 get_signal+0x6bb/0x1650 kernel/signal.c:2482 do_signal+0x84/0x1b70 arch/x86/kernel/signal.c:810 exit_to_usermode_loop+0x155/0x190 arch/x86/entry/common.c:162 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x445/0x4e0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f1c68e74ce8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: fffffffffffffe00 RBX: 000000000071bf80 RCX: 00000000004497b9 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000071bf80 RBP: 000000000071bf80 R08: 0000000000000000 R09: 000000000071bf58 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 0000000000000000 R14: 00007f1c68e759c0 R15: 00007f1c68e75700 This patch tries to use trylock_page to mitigate such deadlock condition for fix. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-25 09:11:03 +00:00
while (!list_empty(&fi->inmem_pages)) {
mutex_lock(&fi->inmem_lock);
__revoke_inmem_pages(inode, &fi->inmem_pages,
true, false, true);
mutex_unlock(&fi->inmem_lock);
}
fi->i_gc_failures[GC_FAILURE_ATOMIC] = 0;
stat_dec_atomic_write(inode);
f2fs: avoid infinite GC loop due to stale atomic files If committing atomic pages is failed when doing f2fs_do_sync_file(), we can get commited pages but atomic_file being still set like: - inmem: 0, atomic IO: 4 (Max. 10), volatile IO: 0 (Max. 0) If GC selects this block, we can get an infinite loop like this: f2fs_submit_page_bio: dev = (253,7), ino = 2, page_index = 0x2359a8, oldaddr = 0x2359a8, newaddr = 0x2359a8, rw = READ(), type = COLD_DATA f2fs_submit_read_bio: dev = (253,7)/(253,7), rw = READ(), DATA, sector = 18533696, size = 4096 f2fs_get_victim: dev = (253,7), type = No TYPE, policy = (Foreground GC, LFS-mode, Greedy), victim = 4355, cost = 1, ofs_unit = 1, pre_victim_secno = 4355, prefree = 0, free = 234 f2fs_iget: dev = (253,7), ino = 6247, pino = 5845, i_mode = 0x81b0, i_size = 319488, i_nlink = 1, i_blocks = 624, i_advise = 0x2c f2fs_submit_page_bio: dev = (253,7), ino = 2, page_index = 0x2359a8, oldaddr = 0x2359a8, newaddr = 0x2359a8, rw = READ(), type = COLD_DATA f2fs_submit_read_bio: dev = (253,7)/(253,7), rw = READ(), DATA, sector = 18533696, size = 4096 f2fs_get_victim: dev = (253,7), type = No TYPE, policy = (Foreground GC, LFS-mode, Greedy), victim = 4355, cost = 1, ofs_unit = 1, pre_victim_secno = 4355, prefree = 0, free = 234 f2fs_iget: dev = (253,7), ino = 6247, pino = 5845, i_mode = 0x81b0, i_size = 319488, i_nlink = 1, i_blocks = 624, i_advise = 0x2c In that moment, we can observe: [Before] Try to move 5084219 blocks (BG: 384508) - data blocks : 4962373 (274483) - node blocks : 121846 (110025) Skipped : atomic write 4534686 (10) [After] Try to move 5088973 blocks (BG: 384508) - data blocks : 4967127 (274483) - node blocks : 121846 (110025) Skipped : atomic write 4539440 (10) So, refactor atomic_write flow like this: 1. start_atomic_write - add inmem_list and set atomic_file 2. write() - register it in inmem_pages 3. commit_atomic_write - if no error, f2fs_drop_inmem_pages() - f2fs_commit_inmme_pages() failed : __revoked_inmem_pages() was done - f2fs_do_sync_file failed : abort_atomic_write later 4. abort_atomic_write - f2fs_drop_inmem_pages 5. f2fs_drop_inmem_pages - clear atomic_file - remove inmem_list Based on this change, when GC fails to move block in atomic_file, f2fs_drop_inmem_pages_all() can call f2fs_drop_inmem_pages(). Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-09-09 12:10:59 +00:00
spin_lock(&sbi->inode_lock[ATOMIC_FILE]);
if (!list_empty(&fi->inmem_ilist))
list_del_init(&fi->inmem_ilist);
if (f2fs_is_atomic_file(inode)) {
clear_inode_flag(inode, FI_ATOMIC_FILE);
sbi->atomic_files--;
}
f2fs: avoid infinite GC loop due to stale atomic files If committing atomic pages is failed when doing f2fs_do_sync_file(), we can get commited pages but atomic_file being still set like: - inmem: 0, atomic IO: 4 (Max. 10), volatile IO: 0 (Max. 0) If GC selects this block, we can get an infinite loop like this: f2fs_submit_page_bio: dev = (253,7), ino = 2, page_index = 0x2359a8, oldaddr = 0x2359a8, newaddr = 0x2359a8, rw = READ(), type = COLD_DATA f2fs_submit_read_bio: dev = (253,7)/(253,7), rw = READ(), DATA, sector = 18533696, size = 4096 f2fs_get_victim: dev = (253,7), type = No TYPE, policy = (Foreground GC, LFS-mode, Greedy), victim = 4355, cost = 1, ofs_unit = 1, pre_victim_secno = 4355, prefree = 0, free = 234 f2fs_iget: dev = (253,7), ino = 6247, pino = 5845, i_mode = 0x81b0, i_size = 319488, i_nlink = 1, i_blocks = 624, i_advise = 0x2c f2fs_submit_page_bio: dev = (253,7), ino = 2, page_index = 0x2359a8, oldaddr = 0x2359a8, newaddr = 0x2359a8, rw = READ(), type = COLD_DATA f2fs_submit_read_bio: dev = (253,7)/(253,7), rw = READ(), DATA, sector = 18533696, size = 4096 f2fs_get_victim: dev = (253,7), type = No TYPE, policy = (Foreground GC, LFS-mode, Greedy), victim = 4355, cost = 1, ofs_unit = 1, pre_victim_secno = 4355, prefree = 0, free = 234 f2fs_iget: dev = (253,7), ino = 6247, pino = 5845, i_mode = 0x81b0, i_size = 319488, i_nlink = 1, i_blocks = 624, i_advise = 0x2c In that moment, we can observe: [Before] Try to move 5084219 blocks (BG: 384508) - data blocks : 4962373 (274483) - node blocks : 121846 (110025) Skipped : atomic write 4534686 (10) [After] Try to move 5088973 blocks (BG: 384508) - data blocks : 4967127 (274483) - node blocks : 121846 (110025) Skipped : atomic write 4539440 (10) So, refactor atomic_write flow like this: 1. start_atomic_write - add inmem_list and set atomic_file 2. write() - register it in inmem_pages 3. commit_atomic_write - if no error, f2fs_drop_inmem_pages() - f2fs_commit_inmme_pages() failed : __revoked_inmem_pages() was done - f2fs_do_sync_file failed : abort_atomic_write later 4. abort_atomic_write - f2fs_drop_inmem_pages 5. f2fs_drop_inmem_pages - clear atomic_file - remove inmem_list Based on this change, when GC fails to move block in atomic_file, f2fs_drop_inmem_pages_all() can call f2fs_drop_inmem_pages(). Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-09-09 12:10:59 +00:00
spin_unlock(&sbi->inode_lock[ATOMIC_FILE]);
}
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 16:20:41 +00:00
void f2fs_drop_inmem_page(struct inode *inode, struct page *page)
f2fs: fix stale ATOMIC_WRITTEN_PAGE private pointer When I forced to enable atomic operations intentionally, I could hit the below panic, since we didn't clear page->private in f2fs_invalidate_page called by file truncation. The panic occurs due to NULL mapping having page->private. BUG: unable to handle kernel paging request at ffffffffffffffff IP: drop_buffers+0x38/0xe0 PGD 5d00c067 PUD 5d00e067 PMD 0 CPU: 3 PID: 1648 Comm: fsstress Tainted: G D OE 4.10.0+ #5 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 task: ffff9151952863c0 task.stack: ffffaaec40db4000 RIP: 0010:drop_buffers+0x38/0xe0 RSP: 0018:ffffaaec40db74c8 EFLAGS: 00010292 Call Trace: ? page_referenced+0x8b/0x170 try_to_free_buffers+0xc5/0xe0 try_to_release_page+0x49/0x50 shrink_page_list+0x8bc/0x9f0 shrink_inactive_list+0x1dd/0x500 ? shrink_active_list+0x2c0/0x430 shrink_node_memcg+0x5eb/0x7c0 shrink_node+0xe1/0x320 do_try_to_free_pages+0xef/0x2e0 try_to_free_pages+0xe9/0x190 __alloc_pages_slowpath+0x390/0xe70 __alloc_pages_nodemask+0x291/0x2b0 alloc_pages_current+0x95/0x140 __page_cache_alloc+0xc4/0xe0 pagecache_get_page+0xab/0x2a0 grab_cache_page_write_begin+0x20/0x40 get_read_data_page+0x2e6/0x4c0 [f2fs] ? f2fs_mark_inode_dirty_sync+0x16/0x30 [f2fs] ? truncate_data_blocks_range+0x238/0x2b0 [f2fs] get_lock_data_page+0x30/0x190 [f2fs] __exchange_data_block+0xaaf/0xf40 [f2fs] f2fs_fallocate+0x418/0xd00 [f2fs] vfs_fallocate+0x157/0x220 SyS_fallocate+0x48/0x80 Signed-off-by: Yunlei He <heyunlei@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> [Chao Yu: use INMEM_INVALIDATE for better tracing] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-03-17 01:55:52 +00:00
{
struct f2fs_inode_info *fi = F2FS_I(inode);
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct list_head *head = &fi->inmem_pages;
struct inmem_pages *cur = NULL;
f2fs_bug_on(sbi, !IS_ATOMIC_WRITTEN_PAGE(page));
mutex_lock(&fi->inmem_lock);
list_for_each_entry(cur, head, list) {
if (cur->page == page)
break;
}
f2fs_bug_on(sbi, list_empty(head) || cur->page != page);
f2fs: fix stale ATOMIC_WRITTEN_PAGE private pointer When I forced to enable atomic operations intentionally, I could hit the below panic, since we didn't clear page->private in f2fs_invalidate_page called by file truncation. The panic occurs due to NULL mapping having page->private. BUG: unable to handle kernel paging request at ffffffffffffffff IP: drop_buffers+0x38/0xe0 PGD 5d00c067 PUD 5d00e067 PMD 0 CPU: 3 PID: 1648 Comm: fsstress Tainted: G D OE 4.10.0+ #5 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 task: ffff9151952863c0 task.stack: ffffaaec40db4000 RIP: 0010:drop_buffers+0x38/0xe0 RSP: 0018:ffffaaec40db74c8 EFLAGS: 00010292 Call Trace: ? page_referenced+0x8b/0x170 try_to_free_buffers+0xc5/0xe0 try_to_release_page+0x49/0x50 shrink_page_list+0x8bc/0x9f0 shrink_inactive_list+0x1dd/0x500 ? shrink_active_list+0x2c0/0x430 shrink_node_memcg+0x5eb/0x7c0 shrink_node+0xe1/0x320 do_try_to_free_pages+0xef/0x2e0 try_to_free_pages+0xe9/0x190 __alloc_pages_slowpath+0x390/0xe70 __alloc_pages_nodemask+0x291/0x2b0 alloc_pages_current+0x95/0x140 __page_cache_alloc+0xc4/0xe0 pagecache_get_page+0xab/0x2a0 grab_cache_page_write_begin+0x20/0x40 get_read_data_page+0x2e6/0x4c0 [f2fs] ? f2fs_mark_inode_dirty_sync+0x16/0x30 [f2fs] ? truncate_data_blocks_range+0x238/0x2b0 [f2fs] get_lock_data_page+0x30/0x190 [f2fs] __exchange_data_block+0xaaf/0xf40 [f2fs] f2fs_fallocate+0x418/0xd00 [f2fs] vfs_fallocate+0x157/0x220 SyS_fallocate+0x48/0x80 Signed-off-by: Yunlei He <heyunlei@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> [Chao Yu: use INMEM_INVALIDATE for better tracing] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-03-17 01:55:52 +00:00
list_del(&cur->list);
mutex_unlock(&fi->inmem_lock);
dec_page_count(sbi, F2FS_INMEM_PAGES);
kmem_cache_free(inmem_entry_slab, cur);
ClearPageUptodate(page);
f2fs_clear_page_private(page);
f2fs: fix stale ATOMIC_WRITTEN_PAGE private pointer When I forced to enable atomic operations intentionally, I could hit the below panic, since we didn't clear page->private in f2fs_invalidate_page called by file truncation. The panic occurs due to NULL mapping having page->private. BUG: unable to handle kernel paging request at ffffffffffffffff IP: drop_buffers+0x38/0xe0 PGD 5d00c067 PUD 5d00e067 PMD 0 CPU: 3 PID: 1648 Comm: fsstress Tainted: G D OE 4.10.0+ #5 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 task: ffff9151952863c0 task.stack: ffffaaec40db4000 RIP: 0010:drop_buffers+0x38/0xe0 RSP: 0018:ffffaaec40db74c8 EFLAGS: 00010292 Call Trace: ? page_referenced+0x8b/0x170 try_to_free_buffers+0xc5/0xe0 try_to_release_page+0x49/0x50 shrink_page_list+0x8bc/0x9f0 shrink_inactive_list+0x1dd/0x500 ? shrink_active_list+0x2c0/0x430 shrink_node_memcg+0x5eb/0x7c0 shrink_node+0xe1/0x320 do_try_to_free_pages+0xef/0x2e0 try_to_free_pages+0xe9/0x190 __alloc_pages_slowpath+0x390/0xe70 __alloc_pages_nodemask+0x291/0x2b0 alloc_pages_current+0x95/0x140 __page_cache_alloc+0xc4/0xe0 pagecache_get_page+0xab/0x2a0 grab_cache_page_write_begin+0x20/0x40 get_read_data_page+0x2e6/0x4c0 [f2fs] ? f2fs_mark_inode_dirty_sync+0x16/0x30 [f2fs] ? truncate_data_blocks_range+0x238/0x2b0 [f2fs] get_lock_data_page+0x30/0x190 [f2fs] __exchange_data_block+0xaaf/0xf40 [f2fs] f2fs_fallocate+0x418/0xd00 [f2fs] vfs_fallocate+0x157/0x220 SyS_fallocate+0x48/0x80 Signed-off-by: Yunlei He <heyunlei@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> [Chao Yu: use INMEM_INVALIDATE for better tracing] Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-03-17 01:55:52 +00:00
f2fs_put_page(page, 0);
trace_f2fs_commit_inmem_page(page, INMEM_INVALIDATE);
}
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 16:20:41 +00:00
static int __f2fs_commit_inmem_pages(struct inode *inode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct f2fs_inode_info *fi = F2FS_I(inode);
struct inmem_pages *cur, *tmp;
struct f2fs_io_info fio = {
.sbi = sbi,
.ino = inode->i_ino,
.type = DATA,
.op = REQ_OP_WRITE,
.op_flags = REQ_SYNC | REQ_PRIO,
.io_type = FS_DATA_IO,
};
struct list_head revoke_list;
bool submit_bio = false;
int err = 0;
INIT_LIST_HEAD(&revoke_list);
list_for_each_entry_safe(cur, tmp, &fi->inmem_pages, list) {
struct page *page = cur->page;
lock_page(page);
if (page->mapping == inode->i_mapping) {
trace_f2fs_commit_inmem_page(page, INMEM);
f2fs_wait_on_page_writeback(page, DATA, true, true);
set_page_dirty(page);
if (clear_page_dirty_for_io(page)) {
inode_dec_dirty_pages(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 16:20:41 +00:00
f2fs_remove_dirty_inode(inode);
}
retry:
fio.page = page;
fio.old_blkaddr = NULL_ADDR;
fio.encrypted_page = NULL;
fio.need_lock = LOCK_DONE;
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 16:20:41 +00:00
err = f2fs_do_write_data_page(&fio);
if (err) {
if (err == -ENOMEM) {
congestion_wait(BLK_RW_ASYNC, HZ/50);
cond_resched();
goto retry;
}
unlock_page(page);
break;
}
/* record old blkaddr for revoking */
cur->old_addr = fio.old_blkaddr;
submit_bio = true;
}
unlock_page(page);
list_move_tail(&cur->list, &revoke_list);
}
if (submit_bio)
f2fs_submit_merged_write_cond(sbi, inode, NULL, 0, DATA);
if (err) {
/*
* try to revoke all committed pages, but still we could fail
* due to no memory or other reason, if that happened, EAGAIN
* will be returned, which means in such case, transaction is
* already not integrity, caller should use journal to do the
* recovery or rewrite & commit last transaction. For other
* error number, revoking was done by filesystem itself.
*/
f2fs: fix to avoid deadlock of atomic file operations Thread A Thread B - __fput - f2fs_release_file - drop_inmem_pages - mutex_lock(&fi->inmem_lock) - __revoke_inmem_pages - lock_page(page) - open - f2fs_setattr - truncate_setsize - truncate_inode_pages_range - lock_page(page) - truncate_cleanup_page - f2fs_invalidate_page - drop_inmem_page - mutex_lock(&fi->inmem_lock); We may encounter above ABBA deadlock as reported by Kyungtae Kim: I'm reporting a bug in linux-4.17.19: "INFO: task hung in drop_inmem_page" (no reproducer) I think this might be somehow related to the following: https://groups.google.com/forum/#!searchin/syzkaller-bugs/INFO$3A$20task$20hung$20in$20%7Csort:date/syzkaller-bugs/c6soBTrdaIo/AjAzPeIzCgAJ ========================================= INFO: task syz-executor7:10822 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D27024 10822 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 schedule_preempt_disabled+0x18/0x30 kernel/sched/core.c:3617 __mutex_lock_common kernel/locking/mutex.c:833 [inline] __mutex_lock+0x5bd/0x1410 kernel/locking/mutex.c:893 mutex_lock_nested+0x1b/0x20 kernel/locking/mutex.c:908 drop_inmem_page+0xcb/0x810 fs/f2fs/segment.c:327 f2fs_invalidate_page+0x337/0x5e0 fs/f2fs/data.c:2401 do_invalidatepage mm/truncate.c:165 [inline] truncate_cleanup_page+0x261/0x330 mm/truncate.c:187 truncate_inode_pages_range+0x552/0x1610 mm/truncate.c:367 truncate_inode_pages mm/truncate.c:478 [inline] truncate_pagecache+0x6d/0x90 mm/truncate.c:801 truncate_setsize+0x81/0xa0 mm/truncate.c:826 f2fs_setattr+0x44f/0x1270 fs/f2fs/file.c:781 notify_change+0xa62/0xe80 fs/attr.c:313 do_truncate+0x12e/0x1e0 fs/open.c:63 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e459c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e45a6cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071bea0 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e45a700 INFO: task syz-executor7:10858 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D28880 10858 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 __rwsem_down_write_failed_common kernel/locking/rwsem-xadd.c:565 [inline] rwsem_down_write_failed+0x5e6/0xc90 kernel/locking/rwsem-xadd.c:594 call_rwsem_down_write_failed+0x17/0x30 arch/x86/lib/rwsem.S:117 __down_write arch/x86/include/asm/rwsem.h:142 [inline] down_write+0x58/0xa0 kernel/locking/rwsem.c:72 inode_lock include/linux/fs.h:713 [inline] do_truncate+0x120/0x1e0 fs/open.c:61 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e3b4c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e3b56cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071c238 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e3b5700 INFO: task syz-executor5:10829 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor5 D28760 10829 6308 0x80000002 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 io_schedule+0x21/0x80 kernel/sched/core.c:5179 wait_on_page_bit_common mm/filemap.c:1100 [inline] __lock_page+0x2b5/0x390 mm/filemap.c:1273 lock_page include/linux/pagemap.h:483 [inline] __revoke_inmem_pages+0xb35/0x11c0 fs/f2fs/segment.c:231 drop_inmem_pages+0xa3/0x3e0 fs/f2fs/segment.c:306 f2fs_release_file+0x2c7/0x330 fs/f2fs/file.c:1556 __fput+0x2c7/0x780 fs/file_table.c:209 ____fput+0x1a/0x20 fs/file_table.c:243 task_work_run+0x151/0x1d0 kernel/task_work.c:113 exit_task_work include/linux/task_work.h:22 [inline] do_exit+0x8ba/0x30a0 kernel/exit.c:865 do_group_exit+0x13b/0x3a0 kernel/exit.c:968 get_signal+0x6bb/0x1650 kernel/signal.c:2482 do_signal+0x84/0x1b70 arch/x86/kernel/signal.c:810 exit_to_usermode_loop+0x155/0x190 arch/x86/entry/common.c:162 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x445/0x4e0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f1c68e74ce8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: fffffffffffffe00 RBX: 000000000071bf80 RCX: 00000000004497b9 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000071bf80 RBP: 000000000071bf80 R08: 0000000000000000 R09: 000000000071bf58 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 0000000000000000 R14: 00007f1c68e759c0 R15: 00007f1c68e75700 This patch tries to use trylock_page to mitigate such deadlock condition for fix. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-25 09:11:03 +00:00
err = __revoke_inmem_pages(inode, &revoke_list,
false, true, false);
/* drop all uncommitted pages */
f2fs: fix to avoid deadlock of atomic file operations Thread A Thread B - __fput - f2fs_release_file - drop_inmem_pages - mutex_lock(&fi->inmem_lock) - __revoke_inmem_pages - lock_page(page) - open - f2fs_setattr - truncate_setsize - truncate_inode_pages_range - lock_page(page) - truncate_cleanup_page - f2fs_invalidate_page - drop_inmem_page - mutex_lock(&fi->inmem_lock); We may encounter above ABBA deadlock as reported by Kyungtae Kim: I'm reporting a bug in linux-4.17.19: "INFO: task hung in drop_inmem_page" (no reproducer) I think this might be somehow related to the following: https://groups.google.com/forum/#!searchin/syzkaller-bugs/INFO$3A$20task$20hung$20in$20%7Csort:date/syzkaller-bugs/c6soBTrdaIo/AjAzPeIzCgAJ ========================================= INFO: task syz-executor7:10822 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D27024 10822 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 schedule_preempt_disabled+0x18/0x30 kernel/sched/core.c:3617 __mutex_lock_common kernel/locking/mutex.c:833 [inline] __mutex_lock+0x5bd/0x1410 kernel/locking/mutex.c:893 mutex_lock_nested+0x1b/0x20 kernel/locking/mutex.c:908 drop_inmem_page+0xcb/0x810 fs/f2fs/segment.c:327 f2fs_invalidate_page+0x337/0x5e0 fs/f2fs/data.c:2401 do_invalidatepage mm/truncate.c:165 [inline] truncate_cleanup_page+0x261/0x330 mm/truncate.c:187 truncate_inode_pages_range+0x552/0x1610 mm/truncate.c:367 truncate_inode_pages mm/truncate.c:478 [inline] truncate_pagecache+0x6d/0x90 mm/truncate.c:801 truncate_setsize+0x81/0xa0 mm/truncate.c:826 f2fs_setattr+0x44f/0x1270 fs/f2fs/file.c:781 notify_change+0xa62/0xe80 fs/attr.c:313 do_truncate+0x12e/0x1e0 fs/open.c:63 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e459c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e45a6cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071bea0 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e45a700 INFO: task syz-executor7:10858 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D28880 10858 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 __rwsem_down_write_failed_common kernel/locking/rwsem-xadd.c:565 [inline] rwsem_down_write_failed+0x5e6/0xc90 kernel/locking/rwsem-xadd.c:594 call_rwsem_down_write_failed+0x17/0x30 arch/x86/lib/rwsem.S:117 __down_write arch/x86/include/asm/rwsem.h:142 [inline] down_write+0x58/0xa0 kernel/locking/rwsem.c:72 inode_lock include/linux/fs.h:713 [inline] do_truncate+0x120/0x1e0 fs/open.c:61 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e3b4c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e3b56cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071c238 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e3b5700 INFO: task syz-executor5:10829 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor5 D28760 10829 6308 0x80000002 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 io_schedule+0x21/0x80 kernel/sched/core.c:5179 wait_on_page_bit_common mm/filemap.c:1100 [inline] __lock_page+0x2b5/0x390 mm/filemap.c:1273 lock_page include/linux/pagemap.h:483 [inline] __revoke_inmem_pages+0xb35/0x11c0 fs/f2fs/segment.c:231 drop_inmem_pages+0xa3/0x3e0 fs/f2fs/segment.c:306 f2fs_release_file+0x2c7/0x330 fs/f2fs/file.c:1556 __fput+0x2c7/0x780 fs/file_table.c:209 ____fput+0x1a/0x20 fs/file_table.c:243 task_work_run+0x151/0x1d0 kernel/task_work.c:113 exit_task_work include/linux/task_work.h:22 [inline] do_exit+0x8ba/0x30a0 kernel/exit.c:865 do_group_exit+0x13b/0x3a0 kernel/exit.c:968 get_signal+0x6bb/0x1650 kernel/signal.c:2482 do_signal+0x84/0x1b70 arch/x86/kernel/signal.c:810 exit_to_usermode_loop+0x155/0x190 arch/x86/entry/common.c:162 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x445/0x4e0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f1c68e74ce8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: fffffffffffffe00 RBX: 000000000071bf80 RCX: 00000000004497b9 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000071bf80 RBP: 000000000071bf80 R08: 0000000000000000 R09: 000000000071bf58 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 0000000000000000 R14: 00007f1c68e759c0 R15: 00007f1c68e75700 This patch tries to use trylock_page to mitigate such deadlock condition for fix. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-25 09:11:03 +00:00
__revoke_inmem_pages(inode, &fi->inmem_pages,
true, false, false);
} else {
f2fs: fix to avoid deadlock of atomic file operations Thread A Thread B - __fput - f2fs_release_file - drop_inmem_pages - mutex_lock(&fi->inmem_lock) - __revoke_inmem_pages - lock_page(page) - open - f2fs_setattr - truncate_setsize - truncate_inode_pages_range - lock_page(page) - truncate_cleanup_page - f2fs_invalidate_page - drop_inmem_page - mutex_lock(&fi->inmem_lock); We may encounter above ABBA deadlock as reported by Kyungtae Kim: I'm reporting a bug in linux-4.17.19: "INFO: task hung in drop_inmem_page" (no reproducer) I think this might be somehow related to the following: https://groups.google.com/forum/#!searchin/syzkaller-bugs/INFO$3A$20task$20hung$20in$20%7Csort:date/syzkaller-bugs/c6soBTrdaIo/AjAzPeIzCgAJ ========================================= INFO: task syz-executor7:10822 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D27024 10822 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 schedule_preempt_disabled+0x18/0x30 kernel/sched/core.c:3617 __mutex_lock_common kernel/locking/mutex.c:833 [inline] __mutex_lock+0x5bd/0x1410 kernel/locking/mutex.c:893 mutex_lock_nested+0x1b/0x20 kernel/locking/mutex.c:908 drop_inmem_page+0xcb/0x810 fs/f2fs/segment.c:327 f2fs_invalidate_page+0x337/0x5e0 fs/f2fs/data.c:2401 do_invalidatepage mm/truncate.c:165 [inline] truncate_cleanup_page+0x261/0x330 mm/truncate.c:187 truncate_inode_pages_range+0x552/0x1610 mm/truncate.c:367 truncate_inode_pages mm/truncate.c:478 [inline] truncate_pagecache+0x6d/0x90 mm/truncate.c:801 truncate_setsize+0x81/0xa0 mm/truncate.c:826 f2fs_setattr+0x44f/0x1270 fs/f2fs/file.c:781 notify_change+0xa62/0xe80 fs/attr.c:313 do_truncate+0x12e/0x1e0 fs/open.c:63 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e459c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e45a6cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071bea0 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e45a700 INFO: task syz-executor7:10858 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor7 D28880 10858 6346 0x00000004 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 __rwsem_down_write_failed_common kernel/locking/rwsem-xadd.c:565 [inline] rwsem_down_write_failed+0x5e6/0xc90 kernel/locking/rwsem-xadd.c:594 call_rwsem_down_write_failed+0x17/0x30 arch/x86/lib/rwsem.S:117 __down_write arch/x86/include/asm/rwsem.h:142 [inline] down_write+0x58/0xa0 kernel/locking/rwsem.c:72 inode_lock include/linux/fs.h:713 [inline] do_truncate+0x120/0x1e0 fs/open.c:61 do_last fs/namei.c:2955 [inline] path_openat+0x2042/0x29f0 fs/namei.c:3505 do_filp_open+0x1bd/0x2c0 fs/namei.c:3540 do_sys_open+0x35e/0x4e0 fs/open.c:1101 __do_sys_open fs/open.c:1119 [inline] __se_sys_open fs/open.c:1114 [inline] __x64_sys_open+0x89/0xc0 fs/open.c:1114 do_syscall_64+0xc4/0x4e0 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f734e3b4c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000002 RAX: ffffffffffffffda RBX: 00007f734e3b56cc RCX: 00000000004497b9 RDX: 0000000000000104 RSI: 00000000000a8280 RDI: 0000000020000080 RBP: 000000000071c238 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff R13: 0000000000007230 R14: 00000000006f02d0 R15: 00007f734e3b5700 INFO: task syz-executor5:10829 blocked for more than 120 seconds. Not tainted 4.17.19 #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. syz-executor5 D28760 10829 6308 0x80000002 Call Trace: context_switch kernel/sched/core.c:2867 [inline] __schedule+0x721/0x1e60 kernel/sched/core.c:3515 schedule+0x88/0x1c0 kernel/sched/core.c:3559 io_schedule+0x21/0x80 kernel/sched/core.c:5179 wait_on_page_bit_common mm/filemap.c:1100 [inline] __lock_page+0x2b5/0x390 mm/filemap.c:1273 lock_page include/linux/pagemap.h:483 [inline] __revoke_inmem_pages+0xb35/0x11c0 fs/f2fs/segment.c:231 drop_inmem_pages+0xa3/0x3e0 fs/f2fs/segment.c:306 f2fs_release_file+0x2c7/0x330 fs/f2fs/file.c:1556 __fput+0x2c7/0x780 fs/file_table.c:209 ____fput+0x1a/0x20 fs/file_table.c:243 task_work_run+0x151/0x1d0 kernel/task_work.c:113 exit_task_work include/linux/task_work.h:22 [inline] do_exit+0x8ba/0x30a0 kernel/exit.c:865 do_group_exit+0x13b/0x3a0 kernel/exit.c:968 get_signal+0x6bb/0x1650 kernel/signal.c:2482 do_signal+0x84/0x1b70 arch/x86/kernel/signal.c:810 exit_to_usermode_loop+0x155/0x190 arch/x86/entry/common.c:162 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x445/0x4e0 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x4497b9 RSP: 002b:00007f1c68e74ce8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: fffffffffffffe00 RBX: 000000000071bf80 RCX: 00000000004497b9 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000071bf80 RBP: 000000000071bf80 R08: 0000000000000000 R09: 000000000071bf58 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 0000000000000000 R14: 00007f1c68e759c0 R15: 00007f1c68e75700 This patch tries to use trylock_page to mitigate such deadlock condition for fix. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-25 09:11:03 +00:00
__revoke_inmem_pages(inode, &revoke_list,
false, false, false);
}
return err;
}
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 16:20:41 +00:00
int f2fs_commit_inmem_pages(struct inode *inode)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct f2fs_inode_info *fi = F2FS_I(inode);
int err;
f2fs_balance_fs(sbi, true);
down_write(&fi->i_gc_rwsem[WRITE]);
f2fs_lock_op(sbi);
set_inode_flag(inode, FI_ATOMIC_COMMIT);
mutex_lock(&fi->inmem_lock);
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 16:20:41 +00:00
err = __f2fs_commit_inmem_pages(inode);
mutex_unlock(&fi->inmem_lock);
clear_inode_flag(inode, FI_ATOMIC_COMMIT);
f2fs_unlock_op(sbi);
up_write(&fi->i_gc_rwsem[WRITE]);
return err;
}
/*
* This function balances dirty node and dentry pages.
* In addition, it controls garbage collection.
*/
void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need)
{
if (time_to_inject(sbi, FAULT_CHECKPOINT)) {
f2fs_show_injection_info(sbi, FAULT_CHECKPOINT);
f2fs_stop_checkpoint(sbi, false);
}
/* balance_fs_bg is able to be pending */
if (need && excess_cached_nats(sbi))
f2fs_balance_fs_bg(sbi);
if (!f2fs_is_checkpoint_ready(sbi))
return;
/*
* We should do GC or end up with checkpoint, if there are so many dirty
* dir/node pages without enough free segments.
*/
if (has_not_enough_free_secs(sbi, 0, 0)) {
mutex_lock(&sbi->gc_mutex);
f2fs_gc(sbi, false, false, NULL_SEGNO);
}
}
void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi)
{
if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
return;
/* try to shrink extent cache when there is no enough memory */
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 16:20:41 +00:00
if (!f2fs_available_free_memory(sbi, EXTENT_CACHE))
f2fs_shrink_extent_tree(sbi, EXTENT_CACHE_SHRINK_NUMBER);
/* check the # of cached NAT entries */
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 16:20:41 +00:00
if (!f2fs_available_free_memory(sbi, NAT_ENTRIES))
f2fs_try_to_free_nats(sbi, NAT_ENTRY_PER_BLOCK);
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 16:20:41 +00:00
if (!f2fs_available_free_memory(sbi, FREE_NIDS))
f2fs_try_to_free_nids(sbi, MAX_FREE_NIDS);
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 16:20:41 +00:00
f2fs_build_free_nids(sbi, false, false);
if (!is_idle(sbi, REQ_TIME) &&
(!excess_dirty_nats(sbi) && !excess_dirty_nodes(sbi)))
return;
/* checkpoint is the only way to shrink partial cached entries */
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 16:20:41 +00:00
if (!f2fs_available_free_memory(sbi, NAT_ENTRIES) ||
!f2fs_available_free_memory(sbi, INO_ENTRIES) ||
excess_prefree_segs(sbi) ||
excess_dirty_nats(sbi) ||
excess_dirty_nodes(sbi) ||
f2fs_time_over(sbi, CP_TIME)) {
if (test_opt(sbi, DATA_FLUSH)) {
struct blk_plug plug;
f2fs: fix to avoid deadloop if data_flush is on As Hagbard Celine reported: [ 615.697824] INFO: task kworker/u16:5:344 blocked for more than 120 seconds. [ 615.697825] Not tainted 5.0.15-gentoo-f2fslog #4 [ 615.697826] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 615.697827] kworker/u16:5 D 0 344 2 0x80000000 [ 615.697831] Workqueue: writeback wb_workfn (flush-259:0) [ 615.697832] Call Trace: [ 615.697836] ? __schedule+0x2c5/0x8b0 [ 615.697839] schedule+0x32/0x80 [ 615.697841] schedule_preempt_disabled+0x14/0x20 [ 615.697842] __mutex_lock.isra.8+0x2ba/0x4d0 [ 615.697845] ? log_store+0xf5/0x260 [ 615.697848] f2fs_write_data_pages+0x133/0x320 [ 615.697851] ? trace_hardirqs_on+0x2c/0xe0 [ 615.697854] do_writepages+0x41/0xd0 [ 615.697857] __filemap_fdatawrite_range+0x81/0xb0 [ 615.697859] f2fs_sync_dirty_inodes+0x1dd/0x200 [ 615.697861] f2fs_balance_fs_bg+0x2a7/0x2c0 [ 615.697863] ? up_read+0x5/0x20 [ 615.697865] ? f2fs_do_write_data_page+0x2cb/0x940 [ 615.697867] f2fs_balance_fs+0xe5/0x2c0 [ 615.697869] __write_data_page+0x1c8/0x6e0 [ 615.697873] f2fs_write_cache_pages+0x1e0/0x450 [ 615.697878] f2fs_write_data_pages+0x14b/0x320 [ 615.697880] ? trace_hardirqs_on+0x2c/0xe0 [ 615.697883] do_writepages+0x41/0xd0 [ 615.697885] __filemap_fdatawrite_range+0x81/0xb0 [ 615.697887] f2fs_sync_dirty_inodes+0x1dd/0x200 [ 615.697889] f2fs_balance_fs_bg+0x2a7/0x2c0 [ 615.697891] f2fs_write_node_pages+0x51/0x220 [ 615.697894] do_writepages+0x41/0xd0 [ 615.697897] __writeback_single_inode+0x3d/0x3d0 [ 615.697899] writeback_sb_inodes+0x1e8/0x410 [ 615.697902] __writeback_inodes_wb+0x5d/0xb0 [ 615.697904] wb_writeback+0x28f/0x340 [ 615.697906] ? cpumask_next+0x16/0x20 [ 615.697908] wb_workfn+0x33e/0x420 [ 615.697911] process_one_work+0x1a1/0x3d0 [ 615.697913] worker_thread+0x30/0x380 [ 615.697915] ? process_one_work+0x3d0/0x3d0 [ 615.697916] kthread+0x116/0x130 [ 615.697918] ? kthread_create_worker_on_cpu+0x70/0x70 [ 615.697921] ret_from_fork+0x3a/0x50 There is still deadloop in below condition: d A - do_writepages - f2fs_write_node_pages - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - f2fs_write_cache_pages - mutex_lock(&sbi->writepages) -- lock once - __write_data_page - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - f2fs_write_data_pages - mutex_lock(&sbi->writepages) -- lock again Thread A Thread B - do_writepages - f2fs_write_node_pages - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - .cp_task = current - f2fs_sync_dirty_inodes - .cp_task = current - filemap_fdatawrite - .cp_task = NULL - filemap_fdatawrite - f2fs_write_cache_pages - enter f2fs_balance_fs_bg since .cp_task is NULL - .cp_task = NULL Change as below to avoid this: - add condition to avoid holding .writepages mutex lock in path of data flush - introduce mutex lock sbi.flush_lock to exclude concurrent data flush in background. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-05-20 09:36:59 +00:00
mutex_lock(&sbi->flush_lock);
blk_start_plug(&plug);
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 16:20:41 +00:00
f2fs_sync_dirty_inodes(sbi, FILE_INODE);
blk_finish_plug(&plug);
f2fs: fix to avoid deadloop if data_flush is on As Hagbard Celine reported: [ 615.697824] INFO: task kworker/u16:5:344 blocked for more than 120 seconds. [ 615.697825] Not tainted 5.0.15-gentoo-f2fslog #4 [ 615.697826] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 615.697827] kworker/u16:5 D 0 344 2 0x80000000 [ 615.697831] Workqueue: writeback wb_workfn (flush-259:0) [ 615.697832] Call Trace: [ 615.697836] ? __schedule+0x2c5/0x8b0 [ 615.697839] schedule+0x32/0x80 [ 615.697841] schedule_preempt_disabled+0x14/0x20 [ 615.697842] __mutex_lock.isra.8+0x2ba/0x4d0 [ 615.697845] ? log_store+0xf5/0x260 [ 615.697848] f2fs_write_data_pages+0x133/0x320 [ 615.697851] ? trace_hardirqs_on+0x2c/0xe0 [ 615.697854] do_writepages+0x41/0xd0 [ 615.697857] __filemap_fdatawrite_range+0x81/0xb0 [ 615.697859] f2fs_sync_dirty_inodes+0x1dd/0x200 [ 615.697861] f2fs_balance_fs_bg+0x2a7/0x2c0 [ 615.697863] ? up_read+0x5/0x20 [ 615.697865] ? f2fs_do_write_data_page+0x2cb/0x940 [ 615.697867] f2fs_balance_fs+0xe5/0x2c0 [ 615.697869] __write_data_page+0x1c8/0x6e0 [ 615.697873] f2fs_write_cache_pages+0x1e0/0x450 [ 615.697878] f2fs_write_data_pages+0x14b/0x320 [ 615.697880] ? trace_hardirqs_on+0x2c/0xe0 [ 615.697883] do_writepages+0x41/0xd0 [ 615.697885] __filemap_fdatawrite_range+0x81/0xb0 [ 615.697887] f2fs_sync_dirty_inodes+0x1dd/0x200 [ 615.697889] f2fs_balance_fs_bg+0x2a7/0x2c0 [ 615.697891] f2fs_write_node_pages+0x51/0x220 [ 615.697894] do_writepages+0x41/0xd0 [ 615.697897] __writeback_single_inode+0x3d/0x3d0 [ 615.697899] writeback_sb_inodes+0x1e8/0x410 [ 615.697902] __writeback_inodes_wb+0x5d/0xb0 [ 615.697904] wb_writeback+0x28f/0x340 [ 615.697906] ? cpumask_next+0x16/0x20 [ 615.697908] wb_workfn+0x33e/0x420 [ 615.697911] process_one_work+0x1a1/0x3d0 [ 615.697913] worker_thread+0x30/0x380 [ 615.697915] ? process_one_work+0x3d0/0x3d0 [ 615.697916] kthread+0x116/0x130 [ 615.697918] ? kthread_create_worker_on_cpu+0x70/0x70 [ 615.697921] ret_from_fork+0x3a/0x50 There is still deadloop in below condition: d A - do_writepages - f2fs_write_node_pages - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - f2fs_write_cache_pages - mutex_lock(&sbi->writepages) -- lock once - __write_data_page - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - f2fs_write_data_pages - mutex_lock(&sbi->writepages) -- lock again Thread A Thread B - do_writepages - f2fs_write_node_pages - f2fs_balance_fs_bg - f2fs_sync_dirty_inodes - .cp_task = current - f2fs_sync_dirty_inodes - .cp_task = current - filemap_fdatawrite - .cp_task = NULL - filemap_fdatawrite - f2fs_write_cache_pages - enter f2fs_balance_fs_bg since .cp_task is NULL - .cp_task = NULL Change as below to avoid this: - add condition to avoid holding .writepages mutex lock in path of data flush - introduce mutex lock sbi.flush_lock to exclude concurrent data flush in background. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-05-20 09:36:59 +00:00
mutex_unlock(&sbi->flush_lock);
}
f2fs_sync_fs(sbi->sb, true);
stat_inc_bg_cp_count(sbi->stat_info);
}
}
static int __submit_flush_wait(struct f2fs_sb_info *sbi,
struct block_device *bdev)
{
struct bio *bio;
int ret;
bio = f2fs_bio_alloc(sbi, 0, false);
if (!bio)
return -ENOMEM;
bio->bi_opf = REQ_OP_WRITE | REQ_SYNC | REQ_PREFLUSH;
bio_set_dev(bio, bdev);
ret = submit_bio_wait(bio);
bio_put(bio);
trace_f2fs_issue_flush(bdev, test_opt(sbi, NOBARRIER),
test_opt(sbi, FLUSH_MERGE), ret);
return ret;
}
static int submit_flush_wait(struct f2fs_sb_info *sbi, nid_t ino)
{
int ret = 0;
int i;
if (!f2fs_is_multi_device(sbi))
return __submit_flush_wait(sbi, sbi->sb->s_bdev);
for (i = 0; i < sbi->s_ndevs; i++) {
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 16:20:41 +00:00
if (!f2fs_is_dirty_device(sbi, ino, i, FLUSH_INO))
continue;
ret = __submit_flush_wait(sbi, FDEV(i).bdev);
if (ret)
break;
}
return ret;
}
static int issue_flush_thread(void *data)
{
struct f2fs_sb_info *sbi = data;
struct flush_cmd_control *fcc = SM_I(sbi)->fcc_info;
wait_queue_head_t *q = &fcc->flush_wait_queue;
repeat:
if (kthread_should_stop())
return 0;
f2fs: make background threads of f2fs being aware of freezing When ->freeze_fs is called from lvm for doing snapshot, it needs to make sure there will be no more changes in filesystem's data, however, previously, background threads like GC thread wasn't aware of freezing, so in environment with active background threads, data of snapshot becomes unstable. This patch fixes this issue by adding sb_{start,end}_intwrite in below background threads: - GC thread - flush thread - discard thread Note that, don't use sb_start_intwrite() in gc_thread_func() due to: generic/241 reports below bug: ====================================================== WARNING: possible circular locking dependency detected 4.13.0-rc1+ #32 Tainted: G O ------------------------------------------------------ f2fs_gc-250:0/22186 is trying to acquire lock: (&sbi->gc_mutex){+.+...}, at: [<f8fa7f0b>] f2fs_sync_fs+0x7b/0x1b0 [f2fs] but task is already holding lock: (sb_internal#2){++++.-}, at: [<f8fb5609>] gc_thread_func+0x159/0x4a0 [f2fs] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (sb_internal#2){++++.-}: __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __sb_start_write+0x11d/0x1f0 f2fs_evict_inode+0x2d6/0x4e0 [f2fs] evict+0xa8/0x170 iput+0x1fb/0x2c0 f2fs_sync_inode_meta+0x3f/0xf0 [f2fs] write_checkpoint+0x1b1/0x750 [f2fs] f2fs_sync_fs+0x85/0x1b0 [f2fs] f2fs_do_sync_file.isra.24+0x137/0xa30 [f2fs] f2fs_sync_file+0x34/0x40 [f2fs] vfs_fsync_range+0x4a/0xa0 do_fsync+0x3c/0x60 SyS_fdatasync+0x15/0x20 do_fast_syscall_32+0xa1/0x1b0 entry_SYSENTER_32+0x4c/0x7b -> #1 (&sbi->cp_mutex){+.+...}: __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __mutex_lock+0x4f/0x830 mutex_lock_nested+0x25/0x30 write_checkpoint+0x2f/0x750 [f2fs] f2fs_sync_fs+0x85/0x1b0 [f2fs] sync_filesystem+0x67/0x80 generic_shutdown_super+0x27/0x100 kill_block_super+0x22/0x50 kill_f2fs_super+0x3a/0x40 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x69/0x80 exit_to_usermode_loop+0x57/0x92 do_fast_syscall_32+0x18c/0x1b0 entry_SYSENTER_32+0x4c/0x7b -> #0 (&sbi->gc_mutex){+.+...}: validate_chain.isra.36+0xc50/0xdb0 __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __mutex_lock+0x4f/0x830 mutex_lock_nested+0x25/0x30 f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_balance_fs_bg+0xb9/0x200 [f2fs] gc_thread_func+0x302/0x4a0 [f2fs] kthread+0xe9/0x120 ret_from_fork+0x19/0x24 other info that might help us debug this: Chain exists of: &sbi->gc_mutex --> &sbi->cp_mutex --> sb_internal#2 Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(sb_internal#2); lock(&sbi->cp_mutex); lock(sb_internal#2); lock(&sbi->gc_mutex); *** DEADLOCK *** 1 lock held by f2fs_gc-250:0/22186: #0: (sb_internal#2){++++.-}, at: [<f8fb5609>] gc_thread_func+0x159/0x4a0 [f2fs] stack backtrace: CPU: 2 PID: 22186 Comm: f2fs_gc-250:0 Tainted: G O 4.13.0-rc1+ #32 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 Call Trace: dump_stack+0x5f/0x92 print_circular_bug+0x1b3/0x1bd validate_chain.isra.36+0xc50/0xdb0 ? __this_cpu_preempt_check+0xf/0x20 __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] __mutex_lock+0x4f/0x830 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] mutex_lock_nested+0x25/0x30 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_balance_fs_bg+0xb9/0x200 [f2fs] gc_thread_func+0x302/0x4a0 [f2fs] ? preempt_schedule_common+0x2f/0x4d ? f2fs_gc+0x540/0x540 [f2fs] kthread+0xe9/0x120 ? f2fs_gc+0x540/0x540 [f2fs] ? kthread_create_on_node+0x30/0x30 ret_from_fork+0x19/0x24 The deadlock occurs in below condition: GC Thread Thread B - sb_start_intwrite - f2fs_sync_file - f2fs_sync_fs - mutex_lock(&sbi->gc_mutex) - write_checkpoint - block_operations - f2fs_sync_inode_meta - iput - sb_start_intwrite - mutex_lock(&sbi->gc_mutex) Fix this by altering sb_start_intwrite to sb_start_write_trylock. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-22 00:52:23 +00:00
sb_start_intwrite(sbi->sb);
if (!llist_empty(&fcc->issue_list)) {
struct flush_cmd *cmd, *next;
int ret;
fcc->dispatch_list = llist_del_all(&fcc->issue_list);
fcc->dispatch_list = llist_reverse_order(fcc->dispatch_list);
cmd = llist_entry(fcc->dispatch_list, struct flush_cmd, llnode);
ret = submit_flush_wait(sbi, cmd->ino);
atomic_inc(&fcc->issued_flush);
llist_for_each_entry_safe(cmd, next,
fcc->dispatch_list, llnode) {
cmd->ret = ret;
complete(&cmd->wait);
}
fcc->dispatch_list = NULL;
}
f2fs: make background threads of f2fs being aware of freezing When ->freeze_fs is called from lvm for doing snapshot, it needs to make sure there will be no more changes in filesystem's data, however, previously, background threads like GC thread wasn't aware of freezing, so in environment with active background threads, data of snapshot becomes unstable. This patch fixes this issue by adding sb_{start,end}_intwrite in below background threads: - GC thread - flush thread - discard thread Note that, don't use sb_start_intwrite() in gc_thread_func() due to: generic/241 reports below bug: ====================================================== WARNING: possible circular locking dependency detected 4.13.0-rc1+ #32 Tainted: G O ------------------------------------------------------ f2fs_gc-250:0/22186 is trying to acquire lock: (&sbi->gc_mutex){+.+...}, at: [<f8fa7f0b>] f2fs_sync_fs+0x7b/0x1b0 [f2fs] but task is already holding lock: (sb_internal#2){++++.-}, at: [<f8fb5609>] gc_thread_func+0x159/0x4a0 [f2fs] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (sb_internal#2){++++.-}: __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __sb_start_write+0x11d/0x1f0 f2fs_evict_inode+0x2d6/0x4e0 [f2fs] evict+0xa8/0x170 iput+0x1fb/0x2c0 f2fs_sync_inode_meta+0x3f/0xf0 [f2fs] write_checkpoint+0x1b1/0x750 [f2fs] f2fs_sync_fs+0x85/0x1b0 [f2fs] f2fs_do_sync_file.isra.24+0x137/0xa30 [f2fs] f2fs_sync_file+0x34/0x40 [f2fs] vfs_fsync_range+0x4a/0xa0 do_fsync+0x3c/0x60 SyS_fdatasync+0x15/0x20 do_fast_syscall_32+0xa1/0x1b0 entry_SYSENTER_32+0x4c/0x7b -> #1 (&sbi->cp_mutex){+.+...}: __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __mutex_lock+0x4f/0x830 mutex_lock_nested+0x25/0x30 write_checkpoint+0x2f/0x750 [f2fs] f2fs_sync_fs+0x85/0x1b0 [f2fs] sync_filesystem+0x67/0x80 generic_shutdown_super+0x27/0x100 kill_block_super+0x22/0x50 kill_f2fs_super+0x3a/0x40 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x69/0x80 exit_to_usermode_loop+0x57/0x92 do_fast_syscall_32+0x18c/0x1b0 entry_SYSENTER_32+0x4c/0x7b -> #0 (&sbi->gc_mutex){+.+...}: validate_chain.isra.36+0xc50/0xdb0 __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __mutex_lock+0x4f/0x830 mutex_lock_nested+0x25/0x30 f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_balance_fs_bg+0xb9/0x200 [f2fs] gc_thread_func+0x302/0x4a0 [f2fs] kthread+0xe9/0x120 ret_from_fork+0x19/0x24 other info that might help us debug this: Chain exists of: &sbi->gc_mutex --> &sbi->cp_mutex --> sb_internal#2 Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(sb_internal#2); lock(&sbi->cp_mutex); lock(sb_internal#2); lock(&sbi->gc_mutex); *** DEADLOCK *** 1 lock held by f2fs_gc-250:0/22186: #0: (sb_internal#2){++++.-}, at: [<f8fb5609>] gc_thread_func+0x159/0x4a0 [f2fs] stack backtrace: CPU: 2 PID: 22186 Comm: f2fs_gc-250:0 Tainted: G O 4.13.0-rc1+ #32 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 Call Trace: dump_stack+0x5f/0x92 print_circular_bug+0x1b3/0x1bd validate_chain.isra.36+0xc50/0xdb0 ? __this_cpu_preempt_check+0xf/0x20 __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] __mutex_lock+0x4f/0x830 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] mutex_lock_nested+0x25/0x30 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_balance_fs_bg+0xb9/0x200 [f2fs] gc_thread_func+0x302/0x4a0 [f2fs] ? preempt_schedule_common+0x2f/0x4d ? f2fs_gc+0x540/0x540 [f2fs] kthread+0xe9/0x120 ? f2fs_gc+0x540/0x540 [f2fs] ? kthread_create_on_node+0x30/0x30 ret_from_fork+0x19/0x24 The deadlock occurs in below condition: GC Thread Thread B - sb_start_intwrite - f2fs_sync_file - f2fs_sync_fs - mutex_lock(&sbi->gc_mutex) - write_checkpoint - block_operations - f2fs_sync_inode_meta - iput - sb_start_intwrite - mutex_lock(&sbi->gc_mutex) Fix this by altering sb_start_intwrite to sb_start_write_trylock. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-22 00:52:23 +00:00
sb_end_intwrite(sbi->sb);
wait_event_interruptible(*q,
kthread_should_stop() || !llist_empty(&fcc->issue_list));
goto repeat;
}
int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino)
{
struct flush_cmd_control *fcc = SM_I(sbi)->fcc_info;
struct flush_cmd cmd;
int ret;
if (test_opt(sbi, NOBARRIER))
return 0;
if (!test_opt(sbi, FLUSH_MERGE)) {
atomic_inc(&fcc->queued_flush);
ret = submit_flush_wait(sbi, ino);
atomic_dec(&fcc->queued_flush);
atomic_inc(&fcc->issued_flush);
return ret;
}
if (atomic_inc_return(&fcc->queued_flush) == 1 ||
f2fs_is_multi_device(sbi)) {
ret = submit_flush_wait(sbi, ino);
atomic_dec(&fcc->queued_flush);
atomic_inc(&fcc->issued_flush);
return ret;
}
cmd.ino = ino;
init_completion(&cmd.wait);
llist_add(&cmd.llnode, &fcc->issue_list);
/* update issue_list before we wake up issue_flush thread */
smp_mb();
if (waitqueue_active(&fcc->flush_wait_queue))
wake_up(&fcc->flush_wait_queue);
if (fcc->f2fs_issue_flush) {
wait_for_completion(&cmd.wait);
atomic_dec(&fcc->queued_flush);
} else {
struct llist_node *list;
list = llist_del_all(&fcc->issue_list);
if (!list) {
wait_for_completion(&cmd.wait);
atomic_dec(&fcc->queued_flush);
} else {
struct flush_cmd *tmp, *next;
ret = submit_flush_wait(sbi, ino);
llist_for_each_entry_safe(tmp, next, list, llnode) {
if (tmp == &cmd) {
cmd.ret = ret;
atomic_dec(&fcc->queued_flush);
continue;
}
tmp->ret = ret;
complete(&tmp->wait);
}
}
}
return cmd.ret;
}
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 16:20:41 +00:00
int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi)
{
dev_t dev = sbi->sb->s_bdev->bd_dev;
struct flush_cmd_control *fcc;
int err = 0;
if (SM_I(sbi)->fcc_info) {
fcc = SM_I(sbi)->fcc_info;
if (fcc->f2fs_issue_flush)
return err;
goto init_thread;
}
fcc = f2fs_kzalloc(sbi, sizeof(struct flush_cmd_control), GFP_KERNEL);
if (!fcc)
return -ENOMEM;
atomic_set(&fcc->issued_flush, 0);
atomic_set(&fcc->queued_flush, 0);
init_waitqueue_head(&fcc->flush_wait_queue);
init_llist_head(&fcc->issue_list);
SM_I(sbi)->fcc_info = fcc;
if (!test_opt(sbi, FLUSH_MERGE))
return err;
init_thread:
fcc->f2fs_issue_flush = kthread_run(issue_flush_thread, sbi,
"f2fs_flush-%u:%u", MAJOR(dev), MINOR(dev));
if (IS_ERR(fcc->f2fs_issue_flush)) {
err = PTR_ERR(fcc->f2fs_issue_flush);
kvfree(fcc);
SM_I(sbi)->fcc_info = NULL;
return err;
}
return err;
}
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 16:20:41 +00:00
void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free)
{
struct flush_cmd_control *fcc = SM_I(sbi)->fcc_info;
if (fcc && fcc->f2fs_issue_flush) {
struct task_struct *flush_thread = fcc->f2fs_issue_flush;
fcc->f2fs_issue_flush = NULL;
kthread_stop(flush_thread);
}
if (free) {
kvfree(fcc);
SM_I(sbi)->fcc_info = NULL;
}
}
int f2fs_flush_device_cache(struct f2fs_sb_info *sbi)
{
int ret = 0, i;
if (!f2fs_is_multi_device(sbi))
return 0;
for (i = 1; i < sbi->s_ndevs; i++) {
if (!f2fs_test_bit(i, (char *)&sbi->dirty_device))
continue;
ret = __submit_flush_wait(sbi, FDEV(i).bdev);
if (ret)
break;
spin_lock(&sbi->dev_lock);
f2fs_clear_bit(i, (char *)&sbi->dirty_device);
spin_unlock(&sbi->dev_lock);
}
return ret;
}
static void __locate_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno,
enum dirty_type dirty_type)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
/* need not be added */
if (IS_CURSEG(sbi, segno))
return;
if (!test_and_set_bit(segno, dirty_i->dirty_segmap[dirty_type]))
dirty_i->nr_dirty[dirty_type]++;
if (dirty_type == DIRTY) {
struct seg_entry *sentry = get_seg_entry(sbi, segno);
enum dirty_type t = sentry->type;
if (unlikely(t >= DIRTY)) {
f2fs_bug_on(sbi, 1);
return;
}
if (!test_and_set_bit(segno, dirty_i->dirty_segmap[t]))
dirty_i->nr_dirty[t]++;
}
}
static void __remove_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno,
enum dirty_type dirty_type)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
if (test_and_clear_bit(segno, dirty_i->dirty_segmap[dirty_type]))
dirty_i->nr_dirty[dirty_type]--;
if (dirty_type == DIRTY) {
struct seg_entry *sentry = get_seg_entry(sbi, segno);
enum dirty_type t = sentry->type;
if (test_and_clear_bit(segno, dirty_i->dirty_segmap[t]))
dirty_i->nr_dirty[t]--;
if (get_valid_blocks(sbi, segno, true) == 0) {
clear_bit(GET_SEC_FROM_SEG(sbi, segno),
dirty_i->victim_secmap);
#ifdef CONFIG_F2FS_CHECK_FS
clear_bit(segno, SIT_I(sbi)->invalid_segmap);
#endif
}
}
}
/*
* Should not occur error such as -ENOMEM.
* Adding dirty entry into seglist is not critical operation.
* If a given segment is one of current working segments, it won't be added.
*/
static void locate_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
unsigned short valid_blocks, ckpt_valid_blocks;
if (segno == NULL_SEGNO || IS_CURSEG(sbi, segno))
return;
mutex_lock(&dirty_i->seglist_lock);
valid_blocks = get_valid_blocks(sbi, segno, false);
ckpt_valid_blocks = get_ckpt_valid_blocks(sbi, segno);
if (valid_blocks == 0 && (!is_sbi_flag_set(sbi, SBI_CP_DISABLED) ||
ckpt_valid_blocks == sbi->blocks_per_seg)) {
__locate_dirty_segment(sbi, segno, PRE);
__remove_dirty_segment(sbi, segno, DIRTY);
} else if (valid_blocks < sbi->blocks_per_seg) {
__locate_dirty_segment(sbi, segno, DIRTY);
} else {
/* Recovery routine with SSR needs this */
__remove_dirty_segment(sbi, segno, DIRTY);
}
mutex_unlock(&dirty_i->seglist_lock);
}
/* This moves currently empty dirty blocks to prefree. Must hold seglist_lock */
void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
unsigned int segno;
mutex_lock(&dirty_i->seglist_lock);
for_each_set_bit(segno, dirty_i->dirty_segmap[DIRTY], MAIN_SEGS(sbi)) {
if (get_valid_blocks(sbi, segno, false))
continue;
if (IS_CURSEG(sbi, segno))
continue;
__locate_dirty_segment(sbi, segno, PRE);
__remove_dirty_segment(sbi, segno, DIRTY);
}
mutex_unlock(&dirty_i->seglist_lock);
}
block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi)
{
f2fs: Lower threshold for disable_cp_again The existing threshold for allowable holes at checkpoint=disable time is too high. The OVP space contains reserved segments, which are always in the form of free segments. These must be subtracted from the OVP value. The current threshold is meant to be the maximum value of holes of a single type we can have and still guarantee that we can fill the disk without failing to find space for a block of a given type. If the disk is full, ignoring current reserved, which only helps us, the amount of unused blocks is equal to the OVP area. Of that, there are reserved segments, which must be free segments, and the rest of the ovp area, which can come from either free segments or holes. The maximum possible amount of holes is OVP-reserved. Now, consider the disk when mounting with checkpoint=disable. We must be able to fill all available free space with either data or node blocks. When we start with checkpoint=disable, holes are locked to their current type. Say we have H of one type of hole, and H+X of the other. We can fill H of that space with arbitrary typed blocks via SSR. For the remaining H+X blocks, we may not have any of a given block type left at all. For instance, if we were to fill the disk entirely with blocks of the type with fewer holes, the H+X blocks of the opposite type would not be used. If H+X > OVP-reserved, there would be more holes than could possibly exist, and we would have failed to find a suitable block earlier on, leading to a crash in update_sit_entry. If H+X <= OVP-reserved, then the holes end up effectively masked by the OVP region in this case. Signed-off-by: Daniel Rosenberg <drosen@google.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-05-30 00:49:03 +00:00
int ovp_hole_segs =
(overprovision_segments(sbi) - reserved_segments(sbi));
block_t ovp_holes = ovp_hole_segs << sbi->log_blocks_per_seg;
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
block_t holes[2] = {0, 0}; /* DATA and NODE */
block_t unusable;
struct seg_entry *se;
unsigned int segno;
mutex_lock(&dirty_i->seglist_lock);
for_each_set_bit(segno, dirty_i->dirty_segmap[DIRTY], MAIN_SEGS(sbi)) {
se = get_seg_entry(sbi, segno);
if (IS_NODESEG(se->type))
holes[NODE] += sbi->blocks_per_seg - se->valid_blocks;
else
holes[DATA] += sbi->blocks_per_seg - se->valid_blocks;
}
mutex_unlock(&dirty_i->seglist_lock);
unusable = holes[DATA] > holes[NODE] ? holes[DATA] : holes[NODE];
if (unusable > ovp_holes)
return unusable - ovp_holes;
return 0;
}
int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable)
{
int ovp_hole_segs =
(overprovision_segments(sbi) - reserved_segments(sbi));
if (unusable > F2FS_OPTION(sbi).unusable_cap)
return -EAGAIN;
if (is_sbi_flag_set(sbi, SBI_CP_DISABLED_QUICK) &&
f2fs: Lower threshold for disable_cp_again The existing threshold for allowable holes at checkpoint=disable time is too high. The OVP space contains reserved segments, which are always in the form of free segments. These must be subtracted from the OVP value. The current threshold is meant to be the maximum value of holes of a single type we can have and still guarantee that we can fill the disk without failing to find space for a block of a given type. If the disk is full, ignoring current reserved, which only helps us, the amount of unused blocks is equal to the OVP area. Of that, there are reserved segments, which must be free segments, and the rest of the ovp area, which can come from either free segments or holes. The maximum possible amount of holes is OVP-reserved. Now, consider the disk when mounting with checkpoint=disable. We must be able to fill all available free space with either data or node blocks. When we start with checkpoint=disable, holes are locked to their current type. Say we have H of one type of hole, and H+X of the other. We can fill H of that space with arbitrary typed blocks via SSR. For the remaining H+X blocks, we may not have any of a given block type left at all. For instance, if we were to fill the disk entirely with blocks of the type with fewer holes, the H+X blocks of the opposite type would not be used. If H+X > OVP-reserved, there would be more holes than could possibly exist, and we would have failed to find a suitable block earlier on, leading to a crash in update_sit_entry. If H+X <= OVP-reserved, then the holes end up effectively masked by the OVP region in this case. Signed-off-by: Daniel Rosenberg <drosen@google.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-05-30 00:49:03 +00:00
dirty_segments(sbi) > ovp_hole_segs)
return -EAGAIN;
return 0;
}
/* This is only used by SBI_CP_DISABLED */
static unsigned int get_free_segment(struct f2fs_sb_info *sbi)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
unsigned int segno = 0;
mutex_lock(&dirty_i->seglist_lock);
for_each_set_bit(segno, dirty_i->dirty_segmap[DIRTY], MAIN_SEGS(sbi)) {
if (get_valid_blocks(sbi, segno, false))
continue;
if (get_ckpt_valid_blocks(sbi, segno))
continue;
mutex_unlock(&dirty_i->seglist_lock);
return segno;
}
mutex_unlock(&dirty_i->seglist_lock);
return NULL_SEGNO;
}
static struct discard_cmd *__create_discard_cmd(struct f2fs_sb_info *sbi,
struct block_device *bdev, block_t lstart,
block_t start, block_t len)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct list_head *pend_list;
struct discard_cmd *dc;
f2fs_bug_on(sbi, !len);
pend_list = &dcc->pend_list[plist_idx(len)];
dc = f2fs_kmem_cache_alloc(discard_cmd_slab, GFP_NOFS);
INIT_LIST_HEAD(&dc->list);
dc->bdev = bdev;
dc->lstart = lstart;
dc->start = start;
dc->len = len;
dc->ref = 0;
dc->state = D_PREP;
dc->queued = 0;
dc->error = 0;
init_completion(&dc->wait);
list_add_tail(&dc->list, pend_list);
spin_lock_init(&dc->lock);
dc->bio_ref = 0;
atomic_inc(&dcc->discard_cmd_cnt);
dcc->undiscard_blks += len;
return dc;
}
static struct discard_cmd *__attach_discard_cmd(struct f2fs_sb_info *sbi,
struct block_device *bdev, block_t lstart,
block_t start, block_t len,
struct rb_node *parent, struct rb_node **p,
bool leftmost)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct discard_cmd *dc;
dc = __create_discard_cmd(sbi, bdev, lstart, start, len);
rb_link_node(&dc->rb_node, parent, p);
rb_insert_color_cached(&dc->rb_node, &dcc->root, leftmost);
return dc;
}
static void __detach_discard_cmd(struct discard_cmd_control *dcc,
struct discard_cmd *dc)
{
if (dc->state == D_DONE)
atomic_sub(dc->queued, &dcc->queued_discard);
list_del(&dc->list);
rb_erase_cached(&dc->rb_node, &dcc->root);
dcc->undiscard_blks -= dc->len;
kmem_cache_free(discard_cmd_slab, dc);
atomic_dec(&dcc->discard_cmd_cnt);
}
static void __remove_discard_cmd(struct f2fs_sb_info *sbi,
struct discard_cmd *dc)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
unsigned long flags;
trace_f2fs_remove_discard(dc->bdev, dc->start, dc->len);
spin_lock_irqsave(&dc->lock, flags);
if (dc->bio_ref) {
spin_unlock_irqrestore(&dc->lock, flags);
return;
}
spin_unlock_irqrestore(&dc->lock, flags);
f2fs_bug_on(sbi, dc->ref);
if (dc->error == -EOPNOTSUPP)
dc->error = 0;
if (dc->error)
printk_ratelimited(
"%sF2FS-fs (%s): Issue discard(%u, %u, %u) failed, ret: %d",
KERN_INFO, sbi->sb->s_id,
dc->lstart, dc->start, dc->len, dc->error);
__detach_discard_cmd(dcc, dc);
}
static void f2fs_submit_discard_endio(struct bio *bio)
{
struct discard_cmd *dc = (struct discard_cmd *)bio->bi_private;
unsigned long flags;
dc->error = blk_status_to_errno(bio->bi_status);
spin_lock_irqsave(&dc->lock, flags);
dc->bio_ref--;
if (!dc->bio_ref && dc->state == D_SUBMIT) {
dc->state = D_DONE;
complete_all(&dc->wait);
}
spin_unlock_irqrestore(&dc->lock, flags);
bio_put(bio);
}
static void __check_sit_bitmap(struct f2fs_sb_info *sbi,
block_t start, block_t end)
{
#ifdef CONFIG_F2FS_CHECK_FS
struct seg_entry *sentry;
unsigned int segno;
block_t blk = start;
unsigned long offset, size, max_blocks = sbi->blocks_per_seg;
unsigned long *map;
while (blk < end) {
segno = GET_SEGNO(sbi, blk);
sentry = get_seg_entry(sbi, segno);
offset = GET_BLKOFF_FROM_SEG0(sbi, blk);
if (end < START_BLOCK(sbi, segno + 1))
size = GET_BLKOFF_FROM_SEG0(sbi, end);
else
size = max_blocks;
map = (unsigned long *)(sentry->cur_valid_map);
offset = __find_rev_next_bit(map, size, offset);
f2fs_bug_on(sbi, offset != size);
blk = START_BLOCK(sbi, segno + 1);
}
#endif
}
static void __init_discard_policy(struct f2fs_sb_info *sbi,
struct discard_policy *dpolicy,
int discard_type, unsigned int granularity)
{
/* common policy */
dpolicy->type = discard_type;
dpolicy->sync = true;
dpolicy->ordered = false;
dpolicy->granularity = granularity;
dpolicy->max_requests = DEF_MAX_DISCARD_REQUEST;
dpolicy->io_aware_gran = MAX_PLIST_NUM;
dpolicy->timeout = 0;
if (discard_type == DPOLICY_BG) {
dpolicy->min_interval = DEF_MIN_DISCARD_ISSUE_TIME;
dpolicy->mid_interval = DEF_MID_DISCARD_ISSUE_TIME;
dpolicy->max_interval = DEF_MAX_DISCARD_ISSUE_TIME;
dpolicy->io_aware = true;
dpolicy->sync = false;
dpolicy->ordered = true;
if (utilization(sbi) > DEF_DISCARD_URGENT_UTIL) {
dpolicy->granularity = 1;
dpolicy->max_interval = DEF_MIN_DISCARD_ISSUE_TIME;
}
} else if (discard_type == DPOLICY_FORCE) {
dpolicy->min_interval = DEF_MIN_DISCARD_ISSUE_TIME;
dpolicy->mid_interval = DEF_MID_DISCARD_ISSUE_TIME;
dpolicy->max_interval = DEF_MAX_DISCARD_ISSUE_TIME;
dpolicy->io_aware = false;
} else if (discard_type == DPOLICY_FSTRIM) {
dpolicy->io_aware = false;
} else if (discard_type == DPOLICY_UMOUNT) {
dpolicy->max_requests = UINT_MAX;
dpolicy->io_aware = false;
/* we need to issue all to keep CP_TRIMMED_FLAG */
dpolicy->granularity = 1;
}
}
static void __update_discard_tree_range(struct f2fs_sb_info *sbi,
struct block_device *bdev, block_t lstart,
block_t start, block_t len);
/* this function is copied from blkdev_issue_discard from block/blk-lib.c */
static int __submit_discard_cmd(struct f2fs_sb_info *sbi,
struct discard_policy *dpolicy,
struct discard_cmd *dc,
unsigned int *issued)
{
struct block_device *bdev = dc->bdev;
struct request_queue *q = bdev_get_queue(bdev);
unsigned int max_discard_blocks =
SECTOR_TO_BLOCK(q->limits.max_discard_sectors);
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct list_head *wait_list = (dpolicy->type == DPOLICY_FSTRIM) ?
&(dcc->fstrim_list) : &(dcc->wait_list);
int flag = dpolicy->sync ? REQ_SYNC : 0;
block_t lstart, start, len, total_len;
int err = 0;
if (dc->state != D_PREP)
return 0;
if (is_sbi_flag_set(sbi, SBI_NEED_FSCK))
return 0;
trace_f2fs_issue_discard(bdev, dc->start, dc->len);
lstart = dc->lstart;
start = dc->start;
len = dc->len;
total_len = len;
dc->len = 0;
while (total_len && *issued < dpolicy->max_requests && !err) {
struct bio *bio = NULL;
unsigned long flags;
bool last = true;
if (len > max_discard_blocks) {
len = max_discard_blocks;
last = false;
}
(*issued)++;
if (*issued == dpolicy->max_requests)
last = true;
dc->len += len;
if (time_to_inject(sbi, FAULT_DISCARD)) {
f2fs_show_injection_info(sbi, FAULT_DISCARD);
err = -EIO;
goto submit;
}
err = __blkdev_issue_discard(bdev,
SECTOR_FROM_BLOCK(start),
SECTOR_FROM_BLOCK(len),
GFP_NOFS, 0, &bio);
submit:
if (err) {
spin_lock_irqsave(&dc->lock, flags);
if (dc->state == D_PARTIAL)
dc->state = D_SUBMIT;
spin_unlock_irqrestore(&dc->lock, flags);
break;
}
f2fs_bug_on(sbi, !bio);
/*
* should keep before submission to avoid D_DONE
* right away
*/
spin_lock_irqsave(&dc->lock, flags);
if (last)
dc->state = D_SUBMIT;
else
dc->state = D_PARTIAL;
dc->bio_ref++;
spin_unlock_irqrestore(&dc->lock, flags);
atomic_inc(&dcc->queued_discard);
dc->queued++;
list_move_tail(&dc->list, wait_list);
/* sanity check on discard range */
__check_sit_bitmap(sbi, lstart, lstart + len);
bio->bi_private = dc;
bio->bi_end_io = f2fs_submit_discard_endio;
bio->bi_opf |= flag;
submit_bio(bio);
atomic_inc(&dcc->issued_discard);
f2fs_update_iostat(sbi, FS_DISCARD, 1);
lstart += len;
start += len;
total_len -= len;
len = total_len;
}
if (!err && len)
__update_discard_tree_range(sbi, bdev, lstart, start, len);
return err;
}
static struct discard_cmd *__insert_discard_tree(struct f2fs_sb_info *sbi,
struct block_device *bdev, block_t lstart,
block_t start, block_t len,
struct rb_node **insert_p,
struct rb_node *insert_parent)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct rb_node **p;
struct rb_node *parent = NULL;
struct discard_cmd *dc = NULL;
bool leftmost = true;
if (insert_p && insert_parent) {
parent = insert_parent;
p = insert_p;
goto do_insert;
}
p = f2fs_lookup_rb_tree_for_insert(sbi, &dcc->root, &parent,
lstart, &leftmost);
do_insert:
dc = __attach_discard_cmd(sbi, bdev, lstart, start, len, parent,
p, leftmost);
if (!dc)
return NULL;
return dc;
}
static void __relocate_discard_cmd(struct discard_cmd_control *dcc,
struct discard_cmd *dc)
{
list_move_tail(&dc->list, &dcc->pend_list[plist_idx(dc->len)]);
}
static void __punch_discard_cmd(struct f2fs_sb_info *sbi,
struct discard_cmd *dc, block_t blkaddr)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct discard_info di = dc->di;
bool modified = false;
if (dc->state == D_DONE || dc->len == 1) {
__remove_discard_cmd(sbi, dc);
return;
}
dcc->undiscard_blks -= di.len;
if (blkaddr > di.lstart) {
dc->len = blkaddr - dc->lstart;
dcc->undiscard_blks += dc->len;
__relocate_discard_cmd(dcc, dc);
modified = true;
}
if (blkaddr < di.lstart + di.len - 1) {
if (modified) {
__insert_discard_tree(sbi, dc->bdev, blkaddr + 1,
di.start + blkaddr + 1 - di.lstart,
di.lstart + di.len - 1 - blkaddr,
NULL, NULL);
} else {
dc->lstart++;
dc->len--;
dc->start++;
dcc->undiscard_blks += dc->len;
__relocate_discard_cmd(dcc, dc);
}
}
}
static void __update_discard_tree_range(struct f2fs_sb_info *sbi,
struct block_device *bdev, block_t lstart,
block_t start, block_t len)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct discard_cmd *prev_dc = NULL, *next_dc = NULL;
struct discard_cmd *dc;
struct discard_info di = {0};
struct rb_node **insert_p = NULL, *insert_parent = NULL;
struct request_queue *q = bdev_get_queue(bdev);
unsigned int max_discard_blocks =
SECTOR_TO_BLOCK(q->limits.max_discard_sectors);
block_t end = lstart + len;
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 16:20:41 +00:00
dc = (struct discard_cmd *)f2fs_lookup_rb_tree_ret(&dcc->root,
NULL, lstart,
(struct rb_entry **)&prev_dc,
(struct rb_entry **)&next_dc,
&insert_p, &insert_parent, true, NULL);
if (dc)
prev_dc = dc;
if (!prev_dc) {
di.lstart = lstart;
di.len = next_dc ? next_dc->lstart - lstart : len;
di.len = min(di.len, len);
di.start = start;
}
while (1) {
struct rb_node *node;
bool merged = false;
struct discard_cmd *tdc = NULL;
if (prev_dc) {
di.lstart = prev_dc->lstart + prev_dc->len;
if (di.lstart < lstart)
di.lstart = lstart;
if (di.lstart >= end)
break;
if (!next_dc || next_dc->lstart > end)
di.len = end - di.lstart;
else
di.len = next_dc->lstart - di.lstart;
di.start = start + di.lstart - lstart;
}
if (!di.len)
goto next;
if (prev_dc && prev_dc->state == D_PREP &&
prev_dc->bdev == bdev &&
__is_discard_back_mergeable(&di, &prev_dc->di,
max_discard_blocks)) {
prev_dc->di.len += di.len;
dcc->undiscard_blks += di.len;
__relocate_discard_cmd(dcc, prev_dc);
di = prev_dc->di;
tdc = prev_dc;
merged = true;
}
if (next_dc && next_dc->state == D_PREP &&
next_dc->bdev == bdev &&
__is_discard_front_mergeable(&di, &next_dc->di,
max_discard_blocks)) {
next_dc->di.lstart = di.lstart;
next_dc->di.len += di.len;
next_dc->di.start = di.start;
dcc->undiscard_blks += di.len;
__relocate_discard_cmd(dcc, next_dc);
if (tdc)
__remove_discard_cmd(sbi, tdc);
merged = true;
}
if (!merged) {
__insert_discard_tree(sbi, bdev, di.lstart, di.start,
di.len, NULL, NULL);
}
next:
prev_dc = next_dc;
if (!prev_dc)
break;
node = rb_next(&prev_dc->rb_node);
next_dc = rb_entry_safe(node, struct discard_cmd, rb_node);
}
}
static int __queue_discard_cmd(struct f2fs_sb_info *sbi,
struct block_device *bdev, block_t blkstart, block_t blklen)
{
block_t lblkstart = blkstart;
if (!f2fs_bdev_support_discard(bdev))
return 0;
trace_f2fs_queue_discard(bdev, blkstart, blklen);
if (f2fs_is_multi_device(sbi)) {
int devi = f2fs_target_device_index(sbi, blkstart);
blkstart -= FDEV(devi).start_blk;
}
mutex_lock(&SM_I(sbi)->dcc_info->cmd_lock);
__update_discard_tree_range(sbi, bdev, lblkstart, blkstart, blklen);
mutex_unlock(&SM_I(sbi)->dcc_info->cmd_lock);
return 0;
}
static unsigned int __issue_discard_cmd_orderly(struct f2fs_sb_info *sbi,
struct discard_policy *dpolicy)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct discard_cmd *prev_dc = NULL, *next_dc = NULL;
struct rb_node **insert_p = NULL, *insert_parent = NULL;
struct discard_cmd *dc;
struct blk_plug plug;
unsigned int pos = dcc->next_pos;
unsigned int issued = 0;
bool io_interrupted = false;
mutex_lock(&dcc->cmd_lock);
dc = (struct discard_cmd *)f2fs_lookup_rb_tree_ret(&dcc->root,
NULL, pos,
(struct rb_entry **)&prev_dc,
(struct rb_entry **)&next_dc,
&insert_p, &insert_parent, true, NULL);
if (!dc)
dc = next_dc;
blk_start_plug(&plug);
while (dc) {
struct rb_node *node;
int err = 0;
if (dc->state != D_PREP)
goto next;
if (dpolicy->io_aware && !is_idle(sbi, DISCARD_TIME)) {
io_interrupted = true;
break;
}
dcc->next_pos = dc->lstart + dc->len;
err = __submit_discard_cmd(sbi, dpolicy, dc, &issued);
if (issued >= dpolicy->max_requests)
break;
next:
node = rb_next(&dc->rb_node);
if (err)
__remove_discard_cmd(sbi, dc);
dc = rb_entry_safe(node, struct discard_cmd, rb_node);
}
blk_finish_plug(&plug);
if (!dc)
dcc->next_pos = 0;
mutex_unlock(&dcc->cmd_lock);
if (!issued && io_interrupted)
issued = -1;
return issued;
}
static int __issue_discard_cmd(struct f2fs_sb_info *sbi,
struct discard_policy *dpolicy)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct list_head *pend_list;
struct discard_cmd *dc, *tmp;
struct blk_plug plug;
int i, issued = 0;
bool io_interrupted = false;
if (dpolicy->timeout != 0)
f2fs_update_time(sbi, dpolicy->timeout);
for (i = MAX_PLIST_NUM - 1; i >= 0; i--) {
if (dpolicy->timeout != 0 &&
f2fs_time_over(sbi, dpolicy->timeout))
break;
if (i + 1 < dpolicy->granularity)
break;
if (i < DEFAULT_DISCARD_GRANULARITY && dpolicy->ordered)
return __issue_discard_cmd_orderly(sbi, dpolicy);
pend_list = &dcc->pend_list[i];
mutex_lock(&dcc->cmd_lock);
if (list_empty(pend_list))
goto next;
if (unlikely(dcc->rbtree_check))
f2fs_bug_on(sbi, !f2fs_check_rb_tree_consistence(sbi,
&dcc->root));
blk_start_plug(&plug);
list_for_each_entry_safe(dc, tmp, pend_list, list) {
f2fs_bug_on(sbi, dc->state != D_PREP);
if (dpolicy->timeout != 0 &&
f2fs_time_over(sbi, dpolicy->timeout))
break;
if (dpolicy->io_aware && i < dpolicy->io_aware_gran &&
!is_idle(sbi, DISCARD_TIME)) {
io_interrupted = true;
break;
}
__submit_discard_cmd(sbi, dpolicy, dc, &issued);
if (issued >= dpolicy->max_requests)
break;
}
blk_finish_plug(&plug);
next:
mutex_unlock(&dcc->cmd_lock);
if (issued >= dpolicy->max_requests || io_interrupted)
break;
}
if (!issued && io_interrupted)
issued = -1;
return issued;
}
static bool __drop_discard_cmd(struct f2fs_sb_info *sbi)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct list_head *pend_list;
struct discard_cmd *dc, *tmp;
int i;
bool dropped = false;
mutex_lock(&dcc->cmd_lock);
for (i = MAX_PLIST_NUM - 1; i >= 0; i--) {
pend_list = &dcc->pend_list[i];
list_for_each_entry_safe(dc, tmp, pend_list, list) {
f2fs_bug_on(sbi, dc->state != D_PREP);
__remove_discard_cmd(sbi, dc);
dropped = true;
}
}
mutex_unlock(&dcc->cmd_lock);
return dropped;
}
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 16:20:41 +00:00
void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi)
{
__drop_discard_cmd(sbi);
}
static unsigned int __wait_one_discard_bio(struct f2fs_sb_info *sbi,
struct discard_cmd *dc)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
unsigned int len = 0;
wait_for_completion_io(&dc->wait);
mutex_lock(&dcc->cmd_lock);
f2fs_bug_on(sbi, dc->state != D_DONE);
dc->ref--;
if (!dc->ref) {
if (!dc->error)
len = dc->len;
__remove_discard_cmd(sbi, dc);
}
mutex_unlock(&dcc->cmd_lock);
return len;
}
static unsigned int __wait_discard_cmd_range(struct f2fs_sb_info *sbi,
struct discard_policy *dpolicy,
block_t start, block_t end)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct list_head *wait_list = (dpolicy->type == DPOLICY_FSTRIM) ?
&(dcc->fstrim_list) : &(dcc->wait_list);
struct discard_cmd *dc, *tmp;
bool need_wait;
unsigned int trimmed = 0;
next:
need_wait = false;
mutex_lock(&dcc->cmd_lock);
list_for_each_entry_safe(dc, tmp, wait_list, list) {
if (dc->lstart + dc->len <= start || end <= dc->lstart)
continue;
if (dc->len < dpolicy->granularity)
continue;
if (dc->state == D_DONE && !dc->ref) {
wait_for_completion_io(&dc->wait);
if (!dc->error)
trimmed += dc->len;
__remove_discard_cmd(sbi, dc);
} else {
dc->ref++;
need_wait = true;
break;
}
}
mutex_unlock(&dcc->cmd_lock);
if (need_wait) {
trimmed += __wait_one_discard_bio(sbi, dc);
goto next;
}
return trimmed;
}
static unsigned int __wait_all_discard_cmd(struct f2fs_sb_info *sbi,
struct discard_policy *dpolicy)
{
struct discard_policy dp;
unsigned int discard_blks;
if (dpolicy)
return __wait_discard_cmd_range(sbi, dpolicy, 0, UINT_MAX);
/* wait all */
__init_discard_policy(sbi, &dp, DPOLICY_FSTRIM, 1);
discard_blks = __wait_discard_cmd_range(sbi, &dp, 0, UINT_MAX);
__init_discard_policy(sbi, &dp, DPOLICY_UMOUNT, 1);
discard_blks += __wait_discard_cmd_range(sbi, &dp, 0, UINT_MAX);
return discard_blks;
}
/* This should be covered by global mutex, &sit_i->sentry_lock */
static void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct discard_cmd *dc;
bool need_wait = false;
mutex_lock(&dcc->cmd_lock);
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 16:20:41 +00:00
dc = (struct discard_cmd *)f2fs_lookup_rb_tree(&dcc->root,
NULL, blkaddr);
if (dc) {
if (dc->state == D_PREP) {
__punch_discard_cmd(sbi, dc, blkaddr);
} else {
dc->ref++;
need_wait = true;
}
}
mutex_unlock(&dcc->cmd_lock);
if (need_wait)
__wait_one_discard_bio(sbi, dc);
}
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 16:20:41 +00:00
void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
if (dcc && dcc->f2fs_issue_discard) {
struct task_struct *discard_thread = dcc->f2fs_issue_discard;
dcc->f2fs_issue_discard = NULL;
kthread_stop(discard_thread);
}
}
/* This comes from f2fs_put_super */
bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct discard_policy dpolicy;
bool dropped;
__init_discard_policy(sbi, &dpolicy, DPOLICY_UMOUNT,
dcc->discard_granularity);
dpolicy.timeout = UMOUNT_DISCARD_TIMEOUT;
__issue_discard_cmd(sbi, &dpolicy);
dropped = __drop_discard_cmd(sbi);
/* just to make sure there is no pending discard commands */
__wait_all_discard_cmd(sbi, NULL);
f2fs_bug_on(sbi, atomic_read(&dcc->discard_cmd_cnt));
return dropped;
}
static int issue_discard_thread(void *data)
{
struct f2fs_sb_info *sbi = data;
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
wait_queue_head_t *q = &dcc->discard_wait_queue;
struct discard_policy dpolicy;
unsigned int wait_ms = DEF_MIN_DISCARD_ISSUE_TIME;
int issued;
set_freezable();
do {
__init_discard_policy(sbi, &dpolicy, DPOLICY_BG,
dcc->discard_granularity);
wait_event_interruptible_timeout(*q,
kthread_should_stop() || freezing(current) ||
dcc->discard_wake,
msecs_to_jiffies(wait_ms));
if (dcc->discard_wake)
dcc->discard_wake = 0;
/* clean up pending candidates before going to sleep */
if (atomic_read(&dcc->queued_discard))
__wait_all_discard_cmd(sbi, NULL);
if (try_to_freeze())
continue;
if (f2fs_readonly(sbi->sb))
continue;
if (kthread_should_stop())
return 0;
if (is_sbi_flag_set(sbi, SBI_NEED_FSCK)) {
wait_ms = dpolicy.max_interval;
continue;
}
if (sbi->gc_mode == GC_URGENT)
__init_discard_policy(sbi, &dpolicy, DPOLICY_FORCE, 1);
f2fs: make background threads of f2fs being aware of freezing When ->freeze_fs is called from lvm for doing snapshot, it needs to make sure there will be no more changes in filesystem's data, however, previously, background threads like GC thread wasn't aware of freezing, so in environment with active background threads, data of snapshot becomes unstable. This patch fixes this issue by adding sb_{start,end}_intwrite in below background threads: - GC thread - flush thread - discard thread Note that, don't use sb_start_intwrite() in gc_thread_func() due to: generic/241 reports below bug: ====================================================== WARNING: possible circular locking dependency detected 4.13.0-rc1+ #32 Tainted: G O ------------------------------------------------------ f2fs_gc-250:0/22186 is trying to acquire lock: (&sbi->gc_mutex){+.+...}, at: [<f8fa7f0b>] f2fs_sync_fs+0x7b/0x1b0 [f2fs] but task is already holding lock: (sb_internal#2){++++.-}, at: [<f8fb5609>] gc_thread_func+0x159/0x4a0 [f2fs] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (sb_internal#2){++++.-}: __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __sb_start_write+0x11d/0x1f0 f2fs_evict_inode+0x2d6/0x4e0 [f2fs] evict+0xa8/0x170 iput+0x1fb/0x2c0 f2fs_sync_inode_meta+0x3f/0xf0 [f2fs] write_checkpoint+0x1b1/0x750 [f2fs] f2fs_sync_fs+0x85/0x1b0 [f2fs] f2fs_do_sync_file.isra.24+0x137/0xa30 [f2fs] f2fs_sync_file+0x34/0x40 [f2fs] vfs_fsync_range+0x4a/0xa0 do_fsync+0x3c/0x60 SyS_fdatasync+0x15/0x20 do_fast_syscall_32+0xa1/0x1b0 entry_SYSENTER_32+0x4c/0x7b -> #1 (&sbi->cp_mutex){+.+...}: __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __mutex_lock+0x4f/0x830 mutex_lock_nested+0x25/0x30 write_checkpoint+0x2f/0x750 [f2fs] f2fs_sync_fs+0x85/0x1b0 [f2fs] sync_filesystem+0x67/0x80 generic_shutdown_super+0x27/0x100 kill_block_super+0x22/0x50 kill_f2fs_super+0x3a/0x40 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x69/0x80 exit_to_usermode_loop+0x57/0x92 do_fast_syscall_32+0x18c/0x1b0 entry_SYSENTER_32+0x4c/0x7b -> #0 (&sbi->gc_mutex){+.+...}: validate_chain.isra.36+0xc50/0xdb0 __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __mutex_lock+0x4f/0x830 mutex_lock_nested+0x25/0x30 f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_balance_fs_bg+0xb9/0x200 [f2fs] gc_thread_func+0x302/0x4a0 [f2fs] kthread+0xe9/0x120 ret_from_fork+0x19/0x24 other info that might help us debug this: Chain exists of: &sbi->gc_mutex --> &sbi->cp_mutex --> sb_internal#2 Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(sb_internal#2); lock(&sbi->cp_mutex); lock(sb_internal#2); lock(&sbi->gc_mutex); *** DEADLOCK *** 1 lock held by f2fs_gc-250:0/22186: #0: (sb_internal#2){++++.-}, at: [<f8fb5609>] gc_thread_func+0x159/0x4a0 [f2fs] stack backtrace: CPU: 2 PID: 22186 Comm: f2fs_gc-250:0 Tainted: G O 4.13.0-rc1+ #32 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 Call Trace: dump_stack+0x5f/0x92 print_circular_bug+0x1b3/0x1bd validate_chain.isra.36+0xc50/0xdb0 ? __this_cpu_preempt_check+0xf/0x20 __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] __mutex_lock+0x4f/0x830 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] mutex_lock_nested+0x25/0x30 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_balance_fs_bg+0xb9/0x200 [f2fs] gc_thread_func+0x302/0x4a0 [f2fs] ? preempt_schedule_common+0x2f/0x4d ? f2fs_gc+0x540/0x540 [f2fs] kthread+0xe9/0x120 ? f2fs_gc+0x540/0x540 [f2fs] ? kthread_create_on_node+0x30/0x30 ret_from_fork+0x19/0x24 The deadlock occurs in below condition: GC Thread Thread B - sb_start_intwrite - f2fs_sync_file - f2fs_sync_fs - mutex_lock(&sbi->gc_mutex) - write_checkpoint - block_operations - f2fs_sync_inode_meta - iput - sb_start_intwrite - mutex_lock(&sbi->gc_mutex) Fix this by altering sb_start_intwrite to sb_start_write_trylock. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-22 00:52:23 +00:00
sb_start_intwrite(sbi->sb);
issued = __issue_discard_cmd(sbi, &dpolicy);
if (issued > 0) {
__wait_all_discard_cmd(sbi, &dpolicy);
wait_ms = dpolicy.min_interval;
} else if (issued == -1){
wait_ms = f2fs_time_to_wait(sbi, DISCARD_TIME);
if (!wait_ms)
wait_ms = dpolicy.mid_interval;
} else {
wait_ms = dpolicy.max_interval;
}
f2fs: make background threads of f2fs being aware of freezing When ->freeze_fs is called from lvm for doing snapshot, it needs to make sure there will be no more changes in filesystem's data, however, previously, background threads like GC thread wasn't aware of freezing, so in environment with active background threads, data of snapshot becomes unstable. This patch fixes this issue by adding sb_{start,end}_intwrite in below background threads: - GC thread - flush thread - discard thread Note that, don't use sb_start_intwrite() in gc_thread_func() due to: generic/241 reports below bug: ====================================================== WARNING: possible circular locking dependency detected 4.13.0-rc1+ #32 Tainted: G O ------------------------------------------------------ f2fs_gc-250:0/22186 is trying to acquire lock: (&sbi->gc_mutex){+.+...}, at: [<f8fa7f0b>] f2fs_sync_fs+0x7b/0x1b0 [f2fs] but task is already holding lock: (sb_internal#2){++++.-}, at: [<f8fb5609>] gc_thread_func+0x159/0x4a0 [f2fs] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (sb_internal#2){++++.-}: __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __sb_start_write+0x11d/0x1f0 f2fs_evict_inode+0x2d6/0x4e0 [f2fs] evict+0xa8/0x170 iput+0x1fb/0x2c0 f2fs_sync_inode_meta+0x3f/0xf0 [f2fs] write_checkpoint+0x1b1/0x750 [f2fs] f2fs_sync_fs+0x85/0x1b0 [f2fs] f2fs_do_sync_file.isra.24+0x137/0xa30 [f2fs] f2fs_sync_file+0x34/0x40 [f2fs] vfs_fsync_range+0x4a/0xa0 do_fsync+0x3c/0x60 SyS_fdatasync+0x15/0x20 do_fast_syscall_32+0xa1/0x1b0 entry_SYSENTER_32+0x4c/0x7b -> #1 (&sbi->cp_mutex){+.+...}: __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __mutex_lock+0x4f/0x830 mutex_lock_nested+0x25/0x30 write_checkpoint+0x2f/0x750 [f2fs] f2fs_sync_fs+0x85/0x1b0 [f2fs] sync_filesystem+0x67/0x80 generic_shutdown_super+0x27/0x100 kill_block_super+0x22/0x50 kill_f2fs_super+0x3a/0x40 [f2fs] deactivate_locked_super+0x3d/0x70 deactivate_super+0x40/0x60 cleanup_mnt+0x39/0x70 __cleanup_mnt+0x10/0x20 task_work_run+0x69/0x80 exit_to_usermode_loop+0x57/0x92 do_fast_syscall_32+0x18c/0x1b0 entry_SYSENTER_32+0x4c/0x7b -> #0 (&sbi->gc_mutex){+.+...}: validate_chain.isra.36+0xc50/0xdb0 __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 __mutex_lock+0x4f/0x830 mutex_lock_nested+0x25/0x30 f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_balance_fs_bg+0xb9/0x200 [f2fs] gc_thread_func+0x302/0x4a0 [f2fs] kthread+0xe9/0x120 ret_from_fork+0x19/0x24 other info that might help us debug this: Chain exists of: &sbi->gc_mutex --> &sbi->cp_mutex --> sb_internal#2 Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(sb_internal#2); lock(&sbi->cp_mutex); lock(sb_internal#2); lock(&sbi->gc_mutex); *** DEADLOCK *** 1 lock held by f2fs_gc-250:0/22186: #0: (sb_internal#2){++++.-}, at: [<f8fb5609>] gc_thread_func+0x159/0x4a0 [f2fs] stack backtrace: CPU: 2 PID: 22186 Comm: f2fs_gc-250:0 Tainted: G O 4.13.0-rc1+ #32 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 Call Trace: dump_stack+0x5f/0x92 print_circular_bug+0x1b3/0x1bd validate_chain.isra.36+0xc50/0xdb0 ? __this_cpu_preempt_check+0xf/0x20 __lock_acquire+0x405/0x7b0 lock_acquire+0xae/0x220 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] __mutex_lock+0x4f/0x830 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] mutex_lock_nested+0x25/0x30 ? f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_sync_fs+0x7b/0x1b0 [f2fs] f2fs_balance_fs_bg+0xb9/0x200 [f2fs] gc_thread_func+0x302/0x4a0 [f2fs] ? preempt_schedule_common+0x2f/0x4d ? f2fs_gc+0x540/0x540 [f2fs] kthread+0xe9/0x120 ? f2fs_gc+0x540/0x540 [f2fs] ? kthread_create_on_node+0x30/0x30 ret_from_fork+0x19/0x24 The deadlock occurs in below condition: GC Thread Thread B - sb_start_intwrite - f2fs_sync_file - f2fs_sync_fs - mutex_lock(&sbi->gc_mutex) - write_checkpoint - block_operations - f2fs_sync_inode_meta - iput - sb_start_intwrite - mutex_lock(&sbi->gc_mutex) Fix this by altering sb_start_intwrite to sb_start_write_trylock. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2017-07-22 00:52:23 +00:00
sb_end_intwrite(sbi->sb);
} while (!kthread_should_stop());
return 0;
}
#ifdef CONFIG_BLK_DEV_ZONED
static int __f2fs_issue_discard_zone(struct f2fs_sb_info *sbi,
struct block_device *bdev, block_t blkstart, block_t blklen)
{
sector_t sector, nr_sects;
block_t lblkstart = blkstart;
int devi = 0;
if (f2fs_is_multi_device(sbi)) {
devi = f2fs_target_device_index(sbi, blkstart);
if (blkstart < FDEV(devi).start_blk ||
blkstart > FDEV(devi).end_blk) {
f2fs_err(sbi, "Invalid block %x", blkstart);
return -EIO;
}
blkstart -= FDEV(devi).start_blk;
}
/* For sequential zones, reset the zone write pointer */
if (f2fs_blkz_is_seq(sbi, devi, blkstart)) {
sector = SECTOR_FROM_BLOCK(blkstart);
nr_sects = SECTOR_FROM_BLOCK(blklen);
if (sector & (bdev_zone_sectors(bdev) - 1) ||
nr_sects != bdev_zone_sectors(bdev)) {
f2fs_err(sbi, "(%d) %s: Unaligned zone reset attempted (block %x + %x)",
devi, sbi->s_ndevs ? FDEV(devi).path : "",
blkstart, blklen);
return -EIO;
}
trace_f2fs_issue_reset_zone(bdev, blkstart);
2019-10-27 14:05:45 +00:00
return blkdev_zone_mgmt(bdev, REQ_OP_ZONE_RESET,
sector, nr_sects, GFP_NOFS);
}
/* For conventional zones, use regular discard if supported */
return __queue_discard_cmd(sbi, bdev, lblkstart, blklen);
}
#endif
static int __issue_discard_async(struct f2fs_sb_info *sbi,
struct block_device *bdev, block_t blkstart, block_t blklen)
{
#ifdef CONFIG_BLK_DEV_ZONED
if (f2fs_sb_has_blkzoned(sbi) && bdev_is_zoned(bdev))
return __f2fs_issue_discard_zone(sbi, bdev, blkstart, blklen);
#endif
return __queue_discard_cmd(sbi, bdev, blkstart, blklen);
}
f2fs: avoid to conduct roll-forward due to the remained garbage blocks The f2fs always scans the next chain of direct node blocks. But some garbage blocks are able to be remained due to no discard support or SSR triggers. This occasionally wreaks recovering wrong inodes that were used or BUG_ONs due to reallocating node ids as follows. When mount this f2fs image: http://linuxtesting.org/downloads/f2fs_fault_image.zip BUG_ON is triggered in f2fs driver (messages below are generated on kernel 3.13.2; for other kernels output is similar): kernel BUG at fs/f2fs/node.c:215! Call Trace: [<ffffffffa032ebad>] recover_inode_page+0x1fd/0x3e0 [f2fs] [<ffffffff811446e7>] ? __lock_page+0x67/0x70 [<ffffffff81089990>] ? autoremove_wake_function+0x50/0x50 [<ffffffffa0337788>] recover_fsync_data+0x1398/0x15d0 [f2fs] [<ffffffff812b9e5c>] ? selinux_d_instantiate+0x1c/0x20 [<ffffffff811cb20b>] ? d_instantiate+0x5b/0x80 [<ffffffffa0321044>] f2fs_fill_super+0xb04/0xbf0 [f2fs] [<ffffffff811b861e>] ? mount_bdev+0x7e/0x210 [<ffffffff811b8769>] mount_bdev+0x1c9/0x210 [<ffffffffa0320540>] ? validate_superblock+0x210/0x210 [f2fs] [<ffffffffa031cf8d>] f2fs_mount+0x1d/0x30 [f2fs] [<ffffffff811b9497>] mount_fs+0x47/0x1c0 [<ffffffff81166e00>] ? __alloc_percpu+0x10/0x20 [<ffffffff811d4032>] vfs_kern_mount+0x72/0x110 [<ffffffff811d6763>] do_mount+0x493/0x910 [<ffffffff811615cb>] ? strndup_user+0x5b/0x80 [<ffffffff811d6c70>] SyS_mount+0x90/0xe0 [<ffffffff8166f8d9>] system_call_fastpath+0x16/0x1b Found by Linux File System Verification project (linuxtesting.org). Reported-by: Andrey Tsyvarev <tsyvarev@ispras.ru> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-04-15 04:57:55 +00:00
static int f2fs_issue_discard(struct f2fs_sb_info *sbi,
block_t blkstart, block_t blklen)
{
sector_t start = blkstart, len = 0;
struct block_device *bdev;
struct seg_entry *se;
unsigned int offset;
block_t i;
int err = 0;
bdev = f2fs_target_device(sbi, blkstart, NULL);
for (i = blkstart; i < blkstart + blklen; i++, len++) {
if (i != start) {
struct block_device *bdev2 =
f2fs_target_device(sbi, i, NULL);
if (bdev2 != bdev) {
err = __issue_discard_async(sbi, bdev,
start, len);
if (err)
return err;
bdev = bdev2;
start = i;
len = 0;
}
}
se = get_seg_entry(sbi, GET_SEGNO(sbi, i));
offset = GET_BLKOFF_FROM_SEG0(sbi, i);
if (!f2fs_test_and_set_bit(offset, se->discard_map))
sbi->discard_blks--;
}
if (len)
err = __issue_discard_async(sbi, bdev, start, len);
return err;
f2fs: avoid to conduct roll-forward due to the remained garbage blocks The f2fs always scans the next chain of direct node blocks. But some garbage blocks are able to be remained due to no discard support or SSR triggers. This occasionally wreaks recovering wrong inodes that were used or BUG_ONs due to reallocating node ids as follows. When mount this f2fs image: http://linuxtesting.org/downloads/f2fs_fault_image.zip BUG_ON is triggered in f2fs driver (messages below are generated on kernel 3.13.2; for other kernels output is similar): kernel BUG at fs/f2fs/node.c:215! Call Trace: [<ffffffffa032ebad>] recover_inode_page+0x1fd/0x3e0 [f2fs] [<ffffffff811446e7>] ? __lock_page+0x67/0x70 [<ffffffff81089990>] ? autoremove_wake_function+0x50/0x50 [<ffffffffa0337788>] recover_fsync_data+0x1398/0x15d0 [f2fs] [<ffffffff812b9e5c>] ? selinux_d_instantiate+0x1c/0x20 [<ffffffff811cb20b>] ? d_instantiate+0x5b/0x80 [<ffffffffa0321044>] f2fs_fill_super+0xb04/0xbf0 [f2fs] [<ffffffff811b861e>] ? mount_bdev+0x7e/0x210 [<ffffffff811b8769>] mount_bdev+0x1c9/0x210 [<ffffffffa0320540>] ? validate_superblock+0x210/0x210 [f2fs] [<ffffffffa031cf8d>] f2fs_mount+0x1d/0x30 [f2fs] [<ffffffff811b9497>] mount_fs+0x47/0x1c0 [<ffffffff81166e00>] ? __alloc_percpu+0x10/0x20 [<ffffffff811d4032>] vfs_kern_mount+0x72/0x110 [<ffffffff811d6763>] do_mount+0x493/0x910 [<ffffffff811615cb>] ? strndup_user+0x5b/0x80 [<ffffffff811d6c70>] SyS_mount+0x90/0xe0 [<ffffffff8166f8d9>] system_call_fastpath+0x16/0x1b Found by Linux File System Verification project (linuxtesting.org). Reported-by: Andrey Tsyvarev <tsyvarev@ispras.ru> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2014-04-15 04:57:55 +00:00
}
static bool add_discard_addrs(struct f2fs_sb_info *sbi, struct cp_control *cpc,
bool check_only)
{
int entries = SIT_VBLOCK_MAP_SIZE / sizeof(unsigned long);
int max_blocks = sbi->blocks_per_seg;
struct seg_entry *se = get_seg_entry(sbi, cpc->trim_start);
unsigned long *cur_map = (unsigned long *)se->cur_valid_map;
unsigned long *ckpt_map = (unsigned long *)se->ckpt_valid_map;
unsigned long *discard_map = (unsigned long *)se->discard_map;
unsigned long *dmap = SIT_I(sbi)->tmp_map;
unsigned int start = 0, end = -1;
bool force = (cpc->reason & CP_DISCARD);
struct discard_entry *de = NULL;
struct list_head *head = &SM_I(sbi)->dcc_info->entry_list;
int i;
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
if (se->valid_blocks == max_blocks || !f2fs_hw_support_discard(sbi))
return false;
if (!force) {
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
if (!f2fs_realtime_discard_enable(sbi) || !se->valid_blocks ||
SM_I(sbi)->dcc_info->nr_discards >=
SM_I(sbi)->dcc_info->max_discards)
return false;
}
/* SIT_VBLOCK_MAP_SIZE should be multiple of sizeof(unsigned long) */
for (i = 0; i < entries; i++)
dmap[i] = force ? ~ckpt_map[i] & ~discard_map[i] :
(cur_map[i] ^ ckpt_map[i]) & ckpt_map[i];
while (force || SM_I(sbi)->dcc_info->nr_discards <=
SM_I(sbi)->dcc_info->max_discards) {
start = __find_rev_next_bit(dmap, max_blocks, end + 1);
if (start >= max_blocks)
break;
end = __find_rev_next_zero_bit(dmap, max_blocks, start + 1);
if (force && start && end != max_blocks
&& (end - start) < cpc->trim_minlen)
continue;
if (check_only)
return true;
if (!de) {
de = f2fs_kmem_cache_alloc(discard_entry_slab,
GFP_F2FS_ZERO);
de->start_blkaddr = START_BLOCK(sbi, cpc->trim_start);
list_add_tail(&de->list, head);
}
for (i = start; i < end; i++)
__set_bit_le(i, (void *)de->discard_map);
SM_I(sbi)->dcc_info->nr_discards += end - start;
}
return false;
}
static void release_discard_addr(struct discard_entry *entry)
{
list_del(&entry->list);
kmem_cache_free(discard_entry_slab, entry);
}
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 16:20:41 +00:00
void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi)
{
struct list_head *head = &(SM_I(sbi)->dcc_info->entry_list);
struct discard_entry *entry, *this;
/* drop caches */
list_for_each_entry_safe(entry, this, head, list)
release_discard_addr(entry);
}
/*
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 16:20:41 +00:00
* Should call f2fs_clear_prefree_segments after checkpoint is done.
*/
static void set_prefree_as_free_segments(struct f2fs_sb_info *sbi)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
unsigned int segno;
mutex_lock(&dirty_i->seglist_lock);
for_each_set_bit(segno, dirty_i->dirty_segmap[PRE], MAIN_SEGS(sbi))
__set_test_and_free(sbi, segno);
mutex_unlock(&dirty_i->seglist_lock);
}
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 16:20:41 +00:00
void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
struct cp_control *cpc)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct list_head *head = &dcc->entry_list;
struct discard_entry *entry, *this;
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
unsigned long *prefree_map = dirty_i->dirty_segmap[PRE];
unsigned int start = 0, end = -1;
unsigned int secno, start_segno;
bool force = (cpc->reason & CP_DISCARD);
bool need_align = test_opt(sbi, LFS) && __is_large_section(sbi);
mutex_lock(&dirty_i->seglist_lock);
while (1) {
int i;
f2fs: issue discard align to section in LFS mode For the case when sbi->segs_per_sec > 1 with lfs mode, take section:segment = 5 for example, if the section prefree_map is ...previous section | current section (1 1 0 1 1) | next section..., then the start = x, end = x + 1, after start = start_segno + sbi->segs_per_sec, start = x + 5, then it will skip x + 3 and x + 4, but their bitmap is still set, which will cause duplicated f2fs_issue_discard of this same section in the next write_checkpoint: round 1: section bitmap : 1 1 1 1 1, all valid, prefree_map: 0 0 0 0 0 then rm data block NO.2, block NO.2 becomes invalid, prefree_map: 0 0 1 0 0 write_checkpoint: section bitmap: 1 1 0 1 1, prefree_map: 0 0 0 0 0, prefree of NO.2 is cleared, and no discard issued round 2: rm data block NO.0, NO.1, NO.3, NO.4 all invalid, but prefree bit of NO.2 is set and cleared in round 1, then prefree_map: 1 1 0 1 1 write_checkpoint: section bitmap: 0 0 0 0 0, prefree_map: 0 0 0 1 1, no valid blocks of this section, so discard issued, but this time prefree bit of NO.3 and NO.4 is skipped due to start = start_segno + sbi->segs_per_sec; round 3: write_checkpoint: section bitmap: 0 0 0 0 0, prefree_map: 0 0 0 1 1 -> 0 0 0 0 0, no valid blocks of this section, so discard issued, this time prefree bit of NO.3 and NO.4 is cleared, but the discard of this section is sent again... To fix this problem, we can align the start and end value to section boundary for fstrim and real-time discard operation, and decide to issue discard only when the whole section is invalid, which can issue discard aligned to section size as much as possible and avoid redundant discard. Signed-off-by: Yunlong Song <yunlong.song@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-07-19 12:58:15 +00:00
if (need_align && end != -1)
end--;
start = find_next_bit(prefree_map, MAIN_SEGS(sbi), end + 1);
if (start >= MAIN_SEGS(sbi))
break;
end = find_next_zero_bit(prefree_map, MAIN_SEGS(sbi),
start + 1);
f2fs: issue discard align to section in LFS mode For the case when sbi->segs_per_sec > 1 with lfs mode, take section:segment = 5 for example, if the section prefree_map is ...previous section | current section (1 1 0 1 1) | next section..., then the start = x, end = x + 1, after start = start_segno + sbi->segs_per_sec, start = x + 5, then it will skip x + 3 and x + 4, but their bitmap is still set, which will cause duplicated f2fs_issue_discard of this same section in the next write_checkpoint: round 1: section bitmap : 1 1 1 1 1, all valid, prefree_map: 0 0 0 0 0 then rm data block NO.2, block NO.2 becomes invalid, prefree_map: 0 0 1 0 0 write_checkpoint: section bitmap: 1 1 0 1 1, prefree_map: 0 0 0 0 0, prefree of NO.2 is cleared, and no discard issued round 2: rm data block NO.0, NO.1, NO.3, NO.4 all invalid, but prefree bit of NO.2 is set and cleared in round 1, then prefree_map: 1 1 0 1 1 write_checkpoint: section bitmap: 0 0 0 0 0, prefree_map: 0 0 0 1 1, no valid blocks of this section, so discard issued, but this time prefree bit of NO.3 and NO.4 is skipped due to start = start_segno + sbi->segs_per_sec; round 3: write_checkpoint: section bitmap: 0 0 0 0 0, prefree_map: 0 0 0 1 1 -> 0 0 0 0 0, no valid blocks of this section, so discard issued, this time prefree bit of NO.3 and NO.4 is cleared, but the discard of this section is sent again... To fix this problem, we can align the start and end value to section boundary for fstrim and real-time discard operation, and decide to issue discard only when the whole section is invalid, which can issue discard aligned to section size as much as possible and avoid redundant discard. Signed-off-by: Yunlong Song <yunlong.song@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-07-19 12:58:15 +00:00
if (need_align) {
start = rounddown(start, sbi->segs_per_sec);
end = roundup(end, sbi->segs_per_sec);
}
f2fs: issue discard align to section in LFS mode For the case when sbi->segs_per_sec > 1 with lfs mode, take section:segment = 5 for example, if the section prefree_map is ...previous section | current section (1 1 0 1 1) | next section..., then the start = x, end = x + 1, after start = start_segno + sbi->segs_per_sec, start = x + 5, then it will skip x + 3 and x + 4, but their bitmap is still set, which will cause duplicated f2fs_issue_discard of this same section in the next write_checkpoint: round 1: section bitmap : 1 1 1 1 1, all valid, prefree_map: 0 0 0 0 0 then rm data block NO.2, block NO.2 becomes invalid, prefree_map: 0 0 1 0 0 write_checkpoint: section bitmap: 1 1 0 1 1, prefree_map: 0 0 0 0 0, prefree of NO.2 is cleared, and no discard issued round 2: rm data block NO.0, NO.1, NO.3, NO.4 all invalid, but prefree bit of NO.2 is set and cleared in round 1, then prefree_map: 1 1 0 1 1 write_checkpoint: section bitmap: 0 0 0 0 0, prefree_map: 0 0 0 1 1, no valid blocks of this section, so discard issued, but this time prefree bit of NO.3 and NO.4 is skipped due to start = start_segno + sbi->segs_per_sec; round 3: write_checkpoint: section bitmap: 0 0 0 0 0, prefree_map: 0 0 0 1 1 -> 0 0 0 0 0, no valid blocks of this section, so discard issued, this time prefree bit of NO.3 and NO.4 is cleared, but the discard of this section is sent again... To fix this problem, we can align the start and end value to section boundary for fstrim and real-time discard operation, and decide to issue discard only when the whole section is invalid, which can issue discard aligned to section size as much as possible and avoid redundant discard. Signed-off-by: Yunlong Song <yunlong.song@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-07-19 12:58:15 +00:00
for (i = start; i < end; i++) {
if (test_and_clear_bit(i, prefree_map))
dirty_i->nr_dirty[PRE]--;
}
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
if (!f2fs_realtime_discard_enable(sbi))
continue;
if (force && start >= cpc->trim_start &&
(end - 1) <= cpc->trim_end)
continue;
if (!test_opt(sbi, LFS) || !__is_large_section(sbi)) {
f2fs_issue_discard(sbi, START_BLOCK(sbi, start),
(end - start) << sbi->log_blocks_per_seg);
continue;
}
next:
secno = GET_SEC_FROM_SEG(sbi, start);
start_segno = GET_SEG_FROM_SEC(sbi, secno);
if (!IS_CURSEC(sbi, secno) &&
!get_valid_blocks(sbi, start, true))
f2fs_issue_discard(sbi, START_BLOCK(sbi, start_segno),
sbi->segs_per_sec << sbi->log_blocks_per_seg);
start = start_segno + sbi->segs_per_sec;
if (start < end)
goto next;
else
end = start - 1;
}
mutex_unlock(&dirty_i->seglist_lock);
/* send small discards */
list_for_each_entry_safe(entry, this, head, list) {
unsigned int cur_pos = 0, next_pos, len, total_len = 0;
bool is_valid = test_bit_le(0, entry->discard_map);
find_next:
if (is_valid) {
next_pos = find_next_zero_bit_le(entry->discard_map,
sbi->blocks_per_seg, cur_pos);
len = next_pos - cur_pos;
if (f2fs_sb_has_blkzoned(sbi) ||
(force && len < cpc->trim_minlen))
goto skip;
f2fs_issue_discard(sbi, entry->start_blkaddr + cur_pos,
len);
total_len += len;
} else {
next_pos = find_next_bit_le(entry->discard_map,
sbi->blocks_per_seg, cur_pos);
}
skip:
cur_pos = next_pos;
is_valid = !is_valid;
if (cur_pos < sbi->blocks_per_seg)
goto find_next;
release_discard_addr(entry);
dcc->nr_discards -= total_len;
}
wake_up_discard_thread(sbi, false);
}
static int create_discard_cmd_control(struct f2fs_sb_info *sbi)
{
dev_t dev = sbi->sb->s_bdev->bd_dev;
struct discard_cmd_control *dcc;
int err = 0, i;
if (SM_I(sbi)->dcc_info) {
dcc = SM_I(sbi)->dcc_info;
goto init_thread;
}
dcc = f2fs_kzalloc(sbi, sizeof(struct discard_cmd_control), GFP_KERNEL);
if (!dcc)
return -ENOMEM;
dcc->discard_granularity = DEFAULT_DISCARD_GRANULARITY;
INIT_LIST_HEAD(&dcc->entry_list);
for (i = 0; i < MAX_PLIST_NUM; i++)
INIT_LIST_HEAD(&dcc->pend_list[i]);
INIT_LIST_HEAD(&dcc->wait_list);
INIT_LIST_HEAD(&dcc->fstrim_list);
mutex_init(&dcc->cmd_lock);
atomic_set(&dcc->issued_discard, 0);
atomic_set(&dcc->queued_discard, 0);
atomic_set(&dcc->discard_cmd_cnt, 0);
dcc->nr_discards = 0;
dcc->max_discards = MAIN_SEGS(sbi) << sbi->log_blocks_per_seg;
dcc->undiscard_blks = 0;
dcc->next_pos = 0;
dcc->root = RB_ROOT_CACHED;
dcc->rbtree_check = false;
init_waitqueue_head(&dcc->discard_wait_queue);
SM_I(sbi)->dcc_info = dcc;
init_thread:
dcc->f2fs_issue_discard = kthread_run(issue_discard_thread, sbi,
"f2fs_discard-%u:%u", MAJOR(dev), MINOR(dev));
if (IS_ERR(dcc->f2fs_issue_discard)) {
err = PTR_ERR(dcc->f2fs_issue_discard);
kvfree(dcc);
SM_I(sbi)->dcc_info = NULL;
return err;
}
return err;
}
static void destroy_discard_cmd_control(struct f2fs_sb_info *sbi)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
if (!dcc)
return;
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 16:20:41 +00:00
f2fs_stop_discard_thread(sbi);
/*
* Recovery can cache discard commands, so in error path of
* fill_super(), it needs to give a chance to handle them.
*/
if (unlikely(atomic_read(&dcc->discard_cmd_cnt)))
f2fs_issue_discard_timeout(sbi);
kvfree(dcc);
SM_I(sbi)->dcc_info = NULL;
}
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
static bool __mark_sit_entry_dirty(struct f2fs_sb_info *sbi, unsigned int segno)
{
struct sit_info *sit_i = SIT_I(sbi);
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
if (!__test_and_set_bit(segno, sit_i->dirty_sentries_bitmap)) {
sit_i->dirty_sentries++;
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
return false;
}
return true;
}
static void __set_sit_entry_type(struct f2fs_sb_info *sbi, int type,
unsigned int segno, int modified)
{
struct seg_entry *se = get_seg_entry(sbi, segno);
se->type = type;
if (modified)
__mark_sit_entry_dirty(sbi, segno);
}
static void update_sit_entry(struct f2fs_sb_info *sbi, block_t blkaddr, int del)
{
struct seg_entry *se;
unsigned int segno, offset;
long int new_vblocks;
bool exist;
#ifdef CONFIG_F2FS_CHECK_FS
bool mir_exist;
#endif
segno = GET_SEGNO(sbi, blkaddr);
se = get_seg_entry(sbi, segno);
new_vblocks = se->valid_blocks + del;
offset = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
f2fs_bug_on(sbi, (new_vblocks >> (sizeof(unsigned short) << 3) ||
(new_vblocks > sbi->blocks_per_seg)));
se->valid_blocks = new_vblocks;
se->mtime = get_mtime(sbi, false);
if (se->mtime > SIT_I(sbi)->max_mtime)
SIT_I(sbi)->max_mtime = se->mtime;
/* Update valid block bitmap */
if (del > 0) {
exist = f2fs_test_and_set_bit(offset, se->cur_valid_map);
#ifdef CONFIG_F2FS_CHECK_FS
mir_exist = f2fs_test_and_set_bit(offset,
se->cur_valid_map_mir);
if (unlikely(exist != mir_exist)) {
f2fs_err(sbi, "Inconsistent error when setting bitmap, blk:%u, old bit:%d",
blkaddr, exist);
f2fs_bug_on(sbi, 1);
}
#endif
if (unlikely(exist)) {
f2fs_err(sbi, "Bitmap was wrongly set, blk:%u",
blkaddr);
f2fs_bug_on(sbi, 1);
se->valid_blocks--;
del = 0;
}
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
if (!f2fs_test_and_set_bit(offset, se->discard_map))
sbi->discard_blks--;
/*
* SSR should never reuse block which is checkpointed
* or newly invalidated.
*/
if (!is_sbi_flag_set(sbi, SBI_CP_DISABLED)) {
if (!f2fs_test_and_set_bit(offset, se->ckpt_valid_map))
se->ckpt_valid_blocks++;
}
} else {
exist = f2fs_test_and_clear_bit(offset, se->cur_valid_map);
#ifdef CONFIG_F2FS_CHECK_FS
mir_exist = f2fs_test_and_clear_bit(offset,
se->cur_valid_map_mir);
if (unlikely(exist != mir_exist)) {
f2fs_err(sbi, "Inconsistent error when clearing bitmap, blk:%u, old bit:%d",
blkaddr, exist);
f2fs_bug_on(sbi, 1);
}
#endif
if (unlikely(!exist)) {
f2fs_err(sbi, "Bitmap was wrongly cleared, blk:%u",
blkaddr);
f2fs_bug_on(sbi, 1);
se->valid_blocks++;
del = 0;
} else if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
/*
* If checkpoints are off, we must not reuse data that
* was used in the previous checkpoint. If it was used
* before, we must track that to know how much space we
* really have.
*/
if (f2fs_test_bit(offset, se->ckpt_valid_map)) {
spin_lock(&sbi->stat_lock);
sbi->unusable_block_count++;
spin_unlock(&sbi->stat_lock);
}
}
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
if (f2fs_test_and_clear_bit(offset, se->discard_map))
sbi->discard_blks++;
}
if (!f2fs_test_bit(offset, se->ckpt_valid_map))
se->ckpt_valid_blocks += del;
__mark_sit_entry_dirty(sbi, segno);
/* update total number of valid blocks to be written in ckpt area */
SIT_I(sbi)->written_valid_blocks += del;
if (__is_large_section(sbi))
get_sec_entry(sbi, segno)->valid_blocks += del;
}
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 16:20:41 +00:00
void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr)
{
unsigned int segno = GET_SEGNO(sbi, addr);
struct sit_info *sit_i = SIT_I(sbi);
f2fs_bug_on(sbi, addr == NULL_ADDR);
if (addr == NEW_ADDR)
return;
f2fs: readahead encrypted block during GC During GC, for each encrypted block, we will read block synchronously into meta page, and then submit it into current cold data log area. So this block read model with 4k granularity can make poor performance, like migrating non-encrypted block, let's readahead encrypted block as well to improve migration performance. To implement this, we choose meta page that its index is old block address of the encrypted block, and readahead ciphertext into this page, later, if readaheaded page is still updated, we will load its data into target meta page, and submit the write IO. Note that for OPU, truncation, deletion, we need to invalid meta page after we invalid old block address, to make sure we won't load invalid data from target meta page during encrypted block migration. for ((i = 0; i < 1000; i++)) do { xfs_io -f /mnt/f2fs/dir/$i -c "pwrite 0 128k" -c "fsync"; } done for ((i = 0; i < 1000; i+=2)) do { rm /mnt/f2fs/dir/$i; } done ret = ioctl(fd, F2FS_IOC_GARBAGE_COLLECT, 0); Before: gc-6549 [001] d..1 214682.212797: block_rq_insert: 8,32 RA 32768 () 786400 + 64 [gc] gc-6549 [001] d..1 214682.212802: block_unplug: [gc] 1 gc-6549 [001] .... 214682.213892: block_bio_queue: 8,32 R 67494144 + 8 [gc] gc-6549 [001] .... 214682.213899: block_getrq: 8,32 R 67494144 + 8 [gc] gc-6549 [001] .... 214682.213902: block_plug: [gc] gc-6549 [001] d..1 214682.213905: block_rq_insert: 8,32 R 4096 () 67494144 + 8 [gc] gc-6549 [001] d..1 214682.213908: block_unplug: [gc] 1 gc-6549 [001] .... 214682.226405: block_bio_queue: 8,32 R 67494152 + 8 [gc] gc-6549 [001] .... 214682.226412: block_getrq: 8,32 R 67494152 + 8 [gc] gc-6549 [001] .... 214682.226414: block_plug: [gc] gc-6549 [001] d..1 214682.226417: block_rq_insert: 8,32 R 4096 () 67494152 + 8 [gc] gc-6549 [001] d..1 214682.226420: block_unplug: [gc] 1 gc-6549 [001] .... 214682.226904: block_bio_queue: 8,32 R 67494160 + 8 [gc] gc-6549 [001] .... 214682.226910: block_getrq: 8,32 R 67494160 + 8 [gc] gc-6549 [001] .... 214682.226911: block_plug: [gc] gc-6549 [001] d..1 214682.226914: block_rq_insert: 8,32 R 4096 () 67494160 + 8 [gc] gc-6549 [001] d..1 214682.226916: block_unplug: [gc] 1 After: gc-5678 [003] .... 214327.025906: block_bio_queue: 8,32 R 67493824 + 8 [gc] gc-5678 [003] .... 214327.025908: block_bio_backmerge: 8,32 R 67493824 + 8 [gc] gc-5678 [003] .... 214327.025915: block_bio_queue: 8,32 R 67493832 + 8 [gc] gc-5678 [003] .... 214327.025917: block_bio_backmerge: 8,32 R 67493832 + 8 [gc] gc-5678 [003] .... 214327.025923: block_bio_queue: 8,32 R 67493840 + 8 [gc] gc-5678 [003] .... 214327.025925: block_bio_backmerge: 8,32 R 67493840 + 8 [gc] gc-5678 [003] .... 214327.025932: block_bio_queue: 8,32 R 67493848 + 8 [gc] gc-5678 [003] .... 214327.025934: block_bio_backmerge: 8,32 R 67493848 + 8 [gc] gc-5678 [003] .... 214327.025941: block_bio_queue: 8,32 R 67493856 + 8 [gc] gc-5678 [003] .... 214327.025943: block_bio_backmerge: 8,32 R 67493856 + 8 [gc] gc-5678 [003] .... 214327.025953: block_bio_queue: 8,32 R 67493864 + 8 [gc] gc-5678 [003] .... 214327.025955: block_bio_backmerge: 8,32 R 67493864 + 8 [gc] gc-5678 [003] .... 214327.025962: block_bio_queue: 8,32 R 67493872 + 8 [gc] gc-5678 [003] .... 214327.025964: block_bio_backmerge: 8,32 R 67493872 + 8 [gc] gc-5678 [003] .... 214327.025970: block_bio_queue: 8,32 R 67493880 + 8 [gc] gc-5678 [003] .... 214327.025972: block_bio_backmerge: 8,32 R 67493880 + 8 [gc] gc-5678 [003] .... 214327.026000: block_bio_queue: 8,32 WS 34123776 + 2048 [gc] gc-5678 [003] .... 214327.026019: block_getrq: 8,32 WS 34123776 + 2048 [gc] gc-5678 [003] d..1 214327.026021: block_rq_insert: 8,32 R 131072 () 67493632 + 256 [gc] gc-5678 [003] d..1 214327.026023: block_unplug: [gc] 1 gc-5678 [003] d..1 214327.026026: block_rq_issue: 8,32 R 131072 () 67493632 + 256 [gc] gc-5678 [003] .... 214327.026046: block_plug: [gc] Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-14 14:37:25 +00:00
invalidate_mapping_pages(META_MAPPING(sbi), addr, addr);
/* add it into sit main buffer */
down_write(&sit_i->sentry_lock);
update_sit_entry(sbi, addr, -1);
/* add it into dirty seglist */
locate_dirty_segment(sbi, segno);
up_write(&sit_i->sentry_lock);
}
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 16:20:41 +00:00
bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr)
{
struct sit_info *sit_i = SIT_I(sbi);
unsigned int segno, offset;
struct seg_entry *se;
bool is_cp = false;
f2fs: introduce DATA_GENERIC_ENHANCE Previously, f2fs_is_valid_blkaddr(, blkaddr, DATA_GENERIC) will check whether @blkaddr locates in main area or not. That check is weak, since the block address in range of main area can point to the address which is not valid in segment info table, and we can not detect such condition, we may suffer worse corruption as system continues running. So this patch introduce DATA_GENERIC_ENHANCE to enhance the sanity check which trigger SIT bitmap check rather than only range check. This patch did below changes as wel: - set SBI_NEED_FSCK in f2fs_is_valid_blkaddr(). - get rid of is_valid_data_blkaddr() to avoid panic if blkaddr is invalid. - introduce verify_fio_blkaddr() to wrap fio {new,old}_blkaddr validation check. - spread blkaddr check in: * f2fs_get_node_info() * __read_out_blkaddrs() * f2fs_submit_page_read() * ra_data_block() * do_recover_data() This patch can fix bug reported from bugzilla below: https://bugzilla.kernel.org/show_bug.cgi?id=203215 https://bugzilla.kernel.org/show_bug.cgi?id=203223 https://bugzilla.kernel.org/show_bug.cgi?id=203231 https://bugzilla.kernel.org/show_bug.cgi?id=203235 https://bugzilla.kernel.org/show_bug.cgi?id=203241 = Update by Jaegeuk Kim = DATA_GENERIC_ENHANCE enhanced to validate block addresses on read/write paths. But, xfstest/generic/446 compalins some generated kernel messages saying invalid bitmap was detected when reading a block. The reaons is, when we get the block addresses from extent_cache, there is no lock to synchronize it from truncating the blocks in parallel. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-04-15 07:26:32 +00:00
if (!__is_valid_data_blkaddr(blkaddr))
return true;
down_read(&sit_i->sentry_lock);
segno = GET_SEGNO(sbi, blkaddr);
se = get_seg_entry(sbi, segno);
offset = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
if (f2fs_test_bit(offset, se->ckpt_valid_map))
is_cp = true;
up_read(&sit_i->sentry_lock);
return is_cp;
}
/*
* This function should be resided under the curseg_mutex lock
*/
static void __add_sum_entry(struct f2fs_sb_info *sbi, int type,
struct f2fs_summary *sum)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
void *addr = curseg->sum_blk;
addr += curseg->next_blkoff * sizeof(struct f2fs_summary);
memcpy(addr, sum, sizeof(struct f2fs_summary));
}
/*
* Calculate the number of current summary pages for writing
*/
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 16:20:41 +00:00
int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra)
{
int valid_sum_count = 0;
int i, sum_in_page;
for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) {
if (sbi->ckpt->alloc_type[i] == SSR)
valid_sum_count += sbi->blocks_per_seg;
else {
if (for_ra)
valid_sum_count += le16_to_cpu(
F2FS_CKPT(sbi)->cur_data_blkoff[i]);
else
valid_sum_count += curseg_blkoff(sbi, i);
}
}
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 12:29:47 +00:00
sum_in_page = (PAGE_SIZE - 2 * SUM_JOURNAL_SIZE -
SUM_FOOTER_SIZE) / SUMMARY_SIZE;
if (valid_sum_count <= sum_in_page)
return 1;
else if ((valid_sum_count - sum_in_page) <=
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 12:29:47 +00:00
(PAGE_SIZE - SUM_FOOTER_SIZE) / SUMMARY_SIZE)
return 2;
return 3;
}
/*
* Caller should put this summary 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 16:20:41 +00:00
struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno)
{
return f2fs_get_meta_page_nofail(sbi, GET_SUM_BLOCK(sbi, segno));
}
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 16:20:41 +00:00
void f2fs_update_meta_page(struct f2fs_sb_info *sbi,
void *src, block_t blk_addr)
{
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 16:20:41 +00:00
struct page *page = f2fs_grab_meta_page(sbi, blk_addr);
memcpy(page_address(page), src, PAGE_SIZE);
set_page_dirty(page);
f2fs_put_page(page, 1);
}
static void write_sum_page(struct f2fs_sb_info *sbi,
struct f2fs_summary_block *sum_blk, block_t blk_addr)
{
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 16:20:41 +00:00
f2fs_update_meta_page(sbi, (void *)sum_blk, blk_addr);
}
static void write_current_sum_page(struct f2fs_sb_info *sbi,
int type, block_t blk_addr)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
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 16:20:41 +00:00
struct page *page = f2fs_grab_meta_page(sbi, blk_addr);
struct f2fs_summary_block *src = curseg->sum_blk;
struct f2fs_summary_block *dst;
dst = (struct f2fs_summary_block *)page_address(page);
memset(dst, 0, PAGE_SIZE);
mutex_lock(&curseg->curseg_mutex);
down_read(&curseg->journal_rwsem);
memcpy(&dst->journal, curseg->journal, SUM_JOURNAL_SIZE);
up_read(&curseg->journal_rwsem);
memcpy(dst->entries, src->entries, SUM_ENTRY_SIZE);
memcpy(&dst->footer, &src->footer, SUM_FOOTER_SIZE);
mutex_unlock(&curseg->curseg_mutex);
set_page_dirty(page);
f2fs_put_page(page, 1);
}
static int is_next_segment_free(struct f2fs_sb_info *sbi, int type)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
unsigned int segno = curseg->segno + 1;
struct free_segmap_info *free_i = FREE_I(sbi);
if (segno < MAIN_SEGS(sbi) && segno % sbi->segs_per_sec)
return !test_bit(segno, free_i->free_segmap);
return 0;
}
/*
* Find a new segment from the free segments bitmap to right order
* This function should be returned with success, otherwise BUG
*/
static void get_new_segment(struct f2fs_sb_info *sbi,
unsigned int *newseg, bool new_sec, int dir)
{
struct free_segmap_info *free_i = FREE_I(sbi);
unsigned int segno, secno, zoneno;
unsigned int total_zones = MAIN_SECS(sbi) / sbi->secs_per_zone;
unsigned int hint = GET_SEC_FROM_SEG(sbi, *newseg);
unsigned int old_zoneno = GET_ZONE_FROM_SEG(sbi, *newseg);
unsigned int left_start = hint;
bool init = true;
int go_left = 0;
int i;
spin_lock(&free_i->segmap_lock);
if (!new_sec && ((*newseg + 1) % sbi->segs_per_sec)) {
segno = find_next_zero_bit(free_i->free_segmap,
GET_SEG_FROM_SEC(sbi, hint + 1), *newseg + 1);
if (segno < GET_SEG_FROM_SEC(sbi, hint + 1))
goto got_it;
}
find_other_zone:
secno = find_next_zero_bit(free_i->free_secmap, MAIN_SECS(sbi), hint);
if (secno >= MAIN_SECS(sbi)) {
if (dir == ALLOC_RIGHT) {
secno = find_next_zero_bit(free_i->free_secmap,
MAIN_SECS(sbi), 0);
f2fs_bug_on(sbi, secno >= MAIN_SECS(sbi));
} else {
go_left = 1;
left_start = hint - 1;
}
}
if (go_left == 0)
goto skip_left;
while (test_bit(left_start, free_i->free_secmap)) {
if (left_start > 0) {
left_start--;
continue;
}
left_start = find_next_zero_bit(free_i->free_secmap,
MAIN_SECS(sbi), 0);
f2fs_bug_on(sbi, left_start >= MAIN_SECS(sbi));
break;
}
secno = left_start;
skip_left:
segno = GET_SEG_FROM_SEC(sbi, secno);
zoneno = GET_ZONE_FROM_SEC(sbi, secno);
/* give up on finding another zone */
if (!init)
goto got_it;
if (sbi->secs_per_zone == 1)
goto got_it;
if (zoneno == old_zoneno)
goto got_it;
if (dir == ALLOC_LEFT) {
if (!go_left && zoneno + 1 >= total_zones)
goto got_it;
if (go_left && zoneno == 0)
goto got_it;
}
for (i = 0; i < NR_CURSEG_TYPE; i++)
if (CURSEG_I(sbi, i)->zone == zoneno)
break;
if (i < NR_CURSEG_TYPE) {
/* zone is in user, try another */
if (go_left)
hint = zoneno * sbi->secs_per_zone - 1;
else if (zoneno + 1 >= total_zones)
hint = 0;
else
hint = (zoneno + 1) * sbi->secs_per_zone;
init = false;
goto find_other_zone;
}
got_it:
/* set it as dirty segment in free segmap */
f2fs_bug_on(sbi, test_bit(segno, free_i->free_segmap));
__set_inuse(sbi, segno);
*newseg = segno;
spin_unlock(&free_i->segmap_lock);
}
static void reset_curseg(struct f2fs_sb_info *sbi, int type, int modified)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
struct summary_footer *sum_footer;
curseg->segno = curseg->next_segno;
curseg->zone = GET_ZONE_FROM_SEG(sbi, curseg->segno);
curseg->next_blkoff = 0;
curseg->next_segno = NULL_SEGNO;
sum_footer = &(curseg->sum_blk->footer);
memset(sum_footer, 0, sizeof(struct summary_footer));
if (IS_DATASEG(type))
SET_SUM_TYPE(sum_footer, SUM_TYPE_DATA);
if (IS_NODESEG(type))
SET_SUM_TYPE(sum_footer, SUM_TYPE_NODE);
__set_sit_entry_type(sbi, type, curseg->segno, modified);
}
static unsigned int __get_next_segno(struct f2fs_sb_info *sbi, int type)
{
/* if segs_per_sec is large than 1, we need to keep original policy. */
if (__is_large_section(sbi))
return CURSEG_I(sbi, type)->segno;
if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
return 0;
if (test_opt(sbi, NOHEAP) &&
(type == CURSEG_HOT_DATA || IS_NODESEG(type)))
return 0;
if (SIT_I(sbi)->last_victim[ALLOC_NEXT])
return SIT_I(sbi)->last_victim[ALLOC_NEXT];
/* find segments from 0 to reuse freed segments */
if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
return 0;
return CURSEG_I(sbi, type)->segno;
}
/*
* Allocate a current working segment.
* This function always allocates a free segment in LFS manner.
*/
static void new_curseg(struct f2fs_sb_info *sbi, int type, bool new_sec)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
unsigned int segno = curseg->segno;
int dir = ALLOC_LEFT;
write_sum_page(sbi, curseg->sum_blk,
GET_SUM_BLOCK(sbi, segno));
if (type == CURSEG_WARM_DATA || type == CURSEG_COLD_DATA)
dir = ALLOC_RIGHT;
if (test_opt(sbi, NOHEAP))
dir = ALLOC_RIGHT;
segno = __get_next_segno(sbi, type);
get_new_segment(sbi, &segno, new_sec, dir);
curseg->next_segno = segno;
reset_curseg(sbi, type, 1);
curseg->alloc_type = LFS;
}
static void __next_free_blkoff(struct f2fs_sb_info *sbi,
struct curseg_info *seg, block_t start)
{
struct seg_entry *se = get_seg_entry(sbi, seg->segno);
int entries = SIT_VBLOCK_MAP_SIZE / sizeof(unsigned long);
unsigned long *target_map = SIT_I(sbi)->tmp_map;
unsigned long *ckpt_map = (unsigned long *)se->ckpt_valid_map;
unsigned long *cur_map = (unsigned long *)se->cur_valid_map;
int i, pos;
for (i = 0; i < entries; i++)
target_map[i] = ckpt_map[i] | cur_map[i];
pos = __find_rev_next_zero_bit(target_map, sbi->blocks_per_seg, start);
seg->next_blkoff = pos;
}
/*
* If a segment is written by LFS manner, next block offset is just obtained
* by increasing the current block offset. However, if a segment is written by
* SSR manner, next block offset obtained by calling __next_free_blkoff
*/
static void __refresh_next_blkoff(struct f2fs_sb_info *sbi,
struct curseg_info *seg)
{
if (seg->alloc_type == SSR)
__next_free_blkoff(sbi, seg, seg->next_blkoff + 1);
else
seg->next_blkoff++;
}
/*
* This function always allocates a used segment(from dirty seglist) by SSR
* manner, so it should recover the existing segment information of valid blocks
*/
static void change_curseg(struct f2fs_sb_info *sbi, int type)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
struct curseg_info *curseg = CURSEG_I(sbi, type);
unsigned int new_segno = curseg->next_segno;
struct f2fs_summary_block *sum_node;
struct page *sum_page;
write_sum_page(sbi, curseg->sum_blk,
GET_SUM_BLOCK(sbi, curseg->segno));
__set_test_and_inuse(sbi, new_segno);
mutex_lock(&dirty_i->seglist_lock);
__remove_dirty_segment(sbi, new_segno, PRE);
__remove_dirty_segment(sbi, new_segno, DIRTY);
mutex_unlock(&dirty_i->seglist_lock);
reset_curseg(sbi, type, 1);
curseg->alloc_type = SSR;
__next_free_blkoff(sbi, curseg, 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 16:20:41 +00:00
sum_page = f2fs_get_sum_page(sbi, new_segno);
f2fs_bug_on(sbi, IS_ERR(sum_page));
sum_node = (struct f2fs_summary_block *)page_address(sum_page);
memcpy(curseg->sum_blk, sum_node, SUM_ENTRY_SIZE);
f2fs_put_page(sum_page, 1);
}
static int get_ssr_segment(struct f2fs_sb_info *sbi, int type)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops;
unsigned segno = NULL_SEGNO;
int i, cnt;
bool reversed = 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 16:20:41 +00:00
/* f2fs_need_SSR() already forces to do this */
if (v_ops->get_victim(sbi, &segno, BG_GC, type, SSR)) {
curseg->next_segno = segno;
return 1;
}
/* For node segments, let's do SSR more intensively */
if (IS_NODESEG(type)) {
if (type >= CURSEG_WARM_NODE) {
reversed = true;
i = CURSEG_COLD_NODE;
} else {
i = CURSEG_HOT_NODE;
}
cnt = NR_CURSEG_NODE_TYPE;
} else {
if (type >= CURSEG_WARM_DATA) {
reversed = true;
i = CURSEG_COLD_DATA;
} else {
i = CURSEG_HOT_DATA;
}
cnt = NR_CURSEG_DATA_TYPE;
}
for (; cnt-- > 0; reversed ? i-- : i++) {
if (i == type)
continue;
if (v_ops->get_victim(sbi, &segno, BG_GC, i, SSR)) {
curseg->next_segno = segno;
return 1;
}
}
/* find valid_blocks=0 in dirty list */
if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
segno = get_free_segment(sbi);
if (segno != NULL_SEGNO) {
curseg->next_segno = segno;
return 1;
}
}
return 0;
}
/*
* flush out current segment and replace it with new segment
* This function should be returned with success, otherwise BUG
*/
static void allocate_segment_by_default(struct f2fs_sb_info *sbi,
int type, bool force)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
if (force)
new_curseg(sbi, type, true);
else if (!is_set_ckpt_flags(sbi, CP_CRC_RECOVERY_FLAG) &&
type == CURSEG_WARM_NODE)
new_curseg(sbi, type, false);
else if (curseg->alloc_type == LFS && is_next_segment_free(sbi, type) &&
likely(!is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
new_curseg(sbi, type, 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 16:20:41 +00:00
else if (f2fs_need_SSR(sbi) && get_ssr_segment(sbi, type))
change_curseg(sbi, type);
else
new_curseg(sbi, type, false);
stat_inc_seg_type(sbi, curseg);
}
f2fs: ioctl for removing a range from F2FS This ioctl shrinks a given length (aligned to sections) from end of the main area. Any cursegs and valid blocks will be moved out before invalidating the range. This feature can be used for adjusting partition sizes online. History of the patch: Sahitya Tummala: - Add this ioctl for f2fs_compat_ioctl() as well. - Fix debugfs status to reflect the online resize changes. - Fix potential race between online resize path and allocate new data block path or gc path. Others: - Rename some identifiers. - Add some error handling branches. - Clear sbi->next_victim_seg[BG_GC/FG_GC] in shrinking range. - Implement this interface as ext4's, and change the parameter from shrunk bytes to new block count of F2FS. - During resizing, force to empty sit_journal and forbid adding new entries to it, in order to avoid invalid segno in journal after resize. - Reduce sbi->user_block_count before resize starts. - Commit the updated superblock first, and then update in-memory metadata only when the former succeeds. - Target block count must align to sections. - Write checkpoint before and after committing the new superblock, w/o CP_FSCK_FLAG respectively, so that the FS can be fixed by fsck even if resize fails after the new superblock is committed. - In free_segment_range(), reduce granularity of gc_mutex. - Add protection on curseg migration. - Add freeze_bdev() and thaw_bdev() for resize fs. - Remove CUR_MAIN_SECS and use MAIN_SECS directly for allocation. - Recover super_block and FS metadata when resize fails. - No need to clear CP_FSCK_FLAG in update_ckpt_flags(). - Clean up the sb and fs metadata update functions for resize_fs. Geert Uytterhoeven: - Use div_u64*() for 64-bit divisions Arnd Bergmann: - Not all architectures support get_user() with a 64-bit argument: ERROR: "__get_user_bad" [fs/f2fs/f2fs.ko] undefined! Use copy_from_user() here, this will always work. Signed-off-by: Qiuyang Sun <sunqiuyang@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Sahitya Tummala <stummala@codeaurora.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-06-05 03:33:25 +00:00
void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
unsigned int start, unsigned int end)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
unsigned int segno;
down_read(&SM_I(sbi)->curseg_lock);
mutex_lock(&curseg->curseg_mutex);
down_write(&SIT_I(sbi)->sentry_lock);
segno = CURSEG_I(sbi, type)->segno;
if (segno < start || segno > end)
goto unlock;
if (f2fs_need_SSR(sbi) && get_ssr_segment(sbi, type))
change_curseg(sbi, type);
else
new_curseg(sbi, type, true);
stat_inc_seg_type(sbi, curseg);
locate_dirty_segment(sbi, segno);
unlock:
up_write(&SIT_I(sbi)->sentry_lock);
if (segno != curseg->segno)
f2fs_notice(sbi, "For resize: curseg of type %d: %u ==> %u",
type, segno, curseg->segno);
f2fs: ioctl for removing a range from F2FS This ioctl shrinks a given length (aligned to sections) from end of the main area. Any cursegs and valid blocks will be moved out before invalidating the range. This feature can be used for adjusting partition sizes online. History of the patch: Sahitya Tummala: - Add this ioctl for f2fs_compat_ioctl() as well. - Fix debugfs status to reflect the online resize changes. - Fix potential race between online resize path and allocate new data block path or gc path. Others: - Rename some identifiers. - Add some error handling branches. - Clear sbi->next_victim_seg[BG_GC/FG_GC] in shrinking range. - Implement this interface as ext4's, and change the parameter from shrunk bytes to new block count of F2FS. - During resizing, force to empty sit_journal and forbid adding new entries to it, in order to avoid invalid segno in journal after resize. - Reduce sbi->user_block_count before resize starts. - Commit the updated superblock first, and then update in-memory metadata only when the former succeeds. - Target block count must align to sections. - Write checkpoint before and after committing the new superblock, w/o CP_FSCK_FLAG respectively, so that the FS can be fixed by fsck even if resize fails after the new superblock is committed. - In free_segment_range(), reduce granularity of gc_mutex. - Add protection on curseg migration. - Add freeze_bdev() and thaw_bdev() for resize fs. - Remove CUR_MAIN_SECS and use MAIN_SECS directly for allocation. - Recover super_block and FS metadata when resize fails. - No need to clear CP_FSCK_FLAG in update_ckpt_flags(). - Clean up the sb and fs metadata update functions for resize_fs. Geert Uytterhoeven: - Use div_u64*() for 64-bit divisions Arnd Bergmann: - Not all architectures support get_user() with a 64-bit argument: ERROR: "__get_user_bad" [fs/f2fs/f2fs.ko] undefined! Use copy_from_user() here, this will always work. Signed-off-by: Qiuyang Sun <sunqiuyang@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Sahitya Tummala <stummala@codeaurora.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-06-05 03:33:25 +00:00
mutex_unlock(&curseg->curseg_mutex);
up_read(&SM_I(sbi)->curseg_lock);
}
void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi, int type)
{
struct curseg_info *curseg;
unsigned int old_segno;
int i;
down_write(&SIT_I(sbi)->sentry_lock);
for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) {
if (type != NO_CHECK_TYPE && i != type)
continue;
curseg = CURSEG_I(sbi, i);
if (type == NO_CHECK_TYPE || curseg->next_blkoff ||
get_valid_blocks(sbi, curseg->segno, false) ||
get_ckpt_valid_blocks(sbi, curseg->segno)) {
old_segno = curseg->segno;
SIT_I(sbi)->s_ops->allocate_segment(sbi, i, true);
locate_dirty_segment(sbi, old_segno);
}
}
up_write(&SIT_I(sbi)->sentry_lock);
}
static const struct segment_allocation default_salloc_ops = {
.allocate_segment = allocate_segment_by_default,
};
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 16:20:41 +00:00
bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
struct cp_control *cpc)
{
__u64 trim_start = cpc->trim_start;
bool has_candidate = false;
down_write(&SIT_I(sbi)->sentry_lock);
for (; cpc->trim_start <= cpc->trim_end; cpc->trim_start++) {
if (add_discard_addrs(sbi, cpc, true)) {
has_candidate = true;
break;
}
}
up_write(&SIT_I(sbi)->sentry_lock);
cpc->trim_start = trim_start;
return has_candidate;
}
static unsigned int __issue_discard_cmd_range(struct f2fs_sb_info *sbi,
struct discard_policy *dpolicy,
unsigned int start, unsigned int end)
{
struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
struct discard_cmd *prev_dc = NULL, *next_dc = NULL;
struct rb_node **insert_p = NULL, *insert_parent = NULL;
struct discard_cmd *dc;
struct blk_plug plug;
int issued;
unsigned int trimmed = 0;
next:
issued = 0;
mutex_lock(&dcc->cmd_lock);
if (unlikely(dcc->rbtree_check))
f2fs_bug_on(sbi, !f2fs_check_rb_tree_consistence(sbi,
&dcc->root));
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 16:20:41 +00:00
dc = (struct discard_cmd *)f2fs_lookup_rb_tree_ret(&dcc->root,
NULL, start,
(struct rb_entry **)&prev_dc,
(struct rb_entry **)&next_dc,
&insert_p, &insert_parent, true, NULL);
if (!dc)
dc = next_dc;
blk_start_plug(&plug);
while (dc && dc->lstart <= end) {
struct rb_node *node;
int err = 0;
if (dc->len < dpolicy->granularity)
goto skip;
if (dc->state != D_PREP) {
list_move_tail(&dc->list, &dcc->fstrim_list);
goto skip;
}
err = __submit_discard_cmd(sbi, dpolicy, dc, &issued);
if (issued >= dpolicy->max_requests) {
start = dc->lstart + dc->len;
if (err)
__remove_discard_cmd(sbi, dc);
blk_finish_plug(&plug);
mutex_unlock(&dcc->cmd_lock);
trimmed += __wait_all_discard_cmd(sbi, NULL);
congestion_wait(BLK_RW_ASYNC, HZ/50);
goto next;
}
skip:
node = rb_next(&dc->rb_node);
if (err)
__remove_discard_cmd(sbi, dc);
dc = rb_entry_safe(node, struct discard_cmd, rb_node);
if (fatal_signal_pending(current))
break;
}
blk_finish_plug(&plug);
mutex_unlock(&dcc->cmd_lock);
return trimmed;
}
int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range)
{
__u64 start = F2FS_BYTES_TO_BLK(range->start);
__u64 end = start + F2FS_BYTES_TO_BLK(range->len) - 1;
unsigned int start_segno, end_segno;
block_t start_block, end_block;
struct cp_control cpc;
struct discard_policy dpolicy;
unsigned long long trimmed = 0;
int err = 0;
bool need_align = test_opt(sbi, LFS) && __is_large_section(sbi);
if (start >= MAX_BLKADDR(sbi) || range->len < sbi->blocksize)
return -EINVAL;
if (end < MAIN_BLKADDR(sbi))
goto out;
if (is_sbi_flag_set(sbi, SBI_NEED_FSCK)) {
f2fs_warn(sbi, "Found FS corruption, run fsck to fix.");
return -EFSCORRUPTED;
}
/* start/end segment number in main_area */
start_segno = (start <= MAIN_BLKADDR(sbi)) ? 0 : GET_SEGNO(sbi, start);
end_segno = (end >= MAX_BLKADDR(sbi)) ? MAIN_SEGS(sbi) - 1 :
GET_SEGNO(sbi, end);
f2fs: issue discard align to section in LFS mode For the case when sbi->segs_per_sec > 1 with lfs mode, take section:segment = 5 for example, if the section prefree_map is ...previous section | current section (1 1 0 1 1) | next section..., then the start = x, end = x + 1, after start = start_segno + sbi->segs_per_sec, start = x + 5, then it will skip x + 3 and x + 4, but their bitmap is still set, which will cause duplicated f2fs_issue_discard of this same section in the next write_checkpoint: round 1: section bitmap : 1 1 1 1 1, all valid, prefree_map: 0 0 0 0 0 then rm data block NO.2, block NO.2 becomes invalid, prefree_map: 0 0 1 0 0 write_checkpoint: section bitmap: 1 1 0 1 1, prefree_map: 0 0 0 0 0, prefree of NO.2 is cleared, and no discard issued round 2: rm data block NO.0, NO.1, NO.3, NO.4 all invalid, but prefree bit of NO.2 is set and cleared in round 1, then prefree_map: 1 1 0 1 1 write_checkpoint: section bitmap: 0 0 0 0 0, prefree_map: 0 0 0 1 1, no valid blocks of this section, so discard issued, but this time prefree bit of NO.3 and NO.4 is skipped due to start = start_segno + sbi->segs_per_sec; round 3: write_checkpoint: section bitmap: 0 0 0 0 0, prefree_map: 0 0 0 1 1 -> 0 0 0 0 0, no valid blocks of this section, so discard issued, this time prefree bit of NO.3 and NO.4 is cleared, but the discard of this section is sent again... To fix this problem, we can align the start and end value to section boundary for fstrim and real-time discard operation, and decide to issue discard only when the whole section is invalid, which can issue discard aligned to section size as much as possible and avoid redundant discard. Signed-off-by: Yunlong Song <yunlong.song@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-07-19 12:58:15 +00:00
if (need_align) {
start_segno = rounddown(start_segno, sbi->segs_per_sec);
end_segno = roundup(end_segno + 1, sbi->segs_per_sec) - 1;
}
cpc.reason = CP_DISCARD;
cpc.trim_minlen = max_t(__u64, 1, F2FS_BYTES_TO_BLK(range->minlen));
cpc.trim_start = start_segno;
cpc.trim_end = end_segno;
if (sbi->discard_blks == 0)
goto out;
mutex_lock(&sbi->gc_mutex);
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 16:20:41 +00:00
err = f2fs_write_checkpoint(sbi, &cpc);
mutex_unlock(&sbi->gc_mutex);
if (err)
goto out;
/*
* We filed discard candidates, but actually we don't need to wait for
* all of them, since they'll be issued in idle time along with runtime
* discard option. User configuration looks like using runtime discard
* or periodic fstrim instead of it.
*/
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
if (f2fs_realtime_discard_enable(sbi))
goto out;
start_block = START_BLOCK(sbi, start_segno);
end_block = START_BLOCK(sbi, end_segno + 1);
__init_discard_policy(sbi, &dpolicy, DPOLICY_FSTRIM, cpc.trim_minlen);
trimmed = __issue_discard_cmd_range(sbi, &dpolicy,
start_block, end_block);
trimmed += __wait_discard_cmd_range(sbi, &dpolicy,
start_block, end_block);
out:
if (!err)
range->len = F2FS_BLK_TO_BYTES(trimmed);
return err;
}
static bool __has_curseg_space(struct f2fs_sb_info *sbi, int type)
{
struct curseg_info *curseg = CURSEG_I(sbi, type);
if (curseg->next_blkoff < sbi->blocks_per_seg)
return true;
return 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 16:20:41 +00:00
int f2fs_rw_hint_to_seg_type(enum rw_hint hint)
{
switch (hint) {
case WRITE_LIFE_SHORT:
return CURSEG_HOT_DATA;
case WRITE_LIFE_EXTREME:
return CURSEG_COLD_DATA;
default:
return CURSEG_WARM_DATA;
}
}
/* This returns write hints for each segment type. This hints will be
* passed down to block layer. There are mapping tables which depend on
* the mount option 'whint_mode'.
*
* 1) whint_mode=off. F2FS only passes down WRITE_LIFE_NOT_SET.
*
* 2) whint_mode=user-based. F2FS tries to pass down hints given by users.
*
* User F2FS Block
* ---- ---- -----
* META WRITE_LIFE_NOT_SET
* HOT_NODE "
* WARM_NODE "
* COLD_NODE "
* ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME
* extension list " "
*
* -- buffered io
* WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME
* WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT
* WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET
* WRITE_LIFE_NONE " "
* WRITE_LIFE_MEDIUM " "
* WRITE_LIFE_LONG " "
*
* -- direct io
* WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME
* WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT
* WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET
* WRITE_LIFE_NONE " WRITE_LIFE_NONE
* WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM
* WRITE_LIFE_LONG " WRITE_LIFE_LONG
*
* 3) whint_mode=fs-based. F2FS passes down hints with its policy.
*
* User F2FS Block
* ---- ---- -----
* META WRITE_LIFE_MEDIUM;
* HOT_NODE WRITE_LIFE_NOT_SET
* WARM_NODE "
* COLD_NODE WRITE_LIFE_NONE
* ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME
* extension list " "
*
* -- buffered io
* WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME
* WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT
* WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_LONG
* WRITE_LIFE_NONE " "
* WRITE_LIFE_MEDIUM " "
* WRITE_LIFE_LONG " "
*
* -- direct io
* WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME
* WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT
* WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET
* WRITE_LIFE_NONE " WRITE_LIFE_NONE
* WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM
* WRITE_LIFE_LONG " WRITE_LIFE_LONG
*/
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 16:20:41 +00:00
enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
enum page_type type, enum temp_type temp)
{
if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER) {
if (type == DATA) {
if (temp == WARM)
return WRITE_LIFE_NOT_SET;
else if (temp == HOT)
return WRITE_LIFE_SHORT;
else if (temp == COLD)
return WRITE_LIFE_EXTREME;
} else {
return WRITE_LIFE_NOT_SET;
}
} else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS) {
if (type == DATA) {
if (temp == WARM)
return WRITE_LIFE_LONG;
else if (temp == HOT)
return WRITE_LIFE_SHORT;
else if (temp == COLD)
return WRITE_LIFE_EXTREME;
} else if (type == NODE) {
if (temp == WARM || temp == HOT)
return WRITE_LIFE_NOT_SET;
else if (temp == COLD)
return WRITE_LIFE_NONE;
} else if (type == META) {
return WRITE_LIFE_MEDIUM;
}
}
return WRITE_LIFE_NOT_SET;
}
static int __get_segment_type_2(struct f2fs_io_info *fio)
{
if (fio->type == DATA)
return CURSEG_HOT_DATA;
else
return CURSEG_HOT_NODE;
}
static int __get_segment_type_4(struct f2fs_io_info *fio)
{
if (fio->type == DATA) {
struct inode *inode = fio->page->mapping->host;
if (S_ISDIR(inode->i_mode))
return CURSEG_HOT_DATA;
else
return CURSEG_COLD_DATA;
} else {
if (IS_DNODE(fio->page) && is_cold_node(fio->page))
return CURSEG_WARM_NODE;
else
return CURSEG_COLD_NODE;
}
}
static int __get_segment_type_6(struct f2fs_io_info *fio)
{
if (fio->type == DATA) {
struct inode *inode = fio->page->mapping->host;
if (is_cold_data(fio->page) || file_is_cold(inode))
return CURSEG_COLD_DATA;
if (file_is_hot(inode) ||
is_inode_flag_set(inode, FI_HOT_DATA) ||
f2fs_is_atomic_file(inode) ||
f2fs_is_volatile_file(inode))
return CURSEG_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 16:20:41 +00:00
return f2fs_rw_hint_to_seg_type(inode->i_write_hint);
} else {
if (IS_DNODE(fio->page))
return is_cold_node(fio->page) ? CURSEG_WARM_NODE :
CURSEG_HOT_NODE;
return CURSEG_COLD_NODE;
}
}
static int __get_segment_type(struct f2fs_io_info *fio)
{
int type = 0;
switch (F2FS_OPTION(fio->sbi).active_logs) {
case 2:
type = __get_segment_type_2(fio);
break;
case 4:
type = __get_segment_type_4(fio);
break;
case 6:
type = __get_segment_type_6(fio);
break;
default:
f2fs_bug_on(fio->sbi, true);
}
if (IS_HOT(type))
fio->temp = HOT;
else if (IS_WARM(type))
fio->temp = WARM;
else
fio->temp = COLD;
return type;
}
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 16:20:41 +00:00
void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
block_t old_blkaddr, block_t *new_blkaddr,
struct f2fs_summary *sum, int type,
struct f2fs_io_info *fio, bool add_list)
{
struct sit_info *sit_i = SIT_I(sbi);
struct curseg_info *curseg = CURSEG_I(sbi, type);
bool put_pin_sem = false;
if (type == CURSEG_COLD_DATA) {
/* GC during CURSEG_COLD_DATA_PINNED allocation */
if (down_read_trylock(&sbi->pin_sem)) {
put_pin_sem = true;
} else {
type = CURSEG_WARM_DATA;
curseg = CURSEG_I(sbi, type);
}
} else if (type == CURSEG_COLD_DATA_PINNED) {
type = CURSEG_COLD_DATA;
}
down_read(&SM_I(sbi)->curseg_lock);
mutex_lock(&curseg->curseg_mutex);
down_write(&sit_i->sentry_lock);
*new_blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
f2fs_wait_discard_bio(sbi, *new_blkaddr);
/*
* __add_sum_entry should be resided under the curseg_mutex
* because, this function updates a summary entry in the
* current summary block.
*/
__add_sum_entry(sbi, type, sum);
__refresh_next_blkoff(sbi, curseg);
stat_inc_block_count(sbi, curseg);
/*
* SIT information should be updated before segment allocation,
* since SSR needs latest valid block information.
*/
update_sit_entry(sbi, *new_blkaddr, 1);
if (GET_SEGNO(sbi, old_blkaddr) != NULL_SEGNO)
update_sit_entry(sbi, old_blkaddr, -1);
if (!__has_curseg_space(sbi, type))
sit_i->s_ops->allocate_segment(sbi, type, false);
/*
* segment dirty status should be updated after segment allocation,
* so we just need to update status only one time after previous
* segment being closed.
*/
locate_dirty_segment(sbi, GET_SEGNO(sbi, old_blkaddr));
locate_dirty_segment(sbi, GET_SEGNO(sbi, *new_blkaddr));
up_write(&sit_i->sentry_lock);
if (page && IS_NODESEG(type)) {
fill_node_footer_blkaddr(page, NEXT_FREE_BLKADDR(sbi, curseg));
f2fs_inode_chksum_set(sbi, page);
}
if (F2FS_IO_ALIGNED(sbi))
fio->retry = false;
if (add_list) {
struct f2fs_bio_info *io;
INIT_LIST_HEAD(&fio->list);
fio->in_list = true;
io = sbi->write_io[fio->type] + fio->temp;
spin_lock(&io->io_lock);
list_add_tail(&fio->list, &io->io_list);
spin_unlock(&io->io_lock);
}
mutex_unlock(&curseg->curseg_mutex);
up_read(&SM_I(sbi)->curseg_lock);
if (put_pin_sem)
up_read(&sbi->pin_sem);
}
static void update_device_state(struct f2fs_io_info *fio)
{
struct f2fs_sb_info *sbi = fio->sbi;
unsigned int devidx;
if (!f2fs_is_multi_device(sbi))
return;
devidx = f2fs_target_device_index(sbi, fio->new_blkaddr);
/* update device state for fsync */
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 16:20:41 +00:00
f2fs_set_dirty_device(sbi, fio->ino, devidx, FLUSH_INO);
/* update device state for checkpoint */
if (!f2fs_test_bit(devidx, (char *)&sbi->dirty_device)) {
spin_lock(&sbi->dev_lock);
f2fs_set_bit(devidx, (char *)&sbi->dirty_device);
spin_unlock(&sbi->dev_lock);
}
}
static void do_write_page(struct f2fs_summary *sum, struct f2fs_io_info *fio)
{
int type = __get_segment_type(fio);
bool keep_order = (test_opt(fio->sbi, LFS) && type == CURSEG_COLD_DATA);
if (keep_order)
down_read(&fio->sbi->io_order_lock);
reallocate:
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 16:20:41 +00:00
f2fs_allocate_data_block(fio->sbi, fio->page, fio->old_blkaddr,
&fio->new_blkaddr, sum, type, fio, true);
f2fs: readahead encrypted block during GC During GC, for each encrypted block, we will read block synchronously into meta page, and then submit it into current cold data log area. So this block read model with 4k granularity can make poor performance, like migrating non-encrypted block, let's readahead encrypted block as well to improve migration performance. To implement this, we choose meta page that its index is old block address of the encrypted block, and readahead ciphertext into this page, later, if readaheaded page is still updated, we will load its data into target meta page, and submit the write IO. Note that for OPU, truncation, deletion, we need to invalid meta page after we invalid old block address, to make sure we won't load invalid data from target meta page during encrypted block migration. for ((i = 0; i < 1000; i++)) do { xfs_io -f /mnt/f2fs/dir/$i -c "pwrite 0 128k" -c "fsync"; } done for ((i = 0; i < 1000; i+=2)) do { rm /mnt/f2fs/dir/$i; } done ret = ioctl(fd, F2FS_IOC_GARBAGE_COLLECT, 0); Before: gc-6549 [001] d..1 214682.212797: block_rq_insert: 8,32 RA 32768 () 786400 + 64 [gc] gc-6549 [001] d..1 214682.212802: block_unplug: [gc] 1 gc-6549 [001] .... 214682.213892: block_bio_queue: 8,32 R 67494144 + 8 [gc] gc-6549 [001] .... 214682.213899: block_getrq: 8,32 R 67494144 + 8 [gc] gc-6549 [001] .... 214682.213902: block_plug: [gc] gc-6549 [001] d..1 214682.213905: block_rq_insert: 8,32 R 4096 () 67494144 + 8 [gc] gc-6549 [001] d..1 214682.213908: block_unplug: [gc] 1 gc-6549 [001] .... 214682.226405: block_bio_queue: 8,32 R 67494152 + 8 [gc] gc-6549 [001] .... 214682.226412: block_getrq: 8,32 R 67494152 + 8 [gc] gc-6549 [001] .... 214682.226414: block_plug: [gc] gc-6549 [001] d..1 214682.226417: block_rq_insert: 8,32 R 4096 () 67494152 + 8 [gc] gc-6549 [001] d..1 214682.226420: block_unplug: [gc] 1 gc-6549 [001] .... 214682.226904: block_bio_queue: 8,32 R 67494160 + 8 [gc] gc-6549 [001] .... 214682.226910: block_getrq: 8,32 R 67494160 + 8 [gc] gc-6549 [001] .... 214682.226911: block_plug: [gc] gc-6549 [001] d..1 214682.226914: block_rq_insert: 8,32 R 4096 () 67494160 + 8 [gc] gc-6549 [001] d..1 214682.226916: block_unplug: [gc] 1 After: gc-5678 [003] .... 214327.025906: block_bio_queue: 8,32 R 67493824 + 8 [gc] gc-5678 [003] .... 214327.025908: block_bio_backmerge: 8,32 R 67493824 + 8 [gc] gc-5678 [003] .... 214327.025915: block_bio_queue: 8,32 R 67493832 + 8 [gc] gc-5678 [003] .... 214327.025917: block_bio_backmerge: 8,32 R 67493832 + 8 [gc] gc-5678 [003] .... 214327.025923: block_bio_queue: 8,32 R 67493840 + 8 [gc] gc-5678 [003] .... 214327.025925: block_bio_backmerge: 8,32 R 67493840 + 8 [gc] gc-5678 [003] .... 214327.025932: block_bio_queue: 8,32 R 67493848 + 8 [gc] gc-5678 [003] .... 214327.025934: block_bio_backmerge: 8,32 R 67493848 + 8 [gc] gc-5678 [003] .... 214327.025941: block_bio_queue: 8,32 R 67493856 + 8 [gc] gc-5678 [003] .... 214327.025943: block_bio_backmerge: 8,32 R 67493856 + 8 [gc] gc-5678 [003] .... 214327.025953: block_bio_queue: 8,32 R 67493864 + 8 [gc] gc-5678 [003] .... 214327.025955: block_bio_backmerge: 8,32 R 67493864 + 8 [gc] gc-5678 [003] .... 214327.025962: block_bio_queue: 8,32 R 67493872 + 8 [gc] gc-5678 [003] .... 214327.025964: block_bio_backmerge: 8,32 R 67493872 + 8 [gc] gc-5678 [003] .... 214327.025970: block_bio_queue: 8,32 R 67493880 + 8 [gc] gc-5678 [003] .... 214327.025972: block_bio_backmerge: 8,32 R 67493880 + 8 [gc] gc-5678 [003] .... 214327.026000: block_bio_queue: 8,32 WS 34123776 + 2048 [gc] gc-5678 [003] .... 214327.026019: block_getrq: 8,32 WS 34123776 + 2048 [gc] gc-5678 [003] d..1 214327.026021: block_rq_insert: 8,32 R 131072 () 67493632 + 256 [gc] gc-5678 [003] d..1 214327.026023: block_unplug: [gc] 1 gc-5678 [003] d..1 214327.026026: block_rq_issue: 8,32 R 131072 () 67493632 + 256 [gc] gc-5678 [003] .... 214327.026046: block_plug: [gc] Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-14 14:37:25 +00:00
if (GET_SEGNO(fio->sbi, fio->old_blkaddr) != NULL_SEGNO)
invalidate_mapping_pages(META_MAPPING(fio->sbi),
fio->old_blkaddr, fio->old_blkaddr);
/* writeout dirty page into bdev */
f2fs: fix to let caller retry allocating block address Configure io_bits with 2 and enable LFS mode, generic/013 reports below dmesg: BUG: unable to handle kernel NULL pointer dereference at 00000104 *pdpt = 0000000029b7b001 *pde = 0000000000000000 Oops: 0002 [#1] PREEMPT SMP Modules linked in: crc32_generic zram f2fs(O) rfcomm bnep bluetooth ecdh_generic snd_intel8x0 snd_ac97_codec ac97_bus snd_pcm snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq pcbc joydev snd_seq_device aesni_intel snd_timer aes_i586 snd crypto_simd cryptd soundcore i2c_piix4 serio_raw mac_hid video parport_pc ppdev lp parport hid_generic psmouse usbhid hid e1000 CPU: 0 PID: 11161 Comm: fsstress Tainted: G O 4.17.0-rc2 #38 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 EIP: f2fs_submit_page_write+0x28d/0x550 [f2fs] EFLAGS: 00010206 CPU: 0 EAX: e863dcd8 EBX: 00000000 ECX: 00000100 EDX: 00000200 ESI: e863dcf4 EDI: f6f82768 EBP: e863dbb0 ESP: e863db74 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 CR0: 80050033 CR2: 00000104 CR3: 29a62020 CR4: 000406f0 Call Trace: do_write_page+0x6f/0xc0 [f2fs] write_data_page+0x4a/0xd0 [f2fs] do_write_data_page+0x327/0x630 [f2fs] __write_data_page+0x34b/0x820 [f2fs] __f2fs_write_data_pages+0x42d/0x8c0 [f2fs] f2fs_write_data_pages+0x27/0x30 [f2fs] do_writepages+0x1a/0x70 __filemap_fdatawrite_range+0x94/0xd0 filemap_write_and_wait_range+0x3d/0xa0 __generic_file_write_iter+0x11a/0x1f0 f2fs_file_write_iter+0xdd/0x3b0 [f2fs] __vfs_write+0xd2/0x150 vfs_write+0x9b/0x190 ksys_write+0x45/0x90 sys_write+0x16/0x20 do_fast_syscall_32+0xaa/0x22c entry_SYSENTER_32+0x4c/0x7b EIP: 0xb7fc8c51 EFLAGS: 00000246 CPU: 0 EAX: ffffffda EBX: 00000003 ECX: 09cde000 EDX: 00001000 ESI: 00000003 EDI: 00001000 EBP: 00000000 ESP: bfbded38 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b Code: e8 f9 77 34 c9 8b 45 e0 8b 80 b8 00 00 00 39 45 d8 0f 84 bb 02 00 00 8b 45 e0 8b 80 b8 00 00 00 8d 50 d8 8b 08 89 55 f0 8b 50 04 <89> 51 04 89 0a c7 00 00 01 00 00 c7 40 04 00 02 00 00 8b 45 dc EIP: f2fs_submit_page_write+0x28d/0x550 [f2fs] SS:ESP: 0068:e863db74 CR2: 0000000000000104 ---[ end trace 4cac79c0d1305ee6 ]--- allocate_data_block will submit all sequential pending IOs sorted by a FIFO list, If we failed to submit other user's IO due to unaligned write, we will retry to allocate new block address for current IO, then it will initialize fio.list again, if fio was in the list before, it can break FIFO list, result in above panic. Thread A Thread B - do_write_page - allocate_data_block - list_add_tail : fioA cached in FIFO list. - do_write_page - allocate_data_block - list_add_tail : fioB cached in FIFO list. - f2fs_submit_page_write : fail to submit IO - allocate_data_block - INIT_LIST_HEAD - f2fs_submit_page_write - list_del <-- NULL pointer dereference This patch adds fio.retry parameter to indicate failure status for each IO, and avoid bailing out if there is still pending IO in FIFO list for fixing. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-28 15:47:18 +00:00
f2fs_submit_page_write(fio);
if (fio->retry) {
fio->old_blkaddr = fio->new_blkaddr;
goto reallocate;
}
f2fs: fix to let caller retry allocating block address Configure io_bits with 2 and enable LFS mode, generic/013 reports below dmesg: BUG: unable to handle kernel NULL pointer dereference at 00000104 *pdpt = 0000000029b7b001 *pde = 0000000000000000 Oops: 0002 [#1] PREEMPT SMP Modules linked in: crc32_generic zram f2fs(O) rfcomm bnep bluetooth ecdh_generic snd_intel8x0 snd_ac97_codec ac97_bus snd_pcm snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq pcbc joydev snd_seq_device aesni_intel snd_timer aes_i586 snd crypto_simd cryptd soundcore i2c_piix4 serio_raw mac_hid video parport_pc ppdev lp parport hid_generic psmouse usbhid hid e1000 CPU: 0 PID: 11161 Comm: fsstress Tainted: G O 4.17.0-rc2 #38 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 EIP: f2fs_submit_page_write+0x28d/0x550 [f2fs] EFLAGS: 00010206 CPU: 0 EAX: e863dcd8 EBX: 00000000 ECX: 00000100 EDX: 00000200 ESI: e863dcf4 EDI: f6f82768 EBP: e863dbb0 ESP: e863db74 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 CR0: 80050033 CR2: 00000104 CR3: 29a62020 CR4: 000406f0 Call Trace: do_write_page+0x6f/0xc0 [f2fs] write_data_page+0x4a/0xd0 [f2fs] do_write_data_page+0x327/0x630 [f2fs] __write_data_page+0x34b/0x820 [f2fs] __f2fs_write_data_pages+0x42d/0x8c0 [f2fs] f2fs_write_data_pages+0x27/0x30 [f2fs] do_writepages+0x1a/0x70 __filemap_fdatawrite_range+0x94/0xd0 filemap_write_and_wait_range+0x3d/0xa0 __generic_file_write_iter+0x11a/0x1f0 f2fs_file_write_iter+0xdd/0x3b0 [f2fs] __vfs_write+0xd2/0x150 vfs_write+0x9b/0x190 ksys_write+0x45/0x90 sys_write+0x16/0x20 do_fast_syscall_32+0xaa/0x22c entry_SYSENTER_32+0x4c/0x7b EIP: 0xb7fc8c51 EFLAGS: 00000246 CPU: 0 EAX: ffffffda EBX: 00000003 ECX: 09cde000 EDX: 00001000 ESI: 00000003 EDI: 00001000 EBP: 00000000 ESP: bfbded38 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b Code: e8 f9 77 34 c9 8b 45 e0 8b 80 b8 00 00 00 39 45 d8 0f 84 bb 02 00 00 8b 45 e0 8b 80 b8 00 00 00 8d 50 d8 8b 08 89 55 f0 8b 50 04 <89> 51 04 89 0a c7 00 00 01 00 00 c7 40 04 00 02 00 00 8b 45 dc EIP: f2fs_submit_page_write+0x28d/0x550 [f2fs] SS:ESP: 0068:e863db74 CR2: 0000000000000104 ---[ end trace 4cac79c0d1305ee6 ]--- allocate_data_block will submit all sequential pending IOs sorted by a FIFO list, If we failed to submit other user's IO due to unaligned write, we will retry to allocate new block address for current IO, then it will initialize fio.list again, if fio was in the list before, it can break FIFO list, result in above panic. Thread A Thread B - do_write_page - allocate_data_block - list_add_tail : fioA cached in FIFO list. - do_write_page - allocate_data_block - list_add_tail : fioB cached in FIFO list. - f2fs_submit_page_write : fail to submit IO - allocate_data_block - INIT_LIST_HEAD - f2fs_submit_page_write - list_del <-- NULL pointer dereference This patch adds fio.retry parameter to indicate failure status for each IO, and avoid bailing out if there is still pending IO in FIFO list for fixing. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-05-28 15:47:18 +00:00
update_device_state(fio);
if (keep_order)
up_read(&fio->sbi->io_order_lock);
}
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 16:20:41 +00:00
void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
enum iostat_type io_type)
{
struct f2fs_io_info fio = {
.sbi = sbi,
.type = META,
.temp = HOT,
.op = REQ_OP_WRITE,
.op_flags = REQ_SYNC | REQ_META | REQ_PRIO,
.old_blkaddr = page->index,
.new_blkaddr = page->index,
.page = page,
.encrypted_page = NULL,
.in_list = false,
};
if (unlikely(page->index >= MAIN_BLKADDR(sbi)))
fio.op_flags &= ~REQ_META;
set_page_writeback(page);
ClearPageError(page);
f2fs_submit_page_write(&fio);
stat_inc_meta_count(sbi, page->index);
f2fs_update_iostat(sbi, io_type, F2FS_BLKSIZE);
}
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 16:20:41 +00:00
void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio)
{
struct f2fs_summary sum;
set_summary(&sum, nid, 0, 0);
do_write_page(&sum, fio);
f2fs_update_iostat(fio->sbi, fio->io_type, F2FS_BLKSIZE);
}
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 16:20:41 +00:00
void f2fs_outplace_write_data(struct dnode_of_data *dn,
struct f2fs_io_info *fio)
{
struct f2fs_sb_info *sbi = fio->sbi;
struct f2fs_summary sum;
f2fs_bug_on(sbi, dn->data_blkaddr == NULL_ADDR);
set_summary(&sum, dn->nid, dn->ofs_in_node, fio->version);
do_write_page(&sum, fio);
f2fs_update_data_blkaddr(dn, fio->new_blkaddr);
f2fs_update_iostat(sbi, fio->io_type, F2FS_BLKSIZE);
}
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 16:20:41 +00:00
int f2fs_inplace_write_data(struct f2fs_io_info *fio)
{
int err;
struct f2fs_sb_info *sbi = fio->sbi;
unsigned int segno;
fio->new_blkaddr = fio->old_blkaddr;
/* i/o temperature is needed for passing down write hints */
__get_segment_type(fio);
segno = GET_SEGNO(sbi, fio->new_blkaddr);
if (!IS_DATASEG(get_seg_entry(sbi, segno)->type)) {
set_sbi_flag(sbi, SBI_NEED_FSCK);
f2fs_warn(sbi, "%s: incorrect segment(%u) type, run fsck to fix.",
__func__, segno);
return -EFSCORRUPTED;
}
stat_inc_inplace_blocks(fio->sbi);
f2fs: add bio cache for IPU SQLite in Wal mode may trigger sequential IPU write in db-wal file, after commit d1b3e72d5490 ("f2fs: submit bio of in-place-update pages"), we lost the chance of merging page in inner managed bio cache, result in submitting more small-sized IO. So let's add temporary bio in writepages() to cache mergeable write IO as much as possible. Test case: 1. xfs_io -f /mnt/f2fs/file -c "pwrite 0 65536" -c "fsync" 2. xfs_io -f /mnt/f2fs/file -c "pwrite 0 65536" -c "fsync" Before: f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65544, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65552, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65560, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65568, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65576, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65584, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65592, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65600, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65608, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65616, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65624, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65632, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65640, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65648, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65656, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65664, size = 4096 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), NODE, sector = 57352, size = 4096 After: f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), DATA, sector = 65544, size = 65536 f2fs_submit_write_bio: dev = (251,0)/(251,0), rw = WRITE(S), NODE, sector = 57368, size = 4096 Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-02-19 08:15:29 +00:00
if (fio->bio)
err = f2fs_merge_page_bio(fio);
else
err = f2fs_submit_page_bio(fio);
if (!err) {
update_device_state(fio);
f2fs_update_iostat(fio->sbi, fio->io_type, F2FS_BLKSIZE);
}
return err;
}
static inline int __f2fs_get_curseg(struct f2fs_sb_info *sbi,
unsigned int segno)
{
int i;
for (i = CURSEG_HOT_DATA; i < NO_CHECK_TYPE; i++) {
if (CURSEG_I(sbi, i)->segno == segno)
break;
}
return i;
}
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 16:20:41 +00:00
void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
block_t old_blkaddr, block_t new_blkaddr,
bool recover_curseg, bool recover_newaddr)
{
struct sit_info *sit_i = SIT_I(sbi);
struct curseg_info *curseg;
unsigned int segno, old_cursegno;
struct seg_entry *se;
int type;
unsigned short old_blkoff;
segno = GET_SEGNO(sbi, new_blkaddr);
se = get_seg_entry(sbi, segno);
type = se->type;
down_write(&SM_I(sbi)->curseg_lock);
if (!recover_curseg) {
/* for recovery flow */
if (se->valid_blocks == 0 && !IS_CURSEG(sbi, segno)) {
if (old_blkaddr == NULL_ADDR)
type = CURSEG_COLD_DATA;
else
type = CURSEG_WARM_DATA;
}
} else {
if (IS_CURSEG(sbi, segno)) {
/* se->type is volatile as SSR allocation */
type = __f2fs_get_curseg(sbi, segno);
f2fs_bug_on(sbi, type == NO_CHECK_TYPE);
} else {
type = CURSEG_WARM_DATA;
}
}
f2fs_bug_on(sbi, !IS_DATASEG(type));
curseg = CURSEG_I(sbi, type);
mutex_lock(&curseg->curseg_mutex);
down_write(&sit_i->sentry_lock);
old_cursegno = curseg->segno;
old_blkoff = curseg->next_blkoff;
/* change the current segment */
if (segno != curseg->segno) {
curseg->next_segno = segno;
change_curseg(sbi, type);
}
curseg->next_blkoff = GET_BLKOFF_FROM_SEG0(sbi, new_blkaddr);
__add_sum_entry(sbi, type, sum);
if (!recover_curseg || recover_newaddr)
update_sit_entry(sbi, new_blkaddr, 1);
f2fs: readahead encrypted block during GC During GC, for each encrypted block, we will read block synchronously into meta page, and then submit it into current cold data log area. So this block read model with 4k granularity can make poor performance, like migrating non-encrypted block, let's readahead encrypted block as well to improve migration performance. To implement this, we choose meta page that its index is old block address of the encrypted block, and readahead ciphertext into this page, later, if readaheaded page is still updated, we will load its data into target meta page, and submit the write IO. Note that for OPU, truncation, deletion, we need to invalid meta page after we invalid old block address, to make sure we won't load invalid data from target meta page during encrypted block migration. for ((i = 0; i < 1000; i++)) do { xfs_io -f /mnt/f2fs/dir/$i -c "pwrite 0 128k" -c "fsync"; } done for ((i = 0; i < 1000; i+=2)) do { rm /mnt/f2fs/dir/$i; } done ret = ioctl(fd, F2FS_IOC_GARBAGE_COLLECT, 0); Before: gc-6549 [001] d..1 214682.212797: block_rq_insert: 8,32 RA 32768 () 786400 + 64 [gc] gc-6549 [001] d..1 214682.212802: block_unplug: [gc] 1 gc-6549 [001] .... 214682.213892: block_bio_queue: 8,32 R 67494144 + 8 [gc] gc-6549 [001] .... 214682.213899: block_getrq: 8,32 R 67494144 + 8 [gc] gc-6549 [001] .... 214682.213902: block_plug: [gc] gc-6549 [001] d..1 214682.213905: block_rq_insert: 8,32 R 4096 () 67494144 + 8 [gc] gc-6549 [001] d..1 214682.213908: block_unplug: [gc] 1 gc-6549 [001] .... 214682.226405: block_bio_queue: 8,32 R 67494152 + 8 [gc] gc-6549 [001] .... 214682.226412: block_getrq: 8,32 R 67494152 + 8 [gc] gc-6549 [001] .... 214682.226414: block_plug: [gc] gc-6549 [001] d..1 214682.226417: block_rq_insert: 8,32 R 4096 () 67494152 + 8 [gc] gc-6549 [001] d..1 214682.226420: block_unplug: [gc] 1 gc-6549 [001] .... 214682.226904: block_bio_queue: 8,32 R 67494160 + 8 [gc] gc-6549 [001] .... 214682.226910: block_getrq: 8,32 R 67494160 + 8 [gc] gc-6549 [001] .... 214682.226911: block_plug: [gc] gc-6549 [001] d..1 214682.226914: block_rq_insert: 8,32 R 4096 () 67494160 + 8 [gc] gc-6549 [001] d..1 214682.226916: block_unplug: [gc] 1 After: gc-5678 [003] .... 214327.025906: block_bio_queue: 8,32 R 67493824 + 8 [gc] gc-5678 [003] .... 214327.025908: block_bio_backmerge: 8,32 R 67493824 + 8 [gc] gc-5678 [003] .... 214327.025915: block_bio_queue: 8,32 R 67493832 + 8 [gc] gc-5678 [003] .... 214327.025917: block_bio_backmerge: 8,32 R 67493832 + 8 [gc] gc-5678 [003] .... 214327.025923: block_bio_queue: 8,32 R 67493840 + 8 [gc] gc-5678 [003] .... 214327.025925: block_bio_backmerge: 8,32 R 67493840 + 8 [gc] gc-5678 [003] .... 214327.025932: block_bio_queue: 8,32 R 67493848 + 8 [gc] gc-5678 [003] .... 214327.025934: block_bio_backmerge: 8,32 R 67493848 + 8 [gc] gc-5678 [003] .... 214327.025941: block_bio_queue: 8,32 R 67493856 + 8 [gc] gc-5678 [003] .... 214327.025943: block_bio_backmerge: 8,32 R 67493856 + 8 [gc] gc-5678 [003] .... 214327.025953: block_bio_queue: 8,32 R 67493864 + 8 [gc] gc-5678 [003] .... 214327.025955: block_bio_backmerge: 8,32 R 67493864 + 8 [gc] gc-5678 [003] .... 214327.025962: block_bio_queue: 8,32 R 67493872 + 8 [gc] gc-5678 [003] .... 214327.025964: block_bio_backmerge: 8,32 R 67493872 + 8 [gc] gc-5678 [003] .... 214327.025970: block_bio_queue: 8,32 R 67493880 + 8 [gc] gc-5678 [003] .... 214327.025972: block_bio_backmerge: 8,32 R 67493880 + 8 [gc] gc-5678 [003] .... 214327.026000: block_bio_queue: 8,32 WS 34123776 + 2048 [gc] gc-5678 [003] .... 214327.026019: block_getrq: 8,32 WS 34123776 + 2048 [gc] gc-5678 [003] d..1 214327.026021: block_rq_insert: 8,32 R 131072 () 67493632 + 256 [gc] gc-5678 [003] d..1 214327.026023: block_unplug: [gc] 1 gc-5678 [003] d..1 214327.026026: block_rq_issue: 8,32 R 131072 () 67493632 + 256 [gc] gc-5678 [003] .... 214327.026046: block_plug: [gc] Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-14 14:37:25 +00:00
if (GET_SEGNO(sbi, old_blkaddr) != NULL_SEGNO) {
invalidate_mapping_pages(META_MAPPING(sbi),
old_blkaddr, old_blkaddr);
update_sit_entry(sbi, old_blkaddr, -1);
f2fs: readahead encrypted block during GC During GC, for each encrypted block, we will read block synchronously into meta page, and then submit it into current cold data log area. So this block read model with 4k granularity can make poor performance, like migrating non-encrypted block, let's readahead encrypted block as well to improve migration performance. To implement this, we choose meta page that its index is old block address of the encrypted block, and readahead ciphertext into this page, later, if readaheaded page is still updated, we will load its data into target meta page, and submit the write IO. Note that for OPU, truncation, deletion, we need to invalid meta page after we invalid old block address, to make sure we won't load invalid data from target meta page during encrypted block migration. for ((i = 0; i < 1000; i++)) do { xfs_io -f /mnt/f2fs/dir/$i -c "pwrite 0 128k" -c "fsync"; } done for ((i = 0; i < 1000; i+=2)) do { rm /mnt/f2fs/dir/$i; } done ret = ioctl(fd, F2FS_IOC_GARBAGE_COLLECT, 0); Before: gc-6549 [001] d..1 214682.212797: block_rq_insert: 8,32 RA 32768 () 786400 + 64 [gc] gc-6549 [001] d..1 214682.212802: block_unplug: [gc] 1 gc-6549 [001] .... 214682.213892: block_bio_queue: 8,32 R 67494144 + 8 [gc] gc-6549 [001] .... 214682.213899: block_getrq: 8,32 R 67494144 + 8 [gc] gc-6549 [001] .... 214682.213902: block_plug: [gc] gc-6549 [001] d..1 214682.213905: block_rq_insert: 8,32 R 4096 () 67494144 + 8 [gc] gc-6549 [001] d..1 214682.213908: block_unplug: [gc] 1 gc-6549 [001] .... 214682.226405: block_bio_queue: 8,32 R 67494152 + 8 [gc] gc-6549 [001] .... 214682.226412: block_getrq: 8,32 R 67494152 + 8 [gc] gc-6549 [001] .... 214682.226414: block_plug: [gc] gc-6549 [001] d..1 214682.226417: block_rq_insert: 8,32 R 4096 () 67494152 + 8 [gc] gc-6549 [001] d..1 214682.226420: block_unplug: [gc] 1 gc-6549 [001] .... 214682.226904: block_bio_queue: 8,32 R 67494160 + 8 [gc] gc-6549 [001] .... 214682.226910: block_getrq: 8,32 R 67494160 + 8 [gc] gc-6549 [001] .... 214682.226911: block_plug: [gc] gc-6549 [001] d..1 214682.226914: block_rq_insert: 8,32 R 4096 () 67494160 + 8 [gc] gc-6549 [001] d..1 214682.226916: block_unplug: [gc] 1 After: gc-5678 [003] .... 214327.025906: block_bio_queue: 8,32 R 67493824 + 8 [gc] gc-5678 [003] .... 214327.025908: block_bio_backmerge: 8,32 R 67493824 + 8 [gc] gc-5678 [003] .... 214327.025915: block_bio_queue: 8,32 R 67493832 + 8 [gc] gc-5678 [003] .... 214327.025917: block_bio_backmerge: 8,32 R 67493832 + 8 [gc] gc-5678 [003] .... 214327.025923: block_bio_queue: 8,32 R 67493840 + 8 [gc] gc-5678 [003] .... 214327.025925: block_bio_backmerge: 8,32 R 67493840 + 8 [gc] gc-5678 [003] .... 214327.025932: block_bio_queue: 8,32 R 67493848 + 8 [gc] gc-5678 [003] .... 214327.025934: block_bio_backmerge: 8,32 R 67493848 + 8 [gc] gc-5678 [003] .... 214327.025941: block_bio_queue: 8,32 R 67493856 + 8 [gc] gc-5678 [003] .... 214327.025943: block_bio_backmerge: 8,32 R 67493856 + 8 [gc] gc-5678 [003] .... 214327.025953: block_bio_queue: 8,32 R 67493864 + 8 [gc] gc-5678 [003] .... 214327.025955: block_bio_backmerge: 8,32 R 67493864 + 8 [gc] gc-5678 [003] .... 214327.025962: block_bio_queue: 8,32 R 67493872 + 8 [gc] gc-5678 [003] .... 214327.025964: block_bio_backmerge: 8,32 R 67493872 + 8 [gc] gc-5678 [003] .... 214327.025970: block_bio_queue: 8,32 R 67493880 + 8 [gc] gc-5678 [003] .... 214327.025972: block_bio_backmerge: 8,32 R 67493880 + 8 [gc] gc-5678 [003] .... 214327.026000: block_bio_queue: 8,32 WS 34123776 + 2048 [gc] gc-5678 [003] .... 214327.026019: block_getrq: 8,32 WS 34123776 + 2048 [gc] gc-5678 [003] d..1 214327.026021: block_rq_insert: 8,32 R 131072 () 67493632 + 256 [gc] gc-5678 [003] d..1 214327.026023: block_unplug: [gc] 1 gc-5678 [003] d..1 214327.026026: block_rq_issue: 8,32 R 131072 () 67493632 + 256 [gc] gc-5678 [003] .... 214327.026046: block_plug: [gc] Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-08-14 14:37:25 +00:00
}
locate_dirty_segment(sbi, GET_SEGNO(sbi, old_blkaddr));
locate_dirty_segment(sbi, GET_SEGNO(sbi, new_blkaddr));
locate_dirty_segment(sbi, old_cursegno);
if (recover_curseg) {
if (old_cursegno != curseg->segno) {
curseg->next_segno = old_cursegno;
change_curseg(sbi, type);
}
curseg->next_blkoff = old_blkoff;
}
up_write(&sit_i->sentry_lock);
mutex_unlock(&curseg->curseg_mutex);
up_write(&SM_I(sbi)->curseg_lock);
}
void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
block_t old_addr, block_t new_addr,
unsigned char version, bool recover_curseg,
bool recover_newaddr)
{
struct f2fs_summary sum;
set_summary(&sum, dn->nid, dn->ofs_in_node, version);
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 16:20:41 +00:00
f2fs_do_replace_block(sbi, &sum, old_addr, new_addr,
recover_curseg, recover_newaddr);
f2fs_update_data_blkaddr(dn, new_addr);
}
void f2fs_wait_on_page_writeback(struct page *page,
enum page_type type, bool ordered, bool locked)
{
if (PageWriteback(page)) {
struct f2fs_sb_info *sbi = F2FS_P_SB(page);
/* submit cached LFS IO */
f2fs_submit_merged_write_cond(sbi, NULL, page, 0, type);
/* sbumit cached IPU IO */
f2fs_submit_merged_ipu_write(sbi, NULL, page);
if (ordered) {
wait_on_page_writeback(page);
f2fs_bug_on(sbi, locked && PageWriteback(page));
} else {
wait_for_stable_page(page);
}
}
}
void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr)
f2fs crypto: fix racing of accessing encrypted page among different competitors Since we use different page cache (normally inode's page cache for R/W and meta inode's page cache for GC) to cache the same physical block which is belong to an encrypted inode. Writeback of these two page cache should be exclusive, but now we didn't handle writeback state well, so there may be potential racing problem: a) kworker: f2fs_gc: - f2fs_write_data_pages - f2fs_write_data_page - do_write_data_page - write_data_page - f2fs_submit_page_mbio (page#1 in inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) - gc_data_segment - move_encrypted_block - pagecache_get_page (page#2 in meta inode's page cache was cached with the invalid datas of physical block located in new blkaddr) - f2fs_submit_page_mbio (page#1 was submitted, later, page#2 with invalid data will be submitted) b) f2fs_gc: - gc_data_segment - move_encrypted_block - f2fs_submit_page_mbio (page#1 in meta inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) user thread: - f2fs_write_begin - f2fs_submit_page_bio (we submit the request to block layer to update page#2 in inode's page cache with physical block located in new blkaddr, so here we may read gabbage data from new blkaddr since GC hasn't writebacked the page#1 yet) This patch fixes above potential racing problem for encrypted inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-10-08 05:27:34 +00:00
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
f2fs crypto: fix racing of accessing encrypted page among different competitors Since we use different page cache (normally inode's page cache for R/W and meta inode's page cache for GC) to cache the same physical block which is belong to an encrypted inode. Writeback of these two page cache should be exclusive, but now we didn't handle writeback state well, so there may be potential racing problem: a) kworker: f2fs_gc: - f2fs_write_data_pages - f2fs_write_data_page - do_write_data_page - write_data_page - f2fs_submit_page_mbio (page#1 in inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) - gc_data_segment - move_encrypted_block - pagecache_get_page (page#2 in meta inode's page cache was cached with the invalid datas of physical block located in new blkaddr) - f2fs_submit_page_mbio (page#1 was submitted, later, page#2 with invalid data will be submitted) b) f2fs_gc: - gc_data_segment - move_encrypted_block - f2fs_submit_page_mbio (page#1 in meta inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) user thread: - f2fs_write_begin - f2fs_submit_page_bio (we submit the request to block layer to update page#2 in inode's page cache with physical block located in new blkaddr, so here we may read gabbage data from new blkaddr since GC hasn't writebacked the page#1 yet) This patch fixes above potential racing problem for encrypted inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-10-08 05:27:34 +00:00
struct page *cpage;
if (!f2fs_post_read_required(inode))
return;
f2fs: introduce DATA_GENERIC_ENHANCE Previously, f2fs_is_valid_blkaddr(, blkaddr, DATA_GENERIC) will check whether @blkaddr locates in main area or not. That check is weak, since the block address in range of main area can point to the address which is not valid in segment info table, and we can not detect such condition, we may suffer worse corruption as system continues running. So this patch introduce DATA_GENERIC_ENHANCE to enhance the sanity check which trigger SIT bitmap check rather than only range check. This patch did below changes as wel: - set SBI_NEED_FSCK in f2fs_is_valid_blkaddr(). - get rid of is_valid_data_blkaddr() to avoid panic if blkaddr is invalid. - introduce verify_fio_blkaddr() to wrap fio {new,old}_blkaddr validation check. - spread blkaddr check in: * f2fs_get_node_info() * __read_out_blkaddrs() * f2fs_submit_page_read() * ra_data_block() * do_recover_data() This patch can fix bug reported from bugzilla below: https://bugzilla.kernel.org/show_bug.cgi?id=203215 https://bugzilla.kernel.org/show_bug.cgi?id=203223 https://bugzilla.kernel.org/show_bug.cgi?id=203231 https://bugzilla.kernel.org/show_bug.cgi?id=203235 https://bugzilla.kernel.org/show_bug.cgi?id=203241 = Update by Jaegeuk Kim = DATA_GENERIC_ENHANCE enhanced to validate block addresses on read/write paths. But, xfstest/generic/446 compalins some generated kernel messages saying invalid bitmap was detected when reading a block. The reaons is, when we get the block addresses from extent_cache, there is no lock to synchronize it from truncating the blocks in parallel. Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-04-15 07:26:32 +00:00
if (!__is_valid_data_blkaddr(blkaddr))
f2fs crypto: fix racing of accessing encrypted page among different competitors Since we use different page cache (normally inode's page cache for R/W and meta inode's page cache for GC) to cache the same physical block which is belong to an encrypted inode. Writeback of these two page cache should be exclusive, but now we didn't handle writeback state well, so there may be potential racing problem: a) kworker: f2fs_gc: - f2fs_write_data_pages - f2fs_write_data_page - do_write_data_page - write_data_page - f2fs_submit_page_mbio (page#1 in inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) - gc_data_segment - move_encrypted_block - pagecache_get_page (page#2 in meta inode's page cache was cached with the invalid datas of physical block located in new blkaddr) - f2fs_submit_page_mbio (page#1 was submitted, later, page#2 with invalid data will be submitted) b) f2fs_gc: - gc_data_segment - move_encrypted_block - f2fs_submit_page_mbio (page#1 in meta inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) user thread: - f2fs_write_begin - f2fs_submit_page_bio (we submit the request to block layer to update page#2 in inode's page cache with physical block located in new blkaddr, so here we may read gabbage data from new blkaddr since GC hasn't writebacked the page#1 yet) This patch fixes above potential racing problem for encrypted inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-10-08 05:27:34 +00:00
return;
cpage = find_lock_page(META_MAPPING(sbi), blkaddr);
if (cpage) {
f2fs_wait_on_page_writeback(cpage, DATA, true, true);
f2fs crypto: fix racing of accessing encrypted page among different competitors Since we use different page cache (normally inode's page cache for R/W and meta inode's page cache for GC) to cache the same physical block which is belong to an encrypted inode. Writeback of these two page cache should be exclusive, but now we didn't handle writeback state well, so there may be potential racing problem: a) kworker: f2fs_gc: - f2fs_write_data_pages - f2fs_write_data_page - do_write_data_page - write_data_page - f2fs_submit_page_mbio (page#1 in inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) - gc_data_segment - move_encrypted_block - pagecache_get_page (page#2 in meta inode's page cache was cached with the invalid datas of physical block located in new blkaddr) - f2fs_submit_page_mbio (page#1 was submitted, later, page#2 with invalid data will be submitted) b) f2fs_gc: - gc_data_segment - move_encrypted_block - f2fs_submit_page_mbio (page#1 in meta inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) user thread: - f2fs_write_begin - f2fs_submit_page_bio (we submit the request to block layer to update page#2 in inode's page cache with physical block located in new blkaddr, so here we may read gabbage data from new blkaddr since GC hasn't writebacked the page#1 yet) This patch fixes above potential racing problem for encrypted inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-10-08 05:27:34 +00:00
f2fs_put_page(cpage, 1);
}
}
void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
block_t len)
{
block_t i;
for (i = 0; i < len; i++)
f2fs_wait_on_block_writeback(inode, blkaddr + i);
}
static int read_compacted_summaries(struct f2fs_sb_info *sbi)
{
struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
struct curseg_info *seg_i;
unsigned char *kaddr;
struct page *page;
block_t start;
int i, j, offset;
start = start_sum_block(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 16:20:41 +00:00
page = f2fs_get_meta_page(sbi, start++);
if (IS_ERR(page))
return PTR_ERR(page);
kaddr = (unsigned char *)page_address(page);
/* Step 1: restore nat cache */
seg_i = CURSEG_I(sbi, CURSEG_HOT_DATA);
memcpy(seg_i->journal, kaddr, SUM_JOURNAL_SIZE);
/* Step 2: restore sit cache */
seg_i = CURSEG_I(sbi, CURSEG_COLD_DATA);
memcpy(seg_i->journal, kaddr + SUM_JOURNAL_SIZE, SUM_JOURNAL_SIZE);
offset = 2 * SUM_JOURNAL_SIZE;
/* Step 3: restore summary entries */
for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) {
unsigned short blk_off;
unsigned int segno;
seg_i = CURSEG_I(sbi, i);
segno = le32_to_cpu(ckpt->cur_data_segno[i]);
blk_off = le16_to_cpu(ckpt->cur_data_blkoff[i]);
seg_i->next_segno = segno;
reset_curseg(sbi, i, 0);
seg_i->alloc_type = ckpt->alloc_type[i];
seg_i->next_blkoff = blk_off;
if (seg_i->alloc_type == SSR)
blk_off = sbi->blocks_per_seg;
for (j = 0; j < blk_off; j++) {
struct f2fs_summary *s;
s = (struct f2fs_summary *)(kaddr + offset);
seg_i->sum_blk->entries[j] = *s;
offset += SUMMARY_SIZE;
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 12:29:47 +00:00
if (offset + SUMMARY_SIZE <= PAGE_SIZE -
SUM_FOOTER_SIZE)
continue;
f2fs_put_page(page, 1);
page = NULL;
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 16:20:41 +00:00
page = f2fs_get_meta_page(sbi, start++);
if (IS_ERR(page))
return PTR_ERR(page);
kaddr = (unsigned char *)page_address(page);
offset = 0;
}
}
f2fs_put_page(page, 1);
return 0;
}
static int read_normal_summaries(struct f2fs_sb_info *sbi, int type)
{
struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
struct f2fs_summary_block *sum;
struct curseg_info *curseg;
struct page *new;
unsigned short blk_off;
unsigned int segno = 0;
block_t blk_addr = 0;
int err = 0;
/* get segment number and block addr */
if (IS_DATASEG(type)) {
segno = le32_to_cpu(ckpt->cur_data_segno[type]);
blk_off = le16_to_cpu(ckpt->cur_data_blkoff[type -
CURSEG_HOT_DATA]);
if (__exist_node_summaries(sbi))
blk_addr = sum_blk_addr(sbi, NR_CURSEG_TYPE, type);
else
blk_addr = sum_blk_addr(sbi, NR_CURSEG_DATA_TYPE, type);
} else {
segno = le32_to_cpu(ckpt->cur_node_segno[type -
CURSEG_HOT_NODE]);
blk_off = le16_to_cpu(ckpt->cur_node_blkoff[type -
CURSEG_HOT_NODE]);
if (__exist_node_summaries(sbi))
blk_addr = sum_blk_addr(sbi, NR_CURSEG_NODE_TYPE,
type - CURSEG_HOT_NODE);
else
blk_addr = GET_SUM_BLOCK(sbi, segno);
}
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 16:20:41 +00:00
new = f2fs_get_meta_page(sbi, blk_addr);
if (IS_ERR(new))
return PTR_ERR(new);
sum = (struct f2fs_summary_block *)page_address(new);
if (IS_NODESEG(type)) {
if (__exist_node_summaries(sbi)) {
struct f2fs_summary *ns = &sum->entries[0];
int i;
for (i = 0; i < sbi->blocks_per_seg; i++, ns++) {
ns->version = 0;
ns->ofs_in_node = 0;
}
} else {
err = f2fs_restore_node_summary(sbi, segno, sum);
if (err)
goto out;
}
}
/* set uncompleted segment to curseg */
curseg = CURSEG_I(sbi, type);
mutex_lock(&curseg->curseg_mutex);
/* update journal info */
down_write(&curseg->journal_rwsem);
memcpy(curseg->journal, &sum->journal, SUM_JOURNAL_SIZE);
up_write(&curseg->journal_rwsem);
memcpy(curseg->sum_blk->entries, sum->entries, SUM_ENTRY_SIZE);
memcpy(&curseg->sum_blk->footer, &sum->footer, SUM_FOOTER_SIZE);
curseg->next_segno = segno;
reset_curseg(sbi, type, 0);
curseg->alloc_type = ckpt->alloc_type[type];
curseg->next_blkoff = blk_off;
mutex_unlock(&curseg->curseg_mutex);
out:
f2fs_put_page(new, 1);
return err;
}
static int restore_curseg_summaries(struct f2fs_sb_info *sbi)
{
struct f2fs_journal *sit_j = CURSEG_I(sbi, CURSEG_COLD_DATA)->journal;
struct f2fs_journal *nat_j = CURSEG_I(sbi, CURSEG_HOT_DATA)->journal;
int type = CURSEG_HOT_DATA;
int err;
if (is_set_ckpt_flags(sbi, CP_COMPACT_SUM_FLAG)) {
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 16:20:41 +00:00
int npages = f2fs_npages_for_summary_flush(sbi, true);
if (npages >= 2)
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 16:20:41 +00:00
f2fs_ra_meta_pages(sbi, start_sum_block(sbi), npages,
META_CP, true);
/* restore for compacted data summary */
err = read_compacted_summaries(sbi);
if (err)
return err;
type = CURSEG_HOT_NODE;
}
if (__exist_node_summaries(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 16:20:41 +00:00
f2fs_ra_meta_pages(sbi, sum_blk_addr(sbi, NR_CURSEG_TYPE, type),
NR_CURSEG_TYPE - type, META_CP, true);
for (; type <= CURSEG_COLD_NODE; type++) {
err = read_normal_summaries(sbi, type);
if (err)
return err;
}
/* sanity check for summary blocks */
if (nats_in_cursum(nat_j) > NAT_JOURNAL_ENTRIES ||
sits_in_cursum(sit_j) > SIT_JOURNAL_ENTRIES) {
f2fs_err(sbi, "invalid journal entries nats %u sits %u\n",
nats_in_cursum(nat_j), sits_in_cursum(sit_j));
return -EINVAL;
}
return 0;
}
static void write_compacted_summaries(struct f2fs_sb_info *sbi, block_t blkaddr)
{
struct page *page;
unsigned char *kaddr;
struct f2fs_summary *summary;
struct curseg_info *seg_i;
int written_size = 0;
int i, j;
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 16:20:41 +00:00
page = f2fs_grab_meta_page(sbi, blkaddr++);
kaddr = (unsigned char *)page_address(page);
memset(kaddr, 0, PAGE_SIZE);
/* Step 1: write nat cache */
seg_i = CURSEG_I(sbi, CURSEG_HOT_DATA);
memcpy(kaddr, seg_i->journal, SUM_JOURNAL_SIZE);
written_size += SUM_JOURNAL_SIZE;
/* Step 2: write sit cache */
seg_i = CURSEG_I(sbi, CURSEG_COLD_DATA);
memcpy(kaddr + written_size, seg_i->journal, SUM_JOURNAL_SIZE);
written_size += SUM_JOURNAL_SIZE;
/* Step 3: write summary entries */
for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) {
unsigned short blkoff;
seg_i = CURSEG_I(sbi, i);
if (sbi->ckpt->alloc_type[i] == SSR)
blkoff = sbi->blocks_per_seg;
else
blkoff = curseg_blkoff(sbi, i);
for (j = 0; j < blkoff; j++) {
if (!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 16:20:41 +00:00
page = f2fs_grab_meta_page(sbi, blkaddr++);
kaddr = (unsigned char *)page_address(page);
memset(kaddr, 0, PAGE_SIZE);
written_size = 0;
}
summary = (struct f2fs_summary *)(kaddr + written_size);
*summary = seg_i->sum_blk->entries[j];
written_size += SUMMARY_SIZE;
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 12:29:47 +00:00
if (written_size + SUMMARY_SIZE <= PAGE_SIZE -
SUM_FOOTER_SIZE)
continue;
set_page_dirty(page);
f2fs_put_page(page, 1);
page = NULL;
}
}
if (page) {
set_page_dirty(page);
f2fs_put_page(page, 1);
}
}
static void write_normal_summaries(struct f2fs_sb_info *sbi,
block_t blkaddr, int type)
{
int i, end;
if (IS_DATASEG(type))
end = type + NR_CURSEG_DATA_TYPE;
else
end = type + NR_CURSEG_NODE_TYPE;
for (i = type; i < end; i++)
write_current_sum_page(sbi, i, blkaddr + (i - type));
}
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 16:20:41 +00:00
void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk)
{
if (is_set_ckpt_flags(sbi, CP_COMPACT_SUM_FLAG))
write_compacted_summaries(sbi, start_blk);
else
write_normal_summaries(sbi, start_blk, CURSEG_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 16:20:41 +00:00
void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk)
{
write_normal_summaries(sbi, start_blk, CURSEG_HOT_NODE);
}
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 16:20:41 +00:00
int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
unsigned int val, int alloc)
{
int i;
if (type == NAT_JOURNAL) {
for (i = 0; i < nats_in_cursum(journal); i++) {
if (le32_to_cpu(nid_in_journal(journal, i)) == val)
return i;
}
if (alloc && __has_cursum_space(journal, 1, NAT_JOURNAL))
return update_nats_in_cursum(journal, 1);
} else if (type == SIT_JOURNAL) {
for (i = 0; i < sits_in_cursum(journal); i++)
if (le32_to_cpu(segno_in_journal(journal, i)) == val)
return i;
if (alloc && __has_cursum_space(journal, 1, SIT_JOURNAL))
return update_sits_in_cursum(journal, 1);
}
return -1;
}
static struct page *get_current_sit_page(struct f2fs_sb_info *sbi,
unsigned int segno)
{
return f2fs_get_meta_page_nofail(sbi, current_sit_addr(sbi, segno));
}
static struct page *get_next_sit_page(struct f2fs_sb_info *sbi,
unsigned int start)
{
struct sit_info *sit_i = SIT_I(sbi);
f2fs: rebuild sit page from sit info in mem This patch rebuild sit page from sit info in mem instead of issue a read io. I test this method and the result is as below: Pre: mmc_perf_test-12061 [001] ...1 976.819992: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [001] ...1 976.856446: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 998.976946: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 999.023269: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1022.060772: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1022.111034: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [002] ...1 1070.127643: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1070.187352: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1095.942124: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1095.995975: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1122.535091: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1122.586521: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [001] ...1 1147.897487: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [001] ...1 1147.959438: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1177.926951: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [002] ...1 1177.976823: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [002] ...1 1204.176087: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [002] ...1 1204.239046: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit Some sit flush consume more than 50ms. Now: mmc_perf_test-2187 [007] ...1 196.840684: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [007] ...1 196.841258: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [007] ...1 219.430582: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [007] ...1 219.431144: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [002] ...1 243.638678: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 243.638980: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [002] ...1 265.392180: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [002] ...1 265.392245: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [000] ...1 290.309051: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 290.309116: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [003] ...1 317.144209: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [003] ...1 317.145913: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [005] ...1 343.224954: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [005] ...1 343.225574: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [000] ...1 370.239846: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 370.241138: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [001] ...1 397.029043: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [001] ...1 397.030750: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [003] ...1 425.386377: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [003] ...1 425.387735: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit Most sit flush consume no more than 1ms. Signed-off-by: Yunlei He <heyunlei@huawei.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-01-25 09:27:11 +00:00
struct page *page;
pgoff_t src_off, dst_off;
src_off = current_sit_addr(sbi, start);
dst_off = next_sit_addr(sbi, src_off);
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 16:20:41 +00:00
page = f2fs_grab_meta_page(sbi, dst_off);
f2fs: rebuild sit page from sit info in mem This patch rebuild sit page from sit info in mem instead of issue a read io. I test this method and the result is as below: Pre: mmc_perf_test-12061 [001] ...1 976.819992: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [001] ...1 976.856446: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 998.976946: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 999.023269: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1022.060772: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1022.111034: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [002] ...1 1070.127643: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1070.187352: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1095.942124: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1095.995975: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1122.535091: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1122.586521: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [001] ...1 1147.897487: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [001] ...1 1147.959438: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1177.926951: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [002] ...1 1177.976823: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [002] ...1 1204.176087: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [002] ...1 1204.239046: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit Some sit flush consume more than 50ms. Now: mmc_perf_test-2187 [007] ...1 196.840684: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [007] ...1 196.841258: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [007] ...1 219.430582: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [007] ...1 219.431144: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [002] ...1 243.638678: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 243.638980: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [002] ...1 265.392180: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [002] ...1 265.392245: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [000] ...1 290.309051: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 290.309116: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [003] ...1 317.144209: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [003] ...1 317.145913: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [005] ...1 343.224954: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [005] ...1 343.225574: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [000] ...1 370.239846: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 370.241138: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [001] ...1 397.029043: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [001] ...1 397.030750: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [003] ...1 425.386377: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [003] ...1 425.387735: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit Most sit flush consume no more than 1ms. Signed-off-by: Yunlei He <heyunlei@huawei.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-01-25 09:27:11 +00:00
seg_info_to_sit_page(sbi, page, start);
f2fs: rebuild sit page from sit info in mem This patch rebuild sit page from sit info in mem instead of issue a read io. I test this method and the result is as below: Pre: mmc_perf_test-12061 [001] ...1 976.819992: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [001] ...1 976.856446: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 998.976946: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 999.023269: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1022.060772: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1022.111034: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [002] ...1 1070.127643: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1070.187352: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1095.942124: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1095.995975: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1122.535091: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1122.586521: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [001] ...1 1147.897487: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [001] ...1 1147.959438: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1177.926951: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [002] ...1 1177.976823: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [002] ...1 1204.176087: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [002] ...1 1204.239046: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit Some sit flush consume more than 50ms. Now: mmc_perf_test-2187 [007] ...1 196.840684: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [007] ...1 196.841258: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [007] ...1 219.430582: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [007] ...1 219.431144: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [002] ...1 243.638678: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 243.638980: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [002] ...1 265.392180: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [002] ...1 265.392245: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [000] ...1 290.309051: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 290.309116: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [003] ...1 317.144209: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [003] ...1 317.145913: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [005] ...1 343.224954: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [005] ...1 343.225574: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [000] ...1 370.239846: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 370.241138: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [001] ...1 397.029043: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [001] ...1 397.030750: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [003] ...1 425.386377: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [003] ...1 425.387735: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit Most sit flush consume no more than 1ms. Signed-off-by: Yunlei He <heyunlei@huawei.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-01-25 09:27:11 +00:00
set_page_dirty(page);
set_to_next_sit(sit_i, start);
f2fs: rebuild sit page from sit info in mem This patch rebuild sit page from sit info in mem instead of issue a read io. I test this method and the result is as below: Pre: mmc_perf_test-12061 [001] ...1 976.819992: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [001] ...1 976.856446: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 998.976946: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 999.023269: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1022.060772: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1022.111034: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [002] ...1 1070.127643: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1070.187352: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1095.942124: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1095.995975: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1122.535091: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [003] ...1 1122.586521: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [001] ...1 1147.897487: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [001] ...1 1147.959438: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [003] ...1 1177.926951: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [002] ...1 1177.976823: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-12061 [002] ...1 1204.176087: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-12061 [002] ...1 1204.239046: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit Some sit flush consume more than 50ms. Now: mmc_perf_test-2187 [007] ...1 196.840684: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [007] ...1 196.841258: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [007] ...1 219.430582: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [007] ...1 219.431144: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [002] ...1 243.638678: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 243.638980: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [002] ...1 265.392180: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [002] ...1 265.392245: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [000] ...1 290.309051: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 290.309116: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [003] ...1 317.144209: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [003] ...1 317.145913: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [005] ...1 343.224954: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [005] ...1 343.225574: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [000] ...1 370.239846: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [000] ...1 370.241138: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [001] ...1 397.029043: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [001] ...1 397.030750: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit mmc_perf_test-2187 [003] ...1 425.386377: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = start flush sit mmc_perf_test-2187 [003] ...1 425.387735: f2fs_write_checkpoint: dev = (259,44), checkpoint for Sync, state = end flush sit Most sit flush consume no more than 1ms. Signed-off-by: Yunlei He <heyunlei@huawei.com> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-01-25 09:27:11 +00:00
return page;
}
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
static struct sit_entry_set *grab_sit_entry_set(void)
{
struct sit_entry_set *ses =
f2fs_kmem_cache_alloc(sit_entry_set_slab, GFP_NOFS);
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
ses->entry_cnt = 0;
INIT_LIST_HEAD(&ses->set_list);
return ses;
}
static void release_sit_entry_set(struct sit_entry_set *ses)
{
list_del(&ses->set_list);
kmem_cache_free(sit_entry_set_slab, ses);
}
static void adjust_sit_entry_set(struct sit_entry_set *ses,
struct list_head *head)
{
struct sit_entry_set *next = ses;
if (list_is_last(&ses->set_list, head))
return;
list_for_each_entry_continue(next, head, set_list)
if (ses->entry_cnt <= next->entry_cnt)
break;
list_move_tail(&ses->set_list, &next->set_list);
}
static void add_sit_entry(unsigned int segno, struct list_head *head)
{
struct sit_entry_set *ses;
unsigned int start_segno = START_SEGNO(segno);
list_for_each_entry(ses, head, set_list) {
if (ses->start_segno == start_segno) {
ses->entry_cnt++;
adjust_sit_entry_set(ses, head);
return;
}
}
ses = grab_sit_entry_set();
ses->start_segno = start_segno;
ses->entry_cnt++;
list_add(&ses->set_list, head);
}
static void add_sits_in_set(struct f2fs_sb_info *sbi)
{
struct f2fs_sm_info *sm_info = SM_I(sbi);
struct list_head *set_list = &sm_info->sit_entry_set;
unsigned long *bitmap = SIT_I(sbi)->dirty_sentries_bitmap;
unsigned int segno;
for_each_set_bit(segno, bitmap, MAIN_SEGS(sbi))
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
add_sit_entry(segno, set_list);
}
static void remove_sits_in_journal(struct f2fs_sb_info *sbi)
{
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_COLD_DATA);
struct f2fs_journal *journal = curseg->journal;
int i;
down_write(&curseg->journal_rwsem);
for (i = 0; i < sits_in_cursum(journal); i++) {
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
unsigned int segno;
bool dirtied;
segno = le32_to_cpu(segno_in_journal(journal, i));
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
dirtied = __mark_sit_entry_dirty(sbi, segno);
if (!dirtied)
add_sit_entry(segno, &SM_I(sbi)->sit_entry_set);
}
update_sits_in_cursum(journal, -i);
up_write(&curseg->journal_rwsem);
}
/*
* CP calls this function, which flushes SIT entries including sit_journal,
* and moves prefree segs to free segs.
*/
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 16:20:41 +00:00
void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc)
{
struct sit_info *sit_i = SIT_I(sbi);
unsigned long *bitmap = sit_i->dirty_sentries_bitmap;
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_COLD_DATA);
struct f2fs_journal *journal = curseg->journal;
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
struct sit_entry_set *ses, *tmp;
struct list_head *head = &SM_I(sbi)->sit_entry_set;
f2fs: ioctl for removing a range from F2FS This ioctl shrinks a given length (aligned to sections) from end of the main area. Any cursegs and valid blocks will be moved out before invalidating the range. This feature can be used for adjusting partition sizes online. History of the patch: Sahitya Tummala: - Add this ioctl for f2fs_compat_ioctl() as well. - Fix debugfs status to reflect the online resize changes. - Fix potential race between online resize path and allocate new data block path or gc path. Others: - Rename some identifiers. - Add some error handling branches. - Clear sbi->next_victim_seg[BG_GC/FG_GC] in shrinking range. - Implement this interface as ext4's, and change the parameter from shrunk bytes to new block count of F2FS. - During resizing, force to empty sit_journal and forbid adding new entries to it, in order to avoid invalid segno in journal after resize. - Reduce sbi->user_block_count before resize starts. - Commit the updated superblock first, and then update in-memory metadata only when the former succeeds. - Target block count must align to sections. - Write checkpoint before and after committing the new superblock, w/o CP_FSCK_FLAG respectively, so that the FS can be fixed by fsck even if resize fails after the new superblock is committed. - In free_segment_range(), reduce granularity of gc_mutex. - Add protection on curseg migration. - Add freeze_bdev() and thaw_bdev() for resize fs. - Remove CUR_MAIN_SECS and use MAIN_SECS directly for allocation. - Recover super_block and FS metadata when resize fails. - No need to clear CP_FSCK_FLAG in update_ckpt_flags(). - Clean up the sb and fs metadata update functions for resize_fs. Geert Uytterhoeven: - Use div_u64*() for 64-bit divisions Arnd Bergmann: - Not all architectures support get_user() with a 64-bit argument: ERROR: "__get_user_bad" [fs/f2fs/f2fs.ko] undefined! Use copy_from_user() here, this will always work. Signed-off-by: Qiuyang Sun <sunqiuyang@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Sahitya Tummala <stummala@codeaurora.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-06-05 03:33:25 +00:00
bool to_journal = !is_sbi_flag_set(sbi, SBI_IS_RESIZEFS);
struct seg_entry *se;
down_write(&sit_i->sentry_lock);
if (!sit_i->dirty_sentries)
goto out;
/*
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
* add and account sit entries of dirty bitmap in sit entry
* set temporarily
*/
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
add_sits_in_set(sbi);
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
/*
* if there are no enough space in journal to store dirty sit
* entries, remove all entries from journal and add and account
* them in sit entry set.
*/
f2fs: ioctl for removing a range from F2FS This ioctl shrinks a given length (aligned to sections) from end of the main area. Any cursegs and valid blocks will be moved out before invalidating the range. This feature can be used for adjusting partition sizes online. History of the patch: Sahitya Tummala: - Add this ioctl for f2fs_compat_ioctl() as well. - Fix debugfs status to reflect the online resize changes. - Fix potential race between online resize path and allocate new data block path or gc path. Others: - Rename some identifiers. - Add some error handling branches. - Clear sbi->next_victim_seg[BG_GC/FG_GC] in shrinking range. - Implement this interface as ext4's, and change the parameter from shrunk bytes to new block count of F2FS. - During resizing, force to empty sit_journal and forbid adding new entries to it, in order to avoid invalid segno in journal after resize. - Reduce sbi->user_block_count before resize starts. - Commit the updated superblock first, and then update in-memory metadata only when the former succeeds. - Target block count must align to sections. - Write checkpoint before and after committing the new superblock, w/o CP_FSCK_FLAG respectively, so that the FS can be fixed by fsck even if resize fails after the new superblock is committed. - In free_segment_range(), reduce granularity of gc_mutex. - Add protection on curseg migration. - Add freeze_bdev() and thaw_bdev() for resize fs. - Remove CUR_MAIN_SECS and use MAIN_SECS directly for allocation. - Recover super_block and FS metadata when resize fails. - No need to clear CP_FSCK_FLAG in update_ckpt_flags(). - Clean up the sb and fs metadata update functions for resize_fs. Geert Uytterhoeven: - Use div_u64*() for 64-bit divisions Arnd Bergmann: - Not all architectures support get_user() with a 64-bit argument: ERROR: "__get_user_bad" [fs/f2fs/f2fs.ko] undefined! Use copy_from_user() here, this will always work. Signed-off-by: Qiuyang Sun <sunqiuyang@huawei.com> Signed-off-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Sahitya Tummala <stummala@codeaurora.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2019-06-05 03:33:25 +00:00
if (!__has_cursum_space(journal, sit_i->dirty_sentries, SIT_JOURNAL) ||
!to_journal)
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
remove_sits_in_journal(sbi);
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
/*
* there are two steps to flush sit entries:
* #1, flush sit entries to journal in current cold data summary block.
* #2, flush sit entries to sit page.
*/
list_for_each_entry_safe(ses, tmp, head, set_list) {
struct page *page = NULL;
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
struct f2fs_sit_block *raw_sit = NULL;
unsigned int start_segno = ses->start_segno;
unsigned int end = min(start_segno + SIT_ENTRY_PER_BLOCK,
(unsigned long)MAIN_SEGS(sbi));
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
unsigned int segno = start_segno;
if (to_journal &&
!__has_cursum_space(journal, ses->entry_cnt, SIT_JOURNAL))
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
to_journal = false;
if (to_journal) {
down_write(&curseg->journal_rwsem);
} else {
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
page = get_next_sit_page(sbi, start_segno);
raw_sit = page_address(page);
}
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
/* flush dirty sit entries in region of current sit set */
for_each_set_bit_from(segno, bitmap, end) {
int offset, sit_offset;
se = get_seg_entry(sbi, segno);
#ifdef CONFIG_F2FS_CHECK_FS
if (memcmp(se->cur_valid_map, se->cur_valid_map_mir,
SIT_VBLOCK_MAP_SIZE))
f2fs_bug_on(sbi, 1);
#endif
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
/* add discard candidates */
if (!(cpc->reason & CP_DISCARD)) {
cpc->trim_start = segno;
add_discard_addrs(sbi, cpc, false);
}
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
if (to_journal) {
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 16:20:41 +00:00
offset = f2fs_lookup_journal_in_cursum(journal,
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
SIT_JOURNAL, segno, 1);
f2fs_bug_on(sbi, offset < 0);
segno_in_journal(journal, offset) =
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
cpu_to_le32(segno);
seg_info_to_raw_sit(se,
&sit_in_journal(journal, offset));
check_block_count(sbi, segno,
&sit_in_journal(journal, offset));
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
} else {
sit_offset = SIT_ENTRY_OFFSET(sit_i, segno);
seg_info_to_raw_sit(se,
&raw_sit->entries[sit_offset]);
check_block_count(sbi, segno,
&raw_sit->entries[sit_offset]);
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
}
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
__clear_bit(segno, bitmap);
sit_i->dirty_sentries--;
ses->entry_cnt--;
}
if (to_journal)
up_write(&curseg->journal_rwsem);
else
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
f2fs_put_page(page, 1);
f2fs_bug_on(sbi, ses->entry_cnt);
release_sit_entry_set(ses);
}
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
f2fs_bug_on(sbi, !list_empty(head));
f2fs_bug_on(sbi, sit_i->dirty_sentries);
out:
if (cpc->reason & CP_DISCARD) {
__u64 trim_start = cpc->trim_start;
for (; cpc->trim_start <= cpc->trim_end; cpc->trim_start++)
add_discard_addrs(sbi, cpc, false);
cpc->trim_start = trim_start;
}
up_write(&sit_i->sentry_lock);
set_prefree_as_free_segments(sbi);
}
static int build_sit_info(struct f2fs_sb_info *sbi)
{
struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
struct sit_info *sit_i;
unsigned int sit_segs, start;
char *src_bitmap, *bitmap;
unsigned int bitmap_size, main_bitmap_size, sit_bitmap_size;
/* allocate memory for SIT information */
sit_i = f2fs_kzalloc(sbi, sizeof(struct sit_info), GFP_KERNEL);
if (!sit_i)
return -ENOMEM;
SM_I(sbi)->sit_info = sit_i;
treewide: Use array_size in f2fs_kvzalloc() The f2fs_kvzalloc() function has no 2-factor argument form, so multiplication factors need to be wrapped in array_size(). This patch replaces cases of: f2fs_kvzalloc(handle, a * b, gfp) with: f2fs_kvzalloc(handle, array_size(a, b), gfp) as well as handling cases of: f2fs_kvzalloc(handle, a * b * c, gfp) with: f2fs_kvzalloc(handle, array3_size(a, b, c), gfp) This does, however, attempt to ignore constant size factors like: f2fs_kvzalloc(handle, 4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ expression HANDLE; type TYPE; expression THING, E; @@ ( f2fs_kvzalloc(HANDLE, - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | f2fs_kvzalloc(HANDLE, - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression HANDLE; expression COUNT; typedef u8; typedef __u8; @@ ( f2fs_kvzalloc(HANDLE, - sizeof(u8) * (COUNT) + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(__u8) * (COUNT) + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(char) * (COUNT) + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(unsigned char) * (COUNT) + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(u8) * COUNT + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(__u8) * COUNT + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(char) * COUNT + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ expression HANDLE; type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * (COUNT_ID) + array_size(COUNT_ID, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * COUNT_ID + array_size(COUNT_ID, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * (COUNT_CONST) + array_size(COUNT_CONST, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * COUNT_CONST + array_size(COUNT_CONST, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * (COUNT_ID) + array_size(COUNT_ID, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * COUNT_ID + array_size(COUNT_ID, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * (COUNT_CONST) + array_size(COUNT_CONST, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * COUNT_CONST + array_size(COUNT_CONST, sizeof(THING)) , ...) ) // 2-factor product, only identifiers. @@ expression HANDLE; identifier SIZE, COUNT; @@ f2fs_kvzalloc(HANDLE, - SIZE * COUNT + array_size(COUNT, SIZE) , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression HANDLE; expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression HANDLE; expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( f2fs_kvzalloc(HANDLE, - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ expression HANDLE; identifier STRIDE, SIZE, COUNT; @@ ( f2fs_kvzalloc(HANDLE, - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products // when they're not all constants... @@ expression HANDLE; expression E1, E2, E3; constant C1, C2, C3; @@ ( f2fs_kvzalloc(HANDLE, C1 * C2 * C3, ...) | f2fs_kvzalloc(HANDLE, - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants. @@ expression HANDLE; expression E1, E2; constant C1, C2; @@ ( f2fs_kvzalloc(HANDLE, C1 * C2, ...) | f2fs_kvzalloc(HANDLE, - E1 * E2 + array_size(E1, E2) , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 21:28:35 +00:00
sit_i->sentries =
f2fs_kvzalloc(sbi, array_size(sizeof(struct seg_entry),
MAIN_SEGS(sbi)),
GFP_KERNEL);
if (!sit_i->sentries)
return -ENOMEM;
main_bitmap_size = f2fs_bitmap_size(MAIN_SEGS(sbi));
sit_i->dirty_sentries_bitmap = f2fs_kvzalloc(sbi, main_bitmap_size,
GFP_KERNEL);
if (!sit_i->dirty_sentries_bitmap)
return -ENOMEM;
#ifdef CONFIG_F2FS_CHECK_FS
bitmap_size = MAIN_SEGS(sbi) * SIT_VBLOCK_MAP_SIZE * 4;
#else
bitmap_size = MAIN_SEGS(sbi) * SIT_VBLOCK_MAP_SIZE * 3;
#endif
sit_i->bitmap = f2fs_kvzalloc(sbi, bitmap_size, GFP_KERNEL);
if (!sit_i->bitmap)
return -ENOMEM;
bitmap = sit_i->bitmap;
for (start = 0; start < MAIN_SEGS(sbi); start++) {
sit_i->sentries[start].cur_valid_map = bitmap;
bitmap += SIT_VBLOCK_MAP_SIZE;
sit_i->sentries[start].ckpt_valid_map = bitmap;
bitmap += SIT_VBLOCK_MAP_SIZE;
#ifdef CONFIG_F2FS_CHECK_FS
sit_i->sentries[start].cur_valid_map_mir = bitmap;
bitmap += SIT_VBLOCK_MAP_SIZE;
#endif
sit_i->sentries[start].discard_map = bitmap;
bitmap += SIT_VBLOCK_MAP_SIZE;
}
sit_i->tmp_map = f2fs_kzalloc(sbi, SIT_VBLOCK_MAP_SIZE, GFP_KERNEL);
if (!sit_i->tmp_map)
return -ENOMEM;
if (__is_large_section(sbi)) {
treewide: Use array_size in f2fs_kvzalloc() The f2fs_kvzalloc() function has no 2-factor argument form, so multiplication factors need to be wrapped in array_size(). This patch replaces cases of: f2fs_kvzalloc(handle, a * b, gfp) with: f2fs_kvzalloc(handle, array_size(a, b), gfp) as well as handling cases of: f2fs_kvzalloc(handle, a * b * c, gfp) with: f2fs_kvzalloc(handle, array3_size(a, b, c), gfp) This does, however, attempt to ignore constant size factors like: f2fs_kvzalloc(handle, 4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ expression HANDLE; type TYPE; expression THING, E; @@ ( f2fs_kvzalloc(HANDLE, - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | f2fs_kvzalloc(HANDLE, - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression HANDLE; expression COUNT; typedef u8; typedef __u8; @@ ( f2fs_kvzalloc(HANDLE, - sizeof(u8) * (COUNT) + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(__u8) * (COUNT) + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(char) * (COUNT) + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(unsigned char) * (COUNT) + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(u8) * COUNT + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(__u8) * COUNT + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(char) * COUNT + COUNT , ...) | f2fs_kvzalloc(HANDLE, - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ expression HANDLE; type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * (COUNT_ID) + array_size(COUNT_ID, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * COUNT_ID + array_size(COUNT_ID, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * (COUNT_CONST) + array_size(COUNT_CONST, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * COUNT_CONST + array_size(COUNT_CONST, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * (COUNT_ID) + array_size(COUNT_ID, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * COUNT_ID + array_size(COUNT_ID, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * (COUNT_CONST) + array_size(COUNT_CONST, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * COUNT_CONST + array_size(COUNT_CONST, sizeof(THING)) , ...) ) // 2-factor product, only identifiers. @@ expression HANDLE; identifier SIZE, COUNT; @@ f2fs_kvzalloc(HANDLE, - SIZE * COUNT + array_size(COUNT, SIZE) , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression HANDLE; expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression HANDLE; expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( f2fs_kvzalloc(HANDLE, - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | f2fs_kvzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ expression HANDLE; identifier STRIDE, SIZE, COUNT; @@ ( f2fs_kvzalloc(HANDLE, - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kvzalloc(HANDLE, - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products // when they're not all constants... @@ expression HANDLE; expression E1, E2, E3; constant C1, C2, C3; @@ ( f2fs_kvzalloc(HANDLE, C1 * C2 * C3, ...) | f2fs_kvzalloc(HANDLE, - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants. @@ expression HANDLE; expression E1, E2; constant C1, C2; @@ ( f2fs_kvzalloc(HANDLE, C1 * C2, ...) | f2fs_kvzalloc(HANDLE, - E1 * E2 + array_size(E1, E2) , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 21:28:35 +00:00
sit_i->sec_entries =
f2fs_kvzalloc(sbi, array_size(sizeof(struct sec_entry),
MAIN_SECS(sbi)),
GFP_KERNEL);
if (!sit_i->sec_entries)
return -ENOMEM;
}
/* get information related with SIT */
sit_segs = le32_to_cpu(raw_super->segment_count_sit) >> 1;
/* setup SIT bitmap from ckeckpoint pack */
sit_bitmap_size = __bitmap_size(sbi, SIT_BITMAP);
src_bitmap = __bitmap_ptr(sbi, SIT_BITMAP);
sit_i->sit_bitmap = kmemdup(src_bitmap, sit_bitmap_size, GFP_KERNEL);
if (!sit_i->sit_bitmap)
return -ENOMEM;
#ifdef CONFIG_F2FS_CHECK_FS
sit_i->sit_bitmap_mir = kmemdup(src_bitmap,
sit_bitmap_size, GFP_KERNEL);
if (!sit_i->sit_bitmap_mir)
return -ENOMEM;
sit_i->invalid_segmap = f2fs_kvzalloc(sbi,
main_bitmap_size, GFP_KERNEL);
if (!sit_i->invalid_segmap)
return -ENOMEM;
#endif
/* init SIT information */
sit_i->s_ops = &default_salloc_ops;
sit_i->sit_base_addr = le32_to_cpu(raw_super->sit_blkaddr);
sit_i->sit_blocks = sit_segs << sbi->log_blocks_per_seg;
sit_i->written_valid_blocks = 0;
sit_i->bitmap_size = sit_bitmap_size;
sit_i->dirty_sentries = 0;
sit_i->sents_per_block = SIT_ENTRY_PER_BLOCK;
sit_i->elapsed_time = le64_to_cpu(sbi->ckpt->elapsed_time);
sit_i->mounted_time = ktime_get_real_seconds();
init_rwsem(&sit_i->sentry_lock);
return 0;
}
static int build_free_segmap(struct f2fs_sb_info *sbi)
{
struct free_segmap_info *free_i;
unsigned int bitmap_size, sec_bitmap_size;
/* allocate memory for free segmap information */
free_i = f2fs_kzalloc(sbi, sizeof(struct free_segmap_info), GFP_KERNEL);
if (!free_i)
return -ENOMEM;
SM_I(sbi)->free_info = free_i;
bitmap_size = f2fs_bitmap_size(MAIN_SEGS(sbi));
free_i->free_segmap = f2fs_kvmalloc(sbi, bitmap_size, GFP_KERNEL);
if (!free_i->free_segmap)
return -ENOMEM;
sec_bitmap_size = f2fs_bitmap_size(MAIN_SECS(sbi));
free_i->free_secmap = f2fs_kvmalloc(sbi, sec_bitmap_size, GFP_KERNEL);
if (!free_i->free_secmap)
return -ENOMEM;
/* set all segments as dirty temporarily */
memset(free_i->free_segmap, 0xff, bitmap_size);
memset(free_i->free_secmap, 0xff, sec_bitmap_size);
/* init free segmap information */
free_i->start_segno = GET_SEGNO_FROM_SEG0(sbi, MAIN_BLKADDR(sbi));
free_i->free_segments = 0;
free_i->free_sections = 0;
spin_lock_init(&free_i->segmap_lock);
return 0;
}
static int build_curseg(struct f2fs_sb_info *sbi)
{
struct curseg_info *array;
int i;
treewide: Use array_size() in f2fs_kzalloc() The f2fs_kzalloc() function has no 2-factor argument form, so multiplication factors need to be wrapped in array_size(). This patch replaces cases of: f2fs_kzalloc(handle, a * b, gfp) with: f2fs_kzalloc(handle, array_size(a, b), gfp) as well as handling cases of: f2fs_kzalloc(handle, a * b * c, gfp) with: f2fs_kzalloc(handle, array3_size(a, b, c), gfp) This does, however, attempt to ignore constant size factors like: f2fs_kzalloc(handle, 4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ expression HANDLE; type TYPE; expression THING, E; @@ ( f2fs_kzalloc(HANDLE, - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | f2fs_kzalloc(HANDLE, - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression HANDLE; expression COUNT; typedef u8; typedef __u8; @@ ( f2fs_kzalloc(HANDLE, - sizeof(u8) * (COUNT) + COUNT , ...) | f2fs_kzalloc(HANDLE, - sizeof(__u8) * (COUNT) + COUNT , ...) | f2fs_kzalloc(HANDLE, - sizeof(char) * (COUNT) + COUNT , ...) | f2fs_kzalloc(HANDLE, - sizeof(unsigned char) * (COUNT) + COUNT , ...) | f2fs_kzalloc(HANDLE, - sizeof(u8) * COUNT + COUNT , ...) | f2fs_kzalloc(HANDLE, - sizeof(__u8) * COUNT + COUNT , ...) | f2fs_kzalloc(HANDLE, - sizeof(char) * COUNT + COUNT , ...) | f2fs_kzalloc(HANDLE, - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ expression HANDLE; type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( f2fs_kzalloc(HANDLE, - sizeof(TYPE) * (COUNT_ID) + array_size(COUNT_ID, sizeof(TYPE)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE) * COUNT_ID + array_size(COUNT_ID, sizeof(TYPE)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE) * (COUNT_CONST) + array_size(COUNT_CONST, sizeof(TYPE)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE) * COUNT_CONST + array_size(COUNT_CONST, sizeof(TYPE)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING) * (COUNT_ID) + array_size(COUNT_ID, sizeof(THING)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING) * COUNT_ID + array_size(COUNT_ID, sizeof(THING)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING) * (COUNT_CONST) + array_size(COUNT_CONST, sizeof(THING)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING) * COUNT_CONST + array_size(COUNT_CONST, sizeof(THING)) , ...) ) // 2-factor product, only identifiers. @@ expression HANDLE; identifier SIZE, COUNT; @@ f2fs_kzalloc(HANDLE, - SIZE * COUNT + array_size(COUNT, SIZE) , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression HANDLE; expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( f2fs_kzalloc(HANDLE, - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression HANDLE; expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( f2fs_kzalloc(HANDLE, - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | f2fs_kzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ expression HANDLE; identifier STRIDE, SIZE, COUNT; @@ ( f2fs_kzalloc(HANDLE, - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kzalloc(HANDLE, - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kzalloc(HANDLE, - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kzalloc(HANDLE, - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kzalloc(HANDLE, - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kzalloc(HANDLE, - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kzalloc(HANDLE, - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | f2fs_kzalloc(HANDLE, - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products // when they're not all constants... @@ expression HANDLE; expression E1, E2, E3; constant C1, C2, C3; @@ ( f2fs_kzalloc(HANDLE, C1 * C2 * C3, ...) | f2fs_kzalloc(HANDLE, - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants. @@ expression HANDLE; expression E1, E2; constant C1, C2; @@ ( f2fs_kzalloc(HANDLE, C1 * C2, ...) | f2fs_kzalloc(HANDLE, - E1 * E2 + array_size(E1, E2) , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 21:28:23 +00:00
array = f2fs_kzalloc(sbi, array_size(NR_CURSEG_TYPE, sizeof(*array)),
GFP_KERNEL);
if (!array)
return -ENOMEM;
SM_I(sbi)->curseg_array = array;
for (i = 0; i < NR_CURSEG_TYPE; i++) {
mutex_init(&array[i].curseg_mutex);
array[i].sum_blk = f2fs_kzalloc(sbi, PAGE_SIZE, GFP_KERNEL);
if (!array[i].sum_blk)
return -ENOMEM;
init_rwsem(&array[i].journal_rwsem);
array[i].journal = f2fs_kzalloc(sbi,
sizeof(struct f2fs_journal), GFP_KERNEL);
if (!array[i].journal)
return -ENOMEM;
array[i].segno = NULL_SEGNO;
array[i].next_blkoff = 0;
}
return restore_curseg_summaries(sbi);
}
static int build_sit_entries(struct f2fs_sb_info *sbi)
{
struct sit_info *sit_i = SIT_I(sbi);
struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_COLD_DATA);
struct f2fs_journal *journal = curseg->journal;
struct seg_entry *se;
struct f2fs_sit_entry sit;
int sit_blk_cnt = SIT_BLK_CNT(sbi);
unsigned int i, start, end;
unsigned int readed, start_blk = 0;
int err = 0;
f2fs: sanity check for total valid node blocks This patch enhances sanity check for SIT entries. syzbot hit the following crash on upstream commit 83beed7b2b26f232d782127792dd0cd4362fdc41 (Fri Apr 20 17:56:32 2018 +0000) Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal syzbot dashboard link: https://syzkaller.appspot.com/bug?extid=bf9253040425feb155ad syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5692130282438656 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5095924598571008 Kernel config: https://syzkaller.appspot.com/x/.config?id=1808800213120130118 compiler: gcc (GCC) 8.0.1 20180413 (experimental) IMPORTANT: if you fix the bug, please add the following tag to the commit: Reported-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com It will help syzbot understand when the bug is fixed. See footer for details. If you forward the report, please keep this part and the footer. F2FS-fs (loop0): invalid crc value F2FS-fs (loop0): Try to recover 1th superblock, ret: 0 F2FS-fs (loop0): Mounted with checkpoint version = d F2FS-fs (loop0): Bitmap was wrongly cleared, blk:9740 ------------[ cut here ]------------ kernel BUG at fs/f2fs/segment.c:1884! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4508 Comm: syz-executor0 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: 0018:ffff8801af526708 EFLAGS: 00010282 RAX: ffffed0035ea4cc0 RBX: ffff8801ad454f90 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82eeb87e RDI: ffffed0035ea4cb6 RBP: ffff8801af526760 R08: ffff8801ad4a2480 R09: ffffed003b5e4f90 R10: ffffed003b5e4f90 R11: ffff8801daf27c87 R12: ffff8801adb8d380 R13: 0000000000000001 R14: 0000000000000008 R15: 00000000ffffffff FS: 00000000014af940(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f06bc223000 CR3: 00000001adb02000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: allocate_data_block+0x66f/0x2050 fs/f2fs/segment.c:2663 do_write_page+0x105/0x1b0 fs/f2fs/segment.c:2727 write_node_page+0x129/0x350 fs/f2fs/segment.c:2770 __write_node_page+0x7da/0x1370 fs/f2fs/node.c:1398 sync_node_pages+0x18cf/0x1eb0 fs/f2fs/node.c:1652 block_operations+0x429/0xa60 fs/f2fs/checkpoint.c:1088 write_checkpoint+0x3ba/0x5380 fs/f2fs/checkpoint.c:1405 f2fs_sync_fs+0x2fb/0x6a0 fs/f2fs/super.c:1077 __sync_filesystem fs/sync.c:39 [inline] sync_filesystem+0x265/0x310 fs/sync.c:67 generic_shutdown_super+0xd7/0x520 fs/super.c:429 kill_block_super+0xa4/0x100 fs/super.c:1191 kill_f2fs_super+0x9f/0xd0 fs/f2fs/super.c:3030 deactivate_locked_super+0x97/0x100 fs/super.c:316 deactivate_super+0x188/0x1b0 fs/super.c:347 cleanup_mnt+0xbf/0x160 fs/namespace.c:1174 __cleanup_mnt+0x16/0x20 fs/namespace.c:1181 task_work_run+0x1e4/0x290 kernel/task_work.c:113 tracehook_notify_resume include/linux/tracehook.h:191 [inline] exit_to_usermode_loop+0x2bd/0x310 arch/x86/entry/common.c:166 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x457d97 RSP: 002b:00007ffd46f9c8e8 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000457d97 RDX: 00000000014b09a3 RSI: 0000000000000002 RDI: 00007ffd46f9da50 RBP: 00007ffd46f9da50 R08: 0000000000000000 R09: 0000000000000009 R10: 0000000000000005 R11: 0000000000000246 R12: 00000000014b0940 R13: 0000000000000000 R14: 0000000000000002 R15: 000000000000658e RIP: update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: ffff8801af526708 ---[ end trace f498328bb02610a2 ]--- Reported-and-tested-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+7d6d31d3bc702f566ce3@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+0a725420475916460f12@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-25 03:34:05 +00:00
block_t total_node_blocks = 0;
do {
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 16:20:41 +00:00
readed = f2fs_ra_meta_pages(sbi, start_blk, BIO_MAX_PAGES,
META_SIT, true);
start = start_blk * sit_i->sents_per_block;
end = (start_blk + readed) * sit_i->sents_per_block;
for (; start < end && start < MAIN_SEGS(sbi); start++) {
struct f2fs_sit_block *sit_blk;
struct page *page;
se = &sit_i->sentries[start];
page = get_current_sit_page(sbi, start);
if (IS_ERR(page))
return PTR_ERR(page);
sit_blk = (struct f2fs_sit_block *)page_address(page);
sit = sit_blk->entries[SIT_ENTRY_OFFSET(sit_i, start)];
f2fs_put_page(page, 1);
err = check_block_count(sbi, start, &sit);
if (err)
return err;
seg_info_from_raw_sit(se, &sit);
f2fs: sanity check for total valid node blocks This patch enhances sanity check for SIT entries. syzbot hit the following crash on upstream commit 83beed7b2b26f232d782127792dd0cd4362fdc41 (Fri Apr 20 17:56:32 2018 +0000) Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal syzbot dashboard link: https://syzkaller.appspot.com/bug?extid=bf9253040425feb155ad syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5692130282438656 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5095924598571008 Kernel config: https://syzkaller.appspot.com/x/.config?id=1808800213120130118 compiler: gcc (GCC) 8.0.1 20180413 (experimental) IMPORTANT: if you fix the bug, please add the following tag to the commit: Reported-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com It will help syzbot understand when the bug is fixed. See footer for details. If you forward the report, please keep this part and the footer. F2FS-fs (loop0): invalid crc value F2FS-fs (loop0): Try to recover 1th superblock, ret: 0 F2FS-fs (loop0): Mounted with checkpoint version = d F2FS-fs (loop0): Bitmap was wrongly cleared, blk:9740 ------------[ cut here ]------------ kernel BUG at fs/f2fs/segment.c:1884! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4508 Comm: syz-executor0 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: 0018:ffff8801af526708 EFLAGS: 00010282 RAX: ffffed0035ea4cc0 RBX: ffff8801ad454f90 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82eeb87e RDI: ffffed0035ea4cb6 RBP: ffff8801af526760 R08: ffff8801ad4a2480 R09: ffffed003b5e4f90 R10: ffffed003b5e4f90 R11: ffff8801daf27c87 R12: ffff8801adb8d380 R13: 0000000000000001 R14: 0000000000000008 R15: 00000000ffffffff FS: 00000000014af940(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f06bc223000 CR3: 00000001adb02000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: allocate_data_block+0x66f/0x2050 fs/f2fs/segment.c:2663 do_write_page+0x105/0x1b0 fs/f2fs/segment.c:2727 write_node_page+0x129/0x350 fs/f2fs/segment.c:2770 __write_node_page+0x7da/0x1370 fs/f2fs/node.c:1398 sync_node_pages+0x18cf/0x1eb0 fs/f2fs/node.c:1652 block_operations+0x429/0xa60 fs/f2fs/checkpoint.c:1088 write_checkpoint+0x3ba/0x5380 fs/f2fs/checkpoint.c:1405 f2fs_sync_fs+0x2fb/0x6a0 fs/f2fs/super.c:1077 __sync_filesystem fs/sync.c:39 [inline] sync_filesystem+0x265/0x310 fs/sync.c:67 generic_shutdown_super+0xd7/0x520 fs/super.c:429 kill_block_super+0xa4/0x100 fs/super.c:1191 kill_f2fs_super+0x9f/0xd0 fs/f2fs/super.c:3030 deactivate_locked_super+0x97/0x100 fs/super.c:316 deactivate_super+0x188/0x1b0 fs/super.c:347 cleanup_mnt+0xbf/0x160 fs/namespace.c:1174 __cleanup_mnt+0x16/0x20 fs/namespace.c:1181 task_work_run+0x1e4/0x290 kernel/task_work.c:113 tracehook_notify_resume include/linux/tracehook.h:191 [inline] exit_to_usermode_loop+0x2bd/0x310 arch/x86/entry/common.c:166 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x457d97 RSP: 002b:00007ffd46f9c8e8 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000457d97 RDX: 00000000014b09a3 RSI: 0000000000000002 RDI: 00007ffd46f9da50 RBP: 00007ffd46f9da50 R08: 0000000000000000 R09: 0000000000000009 R10: 0000000000000005 R11: 0000000000000246 R12: 00000000014b0940 R13: 0000000000000000 R14: 0000000000000002 R15: 000000000000658e RIP: update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: ffff8801af526708 ---[ end trace f498328bb02610a2 ]--- Reported-and-tested-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+7d6d31d3bc702f566ce3@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+0a725420475916460f12@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-25 03:34:05 +00:00
if (IS_NODESEG(se->type))
total_node_blocks += se->valid_blocks;
/* build discard map only one time */
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
if (is_set_ckpt_flags(sbi, CP_TRIMMED_FLAG)) {
memset(se->discard_map, 0xff,
SIT_VBLOCK_MAP_SIZE);
} else {
memcpy(se->discard_map,
se->cur_valid_map,
SIT_VBLOCK_MAP_SIZE);
sbi->discard_blks +=
sbi->blocks_per_seg -
se->valid_blocks;
}
if (__is_large_section(sbi))
get_sec_entry(sbi, start)->valid_blocks +=
se->valid_blocks;
}
start_blk += readed;
} while (start_blk < sit_blk_cnt);
down_read(&curseg->journal_rwsem);
for (i = 0; i < sits_in_cursum(journal); i++) {
unsigned int old_valid_blocks;
start = le32_to_cpu(segno_in_journal(journal, i));
f2fs: sanity check on sit entry syzbot hit the following crash on upstream commit 87ef12027b9b1dd0e0b12cf311fbcb19f9d92539 (Wed Apr 18 19:48:17 2018 +0000) Merge tag 'ceph-for-4.17-rc2' of git://github.com/ceph/ceph-client syzbot dashboard link: https://syzkaller.appspot.com/bug?extid=83699adeb2d13579c31e C reproducer: https://syzkaller.appspot.com/x/repro.c?id=5805208181407744 syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=6005073343676416 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=6555047731134464 Kernel config: https://syzkaller.appspot.com/x/.config?id=1808800213120130118 compiler: gcc (GCC) 8.0.1 20180413 (experimental) IMPORTANT: if you fix the bug, please add the following tag to the commit: Reported-by: syzbot+83699adeb2d13579c31e@syzkaller.appspotmail.com It will help syzbot understand when the bug is fixed. See footer for details. If you forward the report, please keep this part and the footer. F2FS-fs (loop0): Magic Mismatch, valid(0xf2f52010) - read(0x0) F2FS-fs (loop0): Can't find valid F2FS filesystem in 1th superblock F2FS-fs (loop0): invalid crc value BUG: unable to handle kernel paging request at ffffed006b2a50c0 PGD 21ffee067 P4D 21ffee067 PUD 21fbeb067 PMD 0 Oops: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 0 PID: 4514 Comm: syzkaller989480 Not tainted 4.17.0-rc1+ #8 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:build_sit_entries fs/f2fs/segment.c:3653 [inline] RIP: 0010:build_segment_manager+0x7ef7/0xbf70 fs/f2fs/segment.c:3852 RSP: 0018:ffff8801b102e5b0 EFLAGS: 00010a06 RAX: 1ffff1006b2a50c0 RBX: 0000000000000004 RCX: 0000000000000001 RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff8801ac74243e RBP: ffff8801b102f410 R08: ffff8801acbd46c0 R09: fffffbfff14d9af8 R10: fffffbfff14d9af8 R11: ffff8801acbd46c0 R12: ffff8801ac742a80 R13: ffff8801d9519100 R14: dffffc0000000000 R15: ffff880359528600 FS: 0000000001e04880(0000) GS:ffff8801dae00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffed006b2a50c0 CR3: 00000001ac6ac000 CR4: 00000000001406f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: f2fs_fill_super+0x4095/0x7bf0 fs/f2fs/super.c:2803 mount_bdev+0x30c/0x3e0 fs/super.c:1165 f2fs_mount+0x34/0x40 fs/f2fs/super.c:3020 mount_fs+0xae/0x328 fs/super.c:1268 vfs_kern_mount.part.34+0xd4/0x4d0 fs/namespace.c:1037 vfs_kern_mount fs/namespace.c:1027 [inline] do_new_mount fs/namespace.c:2517 [inline] do_mount+0x564/0x3070 fs/namespace.c:2847 ksys_mount+0x12d/0x140 fs/namespace.c:3063 __do_sys_mount fs/namespace.c:3077 [inline] __se_sys_mount fs/namespace.c:3074 [inline] __x64_sys_mount+0xbe/0x150 fs/namespace.c:3074 do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x443d6a RSP: 002b:00007ffd312813c8 EFLAGS: 00000297 ORIG_RAX: 00000000000000a5 RAX: ffffffffffffffda RBX: 0000000020000c00 RCX: 0000000000443d6a RDX: 0000000020000000 RSI: 0000000020000100 RDI: 00007ffd312813d0 RBP: 0000000000000003 R08: 0000000020016a00 R09: 000000000000000a R10: 0000000000000000 R11: 0000000000000297 R12: 0000000000000004 R13: 0000000000402c60 R14: 0000000000000000 R15: 0000000000000000 RIP: build_sit_entries fs/f2fs/segment.c:3653 [inline] RSP: ffff8801b102e5b0 RIP: build_segment_manager+0x7ef7/0xbf70 fs/f2fs/segment.c:3852 RSP: ffff8801b102e5b0 CR2: ffffed006b2a50c0 ---[ end trace a2034989e196ff17 ]--- Reported-and-tested-by: syzbot+83699adeb2d13579c31e@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-24 21:44:16 +00:00
if (start >= MAIN_SEGS(sbi)) {
f2fs_err(sbi, "Wrong journal entry on segno %u",
start);
err = -EFSCORRUPTED;
f2fs: sanity check on sit entry syzbot hit the following crash on upstream commit 87ef12027b9b1dd0e0b12cf311fbcb19f9d92539 (Wed Apr 18 19:48:17 2018 +0000) Merge tag 'ceph-for-4.17-rc2' of git://github.com/ceph/ceph-client syzbot dashboard link: https://syzkaller.appspot.com/bug?extid=83699adeb2d13579c31e C reproducer: https://syzkaller.appspot.com/x/repro.c?id=5805208181407744 syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=6005073343676416 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=6555047731134464 Kernel config: https://syzkaller.appspot.com/x/.config?id=1808800213120130118 compiler: gcc (GCC) 8.0.1 20180413 (experimental) IMPORTANT: if you fix the bug, please add the following tag to the commit: Reported-by: syzbot+83699adeb2d13579c31e@syzkaller.appspotmail.com It will help syzbot understand when the bug is fixed. See footer for details. If you forward the report, please keep this part and the footer. F2FS-fs (loop0): Magic Mismatch, valid(0xf2f52010) - read(0x0) F2FS-fs (loop0): Can't find valid F2FS filesystem in 1th superblock F2FS-fs (loop0): invalid crc value BUG: unable to handle kernel paging request at ffffed006b2a50c0 PGD 21ffee067 P4D 21ffee067 PUD 21fbeb067 PMD 0 Oops: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 0 PID: 4514 Comm: syzkaller989480 Not tainted 4.17.0-rc1+ #8 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:build_sit_entries fs/f2fs/segment.c:3653 [inline] RIP: 0010:build_segment_manager+0x7ef7/0xbf70 fs/f2fs/segment.c:3852 RSP: 0018:ffff8801b102e5b0 EFLAGS: 00010a06 RAX: 1ffff1006b2a50c0 RBX: 0000000000000004 RCX: 0000000000000001 RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff8801ac74243e RBP: ffff8801b102f410 R08: ffff8801acbd46c0 R09: fffffbfff14d9af8 R10: fffffbfff14d9af8 R11: ffff8801acbd46c0 R12: ffff8801ac742a80 R13: ffff8801d9519100 R14: dffffc0000000000 R15: ffff880359528600 FS: 0000000001e04880(0000) GS:ffff8801dae00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffed006b2a50c0 CR3: 00000001ac6ac000 CR4: 00000000001406f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: f2fs_fill_super+0x4095/0x7bf0 fs/f2fs/super.c:2803 mount_bdev+0x30c/0x3e0 fs/super.c:1165 f2fs_mount+0x34/0x40 fs/f2fs/super.c:3020 mount_fs+0xae/0x328 fs/super.c:1268 vfs_kern_mount.part.34+0xd4/0x4d0 fs/namespace.c:1037 vfs_kern_mount fs/namespace.c:1027 [inline] do_new_mount fs/namespace.c:2517 [inline] do_mount+0x564/0x3070 fs/namespace.c:2847 ksys_mount+0x12d/0x140 fs/namespace.c:3063 __do_sys_mount fs/namespace.c:3077 [inline] __se_sys_mount fs/namespace.c:3074 [inline] __x64_sys_mount+0xbe/0x150 fs/namespace.c:3074 do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x443d6a RSP: 002b:00007ffd312813c8 EFLAGS: 00000297 ORIG_RAX: 00000000000000a5 RAX: ffffffffffffffda RBX: 0000000020000c00 RCX: 0000000000443d6a RDX: 0000000020000000 RSI: 0000000020000100 RDI: 00007ffd312813d0 RBP: 0000000000000003 R08: 0000000020016a00 R09: 000000000000000a R10: 0000000000000000 R11: 0000000000000297 R12: 0000000000000004 R13: 0000000000402c60 R14: 0000000000000000 R15: 0000000000000000 RIP: build_sit_entries fs/f2fs/segment.c:3653 [inline] RSP: ffff8801b102e5b0 RIP: build_segment_manager+0x7ef7/0xbf70 fs/f2fs/segment.c:3852 RSP: ffff8801b102e5b0 CR2: ffffed006b2a50c0 ---[ end trace a2034989e196ff17 ]--- Reported-and-tested-by: syzbot+83699adeb2d13579c31e@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-24 21:44:16 +00:00
break;
}
se = &sit_i->sentries[start];
sit = sit_in_journal(journal, i);
old_valid_blocks = se->valid_blocks;
f2fs: sanity check for total valid node blocks This patch enhances sanity check for SIT entries. syzbot hit the following crash on upstream commit 83beed7b2b26f232d782127792dd0cd4362fdc41 (Fri Apr 20 17:56:32 2018 +0000) Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal syzbot dashboard link: https://syzkaller.appspot.com/bug?extid=bf9253040425feb155ad syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5692130282438656 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5095924598571008 Kernel config: https://syzkaller.appspot.com/x/.config?id=1808800213120130118 compiler: gcc (GCC) 8.0.1 20180413 (experimental) IMPORTANT: if you fix the bug, please add the following tag to the commit: Reported-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com It will help syzbot understand when the bug is fixed. See footer for details. If you forward the report, please keep this part and the footer. F2FS-fs (loop0): invalid crc value F2FS-fs (loop0): Try to recover 1th superblock, ret: 0 F2FS-fs (loop0): Mounted with checkpoint version = d F2FS-fs (loop0): Bitmap was wrongly cleared, blk:9740 ------------[ cut here ]------------ kernel BUG at fs/f2fs/segment.c:1884! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4508 Comm: syz-executor0 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: 0018:ffff8801af526708 EFLAGS: 00010282 RAX: ffffed0035ea4cc0 RBX: ffff8801ad454f90 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82eeb87e RDI: ffffed0035ea4cb6 RBP: ffff8801af526760 R08: ffff8801ad4a2480 R09: ffffed003b5e4f90 R10: ffffed003b5e4f90 R11: ffff8801daf27c87 R12: ffff8801adb8d380 R13: 0000000000000001 R14: 0000000000000008 R15: 00000000ffffffff FS: 00000000014af940(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f06bc223000 CR3: 00000001adb02000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: allocate_data_block+0x66f/0x2050 fs/f2fs/segment.c:2663 do_write_page+0x105/0x1b0 fs/f2fs/segment.c:2727 write_node_page+0x129/0x350 fs/f2fs/segment.c:2770 __write_node_page+0x7da/0x1370 fs/f2fs/node.c:1398 sync_node_pages+0x18cf/0x1eb0 fs/f2fs/node.c:1652 block_operations+0x429/0xa60 fs/f2fs/checkpoint.c:1088 write_checkpoint+0x3ba/0x5380 fs/f2fs/checkpoint.c:1405 f2fs_sync_fs+0x2fb/0x6a0 fs/f2fs/super.c:1077 __sync_filesystem fs/sync.c:39 [inline] sync_filesystem+0x265/0x310 fs/sync.c:67 generic_shutdown_super+0xd7/0x520 fs/super.c:429 kill_block_super+0xa4/0x100 fs/super.c:1191 kill_f2fs_super+0x9f/0xd0 fs/f2fs/super.c:3030 deactivate_locked_super+0x97/0x100 fs/super.c:316 deactivate_super+0x188/0x1b0 fs/super.c:347 cleanup_mnt+0xbf/0x160 fs/namespace.c:1174 __cleanup_mnt+0x16/0x20 fs/namespace.c:1181 task_work_run+0x1e4/0x290 kernel/task_work.c:113 tracehook_notify_resume include/linux/tracehook.h:191 [inline] exit_to_usermode_loop+0x2bd/0x310 arch/x86/entry/common.c:166 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x457d97 RSP: 002b:00007ffd46f9c8e8 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000457d97 RDX: 00000000014b09a3 RSI: 0000000000000002 RDI: 00007ffd46f9da50 RBP: 00007ffd46f9da50 R08: 0000000000000000 R09: 0000000000000009 R10: 0000000000000005 R11: 0000000000000246 R12: 00000000014b0940 R13: 0000000000000000 R14: 0000000000000002 R15: 000000000000658e RIP: update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: ffff8801af526708 ---[ end trace f498328bb02610a2 ]--- Reported-and-tested-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+7d6d31d3bc702f566ce3@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+0a725420475916460f12@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-25 03:34:05 +00:00
if (IS_NODESEG(se->type))
total_node_blocks -= old_valid_blocks;
err = check_block_count(sbi, start, &sit);
if (err)
break;
seg_info_from_raw_sit(se, &sit);
f2fs: sanity check for total valid node blocks This patch enhances sanity check for SIT entries. syzbot hit the following crash on upstream commit 83beed7b2b26f232d782127792dd0cd4362fdc41 (Fri Apr 20 17:56:32 2018 +0000) Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal syzbot dashboard link: https://syzkaller.appspot.com/bug?extid=bf9253040425feb155ad syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5692130282438656 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5095924598571008 Kernel config: https://syzkaller.appspot.com/x/.config?id=1808800213120130118 compiler: gcc (GCC) 8.0.1 20180413 (experimental) IMPORTANT: if you fix the bug, please add the following tag to the commit: Reported-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com It will help syzbot understand when the bug is fixed. See footer for details. If you forward the report, please keep this part and the footer. F2FS-fs (loop0): invalid crc value F2FS-fs (loop0): Try to recover 1th superblock, ret: 0 F2FS-fs (loop0): Mounted with checkpoint version = d F2FS-fs (loop0): Bitmap was wrongly cleared, blk:9740 ------------[ cut here ]------------ kernel BUG at fs/f2fs/segment.c:1884! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4508 Comm: syz-executor0 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: 0018:ffff8801af526708 EFLAGS: 00010282 RAX: ffffed0035ea4cc0 RBX: ffff8801ad454f90 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82eeb87e RDI: ffffed0035ea4cb6 RBP: ffff8801af526760 R08: ffff8801ad4a2480 R09: ffffed003b5e4f90 R10: ffffed003b5e4f90 R11: ffff8801daf27c87 R12: ffff8801adb8d380 R13: 0000000000000001 R14: 0000000000000008 R15: 00000000ffffffff FS: 00000000014af940(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f06bc223000 CR3: 00000001adb02000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: allocate_data_block+0x66f/0x2050 fs/f2fs/segment.c:2663 do_write_page+0x105/0x1b0 fs/f2fs/segment.c:2727 write_node_page+0x129/0x350 fs/f2fs/segment.c:2770 __write_node_page+0x7da/0x1370 fs/f2fs/node.c:1398 sync_node_pages+0x18cf/0x1eb0 fs/f2fs/node.c:1652 block_operations+0x429/0xa60 fs/f2fs/checkpoint.c:1088 write_checkpoint+0x3ba/0x5380 fs/f2fs/checkpoint.c:1405 f2fs_sync_fs+0x2fb/0x6a0 fs/f2fs/super.c:1077 __sync_filesystem fs/sync.c:39 [inline] sync_filesystem+0x265/0x310 fs/sync.c:67 generic_shutdown_super+0xd7/0x520 fs/super.c:429 kill_block_super+0xa4/0x100 fs/super.c:1191 kill_f2fs_super+0x9f/0xd0 fs/f2fs/super.c:3030 deactivate_locked_super+0x97/0x100 fs/super.c:316 deactivate_super+0x188/0x1b0 fs/super.c:347 cleanup_mnt+0xbf/0x160 fs/namespace.c:1174 __cleanup_mnt+0x16/0x20 fs/namespace.c:1181 task_work_run+0x1e4/0x290 kernel/task_work.c:113 tracehook_notify_resume include/linux/tracehook.h:191 [inline] exit_to_usermode_loop+0x2bd/0x310 arch/x86/entry/common.c:166 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x457d97 RSP: 002b:00007ffd46f9c8e8 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000457d97 RDX: 00000000014b09a3 RSI: 0000000000000002 RDI: 00007ffd46f9da50 RBP: 00007ffd46f9da50 R08: 0000000000000000 R09: 0000000000000009 R10: 0000000000000005 R11: 0000000000000246 R12: 00000000014b0940 R13: 0000000000000000 R14: 0000000000000002 R15: 000000000000658e RIP: update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: ffff8801af526708 ---[ end trace f498328bb02610a2 ]--- Reported-and-tested-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+7d6d31d3bc702f566ce3@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+0a725420475916460f12@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-25 03:34:05 +00:00
if (IS_NODESEG(se->type))
total_node_blocks += se->valid_blocks;
f2fs: fix to avoid NULL pointer dereference on se->discard_map https://bugzilla.kernel.org/show_bug.cgi?id=200951 These is a NULL pointer dereference issue reported in bugzilla: Hi, in the setup there is a SATA SSD connected to a SATA-to-USB bridge. The disc is "Samsung SSD 850 PRO 256G" which supports TRIM. There are four partitions: sda1: FAT /boot sda2: F2FS / sda3: F2FS /home sda4: F2FS The bridge is ASMT1153e which uses the "uas" driver. There is no TRIM pass-through, so, when mounting it reports: mounting with "discard" option, but the device does not support discard The USB host is USB3.0 and UASP capable. It is the one on RK3399. Given this everything works fine, except there is no TRIM support. In order to enable TRIM a new UDEV rule is added [1]: /etc/udev/rules.d/10-sata-bridge-trim.rules: ACTION=="add|change", ATTRS{idVendor}=="174c", ATTRS{idProduct}=="55aa", SUBSYSTEM=="scsi_disk", ATTR{provisioning_mode}="unmap" After reboot any F2FS write hangs forever and dmesg reports: Unable to handle kernel NULL pointer dereference Also tested on a x86_64 system: works fine even with TRIM enabled. same disc same bridge different usb host controller different cpu architecture not root filesystem Regards, Vicenç. [1] Post #5 in https://bbs.archlinux.org/viewtopic.php?id=236280 Unable to handle kernel NULL pointer dereference at virtual address 000000000000003e Mem abort info: ESR = 0x96000004 Exception class = DABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 Data abort info: ISV = 0, ISS = 0x00000004 CM = 0, WnR = 0 user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000626e3122 [000000000000003e] pgd=0000000000000000 Internal error: Oops: 96000004 [#1] SMP Modules linked in: overlay snd_soc_hdmi_codec rc_cec dw_hdmi_i2s_audio dw_hdmi_cec snd_soc_simple_card snd_soc_simple_card_utils snd_soc_rockchip_i2s rockchip_rga snd_soc_rockchip_pcm rockchipdrm videobuf2_dma_sg v4l2_mem2mem rtc_rk808 videobuf2_memops analogix_dp videobuf2_v4l2 videobuf2_common dw_hdmi dw_wdt cec rc_core videodev drm_kms_helper media drm rockchip_thermal rockchip_saradc realtek drm_panel_orientation_quirks syscopyarea sysfillrect sysimgblt fb_sys_fops dwmac_rk stmmac_platform stmmac pwm_bl squashfs loop crypto_user gpio_keys hid_kensington CPU: 5 PID: 957 Comm: nvim Not tainted 4.19.0-rc1-1-ARCH #1 Hardware name: Sapphire-RK3399 Board (DT) pstate: 00000005 (nzcv daif -PAN -UAO) pc : update_sit_entry+0x304/0x4b0 lr : update_sit_entry+0x108/0x4b0 sp : ffff00000ca13bd0 x29: ffff00000ca13bd0 x28: 000000000000003e x27: 0000000000000020 x26: 0000000000080000 x25: 0000000000000048 x24: ffff8000ebb85cf8 x23: 0000000000000253 x22: 00000000ffffffff x21: 00000000000535f2 x20: 00000000ffffffdf x19: ffff8000eb9e6800 x18: ffff8000eb9e6be8 x17: 0000000007ce6926 x16: 000000001c83ffa8 x15: 0000000000000000 x14: ffff8000f602df90 x13: 0000000000000006 x12: 0000000000000040 x11: 0000000000000228 x10: 0000000000000000 x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000000535f2 x6 : ffff8000ebff3440 x5 : ffff8000ebff3440 x4 : ffff8000ebe3a6c8 x3 : 00000000ffffffff x2 : 0000000000000020 x1 : 0000000000000000 x0 : ffff8000eb9e5800 Process nvim (pid: 957, stack limit = 0x0000000063a78320) Call trace: update_sit_entry+0x304/0x4b0 f2fs_invalidate_blocks+0x98/0x140 truncate_node+0x90/0x400 f2fs_remove_inode_page+0xe8/0x340 f2fs_evict_inode+0x2b0/0x408 evict+0xe0/0x1e0 iput+0x160/0x260 do_unlinkat+0x214/0x298 __arm64_sys_unlinkat+0x3c/0x68 el0_svc_handler+0x94/0x118 el0_svc+0x8/0xc Code: f9400800 b9488400 36080140 f9400f01 (387c4820) ---[ end trace a0f21a307118c477 ]--- The reason is it is possible to enable discard flag on block queue via UDEV, but during mount, f2fs will initialize se->discard_map only if this flag is set, once the flag is set after mount, f2fs may dereference NULL pointer on se->discard_map. So this patch does below changes to fix this issue: - initialize and update se->discard_map all the time. - don't clear DISCARD option if device has no QUEUE_FLAG_DISCARD flag during mount. - don't issue small discard on zoned block device. - introduce some functions to enhance the readability. Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Vicente Bergas <vicencb@gmail.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-09-03 19:52:17 +00:00
if (is_set_ckpt_flags(sbi, CP_TRIMMED_FLAG)) {
memset(se->discard_map, 0xff, SIT_VBLOCK_MAP_SIZE);
} else {
memcpy(se->discard_map, se->cur_valid_map,
SIT_VBLOCK_MAP_SIZE);
sbi->discard_blks += old_valid_blocks;
sbi->discard_blks -= se->valid_blocks;
}
if (__is_large_section(sbi)) {
get_sec_entry(sbi, start)->valid_blocks +=
se->valid_blocks;
get_sec_entry(sbi, start)->valid_blocks -=
old_valid_blocks;
}
}
up_read(&curseg->journal_rwsem);
f2fs: sanity check for total valid node blocks This patch enhances sanity check for SIT entries. syzbot hit the following crash on upstream commit 83beed7b2b26f232d782127792dd0cd4362fdc41 (Fri Apr 20 17:56:32 2018 +0000) Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal syzbot dashboard link: https://syzkaller.appspot.com/bug?extid=bf9253040425feb155ad syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5692130282438656 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5095924598571008 Kernel config: https://syzkaller.appspot.com/x/.config?id=1808800213120130118 compiler: gcc (GCC) 8.0.1 20180413 (experimental) IMPORTANT: if you fix the bug, please add the following tag to the commit: Reported-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com It will help syzbot understand when the bug is fixed. See footer for details. If you forward the report, please keep this part and the footer. F2FS-fs (loop0): invalid crc value F2FS-fs (loop0): Try to recover 1th superblock, ret: 0 F2FS-fs (loop0): Mounted with checkpoint version = d F2FS-fs (loop0): Bitmap was wrongly cleared, blk:9740 ------------[ cut here ]------------ kernel BUG at fs/f2fs/segment.c:1884! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4508 Comm: syz-executor0 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: 0018:ffff8801af526708 EFLAGS: 00010282 RAX: ffffed0035ea4cc0 RBX: ffff8801ad454f90 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82eeb87e RDI: ffffed0035ea4cb6 RBP: ffff8801af526760 R08: ffff8801ad4a2480 R09: ffffed003b5e4f90 R10: ffffed003b5e4f90 R11: ffff8801daf27c87 R12: ffff8801adb8d380 R13: 0000000000000001 R14: 0000000000000008 R15: 00000000ffffffff FS: 00000000014af940(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f06bc223000 CR3: 00000001adb02000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: allocate_data_block+0x66f/0x2050 fs/f2fs/segment.c:2663 do_write_page+0x105/0x1b0 fs/f2fs/segment.c:2727 write_node_page+0x129/0x350 fs/f2fs/segment.c:2770 __write_node_page+0x7da/0x1370 fs/f2fs/node.c:1398 sync_node_pages+0x18cf/0x1eb0 fs/f2fs/node.c:1652 block_operations+0x429/0xa60 fs/f2fs/checkpoint.c:1088 write_checkpoint+0x3ba/0x5380 fs/f2fs/checkpoint.c:1405 f2fs_sync_fs+0x2fb/0x6a0 fs/f2fs/super.c:1077 __sync_filesystem fs/sync.c:39 [inline] sync_filesystem+0x265/0x310 fs/sync.c:67 generic_shutdown_super+0xd7/0x520 fs/super.c:429 kill_block_super+0xa4/0x100 fs/super.c:1191 kill_f2fs_super+0x9f/0xd0 fs/f2fs/super.c:3030 deactivate_locked_super+0x97/0x100 fs/super.c:316 deactivate_super+0x188/0x1b0 fs/super.c:347 cleanup_mnt+0xbf/0x160 fs/namespace.c:1174 __cleanup_mnt+0x16/0x20 fs/namespace.c:1181 task_work_run+0x1e4/0x290 kernel/task_work.c:113 tracehook_notify_resume include/linux/tracehook.h:191 [inline] exit_to_usermode_loop+0x2bd/0x310 arch/x86/entry/common.c:166 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x457d97 RSP: 002b:00007ffd46f9c8e8 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000457d97 RDX: 00000000014b09a3 RSI: 0000000000000002 RDI: 00007ffd46f9da50 RBP: 00007ffd46f9da50 R08: 0000000000000000 R09: 0000000000000009 R10: 0000000000000005 R11: 0000000000000246 R12: 00000000014b0940 R13: 0000000000000000 R14: 0000000000000002 R15: 000000000000658e RIP: update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: ffff8801af526708 ---[ end trace f498328bb02610a2 ]--- Reported-and-tested-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+7d6d31d3bc702f566ce3@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+0a725420475916460f12@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-25 03:34:05 +00:00
if (!err && total_node_blocks != valid_node_count(sbi)) {
f2fs_err(sbi, "SIT is corrupted node# %u vs %u",
total_node_blocks, valid_node_count(sbi));
err = -EFSCORRUPTED;
f2fs: sanity check for total valid node blocks This patch enhances sanity check for SIT entries. syzbot hit the following crash on upstream commit 83beed7b2b26f232d782127792dd0cd4362fdc41 (Fri Apr 20 17:56:32 2018 +0000) Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal syzbot dashboard link: https://syzkaller.appspot.com/bug?extid=bf9253040425feb155ad syzkaller reproducer: https://syzkaller.appspot.com/x/repro.syz?id=5692130282438656 Raw console output: https://syzkaller.appspot.com/x/log.txt?id=5095924598571008 Kernel config: https://syzkaller.appspot.com/x/.config?id=1808800213120130118 compiler: gcc (GCC) 8.0.1 20180413 (experimental) IMPORTANT: if you fix the bug, please add the following tag to the commit: Reported-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com It will help syzbot understand when the bug is fixed. See footer for details. If you forward the report, please keep this part and the footer. F2FS-fs (loop0): invalid crc value F2FS-fs (loop0): Try to recover 1th superblock, ret: 0 F2FS-fs (loop0): Mounted with checkpoint version = d F2FS-fs (loop0): Bitmap was wrongly cleared, blk:9740 ------------[ cut here ]------------ kernel BUG at fs/f2fs/segment.c:1884! invalid opcode: 0000 [#1] SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) Modules linked in: CPU: 1 PID: 4508 Comm: syz-executor0 Not tainted 4.17.0-rc1+ #10 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: 0018:ffff8801af526708 EFLAGS: 00010282 RAX: ffffed0035ea4cc0 RBX: ffff8801ad454f90 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff82eeb87e RDI: ffffed0035ea4cb6 RBP: ffff8801af526760 R08: ffff8801ad4a2480 R09: ffffed003b5e4f90 R10: ffffed003b5e4f90 R11: ffff8801daf27c87 R12: ffff8801adb8d380 R13: 0000000000000001 R14: 0000000000000008 R15: 00000000ffffffff FS: 00000000014af940(0000) GS:ffff8801daf00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f06bc223000 CR3: 00000001adb02000 CR4: 00000000001406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: allocate_data_block+0x66f/0x2050 fs/f2fs/segment.c:2663 do_write_page+0x105/0x1b0 fs/f2fs/segment.c:2727 write_node_page+0x129/0x350 fs/f2fs/segment.c:2770 __write_node_page+0x7da/0x1370 fs/f2fs/node.c:1398 sync_node_pages+0x18cf/0x1eb0 fs/f2fs/node.c:1652 block_operations+0x429/0xa60 fs/f2fs/checkpoint.c:1088 write_checkpoint+0x3ba/0x5380 fs/f2fs/checkpoint.c:1405 f2fs_sync_fs+0x2fb/0x6a0 fs/f2fs/super.c:1077 __sync_filesystem fs/sync.c:39 [inline] sync_filesystem+0x265/0x310 fs/sync.c:67 generic_shutdown_super+0xd7/0x520 fs/super.c:429 kill_block_super+0xa4/0x100 fs/super.c:1191 kill_f2fs_super+0x9f/0xd0 fs/f2fs/super.c:3030 deactivate_locked_super+0x97/0x100 fs/super.c:316 deactivate_super+0x188/0x1b0 fs/super.c:347 cleanup_mnt+0xbf/0x160 fs/namespace.c:1174 __cleanup_mnt+0x16/0x20 fs/namespace.c:1181 task_work_run+0x1e4/0x290 kernel/task_work.c:113 tracehook_notify_resume include/linux/tracehook.h:191 [inline] exit_to_usermode_loop+0x2bd/0x310 arch/x86/entry/common.c:166 prepare_exit_to_usermode arch/x86/entry/common.c:196 [inline] syscall_return_slowpath arch/x86/entry/common.c:265 [inline] do_syscall_64+0x6ac/0x800 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x457d97 RSP: 002b:00007ffd46f9c8e8 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000457d97 RDX: 00000000014b09a3 RSI: 0000000000000002 RDI: 00007ffd46f9da50 RBP: 00007ffd46f9da50 R08: 0000000000000000 R09: 0000000000000009 R10: 0000000000000005 R11: 0000000000000246 R12: 00000000014b0940 R13: 0000000000000000 R14: 0000000000000002 R15: 000000000000658e RIP: update_sit_entry+0x1215/0x1590 fs/f2fs/segment.c:1882 RSP: ffff8801af526708 ---[ end trace f498328bb02610a2 ]--- Reported-and-tested-by: syzbot+bf9253040425feb155ad@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+7d6d31d3bc702f566ce3@syzkaller.appspotmail.com Reported-and-tested-by: syzbot+0a725420475916460f12@syzkaller.appspotmail.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2018-04-25 03:34:05 +00:00
}
return err;
}
static void init_free_segmap(struct f2fs_sb_info *sbi)
{
unsigned int start;
int type;
for (start = 0; start < MAIN_SEGS(sbi); start++) {
struct seg_entry *sentry = get_seg_entry(sbi, start);
if (!sentry->valid_blocks)
__set_free(sbi, start);
else
SIT_I(sbi)->written_valid_blocks +=
sentry->valid_blocks;
}
/* set use the current segments */
for (type = CURSEG_HOT_DATA; type <= CURSEG_COLD_NODE; type++) {
struct curseg_info *curseg_t = CURSEG_I(sbi, type);
__set_test_and_inuse(sbi, curseg_t->segno);
}
}
static void init_dirty_segmap(struct f2fs_sb_info *sbi)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
struct free_segmap_info *free_i = FREE_I(sbi);
unsigned int segno = 0, offset = 0;
unsigned short valid_blocks;
while (1) {
/* find dirty segment based on free segmap */
segno = find_next_inuse(free_i, MAIN_SEGS(sbi), offset);
if (segno >= MAIN_SEGS(sbi))
break;
offset = segno + 1;
valid_blocks = get_valid_blocks(sbi, segno, false);
if (valid_blocks == sbi->blocks_per_seg || !valid_blocks)
continue;
if (valid_blocks > sbi->blocks_per_seg) {
f2fs_bug_on(sbi, 1);
continue;
}
mutex_lock(&dirty_i->seglist_lock);
__locate_dirty_segment(sbi, segno, DIRTY);
mutex_unlock(&dirty_i->seglist_lock);
}
}
static int init_victim_secmap(struct f2fs_sb_info *sbi)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
unsigned int bitmap_size = f2fs_bitmap_size(MAIN_SECS(sbi));
dirty_i->victim_secmap = f2fs_kvzalloc(sbi, bitmap_size, GFP_KERNEL);
if (!dirty_i->victim_secmap)
return -ENOMEM;
return 0;
}
static int build_dirty_segmap(struct f2fs_sb_info *sbi)
{
struct dirty_seglist_info *dirty_i;
unsigned int bitmap_size, i;
/* allocate memory for dirty segments list information */
dirty_i = f2fs_kzalloc(sbi, sizeof(struct dirty_seglist_info),
GFP_KERNEL);
if (!dirty_i)
return -ENOMEM;
SM_I(sbi)->dirty_info = dirty_i;
mutex_init(&dirty_i->seglist_lock);
bitmap_size = f2fs_bitmap_size(MAIN_SEGS(sbi));
for (i = 0; i < NR_DIRTY_TYPE; i++) {
dirty_i->dirty_segmap[i] = f2fs_kvzalloc(sbi, bitmap_size,
GFP_KERNEL);
if (!dirty_i->dirty_segmap[i])
return -ENOMEM;
}
init_dirty_segmap(sbi);
return init_victim_secmap(sbi);
}
static int sanity_check_curseg(struct f2fs_sb_info *sbi)
{
int i;
/*
* In LFS/SSR curseg, .next_blkoff should point to an unused blkaddr;
* In LFS curseg, all blkaddr after .next_blkoff should be unused.
*/
for (i = 0; i < NO_CHECK_TYPE; i++) {
struct curseg_info *curseg = CURSEG_I(sbi, i);
struct seg_entry *se = get_seg_entry(sbi, curseg->segno);
unsigned int blkofs = curseg->next_blkoff;
if (f2fs_test_bit(blkofs, se->cur_valid_map))
goto out;
if (curseg->alloc_type == SSR)
continue;
for (blkofs += 1; blkofs < sbi->blocks_per_seg; blkofs++) {
if (!f2fs_test_bit(blkofs, se->cur_valid_map))
continue;
out:
f2fs_err(sbi,
"Current segment's next free block offset is inconsistent with bitmap, logtype:%u, segno:%u, type:%u, next_blkoff:%u, blkofs:%u",
i, curseg->segno, curseg->alloc_type,
curseg->next_blkoff, blkofs);
return -EFSCORRUPTED;
}
}
return 0;
}
/*
* Update min, max modified time for cost-benefit GC algorithm
*/
static void init_min_max_mtime(struct f2fs_sb_info *sbi)
{
struct sit_info *sit_i = SIT_I(sbi);
unsigned int segno;
down_write(&sit_i->sentry_lock);
sit_i->min_mtime = ULLONG_MAX;
for (segno = 0; segno < MAIN_SEGS(sbi); segno += sbi->segs_per_sec) {
unsigned int i;
unsigned long long mtime = 0;
for (i = 0; i < sbi->segs_per_sec; i++)
mtime += get_seg_entry(sbi, segno + i)->mtime;
mtime = div_u64(mtime, sbi->segs_per_sec);
if (sit_i->min_mtime > mtime)
sit_i->min_mtime = mtime;
}
sit_i->max_mtime = get_mtime(sbi, false);
up_write(&sit_i->sentry_lock);
}
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 16:20:41 +00:00
int f2fs_build_segment_manager(struct f2fs_sb_info *sbi)
{
struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
struct f2fs_sm_info *sm_info;
int err;
sm_info = f2fs_kzalloc(sbi, sizeof(struct f2fs_sm_info), GFP_KERNEL);
if (!sm_info)
return -ENOMEM;
/* init sm info */
sbi->sm_info = sm_info;
sm_info->seg0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
sm_info->main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
sm_info->segment_count = le32_to_cpu(raw_super->segment_count);
sm_info->reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
sm_info->ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
sm_info->main_segments = le32_to_cpu(raw_super->segment_count_main);
sm_info->ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
sm_info->rec_prefree_segments = sm_info->main_segments *
DEF_RECLAIM_PREFREE_SEGMENTS / 100;
if (sm_info->rec_prefree_segments > DEF_MAX_RECLAIM_PREFREE_SEGMENTS)
sm_info->rec_prefree_segments = DEF_MAX_RECLAIM_PREFREE_SEGMENTS;
if (!test_opt(sbi, LFS))
sm_info->ipu_policy = 1 << F2FS_IPU_FSYNC;
sm_info->min_ipu_util = DEF_MIN_IPU_UTIL;
sm_info->min_fsync_blocks = DEF_MIN_FSYNC_BLOCKS;
sm_info->min_seq_blocks = sbi->blocks_per_seg * sbi->segs_per_sec;
sm_info->min_hot_blocks = DEF_MIN_HOT_BLOCKS;
sm_info->min_ssr_sections = reserved_sections(sbi);
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
INIT_LIST_HEAD(&sm_info->sit_entry_set);
init_rwsem(&sm_info->curseg_lock);
if (!f2fs_readonly(sbi->sb)) {
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 16:20:41 +00:00
err = f2fs_create_flush_cmd_control(sbi);
if (err)
return err;
}
err = create_discard_cmd_control(sbi);
if (err)
return err;
err = build_sit_info(sbi);
if (err)
return err;
err = build_free_segmap(sbi);
if (err)
return err;
err = build_curseg(sbi);
if (err)
return err;
/* reinit free segmap based on SIT */
err = build_sit_entries(sbi);
if (err)
return err;
init_free_segmap(sbi);
err = build_dirty_segmap(sbi);
if (err)
return err;
err = sanity_check_curseg(sbi);
if (err)
return err;
init_min_max_mtime(sbi);
return 0;
}
static void discard_dirty_segmap(struct f2fs_sb_info *sbi,
enum dirty_type dirty_type)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
mutex_lock(&dirty_i->seglist_lock);
kvfree(dirty_i->dirty_segmap[dirty_type]);
dirty_i->nr_dirty[dirty_type] = 0;
mutex_unlock(&dirty_i->seglist_lock);
}
static void destroy_victim_secmap(struct f2fs_sb_info *sbi)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
kvfree(dirty_i->victim_secmap);
}
static void destroy_dirty_segmap(struct f2fs_sb_info *sbi)
{
struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
int i;
if (!dirty_i)
return;
/* discard pre-free/dirty segments list */
for (i = 0; i < NR_DIRTY_TYPE; i++)
discard_dirty_segmap(sbi, i);
destroy_victim_secmap(sbi);
SM_I(sbi)->dirty_info = NULL;
kvfree(dirty_i);
}
static void destroy_curseg(struct f2fs_sb_info *sbi)
{
struct curseg_info *array = SM_I(sbi)->curseg_array;
int i;
if (!array)
return;
SM_I(sbi)->curseg_array = NULL;
for (i = 0; i < NR_CURSEG_TYPE; i++) {
kvfree(array[i].sum_blk);
kvfree(array[i].journal);
}
kvfree(array);
}
static void destroy_free_segmap(struct f2fs_sb_info *sbi)
{
struct free_segmap_info *free_i = SM_I(sbi)->free_info;
if (!free_i)
return;
SM_I(sbi)->free_info = NULL;
kvfree(free_i->free_segmap);
kvfree(free_i->free_secmap);
kvfree(free_i);
}
static void destroy_sit_info(struct f2fs_sb_info *sbi)
{
struct sit_info *sit_i = SIT_I(sbi);
if (!sit_i)
return;
if (sit_i->sentries)
kvfree(sit_i->bitmap);
kvfree(sit_i->tmp_map);
kvfree(sit_i->sentries);
kvfree(sit_i->sec_entries);
kvfree(sit_i->dirty_sentries_bitmap);
SM_I(sbi)->sit_info = NULL;
kvfree(sit_i->sit_bitmap);
#ifdef CONFIG_F2FS_CHECK_FS
kvfree(sit_i->sit_bitmap_mir);
kvfree(sit_i->invalid_segmap);
#endif
kvfree(sit_i);
}
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 16:20:41 +00:00
void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi)
{
struct f2fs_sm_info *sm_info = SM_I(sbi);
if (!sm_info)
return;
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 16:20:41 +00:00
f2fs_destroy_flush_cmd_control(sbi, true);
destroy_discard_cmd_control(sbi);
destroy_dirty_segmap(sbi);
destroy_curseg(sbi);
destroy_free_segmap(sbi);
destroy_sit_info(sbi);
sbi->sm_info = NULL;
kvfree(sm_info);
}
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 16:20:41 +00:00
int __init f2fs_create_segment_manager_caches(void)
{
discard_entry_slab = f2fs_kmem_cache_create("discard_entry",
sizeof(struct discard_entry));
if (!discard_entry_slab)
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
goto fail;
discard_cmd_slab = f2fs_kmem_cache_create("discard_cmd",
sizeof(struct discard_cmd));
if (!discard_cmd_slab)
goto destroy_discard_entry;
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
sit_entry_set_slab = f2fs_kmem_cache_create("sit_entry_set",
sizeof(struct sit_entry_set));
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
if (!sit_entry_set_slab)
goto destroy_discard_cmd;
inmem_entry_slab = f2fs_kmem_cache_create("inmem_page_entry",
sizeof(struct inmem_pages));
if (!inmem_entry_slab)
goto destroy_sit_entry_set;
return 0;
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
destroy_sit_entry_set:
kmem_cache_destroy(sit_entry_set_slab);
destroy_discard_cmd:
kmem_cache_destroy(discard_cmd_slab);
destroy_discard_entry:
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
kmem_cache_destroy(discard_entry_slab);
fail:
return -ENOMEM;
}
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 16:20:41 +00:00
void f2fs_destroy_segment_manager_caches(void)
{
f2fs: refactor flush_sit_entries codes for reducing SIT writes In commit aec71382c681 ("f2fs: refactor flush_nat_entries codes for reducing NAT writes"), we descripte the issue as below: "Although building NAT journal in cursum reduce the read/write work for NAT block, but previous design leave us lower performance when write checkpoint frequently for these cases: 1. if journal in cursum has already full, it's a bit of waste that we flush all nat entries to page for persistence, but not to cache any entries. 2. if journal in cursum is not full, we fill nat entries to journal util journal is full, then flush the left dirty entries to disk without merge journaled entries, so these journaled entries may be flushed to disk at next checkpoint but lost chance to flushed last time." Actually, we have the same problem in using SIT journal area. In this patch, firstly we will update sit journal with dirty entries as many as possible. Secondly if there is no space in sit journal, we will remove all entries in journal and walk through the whole dirty entry bitmap of sit, accounting dirty sit entries located in same SIT block to sit entry set. All entry sets are linked to list sit_entry_set in sm_info, sorted ascending order by count of entries in set. Later we flush entries in set which have fewest entries into journal as many as we can, and then flush dense set with merged entries to disk. In this way we can use sit journal area more effectively, also we will reduce SIT update, result in gaining in performance and saving lifetime of flash device. In my testing environment, it shows this patch can help to reduce SIT block update obviously. virtual machine + hard disk: fsstress -p 20 -n 400 -l 5 sit page num cp count sit pages/cp based 2006.50 1349.75 1.486 patched 1566.25 1463.25 1.070 Our latency of merging op is small when handling a great number of dirty SIT entries in flush_sit_entries: latency(ns) dirty sit count 36038 2151 49168 2123 37174 2232 Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2014-09-04 10:13:01 +00:00
kmem_cache_destroy(sit_entry_set_slab);
kmem_cache_destroy(discard_cmd_slab);
kmem_cache_destroy(discard_entry_slab);
kmem_cache_destroy(inmem_entry_slab);
}